Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)
UvA-DARE (Digital Academic Repository)
What goes up must come down: glucose variability and glucose control in diabetes andcritical illness
Siegelaar, S.E.
Link to publication
Citation for published version (APA):Siegelaar, S. E. (2011). What goes up must come down: glucose variability and glucose control in diabetes andcritical illness.
General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.
Download date: 16 Dec 2020
Uitnodiging
Voor het bijwonen van de openbare verdediging
van mijn proefschrift
What goes up must come down
Glucose variability and glucose control in diabetes
and critical illness
Op vrijdag 17 juni 2011om 10:00 uur
in de AgnietenkapelOudezijds Voorburgwal 231
Amsterdam
U bent van harte uitgenodigd voor de receptie ter plaatse
na afloop van de verdediging
Sarah Siegelaar
Jan Olphert Vaillantlaan 157
1086 XZ Amsterdam
06-41430800
Paranimfen
Olivier Siegelaar
Margo Klomp
Wh
at goes u
p m
ust co
me d
own
Sarah E
. Siegelaar
What goes up must come downGlucose variability and glucose control in diabetes and critical illness
Sarah E. Siegelaar
What goes up must come down
Glucose variability and glucose control
in diabetes and critical illness
What goes up must come down: glucose variability and glucose control in diabetes and critical illnessAcademic thesis, University of Amsterdam, Amsterdam, the Netherlands
ISBN: 978-90-9026133-1
Author: Sarah E. SiegelaarLay-out: Barbara ten BrinkCover: Design by Barbara ten BrinkPrint: Gildeprint Drukkerijen, Enschede, The Netherlands
© S.E. Siegelaar, Amsterdam 2011All rights reserved. No part of this publication may be reproduced, stored, or transmitted in any form or by any means, without written permission of the author.
Printing of this thesis was financially supported by: Stichting Asklepios, Universiteit van Amsterdam, AstraZeneca BV, sanofi-aventis Netherlands BV, Boehringer Ingelheim BV.
What goes up must come downGlucose variability and glucose control
in diabetes and critical illness
ACADEMISCH PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam op gezag van de Rector Magnificus
prof.dr. D.C. van den Boomten overstaan van een door het college voor
promoties ingestelde commissie, in het openbaar te verdedigen in de Agnietenkapel
op vrijdag 17 juni 2011, te 10:00 uur
door
Sarah Elaine Siegelaar geboren te Heemstede
PROMOTIECOMMISSIE
Promotores: Prof. dr. J.B.L. Hoekstra
Co-promotor: Dr. J.H. de Vries
Overige leden: Prof. dr. E. Fliers
Prof. dr. E.S. Kilpatrick
Prof. dr. J.A. Romijn
Prof. dr. Y.M. Smulders
Prof. dr. D.F. Zandstra
Faculteit der Geneeskunde
Financial support by the Netherlands Heart Foundation and the Dutch Diabetes Research
Foundation for the publication of this thesis is gratefully acknowledged
CONTENTS
Chapter 1 Introduction
PART I Glucose variability
Chapter 2 Glucose variability; does it matter?
Chapter 3 Mild hyperglycaemia disturbs vascular homeostasis in
humans
Chapter 4 No relevant relationship between glucose variability and
oxidative stress in well regulated type 2 diabetes patients
Chapter 5 A randomised clinical trial comparing the effect of basal
insulin and inhaled mealtime insulin on glucose variability
and oxidative stress
Chapter 6 Glucose variability does not contribute to the development
of peripheral and autonomic neuropathy in type 1 diabetes:
data from the DCCT
Chapter 7 A decrease in glucose variability does not reduce
cardiovascular event rates in type 2 diabetes patients after
acute myocardial infarction: a reanalysis of the HEART2D
study
9
17
41
59
73
85
97
PART II Glucose control in critical illness
Chapter 8 Mean glucose during intensive care unit admission is related
to mortality by a U-shaped curve in surgical and medical
patients: a retrospective cohort study
Chapter 9 Accuracy and reliability of continuous glucose monitoring
in the intensive care unit: a head-to-head comparison of two
subcutaneous glucose sensors in cardiac surgery patients
Chapter 10 Microcirculation and its relation with continuous
subcutaneous glucose sensor accuracy in cardiac surgery
patients in the intensive care unit
Chapter 11 Special considerations for the diabetic patient in the intensive
care unit; targets for treatment and risks of hypoglycaemia
Chapter 12 The effect of diabetes on mortality in critically ill patients:
a systematic review and meta-analysis
Chapter 13 - Summary and future considerations
- Samenvatting en toekomstperspectief
- Authors’ affiliations
- List of publications
- Dankwoord
- Curriculum Vitae
107
123
137
153
171
197
Chapter 1
Introduction
Sarah E. Siegelaar
10
Glucose: essential for life but harmful in excess. Because of this paradox, the glycaemic
balance is a tightly regulated feed-back system in the healthy human body; unrestrained
increases in plasma glucose are prohibited by the action of insulin, and counterregulatory
hormones, such as glucagon, prevent plasma glucose to decrease to dangerously low
levels. As a result, plasma glucose levels in healthy humans almost never exceed 7.8
mmol/l1; the upper limit of what is considered normal. There are, however, conditions
where this equilibrium is being disturbed, resulting in chronic as well as acute
hyperglycaemia. Hyperglycaemia is known to induce endothelial damage, probably
because a mitochondrial glucose overload leads to formation of free oxygen radicals,
so-called oxidative stress 2, but the exact pathophysiology is not fully elucidated yet. This
thesis pictures different types of hyperglycaemia and examines whether hyperglycaemia
is always harmful and, as a result, should by all means be avoided.
Chronic hyperglycaemia is the defining feature of diabetes mellitus, as a result of an
absolute (type 1 diabetes mellitus) or relative (type 2 diabetes mellitus) shortage of
insulin. It affects the micro- and macrovasculature, causing damage to multiple organs 3;4. We know it is beneficial for patients with diabetes to decrease the high glucose
values and aim at HbA1c values below 7% for newly diagnosed patients and perhaps
somewhat less strict for patients with established disease and complications 5. This seems
straightforward, but unfortunately, it is not that easy. Patients with similar mean glucose
or HbA1c levels can have markedly different daily glucose profiles, with differences both
in number and duration of glucose excursions; so called glucose variability, the topic
of Part I of this thesis.
It has been suggested that high glucose variability induces vascular damage independent
from average glycaemia, which would have consequences for diabetes treatment 6. In
Chapter 2 an overview of different glucose variability measures is given and the current
literature regarding its effects in various populations is reviewed. To better understand
the effect of glucose variability, in Chapter 3 we studied whether the effects of mild
hyperglycaemia on vascular homeostasis are glucose-dependent or have a threshold above
which damage starts. The independent effect of glucose variability on oxidative stress
is investigated in Chapter 4 and Chapter 5, including type 2 diabetes patients treated
with oral glucose lowering drugs or insulin, respectively. A more clinical question is
addressed in Chapter 6 where datasets of the large Diabetes Control and Complications
Trial (DCCT) were reanalysed to assess the effect of glucose variability on the development
of neuropathy in type 1 diabetes. In Chapter 7 we describe the first trial that specifically
lowered glucose variability in type 2 diabetes patients assessing the effect on future
cardiovascular event rates.
Introduction
Ch
apte
r 1
11
That glucose homeostasis can be disturbed during critical illness is discussed in
Part II of this thesis. This so called “stress-hyperglycaemia” due to inflammatory and
neuro-endocrine derangements 7, is common in critically ill patients and associated
with mortality 8. In 2001, van den Berghe et al.9 startled the intensive care community
by publishing the results of a randomised controlled trial investigating the effect of
intensive insulin therapy on outcome, which showed that lowering plasma glucose to
normoglycaemic levels dramatically decreased mortality. But again, practice seemed not
to be as easy as proposed: the results from van den Berghe et al. could not be reproduced
and accumulating evidence suggests that the use of intensive insulin therapy is perhaps
even harmful in some patients 10. In Chapter 8 the optimal target range for mean glucose
during intensive care unit admission is explored further.
As heavy the disagreement is on whether strict or less-strict glycaemic control should
be applied in the intensive care unit, all are united that marked hyperglycaemia and
severe hypoglycaemia should be avoided. At present, time-consuming intermittent
blood sampling has to be performed to achieve glycaemic control, and moreover, no
information is available about glucose values in-between measurements. Subcutaneous
continuous glucose monitoring (CGM) could therefore be of value in achieving glycaemic
control. However, accuracy and reliability of the available systems seems uncertain in
critically ill patients 11;12. Therefore, we performed a head-to-head comparison of two
subcutaneous CGM systems in patients admitted to the intensive care unit after cardiac
surgery, presented in Chapter 9. The sometimes decreased accuracy of subcutaneous
CGM systems in the critically ill might result from alterations in the microcirculation
because needle-type glucose sensors measure glucose in the interstitial fluid and not
directly in blood. This has been investigated in Chapter 10, where the microcirculation
and its relation with continuous glucose sensor accuracy were studied in cardiac surgery
patients.
Not all patients with critical illness-induced hyperglycaemia are similar. More and more
the idea evolves that chronic hyperglycaemia in critically ill patients with diabetes is
pathophysiologically different from acute hyperglycaemia in those without previously
diagnosed diabetes. This could mean that treatment targets and strategies should differ
between these populations. The relation between hyperglycaemia and mortality as well as
the effect of intensive insulin therapy in critically ill patients with diabetes is discussed
in Chapter 11. Finally, in Chapter 12 the results of a meta-analysis looking at differences
in mortality between patients with and without diabetes when admitted to the intensive
care unit are shown, which allows us to put acquired knowledge into perspective.
In the end, this thesis is about glucose peaks and their consequences: must all what
goes up come down?
12
References1. Polonsky KS, Given BD, Hirsch LJ, et al (1988) Abnormal patterns of insulin secretion in non-insulin-dependent
diabetes mellitus. N Engl J Med 318: 1231-12392. Brownlee M (2001) Biochemistry and molecular cell biology of diabetic complications. Nature 414: 813-8203. UK Prospective Diabetes Study (UKPDS) Group (1998) Intensive blood-glucose control with sulphonylureas or
insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 352: 837-853
4. Nathan DM, Cleary PA, Backlund JY, et al (2005) Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med 353: 2643-2653
5. Patel A, MacMahon S, Chalmers J, et al (2008) Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med 358: 2560-2572
6. Monnier L, Mas E, Ginet C, et al (2006) Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 295: 1681-1687
7. Dungan KM, Braithwaite SS, Preiser JC (2009) Stress hyperglycaemia. Lancet 373: 1798-18078. Krinsley JS (2003) Association between hyperglycemia and increased hospital mortality in a heterogeneous
population of critically ill patients. Mayo Clin Proc 78: 1471-14789. Van den Berghe G, Wouters P, Weekers F, et al (2001) Intensive insulin therapy in the critically ill patients.
N Engl J Med 345: 1359-136710. Finfer S, Chittock DR, Su SY, et al (2009) Intensive versus conventional glucose control in critically ill patients.
N Engl J Med 360: 1283-129711. Logtenberg SJ, Kleefstra N, Snellen FT, et al (2009) Pre- and postoperative accuracy and safety of a real-time
continuous glucose monitoring system in cardiac surgical patients: a randomized pilot study. Diabetes Technol Ther 11: 31-37
12. Price GC, Stevenson K, Walsh TS (2008) Evaluation of a continuous glucose monitor in an unselected general intensive care population. Crit Care Resusc 10: 209-216
Introduction
Ch
apte
r 1
13
Part I
Glucose variability
Chapter 2
Glucose variability; does it matter?
Sarah E. Siegelaar, Frits Holleman, Joost B.L. Hoekstra and
J. Hans DeVries
Endocrine Reviews 2010; 31(2):171-182
18
Abstract
Overall lowering of glucose is of pivotal importance in the treatment of diabetes,
with proven beneficial effects on microvascular and macrovascular outcomes.
Still, patients with similar glycosylated haemoglobin levels and mean glucose
values can have markedly different daily glucose excursions. The role of this
glucose variability in pathophysiological pathways is the subject of debate. It is
strongly related to oxidative stress in in vitro, animal, and human studies in an
experimental setting. However, in real-life human studies including type 1 and
type 2 diabetes patients, there is neither a reproducible relation with oxidative
stress nor a correlation between short-term glucose variability and retinopathy,
nephropathy, or neuropathy. On the other hand, there is some evidence that long-
term glycaemic variability might be related to microvascular complications in
type 1 and 2 diabetes. Regarding mortality, a convincing relationship with short-
term glucose variability has only been demonstrated in nondiabetic, critically
ill patients. Also, glucose variability may have a role in the prediction of severe
hypoglycaemia. In this review, we first provide an overview of the various
methods to measure glucose variability. Second we review current literature
regarding glucose variability and its relation to oxidative stress, long-term
diabetic complications, and hypoglycaemia. Finally, we make recommendations
on whether and how to target glucose variability, concluding that at present we
lack both the compelling evidence and the means to target glucose variability
separately from all efforts to lower mean glucose while avoiding hypoglycaemia.
OutlineI. Introduction
II. Different methods for glucose variability measurement
III. Contribution of glucose variability to oxidative stress
IV. Contribution of glucose variability to diabetic complications and poor outcomes
in critically ill patients
V. Glucose variability as a predictor of severe hypoglycaemia
VI. Clinical recommendations
A. Should glucose variability be a target for intervention?
B. Available options to target glucose variability
VII. Conclusions and future perspectives
Glucose variability; does it matter?
Ch
apte
r 2
19
I. Introduction
Patients with similar mean glucose or glycosylated haemoglobin (HbA1c) values can have
markedly different daily glucose profiles, with differences both in number and duration
of glucose excursions. Hyperglycaemia is thought to induce oxidative stress and interfere
with normal endothelial function by overproduction of reactive oxygen species, which
results in diabetic complications through several molecular mechanisms 1,2 (Figure 1).
In addition, glucose variability might contribute to these processes as well. Since the
publication of the results of the Diabetes Control and Complications Trial (DCCT) in the
early 1990s 3,4, the topic of glucose variability as a contributor to diabetic complications
has been debated. It was suggested that glucose variability might explain the difference
in microvascular outcome between the intensively and conventionally treated type 1
diabetes patients with the same mean HbA1c throughout the trial 5. Although this
hypothesis was refuted recently by the statisticians of the DCCT/Epidemiology of Diabetes
Interventions and Complications (EDIC) themselves 6, subsequent hypotheses on the
relation of glucose variability to oxidative stress in type 2 diabetes patients and to
mortality in patients with stress hyperglycaemia have been postulated.
Figure 1 Potential mechanism by which hyperglycaemia-induced mitochondrial superoxide overproduction activates four pathways of hyperglycaemic damage (Reproduced with permission from M. Brownlee: Nature 414:813-820, 2001 1, © Macmillan Publishers, Ltd).
20
Glucose variability and lack of predictability are issues that diabetes patients and doctors
encounter in daily practice. In this review article, we will first provide an overview of the
various methods to measure glucose variability. Second, we review the current evidence
for the relation between glucose variability and oxidative stress, long-term diabetic
complications, and severe hypoglycaemia. Lastly, we will make recommendations for
treatment with regard to targeting glucose variability. We performed a structured literature
search using PubMed and EMBASE according to the PICO (patient, intervention, comparison
and outcome) method 7, including relevant literature published online up to March 2009.
Especially in type 2 diabetes, postprandial hyperglycaemia contributes to individual
glucose variability. However, because postprandial hyperglycaemia is different from glucose
variability as defined above, we will not discuss this further, other than to say that the
positive relationship between postprandial hyperglycaemia and cardiovascular risk supports
the possibility that glucose variability may be related to cardiovascular risk as well 8.
II. Measures of glucose variability
There are several methods to quantify glucose variability, but there is no universally accepted
“gold standard”. Table 1 describes the formulas underlying the different measures and
their characteristics. Most authors consider glycaemic variability as a standard of intraday
variation, reflecting the swings of blood glucose in a diabetic patient as a consequence of
diminished or absent autoregulation and the shortcomings of insulin therapy.
The easiest way to get an impression of the glucose variability in an individual patient is
to calculate the SD of glucose measurements and/or the coefficient of variation (CV), if
one wishes to correct for the mean. It is possible to calculate SD and CV from seven-point
glucose curves, facilitating their use in daily practice. On the other hand, when obtaining
seven-point glucose curves, certain peaks or nadirs will always be missed simply because
they occur between two measurements, making this method less accurate. Calculating
SD and CV from continuous glucose measurement (CGM) data seems preferable, but in
daily practice it is impossible to obtain CGM data from each individual patient. Also, the
extent to which CGM assessed SD differs from that calculated from seven-point profiles
has not, to our knowledge, been formally investigated.
In 1964, Schlichtkrull et al. 9 defined a new measure, the M-value, trying to quantify
glycaemic control of diabetes patients. It is a measure of the stability of the glucose
excursions in comparison with an “ideal” glucose value of 6.6 mmol/l (120 mg/dl),
developed using six self-monitored blood glucose (SMBG) values per 24 h in 20 patients with
type 1 diabetes. Later, other “ideal” glucose levels from 4.4 to 5.6 mmol/l (80 to 100 mg/dl)
Glucose variability; does it matter?
Ch
apte
r 2
21
Table 1 Formulas used in describing glucose variability
Variability measure
Formula Explanation of symbolsDiscriminating feature
SDxi = individual observationx– = mean of observationsk = number of observations
easy to determine, extensively used
CVs = standard deviationx– = mean of observations
easy to determine, SD corrected for mean
adjusted M-value
and
MGR = M-value for glucose readingsMw = correction factor for n <24GRt = glucose reading at time tIGV = ideal glucose valueti = time in minutes after start of observations of the ith observationGmax = maximum glucose readingGmin = minimum glucose reading
not a pure variability measure
MAGE if
λ = each blood glucose increase or decrease (nadir-peak or peak-nadir)n = number of observationsv = 1 SD of mean glucose for 24-hr period
used most extensively
CONGA
and
k* = number of observations where there is an observation n x 60 minutes agom = n x 60Dt = difference between glucose reading at time t and t minus n hours ago
specifically developed for CGM
MODD inter-day variation
SD, standard deviation; CV, coefficient of variation; MAGE, mean amplitude of glycaemic excursions; CONGA, continuous overall net glycaemic action; MODD, mean of daily differences; SMBG, self monitored blood glucose; CGM, continuous glucose monitoring. Units are in mmol/l or mg/dl depending on the unity of the glucose values measured. To convert glucose values from mg/dl to mmol/l multiply by 0.0555.
were proposed to obtain the best formula 10. In the final formula, choice of the ideal
glucose value is left up to the investigator, making it difficult to compare different
studies that use different ideal glucose values. The M-value is zero in healthy controls,
rising with increasing glycaemic variability or poorer glycaemic control, making it
difficult to distinguish between patients with either high mean glucose or high glucose
variability. Moreover, because hypoglycaemia has a greater impact on the M-value than
∑(xi – x–)2
k – 1
sx–
MGR + MW
whereMGR =
∑ 10log GRt
IGV
tk
t=t1
3
n
20MW = Gmax – Gmin
∑ λ–nλ v
∑tk*
t=t1
(Dt – D–)2
k* – 1
whereDt = GRt – GRt–m
∑tk*
t=t1
Dt
k*D– =
k*
GR1 – GRt–1440∑tk*
t=t1
22
hyperglycaemia, it is more a clinical than a mathematical indicator of glycaemic control.
In 1970, Service et al. 11 described a method that is is widely used nowadays: the mean
amplitude of glycaemic excursions (MAGE). Developed using hourly blood glucose
sampling for 48 hrs, this method generates a value for the variation around a mean
glucose value by summating the absolute rises or falls encountered in a day. The reference
point here is the mean glucose value rather than an arbitrarily chosen ideal value.
Arbitrarily, it ignores excursions of less than 1 SD. This may incorrectly disregard possibly
important smaller excursions. MAGE was originally defined from hourly glucose sampling
for 48 hrs in 14 patients. Thus, it has never been formally validated for calculation from
seven-point glucose profiles; neither do we know how the MAGE calculated from CGM
data corresponds to the originally developed value.
An intraday measurement of glycaemic variability specifically developed for use on CGM
data was proposed in 1999 by McDonnell et al. 12, i.e., continuous overlapping net glycaemic
action (CONGA-n). It is calculated as the SD of the summated differences between a current
observation and an observation n hours previously. Because CONGA does not require
arbitrary glucose cutoffs or arbitrarily defined rises and falls, it seems to be a more
objective manner to define glucose variability than M-value or MAGE. It is proposed for
CONGA-1, CONGA-2, and CONGA-4, but it is unknown which, if any, of these is preferable.
When examining glucose variability, the interday variation in blood glucose is also of
interest. In 1972, Molnar et al. 13 observed different day to day glucose patterns in patients
with a similar MAGE. They proposed the absolute mean of daily differences (MODD) as
a supplement to the MAGE and mean blood glucose. The MODD is the mean absolute
value of the differences between glucose values on 2 consecutive days at the same time.
In daily practice, eating habits play an interfering role because different mealtimes will
influence MODD. Developed using hourly blood sampling during 48 hrs, the validity of
its use on seven-point glucose curve data or CGM is unknown.
The most straightforward and easy way to measure interday variability is calculating the
SD of fasting blood glucose concentrations 14. However, it is more a measure of long-term
glucose variability because it takes values of at least 2 consecutive days to calculate. Above
all, fasting glucose variability neglects the variability in al other intraday glucose values.
Besides the commonly used measurements described above, several other methods have
been proposed that have not gained widespread use: the blood glucose rate of change,
computed for CGM, describing the magnitude of temporal fluctuations of blood glucose
levels using logarithmically transformed glucose data 15-17; the mean absolute difference
of consecutive glucose values, validated for SMBG curves 18; the “J”-index, defined as the
Glucose variability; does it matter?
Ch
apte
r 2
23
square of the mean plus SD of glucose measurements, excluding severe and persistent
hypoglycaemia, which is validated for SMBG curves 19; and the lability index, based on the
change in glucose levels over time 20. The complexity of the calculations 17 or substantial
similarity with other measures 18,19 probably underlie their limited use.
The MAGE is most commonly used for CGM data and SD/CV for SMBG curves. It has to be
mentioned that blood glucose values are seldom normally distributed, a mathematical
condition for use of the SD 16. In literature, this limitation is mostly ignored. However,
for SMBG strong correlations between variability measures, expressed as SD and mean
absolute difference, have been described 18. Using data from a previous study 21, we also
identified strong and significant correlations between cited variability measures (r =
0.63-0.93; P = 0.01; our unpublished data), suggesting a high degree of overlap between
the different measures when using CGM data. Because the SD correlates highly with
all other variability measures, it seems of little concern that the SD does not take the
number of glycaemic swings into account (Figure 2), whereas the calculation of MAGE,
MODD and CONGA is based on this. Whether calculating MAGE, MODD, CONGA or other
measures simultaneously helps to get additional insight in pathophysiological processes
needs further investigation. A further complication is that the time needed to reliably
assess a given standard of variability is not known. Preliminary results suggest this may
take several days of CGM measurements 22.
In addition to methods to quantify glucose variability derived from direct glucose
measurements, serum determination of 1,5-anhydroglucitol (1,5-AG) has been suggested
as a measure of glycaemic excursions 23. 1,5-AG is a polyol kept within stable limits
in subjects with glucose values in the normal range. Its reabsorption in the kidney
is inhibited by excessive excretion of urinary glucose; the higher the plasma glucose
concentration, the lower the plasma 1,5-AG concentration 24. Urinary glucose appears
at a plasma glucose concentration of approximately 8.8-10.5 mmol/l (160-190 mg/dl),
so despite a very quick response of this marker to changes in plasma glucose levels, it
seems of little use detecting glucose fluctuations below this range. Also, the correlation
between glucose variability and 1,5-AG is weak when HbA1c values are above 8% 25,26.
Measurement of 1,5-AG concentrations seems therefore only of use when looking at
hyperglycaemic excursions, i.e., postprandial hyperglycaemia in patients with an HbA1c
below 8%.
In summary, we suggest SD as the preferable method when quantifying variability from
CGM data since this is the easiest and best validated measure. Also, as further explained
below, SD was the measure used in the only field so far where a relation between glucose
variability and hard outcomes could be demonstrated, i.e., mortality in intensive care
unit (ICU) patients.
24
Figure 2 Two fictitious patients with identical mean glucose and SD, but different patterns of variability A and B are two different patients with different patterns of variability but the same mean glucose (6.0 mmol/l)
and SD (2.1 mmol/l). SD is calculated as the square root of the variance: , where xi is the sample
of the ith observation, x̄ the mean of all observations, and k the number of observations.
III. Contribution of glucose variability to oxidative stress
The current hypothesis about the link between hyperglycaemia and diabetic
complications suggests that the hyperglycaemia-driven formation of reactive oxygen
species enhances four mechanisms of tissue damage: the polyol pathway, the hexosamine
pathway, protein kinase C (PKC) activation, and formation of advanced glycation end-
products 1 (Figure 1). It should, however, be noted that at this time no human intervention
studies have been published that establish a causal relation between oxidative stress and
micro- or macrovascular complications 27. Moreover, daily antioxidant supplementation
does not reduce the risk of cardiovascular events and microvascular complications 28.
However strong the evidence supporting the concept of hyperglycaemia-induced oxidative
stress may be, the role of glycaemic variability in the formation of oxidative stress is
much more controversial. In vitro, animal and human studies in experimental settings
consistently report a deleterious effect of intermittent high glucose, either larger than or
as large as constant high glucose, despite less total glucose exposure, but these findings
cannot be reproduced in real-life human studies.
Quagliaro et al. 29 and Piconi et al. 30 demonstrated that intermittent high glucose
levels stimulate reactive oxygen species overproduction leading to increased cellular
apoptosis in human umbilical vein endothelial cells compared to a stable high glucose
environment. In these studies, three groups of cells were compared, each group receiving
a different fresh medium every 24 hrs for 14 days: a continuously normal glucose medium
∑(xi – x–)2
k – 1
Glucose variability; does it matter?
Ch
apte
r 2
25
(5 mmol/l), a continuously high glucose medium (20 mmol/l) and normal and high
glucose media alternating every 24 hrs (5 mmol/l and 20 mmol/l, respectively).
The effect of glycaemic variation vs. constant high glucose was also studied in cells of
the kidney. Takeuchi et al. 31 examined the effects of periodic changes in extracellular
glucose concentration on matrix production and proliferation of cultured rat mesangial
cells. Mesangial cell matrix production, measured as collagen III and IV protein
production and DNA level, was examined as a marker of cell proliferation and
nephropathy development 32,33. Three groups of cells were used, receiving a different
glucose medium every 24 hrs (5 mmol/l, alternating 5 and 25 mmol/l, and 25 mmol/l,
respectively) for 10 days. They reported a significantly larger collagen III and IV protein
and DNA production in the alternating glucose group compared with the continuous
high glucose group. No mechanism for these effects was demonstrated.
Jones et al. 34 investigated the effects of constant and intermittently increased glucose
on human kidney proximal tubule cells (PTC) and cortical fibroblasts (CF). In this study,
cell growth was assessed by thymidine uptake as an index of DNA synthesis, collagen
synthesis as a marker of extracellular matrix production, and protein content. They
exposed three groups of cells for 4 days to 6.1 mmol/l, 25 mmol/l or alternating 6.1 and
25 mmol/l glucose with daily medium change. Overall, the alternating glucose cells
showed larger thymidine uptake (PTC and CF) and more collagen synthesis (CF) than the
cells exposed to a stable high glucose medium. Nevertheless, no differences between
the high and intermittent glucose groups were found in cell protein content in both
PTC and CF. On the cytokine level, alternating high glucose activated more TGF-β1 and
IGF binding protein-3 than stable high glucose, suggesting more collagen synthesis,
potential apoptosis, and biological activity of IGF-1, which has been implicated in the
development of diabetic nephropathy 35,36.
Horváth et al. 37 built on these findings and compared the effect of nontreated diabetes
(continuous high glucose) with intermittently insulin-treated diabetes (oscillating
glucose) on the development of endothelial dysfunction in 19 male Wistar rats. After
10 days of insulin treatment, they monitored blood glucose levels every 6 hrs for 48
hrs in total. After these 48 hrs the rats were killed, and organs were harvested. Their
main finding was that the intermittently treated rats showed a significantly larger
impairment in endothelial function compared with the nontreated animals despite
lower total glucose exposure, with indications for an effect of the poly (ADP-ribose)
polymerase pathway.
The human studies performed are less consistent in their findings. Ceriello et al. 38
performed a normoinsulinemic hyperglycaemic glucose clamp study investigating the
relation between glucose variability, oxidative stress (assessed as plasma 3-nitrotyrosine
26
and 24-hr excretion rates of free 8-iso-prostaglandin F2α [8-iso-PGF2α]), and endothelial
function, measured by flow-mediated dilatation. Type 2 diabetic patients as well as
healthy controls were studied. They suggested that an oscillating glucose level has more
deleterious effects on endothelial function and enhances oxidative stress more than a
constant high glucose level. To mimic glucose variability, glycaemia was increased from
5 to 15 mmol/l and back every 6 hrs for 24 hrs. Stable hyperglycaemia conditions at 10
and 15 mmol/l for 24 hrs were the comparators.
It can be debated how many consecutive periods with alternating degrees of glycaemia
are necessary to reliably assess glycaemic variability rather than the effect of repeated
stimuli. From the field of pituitary function assessment, it is known that repeated
stimuli can result either in extinction of the response or exaggerated response 39. Also,
in everyday life, glucose swings of a patient with diabetes have a duration of less than 6
hrs and occur more frequently than the two 6-hr cycles used in the study performed by
Ceriello et al. 38. As already acknowledged in one of these manuscripts 31, the duration
of alternating glycaemia is also an important comment on the in vitro studies described
earlier since they alternate their glucose media every 24 hrs.
Three studies investigated the correlation between glucose variability assessed using
CGM and oxidative stress in a nonintervention design (Figure 3). These studies
calculated the MAGE to assess glucose variability and 24-hr urinary excretion rates of
8-iso-PGF2α to assess oxidative stress. The first study was performed by Monnier et al. 40
in 21 type 2 diabetes patients. They found a strong correlation between glucose
variability and oxidative stress (r = 0.86; P < 0.001). The second study was performed
by Wentholt et al. 21 in 25 type 1 diabetes patients. They expected to find an even stronger
correlation because of the greater glucose variability in type 1 diabetes patients, but
they could not confirm the findings of Monnier (r = 0.28; no P-value reported). A possible
explanation for this discrepancy is that the studies used a different method to quantify
8-iso-PGF2α excretion rates. Tandem mass spectrometry, used by Wentholt, is not
hampered by cross-reactivity of structurally (un)related components of 8-iso-PGF2α,
whereas the immunoassay used by Monnier is more susceptible to interference 40.
To solve this contradiction, our group reexamined this relationship in 24 type 2
diabetes patients quantifying urinary 8-iso-PGF2α excretion rates with tandem mass
spectrometry 41. We could not reproduce a relationship between glucose variability
and oxidative stress (r = 0.12; P = 0.53).
One intervention trial has been performed to assess the effect of lowering glucose
variability on oxidative stress 42. This crossover trial compared the effect of a basal insulin
regimen and a mealtime insulin regimen on glucose variability and oxidative stress in
type 2 diabetes using CGMS data (n = 40). Although glucose variability tended to be lower
Glucose variability; does it matter?
Ch
apte
r 2
27
Figure 3 Different relations between glucose variability and oxidative stress in type 2 and type 1 diabetesCorrelation between glucose variability, expressed as MAGE, and oxidative stress, expressed as urinary excretion rate of 8-iso-PGF2α, in type 2 (A) and type 1 (B) diabetes patients. A, r = 0.86; B, r = 0.26. (Panel A is reproduced with permission from L. Monnier et al.: JAMA 295:1681-1687, 2006 40, © American Medical Association. Panel B is reproduced from Figure 3 with kind permission from I.M. Wentholt et al.: Diabetologia 51:183-190, 2008 21, © Springer Science + Business media.)
(9%; P = nonsignificant) in the mealtime insulin group, no difference in oxidative stress
was found. If anything, there was more oxidative stress in the mealtime insulin group.
Again, no correlation between glucose variability and oxidative stress determined by 24-hr
urinary excretion rates of 8-iso-PGF2α was seen in these insulin-treated type 2 patients.
In this study, 8-iso-PGF2α was quantified by tandem mass spectrometry.
Summarizing, in vitro studies do show a relationship between glycaemic variability and
oxidative stress-induced apoptosis and renal cell proliferation in cultured human or rat
cells. These findings are confirmed in an animal study, but this relation could not be
consistently reproduced in human studies. Differences in duration and frequency of the
periods with alternating glycaemia as well as differences in methods used for oxidative
stress quantification are possible explanations for these discrepant findings.
IV. Contribution of glucose variability to diabetic complications and poor outcomes in critically ill patients
The most important issue for clinical practice is whether glucose variability contributes
to morbidity and mortality irrespective of the pathophysiological mechanism. This issue
was studied retrospectively in type 1 diabetes patients 6,43-47 and in critically ill patients
at the adult 48-50 and pediatric 51,52 ICU.
The DCCT, a randomised controlled trial which included 1,441 patients with type
1 diabetes, presented statistical models in 1995 suggesting a connection between
28
variability in blood glucose and the occurrence of microvascular complications 4. At
similar HbA1c levels throughout the study, patients from the conventionally treated
group were thought to be at higher risk for microvascular complications, particularly
progression of retinopathy, than those in the intensively treated group. Kilpatrick et al. 43,44
independently performed analyses of the data of the DCCT showing that the variability
in blood glucose around a patient’s mean value (SD) was not related to the development
or progression of either retinopathy or nephropathy in type 1 diabetes patients. More
than 10 years later, the DCCT statisticians themselves corrected their previous findings
and refuted the relation suggested earlier 6. As opposed to short-term glucose variability,
long-term fluctuations in glycaemia, expressed as HbA1c variability, may contribute to
the development of retinopathy and nephropathy in the DCCT group 45.
Bragd et al. 46 performed a prospective observational study in 100 type 1 diabetes patients,
collecting five-point self monitoring glucose data for 4 weeks. Data on the incidence and
prevalence of micro- and macrovascular complications as well as peripheral neuropathy
were obtained during an 11-year follow-up. This study confirmed the findings of the
studies mentioned previously in this section, finding no relationship between short-term
glucose variability measured as SD and microvascular complications. However, they
found that glucose variability was significantly related to the presence of peripheral
neuropathy and was a borderline predictor of its incidence (hazard ratio, 1.73; P = 0.07),
suggesting that the nervous system may be vulnerable to glycaemic variability. On the
other hand, recent analysis of the more extensive DCCT datasets did not show any
relation between glucose variability and the prevalence of diabetic peripheral as well
as autonomic neuropathy 47.
A single study in type 2 diabetes patients examined the effect of glucose variability on
retinopathy 53. The coefficient of variation of fasting plasma glucose was retrospectively
calculated in 130 patients without retinopathy at baseline with an average follow-up
of 5.2 years. The frequency of glucose measurements ranged from quarterly to yearly,
so long-term variability of fasting plasma glucose was assessed. The highest quartile of
variation in fasting plasma glucose contributed to diabetic retinopathy independently
from and in addition to HbA1c (odds ratio, 3.68; P = 0.049). This finding is in line with
the above-noted relation of long-term fluctuations in glycaemia to the development of
retinopathy in type 1 diabetes 45.
Recently, a randomised controlled trial was published comparing the effects of a prandial
and a basal insulin regimen with respect to cardiovascular outcomes in type 2 diabetes
patients after acute myocardial infarction (HEART2D Trial 54). The authors concluded
that a significant difference in postprandial glucose values, while achieving comparable
HbA1c values, was not associated with a difference in cardiovascular outcome. Glucose
Glucose variability; does it matter?
Ch
apte
r 2
29
variability was not separately assessed, but visual evaluation of the mean glucose profiles
collected during the study seems to show a difference in glucose variability in favour
of the prandial insulin group that did not translate into improved outcome (Figure 4).
Glucose variability has also been studied in critically ill patients. Three different groups
performed retrospective analyses of glucose variability as a predictor of mortality at the
adult ICU 48-50. All three groups concluded that glucose variability measured as SD was
a significant predictor of mortality in critically ill patients independently from severity
of illness. The finding that mortality significantly increased with variability in different
strata of mean glucose level 50, contributes to the suggestion that variability is a predictor
of mortality independent from mean glucose level (Figure 5). Egi et al. 49 performed a
subgroup analysis of patients with diabetes. Interestingly, in this group glucose control,
as assessed by the SD and mean glucose, displayed no relation with survival in contrast
to patients without diabetes. These results may suggest that patients with diabetes “get
accustomed” to fluctuating glucose levels, making them less devastating.
Figure 4 Glycaemic measures in a randomised controlled trial comparing a prandial with a basal insulin regimen (A) Mean (SD) HbA1c by treatment strategy. (B) Seven-point mean SMBG profiles at baseline (dotted line) and throughout the study (solid line) by treatment strategy. (Reproduced from Figure 2 with permission from I. Raz et al.: Diabetes Care 32:381-386, 2009 54 © American Diabetes Association in the format Journal via Copyright Clearance Center.
30
Figure 5 Hospital mortality related to mean glucose and glycaemic variability Q1, lowest quartile of glycaemic variability; Q4, highest quartile of glycaemic variability. To convert mean glucose from mg/dl to mmol/l, multiply by 0.0555. (Reproduced from Figure 1 with permission from J.S. Krinsley: Crit Care Med 36:3008-3013, 2008 50, © Wolters Kluwer Health.)
Not only in the adult ICU, but also in two different pediatric ICUs (PICUs), the influence
of glycaemic variability was studied. Wintergerst et al. 52 retrospectively reviewed all PICU
admissions of 1 year, excluding patients with a known diagnosis of diabetes mellitus
(n = 1094). Glucose variability was assessed as the mean of the absolute differences
between sequential glucose values divided by the differences in collection time. The
second retrospective cohort analysis was performed by Hirshberg et al. 51. They included
all PICU admissions with a length of stay of more than 24 hrs in 1 year, excluding patients
above 18 years of age, patients with known diabetes mellitus or when insulin therapy
was administered during PICU stay (n = 863). Glucose variability was described as a
patient who suffered from both hyperglycaemia (≥8.3 mmol/l) and hypoglycaemia (≤3.3
mmol/l) during PICU stay, which occurred in 6.8% of all patients. Both of these studies
confirmed the earlier described adult data showing that glucose variability is associated
with mortality and increased length of stay in this population, and they even show a
stronger association than hyperglycaemia, although only the latter study adjusted for
severity of illness in multivariate analysis.
Van den Berghe et al. 55 published a landmark trial in 2001 showing a dramatic 42%
relative reduction in mortality in the surgical ICU when blood glucose was normalised
to 4.4-6.1 mmol/l compared to 9.9-11.0 mmol/l. Recently, the purported benefits of tight
glycaemic control in the ICU have been challenged. The NICE-SUGAR study 56 showed that
intensive glucose control (4.5-6.0 mmol/l compared with <10 mmol/l) increased mortality
among adults in the ICU (odds ratio, 1.14; confidence interval, 1.02-1.28). One possible
Glucose variability; does it matter?
Ch
apte
r 2
31
explanation for these conflicting results is a differential effect on glucose variability
in these studies because this is strongly associated with mortality in this population 48-50. The results of the van den Berghe study showed a substantially lower SD in the
intensively treated group (SD of morning blood glucose, 19 vs. 33 mg/dl in the intensively
vs. conventionally treated groups, respectively) as opposed to the NICE-SUGAR study
where SD of morning blood glucose was equal in both groups, 25 and 26 mg/dl in the
intensively and conventionally treated groups, respectively.
We can draw a few conclusions from these studies. First, a relation between short-term
glucose variability and microvascular or neurological complications has not been proven
in type 1 diabetes patients and has not been investigated in type 2 diabetes. Second,
no studies have been performed investigating the influence of glucose variability on
macrovascular complications and death in either type 1 or type 2 diabetes patients,
but the HEART2D trial suggests that lowering glucose variability does not improve
cardiovascular outcome in type 2 diabetes patients after acute myocardial infarction.
In contrast, glucose variability is clearly related to mortality in critically ill patients
without diabetes, but intervention trials are still lacking.
V. Glucose variability as a predictor of severe hypoglycaemia
Hypoglycaemia is a complication of diabetes treatment with sometimes severe
consequences, such as seizures, accidents, coma, and death. The frequency of severe
hypoglycaemia increases exponentially when lowering blood glucose 3. Because lowering
blood glucose is the main goal of the treatment of diabetes, occurrence of hypoglycaemia
is a frequent problem. Much harm could be avoided if it were possible to predict severe
hypoglycaemia. Unfortunately, only a modest percentage of future severe hypoglycaemic
episodes can be predicted from known variables such as history of severe hypoglycaemia
and hypoglycaemia awareness 57,58.
In the search for possible predictors, glucose variability is a plausible candidate because
severe hypoglycaemia is preceded by blood glucose disturbances 59, and several studies
reported a decline in the occurrence of hypoglycaemia coinciding with lower glucose
variability 60-62. In 1994, Cox et al. 63 described glucose variability as a more powerful
predictor of future severe hypoglycaemia than HbA1c. In this study, 87 type 1 diabetes
patients prone to severe hypoglycaemia were included. Fifty SMBG readings were
collected during 2 to 3 weeks, and severe hypoglycaemia occurrence was recorded for
the subsequent 6 months.
32
The Diabetes Outcomes in Veterans Study (DOVES) 64 developed and subsequently validated
a model for predicting hypoglycaemia based on the idea that hypoglycaemia is more
likely if the mean blood glucose is low or if negative deviations from the mean are large.
The 195 insulin-treated type 2 diabetes patients included collected SMBG glucose values
four times daily for 8 weeks and had three follow-up visits in 1 year. In this model, the
risk of hypoglycaemia of any severity (blood glucose ≤3.33 mmol/l) appeared to be unique
to each subject and was as much related to glucose variability as to the mean glucose
value. The authors suggested that minimizing glucose variability is a plausible method
for offsetting the increased risk of hypoglycaemia associated with tight glycaemic control.
Unfortunately, how glycaemic variability could be targeted separately remains unclear.
Kilpatrick et al. 65 used the datasets of the DCCT to establish whether mean blood glucose
and/or glucose variability add to the predictive value of HbA1c for hypoglycaemia risk
in type 1 diabetes. This is the only study aiming to predict hypoglycaemia within 24 hrs
after SMBG collection. In this model, glucose variability, calculated as the SD of daily
blood glucose and MAGE, was independently predictive of hypoglycaemia just like mean
blood glucose. Concerning night-time hypoglycaemic events, variability at the end of
the day seemed predictive, suggesting that patients who suffer from this complication
could aim at reducing glycaemic fluctuations rather than let their blood glucose run
higher at bedtime.
From the above, it can be concluded that glucose variability is larger in patients with
diabetes who suffer from hypoglycaemia, in particular severe hypoglycaemia. Also,
glucose variability seems a predictor of severe hypoglycaemia, but it is more difficult to
answer the question whether it is an independent predictor of future hypoglycaemia.
None of the studies reviewed here performed an analysis to examine whether glucose
variability remains a predictor of hypoglycaemia when correcting for known predictors
such as history of severe hypoglycaemia and hypoglycaemia unawareness. It may be
useful to aim at lower glucose variability in those who suffer from severe hypoglycaemia
while at the same time trying to prevent a rise in mean blood glucose and HbA1c, but
a specific intervention trial is lacking.
VI. Clinical recommendations
A. Should glucose variability be a target for intervention?According to the reviewed literature, glucose variability could be investigated as a
separate treatment target in nondiabetic, critically ill patients, but with the introduction
of strict glucose regulation at the ICU, diminishing hyperglycaemic glucose excursions is
already a goal of therapy 55,66. Also, prevention and treatment of hypoglycaemia will be a
Glucose variability; does it matter?
Ch
apte
r 2
33
target anyway, although data on whether hypoglycaemia in the ICU is related to increased
mortality are conflicting 55,56,66-70. Glucose regulation with alertness for hypoglycaemia
should remain the intervention of choice until interventions specifically targeting
variability become available and are shown to result in improved outcome.
In insulin-treated diabetes patients with severe hypoglycaemia, it is often unavoidable
to reduce insulin doses to avoid subsequent episodes. However, a reduction in insulin
potentially leads to a deterioration of glucose control 71. Theoretically, therapies specifically
aiming to lower glucose variability might prevent severe hypoglycaemia while leaving
general glucose regulation unaffected. Again, trials supporting this notion are lacking.
As described above, there is little evidence to target glucose variability in general for
its limited effects on outcome. But one could think of other reasons to treat glucose
variability on an individual basis. It has been shown that within-day variability is an
independent predictor of the HbA1c achieved in type 1 diabetes patients receiving
multiple daily insulin therapy, with the largest variability correlating with the highest
HbA1c levels 72. One of the possible explanations for this is that glucose variability
reflects unexpected hypoglycaemic episodes due to a variable response to insulin
injections. This might lead to patient fear of hypoglycaemia and a possible deterioration
of glycaemic control when avoiding hypoglycaemia by resisting raising insulin dosage
or physical activity and a subsequent reduction in the patients’ quality of life 73. Clinical
investigations correlating glycaemic variability and quality of life are lacking, however.
Another important consequence of large intraindividual glucose variability is that the
patient has to perform SMBG more frequently, which is a burden for most diabetes
patients both from a psychological and a financial point of view.
B. Available options to target glucose variabilityAs for outpatients with type 1 or type 2 diabetes, long-acting insulin analogues seem
to improve glucose stability; treatment with long-acting analogues has been shown to
diminish hypoglycaemia and glucose variability 74-76. Prandial insulins, and even more
short-acting analogues, diminish postprandial hyperglycaemia and consequently glucose
variability specifically in type 2 diabetes patients 77,78. In comparison to the long-acting
analogue insulin glargine, the glucagon-like peptide-1 receptor agonist exenatide reduced
glucose variability with a similar reduction in HbA1c 79. Furthermore, compared to
multiple daily insulin injections, the use of continuous subcutaneous insulin infusion
is in type 1 diabetes associated with a decrease in glucose variability 60,80,81. Whether
diminishing glycaemic variability in these patient groups translates into improved
outcome is unknown, although it has been shown that patients with the largest glucose
variability benefit the most from switching from multiple daily insulin to continuous
subcutaneous insulin infusion, achieving significant lower HbA1c values 72.
34
VII. Conclusions and future perspectives
According to the literature we may conclude that glucose variability seems related to oxidative
stress in in vitro and animal studies and, although not consistently, in an experimental setting
in type 2 diabetes patients. In a clinical setting, glucose variability is related to mortality in
non-diabetic, critically ill subjects and is associated with (severe) hypoglycaemia in insulin-
treated diabetes patients. However, at this time there is no supportive evidence for targeting
glucose variability separately from mean glucose and/or HbA1c values.
There is no “gold” standard for determining glucose variability. Until added value for
other measures is shown, a simple SD seems the best way to quantify glucose variability.
CGM readings seem preferable to SMBG measurements to capture all variability, but no
data are available comparing these two methods in assessing glucose variability.
The only way to establish the utility of targeting glycaemic variability would be further
studies specifically aimed at lowering glucose variability, to investigate its influence on
hard outcomes.
Glucose variability; does it matter?
Ch
apte
r 2
35
References1. Brownlee M (2001) Biochemistry and molecular cell biology of diabetic complications. Nature 414:813-8202. Brownlee M (2005) The pathobiology of diabetic complications: a unifying mechanism. Diabetes 54:1615-16253. The Diabetes Control and Complications Trial Research Group (1993) The effect of intensive treatment of
diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 329:977-986
4. The Diabetes Control and Complications Trial Research Group (1995) The relationship of glycemic exposure (HbA1c) to the risk of development and progression of retinopathy in the diabetes control and complications trial. Diabetes 44:968-983
5. Brownlee M, Hirsch IB (2006) Glycemic variability: a hemoglobin A1c-independent risk factor for diabetic complications. JAMA 295:1707-1708
6. Lachin JM, Genuth S, Nathan DM, Zinman B, Rutledge BN, for the DCCT/EDIC Research Group (2008) Effect of glycemic exposure on the risk of microvascular complications in the diabetes control and complications trial--revisited. Diabetes 57:995-1001
7. Santos CMdC, Pimenta CAdM, Nobre MRC (2007) The PICO strategy for the research question construction and evidence search. Revista Latino-Americana de Enfermagem 15:508-511
8. Ceriello A, Piconi L, Quagliaro L, Wang Y, Schnabel CA, Ruggles JA, Gloster MA, Maggs DG, Weyer C (2005) Effects of pramlintide on postprandial plucose excursions and measures of oxidative stress in patients with type 1 diabetes. Diabetes Care 28:632-637
9. Schlichtkrull J, Munck O, Jersild M (1964) the M-value, an index of blood sugar control in diabetics. Acta Med Scand 177:95-102
10. Wojcicki JM (1993) Mathematical descriptions of the glucose control in diabetes therapy. Analysis of the Schlichtkrull “M”-value. Horm Metab Res 27:1-5
11. Service FJ, Molnar GD, Rosevear JW, Ackerman E, Gatewood LC, Taylor WF (1970) Mean amplitude of glycemic excursions, a measure of diabetic instability. Diabetes 19:644-655
12. McDonnell CM, Donath SM, Vidmar SI, Werther GA, Cameron FJ (2005) A novel approach to continuous glucose analysis utilizing glycemic variation. Diabetes Technology & Therapeutics 7:253-263
13. Molnar GD, Taylor WF, Ho MM (1972) Day-to-day variation of continuously monitored glycaemia. Diabetologia 8:342-348
14. Shima K, Tanaka R, Morishita S, Tarui S, Kumahara Y (1977) Studies on the etiology of “brittle diabetes”. Relationship between diabetic instability and insulinogenic reserve. Diabetes 26:717-725
15. McCall AL, Cox DJ, Crean J, Gloster M, Kovatchev BP (2006) A novel analytical method for assessing glucose variability: using CGMS in type 1 diabetes mellitus. Diabetes Technology & Therapeutics 8:644-653
16. Kovatchev BP, Cox DJ, Gonder-Frederick LA, Clarke W (1997) Symmetrization of the blood glucose measurement scale and its applications. Diabetes Care 20:1655-1658
17. Kovatchev BP, Clarke WL, Breton M, Brayman K, McCall A (2005) Quantifying temporal glucose variability in diabetes via continuous glucose monitoring: mathematical methods and clinical application. Diabetes Technology & Therapeutics 7:849-862
18. Moberg E, Kollind M, Lins P, Adamson U (1993) Estimation of blood-glucose variability in patients with insulin-dependent diabetes mellitus. Scand J Clin Lab Invest 53:507-514
19. Wojcicki JM (1995) “J”-index. A new proposition of the assessment of current glucose control in diabetic patients. Horm Metab Res 27:41-42
20. Ryan EA, Shandro T, Green K, Paty BW, Senior PA, Bigam D, Shapiro AMJ, Vantyghem MC (2004) Assessment of the severity of hypoglycemia and glycemic lability in type 1 diabetic subjects undergoing islet transplantation. Diabetes 53:955-962
21. Wentholt IME, Kulik W, Hoekstra JBL, DeVries JH (2008) Glucose fluctuations and activation of oxidative stress in type 1 diabetes patients. Diabetologia 51:183-190
22. Bugler J, Buell H (2008) Measuring glycemic variability - an estimation of data required based on a 90 day user study. Workshop of the AIDPIT study group 2008 (abstract S16)
23. Kishimoto M, Yamasaki Y, Kubota M, Arai K, Morishima T, Kawamori R, Kamada T (1995) 1,5-Anhydro-D-glucitol evaluates daily glycemic excursions in well-controlled NIDDM. Diabetes Care 18:1156-1159
24. Buse JB, Freeman JLR, Edelman SV, Jovanovic L, McGill JB (2003) Serum 1,5-anhydroglucitol (GlycoMark): a short-term glycemic marker. Diabetes Technology & Therapeutics 5:355-363
25. Mehta S, Bucey N, Volkening L, Svoren B, Wood J, Laffel L (2009) Utility of 1,5-anhydroglucitol (1,5-AG) in assessing glycemia among youth and young adults with type 1 diabetes (T1D). Diabetes Care 58 suppl 1:A463
26. Kuenen J, Borg R, Button E, Nathan D, Zheng H, Kostense P, Heine R, Diamant M (2009) 1,5-anhydroglucitol
36
concentrations and measures of glucose control and glucose variability in T1DM and T2DM Patients. Diabetes Care 58 suppl 1:A233-234
27. Jay D, Hitomi H, Griendling KK (2006) Oxidative stress and diabetic cardiovascular complications. Free Radical Biology and Medicine 40:183-192
28. Lonn E, Yusuf S, Hoogwerf B, Pogue J, Yi Q, Zinman B, Bosch J, Dagenais G, Mann JFE, Gerstein HC (2002) Effects of vitamin E on cardiovascular and microvascular outcomes in high-risk patients with diabetes: results of the HOPE Study and MICRO-HOPE substudy. Diabetes Care 25:1919-1927
29. Quagliaro L, Piconi L, Assaloni R, Martinelli L, Motz E, Ceriello A (2003) Intermittent high glucose enhances apoptosis related to oxidative stress in human umbilical vein endothelial cells: the role of protein kinase C and NAD(P)H-oxidase activation. Diabetes 52:2795-2804
30. Piconi L, Quagliaro L, Assaloni R, Da Ros R, Maier A, Zuodar G, Ceriello A (2006) Constant and intermittent high glucose enhances endothelial cell apoptosis through mitochondrial superoxide overproduction. Diabetes Metab Res Rev 22:198-203
31. Takeuchi A, Throckmorton DC, Brogden AP, Yoshizawa N, Rasmussen H, Kashgarian M (1995) Periodic high extracellular glucose enhances production of collagens III and IV by mesangial cells. Am J Physiol Renal Physiol 268:F13-F19
32. Mauer SM, Steffes MW, Ellis EN, Sutherland DE, Brown DM, Goetz FC (1984) Structural-functional relationships in diabetic nephropathy. J Clin Invest 74:1143-1155
33. Steffes MW, Bilous RW, Sutherland DE, Mauer SM (1992) Cell and matrix components of the glomerular mesangium in type I diabetes. Diabetes 41:679-684
34. Jones SC, Saunders HJ, Qi W, Pollock CA (1999) Intermittent high glucose enhances cell growth and collagen synthesis in cultured human tubulointerstitial cells. Diabetologia 42:1113-1119
35. Esposito C, Liu ZH, Striker GE et al. (1996) Inhibition of diabetic nephropathy by a GH antagonist: a molecular analysis. Kidney Int 50:506-514
36. Flyvbjerg A, Orskov H (1990) Kidney tissue insulin-like growth factor I and initial renal growth in diabetic rats: relation to severity of diabetes. Acta Endocrinol (Copenh) 122:374-378
37. Horvath EM, Benko R, Kiss L et al. (2009) Rapid ‘glycaemic swings’ induce nitrosative stress, activate poly(ADP-ribose) polymerase and impair endothelial function in a rat model of diabetes mellitus. Diabetologia 52:952-961
38. Ceriello A, Esposito K, Piconi L et al. (2008) Oscillating glucose is more deleterious to endothelial function and oxidative stress than mean glucose in normal and type 2 diabetic patients. Diabetes 57:1349-1354
39. de Vries JH, Noorda RJP, Voetberg GA, van der Veen EA (1991) Growth hormone release after the sequential use of growth hormone releasing factor and exercise. Horm Metab Res 23:397-398
40. Monnier L, Mas E, Ginet C, Michel F, Villon L, Cristol JP, Colette C (2006) Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 295:1681-1687
41. Siegelaar SE, Barwari T, Kulik W, Hoekstra JB, DeVries JH (2009) No relationship between glucose variability and oxidative stress in type 2 diabetes patients. Diabetologia 52 suppl 1:S500 (now published in J Diabetes Sci Tech 2011; 5:86-92)
42. Siegelaar SE, Kulik W, van Lenthe H, Mukherjee R, Hoekstra JB, DeVries JH (2009) A randomized controlled trial comparing the effect of basal insulin and inhaled mealtime insulin on glucose variability and oxidative stress. Diabetes, Obesity and Metabolism 11:709-714
43. Kilpatrick ES, Rigby AS, Atkin SL (2006) The effect of glucose variability on the risk of microvascular complications in type 1 diabetes. Diabetes Care 29:1486-1490
44. Kilpatrick ES, Rigby AS, Atkin SL (2007) Variability in the relationship between mean plasma glucose and HbA1c: implications for the assessment of glycemic control. Clin Chem 53:897-901
45. Kilpatrick ES, Rigby AS, Atkin SL (2008) A1c variability and the risk of microvascular complications in type 1 diabetes: data from the DCCT. Diabetes Care 31:2198-2202
46. Bragd J, Adamson U, Backlund LB, Lins PE, Moberg E, Oskarsson P (2008) Can glycaemic variability, as calculated from blood glucose self-monitoring, predict the development of complications in type1 diabetes over a decade? Diabetes & Metabolism 34:612-616
47. Siegelaar SE, Kilpatrick ES, Rigby AS, Atkin SL, Hoekstra JB, DeVries JH (2009) Glucose variability does not contribute to the development of peripheral and autonomic neuropathy in type 1 diabetes: data from the DCCT. Diabetologia 11:709-714
48. Dossett LA, Cao H, Mowery NT, Dortch MJ, Morris J, May AK (2008) Blood glucose variability is associated with mortality in the surgical intensive care unit. American Surgeon 74:679-685
49. Egi M, Bellomo R, Stachowski E, French C, Hart G (2006) Variability of blood glucose concentration and short-term mortality in critically ill patients. Anesthesiology 105:244-252
50. Krinsley J (2008) Glycemic variability: A strong independent predictor of mortality in critical ill patients.
Glucose variability; does it matter?
Ch
apte
r 2
37
Crit Care Med 36:3008-301351. Hirshberg E, Larsen G, van Duker H. (2008) Alterations in glucose homeostasis in the pediatric intensive care
unit: hyperglycemia and glucose variability are associated with increased mortality and morbidity. Pediatr Crit Care Med 9:361-366
52. Wintergerst KA, Buckingham B, Gandrud L, Wong BJ, Kache S, Wilson DM (2006) Association of hypoglycemia, hyperglycemia, and glucose variability with morbidity and death in the pediatric intensive care unit. Pediatrics 118:173-179
53. Gimeno-Orna JA, Castro-Alonso FJ, Boned-Juliani B, Lou-Arnal LM (2003) Fasting plasma glucose variability as a risk factor of retinopathy in Type 2 diabetic patients. J Diabetes Complications 17:78-81
54. Raz I, Wilson PWF, Strojek K et al. (2009) Effects of prandial versus fasting glycemia on cardiovascular outcomes in type 2 diabetes: the HEART2D trial. Diabetes Care 32:381-386
55. Van den Berghe G, Wouters P, Weekers F, et al. (2001) Intensive insulin therapy in critically ill patients. N Engl J Med 345:1359-1367
56. Finfer S, Chittock DR, Su SY, et al. (2009) Intensive versus conventional glucose control in critically ill patients. N Engl J Med 360:1283-1297
57. The Diabetes Control and Complications Trial Research Group (1997) Hypoglycemia in the diabetes control and complications trial. Diabetes 46:271-286
58. Gold AE, Frier BM, MacLeod KM, Deary IJ (1997) A structural equation model for predictors of severe hypoglycaemia in patients with insulin-dependent diabetes mellitus. Diabetic Medicine 14:309-315
59. Kovatchev BP, Cox DJ, Farhy LS, Straume M, Gonder-Frederick L, Clarke WL (2000) Episodes of severe hypoglycemia in type 1 diabetes are preceded and followed within 48 hours by measurable disturbances in blood glucose. J Clin Endocrinol Metab 85:4287-4292
60. Jeha G, Karaviti L, Anderson B, Smith E, Donaldson S, McGirk T, Haymond M (2005) Insulin pump therapy in preschool children with type 1 diabetes mellitus improves glycemic control and decreases glucose excursions and the risk of hypoglycemia. Diabetes Technology & Therapeutics 7:876-884
61. Kudva Y, Basu A, Jenkins G et al. (2007) Glycemic variation and hypoglycemia in patients with well-controlled type 1 diabetes on a multiple daily insulin injection program with use of glargine and ultralente as basal insulin. Endocr Pract 13:244-250
62. Saudek C, Duckworth WC, Giobbie-Hurder A et al. (2006) Implantable insulin pump vs multiple-dose insulin for non-insulin-dependent diabetes mellitus: a randomized clinical trial. Department of Veterans Affairs Implantable Insulin Pump Study Group. JAMA 276:1322-1327
63. Cox DJ, Kovatchev BP, Julian DM, Gonder-Frederick LA, Polonsky WH, Schlundt DG, Clarke WL (1994) Frequency of severe hypoglycemia in insulin-dependent diabetes mellitus can be predicted from self-monitoring blood glucose data. J Clin Endocrinol Metab 79:1659-1662
64. Murata GH, Hoffman RM, Shah JH, Wendel CS, Duckworth WC (2004) A probabilistic model for predicting hypoglycemia in type 2 diabetes mellitus: the diabetes outcomes in veterans study (DOVES). Arch Intern Med 164:1445-1450
65. Kilpatrick, Rigby, Goode, Atkin (2007) Relating mean blood glucose and glucose variability to the risk of multiple episodes of hypoglycaemia in type 1 diabetes. Diabetologia 50:2553-2561
66. Van den Berghe G, Wilmer A, Hermans G et al. (2006) Intensive insulin therapy in the medical ICU. N Engl J Med 354:449-461
67. Krinsley JS (2004) Effect of an intensive glucose management protocol on the mortality of critically ill adult patients. Mayo Clin Proc 79:992-1000
68. Vriesendorp TM, DeVries JH, van Santen S et al. (2006) Evaluation of short-term consequences of hypoglycemia in the intensive care unit. Crit Care Med 34:2714-2718
69. Brunkhorst FM, Engel C, Bloos F et al. The German Competence Network Sepsis (SepNet) (2008) Intensive insulin therapy and pentastarch resuscitation in severe sepsis. N Engl J Med 358:125-139
70. Devos P, Preiser JC, Melot C (2007) Impact of tight glucose control by intensive insulin therapy on ICU mortality and the rate of hypoglycemia: final results of the Glucontrol study. Intensive Care Medicine 33: Suppl 2:S189
71. The Diabetes Control and Complications Trial Research Group (1996) The absence of a glycemic threshold for the development of long-term complications: the perspective of the Diabetes Control and Complications Trial. Diabetes 45:1289-1298
72. Pickup J, Kidd J, Burmiston S, Yemane N (2006) Determinants of glycaemic control in type 1 diabetes during intensified therapy with multiple daily insulin injections or continuous subcutaneous insulin infusion: importance of blood glucose variability. Diabetes/Metabolism Research and Reviews 22:232-237
73. Hartman I (2008) Insulin analogs: impact on treatment success, satisfaction, quality of life, and adherence. Clinical Medicine & Research 6:54-67
74. Riddle MC, Rosenstock J, Gerich J (2003) The treat-to-target trial: randomized addition of glargine or human
38
NPH insulin to oral therapy of type 2 diabetic patients. Diabetes Care 26:3080-308675. Hermansen K, Davies M, Derezinski T, Martinez Ravn G, Clauson P, Home P, on behalf of the Levemir Treat-to-
Target Study Group (2006) A 26-Week, randomized, parallel, treat-to-target trial comparing insulin detemir with NPH insulin as add-on therapy to oral glucose-lowering drugs in insulin-naive people with type 2 diabetes. Diabetes Care 29:1269-1274
76. White NH, Chase HP, Arslanian S, Tamborlane WV (2009) Comparison of glycemic variability associated with insulin glargine and intermediate-acting insulin when used as the basal component of multiple daily injections for adolescents with type 1 diabetes. Diabetes Care 32:387-93
77. Anderson JJ, Brunelle R, Keohane P, Koivisto V, Trautmann M, Vignati L, DiMarchi R (1997) Mealtime treatment with insulin analog improves postprandial hyperglycemia and hypoglycemia in patients with non-insulin-dependent diabetes mellitus. Multicenter Insulin Lispro Study Group. Arch Intern Med 157:1249-1255
78. Kang S, Creagh FM, Peters JR, Brange J, Volund A, Owens DR (1991) Comparison of subcutaneous soluble human insulin and insulin analogues (AspB9, GluB27; AspB10; AspB28) on meal-related plasma glucose excursions in type I diabetic subjects. Diabetes Care 14:571-577
79. McCall AL, Cox DJ, Brodows R, Crean J, Johns D, Kovatchev B (2009) Reduced daily risk of glycemic variability: comparison of exenatide with insulin glargine. Diabetes Technology & Therapeutics 11:339-344
80. Bruttomesso D, Crazzolara D, Maran A et al. (2008) In type 1 diabetic patients with good glycaemic control, blood glucose variability is lower during continuous subcutaneous insulin infusion than during multiple daily injections with insulin glargine. Diabet Med 25:326-332
81. Alemzadeh R, Palma-Sisto P, Holzum M, Parton E, Kicher J (2007) Continuous subcutaneous insulin infusion attenuated glycemic instability in preschool children with type 1 diabetes mellitus. Diabetes Technol Ther 9:339-347
Glucose variability; does it matter?
Ch
apte
r 2
39
Chapter 3
Mild hyperglycaemia disturbs vascular homeostasis in humans
Sarah E. Siegelaar, Bregtje A. Lemkes, Max Nieuwdorp, Wim Kulik,
Joost C. Meijers, Joost B.L. Hoekstra and Frits Holleman
Submitted for publication
42
Abstract
Hyperglycaemia induces oxidative stress, disturbs endothelial function, damages
the endothelial glycocalyx and causes a prothrombotic shift in coagulation and
fibrinolysis. Little is known about the exact blood glucose level necessary to
start these processes. The aim of this study was to determine at which level of
glycaemia these changes occur. A stepwise hyperglycaemic clamp was performed
in eleven healthy human males at blood glucose (BG) levels of 6, 8 and 10 mmol/l
for two hours each while suppressing endogenous insulin release. Oxidative
stress, assessed by malondialdehyde, showed a gradual increase highly correlating
with BG. Coagulation, assessed by prothrombin fragments F1+2, significantly
increased at 6 mmol/l and was followed by an increase in both plasmin-
antiplasmin complexes and d-dimer levels at 8 mmol/l, indicating fibrinolysis
activation. The endothelial glycocalyx, measured by hyaluronic acid levels,
showed no relevant change during the clamp. Hyaluronidase showed a gradual
decrease indicating increased hyaluronidase substrate binding by shedding of
glycocalyx constituents, significant at 10 mmol/l. There was no indication of a
cumulative effect of glycaemia over time on any of the parameters. These data
indicate that oxidative stress as well as coagulation activation already starts at
near normal BG levels, while endothelial glycocalyx changes occur at 10 mmol/l.
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
43
Introduction
Patients with diabetes mellitus are at high risk of developing cardiovascular disease. A
combination of accelerated atherosclerosis and a shift towards a pro-coagulant state leads to
atherothrombotic events in nearly two thirds of all patients with diabetes 1. Hyperglycaemia,
its defining feature, has been shown to cause both endothelial dysfunction, a precursor
of atherosclerosis, and activation of the coagulation system 2;3. Endothelial dysfunction is
paired with damage to the endothelial glycocalyx, the protective layer of proteoglycans and
glycosaminoglycans lining the luminal side of all blood vessels. Hyperglycaemia induced
disruption of the endothelial glycocalyx results in a pro-atherogenic state, characterised
by increased vascular permeability, coagulation activation and increased cellular adhesion
and migration 4. The formation of reactive oxygen species (ROS) is an important mechanism
by which hyperglycaemia leads to endothelial (glycocalyx) damage, since high doses of
anti-oxidants are able to attenuate this damage 5. Furthermore, hyperglycaemia induced
ROS formation may also affect the coagulation system, by influencing gene transcription
of coagulation and fibrinolytic factors 6;7.
It is unclear, however, at which glucose level these vascular changes first occur and
whether this is an on-off phenomenon with a threshold or a continuous relationship. This
distinction is also of importance given the recent debate about the impact of glycaemic
variability on the development of complications in patients with diabetes 8;9. Some have
argued that a high variation in blood glucose levels throughout the day has a greater
impact on pro-atherogenic processes than a stable high glucose 10-12. If so, a threshold
phenomenon should exist for these processes, since a dose-dependent effect would lead
to comparable outcome when the mean blood glucose levels are equal.
Most studies investigating the effects of hyperglycaemia on vascular homeostasis have
described effects of a glucose level of 10 mmol/l or higher 2;5, but epidemiological studies
suggest that vascular damage actually starts at near normal glucose levels. Even in the
lower glucose ranges a linear relationship between HbA1c, fasting plasma glucose and
vascular complications of diabetes was demonstrated in patients with both type 1 and
type 2 diabetes 13;14. Moreover, impaired glucose tolerance and impaired fasting glucose,
both representing only mildly elevated glucose levels, already carry an increased risk
for macrovascular disease 15.
In the present study we describe the effects of only mildly elevated glucose levels on
oxidative stress, the endothelial glycocalyx and the thrombotic system in healthy males,
studied by performing a stepwise glucose clamp while suppressing endogenous insulin
levels.
44
Results
ProtocolWe investigated whether near-normal glucose levels were associated with endothelial
dysfunction by means of a stepwise hyperglycaemic clamp while suppressing endogenous
insulin production by octreotide infusion. Blood glucose (BG) levels were maintained at
6, 8 and 10 mmol/l successively for 2 hrs per level (Figure 1). We obtained blood samples
every 30 minutes during the clamp. Also, the day after the clamp a fasting blood sample
was obtained to assess the recovery after the glucose load.
PatientsIn total, 14 healthy non-smoking Caucasian males with a fasting plasma glucose level
≤5 mmol/l without risk factors for macrovascular disease as measured by BMI, blood
pressure, cholesterol and triglyceride levels were included in the study. Of those, one
dropped out due to febrile illness before the study day and two subjects were excluded
before analysis of the blood samples due to poor performance of the hyperglycaemic
clamp, resulting in 11 subjects who were included in the final analyses. Baseline
characteristics of the included subjects are listed in Table 1.
Table 1 Baseline characteristics
n = 11
Age, years 24.3 (3.6)
BMI, kg/m2 21.6 (1.9)
Systolic blood pressure, mmHg 115.1 (12.3)
Diastolic blood pressure, mmHg 68.9 (8.5)
Fasting plasma glucose, mmol/l 4.7 (0.2)
HbA1c, % 5.3 (0.2)
Total cholesterol, mmol/l 4.0 (0.9)
LDL cholesterol, mmol/l 2.2 (0.9)
HDL cholesterol, mmol/l 1.5 (0.3)
Triglycerides, mmol/l 0.7 (0.4)
Values are expressed as means (SD). BMI, body mass index; HDL, high density lipoprotein; LDL, low density lipoprotein
Hyperglycaemic clamp During the clamp endogenous insulin levels were adequately suppressed. Plasma insulin
levels after 1 hr of octreotide infusion and at the end of the clamp were comparable with
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
45
fasting levels in all patients (median <15 pmol/l, maximum 43 pmol/l). Mean glucose
levels of all included time points are depicted in Figure 1.
Figure 1 Glucose clamp Glucose values obtained during the clamp. Data are expressed as mean (SD) of the previous 30 minutes. Dotted lines represent the time points where glucose infusion was increased to go to the next glucose level.
Oxidative stressOxidative stress was assessed by quantitative determination of malondialdehyde (MDA)
in plasma using high performance liquid chromatography tandem mass spectrometry
(HPLC-MS/MS). MDA is a reactive and potentially mutagenic aldehyde which is formed
as a result of lipid peroxidation caused by hyperglycaemia induced formation of
ROS. Lipid peroxidation is thought to be an important part of the pathogenesis of
atherosclerosis 16 and the metabolites are frequently used as biomarkers for oxidative
stress 17.
Plasma MDA levels during the glucose clamp are depicted in Figure 2 and Table 2.
Plasma MDA levels did not increase after 1 hr octreotide infusion (median [IQR]
6.6 [6.2-7.7] μmol/l to 6.8 [6.0-8.0] μmol/l, P = 0.89). After the start of the glucose
infusion, plasma MDA increased gradually accompanying the increase in blood
glucose with a strong correlation between MDA and blood glucose levels (ρ = 0.82, P
<0.001, Spearman correlation). Median MDA levels at the 8 mmol/l glucose plateau
were significantly higher than after 1 hr octreotide infusion (9.9 [9.3-10.6] μmol/l,
P = 0.02) and were further increased substantially at the 10 mmol/l plateau (11.8
[10.8-12.5] μmol/l, P = 0.01). No cumulative effect of glucose over time during each
plateau was observed (Friedman test for repeated measures). The day after the clamp
median plasma MDA levels had returned to baseline (6.4 [5.6-6.9] μmol/l, P = 0.24).
46
Tab
le 2
Pla
sma
leve
ls o
f th
e p
aram
eter
s o
f in
tere
st d
uri
ng
the
glu
cose
cla
mp
Mar
ker
Un
itB
asel
ine
t =
1 h
r6
mm
ol/
l8
mm
ol/
l10
mm
ol/
lt
= 24
hrs
MD
Aμm
ol/l
6.6
(6.2
-7.7
)6.
8 (6
.0-8
.0)
8.0
(7.6
-9.6
)9.
9 (9
.3-1
0.6)
*11
.8 (1
0.8-
12.5
)*6.
4 (5
.6-6
.9)
F1+2
pm
ol/l
140
(118
-151
)17
0 (1
24-4
74)
508
(275
-171
4)*
731
(446
-109
5)*
608
(445
-817
)*13
6 (1
23-1
69)
vWF
%66
(50-
127)
63 (5
3-11
3)59
(39-
86)*
58 (3
9-10
7)*
55 (4
2-78
)*76
(58-
130)
ETP
nM
.min
1271
(116
7-14
15)
1237
(107
4-14
29)
1297
(108
2-14
25)
1263
(113
7-14
40)
1296
(114
3-14
86)*
1328
(117
0-14
81)^
Peak
th
rom
bin
nM
222
(200
-264
)20
9 (1
86-2
44)
208
(180
-245
)18
7 (1
74-2
45)
209
(189
-243
)22
6 (2
11-2
65)
PAP
µg/l
366
(268
-460
)33
5 (2
44-9
34)
540
(271
-854
)60
3 (3
98-1
060)
*61
5 (5
08-7
11)
426
(277
-650
)
d-d
imer
mg/
l0.
00 (0
-0.1
8)0.
00 (0
-0.0
5)0.
08 (0
-0.2
3)0.
28 (0
-0.5
1)*
0.36
(0.0
4-0.
51)*
0.19
(0-0
.39)
^
HA
ng/
ml
49.6
(48.
1-50
.2)
49.5
(47.
9-50
.2)
49.9
(48.
8-50
.9)
49.9
(47.
0-51
.5)*
50.3
(47.
2-51
.2)*
51.6
(50.
4-54
.5)^
Hya
luro
nid
ase
U/m
l51
.6 (4
3.6-
55.3
)49
.4 (4
4.0-
58.3
)45
.4 (3
6.3-
48.6
)48
.3 (3
5.4-
55.2
)36
.0 (3
3.4-
41.4
)*49
.5 (4
4.1-
53.0
)
At
each
glu
cose
pla
teau
th
e m
edia
n (
IQR
) va
lues
of
the
mea
n v
alu
e p
er p
atie
nt
are
dep
icte
d. M
DA
, m
alon
dia
ldeh
yde;
F1+
2, p
roth
rom
bin
fra
gmen
t 1+
2; v
WF,
von
W
ille
bran
d f
acto
r; P
AP,
pla
smin
-an
tip
lasm
in c
omp
lex;
ETP
, en
dog
enou
s th
rom
bin
pot
enti
al; H
A, h
yalu
ron
ic a
cid
. ^
P <0
.05
com
par
ed t
o ba
seli
ne;
* P
<0.
05 c
omp
ared
to
t =
1
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
47
Figure 2 Oxidative stress Oxidative stress was assessed by plasma malondialdehyde (MDA) levels during the glucose clamp. Data are depicted as medians with interquartile ranges. Dotted lines represent the timepoints where glucose infusion was increased to go to the next glucose level. *P <0.05 compared to t = 1, after 1 hr of octreotide infusion.
Coagulation We determined the effects of increasing glucose levels on coagulation by measuring
prothrombin fragment 1+2 (F1+2) and von Willebrand factor (vWF). F1+2 are released
when thrombin is formed from prothrombin and therefore provide an in vivo measure
of thrombin formation. VWF plays a major role in haemostasis by ensuring the arrest
of blood platelets at sites of injury, and by binding of coagulation factor VIII, but it is
also an established marker of endothelial dysfunction.
The effects of the stepwise increase in blood glucose on these markers of coagulation
are depicted in Figure 4 and Table 2. No significant effects on F1+2 levels or vWF were
detected after octreotide infusion. Median F1+2 levels showed a significant increase
from 170 (124-474) pmol/l to 508 (275-1714) pmol/l when the glucose level was raised to 6
mmol/l, further increased to 731 (446-1095) pmol/l at 8 mmol/l and remained at a stable
high level at 10 mmol/l with no significant differences between the glucose plateaus.
The following day F1+2 levels had returned to baseline. After raising the glucose level to
6 mmol/l vWF levels dropped to 59% (39-86, P = 0.02). An increase of glucose to 8 mmol/l
led to a further decrease of vWF levels to 58% (39-107, P = 0.05), but raising the glucose
level to 10 mmol/l did not cause further significant changes. After 24 hrs, vWF levels
had returned to baseline values.
Finally, we determined the endogenous thrombin potential (ETP), which represents the
48
balance between pro- and anti-coagulant processes in plasma and provides an ex vivo
measure for overall coagulability. No significant effects of the octreotide run-in period
or any of the blood glucose levels on peak thrombin values could be detected (Table
2). ETP showed a significant decrease from 1271 (1167-1415) nM.min to 1237 (1074-1429)
nM.min after the 1 hr octreotide period (P = 0.01) and a subsequent small increase to
1296 (1143-1486) nM.min at the 10 mmol/l glucose plateau (P = 0.01). The following day,
ETP remained higher than baseline at 1328 (1170-1481) nM.min (P = 0.05).
Figure 3 Endothelial glycocalyx Shedding of endothelial glycocalyx components was assessed by plasma hyaluronan levels (left panel) and activity of the regulatory enzyme hyaluronidase (right panel). Data are depicted as medians with interquartile ranges. Dotted lines represent the time points where glucose infusion was increased to go to the next glucose level. *P <0.05 compared to t = 1, after 1 hr of octreotide infusion.
Figure 4 Coagulation Coagulation was assessed by prothrombin fragment 1+2 (F1+2; left panel) and von Willebrand factor (vWF; right panel) plasma levels. Data are depicted as medians with interquartile ranges. Dotted lines represent the time points where glucose infusion was increased to go to the next glucose level. *P <0.05 compared to t = 1, after 1 hr of octreotide infusion.
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
49
Figure 5 Fibrinolysis Fibrinolysis was assessed by plasmin-alpha-antiplasmin (PAP) complexes (left panel) and d-dimer (right panel) plasma levels. Data are depicted as medians with interquartile ranges. Dotted lines represent the time points where glucose infusion was increased to go to the next glucose level. *P <0.05 compared to t = 1, after 1 hr of octreotide infusion.
Fibrinolysis Fibrinolysis was assessed by measuring plasmin-alpha2-antiplasmin (PAP) complexes
and d-dimer (Figure 5 and Table 2). PAP complexes serve as an indicator of recent in
vivo fibrinolytic activity, since alpha2 antiplasmin is the most important circulating
inhibitor of plasmin, the main enzyme in the fibrinolytic system. D-dimer is a fibrin
degradation product, which is dependent on the amount of fibrin that is generated
(coagulation) as well as the ability of the fibrinolytic system to degrade the generated
fibrin (fibrinolysis).
Neither PAP complexes nor d-dimer had changed significantly after the 1-hr octreotide
infusion period (t = 1; Figure 5). Median PAP levels were not significantly different at
a blood glucose level of 6 mmol/l when compared to t = 1, but did show a significant
increase at a blood glucose of 8 mmol/l (335 [244-934] μg/l to 603 [398-1060] μg/l, P = 0.01).
PAP levels remained at a stable high level when blood glucose was further increased to
10 mmol/l and returned to baseline the following day. Median d-dimer levels showed an
increasing trend from 0.00 (0.00-0.05) mg/l to 0.08 (0.00-0.23) mg/l when the glucose level
was raised to 6 mmol/l (P = 0.07). At 8 mmol/l d-dimer levels had risen to 0.28 (0.00-0.51)
mg/l (P = 0.04, compared to t = 1) and increased further to 0.36 (0.004-0.51) mg/l when
blood glucose was raised to 10 mmol/l (P = 0.02, compared to t = 1). After 24 hrs, d-dimer
levels remained higher than at t = 0, at 0.19 (0.00-0.39) mg/l (P = 0.04).
No cumulative effect of any of the coagulation and fibrinolysis parameters was detected
at the three examined blood glucose levels, except for F1+2 at a blood glucose level of 6
mmol/l (p=0.02, Friedman test for repeated measures).
50
Endothelial glycocalyxThe effect of elevation of blood glucose levels on the endothelial glycocalyx was
assessed by plasma measurement of its main component hyaluronic acid (HA) and its
regulatory enzyme hyaluronidase to detect shedding from the glycocalyx. Plasma HA
and hyaluronidase levels were unaffected by 1 hr octreotide infusion (t = 1). When raising
the blood glucose level to 6 mmol/l, median HA levels remained unaffected but a raise
to 8 and 10 mmol/l showed a significant, but small increase compared to t = 1 (from
49.5 [47.9-50.2] ng/ml to 49.9 [47.0-51.5] ng/ml, P = 0.038, and to 50.3 [47.2-51.2] ng/ml, P
= 0.008). This increase persisted after 24 hrs. Plasma hyaluronidase activity showed a
gradual decrease during the clamp, with significantly lower activity at a blood glucose
level of 10 mmol/l (36.0 [33.4-41.4] U/ml compared to 51.6 [43.6-55.3] U/ml at t = 1, P =
0.005). After 24 hrs, these levels had returned to baseline values (Figure 3 and Table 2).
Discussion
In this study we show that oxidative stress, represented by MDA levels, showed a stepwise
increase during the clamp, mimicking the glucose curve. The coagulation system was
activated even at near normal glucose levels of 6 mmol/l, resulting in a significant
increase in prothrombin fragments 1+2 (F1+2) indicating thrombin formation. This was
followed by activation of the fibrinolytic system, as measured by PAP complexes and
d-dimer, at a glucose level of 8 mmol/l. Relevant endothelial glycocalyx changes were not
detected using biochemistry techniques, except for a decrease in hyaluronidase activity
when the glucose concentration was raised to 10 mmol/l.
To our knowledge this is the first study examining the effects of isolated mild
hyperglycaemia, with a maximum of 10 mmol/l, on vascular homeostasis. Previous studies
on oxidative stress show glucose dependent formation of reactive oxygen species (ROS)
with blood glucose levels above 10 mmol/l 2;10, which is comparable with our findings and
in vitro studies 18. Our data suggest that hyperglycaemia dependent ROS formation is dose-
dependent rather than an on-off phenomenon. This is depicted in Figure 2, showing no
cumulative effect within the different glucose plateaus but only an increase in oxidative
stress when blood glucose is increased to the next level. Data reported by Ceriello et al. 10
support this finding showing higher plasma nitrotyrosine levels at a plasma glucose of
15 mmol/l compared to 11 mmol/l as well as no further increase in plasma nitrotyrosine
levels when stabilizing plasma glucose. Also, in type 2 diabetes patients an impressive
correlation between MDA and blood glucose (ranging from 6 to 14 mmol/l) was found
after a mixed-meal test suggesting an insulin-independent effect 19.
Our results are in line with previous observations, which have shown that hyperglycaemia
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
51
activates the coagulation 5;20 as well as the fibrinolytic system 5. Unlike oxidative stress,
our results indicate that the glucose induced-activation of the coagulation system is an
on-off phenomenon showing a more than threefold increase in thrombin generation,
measured by F1+2 levels, triggered by a blood glucose level of only 6 mmol/l. This
hypothesis is supported by the observation that maximum levels are reached quickly
and show no increase, perhaps even a decrease, at the highest glucose level. Moreover,
the maximum levels of F1+2 and d-dimer are comparable with the levels reached during
a hyperglycaemic clamp at a blood glucose of 15 mmol/l previously performed by our
group 5. The timing of the increase in fibrinolytic activity, closely following the coagulant
activity, suggests that the increased fibrinolytic activity is secondary to the coagulation
activation. Conversely, diabetes mellitus is associated with impairment in fibrinolysis 21,
which we did not detect in our study. However, Stegenga et al. 20 showed that fibrinolysis
was mainly affected by hyperinsulinemia as opposed to hyperglycaemia, and insulin was
suppressed throughout our experiments.
ETP changed only minimally during and after the clamp. This indicates no relevant
change in the thrombin generating capacity of the coagulation system itself, but
rather suggests that glucose is a trigger for the in vivo activation of coagulation. VWF
levels showed a maximal decrease of 5%. This modest change could be due to increased
binding to blood platelets, known to be activated by hyperglycaemia, or caused by
physical inactivity of the participants. VWF levels have been shown to increase after
physical exercise 22 and previous control experiments by our group have shown a similar
decreasing effect of a 6-hr saline infusion in healthy males (M. Nieuwdorp, unpublished
data).
We did not detect relevant changes in plasma HA levels and a decrease in hyaluronidase
activity was found only at a glucose level of 10 mmol/l. Previous investigations show
marked increase in HA shedding from 70 ng/ml at baseline to 112 ng/ml with blood
glucose levels of 15 mmol/l 5, suggesting that the trigger for direct endothelial damage as
reflected by loss of glycocalyx lies above a blood glucose level of 10 mmol/l. Statistically,
there was a change in plasma HA levels at 8 and 10 mmol/l. However, the maximum
increase was only 1.6% indicating no significant biological effect. This is supported by
the limited effects on vWF levels, which are also considered a marker for endothelial
damage. The decrease in hyaluronidase activity at 10 mmol/l does indicate substrate
binding to this enzyme. This substrate may consist of other glycosaminoglycans than
HA shed from the glycocalyx, such as heparan sulphate or chondroitin sulphate, since
these are also bound by hyaluronidase.
The results of our study are in line with epidemiological data, which show that the
increase in cardiovascular risk already starts at mildly elevated glucose levels 13-15.
52
Nonetheless, our results indicate that glucose-induced activation of the coagulation
system and ROS formation are completely reversible after 24 hrs. Therefore, these changes
may not be considered to be pathological in healthy subjects who spent the greater part
of the day with glucose levels below 6.1 mmol/l 23. Conversely, patients with diabetes or
pre-diabetes by definition have a fasting glucose level of >5.6 mmol/l if untreated 24, and
are exposed to glucose levels above 6 mmol/l throughout the day. This may interfere
with the reversibility of the changes in coagulation and oxidative stress, and translate
to pathological effects. Moreover, in diabetes inappropriate activation of the coagulation
system may not be counterbalanced because of the fibrinolytic impairment associated
with this disease 21. Our results do not support a role for glucose variability in coagulation
activation and ROS formation, since coagulation activation occurred even at a blood
glucose of 6 mmol/l and the relationship between blood glucose and oxidative stress
was continuous.
Several aspects of our study need comment. First, this study was specifically designed to
assess the effects of mild hyperglycaemia on several components of vascular homeostasis
and was therefore performed under full suppression of insulin levels. In disease states,
such as type 2 diabetes or stress-hyperglycaemia during severe illness, high glucose
levels are accompanied by high insulin levels and therefore our results cannot be
extrapolated directly to these settings. However, glucose levels are highly predictive of
vascular complications 14;25 although insulin levels are highly variable in these patients.
Second, given the design of our experiment, it can be argued that the effects we detected
may not be glucose specific, but rather result from the osmotic effect of raising blood
glucose or from prolonged administration of octreotide. However, previous work from
our group has shown no effect on coagulation or fibrinolysis in a control experiment
during which octreotide was administered in combination with mannitol infusion for six
hours, serving as a time and osmolality control 5. Moreover, in our study no significant
effect on any of the parameters after one hour of octreotide infusion was detected. This
is supported by literature, showing no significant vaso-active or haemostatic effects of
this dose of octreotide 26;27.
In conclusion, our results show that glucose-induced changes to vascular homeostasis
already start at near normal glucose levels. Furthermore our study reveals a dose-
dependent effect of glucose on MDA formation and an on-off phenomenon for glucose
induced coagulation activation, while changes to the endothelial glycocalyx occur at
glucose levels of 10 mmol/l or higher. These results give us more insight in the glucose
driven mechanisms of vascular complications in humans. To elucidate the difference
between acute and chronic mild hyperglycaemia on vascular homeostasis, further studies
are needed.
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
53
Methods
PatientsThe study was approved by the institutional medical ethical committee and conducted
according to Declaration of Helsinki principles. Participants signed informed consent
prior to inclusion after oral and written explanation of the study.
Stepwise hyperglycaemic clamp protocol (Figure 6) After an overnight fast, two catheters for venous access were placed, one in every arm.
First, basal measurements of haemostasis and ROS formation were performed. Octreotide
was dissolved in saline 0.9% and albumin 20% (proportion 59:1 in a 60 ml syringe)
and administered at a final concentration of 30 ng/kg/min octreotide, to suppress
endogenous insulin production 5. To confirm that this infusion did not influence the
parameters of interest, the basal measurements were repeated after 1 hr of octreotide
infusion. Hereafter, glucose infusion with 20% glucose solution was started to reach the
desired glucose concentration based on a steady state principle 28. The plasma glucose
concentration was held constant at the desired plateau for 2 hrs by determination of
the plasma glucose concentration every 5 minutes and adjusting the glucose solution
accordingly. When a stable glucose concentration was reached, glycocalyx dimension,
ROS formation, and haemostasis parameters were measured every 30 minutes; 4 times
per plateau, the last measurement being the baseline value of the next step. Glucose
infusion was then increased to reach the next level of glycaemia and measurements were
repeated. In total, the actual clamp took 7 hrs. Blood samples were centrifuged within
1 hr after collection and stored at -80°C.
Figure 6 Glucose clamp protocol Depicted in the lower boxes are the times from baseline where the several actions were performed. An arrow indicates an assessment point for oxidative stress, glycocalyx and coagulation/fibrinolysis parameters.
54
Oxidative stressPlasma MDA concentration was determined using high HLPC-MS/MS as described by Pilz 29,
with minor modifications. Total (free and bound) malondialdehyde (MDA) in human
plasma samples was determined as the 2,4-dinitrophenylhydrazine (DNPH) derivative.
After addition of the stable isotopically labeled analogue (2H2-MDA) as internal standard
(IS), alkaline hydrolyzation, deproteinization and derivatization with DNPH, MDA-
hydrazone was analyzed by HPLC-MS/MS and positive electrospray ionization. Using
an Acquity UPLC system (Waters Corporation, Milford, MA), samples were injected on
a LC-18-DB analytical column (250 · 4.6 mm, 5 μm particles, Supelco) hyphenated to a
Quattro Premier XE mass spectrometer (Waters Corporation, Milford, MA). Analytes and
IS were eluted with acetonitrile/water/acetic acid (50/50/0.2) and detected in multiple
reaction monitoring (MRM) mode for the transitions m/z 235 m/z 159; m/z 237 m/z
161. Samples were quantified against calibration standards.
Endothelial glycocalyxHyaluronic acid was measured by a commercially available ELISA kit (Corgenix, Inc.,
Broomfield, Colorado, USA). In short, HA reacted with hyaluronic acid binding protein.
Thereafter horseradish peroxidase was added to form complexes with bound HA.
After addition of a chromogenic substrate the intensity of the colour was measured in
optical density units with a spectrophotometer at 450 nm. Hyaluronidase activity was
determined by a previously described assay 30;31.
Coagulation and fibrinolysisD-dimer was measured on an automated coagulation analyzer (Behring Coagulation
System, BSC) using protocols and reagents from the manufacturer (Siemens Healthcare
Diagnostics, Marburg, Germany). Antigen levels of vWF were assayed by ELISA using
antibodies from Dako (Glostrup, Denmark). F1+2 and PAP were determined by ELISA
from Siemens Healthcare Diagnostics and DRG (Marburg, Germany), respectively. The ETP
was determined using the Calibrated Automated Thrombogram as described by Hemker
et al. 32 and the Thrombinoscope manual (Maastricht, the Netherlands). Coagulation
was triggered by recalcification in the presence of 5 pM recombinant human tissue
factor (Innovin, Siemens Healthcare Diagnostics), 4 μM phospholipids, and 417 μM
fluorogenic substrate Z-Gly-Gly-Arg-AMC (Bachem, Bubendorf, Switzerland). Fluorescence
was monitored using the Fluoroskan Ascent fluorometer (ThermoLabsystems, Helsinki,
Finland), and the ETP and peak thrombin were calculated using the Thrombinoscope
software.
Data interpretationThe study was conducted to assess the influence of a certain level of plasma glucose on
the parameters described above. We excluded samples taken at a certain glucose plateau
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
55
when the desired glucose level was exceeded by more than 1 mmol/l since crossing the
desired glucose level could interfere with the study results. For example, when a plasma
glucose level of 7.1 mmol/l occurred at any point during the 6 mmol/l plateau phase,
all subsequent samples taken at the 6 mmol/l plateau were excluded from analysis.
Moreover, samples were only included in the analysis when the desired glucose level was
truly reached. This was determined by calculation of the mean glucose level of the 30
minutes before sampling which had to be within 0.5 mmol/l of the desired glucose level.
Statistical analysisBaseline characteristics are expressed as mean (SD) and outcome parameters as median
(IQR). Differences between plateaus were assessed by a Wilcoxon signed ranks test for
paired data. The influence of time on the measurements at each glucose level was
assessed using the Friedman test. All analyses were performed using Predictive Analytics
Software (PASW) statistics version 18.0 (SPSS Inc., Chicago, IL, USA). A P-value <0.05 was
considered statistically significant.
AcknowledgementsWe would like to thank Jeroen Sierts from the Laboratory of Experimental Vascular
Medicine and Arno van Cruchten from the Laboratory Genetic Metabolic Diseases,
both at the Academic Medical Centre, Amsterdam, the Netherlands, for their excellent
laboratory support.
56
References1. Pignone M, Alberts MJ, Colwell JA, et al (2010) Aspirin for primary prevention of cardiovascular events in
people with diabetes: a position statement of the American Diabetes Association, a scientific statement of the American Heart Association, and an expert consensus document of the American College of Cardiology Foundation. Circulation 121: 2694-2701
2. Ceriello A, Esposito K, Piconi L, et al (2008) Glucose “peak” and glucose “spike”: Impact on endothelial function and oxidative stress. Diabetes Res Clin Pract 82: 262-267
3. Lemkes BA, Hermanides J, DeVries JH, Holleman F, Meijers JC, Hoekstra JB (2010) Hyperglycemia: a prothrombotic factor? J Thromb Haemost 8: 1663-1669
4. Nieuwdorp M, Meuwese MC, Vink H, Hoekstra JB, Kastelein JJ, Stroes ES (2005) The endothelial glycocalyx: a potential barrier between health and vascular disease. Curr Opin Lipidol 16: 507-511
5. Nieuwdorp M, van Haeften TW, Gouverneur MC, et al (2006) Loss of endothelial glycocalyx during acute hyperglycemia coincides with endothelial dysfunction and coagulation activation in vivo. Diabetes 55: 480-486
6. Iwasaki Y, Kambayashi M, Asai M, Yoshida M, Nigawara T, Hashimoto K (2007) High glucose alone, as well as in combination with proinflammatory cytokines, stimulates nuclear factor kappa-B-mediated transcription in hepatocytes in vitro. J Diabetes Complications 21: 56-62
7. Min C, Kang E, Yu SH, Shinn SH, Kim YS (1999) Advanced glycation end products induce apoptosis and procoagulant activity in cultured human umbilical vein endothelial cells. Diabetes Res Clin Pract 46: 197-202
8. Siegelaar SE, Holleman F, Hoekstra JB, DeVries JH (2010) Glucose variability; does it matter? Endocr Rev 31: 171-182
9. Kilpatrick ES (2009) Arguments for and against the role of glucose variability in the development of diabetes complications. J Diabetes Sci Technol 3: 649-655
10. Ceriello A, Esposito K, Piconi L, et al (2008) Oscillating Glucose Is More Deleterious to Endothelial Function and Oxidative Stress Than Mean Glucose in Normal and Type 2 Diabetic Patients. Diabetes 57: 1349-1354
11. Monnier L, Colette C, Mas E, et al (2010) Regulation of oxidative stress by glycaemic control: evidence for an independent inhibitory effect of insulin therapy. Diabetologia 53: 562-571
12. Siegelaar SE, Barwari T, Kulik W, Hoekstra JB, DeVries JH (2011) No relevant relationship between glucose variability and oxidative stress in well-regulated type 2 diabetes patients. J Diabetes Sci Technol 5: 86-92
13. Cheng YJ, Gregg EW, Geiss LS, et al (2009) Association of A1C and fasting plasma glucose levels with diabetic retinopathy prevalence in the U.S. population: Implications for diabetes diagnostic thresholds. Diabetes Care 32: 2027-2032
14. The Diabetes Control and Complications Trial Research Group (1996) The absence of a glycemic threshold for the development of long-term complications: the perspective of the Diabetes Control and Complications Trial. Diabetes 45: 1289-1298
15. Haffner SM, Stern MP, Hazuda HP, Mitchell BD, Patterson JK (1990) Cardiovascular risk factors in confirmed prediabetic individuals. Does the clock for coronary heart disease start ticking before the onset of clinical diabetes? JAMA 263: 2893-2898
16. Uchida K (2000) Role of reactive aldehyde in cardiovascular diseases. Free Radical Biology and Medicine 28: 1685-1696
17. Nielsen F, Mikkelsen BB, Nielsen JB, Andersen HR, Grandjean P (1997) Plasma malondialdehyde as biomarker for oxidative stress: reference interval and effects of life-style factors. Clin Chem 43: 1209-1214
18. Brownlee M (2001) Biochemistry and molecular cell biology of diabetic complications. Nature 414: 813-82019. Bunck MC, Cornér A, Eliasson B, et al. (2010) One year treatment with exenatide vs. Insulin Glargine: effects
on postprandial glycemia, lipid profiles, and oxidative stress. Atherosclerosis 212:223-22920. Stegenga ME, van der Crabben SN, Levi M, et al (2006) Hyperglycemia stimulates coagulation, whereas
hyperinsulinemia impairs fibrinolysis in healthy humans. Diabetes 55: 1807-181221. Grant PJ (2007) Diabetes mellitus as a prothrombotic condition. J Intern Med 262: 157-17222. Lippi G, Maffulli N (2009) Biological influence of physical exercise on hemostasis. Semin Thromb Hemost
35: 269-27623. Borg R, Kuenen JC, Carstensen B, et al (2010) Real-life glycaemic profiles in non-diabetic individuals with
low fasting glucose and normal HbA1c: the A1C-Derived Average Glucose (ADAG) study. Diabetologia 53: 1608-1611
24. American Diabetes Association (2011) Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 34: S62-S69
Mild hyperglycaemia disturbs vascular homeostasis in humans
Ch
apte
r 3
57
25. UK Prospective Diabetes Study (UKPDS) Group (1998) Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet 352: 837-853
26. Williams SB, Goldfine AB, Timimi FK, et al (1998) Acute hyperglycemia attenuates endothelium-dependent vasodilation in humans in vivo. Circulation 97: 1695-1701
27. Witzig TE, Kvols LK, Moertel CG, Bowie EJ (1991) Effect of the somatostatin analogue octreotide acetate on hemostasis in humans. Mayo Clin Proc 66: 283-286
28. Gottesman I, Mandarino L, Gerich J (1983) Estimation and kinetic analysis of insulin-independent glucose uptake in human subjects. Am J Physiol 244: E632-E635
29. Pilz J, Meineke I, Gleiter CH (2000) Measurement of free and bound malondialdehyde in plasma by high-performance liquid chromatography as the 2,4-dinitrophenylhydrazine derivative. J Chromatogr B Biomed Sci Appl 742: 315-325
30. Frost GI, Stern R (1997) A microtiter-based assay for hyaluronidase activity not requiring specialized reagents. Anal Biochem 251: 263-269
31. Nieuwdorp M, Mooij HL, Kroon J, et al (2006) Endothelial glycocalyx damage coincides with microalbuminuria in type 1 diabetes. Diabetes 55: 1127-1132
32. Hemker HC, Giesen P, al DR, et al (2003) Calibrated automated thrombin generation measurement in clotting plasma. Pathophysiol Haemost Thromb 33: 4-15
Chapter 4
No relevant relationship between glucose variability and oxidative stress in well-regulated type 2 diabetes patients
Sarah E. Siegelaar, Temo Barwari, Wim Kulik, Joost B.L. Hoekstra
and J. Hans DeVries
Journal of Diabetes Science and Technology 2011; 5(1):86-92
60
Abstract
Background: A strong relationship between glycaemic variability and oxidative
stress in poorly regulated type 2 diabetes on oral medication has been reported.
However, this relationship was not seen in type 1 diabetes. The purpose of this
study is to re-examine the relation between glycaemic variability and oxidative
stress in a cohort of type 2 diabetes patients on oral medication.
Methods: Twenty-four patients with type 2 diabetes on oral glucose lowering
treatment underwent 48 hrs of continuous glucose monitoring (CGMS® System
GoldTM, Medtronic MiniMed) and simultaneous collection of two consecutive
24-hr urine samples for determination of 15(S)-8-iso-prostaglandin F2α (PGF2α)
using high-performance liquid chromatography tandem mass spectrometry.
Standard deviation (SD) and mean amplitude of glycaemic excursions (MAGE)
were calculated as markers of glycaemic variability.
Results: Included in the study were 66.7% males with a mean age (range) of 59
(36-76) years and a mean (SD) HbA1c of 6.9% (0.7). Median (interquartile range
[IQR]) urinary 15(S)-8-iso-PGF2α excretion was 176.1 (113.6-235.8) pg/mg creatinine.
Median (IQR) SD was 1.7 (1.3-2.2) mmol/l and MAGE 4.7 (3.1-5.9) mmol/l. Spearman
correlation did not show a significant relation for SD (ρ = 0.15, P = 0.49) or MAGE
(ρ = 0.23, P = 0.29) with 15(S)-8-iso-PGF2α excretion. Multivariate regression analysis
adjusted for age, sex, HbA1c, and exercise did not alter this observation.
Conclusions: We did not find a relevant relationship between glucose variability
and 15(S)-8-iso-PGF2α excretions in type 2 diabetes patients well-regulated with
oral medication that would support an interaction between hyperglycaemia and
glucose variability with respect to the formation of reactive oxygen species.
No effect of glucose variability on oxidative stress in well regulated T2DM
Ch
apte
r 4
61
Introduction
The main cause of morbidity and mortality in both type 1 and type 2 diabetes is the
development of micro- and macrovascular complications. Several studies have shown a
direct relation between glucose exposure, as measured by haemoglobin A1c (HbA1c) level,
and complications of diabetes 1;2. The molecular mechanisms that explain this relation
between hyperglycaemia and diabetic complications have not been fully elucidated. A
suggested common pathway by which hyperglycaemia leads to these complications is
the formation of reactive oxygen species (ROS) by glucose overload in the mitochondria.
This could lead to vascular damage through several molecular mechanisms 3. However,
HbA1c only explains around 25% of the variation in risk of developing complications 4,
suggesting the contribution of other factors. It has been suggested that glycaemic
variability contributes to diabetes complications via formation of ROS 5.
Monnier et al. 6 were the first to study the role of glucose variability in oxidative stress
formation in type 2 diabetes. They measured 24-hr urinary excretion rates of free 8-iso-
prostaglandin F2α (PGF2α), which is considered a good marker of oxidative stress 7;8.
Continuous glucose monitoring (CGM) was performed for determination of glycaemic
variability. They found a strong correlation between 24-hr PGF2α excretion rates and
glucose variability in type 2 diabetes patients on diet and/or oral antihyperglycaemic
drugs. Subsequently, Wentholt et al. 9 investigated the role of glucose variability in
activation of oxidative stress in type 1 diabetes patients using the same oxidative stress
marker and the same method to determine glucose variability. Despite more pronounced
glycaemic variability, no relationship was found between oxidative stress and glucose
variability. They found oxidative stress to be increased in type 1 diabetes.
In type 1 diabetes patients, other pathways than glucose variability might be involved in
activation of oxidative stress, perhaps explaining these conflicting findings. Moreover,
different techniques were used to assess the PGF2α excretion rate. The immunoassay is less
specific than the tandem mass spectrometry technique used by Wentholt et al., as already
acknowledged by Monnier et al. 10;11. Interestingly, a study in 2010 by Monnier confirmed
the findings of Wentholt regarding the absence of a relation between glucose variability
and oxidative stress in type 1 diabetes patients. Even more, type 1 diabetes patients and
healthy controls did not differ in 8-iso-PGF2α excretion 12. Also their results suggest that
the relation between glucose variability and ROS formation in type 2 diabetes patients
depends on the HbA1c level, a relation only seen at higher HbA1c levels.
The aim of this study is to evaluate the role of acute glucose fluctuations in the activation
of oxidative stress in type 2 diabetes patients using oral glucose-lowering agents, using
tandem mass spectrometry for the assessment of PGF2α excretion rates.
62
Methods
PatientsTwenty-eight patients with type 2 diabetes were randomly recruited from the outpatient
clinic of the Academic Medical Centre in Amsterdam, The Netherlands, between April 2008
and February 2009. Inclusion criteria were a diagnosis of type 2 diabetes for more than 6
months and treatment with oral glucose-lowering agents or diet. Patients were excluded in
case of use of steroid or nonsteroidal anti-inflammatory drugs, insulin treatment, an acute
illness during the 3-month period prior to the investigation, or an estimated glomerular
filtration rate (GFR) of less than 60 ml/min/1.73m2 according to the Cockcroft-Gault formula 13 because of a potential influence on oxidative stress production. Also the use of heparin
or oral anticoagulants (except for aspirin) was not allowed as this could cause bleeding
during sensor insertion. The study was approved by the local ethics committee and patients
gave written informed consent after written and oral explanation of the study.
Study designOn day 1, after giving written informed consent, the continuous glucose monitor (CGMS®
System GoldTM, Medtronic MiniMed, Northridge, CA, USA) was inserted subcutaneously in
the abdominal wall. Patients were provided with home blood glucose meters (OneTouch®
Ultra®, LifeScan, Inc., Milpitas, CA, USA) and were instructed to perform the required
sensor calibration procedure four times daily according to manufacturers instructions.
Urine was collected for two consecutive 24-hr periods. Patients were asked to store
the urine jars in the refrigerator, to avoid ex vivo formation of isoprostanes. Systolic
and diastolic blood pressure was measured and venous blood samples were drawn for
the following laboratory measurements: fasting plasma glucose (FPG), HbA1c, total
cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL)
cholesterol, triglycerides, and creatinine. Relevant patient characteristics were recorded
(e.g., medical history, current medication use, body mass index [BMI], smoking habits).
The reported amount of physical exercise was graded as sedentary, moderately active,
active, and fit. On day 4, patients returned to the clinical trial room with the two urine
collection jars. From each 24-hr period (day 2 and day 3 of the study) a 7-ml urine sample
was stored in a freezer at -80 °C until analysis of all urine samples in one run. For the
analyses, the 15(S)-8-iso-PGF2α excretion rates over 48 hrs were used calculated by averaging
the first- and second-day samples. Lastly, the continuous glucose monitor was removed
and data was downloaded and stored electronically. Only glucose data obtained during
the 48 hrs of urine sampling was used for further analysis.
Laboratory MeasurementsHaemoglobin A1c was measured using an high-performance liquid chromatography
(HPLC) assay (Variant II, Bio-Rad Laboratories, Montreal, Quebec, Canada). Activation of
No effect of glucose variability on oxidative stress in well regulated T2DM
Ch
apte
r 4
63
oxidative stress was estimated by determining the urinary isoprostane excretion (15(S)-8-iso-
PGF2α), using HPLC tandem mass spectrometry (HPLC-MS/MS). Urine samples were collected
and stored without additives as 7-ml aliquots at -80 °C. Initial creatinine concentrations
were established by colourimetric Jaffé assay. Two millilitres of sample were mixed with 2H4-
labelled 15(S)-8-iso-PGF2α as internal standard and applied to an 8-iso-PGF2α immunoaffinity
column (Cayman Chemical Company, Ann Arbor, MI, USA). After washing, the extract was
eluted, evaporated to dryness (60 °C, N2) and reconstituted in 200-μl 0.05 mol/l formic
acid-ethanol (75:25, v/v); 50 μl was injected on the HPLC-MS/MS system. Chromatographic
separation was achieved on a modular HPLC system (Surveyor®, Thermo Finnigan, San Jose,
CA, USA) consisting of a cooled autosampler (T = 12 °C), a low-flow quaternary MS pump
and analytical HPLC column: Alltima C8, 2.1 x 150 mm, 5 μm (Alltech, Lexinton, KY, USA).
Samples were eluted with a flow rate of 200 μl/min and a programmed linear gradient
between A (0.01% HCOOH in H2O; v/v) and B (CH3CN): from t = 0 min 45% A, 55% B toward t = 3
min 30% A, 70% B toward t = 3.1 min 100% A until t = 6 min. MS/MS analyses were performed
on a TSQ Quantum AM (Thermo Finnigan, San Jose, CA, USA) operated in the negative ion
electrospray ionisation mode. The surface-induced dissociation was set at 2 V; spray voltage
was 3500 V, and the capillary temperature was 400°C. In the MS/MS experiments argon was
used as collision gas at a pressure of 0.2 Pa; collision energy was 26 eV for the optimised
transitions: m/z 353.24 m/z 193.10 and m/z 357.24 m/z 197.10. The interassay (n = 5) and
intraassay (average of 5 days, n = 3) variability allowed for determination at physiological
concentrations with a coefficient of variation of <7%.
Assessment of glycaemic variabilityIn literature, there is no universally accepted “gold” standard to measure variability
in glucose values 14. We calculated the SD of the glucose measurements and the mean
amplitude of glycaemic excursions (MAGE) as described by Service et al. 15 because SD is
the best mathematically validated measure and both are commonly used in literature.
The MAGE over 48 hrs is the mean of the absolute differences between peak and nadir
values over 48 hrs. The peaks and nadirs are defined as glucose values preceded and
followed by an increase and decrease or a decrease and increase, respectively. Only
increases or decreases larger than 1 SD of the mean glucose are taken into account. If
a decrease of more than 1 SD was the first excursion, only peak-to-nadir excursions (>1
SD) were included in the calculation of the MAGE and vice versa.
Assessment of postprandial glucose excursionsWe determined the incremental areas above preprandial glucose values (breakfast, lunch,
dinner) over a 4-hr period following the beginning of each meal using the trapezoidal
rule 16. The six incremental areas of each patient during the 48 hrs of continuous glucose
monitoring were summed and averaged to calculate the mean postprandial incremental
area under the curve (AUCpp) 6.
64
Statistical analysisMeans and standard deviations of patient characteristics, urinary excretion of 15(S)-8-iso-
PGF2α, and glucose variability measures were assessed using standard statistics. Levels of
urinary 15(S)-8-iso-PGF2α in the samples from both days were compared using a Wilcoxon
signed-rank test. Spearman correlation was calculated in order to evaluate the relation
between glycaemic variability measures, HbA1c, FPG, postprandial glucose excursions,
and the excretion of urinary 15(S)-8-iso-PGF2α. The effect of glucose variability on 15(S)-
8-iso-PGF2α excretion was also assessed in a multivariate regression model adjusting for
variables that have been reported to be involved in oxidative stress activation, i.e., sex,
age, smoking, and HbA1c 17, and for variables that showed a relation with urinary 15(S)-
8-iso-PGF2α excretion in Spearman correlation analysis. Analyses were performed using
SPSS version 16.0.2.
Results
From the 28 recruited patients, 4 patients were excluded from data analysis; 2 patients
showed an estimated GFR of less than 60 ml/min/1.73m2, in 1 patient the glucose sensor
failed to record any data and 1 patient did not perform the urine collection adequately.
From the remaining 24 patients, 16 were male, mean (range) age was 58.9 years (36-76)
and mean (SD) HbA1c was 6.9% (0.7). Median (interquartile range [IQR]) urinary 15(S)-
8-iso-PGF2α excretion was 176.1 (113.6-235.8) pg/mg creatinine. There was no significant
difference between the first- and second-day urine samples (Wilcoxon signed-rank test,
P = 0.95). Median (IQR) SD and MAGE were 1.7 (1.3-2.2) and 4.7 (3.1-5.9) mmol/l, respectively.
Patient characteristics are listed in Table 1.
Spearman correlation did not reveal a significant relation for any glucose variability
parameter with 15(S)-8-iso-PGF2α excretion (SD, ρ = 0.15, P = 0.49; MAGE, ρ = 0.23, P = 0.29,
Figure 1]. Multivariate regression analysis was performed to adjust for sex, age, smoking,
and HbA1c 17, and for the variable that showed a significant correlation with 15(S)-8-iso-
PGF2α, i.e., the amount of physical exercise. This did not alter the results from Spearman
correlation described above (SD, r2 = 0.003, P = 0.77; MAGE, r2 = <0.001, P = 0.98).
No significant correlation was found between 15(S)-8-iso-PGF2α excretion and HbA1c (ρ =
0.27, P = 0.20) or AUCpp (ρ = 0.14, P = 0.51), FPG (ρ = 0.27, P = 0.20) and the mean of sensor
glucose measurements (ρ = 0.23, P = 0.29), Table 2. The plots do not reveal a pattern that
No effect of glucose variability on oxidative stress in well regulated T2DM
Ch
apte
r 4
65
Table 1 Baseline characteristics
Characteristics Patients (n = 24)
Age, years 58.9 (36-76)
Men/women, n 16/8
Diabetes duration, years 7.2 (4.2)
Diabetes treatment, n (%)MetforminSulfonylureaRosiglitazone
23 (96)15 (63)2 (8)
Other treatments, n (%)ACE inhibitorStatinAspirin
9 (38)19 (79)7 (29)
Cigarette smoking, n (%) 2 (8)
BMI, kg/m2 30.5 (5.5)
Systolic blood pressure, mmHg 135 (17)
Diastolic blood pressure, mmHg 82 (10)
Plasma creatinine, μmol/l 76.3 (13.1)
Total cholesterol, mmol/l 4.18 (0.80)
HDL cholesterol, mmol/l 1.10 (0.20)
LDL cholesterol, mmol/l 2.31 (0.71)
Triglycerides, mmol/l 1.71 (0.72)
HbA1c, % 6.9 (0.7)
FPG, mmol/l 8.0 (1.8)
Mean sensor glucose, mmol/l 8.1 (1.5)
AUCpp, mmol/l/hr 7.2 (3.0)
Markers of glucose variability (median [IQR])SD, mmol/lMAGE, mmol/l
1.7 (1.3-2.2)4.7 (3.1-5.9)
Urinary 15(S)-8-iso-PGF2α, pg/mg creatinine (median [IQR]) 176.1 (113.6-235.8)
Data are means (SD) or means (range), unless stated otherwise in parentheses. BMI, body mass index; FPG, fasting plasma glucose; AUCpp, 4-hour postprandial incremental area under the curve; SD, standard deviation; MAGE, mean amplitude of glycaemic excursions; IQR, Interquartile Range. To convert mean glucose, AUCpp, MAGE and SD from mmol/l to mg/dl divide by 0.0555.
66
would suggest a threshold phenomenon. Also, no significant correlations were found
for sex, age, BMI, smoking, systolic or diastolic blood pressure, or lipid concentrations
(Table 2). A multivariate regression model with age, sex, smoking, exercise, and HbA1c
as covariates did show significant inverse relations between 15(S)-8-iso-PGF2α excretion
rates and age (r = -0.49, P = 0.01) and the amount of physical exercise (r = -0.61, P = 0.004).
Figure 1 Correlations between glycaemic markers and oxidative stress X-axis: glycaemic markers over 48 h of glucose measurements expressed as SD, MAGE, AUCpp and mean blood glucose. Y-axis: oxidative stress, expressed as 15(S)-8-iso-PGF2α in pg/mg creatinine, over 48 hrs of urine collection.
No effect of glucose variability on oxidative stress in well regulated T2DM
Ch
apte
r 4
67
Table 2 Spearman correlation coefficients (ρ) between urinary excretion rates of
15(S)-8-iso-PGF2α, clinical characteristics and glycaemic markers
Sex Age BMI Exercise HbA1c FPG MBG SD MAGE AUCpp
FPG 0.09 0.33 -0.12 -0.15 0.48 a
MBG -0.08 0.36 -0.22 -0.20 0.63b 0.85 b
SD -0.29 -0.04 -0.45 a -0.04 0.43 a 0.34 0.45 a
MAGE -0.27 -0.13 -0.41 a 0.01 0.44a 0.38 0.44 a 0.95 b
AUCpp -0.41 a -0.18 -0.28 0.03 0.39 0.29 0.35 0.85 b 0.86 b
8-isoPGF -0.09 -0.27 0.23 -0.50a 0.27 0.27 0.23 0.15 0.23 0.14
a P <0.05; b P <0.01
Discussion
We did not find a relevant association between markers of glycaemic variability and
oxidative stress, estimated by 15(S)-8-iso-PGF2α excretion rates, in patients with type 2
diabetes who were well-regulated with oral glucose lowering agents only. At first, these
results seem in contrast with earlier data, showing a firm correlation between glycaemic
variability and oxidative stress in type 2 diabetes patients using oral hypoglycaemic
agents 6, despite use of the same continuous glucose monitor and mathematical methods
to assess glucose variability.
A possible explanation for the seemingly opposing findings in type 2 diabetes patients
might be the difference in glucose regulation between the study populations. The mean
glucose and HbA1c of the original Monnier population was markedly higher than in
our population (mean glucose 10.5 and 8.1 mmol/l, HbA1c 9.6 and 6.9%, respectively)
though the mean MAGE was roughly comparable (4.2 and 4.7 mmol/l respectively). In a
paper published by the same group 12 there seemed to be a modifying effect of HbA1c
on the relation between glucose variability and ROS formation in patients with type 2
diabetes on oral medication. No effect of MAGE on the formation of 15(S)-8-iso-PGF2α was
seen in the group with an HbA1c below the median of 8.2%, while people with higher
MAGE values had more oxidative stress in the subgroup with an HbA1c above the median.
Thus, the absence of a relation between glucose variability and oxidative stress in our
population with a mean HbA1c of 6.9% is in agreement with their findings. It might be
possible that the fluctuations observed in patients with a high HbA1c level reach a higher
maximum glucose value and in that way cross a threshold for oxidative stress. In line
68
with this is the observation that the AUCpp in the Monnier population is much larger
than in our well regulated population (22.6 and 7.2 mmol/l/hr respectively) suggesting
that the variability in the Monnier population depended more on large postprandial
excursions which are known to enhance oxidative stress 18.
Another difference between our results and those reported by Monnier et al. is a
difference in technique for quantification of oxidative stress. Monnier et al. used an
enzyme immunoassay (EIA) to quantify urinary excretion of 15(S)-8-iso-PGF2α, whereas
we used HPLC-MS/MS, the same method used by Wentholt et al. Measurement of F2-
isoprostanes by way of mass spectrometry has been reported to be more specific and
sensitive than measurement by immunoassay 10;11. Studies have shown that MS is not
hampered by cross reactivity of structurally (un)related components of 8-iso-PGF2α in that
way selectively determining 15(S)-8-iso-PGF2α, whereas several substances in biological
fluids that are not products of lipid peroxidation interfere with the immunoassay that
includes its enantiomer ent-15(S)-8-iso-PGF2α in the quantification of oxidative stress 8;11.
This is also why in literature 8-iso-PGF2α levels are often reported to be higher when
assessed by EIA compared to MS.
The excretion rates of 15(S)-8-iso-PGF2α in our type 2 population (median [IQR] 176.1 [113.6-
235.8] pg/mg creatinine) are of the same magnitude as the excretion rates in the type 1
diabetes patient group reported by Wentholt and colleagues (median [IQR] 161 [140–217]
pg/mg creatinine) that were significantly higher than in their healthy control group
(median [IQR] 118 [101–146] pg/mg creatinine). Also, when we match 9 of our patients with
9 of the controls used in the Wentholt study according to age and sex, the PGF excretion
rate in our well-regulated type 2 diabetes patients is significantly larger (median [IQR]
controls 108.2 [90.3-144.0] and patients 243.3 [147.8-287.5], P = 0.006, Mann-Whitney U Test).
The excretion rates of the patient group reported by Monnier et al. are nearly threefold
higher (mean [SD] 482 [206] pg/mg creatinine) than our patient group. This disparity is
likely to be caused by the different methods used to assess 8-iso-PGF2α excretion rates as
explained earlier, together with the higher mean glucose and HbA1c of the Monnier
population, as a higher mean glucose is likely to cause more oxidative stress 3;19.
Our findings are supported by the only intervention trial assessing the effect of lowering
glucose variability on oxidative stress in type 2 diabetes patients 20. This crossover trial
comparing prandial vs. basal insulin did not find any influence of lowering glucose
variability on oxidative stress, assessed by the 24-hr excretion rates of 8-iso-PGF2α,
measured by HPLC-MS/MS. Additionally, no correlation between glucose variability (MAGE)
and oxidative stress was found in this insulin-treated population. Again, this is in line
with the recent findings of Monnier, who reported no relation between glucose variability
and urinary excretion of 15(S)-8-iso-PGF2α in insulin-treated type 2 diabetes patients.
No effect of glucose variability on oxidative stress in well regulated T2DM
Ch
apte
r 4
69
Despite the vast amount of studies using F2-isoprostanes as a biomarker of oxidative stress,
not all (patho) physiological factors contributing to the formation of F2-isoprostanes are
known 17. Multivariate regression analysis showed the amount of physical exercise and
age to be inversely related to oxidative stress activation in our study population. This has
been reported in another study, although for both parameters, also positive relations
and no relations have been reported (for review, see Basu and Helmersson 17).
A limitation of this study is that from the broad spectrum of markers of oxidative stress
we only measured the urinary excretion of F2-isoprostanes. Measurement of another
marker(s) could have complemented the present data. On the other hand, measurement
of F2-isoprostanes is regarded the best option for addressing lipid peroxidation 7;8. A
further limitation is the absence of people with higher HbA1c values on oral medication
in our cohort. We simply could not find such patients, who apparently had already
received treatment intensification according to current treatment guidelines.
Conclusions
We found no relevant relationship between glycaemic variability, assessed by continuous
glucose monitoring, and 15(S)-8-iso-PGF2α excretion, assessed by HPLC-MS/MS in a
population of well regulated type 2 diabetes patients. These results argue against a role
of glycaemic variability in the activation of oxidative stress in this group of patients,
and together with findings from literature suggest that the relation between glucose
variability and oxidative stress is only seen at higher HbA1c levels.
AcknowledgementsWe thank H. van Lenthe, Laboratory Genetic Metabolic Diseases, Academic Medical
Centre, Amsterdam, the Netherlands, who performed all the laboratory analyses of the
urine samples for determination of 15(S)-8-iso-PGF2α.
70
References1. The Diabetes Control and Complications Trial Research Group (1993) The Effect of Intensive Treatment of
Diabetes on the Development and Progression of Long-Term Complications in Insulin-Dependent Diabetes Mellitus. N Engl J Med 329: 977-986
2. Stratton IM, Adler AI, Neil HA, et al (2000) Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 321: 405-412
3. Brownlee M (2001) Biochemistry and molecular cell biology of diabetic complications. Nature 414: 813-8204. The Diabetes Control and Complications Trial Research Group (1995) The relationship of glycemic exposure
(HbA1c) to the risk of development and progression of retinopathy in the diabetes control and complications trial. Diabetes 44: 968-983
5. Hirsch IB, Brownlee M (2005) Should minimal blood glucose variability become the gold standard of glycemic control? J Diabetes Complications 19: 178-181
6. Monnier L, Mas E, Ginet C, et al (2006) Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 295: 1681-1687
7. Morrow JD, Hill KE, Burk RF, Nammour TM, Badr KF, Roberts LJ II (1990) A Series of Prostaglandin F2-Like Compounds are Produced in vivo in Humans by a Non-Cyclooxygenase, Free Radical-Catalyzed Mechanism. Proc Natl Acad Sci 87: 9383-9387
8. Roberts LJ, Morrow JD (2000) Measurement of F2-isoprostanes as an index of oxidative stress in vivo. Free Radic Biol Med 28: 505-513
9. Wentholt IME, Kulik W, Hoekstra JBL, de Vries JH (2008) Glucose fluctuations and activation of oxidative stress in type 1 diabetes patients. Diabetologia 51: 183-190
10. Proudfoot J, Barden A, Mori TA, et al (1999) Measurement of urinary F(2)-isoprostanes as markers of in vivo lipid peroxidation-A comparison of enzyme immunoassay with gas chromatography/mass spectrometry. Anal Biochem 272: 209-215
11. Saenger AK, Laha TJ, Edenfield MJ, Sadrzadeh SM (2007) Quantification of urinary 8-iso-PGF2alpha using liquid chromatography-tandem mass spectrometry and association with elevated troponin levels. Clin Biochem 40: 1297-1304
12. Monnier L, Colette C, Mas E, et al (2010) Regulation of oxidative stress by glycaemic control: evidence for an independent inhibitory effect of insulin therapy. Diabetologia 53: 562-571
13. Cockcroft DW, Gault MH (1976) Prediction of creatinine clearance from serum creatinine. Nephron 16: 31-4114. Siegelaar SE, Holleman F, Hoekstra JB, DeVries JH (2010) Glucose Variability; Does It Matter? Endocr.Rev. 31:
171-18215. Service FJ, Molnar GD, Rosevear JW, Ackerman E, Gatewood LC, Taylor WF (1970) mean amplitude of glycemic
excursions, a measure of diabetic instability. Diabetes 19: 644-65516. Rohlfing CL, Wiedmeyer HM, Little RR, England JD, Tennill A, Goldstein DE (2002) Defining the Relationship
Between Plasma Glucose and HbA1c: Analysis of glucose profiles and HbA1c in the Diabetes Control and Complications Trial. Diabetes Care 25: 275-278
17. Basu S, Helmersson J (2005) Factors regulating isoprostane formation in vivo. Antioxid Redox Signal 7: 221-235
18. Ceriello A, Bortolotti N, Motz E, et al (1998) Meal-generated oxidative stress in type 2 diabetic patients. Diabetes Care 21: 1529-1533
19. Davi G, Ciabattoni G, Consoli A, et al (1999) In Vivo Formation of 8-Iso-Prostaglandin F2α and Platelet Activation in Diabetes Mellitus : Effects of Improved Metabolic Control and Vitamin E Supplementation. Circulation 99: 224-229
20. Siegelaar SE, Kulik W, van Lenthe H, Mukherjee R, Hoekstra JB, DeVries JH (2009) A randomized controlled trial comparing the effect of basal insulin and inhaled mealtime insulin on glucose variability and oxidative stress. Diabetes, Obesity and Metabolism 11: 709-714
No effect of glucose variability on oxidative stress in well regulated T2DM
Ch
apte
r 4
71
Chapter 5
A randomised clinical trial comparing the effect of basal insulin and inhaled mealtime insulin on glucose variability and oxidative stress
Sarah E. Siegelaar, Wim Kulik, Henk van Lenthe, Robin Mukherjee,
Joost B.L. Hoekstra and J. Hans DeVries
Diabetes, Obesity and Metabolism 2009; 11(7):709-714
74
Abstract
Aim: To assess the effect of three times daily mealtime inhaled insulin therapy
compared with once daily basal insulin glargine therapy on 72-hr glucose
profiles, glucose variability and oxidative stress in type 2 diabetes patients.
Methods: In an inpatient crossover study, 40 subjects with type 2 diabetes
were randomised to receive 9 days of inhaled insulin three times daily before
meals or 9 days of glargine administered in the morning before breakfast in a
randomised order. During the last 72 hrs in each phase, glucose was measured
with continuous glucose monitoring. Activation of oxidative stress was measured
by determining the 15(S)-8-iso-PGF2α secretion in 24-hr urine samples.
Results: Inhaled insulin improved overall and postprandial glucose control
significantly better than insulin glargine (P <0.0001). There was a trend towards a
greater reduction in glucose variability (8-9%) in the inhaled group (P = 0.1430 and
P = 0.3298 for mean amplitude of glycaemic excursions (MAGE) and mean of daily
differences, respectively). Oxidative stress, estimated by determining the urinary
isoprostane excretion (15(S)-8-iso-PGF2α), was equally reduced from baseline by
both treatments. No correlation was found between glucose variability and
oxidative stress in both groups.
Conclusions: This study showed a mealtime insulin approach to improve
glycaemic control more than a basal insulin approach. These findings indicate
also that lowering glucose using insulin treatment lowers oxidative stress over
time, at least for the study period of 9 days, in type 2 diabetes patients. Contrary
to earlier data, we found no correlation between glucose variability (MAGE) and
oxidative stress (15(S)-8-iso-PGF2α) in this study.
Effect of basal and inhaled mealtime insulin on glucose variability and oxidative stress
Ch
apte
r 5
75
Introduction
It has been suggested that glucose variability, considered in combination with
haemoglobin A1c (HbA1c), is a more reliable indicator of blood glucose control and
the risk for long-term complications than HbA1c alone 1-3, and if so, this has important
bearings for the people with diabetes. Still the impact of glucose variability on
macrovascular complications and its effect in type 2 diabetic patients has not been
firmly established.
In an attempt to provide evidence for this hypothesis, Monnier et al. 4 showed a strong
correlation between glucose variability (expressed as mean amplitude of glycaemic
excursions [MAGE]) and oxidative stress (measured as 8-isoprostane excretion) in type
2 diabetes patients, suggesting a relationship between glucose variability and diabetic
complications. However, in type 1 diabetes patients, a disease with more pronounced
glucose variability, we could not find a correlation between MAGE and urinary
8-isoprostane excretion 5. To further test this hypothesis, an intervention study aiming
at reducing variability specifically in the intervention group without affecting mean
glycaemia more than in the control group is rational. Hirsch and Brownlee 1 suggested
to perform a randomised controlled trial comparing a regimen of mealtime and basal
insulin with a regimen of basal insulin alone in newly diagnosed type 2 diabetes patients.
The present study compared a mealtime insulin regimen (inhaled insulin) with a basal
insulin regimen (insulin glargine) in a crossover design in type 2 diabetes patients failing
on oral medication. In a study by Bretzel et al. 6 using a similar design with a rapid-acting
analogue rather than inhaled insulin, indeed overall glycaemia was reduced by both
treatments to a similar extent, while postprandial values were lower in the mealtime
insulin group and fasting glucose was lower in the basal insulin group. Thus, glucose
variability was by necessity more reduced in the mealtime insulin group, making it
possible to study glucose variability in both groups independent from the influence of
overall glycaemia. We report the first intervention study specifically targeting glucose
variability in insulin-treated type 2 diabetes patients.
Methods
Subjects The study population consisted of patients with type 2 diabetes mellitus poorly controlled
on a combination of two oral agents. Patients had to be at least 18 years old with type 2
diabetes diagnosed ³6 months prior to study entry and currently be treated on a stable
dose of at least 2 oral hypoglycaemic agents for at least 1 month prior study entry,
76
with or without adjunct subcutaneous insulin. Patients had to have a screening HbA1c
³6.5% and ≤10.5% and a BMI ≤44 kg/m2. A complete list of in- and exclusion criteria can
be found in the paper describing the glycaemic outcome of this study. (Hompesch et
al.7). The study was carried out in accordance with the principles of the Declaration of
Helsinki and of Good Clinical Practice. All local regulatory requirements were followed.
Before entering the study, patients gave written informed consent after a detailed oral
and written explanation of the study procedures.
Study design and procedureThis was a prospective, open-label, randomised, two-period crossover trial. The study
consisted of a screening visit and two 9-day inpatient periods separated by a 7- to 10-day
washout period. Subjects agreed to participate in this study by signing an informed
consent. Subjects were randomised 1:1 to inhaled insulin (Exubera®, Pfizer, New York,
NY, USA) three times daily before meals or glargine (Lantus®, sanofi-aventis, Paris,
France) administered in the morning of the second day before breakfast, then switched
treatments for the second phase (Figure 1). Patients who had existing subcutaneous
insulin regimens discontinued those therapies prior to each inpatient stay but resumed
them during washout. All patients did continue their pre-study oral diabetic treatments
throughout the study.
Figure 1 Study design
On the fifth day of each inpatient stay, vascular access for frequent venous blood
sampling was secured and subjects were subsequently connected to an automated
glucose monitoring system (CGMS® system gold, Medtronic MiniMed, Northridge, CA, USA)
Effect of basal and inhaled mealtime insulin on glucose variability and oxidative stress
Ch
apte
r 5
77
inserted in the abdominal wall for 72 hrs. Average interstitial glucose (IG) levels calibrated
to blood glucose were stored at 5-min intervals. The CGMS device was calibrated according
to the manufacturer’s instructions. For analysis, only data generated in the 72 hrs between
06:00 hrs on day 6 and 06:00 hrs on day 9 for each phase were used. During both inpatient
CGMS glucose profile periods, urine and blood samples were collected. Urinary 8-iso-PGF2α
was sampled from 24-hr urine collection on days 1 and 8 of each inpatient period.
Laboratory measurements Activation of oxidative stress was estimated by determining the urinary isoprostane
excretion (15(S)-8-iso-PGF2α), using high performance liquid chromatography tandem
mass spectrometry (HPLC-MS/MS). Urine samples were collected and stored without
additives as 7 ml aliquots at -80 °C. Initial creatinine concentrations were established
by colourimetric Jaffé assay. Two millilitres of sample were mixed with 2H4-labelled 15(S)-
8-iso-PGF2α as internal standard and applied to an 8-iso-PGF2α immunoaffinity column
(Cayman Chemical Company, Ann Arbor, MI, USA). After washing, the extract was eluted,
evaporated to dryness (60 °C, N2) and reconstituted in 200 μl 0.05 mol/l formic acid-
ethanol (75:25, v/v); 50 μl was injected on the HPLC-MS/MS system. Chromatographic
separation was achieved on a modular HPLC system (Surveyor, Thermo Finnigan, San
Jose, CA, USA) consisting of a cooled autosampler (T = 12°C), a low-flow quaternary MS
pump and analytical HPLC column: Alltima C8, 2.1 x 150 mm, 5 μm (Alltech, Lexinton,
KY, USA). Samples were eluted with a flow rate of 200 μl/min and a programmed linear
gradient between A (0.01% HCOOH in H2O; v/v) and B (CH3CN): from t = 0 min 45% A,
55% B towards t = 3 min 30% A, 70% B towards t = 3.1 min 100% A until t = 6 min. MS/
MS analyses were performed on a TSQ Quantum AM (Thermo Finnigan) operated in the
negative ion electrospray ionisation mode. The surface-induced dissociation was set at
2 V; spray voltage was 3500 V and the capillary temperature was 400°C. In the MS/MS
experiments argon was used as collision gas at a pressure of 0.2 Pa; collision energy was
26 eV for the optimised transitions: m/z 353.24 m/z 193.10. The interassay (n = 5) and
intraassay (average of 5 days, n = 3) variability allowed for determination at physiological
concentrations with a coefficient of variation of <7%.
Assessment of glycaemic variabilityInterday glycaemic variability
The day-to-day variation of the glucose pattern was calculated with the mean of the daily
differences (MODD), which is defined as the mean of the absolute differences between
glucose values on day 2 and the corresponding values on day 1, at the same time 8;9.
Intraday glycaemic variability
This was calculated by the MAGE 10. The MAGE over 24 hrs is the mean of the absolute
differences between peak and nadir values over 24 hrs, with peaks (nadirs) defined as
78
glucose values preceded and followed by an increase (decrease) and decrease (increase),
respectively, in excess of at least 1 SD of the mean glucose. If a decrease of more than 1
SD was the first excursion, only peak-to-nadir excursions (>1 SD) were included in the
calculation of the MAGE and vice versa.
Statistical analysisNo sample size calculations were performed since this was an exploratory study.
Pharmacodynamic endpoints, including the mean 72-hr glucose profiles; the area
under the IG concentration-time curve over 72 hrs (IG-AUC0-72h); the 3-day mean IG-AUC0-
24h; postmeal IG-AUC0-4h for breakfast, lunch, and dinner; the 3-day mean IG-AUC0-6h at
night time; the 3-day mean maximum glucose concentration after breakfast IG-Cmax;
and the time to IG-Cmax (IG-tmax) were derived from the two glucose exposure profiles
measured with the CGMS. Glycaemic exposure measures (AUCs) were calculated using the
trapezoidal rule. To evaluate the variability of glucose exposure, the following parameters
were calculated: the SD of the mean 72-hr glucose concentration, the MAGE over 72 hrs 10, and the MODD of paired glucose levels over 72 hrs 8;9 using standard statistics. The
differences in 15(S)-8-iso-PGF2α levels are analyzed using a prespecified mixed effects model
procedure in SAS (version 8.02, SAS Institute, Cary, NC, USA) with treatment, sequence
and period as fixed effects and patient within sequence considered a random effect.
Normality assumptions were checked as necessary, with log transformation to improve
this. Correlation was calculated and univariate regression analysis was performed to
evaluate the relation between the excretion rate of 15(S)-8-iso-PGF2α and the markers
of glycaemic variability. Locally weighted polynomial regression (LOESS) curves were
added in Figure 2. Statistical significance for all results is expressed through 95% CIs,
whereby a finding is deemed significant when neither side of the confidence interval
crosses 1.0 (or 100%).
Results
A total of 40 patients with type 2 diabetes (male, 29; age, 57 ± 10 y; BMI, 31.9 ± 4.6 kg/m²;
HbA1C, 7.9 ± 1.0%; mean prebreakfast capillary glucose at randomization 8.1 ± 2.2 mmol/l;
type 2 diabetes duration 10.2 years [range 0.8 – 27.0]) were enrolled. Diabetes was treated
with oral antidiabetic medication (39 patients received metformin, 28 sulfonylurea, 11
rosiglitazone, and 5 pioglitazone); five patients were also treated with glargine. During
the study, two patients did not complete the two treatment sequences. Two patients
discontinued during glargine treatment, one during the first period (and consequently
did not receive inhaled insulin in the second period), and one during the second period
(having completed treatment on inhaled insulin in the first period).
Effect of basal and inhaled mealtime insulin on glucose variability and oxidative stress
Ch
apte
r 5
79
Figure 2 Correlation between oxidative stress and glucose variability in both groups Scatter plot showing correlation between 24-hr urinary excretion rates of 8-iso-PGF2α and mean amplitude of glycaemic excursions (MAGE) in both groups. This figure shows no correlation between 8-iso-PGF2α and MAGE: r = 0.18 (inhaled) and r = 0.28 (glargine). The lines are LOESS curves; 8-iso-PGF2α is expressed in nmol/mmol creatinine.
The mealtime targeted approach with inhaled insulin improved overall and postprandial
glucose control (expressed as total glycaemic exposure over 72 hrs; mean glucose
concentration over the final 72-hr period; the 3-day mean glycaemic exposure; 4-hr
postmeal glycaemic exposure after breakfast, lunch, dinner; and the 3-day mean
maximum glucose levels after breakfast) significantly better than the basal insulin
approach using glargine (P <0.0001; Table 1).
As hypothesised, glucose variability was more reduced in the inhaled group compared
to the glargine group, however not significantly (8-9% reduction; P = 0.1430, P = 0.3298
and P = 0.1613 for MAGE, MODD and SD respectively) (Table 1).
Oxidative stress, estimated by determining the urinary isoprostane excretion (15(S)-8-
iso-PGF2α), was reduced by both treatments (Table 2). There was no evidence of a period
effect, looking at the baseline 8-isoprostane values after the washout periods that were
higher than after finishing treatment. There was a trend towards a somewhat greater
reduction in oxidative stress in the glargine group compared to the inhaled group.
We found no correlation between oxidative stress (urinary isoprostane excretion) and
glucose variability (expressed as MAGE and MODD) (Figure 2). The correlation coefficients
(r) for the inhaled and glargine group were 0.18 and 0.28 for MAGE and 0.17 and 0.20
for MODD respectively.
80
Table 1 Glycaemic exposure and variability of glycaemic exposure
Inhaled (n = 38)
Glargine (n = 38)
Inh/Glap
(95% CI) P-valueq
Glycaemic exposure (mean ± SD)
Mean 72 hrs (mmol/l) 5.3 ± 0.6 6.0 ± 1.1 88 (84-93) <0.0001
AUC72h (mmol/l·hr) 380.3 ± 45.3 426.3 ± 89.2 89 (84-93) <0.0001
AUC0-24h 3-day mean (mmol/l·hr) 126.5 ± 15.1 142.5 ± 27.9 89 (84-93) <0.0001
AUC0-4h post breakfast (mmol/·l·hr) 25.9 ± 4.8 27.7 ± 52 93 (87-99) 0.0318
AUC0-4h post lunch (mmol/·l·hr) 19.0 ± 4.0 24.6 ± 5.7 78 (72-84) <0.0001
AUC0-4h post dinner (mmol/l·hr) 20.5 ± 4.0 26.3 ± 5.7 78 (72-84) <0.0001
AUC0-6h night time (mmol/l·hr) 29.6 ± 4.6 31.0 ± 6.4 96 (91-103) 0.2345
Cmax post breakfast (mmol/l·hr) 8.3 ± 1.8 8.8 ± 1.7 94 (87-100) 0.0572
tmax post breakfast (hrs) 1.5 ± 0.6 1.8 ± 0.7 83 (71-97) 0.0180
Variability of glycaemic exposure (mean ± SD)
SD 72 hrs (mmol/l) 1.5 ± 0.6 1.6 ± 0.6 92 (81-104) 0.1613
MAGE 72 hrs (mmol/l) 3.5 ± 1.4 3.7 ± 1.3 91 (79-104) 0.1430
MODD 72 hrs (mmol/l) 1.4 ± 0.5 1.5 ± 0.6 93 (81-107) 0.3289
p Treatment/reference ratio (%) of the estimates of the geometric means from the mixed model fitted to the natural-log transformed endpoint data. q P-value is for the estimated treatment effect between inhaled and glargine from the mixed model fitted to the natural-log transformed endpoint data. AUC, area under concentration-time curve; Cmax, maximum concentration; tmax, time to maximum concentration; SD, standard deviation; MAGE, mean amplitude of glycaemic excursions; MODD, mean of daily differences.
Table 2 15(S)-8-iso-PGF2α of creatinine
BaselineMean (SD)
End of treatmentMean (SD)
∆r
Mean (SD)99% CIs
Inhaled 0.06 (0.03) 0.05 (0.03) 0.01 (0.02) -0.020, 0.000
Glargine 0.06 (0.03) 0.05 (0.02) 0.02 (0.02) -0.030, -0.010
15(S)-8-iso-PGF2α of creatinine is expressed in nmol/mmol creatinine. r ∆ represents the difference in 15(S)-8-iso-PGF2α concentrations from baseline to end of treatment. s 99% confidence interval is chosen because of the small design of the study.
Effect of basal and inhaled mealtime insulin on glucose variability and oxidative stress
Ch
apte
r 5
81
Discussion
The present study assessed the effect of three times daily mealtime insulin (inhaled)
therapy compared to once daily basal insulin (glargine) therapy on 72-hr glucose profiles,
glucose variability and oxidative stress in type 2 diabetes patients.
At least two important conclusions can be drawn. First, this study reveals that glucose
lowering using insulin treatment lowers oxidative stress (Table 2). In both groups,
there was a significant decline in 8-isoprostane production during treatment. As there
was no indication of a period effect, a time effect can be excluded. Lowering oxidative
stress as a result of lowering glycaemia was so far only reported as a momentary effect.
Ceriello et al. 11 described earlier a significant reduction of postprandial oxidative
stress in type 1 diabetes patients when reducing postprandial glucose excursions with
pramlintide, an amylin analogue. They examined nitrotyrosine, oxidised LDL and total
radical-trapping antioxidant parameter during a 4-hr postprandial period. They found a
correlation between the extent of postprandial glycaemia and the oxidative stress. Our
data strengthen their findings and extend the hypothesis of an oxidative stress lowering
effect by lowering glucose to a period of 8 days.
Second, we found no correlation between oxidative stress, measured as 8-iso-PGF2α in
urine, and glucose variability, defined as MAGE (Table 1, Figure 2). This is in accordance
with an earlier study in type 1 diabetes patients 5. On the other hand, Monnier et al. 4
reported a strong relationship between glucose variability and oxidative stress in type 2
diabetes patients. An explanation for the lack of correlation between glucose variability
and oxidative stress can be a methodological issue. We used next to immunoaffinity
isolation highly selective HPLC tandem MS for detection instead of the less specific
enzyme immunoassay to quantify 8-isoprostanes: HPLC-MS/MS is not hampered by cross-
reactivity of structurally (un)related components of 8-iso-PGF2α, whereas the immunoassay
is more susceptible to interference, as acknowledged in the earlier Monnier report 4. An
alternative possibility is that a relationship between glucose variability and oxidative
stress only exists in non-insulin treated type 2 diabetes patients. Recently, Ceriello
reported a clamp study 3 suggesting that oscillating glucose can have more deleterious
effects than constant high glucose on endothelial function and oxidative stress. However,
in this study only two glucose excursions were elicited, what in our opinion shows that
a second repetitive episode of acute hyperglycaemia elicits more oxidative stress than
the first 12, which is somewhat different than glucose variability over the whole day.
Looking at the consequence of oxidative stress, that is vascular complications, Gordin
et al. 13 detected no correlation between glucose variability (expressed as MAGE) and
arterial stiffness, considered an early sign of macrovascular complications, in type 1
diabetes patients.
82
Overall glycaemic control was significantly better with a mealtime insulin approach
using inhalation insulin compared to a once daily approach using glargine. Literature
is not in agreement on this subject, varying from no difference between both regimens 6 to significantly better glycaemic control (expressed as decrease in HbA1c) when using
a mealtime approach in patients with type 2 diabetes starting with insulin therapy 14.
The overall better glycaemic control in the mealtime insulin group confounded the
comparison in oxidative stress between both groups, as lowering glycaemia lowers
oxidative stress.
In our study, glucose variability in the mealtime insulin group was somewhat lower
than in the once daily insulin group (Table 1) albeit not significantly. We think that the
non-significance is likely mainly explained by the small and therefore underpowered
study group. As postprandial hyperglycaemia accounts for the major part of glucose
variability 15, certainly in type 2 diabetes patients who experience few hypoglycaemic
episodes, one would expect glucose variability to be smaller with a mealtime insulin
approach. Glucose variability in the inhaled group was enlarged by the twofold higher
incidence of mild and moderate hypoglycaemia in the inhaled group 7. It is therefore
likely that the variability in the mealtime insulin group would have been lower if overall
glycaemic control would have been the same. It is also possible that glucose variability in
the basal insulin group is lower than expected. This could be explained by the residual
beta-cell function of the patients treated until recently with oral medication, mitigating
the postprandial glucose to rise even without mealtime insulin administration.
We have no clear explanation why oxidative stress seemed somewhat more lowered in
the glargine group. From the literature one would have expected lower oxidative stress in
the mealtime insulin group, resulting either from better overall glycaemic control in the
inhaled group or from reduced variability 16;17 . Possibly, the lower oxidative stress level
in the basal insulin group is a spurious finding. Again, a larger trial would be needed
to answer this question more definitively.
In conclusion, this study shows that lowering glucose lowers oxidative stress in type 2
diabetes patients not only as a momentary effect, extending the existing data of Ceriello
to a longer period 11. Second, we found no correlation between glucose variability and
oxidative stress in insulin-treated type 2 diabetes patients. Finally, a non-significant
almost 10% decline in glucose variability did not result in lower oxidative stress in
insulin-treated type 2 diabetes patients.
AcknowledgementsThis study was supported financially by Pfizer Inc.
Effect of basal and inhaled mealtime insulin on glucose variability and oxidative stress
Ch
apte
r 5
83
References1. Hirsch IB, Brownlee M (2005) Should minimal blood glucose variability become the gold standard of glycemic
control? J Diabetes Complications 19: 178-1812. Brownlee M, Hirsch IB (2006) Glycemic variability: a hemoglobin A1c-independent risk factor for diabetic
complications. JAMA 295: 1707-17083. Ceriello A, Esposito K, Piconi L, et al (2008) Oscillating glucose is more deleterious to endothelial function
and oxidative stress than mean glucose in normal and type 2 diabetic patients. Diabetes 57: 1349-13544. Monnier L, Mas E, Ginet C, et al (2006) Activation of oxidative stress by acute glucose fluctuations compared
with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 295: 1681-16875. Wentholt IME, Kulik W, Hoekstra JBL, DeVries JH (2008) Glucose fluctuations and activation of oxidative
stress in type 1 diabetes patients. Diabetologia 51: 183-1906. Bretzel RG, Nuber U, Landgraf W, Owens DR, Bradley C, Linn T (2008) Once-daily basal insulin glargine versus
thrice-daily prandial insulin lispro in people with type 2 diabetes on oral hypoglycaemic agents (APOLLO): an open randomised controlled trial. Lancet 371: 1073-1084
7. Hompesch M, Kollmeier A, Rave K, et al (2009) Glycemic exposure is affected favorably by inhaled human insulin (Exubera) as compared with subcutaneous insulin glargine (Lantus) in patients with type 2 diabetes. Diabetes Technol Ther 11: 307-313
8. McDonnell CM, Donath SM, Vidmar SI, Werther GA, Cameron FJ (2005) A novel approach to continuous glucose analysis utilizing glycemic variation. Diabetes Technol Ther 7: 253-263
9. Molnar GD, Taylor WF, Ho MM (1972) Day-to-day variation of continuously monitored glycaemia. Diabetologia 8: 342-348
10. Service FJ, Molnar GD, Rosevear JW, Ackerman E, Gatewood LC, Taylor WF (1970) Mean amplitude of glycemic excursions, a measure of diabetic instability. Diabetes 19: 644-655
11. Ceriello A, Piconi L, Quagliaro L, et al (2005) Effects of pramlintide on postprandial glucose excursions and measures of oxidative stress in patients with type 1 diabetes. Diabetes Care 28: 632-637
12. Brownlee M (2005) The pathobiology of diabetic complications: a unifying mechanism. Diabetes 54: 1615-1625
13. Gordin D, Rönnback M, Forsblom C, Mäkinen V, Saraheimo M, Groop P-H (2008) Glucose variability, blood pressure and arterial stiffness in type 1 diabetes. Diabetes Res Clin Pract 80: e4-e7
14. Kazda C, Hulstrunk H, Helsberg K, Langer F, Forst T, Hanefeld M (2006) Prandial insulin substitution with insulin lispro or insulin lispro mid mixture vs. basal therapy with insulin glargine: A randomized controlled trial in patients with type 2 diabetes beginning insulin therapy. J Diabetes Complications 20: 145-152
15. McCall AL, Cox DJ, Crean J, Gloster M, Kovatchev BP (2006) A novel analytical method for assessing glucose variability: using CGMS in type 1 diabetes mellitus. Diabetes Technol Ther 8: 644-653
16. Flores L, Rodela S, Abian J, Claria J, Esmatjes E (2004) F2 isoprostane is already increased at the onset of type 1 diabetes mellitus: effect of glycemic control. Metabolism 53: 1118-1120
17. Shamir R, Kassis H, Kaplan M, Naveh T, Shehadeh N (2008) Glycemic control in adolescents with type 1 diabetes mellitus improves lipid serum levels and oxidative stress. Pedriatics Diabetes 9: 104-109
Chapter 6
Glucose variability does not contribute to the development of peripheral and autonomic neuropathy in type 1 diabetes: data from the DCCT
Sarah E. Siegelaar, Eric S. Kilpatrick, Alan S. Rigby, Steven L. Atkin,
Joost B.L. Hoekstra and J. Hans DeVries
Published in abbreviated form, Diabetologia 2009; 52(10):2229-2232
86
Abstract
Aims/hypothesis: While the presence of an effect of glycaemic variability
on retinopathy and nephropathy has been negated, it is unknown whether
glycaemic variability may influence neuropathy. We analysed data from the
Diabetes Control and Complications Trial (DCCT) dataset to assess whether
glycaemic variability is a risk factor for the development of diabetic neuropathy.
Methods: Seven-point glucose profiles were collected quarterly during the DCCT
in 1,441 type 1 diabetes patients. Peripheral and autonomic neuropathies were
assessed at baseline and at 5 and 4 years follow-up, respectively. The effect of
glycaemic variability, expressed as standard deviation (SD) and mean amplitude
of glycaemic excursions (MAGE), on the development of neuropathy in addition to
HbA1c and mean glucose was assessed using a logistic regression model, adjusted
for age, sex, disease duration, treatment group and prevention cohort.
Results: Glucose variability had no significant effect on the incidence of clinical
neuropathy confirmed by autonomic or electromyography abnormalities (SD,
odds ratio [OR] 1.07, 95% confidence interval [CI] 0.83-1.35; MAGE, 1.06 [0.96-
1.20]) or clinical neuropathy alone (SD, 0.95 [0.77-1.18]; MAGE, 1.01 [0.91-1.11]). It
appeared to have a significant effect on overall autonomic dysfunction but not
when adjusting for HbA1c or mean glucose (SD, OR 1.08 and 1.09, CI 0.82-1.44
and 0.80-1.48, respectively; MAGE, OR 1.09 and 1.10, CI 0.96-1.23 and 0.97-1.25
respectively).
Conclusions: Glucose variability was not an additional risk factor in the
development of diabetic peripheral or autonomic neuropathy over and above
HbA1c or mean glucose in the DCCT.
No contribution of glucose variability to development of neuropathy in T1DM
Ch
apte
r 6
87
Introduction
Diabetes is one of the most common causes of small fibre neuropathy causing sensory
symptoms as pain and numbness as well as autonomic dysfunction. Large fibre
involvement is frequent, which makes neuropathy an important complication of
diabetes, potentially leading to disability and premature deaths 1;2. It is thought that
hyperglycaemia causes direct damage to the nerve parenchyma as well as indirect
hyperglycaemia-induced neuronal ischemia by decreases in neuronal flow 3. Good
glycaemic control is a proven robust measure to delay or prevent the development of
diabetic polyneuropathy 4;5. Other cardiovascular risk-factors such as body mass index and
hypertension are associated with this complication in type 1 diabetes and are therefore
assessed in prevention programmes 4.
It is suggested that, in addition to hyperglycaemia, glucose variability can contribute to
the severity and development of diabetic neuropathy because the nervous system may be
particularly susceptible to glycaemic fluctuations 6. On the other hand, glucose variability
is not related to the development of retinopathy and nephropathy in type 1 diabetes 6;7.
To determine any additional effect of glucose variability, above that assessed by HbA1c
and mean glucose, on peripheral and autonomic diabetic neuropathy, we analyzed the
data from the Diabetes Control and Complications Trial (DCCT).
Methods
The datasetsWe used for this study the datasets collected during the DCCT (publicly accessible at,
www.gcrc.med.umn.edu/gcrc/downloads/dcct.html, accessed 23-27 January 2009). The
DCCT was a 9-year follow-up study of 1,441 patients with type 1 diabetes comparing the
effects of intensive versus conventional treatment on the development of microvascular
complications and neuropathy 8. A standardised neurologic history, physical examination
and nerve conduction studies were done by DCCT neurologists at baseline, 5 years, and
study end. Autonomic nervous system tests were performed at baseline and biennially
thereafter 9. We included only data from baseline to 4 years (autonomic function data)
or 5 years of follow-up in the analyses as more than 50% of the patients did not have
records of glucose data after 5 years of follow-up.
Definition of eventsClinical neuropathy was defined as abnormal findings in two or more of the following
categories in the absence of other known causes of neuropathy: neuropathic symptoms
(dysesthesias, paresthesias, hypersensitivity to touch and burning pain), sensory deficits
88
(light touch, position, temperature and pin-prick) or deep tendon reflexes 9. The nerve
conduction studies consisted of median motor and sensory, peroneal motor and sural
sensory nerve conduction velocities; distal latencies and amplitudes; and median and
peroneal motor F-wave latencies using a standard protocol 9. Abnormal nerve conduction
was considered present when at least one measured attribute was abnormal in at least
two anatomically distinct nerves. Autonomic nervous system function was assessed using
three tests: beat-to-beat heart rate variation (R-R variation) during deep breathing and
during a standardised Valsalva maneuver, and postural blood pressure testing 9. Abnormal
autonomic function was determined as at least one abnormal autonomic function
test. The main neurological endpoint of the DCCT was the development of confirmed
clinical neuropathy, defined as clinical neuropathy confirmed by either abnormal nerve
conduction or autonomic nervous system testing.
We studied the effect of glucose variability on the main neurological endpoint, i.e.
confirmed clinical neuropathy, and on the DCCT-defined secondary endpoints separately:
clinical neuropathy, abnormal nerve conduction studies, and abnormal autonomic
function. In addition, we determined its effect on the subvariables median motor F-wave
latency, sural amplitude, sensory signs, and beat-to-beat heart-rate variation (with Valsalva
ratio <1.5), as these variables tend to be the first affected by diabetes.
Glycaemic variablesDuring the DCCT a seven-point blood glucose profile was collected every 3 months (pre
breakfast, post breakfast, pre lunch, post lunch, pre supper, post supper and bedtime). An
additional data point was collected during the night, but since this was only measured
in <1% of the subjects, it is left out of further analysis. We included all profiles with five
observations or more during the 24-hr period, extrapolating missing values from the
surrounding points 10. Mean blood glucose was calculated by the area under the curve
(AUC) using the trapezoidal rule 11. Variability of blood glucose was calculated as the SD of
daily blood glucose around the mean from each quarterly visit (within-day SD) 12 and the
mean amplitude of glycaemic excursions (MAGE) 13. Last, we calculated the mean SD from
individual glucose data transformed to a symmetric distribution according to Kovatchev 14. Glucose variability from baseline to 4 or 5 years was assessed as the mean SD and
mean MAGE from the first quarter till the 16th or 20th quarter of follow-up, respectively.
Statistical analysisThe relationship between glucose variability and the development of each diabetic
neuropathy variable was assessed by a logistic regression model from which odds ratios
(OR) and 95% confidence intervals (CI) were calculated. Patients with a positive baseline
score on the neuropathy parameter studied were excluded from analysis. All regression
models were adjusted for the following baseline covariates: age (years), sex, disease
No contribution of glucose variability to development of neuropathy in T1DM
Ch
apte
r 6
89
duration (years), randomization treatment (conventional vs. intensive) and prevention
cohort (primary vs. secondary). Finally, the additional effect of glucose variability on
neuropathy separate from the effect of HbA1c and mean glucose (AUC) was computed
using the same technique. Statistical analysis was performed using SPSS version 16.0.2.
A P-value <0.05 was considered significant.
Results
The main characteristics of the patients in the group analysed for confirmed clinical
neuropathy are listed in Table 1. Of the 1,441 patients in total, 1,160 were included in
this specific analysis. Ninety-two patients were excluded from the analysis because they
had a positive score at baseline, and 189 patients had missing data on confirmed clinical
neuropathy at baseline (n = 3) or at 5 years (n = 186). The numbers of participants with
data analysed in the other specific neuropathy groups are listed in Table 2.
Table 1 Patient characteristics in the group analysed for confirmed clinical
neuropathy
Confirmed clinical neuropathy
Yes (n = 108) No (n = 1052) P - value
Age at baseline, years 28.28 (6.77) 26.40 (7.10) 0.008
Male sex, n (%) 52 (48) 533 (53) 0.38
Diabetes duration at baseline, months 79.59 (45.62) 68.60 (49.79) 0.02
Conventional treatment, n (%) 80 (74) 517 (49) <0.001
Primary prevention cohort, n (%) 35 (32) 503 (48) 0.002
HbA1c (%) 9.10 (1.58) 8.08 (1.43) <0.001
Mean glucose, mmol/l 13.51 (3.33) 11.52 (3.78) <0.001
MAGE, mmol/l 8.00 (1.96) 7.55 (1.90) 0.02
SD, mmol/l 4.24 (0.89) 4.05 (0.93) 0.04
SD TF, mmol/l 0.75 (0.17) 0.81 (0.16) <0.001
Data are means (SD), unless stated otherwise in parentheses. For this analysis patients with a positive or missing score for confirmed clinical neuropathy at baseline were excluded (neuropathy, n = 92; missing, n = 3). Patients with a missing score at 5 years are also excluded from the analysis (n = 186). P-values are comparisons between groups (independent samples t-test). MAGE, mean amplitude of glycaemic excursions; SD, standard deviation; SD TF, standard deviation obtained from glucose data transformed according to Kovatchev et al.: transformed blood glucose = 1.794*([log{BG}]1.026 - 1.861) 14.
90
Tab
le 2
Bin
ary
log
isti
c re
gre
ssio
n a
nal
ysis
rel
atin
g th
e ef
fect
of
dif
fere
nt
gly
caem
ic v
aria
ble
s to
neu
rolo
gic
al
com
pli
cati
on
s, a
s d
efin
ed b
y th
e D
CC
T
Co
nfi
rmed
cli
nic
al n
euro
pat
hya
Cli
nic
al n
euro
pat
hy
aA
uto
no
mic
neu
rop
ath
y b
Ab
no
rmal
ner
ve c
on
du
ctio
n a
n =
108
/116
0 c
n =
148
/111
3 c
n =
79/
1258
cn
= 2
07/8
13 c
mod
elO
R (
95 %
CI)
PO
R (
95%
CI)
PO
R (
95%
CI)
PO
R (
95%
CI)
P
Hb
A1c
1.64
(1.3
7-1.
95)
<0.0
011.
40 (1
.20-
1.64
)<0
.001
1.51
(1.2
3-1.
85)
<0.0
011.
62 (1
.39-
1.90
)<0
.001
AU
C1.
17 (1
.09-
1.26
)<0
.001
1.13
(1.0
5-1.
21)
0.00
11.
15 (1
.05-
1.26
)0.
003
1.15
(1.0
7-1.
23)
<0.0
01
SD1.
07 (0
.83-
1.35
)0.
670.
95 (0
.77-
1.18
)0.
661.
30 (0
.99-
1.70
)0.
061.
17 (0
.96-
1.42
)0.
12
SD (H
bA
1c)
0.86
(0.6
7-1.
10)
0.24
0.82
(0.6
6-1.
03)
0.08
1.08
(0.8
2-1.
44)
0.58
0.94
(0.7
5-1.
14)
0.46
SD (A
UC
)0.
85 (0
.65-
1.10
)0.
210.
78 (0
.61-
0.99
)0.
041.
09 (0
.80-
1.48
)0.
590.
95 (0
.76-
1.19
)0.
68
MA
GE
1.06
(0.9
6-1.
20)
0.23
1.01
(0.9
1-1.
11)
0.90
1.16
(1.0
3-1.
30)
0.01
1.07
(0.9
8-1.
17)
0.14
MA
GE
(Hb
A1c
)1.
00 (0
.89-
1.12
)0.
950.
96 (0
.86-
1.06
)0.
371.
09 (0
.96-
1.23
)0.
180.
98 (0
.90-
10.8
)0.
74
MA
GE
(AU
C)
1.00
(0.8
8-1.
12)
0.93
0.94
(0.8
5-1.
05)
0.29
1.10
(0.9
7-1.
25)
0.13
1.00
(0.9
1-1.
10)
0.93
SD T
F0.
14 (0
.04-
0.52
)0.
003
0.15
(0.0
5-0.
48)
0.00
10.
63 (0
.13-
3.07
)0.
570.
16 (0
.06-
0.47
)0.
001
SD T
F (H
bA
1c)
0.36
(0.0
9-1.
37)
0.13
0.25
(0.0
7-0.
86)
0.03
1.42
(0.2
6-7.
71)
0.68
0.62
(0.2
0-1.
87)
0.39
SD T
F (A
UC
)0.
33 (0
.08-
1.39
)0.
130.
26 (0
.08-
0.87
)0.
031.
27 (0
.25-
6.29
)0.
770.
61 (0
.19-
1.94
)0.
40
HbA
1c, A
UC
, SD
, MA
GE
and
SD
TF
rep
rese
nt
mea
ns
from
qu
arte
rly
visi
t 1-
16b
or 1
-20a . c
Pat
ien
ts w
ith
a p
osit
ive
neu
rop
ath
y sc
ore
at fi
ve y
ears
/com
ple
te a
nal
ysis
gro
up
p
er p
aram
eter
. All
mod
els
are
adju
sted
for
base
lin
e co
vari
ates
(sex
, age
, dis
ease
du
rati
on, p
reve
nti
on c
ohor
t, r
and
omiz
atio
n t
reat
men
t). S
D (H
bA1c
), M
AG
E (H
bA1c
), SD
TF
(HbA
1c),
SD (A
UC
), M
AG
E (A
UC
) an
d S
D T
F (A
UC
) are
six
dis
tin
ct m
odel
s ad
dit
ion
ally
ad
just
ed f
or H
bA1c
or
AU
C a
par
t fr
om t
he
base
lin
e co
vari
ates
. OR
, od
ds
rati
o;
CI,
con
fid
ence
in
terv
al; A
UC
, are
a u
nd
er t
he
curv
e; S
D, s
tan
dar
d d
evia
tion
; MA
GE,
mea
n a
mp
litu
de
of g
lyca
emic
exc
urs
ion
s; S
D T
F, s
tan
dar
d d
evia
tion
obt
ain
ed f
rom
gl
uco
se d
ata
tran
sfor
med
acc
ord
ing
to K
ovat
chev
: tra
nsf
orm
ed b
lood
glu
cose
= 1
.794
*([l
og{B
G}]
1.02
6 -1.8
61) 14
.
No contribution of glucose variability to development of neuropathy in T1DM
Ch
apte
r 6
91
Logistic regression analysis showed no effect of glucose variability, computed as the mean
SD and mean MAGE from the seven-point glucose profiles from quarterly visit 1-20 (first
5 years), on confirmed clinical neuropathy, the main neuropathy endpoint of the DCCT
(Table 2). Dividing the variability parameters in quartiles and performing the analysis
per randomization group did not change the outcome (data not shown).
No effect of glycaemic variability on clinical neuropathy was seen, with exception of
a small protective effect of the SD adjusted for AUC (Table 2). In addition, no effect of
glycaemic variability was seen on the incidence of sensory signs (SD, 1.00 [0.82-1.22], P =
0.99; MAGE, 1.02 [0.93-1.12], P = 0.69) as well as in separate analysis of the F-wave latency
of the median nerve (SD, 1.12 [0.84-1.49], P = 0.44; MAGE, 1.03 [0.90-1.17], P = 0.67) and the
amplitude of the sural nerve (SD, 1.27 [1.00-1.60], P = 0.05; MAGE, 1.05 [0.95-1.17], P = 0.34).
Glycaemic variability seemed to have an effect on autonomic neuropathy, but this effect
disappeared when adjusting the model for HbA1c or AUC (Table 2). Analysing both
randomization groups separately also did not reveal a relation over HbA1c (data not
shown). Separate examination of the three autonomic function parameters showed that
only for beat-to-beat heart rate variation during a Valsalva manoeuvre did the effect
remain significant when adjusting for mean glucose (SD, 2.64 [1.17-5.94], P = 0.02; MAGE,
1.42 [1.07-1.90], P = 0.02), but not when adjusting for HbA1c (SD, 1.84 [0.90-3.76], P = 0.09;
MAGE 1.30 [0.98-1.72], P = 0.07). There was no effect of glycaemic variability on beat-to-
beat heart-rate variation during deep breathing and postural blood pressure testing
(data not shown).
HbA1c and AUC itself were strong predictors of any form of neuropathy as described
above and transformation of the individual glucose data according to Kovatchev 14 did
not alter the results (Table 2).
Discussion
In this study, glycaemic variability did not influence the development of neuropathy
over HbA1c or mean glucose. HbA1c and mean glucose itself were strong predictors for
the development of diabetic neuropathy. These results are in line with earlier analysis
of DCCT data describing no influence of glycaemic variability on the development or
progression of retinopathy and nephropathy 7.
Bragd et al. 6 found that glucose variability (SD) was a borderline predictor of the
incidence of peripheral neuropathy in 100 type 1 diabetes and with a follow-up period
of 11 years (P = 0.07; HR 1.73, range 0.94-3.19). Peripheral neuropathy in their study was
92
defined as sensory neuropathy, as indicated by monofilament testing, and an abnormal
EMG and/or vibration test. This same study showed a significant relationship between
SD and the prevalence of peripheral neuropathy (P = 0.03; OR 2.34, range 1.06-5.20),
perhaps suggesting that the nervous system may be particularly susceptible to glycaemic
fluctuations 6. Another cross-sectional study investigated the relation between glucose
variability and the presence of pain in 20 type 1 diabetes patients with established
peripheral neuropathy. Compared to the group without symptoms (n = 10), the group with
painful symptoms had more glycaemic excursions, although there was no difference in
MAGE 15. Since the groups were neither matched nor the effect adjusted for mean glucose,
the significantly larger mean glucose in the painful group is more likely to explain the
difference between the groups. In the DCCT no separate distinction was made for pain
as a symptom so the outcome measure is not exactly comparable.
We did find a relation between glucose variability and two neuropathy parameters: the
autonomous parameter beat-to-beat heart rate variation <20 combined with a Valsalva
ratio >1.5 as well as clinical neuropathy, both independent from AUC. These results are
likely the consequence of multiple testing. When adjusting for multiple testing using the
Holm method 16 a P-value of 0.0125 would be needed to reject the H0 hypothesis, which
is smaller than the P-value of 0.02 and 0.04 we found for the autonomic neuropathy
parameter and clinical neuropathy respectively. Multiple testing is also the most likely
explanation for the odds ratio’s smaller than 1 found for some of the clinical neuropathy
parameters (Table 1).
What strengthens our results is that we did not find a relationship between glucose
variability and sensory signs or median motor F-wave latency and sural amplitude, the
earliest indicators of diabetic neuropathy. As diabetic neuropathy is mostly a small-fibre
disease, sensory signs are usually the presenting sign of the disease and they are a stable
and reliable measure of disease status or progression 17. Although EMG studies measure
large-fibre function, median motor F-wave latency and sural amplitude are the most
sensitive of all EMG measures to detect diabetic neuropathy 18;19.
A limitation of this study is that the variability parameters are calculated from seven-
point glucose curves by self-monitoring. Continuous glucose monitoring (CGM) might
detect fluctuations occurring between two measurements that would be missed by self-
monitoring of blood glucose. Also the DCCT participants did not collect all profiles as
required, resulting in missing values. However, they were highly motivated, thus limiting
missing data to a minimum. Another difficulty is that the neuropathy variables were
infrequently scored. We decided to focus on events up to 4 (autonomic neuropathy) and
5 (clinical neuropathy) years because after 5 years of follow-up in more than 50% of the
patients the glucose data has not been recorded. It might be possible that the analysis
No contribution of glucose variability to development of neuropathy in T1DM
Ch
apte
r 6
93
at this point has been hampered by a power problem due to too few events. Possibly the
DCCT/Epidemiology of Diabetes Interventions and Complications (EDIC) follow-up will
provide more endpoints as the same neuropathy parameters assessed in the DCCT are
measured in years 13 or 14 of its follow-up (2007-8; www.niddkrepository.org). These
data have not yet been released.
In conclusion, glucose variability was not a risk factor separate from HbA1c or mean
glucose in the development of diabetic peripheral neuropathy in the DCCT.
AcknowledgementsThe Diabetes Control and Complications Trial (DCCT) and its follow-up the Epidemiology
of Diabetes Interventions and Complications (EDIC) study were conducted by the DCCT/
EDIC Research Group and supported by National Institute of Health (NIH) grants and
contracts and by the General Clinical Research Center Program, NCRR. This manuscript
was not prepared under the auspices of the DCCT/EDIC study and does not represent
analyses nor conclusions of the DCCT/EDIC study group nor the NIH.
94
References1. Lauria G (2005) Small fibre neuropathies. Current Opinion in Neurology 18: 591-5972. Freeman R (2005) Autonomic peripheral neuropathy. Lancet 365: 1259-12703. Sugimoto K, Murakawa Y, Sima AAF (2000) Diabetic neuropathy - a continuing enigma. Diabetes Metab Res
Rev 16: 408-4334. Tesfaye S, Chaturvedi N, Eaton SEM, et al (2005) Vascular Risk Factors and Diabetic Neuropathy.
N Engl J Med 352: 341-3505. The Diabetes Control and Complications Trial (DCCT) Research Group (1995) The effect of intensive diabetes
therapy on the development and progression of neuropathy. Ann Intern Med 122: 561-5686. Bragd J, Adamson U, Backlund LB, Lins PE, Moberg E, Oskarsson P (2008) Can glycaemic variability, as
calculated from blood glucose self-monitoring, predict the development of complications in type1 diabetes over a decade? Diabetes & Metabolism 34: 612-616
7. Kilpatrick ES, Rigby AS, Atkin SL (2006) The effect of glucose variability on the risk of microvascular complications in type 1 diabetes. Diabetes Care 29: 1486-1490
8. The DCCT Research Group (1993) The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus.N Engl J Med 329:977-986
9. The Diabetes Control and Complications Trial Research Group (1988) Factors in development of diabetic neuropathy. Baseline analysis of neuropathy in feasibility phase of the Diabetes Control and Complications Trial (DCCT). Diabetes 37: 476-481
10. Kilpatrick ES, Rigby AS, Atkin SL (2008) Mean blood glucose compared with HbA1c in the prediction of cardiovascular disease in patients with type 1 diabetes. Diabetologia 51: 365-371
11. Rohlfing CL, Wiedmeyer HM, Little RR, England JD, Tennill A, Goldstein DE (2002) Defining the relationship between plasma glucose and HbA1c: analysis of glucose profiles and HbA1c in the DCCT. Diabetes Care 25: 275-278
12. Moberg E, Kollind M, Lins P, Adamson U (1993) Estimation of blood-glucose variability in patients with insulin-dependent diabetes mellitus. Scand J Clin Lab Invest 53: 507-514
13. Service FJ, Molnar GD, Rosevear JW, Ackerman E, Gatewood LC, Taylor WF (1970) Mean amplitude of glycemic excursions, a measure of diabetic instability. Diabetes 19: 644-655
14. Kovatchev BP, Cox DJ, Gonder-Frederick LA, Clarke W (1997) Symmetrization of the blood glucose measurement scale and its applications. Diabetes Care 20: 1655-1658
15. Oyibo SO, Prasad YDM, Jackson NJ, Jude EB, Boulton AJM (2002) The relationship between blood glucose excursions and painful diabetic peripheral neuropathy:a pilot study.Diabetic Medicine 19:870-873
16. Aickin M, Gensler H (1996) Adjusting for multiple testing when reporting research results: the Bonferroni vs Holm methods. Am J Public Health 86: 726-728
17. Bril V (1999) NIS-LL: The primary measurement scale for clinical trial endpoints in diabetic peripheral neuropathy. European Neurology 41: 8-13
18. Sima AAF (1992) Structure-function interactions in the therapeutic response of diabetic neuropathy. Journal of Diabetes and its Complications 6: 64-68
19. Perkins BA, Bril V (2003) Diabetic neuropathy: a review emphasizing diagnostic methods. Clinical Neurophysiology 114: 1167-1175
No contribution of glucose variability to development of neuropathy in T1DM
Ch
apte
r 6
95
Chapter 7
A decrease in glucose variability does not reduce cardiovascular event rates in type 2 diabetes patients after acute myocardial infarction: a reanalysis of the HEART2D study
Sarah E. Siegelaar, Lisa Kerr, Scott J. Jacober and J. Hans DeVries
Diabetes Care 2011; 34(4): 855-857
98
Abstract
Objective: To assess the effect of intraday glucose variability (GV) on
cardiovascular outcomes in a reanalysis of Hyperglycaemia and Its Effect After
Acute Myocardial Infarction on Cardiovascular Outcomes in Patients With Type
2 Diabetes Mellitus (HEART2D) study data.
Research Design and Methods: Type 2 diabetes patients after acute myocardial
infarction were randomised to an insulin treatment strategy targeting
postprandial (PRANDIAL; n = 557) or fasting/interprandial (BASAL; n = 558)
hyperglycaemia. GV was calculated as mean amplitude of glycaemic excursions
(MAGE), mean absolute glucose (MAG) change, and standard deviation (SD).
Results: The PRANDIAL strategy resulted in an 18% lower MAG than BASAL (mean
[SEM] difference 0.09 [0.04] mmol/l/hr, P = 0.02). Also, MAGE and SD were lower in
the PRANDIAL group, however not significantly. HbA1c levels and cardiovascular
event rates were comparable between groups.
Conclusions: A PRANDIAL strategy demonstrated lower intraday GV vs. a BASAL
strategy with similar overall glycaemic control but did not result in a reduction
in cardiovascular outcomes. This does not support the hypothesis that targeting
GV would be beneficial in reducing subsequent secondary cardiovascular events.
Decreasing glucose variability does not reduce cardiovascular event rates in T2DM after MI
Ch
apte
r 7
99
Introduction
Short-term variation in blood glucose (BG) levels is a daily challenge to patients with diabetes.
It confers a possible increased risk for hypoglycaemia, and it has been suggested that glucose
variability (GV) is related to cardiovascular risk 1-3 However, reanalysis of the Diabetes Control
and Complications Trial (DCCT) and DCCT/Epidemiology of Diabetes Interventions and
Complications (EDIC) dataset examining the predictive value of GV on microvascular and
neurological complications did not show an effect of GV independent from mean glucose
and HbA1c 4-6, and randomised controlled trials (RCTs) specifically targeting GV are lacking 7;8.
To assess the effect of intraday GV on cardiovascular outcomes we re-examined data
from Hyperglycaemia and Its Effect After Acute Myocardial Infarction on Cardiovascular
Outcomes in Patients With Type 2 Diabetes Mellitus study (HEART2D; clinical trial
registry number NCT00191282, clinicaltrials.gov) 9.
Research Design and Methods
The HEART2D study included 1,115 type 2 diabetic patients who had had a recent
myocardial infarction; patients well-controlled with diet or treated with intensive
insulin therapy were excluded. It was designed to investigate possible differences
between two insulin treatment strategies on time until first combined cardiovascular
event (a composite of cardiovascular death, nonfatal myocardial infarction, nonfatal
stroke, coronary revascularization, or hospitalization for acute coronary syndrome) 9.
Within 21 days after admission for acute myocardial infarction, patients were randomised
to one of two insulin treatment strategies: one targeted postprandial hyperglycaemia
with thrice-daily premeal insulin lispro (PRANDIAL; n = 557), and the other targeted
fasting/interprandial hyperglycaemia with once-daily insulin glargine or twice-daily
NPH (BASAL; n = 558). The study succeeded in achieving similar HbA1c levels in both
strategies, which allowed the authors to look at effects of targeting postprandial glucose
values independent from glycaemic control. There was no significant difference between
groups in time to first combined cardiovascular event (hazard ratio 0.98 [95% CI 0.8-
1.21]). Though the PRANDIAL group achieved significantly lower mean postprandial
blood glucose values, the between-group difference was less than expected. The trial was
halted early due to futility with a lower than expected overall number of cardiovascular
events. We evaluated the effect of glycaemic variability to help further the interpretation
of HEART2D results.
In the present analysis we calculated mean GV in both strategies from seven-point self-
measured BG profiles collected prior to study visits during the study period, obtained
100
over 24 hrs from breakfast to breakfast the next morning. This adds to the original
analysis since postprandial excursions contribute to GV, but GV encompasses more than
postprandial excursions alone. Since no gold standard for quantifying GV exists 7 we
calculated mean absolute glucose (MAG) change, mean amplitude of glycaemic excursions
(MAGE) and standard deviation (SD). MAG is the summated change in glucose per unit
of time (MAG = |ΔGlucose|/ΔTime) (Figure 1), which showed in the intensive care unit
a stronger association with mortality than SD 10. This is the first time MAG is used in a
diabetic population. MAGE is the average of all BG increases or decreases that are >1 SD
of all BG measures 11. Differences between regimens and those experiencing vs. those
not experiencing a cardiovascular event were assessed using a pattern mixed-model
repeated-measurement analysis. The model included strategy, baseline GV, randomisation
factors, and an additional factor for study duration (defined as ≤30, >30 and ≤42, and
>42 months) 9.
Results
The original HEART2D study analysis showed that HbA1c did not differ between groups
during the study (mean [SEM] PRANDIAL 7.7% [0.1], BASAL 7.8% [0.1], P = 0.4). We found
that the PRANDIAL strategy resulted in a significantly lower MAG compared with the
BASAL strategy (mean [SEM] PRANDIAL 0.40 [0.03] mmol/l/hr, BASAL 0.49 [0.02] mmol/
l/h, difference 0.09 [0.04] mmol/l/hr, P = 0.02). Also MAGE and SD were lower in the
PRANDIAL group, however not significantly (MAGE, PRANDIAL 3.14 [0.22] mmol/l, BASAL
3.32 [0.17] mmol/l, difference 0.18 [0.27] mmol/l, P = 0.50; SD, PRANDIAL 1.42 [0.09] mmol/l,
BASAL 1.58 [0.07] mmol/l, difference 0.16 [0.11] mmol/l, P = 0.15). Additionally, taking
the two randomization groups together, there was no difference in GV between those
experiencing vs. those not experiencing a cardiovascular event (mean [SEM] MAG 0.47
[0.03] mmol/l/hr vs. 0.44 [0.02] mmol/l/hr, P = 0.57; MAGE 3.35 [0.23] mmol/l vs. 3.16 [0.16]
mmol/l, P = 0.49; SD 1.56 [0.10] mmol/l vs. 1.48 [0.07] mmol/l, P = 0.52).
Conclusions
We found that in type 2 diabetes patients after acute myocardial infarction an insulin
regimen targeting postprandial hyperglycaemia lowered intraday GV compared with
a basal insulin strategy, with similar overall glycaemic control. However, GV decreases
shown in the PRANDIAL strategy did not translate into a reduction in cardiovascular
outcomes compared with treatment with BASAL strategy.
Decreasing glucose variability does not reduce cardiovascular event rates in T2DM after MI
Ch
apte
r 7
101
It might be possible that these negative results are explained by the patient group
studied, i.e., type 2 diabetic patients with advanced atherosclerosis. Another possibility
is that patients with diabetes, in contrast with critically ill patients without previously
diagnosed diabetes 10, are not affected by GV because of the ability of cells to adapt to
the harmful effects of changing ambient glucose.
A strong point of the present study is the assessment of GV by MAG. MAG takes GV
into account to its fullest extent as opposed to MAGE, which neglects glycaemic swings
smaller than 1 SD and assesses only the increases or decreases of the glucose profile 11.
Furthermore, MAG calculates glucose change over time, whereas SD does not take time
into account (Figure 1).
In conclusion, the results of the present analysis showed that targeting postprandial
glucose decreased intraday GV by 18% without a corresponding reduction in subsequent
secondary cardiovascular events at least in this population. Further studies looking at
different groups of patients are needed to investigate whether reducing GV will reduce
cardiovascular risk independently from HbA1c.
Figure 1 Two fictitious patients with identical mean glucose, SD, and mean amplitude of glycaemic excursions (MAGE), but different patterns of variability expressed by mean absolute glucose change (MAG).
102
References1. Borg R, Kuenen JC, Carstensen B, et al (2011) HbA(1c) and mean blood glucose show stronger associations
with cardiovascular disease risk factors than do postprandial glycaemia or glucose variability in persons with diabetes: the A1C-Derived Average Glucose (ADAG) study. Diabetologia 54: 69-72
2. Nalysnyk L, Hernandez-Medina M, Krishnarajah G (2010) Glycaemic variability and complications in patients with diabetes mellitus: evidence from a systematic review of the literature. Diabetes Obes Metab 12: 288-298
3. Monnier L, Colette C, Mas E, et al (2010) Regulation of oxidative stress by glycaemic control: evidence for an independent inhibitory effect of insulin therapy. Diabetologia 53: 562-571
4. Kilpatrick ES, Rigby AS, Atkin SL (2006) The effect of glucose variability on the risk of microvascular complications in type 1 diabetes. Diabetes Care 29: 1486-1490
5. Siegelaar SE, Kilpatrick ES, Rigby AS, Atkin SL, Hoekstra JB, DeVries JH (2009) Glucose variability does not contribute to the development of peripheral and autonomic neuropathy in type 1 diabetes: data from the DCCT. Diabetologia 52: 2229-2232
6. Kilpatrick ES, Rigby AS, Atkin SL (2009) Effect of glucose variability on the long-term risk of microvascular complications in type 1 diabetes. Diabetes Care 32: 1901-1903
7. Siegelaar SE, Holleman F, Hoekstra JB, DeVries JH (2010) Glucose variability; does it matter? Endocr Rev 31: 171-182
8. Siegelaar SE, Kulik W, van Lenthe H, Mukherjee R, Hoekstra JB, DeVries JH (2009) A randomized clinical trial comparing the effect of basal insulin and inhaled mealtime insulin on glucose variability and oxidative stress. Diabetes Obes Metab 11: 709-714
9. Raz I, Wilson PW, Strojek K, et al (2009) Effects of prandial versus fasting glycemia on cardiovascular outcomes in type 2 diabetes: the HEART2D trial. Diabetes Care 32: 381-386
10. Hermanides J, Vriesendorp TM, Bosman RJ, Zandstra DF, Hoekstra JB, DeVries JH (2010) Glucose variability is associated with intensive care unit mortality. Crit Care Med 38: 838-842
11. Service FJ, Molnar GD, Rosevear JW, Ackerman E, Gatewood LC, Taylor WF (1970) mean amplitude of glycemic excursions, a measure of diabetic instability. Diabetes 19: 644-655
Decreasing glucose variability does not reduce cardiovascular event rates in T2DM after MI
Ch
apte
r 7
103
Part II
Glucose control in critical illness
Chapter 8
Mean glucose during intensive care unit admission is related to mortality by a U-shaped curve in surgical and medical patients: a retrospective cohort study
Sarah E. Siegelaar, Jeroen Hermanides, Heleen M. Oudemans- van Straaten,
Peter H.J. van der Voort, Robert J. Bosman, Durk F. Zandstra and
J. Hans DeVries
Critical Care 2010; 14(6):R224
108
Abstract
Introduction: Lowering of hyperglycaemia in the intensive care unit (ICU) is
widely practised. We investigated in which way glucose regulation, defined as
mean glucose concentration during admission, is associated with ICU mortality
in a medical and a surgical cohort.
Methods: Retrospective database cohort study including patients admitted
between January 2004 and December 2007 in a 20-bed medical/surgical ICU in
a teaching hospital. Hyperglycaemia was treated using a computerised algorithm
targeting for glucose levels of 4.0-7.0 mmol/l. Five thousand eight hundred
twenty-eight patients were eligible for analyses, of whom 1,339 patients had a
medical and 4,489 had a surgical admission diagnosis.
Results: The cohorts were subdivided in quintiles of increasing mean glucose.
We examined the relation between these mean glucose strata and mortality. In
both cohorts we observed the highest mortality in the lowest and highest strata.
Logistic regression analysis adjusted for age, sex, Acute Physiology and Chronic
Health Evaluation II (APACHE II) score, admission duration and occurrence of
severe hypoglycaemia showed that in the medical cohort mean glucose levels
<6.7 mmol/l and >8.4 mmol/l and in the surgical cohort mean glucose levels
<7.0 mmol/l and >9.4 mmol/l were associated with significantly increased ICU
mortality (OR 2.4-3.0 and 4.9-6.2 respectively). Limitations of the study were its
retrospective design and possible incomplete correction for severity of disease.
Conclusions: Mean overall glucose during ICU admission is related to mortality
by a U-shaped curve in medical and surgical patients. In this cohort of patients a
“safe range” of mean glucose regulation might be defined approximately between
7.0 and 9.0 mmol/l.
Mean glucose during ICU admission is related to mortality by a U-shaped curve
Ch
apte
r 8
109
Introduction
Owing to inflammatory and neuro-endocrine derangements in critically ill patients,
stress hyperglycaemia associated with high hepatic glucose output and insulin resistance
is common in the intensive care unit (ICU) 1. This stress hyperglycaemia is associated
with poor outcome 2. Moreover, several studies report a deleterious effect of glycaemic
variability over and above mean glucose after correction for severity of disease 3-6.
In 2001, van den Berghe et al. 7 published the first randomised controlled trial (RCT)
comparing normalization of glycaemia by intensive insulin treatment (IIT) with
conventional glycaemic control in a surgical ICU (glucose target: 4.4 to 6.1 mmol/l vs.
10.0 to 11.1 mmol/l). The authors reported an impressive reduction in mortality with IIT.
The same group failed to reproduce these findings in the entire population of patients
in their medical ICU 8; however, mortality was lower in the predefined subgroup of
patients receiving IIT for more than 3 days. After the data were pooled from both RCT’s,
IIT seemed to be associated with a reduction in mortality 9. On the basis of these “Leuven
trials”, many hospitals decided to implement protocols and target normalization of
glucose levels to improve patient care.
Recently, after the publication of two inconclusive multicentre studies (the Volume
Substitution and Insulin Therapy in Severe Sepsis [VISEP] 10 and the GluControl 11;12 studies)
followed by the NICE-SUGAR (Normoglycaemia in Intensive Care Evaluation- Survival
Using Glucose Algorithm Regulation) trial 13, doubt was cast upon the benefits of tight
glycaemic control; the NICE-SUGAR trial investigators reported an absolute increase in
deaths at 90 days with IIT (glucose target: 4.5 to 6.0 mmol/l versus 8.0 to10.0 mmol/l). A
recently published meta-analysis including this latter trial showed that intensive insulin
therapy significantly increased the risk of hypoglycaemia and conferred no overall
mortality benefit among critically ill patients 14. The goal of this study is to report glucose
and mortality data from cohorts of patients with a medical and a surgical admission
diagnosis from a general ICU of a teaching hospital in The Netherlands.
Materials and methods
Cohorts, setting, and data collectionWe collected information about patients admitted between January 2004 and December
2007 in a 20-bed medical/surgical ICU in a teaching hospital (Onze Lieve Vrouwe Gasthuis
[OLVG], Amsterdam, the Netherlands) (the OLVG cohort). All data was anonymous and
collected retrospectively, so no ethical approval was necessary. On average, one nurse
took care of two patients, depending on the severity of disease. All beds were equipped
110
with a clinical information system (MetaVision; iMDsoft, Tel Aviv, Israel) from which
all clinical and laboratory data were extracted. The glucose regulation algorithm was
implemented successfully in 2001 15, targeting for glucose values of between 4.0 and 7.0
mmol/l. The glucose protocol was started for every patient at the time of arrival at the
ICU. Insulin infusion was started when admission blood glucose exceeded 7.0 mmol/l.
When admission glucose was lower than 7.0 mmol/l, blood glucose was further measured
every 2 hrs and insulin was started when necessary (that is, when blood glucose exceeded
7.0 mmol/l). The nursing staff was instructed to use a dynamic computerised algorithm
to adjust the insulin infusion rate, depending on the current glucose value and the rate
of glucose change (based on the previous five measurements). The software also provided
the time the next glucose measurement was due, which could vary from 15 min up to 4
hrs. Routinely, enteral feeding was started within 24 hrs after admission, aiming at 1,500
kcal per 24 hrs, and subsequently adjusted to the patient’s requirements, except for the
uncomplicated cardiac surgery patients who do not receive enteral feeding if extubated
within 24 hrs. A duodenal feeding tube was inserted in case of persistent gastric retention.
The tight glucose algorithm was deactivated when patients resumed normal eating.
We excluded readmissions, patients with a withholding care policy, and patients with
only one glucose value measured during admission. From the clinical information system,
we collected demographic variables, mortality rates in the ICU, and glucose values. As
severity of disease measure, we used the Acute Physiology and Chronic Health Evaluation II
(APACHE II) score 16. Informed consent was not required according to Dutch Ethical Review
Board regulations, because a retrospective analysis of anonymous data was performed.
Glucose measuresFor each patient, we calculated the mean overall glucose during admission from all
glucose values measured during admission and the mean morning glucose from the first
value available between 5:00 and 7:00 hrs per patient per day. Glucose values mentioned
in this paper stand for mean overall glucose unless stated otherwise. We calculated the
standard deviation (SD) and the mean absolute glucose (MAG) change 6 per patient as
markers of glycaemic variability. Glucose was obtained from arterial blood samples by
means of a handheld glucose measurement device (AccuChek; Roche/Hitachi, Basel,
Switzerland). Results were automatically stored in the clinical information system.
Data interpretationThe cohort characteristics are presented as mean (SD) or as median and interquartile range
(IQR), depending on the distribution of the data. The mean glucose values and SD’s were
divided into five strata with equal numbers of patients per group. For each stratum, the
ICU mortality was calculated. Subsequently, we performed a logistic regression analysis
to calculate the odds ratio (OR) with 95% confidence intervals (CI) for ICU mortality per
Mean glucose during ICU admission is related to mortality by a U-shaped curve
Ch
apte
r 8
111
glucose stratum. The stratum with the lowest mortality incidence was used as a reference.
In this model we adjusted for age, sex, severity of disease (APACHE II score), occurrence of
severe hypoglycaemia (≤2.2 mmol/l), and admission duration (that is, ≤ or > 24 hrs). The last
adjustment was done because glucose values are higher and have a wider range in the first 24
hrs of admission, biasing the patients with longer admission times and corresponding lower
mean glucose values. In a second model, adjustment for occurrence of mild hypoglycaemia
(≤4.7 mmol/l), which is also independently associated with mortality 17, was made.
Results
In total, 5,828 patients were eligible for analyses of the mean glucose for the OLVG
population after excluding 656 readmissions, 86 patients with a withholding care policy,
and 160 patients with only one glucose value measured. This cohort consisted of 1,339
patients with a medical admission diagnosis (the “medical” population) and 4,489 patients
with a surgical admission diagnosis (the “surgical” population). In the medical cohort, a
median (IQR) of 34 (15-65) glucose values per patient were collected and in the surgical
cohort a median (IQR) of 10 (5-14) values. The median (IQR) admission duration was 64 (30-
129) hrs in the medical and 22 (18-28) hrs in the surgical cohort.
Mean glucoseThe overall mean (SD) glucose values of the medical and surgical populations were
7.9 (2.7) and 8.1 (1.6) mmol/l (Table 1). The mean glucose values of the first 24 hrs of
admission were higher and had a wider range than did the mean glucose values after
24 hrs (medical: mean [SD] 8.4 [3.3] mmol/l, range 3.7-40.2 mmol/l and 7.0 [1.4] mmol/l,
range 3.2-31.1 mmol/l; surgical: mean [SD] 8.3 [1.9] mmol/l, range 0.6-27.5 mmol/l and 7.6
[1.7] mmol/l, range 3.2-15.7 mmol/l). The mean morning glucose was 7.4 [2.6] mmol/l in
the medical population and 7.7 [2.3] mmol/l in the surgical population. After dividing
the mean glucose of both populations into five equally sized strata, the lowest mean
glucose stratum ranged from 6.7 mmol/l and lower in the medical cohort and from 7.0
mmol/l and lower in the surgical cohort. The highest stratum ranged 8.5 mmol/l and
higher in the medical cohort and 9.5 mmol/l and higher in the surgical cohort. Mean
glucose ranges per stratum and corresponding mortality rates per cohort are displayed
in Figure 1. This results in a U-shaped curve relationship between mean glucose and
mortality in both cohorts, with high ICU mortality in the lowest and highest glucose
strata (medical: 26.9% and 35.6%; surgical: 3.6% and 1.4%). Logistic regression analysis
showed that in both populations mean glucose values in the lowest and highest strata
were associated with a significant higher OR for ICU mortality compared to the stratum
with the lowest mortality (Figure 2). This results in “safe ranges” of 6.7 to 8.5 mmol/l
in the medical, and 7.0 to 9.5 mmol/l in the surgical cohort. The non-linear U-shaped
112
Tab
le 1
Ch
arac
teri
stic
s o
f th
e st
ud
ied
co
ho
rts,
div
ided
by
mea
n g
luco
se r
ange
s
Med
ical
pop
ula
tion
Su
rgic
al p
opu
lati
on
Tota
ln
= 1
,339
≤ 6.
6 m
mol
/ln
= 2
68“s
afe
ran
ge”
n =
804
≥ 8.
5 m
mol
/ln
= 2
67To
tal
n =
4,4
89≤
6.9
mm
ol/l
n =
898
“saf
e ra
nge
”n
= 2
,694
≥ 9.
5 m
mol
/ln
= 8
97
Age
, yea
rs (m
ean
± S
D)
61.8
± 1
6.9
59.0
± 1
8.4
62.5
± 1
6.2
62.4
± 1
7.0
66.0
± 1
2.0
66.8
± 1
2.5
65.4
± 1
2.1
67.2
± 1
1.3
Gen
der,
fem
ale
(%)
38.2
37.3
37.7
40.4
33.2
36.6
32.0
33.4
APA
CH
E II
sco
re (m
ean
± S
D)
24.6
± 8
.824
.8 ±
9.1
24.1
± 8
.125
.8 ±
10.
215
.1 ±
4.6
16.3
± 5
.214
.8 ±
4.5
14.7
± 4
.2
Dia
bete
s M
elli
tus
(%)
0.6
0.4
0.5
1.1
15.4
23.7
16.4
4.1
Die
d IC
U (%
)20
.926
.914
.135
.61.
63.
61.
01.
4
Die
d h
ospi
tal (
%)
31.3
35.4
26.6
41.2
4.3
7.5
3.9
2.7
Mor
nin
g gl
uco
se, m
mol
/l (m
ean
± S
D)
7.4
± 2.
65.
9 ±
1.0
7.1
± 1.
210
.3 ±
4.5
7.7
± 2.
35.
8 ±
1.2
7.3
± 1.
710
.6 ±
1.9
Ove
rall
glu
cose
, mm
ol/l
(mea
n ±
SD
)7.
9 ±
2.7
6.0
± 0.
67.
3 ±
0.5
11.6
± 4
.18.
1 ±
1.6
6.4
± 0.
57.
9 ±
0.7
10.7
± 1
.1
Hyp
ogly
caem
ia in
cide
nce
(%)
9.9
18.7
8.8
4.5
1.8
4.8
1.3
0.1
SD, m
mol
/l (m
edia
n [I
QR
])2.
0 [1
.5-2
.9]
1.6
[1.2
-1.9
]2.
0 [1
.6-2
.6]
3.8
[2.7
-5.4
]1.
8 [1
.3-2
.3]
1.6
[1.3
-2.0
]1.
8 [1
.4-2
.4]
1.9
[1.4
-2.6
]
MA
G, m
mol
/l/h
r (m
edia
n [I
QR
])0.
8 [0
.5-1
.1]
0.5
[0.3
-0.8
]0.
8 [0
.6-1
.0]
1.4
[0.9
-2.0
]0.
6 [0
.4-0
.8]
0.5
[0.4
-0.7
]0.
6 [0
.4-0
.9]
0.5
[0.3
-0.7
]
Cal
oric
inta
ke p
er 2
4 h
rs (m
ean
± S
D)
1103
.0 ±
758
.411
59.3
± 1
108.
611
07.1
± 5
07.2
1033
.6 ±
944
.531
5.0
± 39
2.3
427.
7 ±
466.
632
2.8
± 38
7.5
181.
5 ±
268.
9
Use
of i
nsu
lin
(%)
88.5
79.5
93.3
82.8
64.0
93.1
71.8
11.6
Insu
lin
dos
e, IU
/hou
r (m
edia
n [I
QR
])1.
4 [0
.8-2
.4]
0.6
[0.4
-1.0
]1.
4 [0
.9-2
.1]
3.4
[2.0
-6.2
]1.
2 [0
.7-1
.9]
1.0
[0.7
-1.5
]1.
3 [0
.8-2
.0]
1.5
[0.7
-3.2
]
Use
of v
asop
ress
or d
rugs
(%)
86.0
19.4
11.8
15.4
94.8
94.1
94.2
97.0
Use
of c
orti
coid
s (%
)92
.591
.094
.886
.999
.199
.099
.199
.1
Mec
han
ical
ven
tila
tion
(%)
81.6
81.7
85.0
71.2
97.9
97.3
97.9
98.6
CV
VH
(%)
16.7
20.1
17.4
11.2
2.6
7.0
1.8
0.8
The
“saf
e ra
nge
” re
fers
to
the
mea
n g
luco
se l
evel
s as
soci
ated
wit
h t
he
low
est
mor
tali
ty r
ates
: 6.7
to
8.4
mm
ol/l
in
th
e m
edic
al a
nd
7.0
to
9.4
mm
ol/l
in
th
e su
rgic
al
coh
ort.
Hyp
ogly
caem
ia w
as d
efin
ed a
s at
lea
st o
ne
glu
cose
val
ue
of n
ot m
ore
than
2.2
mm
ol/l
. APA
CH
E II
, Acu
te P
hys
iolo
gy a
nd
Ch
ron
ic H
ealt
h E
valu
atio
n I
I; C
VV
H,
con
tin
uou
s ve
no-
ven
ous
hae
mofi
ltra
tion
; IC
U, I
nte
nsi
ve C
are
Un
it; M
AG
, mea
n a
bsol
ute
glu
cose
ch
ange
; SD
, sta
nd
ard
dev
iati
on
Mean glucose during ICU admission is related to mortality by a U-shaped curve
Ch
apte
r 8
113
relationship between mean glucose and ICU mortality was supported by significance
of the quadratic transformation of the mean glucose levels in this logistic regression
model (P<0.001). The characteristics of our populations, also subdivided in groups with
low, “safe range” and high glucose values, are displayed in Tables 1 and 2.
Figure 1 ICU mortality (y-axis) per mean glucose stratum (x-axis) (A) Medical population. (B) Surgical population.
Other glycaemic measuresOverall, 9.9% and 1.8% of the medical and surgical patients, respectively, sustained at
least one hypoglycaemic episode, defined as a glucose value of not more than 2.2 mmol/l,
during ICU admission. Seventeen point five percent of all deaths during ICU admission
concerned patients who had experienced severe hypoglycaemia (both groups). Twenty-
eight percent of the patients who were in the lowest mean glucose strata and who died in
the ICU experienced hypoglycaemia, and 72% did not. The incidence of severe and mild
(≤4.7 mmol/l) hypoglycaemia in the different mean glucose strata is reported in Figure 3.
When we adjusted the logistic regression model for occurrence of mild hypoglycaemia
with a cutoff value of 4.7 mmol/l, which is also independently associated with mortality 17,
the OR (CI) for ICU mortality in the lowest glucose stratum remained significant (medical:
2.6 [1.6-4.4], P <0.001; surgical: 4.9 [1.1-22.1], P = 0.04).
In the medical cohort, glucose variability, both when expressed as the median of
individual SD’s and MAG changes 6, linearly increased with increasing glucose strata
(SD median [IQR] 1.6 [1.2-1.9] to 3.8 [2.7-5.4] mmol/l, P for trend <0.001; MAG 0.5 [0.3-0.8]
to 1.4 [0.9-2.0] mmol/l/h, P for trend 0.007). However, in the surgical cohort, no consistent
trend in glucose variability across the glucose strata was seen (SD median [IQR] 1.8 [1.3-
2.3] mmol/l; MAG 0.6 [0.4-0.8] mmol/l/hr). Adjusting the logistic regression model for
variability did not change the above-described relationship between mean glucose and
mortality (data not shown).
114
Figure 2 Odds ratio (OR) for mortality (y-axis) per glucose stratum (x-axis) with the highest OR in the lowest and highest strata (A) Medical population. (B) Surgical population. Logistic regression model was adjusted for age, sex, APACHE II (Acute Physiology and Chronic Health Evaluation II) score, admission duration (≤ and > 24 hrs), and occurrence of severe hypoglycaemia. *P <0.05, **P <0.001. CI, confidence interval
Figure 3 Hypoglycaemia incidence (y-axis) per mean glucose stratum (x-axis) (A) Medical population. (B) Surgical population. The y-axis represents the percentage of patients experiencing at least one severe (≤2.2 mmol/l, left bars) and mild (≤4.7 mmol/l, right bars) hypoglycaemic event.
Discussion
The salient finding of this investigation is that in this mixed medical and surgical cohort
of critically ill patients, mean glucose values of between approximately 7.0 and 9.0 mmol/l
during ICU stay were associated with the lowest OR for ICU mortality, while mean values
of below 7.0 and greater than 9.0 mmol/l confer significantly higher OR’s. These results
were attained while using a dynamic glucose algorithm that targeted for glucose values of
between 4.0 and 7.0 mmol/l. The finding that hyperglycaemia is associated with increased
mortality is in accordance with published literature 2;18;19. Also, the U-shaped curve we
found, with increased mortality in the lower and upper parts, is described earlier in
Mean glucose during ICU admission is related to mortality by a U-shaped curve
Ch
apte
r 8
115
patients with myocardial infarction during admission 20-22, more generally in patients
with type 2 diabetes mellitus 23, and in the ICU setting 24-26, corroborating this finding.
The optimum glucose levels in the ICU setting reported previously are somewhat lower
than we found. This is possibly due to differences in inclusion criteria or uncertainty
about the practice of tight glycaemic control 26, lack of regression analysis between the
strata 25, or a different method to assess mean glucose 24. Another difference between our
and other ICU cohorts is the high percentage of patients admitted after cardiac arrest
(Table 2), a population with a high mortality rate. Also, the percentage of patients with
diabetes in our cohort might be underestimated since we scored diabetes only when the
patient used anti-hyperglycaemic drugs. However, how these factors might influence the
position of the U-curve in relation to the x-axis is not known.
Hypoglycaemia is associated with increased risk of ICU and hospital mortality 17;27-29.
In our population, the incidence of hypoglycaemia was highest in the lowest mean
glucose cohorts in which mortality was higher as well. In addition, a significant
percentage of the patients who died had experienced a hypoglycaemic episode. However,
hypoglycaemia can account only partially for the high mortality rate in the lowest mean
overall glucose stratum since 72.0% of the non-survivors did not experience severe
hypoglycaemia. Also, when the logistic regression model was adjusted for occurrence
of severe or mild hypoglycaemia, the OR for mortality remained significantly higher for
those patients with a mean glucose in the lowest quintile. However, it might be possible
that some hypoglycaemic episodes were not recorded due to intermittent sampling,
or were underestimated because of the AccuChek point-of-care meter used for glucose
measurements, the results of which tend to be higher than those obtained from the
laboratory 30;31. Therefore, the contribution of hypoglycaemia to ICU death could be
underestimated and needs further research using continuous glucose measurement. An
alternative explanation for increased mortality at lower glucose values might be that
tissues with insulin-independent glucose uptake may suffer from insufficient glucose
availability at lower concentrations. In our cohort, glucose variability increased with
increasing glucose strata in the medical cohort. In the surgical cohort, no consistent
relationship was found. Since glucose variability is associated with mortality 6, it is
unlikely that this contributes to the higher mortality in the lower glucose strata.
In the NICE-SUGAR study, the mean glucose of the IIT group (6.4 mmol/l) falls into the
stratum with increased mortality compared to the conventional group (8.0 mmol/),
which lies in the safe range of both OLVG populations (Figure 1) 13. Thus, the findings
of the NICE-SUGAR trial are in accordance with the mortality data from our cohort.
This is in contrast with the data of both Leuven studies. The means of the IIT groups of
both the Leuven studies (6.1 mmol/l in the medical population 8 and 5.7 mmol/l in the
surgical population 7) fall into the lowest mean glucose stratum in the corresponding
116
Tab
le 2
Per
cen
tage
of
pat
ien
ts p
er A
PAC
HE
II
adm
issi
on
cat
ego
ry
Med
ical
po
pu
lati
on
Surg
ical
po
pu
lati
on
Tota
ln
= 1
,339
≤ 6.
6 m
mo
l/l
n =
268
“saf
e ra
nge
”n
= 8
04≥
8.5
mm
ol/
ln
= 2
67To
tal
n =
4,4
89≤
6.9
mm
ol/
ln
= 8
98“s
afe
ran
ge”
n =
2,6
94≥
9.5
mm
ol/
ln
= 8
97
Car
dio
vasc
ula
r18
.011
.619
.918
.788
.281
.088
.395
.1
Sep
sis
16.5
22.8
16.0
11.6
1.2
2.8
1.0
0.1
Aft
er c
ard
iac
arre
st21
.611
.921
.531
.50.
20.
60.
10.
1
Gas
troi
nte
stin
al4.
34.
14.
24.
95.
38.
75.
02.
8
Hae
mat
olog
ical
0.6
0.7
0.7
00.
20.
40.
10.
1
Ren
al1.
91.
51.
05.
20.
30.
60.
20.
1
Met
abol
ic3.
63.
02.
76.
70.
20.
10.
20.
1
Neu
rolo
gica
l11
.518
.310
.38.
20.
91.
11.
00.
3
Res
pir
ator
y22
.026
.123
.513
.13.
64.
84.
01.
2
The
“saf
e ra
nge
” re
fers
to
the
mea
n g
luco
se le
vels
ass
ocia
ted
wit
h t
he
low
est
mor
tali
ty r
ates
: 6.7
to
8.4
mm
ol/l
in t
he
med
ical
, an
d 7
.0 t
o 9.
4 m
mol
/l in
th
e su
rgic
al c
ohor
t.
Mean glucose during ICU admission is related to mortality by a U-shaped curve
Ch
apte
r 8
117
OLVG cohorts, in which mortality is highest. The means of the conventional groups in
the Leuven studies (8.5 mmol/l in the medical as well as in the surgical population 7;8)
lie in the safe ranges of both OLVG populations (Figure 1).
A possible explanation for the low mortality of the Leuven IIT group might be the way
of feeding. In a recent paper, Marik and Preiser 32 suggested that the use of intravenous
calories could explain differences between populations treated with IIT, with a
positive effect of IIT in patients who receive most of their calories intravenously. In our
population, as opposed to the Leuven studies, only 0.7% of carbohydrates were given
parenterally. In populations predominantly fed parenterally, the relationship between
mean overall glucose and mortality might be different. Also, glycaemic swings are a
known risk factor of ICU death and might contribute to differences in mortality rate 4;5.
However, it is unlikely that differences in glucose variability explain the higher mortality
in our cohort compared with the Leuven IIT group as the medians [IQR] of the individual
median SD’s are roughly comparable (Leuven medical 1.99 [1.57-2.66] mmol/l 33 and OLVG
medical 2.03 [1.54-2.86] mmol/l). In addition, other explanations have been proposed to
explain the diverging outcomes of Leuven and NICE-SUGAR 34.
The mean glucose of the OLVG population (medical: 7.9 mmol/l; surgical: 8.1 mmol/l)
was higher than the target range, which was between 4.0 and 7.0 mmol/l. Other studies
of IIT also did not reach their target range, illustrating the difficult implementation of
this therapy 10;12;13. The high percentage of corticosteroid treatment in our population
might have contributed (Table 1). Also, the relatively short ICU duration of stay in the
predominantly surgical population of the OLVG explains that mean glucose is slightly
higher than the target (median ICU stay was 22 hrs in our cohort compared to 3 days in
the Leuven cohort and 4.2 days “on algorithm” in the NICE SUGAR study) because of the
time needed to reach target. Glucose values were indeed higher and had a wider range
in the first 24 hrs of admission. Furthermore, our patients were treated in a normal-care
setting without the extra stimuli of a trial setting to achieve the target. It should be noted
that mean glucose does not equal time in target range, since the protocol requires more
frequent sampling when not in target, thus falsely inflating the mean.
In our logistic regression model, we adjusted for severity of disease and admission
duration less or more than 24 hrs, since both high and low glucose levels could be a
manifestation, rather than a cause, of severe disease. Glucose values are higher and have
a wider range in the first 24 hrs of admission, biasing the patients with longer admission
times and corresponding lower mean glucose values. A limitation of our correction for
severity of disease is the use of the APACHE II score, because the use of APACHE II to
predict mortality is not validated for cardiac surgery patients. However, this adjustment
is the best available method 35.
118
Conclusions
In our mixed cohort of surgical and medical patients, the mean glucose during ICU
stay was related to mortality by a U-shaped curve; a “safe range” for mean glucose
can be defined as between approximately 7.0 and 9.0 mmol/l, while both higher and
lower mean values are associated with higher mortality. This finding applied to the
surgical as well as the medical patients. Hypoglycaemia seems to only partially explain
the high mortality rate in the lowest mean glucose quintile, and glucose variability
does not. Second, comparison of the combined Leuven, NICE-SUGAR, and our cohorts
demonstrates that the increased mortality in the IIT group of NICE-SUGAR is in line with
our U-shaped curve but that the low mortality in the intensively treated Leuven group
is not. The percentage of calories given parenterally may influence the relationship
between mean glucose and mortality. We await further studies, but according to these
findings, we recommend treating hyperglycaemia at the ICU in a moderately intensive
way in both medical and surgical patients, targeting for mean glucose values of between
approximately 7.0 and 9.0 mmol/l and avoiding hypoglycaemia. This “safe range” should
be studied prospectively in randomised clinical trials.
Key Messages
- During ICU admission, mean glucose relates to mortality by a U-shaped curve.
- A mean glucose range of 7.0 to 9.0 mmol/l is associated with the lowest mortality in
our cohort.
- Occurrence of hypoglycaemia does not fully explain the high mortality in the lower
glucose strata.
Mean glucose during ICU admission is related to mortality by a U-shaped curve
Ch
apte
r 8
119
References1. Dungan KM, Braithwaite SS, Preiser JC (2009) Stress hyperglycaemia. Lancet 373: 1798-18072. Krinsley JS (2003) Association between hyperglycemia and increased hospital mortality in a heterogeneous
population of critically ill patients. Mayo Clin Proc 78: 1471-14783. Dossett LA, Cao H, Mowery NT, Dortch MJ, Morris J, May AK (2008) Blood Glucose Variability Is Associated
with Mortality in the Surgical Intensive Care Unit. American Surgeon 74: 679-6854. Egi M, Bellomo R, Stachowski E, French CJ, Hart G (2006) Variability of blood glucose concentration and
short-term mortality in critically ill patients. Anesthesiology 105: 244-2525. Krinsley JS (2008) Glycemic variability: A strong independent predictor of mortality in critical ill patients.
Crit Care Med 36: 3008-30136. Hermanides J, Vriesendorp TM, Bosman RJ, Zandstra DF, Hoekstra JB, DeVries JH (2010) Glucose variability
is associated with intensive care unit mortality. Crit Care Med 38: 838-8427. Van den Berghe G, Wouters P, Weekers F, et al (2001) Intensive Insulin Therapy in Critically Ill Patients. N
Engl J Med 345: 1359-13678. Van den Berghe G, Wilmer A, Hermans G, et al (2006) Intensive Insulin Therapy in the Medical ICU. N Engl
J Med 354: 449-4619. Van den Berghe G, Wilmer A, Milants I, et al (2006) Intensive Insulin Therapy in Mixed Medical/Surgical
Intensive Care Units. Diabetes 55: 3151-315910. Brunkhorst FM, Engel C, Bloos F, et al (2008) Intensive Insulin Therapy and Pentastarch Resuscitation in
Severe Sepsis. N Engl J Med 358: 125-13911. Devos P, Preiser JC, Melot C (2007) Impact of tight glucose control by intensive insulin therapy on ICU
mortality and the rate of hypoglycemia: final results of the Glucontrol study. Intensive Care Medicine 33: Suppl 2: S189
12. Preiser JC, Devos P, Ruiz-Santana S, et al (2009) A prospective randomised multi-centre controlled trial on tight glucose control by intensive insulin therapy in adult intensive care units: the Glucontrol study. Intensive Care Med 35: 1738-1748
13. The NICE-SUGAR Study Investigators (2009) Intensive versus Conventional Glucose Control in Critically Ill Patients. N Engl J Med 360: 1283-1297
14. Griesdale DEG, de Souza RJ, van Dam RM, et al (2009) Intensive insulin therapy and mortality among critically ill patients: a meta-analysis including NICE-SUGAR study data. CMAJ 180: 821-827
15. Rood E, Bosman RJ, van der Spoel JI, Taylor P, Zandstra DF (2005) Use of a computerized guideline for glucose regulation in the Intensive Care Unit improved both guideline adherence and glucose regulation. J Am Med Inform Assoc 12: 172-180
16. Knaus WA, Draper EA, Wagner DP, Zimmerman JE (1985) APACHE II: A severity of disease classification system. Crit Care Med 13: 818-829
17. Hermanides J, Bosman RJ, Vriesendorp TM, et al (2010) Hypoglycaemia is related with intensive care unit mortality. Crit Care Med 38: 1430-1434
18. Umpierrez GE, Isaacs SD, Bazargan N, You X, Thaler LM, Kitabchi AE (2002) Hyperglycemia: An Independent Marker of In-Hospital Mortality in Patients with Undiagnosed Diabetes. J Clin Endocrinol Metab 87: 978-982
19. Krinsley JS (2006) Glycemic control, diabetic status, and mortality in a heterogeneous population of critically ill patients before and during the era of intensive glycemic management: six and one-half years experience at a university-affiliated community hospital. Semin Thorac Cardiovasc Surg 18: 317-325
20. Kosiborod M, Inzucchi SE, Krumholz HM, et al (2008) Glucometrics in Patients Hospitalized With Acute Myocardial Infarction: Defining the Optimal Outcomes-Based Measure of Risk. Circulation 117: 1018-1027
21. Pinto DS, Skolnick AH, Kirtane AJ, et al (2005) U-Shaped Relationship of Blood Glucose With Adverse Outcomes Among Patients With ST-Segment Elevation Myocardial Infarction. J Am Coll Cardiol 46: 178-180
22. Pinto DS, Kirtane AJ, Pride YB, et al (2008) Association of Blood Glucose With Angiographic and Clinical Outcomes Among Patients With ST-Segment Elevation Myocardial Infarction (from the CLARITY-TIMI-28 Study). Am J Cardiol 101: 303-307
23. Currie CJ, Peters JR, Tynan A, et al (2010) Survival as a function of HbA1c in people with type 2 diabetes: a retrospective cohort study. Lancet 375: 481-489
24. Bagshaw SM, Egi M, George C, Bellomo R (2009) Early blood glucose control and mortality in critically ill patients in Australia. Crit Care Med 37: 463-470
25. Egi M, Bellomo R, Stachowski E, et al (2008) Blood glucose concentration and outcome of critical illness: the impact of diabetes. Crit Care Med 36: 2249-2255
26. Falciglia M, Freyberg RW, Almenoff PL, D’Alessio DA, Render ML (2009) Hyperglycemia-related mortality in
120
critically ill patients varies with admission diagnosis. Crit Care Med 37: 3001-300927. Bagshaw SM, Bellomo R, Jacka M, et al (2009) The impact of early hypoglycemia and blood glucose variability
on outcome in critical illness. Critical Care 13: R9128. Krinsley JS, Grover A (2007) Severe hypoglycemia in critically ill patients: Risk factors and outcomes. Crit
Care Med 35: 2262-226729. Vriesendorp TM, DeVries JH, van Santen S, et al (2006) Evaluation of short-term consequences of hypoglycemia
in an intensive care unit. Crit Care Med 34: 2714-271830. Hoedemaekers CWE, Klein Gunnewiek JMT, Prinsen MA, Willems JL, Van der Hoeven JG (2008) Accuracy of
bedside glucose measurement from three glucometers in critically ill patients *. Crit Care Med 36:3062-3066 31. Karon BS, Gandhi GY, Nuttall GA, et al (2007) Accuracy of Roche Accu-Chek Inform Whole Blood Capillary,
Arterial, and Venous Glucose Values in Patients Receiving Intensive Intravenous Insulin Therapy After Cardiac Surgery. Am J Clin Pathol 127: 919-926
32. Marik PE, Preiser JC (2010) Toward understanding tight glycemic control in the ICU: a systematic review and metaanalysis. Chest 137: 544-551
33. Meyfroidt G, Keenan DM, Wang X, Wouters PJ, Veldhuis JD, Van den Berghe G (2010) Dynamic characteristics of blood glucose time series during the course of critical illness: effects of intensive insulin therapy and relative association with mortality. Crit Care Med 38: 1021-1029
34. Van den Berghe G, Schetz M, Vlasselaers D, et al (2009) Intensive insulin therapy in critically ill patients: NICE-SUGAR or Leuven blood glucose target? J Clin Endocrinol Metab 94: 3163-3170
35. Kramer AA, Zimmerman JE (2008) Predicting Outcomes for Cardiac Surgery Patients After Intensive Care Unit Admission. Semin Cardiothorac Vasc Anesth 12: 175-183
Mean glucose during ICU admission is related to mortality by a U-shaped curve
Ch
apte
r 8
121
Chapter 9
Accuracy and reliability of continuous glucose monitoring in the intensive care unit: a head-to-head comparison of two subcutaneous glucose sensors in cardiac surgery patients
Sarah E. Siegelaar, Temo Barwari, Jeroen Hermanides,
Peter H.J. van der Voort and J. Hans DeVries
Published in abbreviated form, Diabetes Care 2011; 34(3): e31
124
Abstract
Objective: To investigate accuracy and reliability of two different continuous
glucose monitoring (CGM) devices in patients who underwent cardiac surgery.
Methods: We performed a prospective, observational, investigator-initiated
study in a 20-bed intensive care unit (ICU) of a teaching hospital. We studied 60
consecutive patients who underwent cardiac surgery. Two CGM devices (Guardian
Real-Time, Medtronic Minimed; FreeStyle Navigator, Abbott Diabetes Care) were
placed subcutaneously in the abdominal wall before surgery. Both devices
were calibrated simultaneously upon arrival at the ICU and further according
to manufacturers’ instructions. An arterial reference blood glucose value was
measured every two hours. Relative absolute deviation (RAD) between reference
and sensor glucose was calculated in six 5-minute intervals from the reference
glucose, to assess a possible delay.
Results: Of the 1,017 reference glucose values measured 77.7% could be paired
with a Guardian and 91.8% with a Navigator glucose value, missing values
indicating technical problems with the devices: unintentional signal loss (both
systems) or an interruption of real-time representation of glucose values after
delayed recalibration (Guardian). Median [IQR] RAD was significantly smaller for
Navigator than for Guardian measurements at the first and second interval (11%
[8-16] and 10% [8-16] compared to 14% [11-18] and 14% [11-17], P = 0.05 and 0.001).
The delay was estimated to be 5-9 minutes for the Navigator and 15-19 minutes
for the Guardian.
Conclusions: FreeStyle Navigator performed better regarding accuracy and
reliability than the Guardian Real-Time in cardiac surgery patients at the ICU.
Use of this device seems feasible in these patients.
Accuracy and reliability of CGM in the ICU
Ch
apte
r 9
125
Introduction
Occurrence of hyperglycaemia in the intensive care unit (ICU) is common, also in patients
without a known history of diabetes. Severe illness causes hormonal changes resulting
in hyperglycaemia through increased gluconeogenesis in the liver and increased insulin
resistance. This transient so called stress-hyperglycaemia is associated with increased
mortality 1. Several trials assessed the effect of intensive insulin therapy on outcome in
this patient group with conflicting outcomes 2;3. However, glycaemic control remains a
widespread practice, although the target range is unclear. Besides hyperglycaemia also
hypoglycaemia, as a consequence of intensive insulin therapy or due to severe illness,
and glucose variability are independently associated with mortality 4;5.
At present, intermittent manual blood sampling has to be performed to achieve glycaemic
control. This method is time consuming, certainly when the patients’ glucose levels
fluctuate. Moreover, no information is available for the period in-between measurements
with perhaps unnoticed hypoglycaemic episodes. Continuous glucose monitoring (CGM)
could therefore be of value in achieving glycaemic control, providing real-time glucose
values as well as an alarm function to alert for glucose values outside a predefined range,
and information on rapid increases or decreases in glucose levels 6.
Promising as CGM in the ICU may be, the accuracy and reliability of these devices is
uncertain in critically ill patients 7-9. Different studies report “acceptable” differences
between sensor and reference glucose values, but it can be debated how large an
acceptable deviation in the ICU may be, also because it is known from outpatient
data that accuracy is worse in the hypoglycaemic range 10. CGM measurements reflect
interstitial rather than plasma glucose levels, so microcirculatory changes seen in the
critically ill might influence CGM function. However, in patients who underwent cardiac
surgery a good correlation between arterial and interstitial glucose was found using an
experimental micro dialysis system 11, suggesting that this patient group lends itself for CGM.
In this study we investigated the accuracy and reliability of two different CGM devices,
the Guardian® Real-Time (Medtronic Minimed, Northridge, CA) and FreeStyle Navigator®
(Abbott Diabetes Care, Alameda, CA), postoperatively in cardiac surgery patients in an
investigator-initiated trial.
Methods
PatientsWe performed a prospective observational study in a 20-bed medical/surgical ICU in
126
the Onze Lieve Vrouwe Gasthuis (OLVG; Amsterdam, the Netherlands) to obtain glucose
monitoring data from the two devices. The study was approved by the Institutional Ethical
Review Board according to the declarations of Helsinki. We included subsequent patients
above the age of 18 who were planned to undergo elective cardiac surgery; coronary artery
bypass grafting (CABG) and/or valve surgery. We excluded patients with an abdominal
condition which would prohibit sensor insertion. Eligible patients received an information
letter at least one week before hospital admission. Before the planned surgery patients
were asked to give written informed consent after oral explanation of the study. During
ICU admittance Acute Physiology and Chronic Health Evaluation (APACHE) IV predicted
mortality score was calculated for the first 24 hrs of admission and Sequential Organ
Failure (SOFA) score was obtained daily. Also the European System for Cardiac Operative
Risk Evaluation (euroSCORE) score, a method of calculating predicted operative mortality
risk for patients undergoing cardiac surgery, was recorded for every patient.
Glucose monitoringTwo needle-type sensors, Guardian® Real-Time (Guardian; Medtronic Minimed,
Northridge, CA) and FreeStyle Navigator® (Navigator; Abbott Diabetes Care, Alameda,
CA), were inserted in the abdominal wall on either side of the umbilicus and calibrated
before surgery to allow stabilization of the signal. Upon arrival at the ICU after surgery,
the device’s internal clock was matched with the bedside computer and both devices
were calibrated simultaneously. Further calibrations were performed according to
manufacturers’ instructions. Except from the calibrations, all sensor dealings were
performed solely by the investigators. The devices were removed after 48 hrs of ICU
admission or earlier when the patient was discharged.
During the study an arterial blood glucose value was measured with the AccuChek
handheld glucose measurement device (Performa II, lot 320098, Roche/Hitachi®, Basel,
Switzerland) as a reference value every two hours. These samples were used as calibration
when needed and as reference otherwise. An in-house quality assurance study showed
that the slope of the regression between this point-of-care measurement method and
arterial glucose measurement by blood gas analysis was 1.0 (95% CI 1.00-1.02, n = 1393,
sample range 2.9-30.0 mmol/l; Passing-Bablok regression) and 95% of the absolute
differences between reference and point-of-care measurement were lower than 15%,
thereby meeting the ISO 15197 guideline. All results were stored in the ICU’s clinical
information system (iMD-Soft; MetaVision, Tel Aviv, Israel). Using a dynamic computerised
algorithm implemented in 2001 12, glucose values between 5.0 and 8.0 mmol/l were
targeted. The glucose protocol was started for every patient at time of arrival at the ICU.
Insulin infusion was started when admission blood glucose exceeded 8.0 mmol/l. When
admission glucose was lower than 8.0 mmol/l, blood glucose was measured every 2 hrs
and insulin was started when blood glucose exceeded 8.0 mmol/l. The nursing staff was
Accuracy and reliability of CGM in the ICU
Ch
apte
r 9
127
instructed to adjust the insulin infusion rate depending on the current glucose value and
the rate of glucose change based on the previous five measurements. The 2-hr reference
glucose values were used as input in this algorithm. The nursing staff did not act upon
the sensor glucose values.
Data interpretation and statisticsThe Guardian device displays the average sensor glucose every five minutes and stores
all these values. The Navigator device refreshes the displayed glucose every minute and
stores the value of every tenth minute. The relative absolute deviation (RAD) [|sensor
value-reference glucose|/reference glucose] between reference and sensor glucose values
was calculated to assess the accuracy of the devices. For this purpose we linked the
reference value to the first available sensor value after the reference value using the
exact sampling times of both devices obtained after downloading the individual data. To
assess a possible delay of the CGM devices the reference value was linked to subsequent
sensor values up to 30 minutes after the reference value. We calculated the interval
between each reference-sensor pair in minutes and created six five-minute intervals (0-4,
5-9, 10-14, 15-19, 20-24, 25-29 minutes) in which we could match reference with sensor
glucose values. These five-minute intervals permit a fair comparison between both sensors
independent from the sampling frequency, since the Navigator stores data only every
tenth minute and the Guardian every five minutes. The median RAD per patient per
interval was calculated for each sensor and subsequently both sensors were compared
using a Wilcoxon signed ranks test for not normally distributed paired data. Assessment
of the lag time on the RAD per sensor was calculated using repeated measures ANOVA.
All analyses were performed using SPSS version 16.0.
For each sensor, all paired samples of reference glucose values and matching next sensor
values were plotted in a Clarke error grid 13. Also, the absolute differences between sensor
readings and reference glucose measurements were plotted against the average of the
two in a Bland-Altman plot 14.
Results
We included 61 patients in the study, of whom 1 patient dropped out due to cancellation
of surgery because of intercurrent febrile illness. In total we included 60 patients in
the final analysis of whom 48 were males. The median (range) age was 65 (25-85) years
and 26.7% of the patients were previously diagnosed with diabetes. The majority of the
patients underwent only a CABG procedure (53.3%). Median (IQR) APACHE IV PM and
maximum SOFA scores were 0.01 (0.003-0.02) and 6.0 (5.3-7.0). Patient characteristics are
reported in Table 1.
128
Table 1 Baseline characteristics
Patients, n=60
Male sex, n (%) 48 (80.0)
Age, years 65.0 (59.0-73.8)
Diabetes, n (%) 16 (26.7)
Procedure, n (%)
CABGValve surgeryCABG + valve surgery
32 (53.3)16 (26.7)12 (20.0)
APACHE IV PM 0.01 (0.003-0.02)
SOFA max 6.0 (5.3-7.0)
euroSCORE 4.0 (2.0-5.0)
ICU stay, hours 23.0 (19.0-45.8)
ICU readmission, n (%) 6 (10.0)
Death in ICU/hospital, n 0
Glucose ICU, mean (SD) 8.2 (2.1)
Data are given in median (IQR) unless stated otherwise. CABG, coronary artery bypass grafting; ICU, intensive care unit; APACHE, Acute Physiology and Chronic Health Evaluation score; SOFA, sequential organ failure assessment score; euroSCORE, European System for Cardiac Operative Risk Evaluation. Valve surgery includes mitral valve plasty, tricuspid valve plasty, aortic valve replacement or a combination of these.
ReliabilityDuring the study 1,017 reference glucose values were collected. Of these 91.8% could be
paired with a Navigator and 77.7% with a Guardian glucose value in the first data storage
interval. Missing values indicated technical problems with the device: unintentional
signal loss (Guardian: 19 patients; Navigator: 1 patient), interruption of real-time
representation of glucose values after delayed recalibration (Guardian) or temporary
failure of data-recording (Navigator: 4 patients). In 7 patients a new Guardian sensor
had to be placed due to sensor failure. In 2 patients a new Navigator sensor was placed
due to a disconnection between the actual sensor and fixation plate.
AccuracyMedian (IQR) RAD was significantly smaller for Navigator compared with Guardian
glucose measurements at intervals 0-4 and 5-9 minutes after the reference glucose (11%
[8-16] versus 14% [11-18], P = 0.05 and 10% [8-14] versus 14% [11-17], P = 0.001; Figure 1). The
lowest RAD of the Navigator was observed 5-9 minutes after the reference glucose, but
no significant effect of time was seen (P = 0.74, repeated measures ANOVA). The accuracy
of the Guardian did show a delay with the lowest RAD after 15-19 minutes (11% [8-13], P =
0.01; Figure 1). The results did not differ among subgroups of patients with or without
Accuracy and reliability of CGM in the ICU
Ch
apte
r 9
129
diabetes mellitus (Mann-Whitney U Test; data not shown).
Clarke error grids of reference glucoses with corresponding next sensor values are shown
in Figure 2 for each CGM device separately. For the comparisons with arterial reference
glucose values 81.8% of the Navigator and 73.2% of the available Guardian glucose values
fell in zone A. 17.7% of the Navigator and 25.2% of the Guardian glucose values fell in
zone B. Five of the 934 Navigator values fell in zone C or D (0.5%) and none in zone E
compared to 13 of the 790 Guardian values in zone C, D or E (1.3%).
To evaluate variations in accuracy over the range of measured glucose concentrations,
absolute differences between sensor readings and reference glucose values were plotted
(Figure 3). The variation in accuracy was larger with the Guardian looking at the range
between the 5th and 95th percentile (Guardian -3.03 to 2.27 mmol/l, range 5.30 mmol/l;
Navigator -1.83 to 2.53 mmol/l, range 4.36 mmol/l). There was no consistency in direction
of the error. Both positive and negative differences were seen, resulting in median (IQR)
differences coming close to zero (Navigator 0.10 [-0.60-0.90] mmol/l, Guardian 0.24 [-0.75-
1.07] mmol/l). No trend was observed visually for more inaccuracy in the hypo- and
hyperglycaemic ranges (Figure 3). We did not perform separate analyses to assess accuracy
during hypoglycaemia due to too few hypoglycaemic events: no severe hypoglycaemic
events (≤2.2 mmol/l) were measured and only 34 of 1,017 reference glucose values were
mildly hypoglycaemic (≤4.7 mmol/l) 5.
Figure 1 Head-to-head comparison of the accuracy of both sensors Relative Absolute Deviation (RAD) between reference and sensor glucose values at different intervals after the reference glucose measurement. RAD’s are displayed in medians. *P = 0.05, **P = 0.001
130
Figure 2 Clarke error grids of glucose measurements (A) Navigator. (B) Guardian. Each grid shows data pairs of reference glucose values at the x-axis with proximate sensor values (within 10 minutes for the Navigator and within 5 minutes for the Guardian) at the y-axis.
Figure 3 Bland-Altman plots of glucose measurements (A) Navigator. (B) Guardian. The x-axis represents the average of sensor and reference glucose values in mmol/l. The y-axis represents the absolute difference between sensor and reference glucose values in mmol/l. The dashed line represents the median difference (Navigator 0.10 and Guardian 0.24 mmol/l). The dotted lines represent the 5th and 95th percentile (Navigator -1.83-2.53 and Guardian -3.03-2.27 mmol/l).
Discussion
We report that the FreeStyle Navigator CGM system performed better than the Guardian
Real-Time in accuracy, defined by MAD in comparison to AccuChek arterial glucose
measurements, as well as reliability, determined by the technical error rates, in
postoperative cardiac surgery patients during ICU stay.
To our knowledge, the only study comparing these two devices is a clamp study by
Kovatchev and colleagues 15 in type 1 diabetes patients, concluding that the numerical
Accuracy and reliability of CGM in the ICU
Ch
apte
r 9
131
accuracy was comparable during normoglycaemia. However, the Navigator performed
significantly better during hypoglycaemia. It has to be noted that in our study we used
a new version of the Navigator with 1-hr initiation duration instead of the older version
with 10-hr initiation duration.
Regarding accuracy, our Guardian results are comparable with those from Logtenberg
et al. 7 who also performed a study in cardiac patients at the ICU, and reported a median
RAD of 12.3% during the ICU period. No studies using the Navigator at the ICU have
been published so far. Looking at the accuracy of both devices in our ICU population
compared to data in type 1 diabetes patients, this is comparable for the Guardian 16 and
even somewhat better for the Navigator 17. This suggests that at least in the population
of cardiac surgery patients CGM use seems feasible.
It is subject of debate whether sensor accuracy in the range that is acceptable for patients
with diabetes is also accurate enough for the critically ill patient. Hypoglycaemia is to be
avoided since already a single episode of low plasma glucose is independently associated
with mortality 5 and sedation makes it difficult to rely on hypoglycaemic symptoms.
Potentially dangerous sensor readings are those in the higher range while the reference
glucose is in the (near) hypoglycaemic range (Clark error grid zones D and E; Figure 2).
For the Navigator this occurred 3 times (0.3% of all sensor readings) and for the Guardian
5 times (0.6% of all sensors readings). We think that this low percentage of potentially
dangerous sensor readings is likely to be outweighed by hypoglycaemic episodes that
are likely to be prevented by continuous glucose monitoring 18. Of note, in the present
study the number of low glucose measurements is too small to draw conclusions on the
accuracy of the devices during hypoglycaemia.
We found a significant time-lag of the Guardian with optimal accuracy 15-19 minutes
following reference glucose. This is in accordance with Wei et al. 19 who found a median
delay of 16 minutes in their population of type 1 diabetes patients. The optimal
accuracy of the Navigator was reached 5-10 minutes after the reference glucose, however
no significant effect of time was found. All subcutaneous sensor measurements are
accompanied by a physiological delay of 0-10 minutes required for glucose to equilibrate
across the capillary endothelial barrier 20 suggesting the Navigator has a minimal
technological delay due to data processing and filtering. Our findings are different from
Garg et al. 21 who found a system time-lag of 15 minutes for the Navigator in adults
with type 1 diabetes. However, they used an earlier version of the Navigator with 10-hr
initiation time in their study which might explain these different findings and suggest
improvement of the new system with 1-hr initiation time. The differences between the
Guardian and Navigator and between different versions of the same device are intriguing;
however information on data processing is proprietary. The time-lag found favours the
132
Navigator for clinical use, as decision making will depend on real-time sensor values.
Also regarding reliability, the Navigator performed significantly better. Moreover, the
technical failure rates might be reduced when these systems attended to on a 24/24 hour
basis; most of the connection problems occurred at night when no member of the study
team was available. The nurses were blinded to the sensor glucose values and therefore
the alarms were set off, so a connection problem at night was only noticed when there
was a need for recalibration. However, the high frequency of connection problems of
the Guardian system suggests some kind of interference with other ICU equipment not
occurring with the Navigator system. The technical problems we experienced with the
Guardian device were described earlier, although not quantified 22.
As a reference method we used an arterial glucose measurement performed by the
AccuChek handheld device, since clinical decision rules in our and many other ICUs are
based on this measurement. Mortality is decreased by targeting hyperglycaemia relying on
point-of-care measurements 3;23, so we compared a new method with one proven effective.
Recently it is debated however how accurate this point-of-care meter is in comparison
with laboratory glucose measurements. Accuracy seems unacceptably decreased in older
patients with high disease severity scores and high ICU mortality 24. On the other hand
Meynaar et al. 25 concluded that AccuChek measurement has acceptable accuracy for use
in the ICU. As our patient group is characterised by relatively low mortality rates and
severity scores, AccuChek measurements should have been sufficiently accurate. This
is further substantiated by our in-house study comparing the AccuChek with a robust
laboratory reference method, which showed agreement as required by the ISO guideline
between the two methods. But even with the availability of a possible superior reference
method, the inaccuracy of the AccuChek seems random, so that the outcome of our
comparison between the two sensors would be unaffected.
In conclusion, we report that the FreeStyle Navigator CGM system performed better in
accuracy as well as reliability compared to the Guardian Real-Time in cardiac surgery
patients at the ICU. Remarkably, the RAD of both sensors was quite good as compared
to known data for outpatients. We think that this device can be used in this group
of ICU patients characterised by low disease severity scores and low mortality rates.
Further studies will concentrate on patient factors influencing sensor performance and
different populations of critically ill patients to allow an even better definition of the
ICU population who might benefit from the now available systems. Also, whether or not
the use of CGM truly improves glycaemic control and mortality has to be the subject of
further research.
Accuracy and reliability of CGM in the ICU
Ch
apte
r 9
133
AcknowledgementsThis study was supported by a European Foundation for the Study of Diabetes (EFSD)/
LifeScan research grant. The sensors used were provided free of charge by Medtronic
Minimed and at a discounted rate by Abbott Diabetes Care.
134
References1. Krinsley JS (2003) Association between hyperglycemia and increased hospital mortality in a heterogeneous
population of critically ill patients. Mayo Clin Proc 78: 1471-14782. Finfer S, Chittock DR, Su SY, et al (2009) Intensive versus conventional glucose control in critically ill patients.
N Engl J Med 360: 1283-12973. Van den Berghe G, Wilmer A, Milants I, et al (2006) Intensive insulin therapy in mixed medical/surgical
intensive care units: benefit versus harm. Diabetes 55: 3151-31594. Hermanides J, Vriesendorp TM, Bosman RJ, Zandstra DF, Hoekstra JB, Devries JH (2010) Glucose variability
is associated with intensive care unit mortality. Crit Care Med 38: 838-8425. Hermanides J, Bosman RJ, Vriesendorp TM, et al (2010) Hypoglycaemia is related with ICU mortality. Crit
Care Med 38: 1430-14346. De Block C, Manuel-Y-Keenoy B, Van Gaal L, Rogiers P (2006) Intensive Insulin Therapy in the Intensive Care
Unit. Diabetes Care 29: 1750-17567. Logtenberg SJ, Kleefstra N, Snellen FT, et al (2009) Pre- and postoperative accuracy and safety of a real-time
continuous glucose monitoring system in cardiac surgical patients: a randomized pilot study. Diabetes Technol Ther 11: 31-37
8. Price GC, Stevenson K, Walsh TS (2008) Evaluation of a continuous glucose monitor in an unselected general intensive care population. Crit Care Resusc 10: 209-216
9. Rabiee A, Andreasik RN, Abu-Hamdah R, et al (2009) Numerical and clinical accuracy of a continuous glucose monitoring system during intravenous insulin therapy in the surgical and burn intensive care units. J Diabetes Sci Technol 3: 951-959
10. Wentholt IM, Hoekstra JB, Devries JH (2007) Continuous glucose monitors: the long-awaited watch dogs? Diabetes Technol Ther 9: 399-409
11. Ellmerer M, Haluzik M, Blaha J, et al (2006) Clinical evaluation of alternative-site glucose measurements in patients after major cardiac surgery. Diabetes Care 29: 1275-1281
12. Rood E, Bosman RJ, van der Spoel JI, Taylor P, Zandstra DF (2005) Use of a computerized guideline for glucose regulation in the Intensive Care Unit improved both guideline adherence and glucose regulation. J Am Med Inform Assoc 12: 172-180
13. Clarke WL, Cox D, Gonder-Frederick LA, Carter W, Pohl SL (1987) Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care 10: 622-628
14. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307-310
15. Kovatchev B, Anderson S, Heinemann L, Clarke W (2008) Comparison of the numerical and clinical accuracy of four continuous glucose monitors. Diabetes Care 31: 1160-1164
16. Mazze RS, Strock E, Borgman S, Wesley D, Stout P, Racchini J (2009) Evaluating the accuracy, reliability, and clinical applicability of continuous glucose monitoring (CGM): Is CGM ready for real time? Diabetes Technol Ther. 11: 11-18
17. Weinstein RL, Schwartz SL, Brazg RL, Bugler JR, Peyser TA, McGarraugh GV (2007) Accuracy of the 5-day FreeStyle Navigator Continuous Glucose Monitoring System: comparison with frequent laboratory reference measurements. Diabetes Care 30: 1125-1130
18. Holzinger U, Warszawska J, Kitzberger R, et al (2010) Real time continuous glucose monitoring in critically ill patients - a prospective, randomized trial. Diabetes Care 33: 467-472
19. Wei C, Lunn DJ, Acerini CL, et al (2010) Measurement delay associated with the Guardian(R) RT continuous glucose monitoring system. Diabet Med 27: 117-122
20. Wentholt IM, Vollebregt MA, Hart AA, Hoekstra JB, Devries JH (2005) Comparison of a needle-type and a microdialysis continuous glucose monitor in type 1 diabetic patients. Diabetes Care 28: 2871-2876
21. Garg SK, Voelmle M, Gottlieb PA (2010) Time lag characterization of two continuous glucose monitoring systems. Diabetes Res Clin Pract 87: 348-353
22. Jacobs B, Phan K, Bertheau L, Dogbey G, Schwartz F, Shubrook J (2010) Continuous glucose monitoring system in a rural intensive care unit: a pilot study evaluating accuracy and acceptance. J.Diabetes Sci Technol 4: 636-644
23. Umpierrez GE, Isaacs SD, Bazargan N, You X, Thaler LM, Kitabchi AE (2002) Hyperglycemia: an independent marker of in-hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab 87: 978-982
Accuracy and reliability of CGM in the ICU
Ch
apte
r 9
135
24. Hoedemaekers CW, Klein Gunnewiek JM, Prinsen MA, Willems JL, Van der Hoeven JG (2008) Accuracy of bedside glucose measurement from three glucometers in critically ill patients. Crit Care Med 36: 3062-3066
25. Meynaar IA, van SM, Tangkau PL, et al (2009) Accuracy of AccuChek glucose measurement in intensive care patients. Crit Care Med 37: 2691-2696
Chapter 10
Microcirculation and its relation with continuous subcutaneous glucose sensor accuracy in cardiac surgery patients in the intensive care unit
Sarah E. Siegelaar, Temo Barwari, Jeroen Hermanides,
Peter H.J. van der Voort, Joost B.L. Hoekstra and J. Hans DeVries
Submitted for publication
138
Abstract
Objective: Continuous glucose monitoring (CGM) could be helpful in glucose
regulation in critically ill patients but accuracy of the systems is uncertain.
The accuracy might be influenced by impaired microcirculation. Therefore, we
investigated the microcirculation and its relation with accuracy of two CGM
devices in patients after cardiac surgery.
Methods: In a prospective, observational study in a 20-bed intensive care unit
(ICU) we included 60 patients who were about to undergo cardiac surgery. Two
CGM devices (Guardian Real-Time, Medtronic Minimed; FreeStyle Navigator,
Abbott Diabetes Care) were placed before surgery. Relative absolute deviation
(RAD) between CGM and arterial reference glucose was calculated to assess
accuracy. Microcirculation was measured by microvascular flow index (MFI),
perfused vessel density (PVD) and proportion of perfused vessels (PPV) using
sublingual sidestream dark-field imaging, and tissue oxygenation (StO2) obtained
with near-infrared spectroscopy.
Results: Thirty-two patients underwent only a CABG procedure. Median (IQR)
APACHE IV PM was 0.01 (0.003-0.02). StO2 significantly increased during ICU
admission (max 91.2% [3.9] after 6 hrs) and decreased thereafter, stabilizing after
20 hrs. The increase in StO2 was accompanied by a decrease in PVD. MFI and PPV
did not show a time effect. Microcirculatory variables were not associated with
sensor accuracy. For the Navigator lower peripheral temperature (b = -0.008, P=
0.003), higher APACHE IV PM (b = 0.017, P <0.001) and age (b = 0.002, P = 0.037)
and for the Guardian lower peripheral temperature (b = -0.006, P = 0.048) were
significantly associated with decreased sensor accuracy.
Conclusions: This study showed that microcirculation was impaired in patients
after cardiac surgery but to a limited extent only compared with septic patients
and healthy controls. The impairment in microcirculatory variables was not
related to sensor accuracy but peripheral temperature (both sensors), patient
age and APACHE IV PM (Navigator) were.
Microcirculation and its relation with subcutaneous CGM accuracy
Ch
apte
r 10
139
Introduction
Intensive glucose control is widely practised in the ICU 1. However, the associated
frequent glucose measurements are time consuming for the nursing staff and there
is no information available about glucose values in between those measurements.
Occurrence of hypoglycaemia is independently associated with mortality in the ICU 2.
Continuous glucose monitoring could therefore be a step forward by decreasing severe
hypoglycaemia frequency 3 and possibly by increasing time within blood glucose target
range. Although we recently reported promising results of continuous glucose sensor
accuracy in cardiac surgery patients 4, a patient population that seems to benefit most
from intensive insulin treatment 5;6, other studies reported suboptimal accuracy of the
commercially available systems 7;8.
A part of the accuracy problem in critically ill patients could result from the fact
that current commercially available needle-type continuous glucose sensors measure
glucose concentrations in the interstitial fluid and not directly in blood. The transport
of molecules such as glucose to the interstitial fluid is dependent on glucose supply
to the tissue and therefore on microcirculatory function. In critically ill patients the
microcirculatory function is altered 9;10 which might negatively affect sensor performance.
In this investigator-initiated study we investigated the microcirculation in patients
after cardiac surgery in the ICU and assessed whether microcirculatory variables and
other patient-related factors were associated with accuracy of the Guardian® Real-Time
(Medtronic Minimed, Northridge, CA) and FreeStyle Navigator® (Abbott Diabetes Care,
Alameda, CA) continuous glucose monitoring systems.
Materials and Methods
PatientsWe performed a prospective observational study in a 20-bed mixed ICU in the Onze
Lieve Vrouwe Gasthuis (OLVG; Amsterdam, the Netherlands). The study was approved
by the Institutional Review Board. We included patients above the age of 18 who were
to undergo elective cardiac surgery; coronary artery bypass grafting (CABG) and/or
valve surgery. We excluded patients with an abdominal condition which would impair
sensor insertion. Eligible patients received an information letter at least one week before
hospital admission and were asked to give written informed consent after additional
explanation of the study the day before surgery.
140
Glucose monitoringAfter we obtained informed consent the two needle-type sensors (the Guardian® Real-
Time, Medtronic Minimed, Northridge, CA and the FreeStyle Navigator®, Abbott Diabetes
Care, Alameda, CA) were inserted in the abdominal wall on either side of the umbilicus
the day before surgery and calibrated to allow stabilization of the signal. Upon arrival
at the ICU after surgery, the device’s internal clock was synchronised with the bedside
computer and both devices were calibrated simultaneously. Further calibrations were
performed according to manufacturers’ instructions. The devices were removed after
48 h of ICU admission or earlier when the patient was discharged.
During the study a reference arterial blood glucose value was measured with the
AccuChek handheld glucose measurement device (Performa II, lot 320098, Roche/Hitachi®,
Basel, Switzerland) every two hours. These samples were used as calibration when needed
and as reference otherwise. An in-house quality assurance study showed that the slope
of the regression between this point-of-care measurement method and arterial glucose
measurement by blood gas analysis was 1.0 (95% CI 1.00-1.02, n = 1,393, sample range
2.9-30.0 mmol/l; Passing-Bablok regression) and 95% of the absolute differences between
reference and point-of-care measurement were lower than 15%, thereby meeting the ISO
15197 guideline 11. All results were stored in the ICU’s clinical information system (iMDsoft;
MetaVision®, Tel Aviv, Israel). Using a dynamic computerised algorithm implemented in
2001 12, glucose values between 5.0 and 8.0 mmol/l were targeted. The nursing staff was
instructed to adjust the insulin infusion rate depending on the current glucose value
and the rate of glucose change based on the previous five measurements. The nurses
did not act upon the sensor glucose values.
Measurement of the microcirculationWe assessed the microcirculation by recording the sublingual microcirculation, after
gentle removal of excess saliva, using a handheld sidestream dark-field (SDF) camera.
This method has been described in detail previously 13. Imaging was performed as soon
as possible after the patient arrived at the ICU and 2 and 4 hours thereafter, as well as
at 8:00, 12:00 and 16:00 hrs the next days until the patientwas discharged from the ICU
with a maximum of 48 hrs after ICU admission. At each timepoint SDF recordings of at
least three different sublingual sites were recorded, stored and later scored in random
order to prevent bias. During the scoring procedure recordings were excluded when there
was no flow visible in the large vessels indicating a pressure artefact, presence of excess
saliva making it impossible to reliably visualise all vessels, and/or excess movement of
the recording. We analyzed small vessels with a diameter <20 μm in line with previous
literature 14 since these capillaries and small venules are considered to be most important
in nutrient, i.e. glucose, transport. Together with every SDF recording a reference plasma
glucose measurement was obtained to calculate sensor accuracy.
Microcirculation and its relation with subcutaneous CGM accuracy
Ch
apte
r 10
141
We scored the microvascular flow index (MFI), the proportion of perfused vessels (PPV)
and the perfused vessel density (PVD) per sublingual site per patient. The three outcomes
per timepoint per patient were then averaged according to the consensus statement
on how to evaluate the microcirculation 15 using the automated vascular analysis (AVA)
programme version 3.0 (Department of Medical Technological Development, Academic
Medical Centre, and MicroVision Medical, Amsterdam, the Netherlands). In brief, for PVD
and PPV the vessel density was calculated as the number of vessels crossing 3 horizontal
and 3 vertical equidistant lines divided by the total length of the lines. Perfusion of the
crossing vessels was scored as follows: 0 = no flow, 1 = intermittent flow (flow present
<50% of the recording), 2 = sluggish flow (flow present >50% but <100% of the recording
or continuous very slow flow), 3 = continuous flow. PVD was then calculated as the
number of crossing vessels with flow present (scores 2 or 3) reported in n/mm. PPV is
the proportion of vessels with flow present (scores 2 or 3) 9. For the MFI the predominant
type of flow in four quadrants was determined according to the same scoring system.
The MFI is the sum of these flow scores divided by the number of quadrants where the
vessel type is visible 16;17. The heterogeneity index (HI) per timepoint for the PVD was
calculated by dividing the difference between the lowest and the highest value by the
mean to objectify intra-site differences per timepoint 18.
The MFI was scored by two different investigators (SES and TB), compared and revised
by consensus when the separate scores differed more than one point. The final score
was obtained by averaging the separate scores. The PPV and PVD were scored by one
investigator (SES) and intra-observer variability was determined by reanalyzing 17
randomly chosen movie sequences again after 4 weeks. The intraclass correlation
coefficient (ICC) was calculated using a two-way random model to determine absolute
agreement. The ICC (95% CI) for PVD was 0.94 (0.84-0.98) and for PPV 0.94 (0.84-0.98),
indicating good intra-observer agreement.
Additionally, tissue oxygenation (StO2) was measured with near-infrared spectroscopy
to assess microcirculatory function (InSpectratm StO2 Tissue Oxygenation Monitor,
Hutchinson Technology, Hutchinson, MN, USA). The StO2 is a measure of oxygen
consumption of the tissue and a predictor of organ dysfunction 19. It reflects the ratio of
oxygenated haemoglobin to total haemoglobin in the microcirculation measured by the
absorption of near infrared light (wavelength 700-1000 nm). The probe was placed on the
thenar muscle at ICU arrival and removed after 48 hrs or upon discharge from the ICU.
Other variablesBesides microcirculatory variables, other factors possibly influencing CGM accuracy
were obtained at the time of every reference glucose measurement: mean arterial
pressure (MAP), peripheral and rectal temperature, pulse oximeter oxygen saturation
142
and dosing of vasoactive medication. The Acute Physiology and Chronic Health Evaluation
IV predicted mortality (APACHE IV PM) was calculated for the first 24 hrs of admission
and the Sequential Organ Failure (SOFA) score was obtained daily. Also the European
System for Cardiac Operative Risk Evaluation (euroSCORE) score, a method of calculating
predicted operative mortality risk for patients undergoing cardiac surgery, was recorded
for every patient.
Data interpretation and statisticsWe calculated the relative absolute deviation (RAD) [|sensor value-reference glucose|/
reference glucose] between reference glucose and the next available sensor glucose value
within 5 minutes to assess the real-time accuracy of the devices. To describe patient
characteristics and overall circulatory function the median (IQR) or mean (SD) of the
mean values per patient were calculated (Table 1 and 2). First, associations between sensor
accuracy and individual circulatory variables were assessed using a linear mixed-effects
model for repeated measures. Second, we built a multivariable model for each sensor
with those variables having a significant association with the accuracy of that specific
sensor. Last, the microcirculatory variables (MFI, PVD, PPV and StO2) were forced in turn
into the definite model to assess a possible independent effect on sensor accuracy. All
analyses were performed using Predictive Analytics Software (PASW) statistics version
18.0 (SPSS Inc., Chicago, IL, USA). A P-value <0.05 was considered statistically significant.
Results
We included 61 patients in the study, of whom 1 patient dropped out due to cancellation
of surgery because of intercurrent febrile illness. In total 60 patients were available for
the final analysis of whom 48 were males. The median (range) age was 65 (25-85) years
and 16 of the patients were previously diagnosed with type 2 diabetes. The majority of
the patients underwent only a CABG procedure (53.3%). The median (IQR) euroSCORE
score was 4.0 (2.0-5.0). Median (IQR) APACHE IV PM and maximum SOFA scores were 0.01
(0.003-0.02) and 6.0 (5.3-7.0). Patient characteristics are reported in Table 1.
MicrocirculationIn total, SDF recordings at 246 time points met our quality criteria for MFI assessment,
with a median (IQR) of 4 (3-5) time points per patient. For PVD and PPV scoring only 178
time points were suitable due to excessive movement of one of the three recordings of
that timepoint (median [IQR] 3 [2-4] per patient). The median (IQR) MFI was 2.8 (2.7-2.9),
PVD 9.3 (8.4-10.0) vessels/mm, PPV 0.97 (0.96-0.99), HIpvd 0.17 (0.13-0.21) and StO2 90.3%
(87.1-92.0) (Table 2). Microcirculation variables were not different between patients with
or without diabetes. The mean (SD) StO2 significantly increased during the first hours of
Microcirculation and its relation with subcutaneous CGM accuracy
Ch
apte
r 10
143
Table 1 Patient characteristics
Patients, n = 60
Male sex, n (%) 48 (80.0)
Age, years 65.0 (59.0-73.8)
Diabetes, n (%) 16 (26.7)
Procedure, n (%)
CABGValve surgeryCABG + valve surgery
32 (53.3)16 (26.7)12 (20.0)
APACHE IV PM 0.01 (0.003-0.02)
SOFA max 6.0 (5.3-7.0)
euroSCORE 4.0 (2.0-5.0)
ICU stay, hours 23.0 (19.0-45.8)
ICU readmission, n (%) 6 (10.0)
Death in ICU/hospital, n 0
Glucose ICU, mmol/l (mean [SD]) 8.2 (2.1)
RAD Navigator, % 11 (8-15)
RAD Guardian, % 14 (11-18)
Dopamine, μg/kg/min (n=60) 1.62 (1.03-2.34)
Nitroglycerine, μg/kg/min (n=57) 0.22 (0.16-0.34)
Enoximon, μg/kg/min (n=40) 1.19 (0.80-1.71)
Ketanserine, μg/kg/min (n=4) 0.26 (0.08-0.39)
Noradrenaline, μg/kg/min (n=1) 0.09
Values are depicted as median (IQR) unless stated otherwise, calculated from the mean per patient during ICU admission. APACHE IV PM, Acute Physiology and Chronic Health Evaluation IV predicted mortality; CABG, coronary artery bypass grafting; euroSCORE, European System for Cardiac Operative Risk Evaluation; RAD, relative absolute deviation; SOFA max, maximum sequential organ failure assessment score per patient during the study period; Valve surgery includes mitral valve plasty, tricuspid valve plasty, aortic valve replacement or a combination of these.
ICU admission, from 88.6% (5.6) to a maximum of 91.2% (3.9) after 6 hrs, and decreased
thereafter, stabilizing after 20 hrs (Figure 1). This increase in StO2 was accompanied with
a decrease in PVD (b= -0.697, SE=0.267, p=0.01, linear mixed-effects model for repeated
measures). In line with this observation we found that the PVD was lowest in the first
eight hours after ICU admission compared with the next day (median [IQR] 9.2 [8.1-9.9]
and 9.4 [8.6-10.5], P = 0.049 Wilcoxon signed ranks test). The MFI, PPV and HIpvd did not
change over time. The arterial oxygen content (CaO2; haemoglobin concentration*pulse
144
oximeter oxygen saturation*1.34) was stable during ICU admission (mean [SD] 12.6 [1.3]
ml O2/100 ml).
Table 2 Haemodynamic and microvascular variables
Patients, n = 60
MFI small vessels 2.8 (2.7-2.9)
PPV small vessels 0.97 (0.96-0.99)
PVD small vessels, n/mm 9.3 (8.4-10.0)
Heterogeneity index PVD 0.17 (0.13-0.21)
StO2, % 90.3 (87.1-92.0)
SpO2, % 97.2 (95.7-98.6)
Temperature rectal, ◦C 36.8 (36.4-37.1)
Temperature peripheral, ◦C 32.8 (31.6-33.6)
HF, beats/min 84.2 (74.4-90.0)
MAP, mmHg 73.6 (69.3-81.0)
Values are depicted as median (IQR), calculated from the mean per patient during ICU admission. HF, heart frequency; MAP, mean arterial pressure; MFI, microvascular flow index; PPV, proportion of perfused vessels; PVD, perfused vessel density; SpO2, peripheral oxygenation; StO2, tissue oxygenation.
CGM accuracyOverall, the Navigator sensor performed significantly better than the Guardian
sensor (median [IQR] RAD 11% [8-16] and 14% [11-18], p=0.05 4). Results of this head-to-
head comparison are described in Chapter 9 of this thesis. In a linear mixed-effects
model for repeated measures, variables that were individually associated with worse
Navigator sensor accuracy were higher age, a diagnosis of diabetes, decreased peripheral
temperature, increase in ketanserine, dopamine or enoximon dose and higher APACHE IV
PM. Only decreased peripheral temperature and increasing dopamine use were associated
with decreased accuracy of the Guardian sensor (Table 3A). None of the microcirculation
variables nor sex, MAP, norepinephrine dose and nitroglycerine dose showed significant
associations with accuracy of one of the sensors.
Subsequently, we built a multivariable model per sensor with the significantly associated
factors (Table 3A). Due to co-linearity between APACHE IV PM and ketanserine dose we
only included APACHE IV PM in the final model. This model showed that for the Navigator
only higher APACHE IV PM (b=0.017, SE= 0.004, p<0.001), lower peripheral temperature
(b= -0.008, SE= 0.003, p=0.003) and higher age (b=0.002, SE=0.001, p=0.037) remained
significantly associated with an increase in sensor RAD, i.e. worse sensor accuracy.
Microcirculation and its relation with subcutaneous CGM accuracy
Ch
apte
r 10
145
Figure 1 Tissue oxygenation in relation with arterial oxygen content during ICU admission Development of mean (SE) tissue oxygenation (StO2, left y-axis) and arterial oxygen content (CaO2, right y-axis) during intensive care unit admission (x-axis).
Figure 2 Sensor accuracy in relation with peripheral temperature Relative absolute deviation (RAD) of both sensors stratified by peripheral temperature quartiles, depicted as median (IQR). The highest RAD is seen below 31.00 degrees Celsius and the lowest RAD above 34.11 degrees Celsius. *P <0.05
For the Guardian a decrease in peripheral temperature remained associated with
an increase in RAD (b= -0.006, SE= 0.003, p=0.048). Figure 2 shows crude accuracy of
both sensors stratified by peripheral temperature. Lastly, we successively forced the
microcirculatory variables in these final models to assess a possible independent effect on
sensor accuracy. Forcing these variables in turns did not reveal a significant association
with sensor accuracy (Table 3B).
146
Table 3 Covariates associated with sensor accuracy
A
Univariate Multivariate
Navigator b SE p b SE p
APACHE IV PM 0.017 0.010 <0.001 0.017 0.004 <0.001
Age 0.002 0.001 <0.001 0.002 0.001 0.037
Peripheral temperature -0.009 0.003 0.001 -0.008 0.003 0.003
Dopamine dose 0.008 0.003 0.006 -0.001 0.004 0.785
Enoximon dose 0.016 0.006 0.008 0.009 0.007 0.207
DM no vs. yes -0.038 0.016 0.019 -0.024 0.018 0.177
Ketanserine dose 0.353 0.058 <0.001
Guardian
Peripheral temperature -0.006 0.003 0.036 -0.006 0.003 0.048
Dopamine dose 0.007 0.003 0.046 0.004 0.004 0.336
B
Navigator b SE p
+ MFIs -0.021 0.041 0.61
+ PVD -0.003 0.007 0.63
+ PPV 0.147 0.307 0.63
+ StO2 0.002 0.001 0.12
Guardian
+ MFIs -0.101 0.063 0.11
+ PVD 0.009 0.009 0.33
+ PPV -0.256 0.448 0.57
+ StO2 -0.001 0.001 0.61
Linear mixed-effects model for repeated measures, with the different patients as subjects and the two hour reference glucose measurements as repeated measures, using the AR(1) repeated covariance structure. Sensor accuracy, i.e. relative absolute deviation was the dependent variable. Covariates were put into the model as fixed effects. A: individual associations of variables with sensor accuracy (univariate) and basic model of both sensors including the variables with individual significant association with sensor accuracy (multivariate). B: forced introduction of the microcirculatory variables in turns does not show a significant association with sensor accuracy. APACHE IV PM, acute physiology and chronic health evaluation IV predicted mortality; DM, diabetes mellitus; MFIs, microvascular flow index of small vessels; PVD, perfused vascular density; PPV, proportion of perfused vessels; SE, standard error; StO2, tissue oxygen saturation.
Microcirculation and its relation with subcutaneous CGM accuracy
Ch
apte
r 10
147
Discussion
In the current study we found microcirculatory impairment to a limited extent in cardiac
surgery patients in the first hours of ICU admission after surgery during a median
follow-up of 23 hrs. Microcirculatory impairment was reflected by a decrease in PVD
and a transient increase in StO2 but no differences in PPV and MFI. The microcirculatory
variables showed no relationship with continuous glucose sensor accuracy. Variables
associated with worse sensor accuracy were higher age, lower peripheral temperature
and higher APACHE IV PM for the Navigator sensor and lower peripheral temperature
only for the Guardian sensor.
Previous studies of microcirculatory function in postoperative cardiac surgery patients
show various results. Some found normal MFI values at ICU arrival 20 and PVD one
hour after cardiopulmonary bypass 21, whereas De Backer et al. 22 found a decrease in
PPV at the end of on-pump surgery persisting until 24 hrs after surgery. Of note, the
latter is the only study besides the present study with more than 1 hr follow-up during
postoperative ICU admission of cardiac surgery patients. Looking at the absolute values
of the measured microcirculation variables, the MFI, PPV and HI found in our population
are roughly comparable with healthy controls measured by Trzeciak 18. PVD in our group
was somewhat smaller than in this control group and not as low as in sepsis patients 9.
Unfortunately, we did not have the opportunity to measure microcirculatory function
pre-operatively and therefore we are unable to assess pre- to postoperative changes.
We found an increase in StO2 after ICU admission with a peak after 6-12 hrs (Figure 1) and
a gradual decrease thereafter stabilizing to normal levels under 90% after 20 hrs of ICU
stay. No studies measuring StO2 in patients after cardiac surgery are known to us. Together
with the increase in StO2, the lowest PVD values were found in the first 8 hrs of ICU
admission and the increase in StO2 was significantly associated with the decrease in PVD
in mixed model analysis. Endotoxin release and subsequent triggering of the inflammatory
response observed in on-pump surgery could explain these findings. Endotoxin levels
peak shortly after surgery 23;24 and the inflammatory response, as represented by plasma
IL-6 concentration, is found to be highest after 6 hrs 23 which is comparable with the
observed peak in StO2. A much larger inflammatory response is seen in sepsis and this is
accompanied by large microcirculatory changes 9 and also a possible increase in StO2 due
to impaired oxygen offloading in a hyperdynamic flow state 25. The changes seen in our
population could thus be due to similar but much smaller changes as those seen in sepsis
caused by distributive shock, due to arterial-venous shunting of the microcirculation
reducing the number of perfused vessels thereby disabling oxygen offloading. This
hypothesis is supported by the stable CaO2 during ICU admission (Figure 1) which makes
it unlikely that the increased StO2 could be explained by an increase in oxygen supply.
148
Microcirculatory variables did not influence accuracy of both sensors. As far as we
know no previous studies assessing this relation have been performed. This finding
suggests that microcirculatory status does not impair glucose transport in cardiac
surgery patients; at least not to the extent that it has impact on subcutaneous sensor
performance. The patients included in this study had a relatively good microcirculation
as measured by MFI, PPV and PVD; although differences between patients were quite
large (PVD ranged 5.9-13.8 vessels/mm). It remains to be established whether in patients
with worse overall microcirculatory variables, i.e. patients with the sepsis syndrome,
sensor performance is affected.
A decrease in peripheral temperature did influence the accuracy of both sensors
negatively. This finding is in contrast with one other study performed in pediatric cardiac
surgery patients using the older Guardian RT CGM system not showing a relationship 26.
It is however plausible that the temperature of the skin influences sensor accuracy since
the optimal reaction temperature for the glucose oxidase enzyme incorporated in the
sensors is 30-40 degrees Celsius 27 and skin temperature during and shortly after cardiac
surgery is often below that lower limit. Nevertheless, the decreased sensor accuracy
needs to be put into clinical perspective as the median (IQR) RAD of both sensors during
peripheral temperatures under 31 degrees Celsius is still relatively low (Navigator: 10.5%
[4.8-20.0], Guardian: 14.7% [6.6-27.7]; Figure 2).
The performance of the Navigator was also negatively influenced by higher age and higher
APACHE IV PM. As we hypothesised, the highest RAD was found in the more severely ill
patients, but apparently the used microcirculatory variables did not mediate this effect
in those with the highest APACHE IV PM. Also APACHE IV PM in our patient group was
relatively low; therefore these results cannot be extrapolated to patients with higher
severity of disease scores. Notably, the accuracy of the Navigator was influenced by more
variables than the accuracy of the Guardian. Possibly the inaccuracy of the Guardian is
largely dependent on sensor-related factors overruling the influence of patient-related
variables.
Conclusions
This study showed that in cardiac surgery patients microcirculation was impaired after
surgery, reflected by a transient increase in StO2 and decrease in PVD, but was overall
quite good compared with septic patients and healthy controls. The impairment in
microcirculatory variables was not related to sensor accuracy but peripheral temperature
(both sensors) and age as well as APACHE IV PM (Navigator) was. These results support
CGM use in cardiac surgery patients characterised by low severity of illness. Further
Microcirculation and its relation with subcutaneous CGM accuracy
Ch
apte
r 10
149
studies need to assess the influence of microcirculatory changes on sensor accuracy in
more severely ill patients.
AcknowledgementsThis study was supported by a European Foundation for the Study of Diabetes (EFSD)/
LifeScan research grant. The sensors used were provided free of charge by Medtronic
Minimed and at a discounted rate by Abbott Diabetes Care.
150
References1. Inzucchi SE, Siegel MD (2009) Glucose control in the ICU--how tight is too tight? N Engl J Med 360: 1346-13492. Hermanides J, Bosman RJ, Vriesendorp TM, et al (2010) Hypoglycaemia is related with ICU mortality. Crit
Care Med 38: 1430-14343. Holzinger U, Warszawska J, Kitzberger R, et al (2010) Real time continuous glucose monitoring in critically
ill patients - a prospective, randomized trial. Diabetes Care 33: 467-4724. Siegelaar SE, Barwari T, Hermanides J, Stooker W, van der Voort PHJ, Devries JH (2011) Accuracy and reliability
of continuous glucose monitoring at the ICU; a head to head comparison of two subcutaneous glucose sensors in cardiac surgery patients. Diabetes Care in press
5. Van den Berghe G, Wouters P, Weekers F, et al (2001) Intensive Insulin Therapy in Critically Ill Patients. N Engl J Med 345: 1359-1367
6. Furnary AP, Gao G, Grunkemeier GL, et al (2003) Continuous insulin infusion reduces mortality in patients with diabetes undergoing coronary artery bypass grafting. J Thorac Cardiovasc Surg 125: 1007-1021
7. Logtenberg SJ, Kleefstra N, Snellen FT, et al (2009) Pre- and postoperative accuracy and safety of a real-time continuous glucose monitoring system in cardiac surgical patients: a randomized pilot study. Diabetes Technol Ther 11: 31-37
8. Rabiee A, Andreasik RN, Abu-Hamdah R, et al (2009) Numerical and clinical accuracy of a continuous glucose monitoring system during intravenous insulin therapy in the surgical and burn intensive care units. J Diabetes Sci Technol 3: 951-959
9. De Backer D, Creteur J, Preiser JC, Dubois MJ, Vincent JL (2002) Microvascular blood flow is altered in patients with sepsis. Am.J.Respir.Crit Care Med 166: 98-104
10. De Backer D, Creteur J, Dubois MJ, Sakr Y, Vincent JL (2004) Microvascular alterations in patients with acute severe heart failure and cardiogenic shock. Am Heart J 147: 91-99
11. International Organization for Standardization (2003) In vitro diagnostic test systems - Requirements for blood-glucose monitoring systems for self-testing in managing diabetes mellitus. ISO 15197
12. Rood E, Bosman RJ, van der Spoel JI, Taylor P, Zandstra DF (2005) Use of a computerized guideline for glucose regulation in the Intensive Care Unit improved both guideline adherence and glucose regulation. J Am Med Inform Assoc 12: 172-180
13. Goedhart PT, Khalilzada M, Bezemer R, Merza J, Ince C (2007) Sidestream Dark Field (SDF) imaging: a novel stroboscopic LED ring-based imaging modality for clinical assessment of the microcirculation. Opt Express 15: 15101-15114
14. Elbers PW, Ozdemir A, Heijmen RH, et al (2010) Microvascular hemodynamics in human hypothermic circulatory arrest and selective antegrade cerebral perfusion. Crit Care Med 38: 1548-1553
15. De Backer D, Hollenberg S, Boerma C, et al (2007) How to evaluate the microcirculation: report of a round table conference. Crit Care 11: R101
16. Boerma EC, Mathura KR, van der Voort PH, Spronk PE, Ince C (2005) Quantifying bedside-derived imaging of microcirculatory abnormalities in septic patients: a prospective validation study. Crit Care 9: R601-R606
17. Spronk PE, Ince C, Gardien MJ, Mathura KR, Oudemans-van Straaten HM, Zandstra DF (2002) Nitroglycerin in septic shock after intravascular volume resuscitation. Lancet 360: 1395-1396
18. Trzeciak S, Dellinger RP, Parrillo JE, et al (2007) Early microcirculatory perfusion derangements in patients with severe sepsis and septic shock: relationship to hemodynamics, oxygen transport, and survival. Ann Emerg Med 49: 88-98
19. Cohn SM, Nathens AB, Moore FA, et al (2007) Tissue oxygen saturation predicts the development of organ dysfunction during traumatic shock resuscitation. J Trauma 62: 44-54
20. den Uil CA, Lagrand WK, Spronk PE, et al (2008) Impaired sublingual microvascular perfusion during surgery with cardiopulmonary bypass: a pilot study. J Thorac Cardiovasc Surg 136: 129-134
21. Bauer A, Kofler S, Thiel M, Eifert S, Christ F (2007) Monitoring of the sublingual microcirculation in cardiac surgery using orthogonal polarization spectral imaging: preliminary results. Anesthesiology 107: 939-945
22. De Backer D, Dubois MJ, Schmartz D, et al (2009) Microcirculatory alterations in cardiac surgery: effects of cardiopulmonary bypass and anesthesia. Ann Thorac Surg 88: 1396-1403
23. Boelke E, Storck M, Buttenschoen K, Berger D, Hannekum A (2000) Endotoxemia and mediator release during cardiac surgery. Angiology 51: 743-749
24. Oudemans-van Straaten HM, Jansen PG, Hoek FJ, et al (1996) Intestinal permeability, circulating endotoxin, and postoperative systemic responses in cardiac surgery patients. J Cardiothorac Vasc Anesth 10: 187-194
25. Elbers PW, Ince C (2006) Mechanisms of critical illness--classifying microcirculatory flow abnormalities in distributive shock. Crit Care 10: 221
Microcirculation and its relation with subcutaneous CGM accuracy
Ch
apte
r 10
151
26. Piper HG, Alexander JL, Shukla A, et al (2006) Real-time continuous glucose monitoring in pediatric patients during and after cardiac surgery. Pediatrics 118: 1176-1184
27. Takamatsu S, Takano H, Binh-Khiem N, et al (2010) Liquid-phase packaging of a glucose oxidase solution with parylene direct encapsulation and an ultraviolet curing adhesive cover for glucose sensors. Sensors 10: 5888-5898
Chapter 11
Special considerations for the diabetic patient in the intensive care unit: targets for treatment and risks of hypoglycaemia
Sarah E. Siegelaar, Joost B.L. Hoekstra and J. Hans DeVries
Best Practice & Research: Clinical Endocrinology and Metabolism 2011; in press
154
Abstract
Due to the diabetes pandemic the number of diabetic patients admitted to the
ICU increases. Diabetic patients admitted to the ICU are more vulnerable for
developing complications as compared to non-diabetic patients, but this does
not directly translate into higher mortality rates. However, mortality might
differ per admission diagnosis. Hyperglycaemia is common in diabetic as well
as non-diabetic critically ill patients, but probably chronic hyperglycaemia
is pathophysiologically different from acute hyperglycaemia. As opposed
to non-diabetic patients, there is discussion about the association between
hyperglycaemia and mortality in diabetic patients. They do not seem to benefit
from strict glycaemic control and also glucose variability appears less harmful,
although clinical trials in diabetic populations have not been performed yet.
Diabetes is a risk factor for hypoglycaemia and evidence suggests that even near-
normal glucose levels are associated with worse outcome. Taking this together,
it is suggested to strive for moderate targets when treating hyperglycaemia in
critically ill diabetic patients.
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
155
I. Introduction
Hyperglycaemia is common in all critically ill patients, not only in those with a prior
diagnosis of diabetes mellitus. Hyperglycaemia in the critically ill is associated with
increased mortality 1 but large intervention studies evaluating the treatment of intensive
care unit (ICU) hyperglycaemia by intensive insulin therapy, show conflicting results 2-4.
There is no definite answer to the question whether and how tight hyperglycaemia of
critically ill patients has to be treated. Also, there may be a difference in outcome between
subgroups of patients. More and more the idea evolves that chronic hyperglycaemia
in critically ill patients with diabetes is pathophysiologically different from acute
hyperglycaemia in those without previously diagnosed diabetes, with consequences in the
hyperglycaemic as well as in the hypoglycaemic range. This could mean that treatment
targets and strategies in patients with diabetes should differ from those without diabetes.
If this hypothesis is true, the next question is whether undiagnosed diabetes should
be treated as patients with known diabetes or as patients without diabetes and a big
challenge for ICU physicians would be to unmask all patients with diabetes admitted
at the ICU. A large proportion of the patients admitted are unconscious hampering
adequate history taking and the medical history may be incomplete. Plasma glucose
values will not distinguish between those with and without diabetes due to the fact
that hyperglycaemia is also very common in non-diabetic critically ill patients. Therefore
some plead to measure HbA1c in all admitted patients to diagnose pre-existing diabetes 5, but this discussion lies outside the scope of this review.
In this review we will give an overview of the current literature on morbidity and
mortality in diabetic ICU patients with special consideration to glucose regulation,
insulin treatment and hypoglycaemia. For this purpose we searched MEDLINE for
studies conducted at any ICU concerning patients with diabetes, solely or in comparison
with patients without diabetes, without any restriction with respect to publication
date. Randomised controlled trials as well as observational studies were included.
We distinguished between studies regarding glucose regulation and studies looking
at morbidity and mortality of patients with diabetes at the ICU independent from
glucose regulation. Where possible the results are presented separately for different
subpopulations of ICU patients, for example cardiac surgery patients.
II. The diabetic patient in the ICU
A. Are diabetic patients at higher risk for morbidity?As a result of the diabetes pandemic in the western world, the number of diabetic patients
156
admitted at the ICU also increases. Are there any consequences of this increase? It is
known that patients with diabetes are vulnerable for the development of complications
during ICU admission: immune cell functions are hampered 6;7 which theoretically
promotes the incidence of various kinds of infections and there is a prothrombotic shift
in coagulation and fibrinolysis 8. Also the presence of diabetic complications as micro-
and macrovascular damage might have an effect on morbidity and perhaps mortality
as compared to non-diabetic individuals.
Nearly all studies looking at the incidence of complications in diabetic patients admitted
at the ICU support the hypothesis that they have increased morbidity. In cardiac surgery
patients all studies confirm that diabetes is a risk factor for postoperative complications
such as infections and, perhaps as a consequence, longer ICU and hospital stay 9-11. The
same goes for diabetic trauma patients 12;13. Also in mixed ICU populations diabetes seems
to be a risk factor for the development of (severe) infections 14-16 and acute organ failure 17. Interestingly, diabetic patients seem to have a decreased risk of developing acute
respiratory distress syndrome (ARDS) 18;19. This could be explained by the abovementioned
impairment in immune function promoting infections, as the overzealous activation
and recruitment of circulating neutrophils into the lung is involved in the pathogenesis
of ARDS 19. However, although the evidence points to an increased complication risk in
diabetic patients admitted at the ICU, there are also studies which could not confirm
this for development of nocosomial pneumonia 20 and bacteriuria 21 and as far as the
authors are aware no meta-analysis has been performed so far.
B. Are diabetic patients at higher risk for mortality? With a probably higher complication risk and increased length of stay, one would also
expect an increased mortality risk for diabetic ICU patients. Studies in cardiac surgery
patients indeed show increased mortality rates for all diabetic patients 9;22 or in those
with long-term complications 23. On the other hand, medical diabetic patients or patients
admitted in mixed ICU’s having an infection do not seem to have increased mortality 24-27. Data from mixed ICU populations without further specification are less conclusive.
Egi et al. 28showed even a lower adjusted mortality risk for diabetic patients which was
confirmed by the publication of the largest cohorts reported so far, 1,509,890 and 36,414
diabetic patients 29. But also increased 17;30;31 and similar mortality rates 1;18;32 have been
described.
It is intriguing to conclude that some diabetic ICU patients are perhaps protected and
others have a higher mortality risk. It might be that for example diabetic patients
undergoing cardiac surgery have a larger number of affected coronary vessels contributing
to the higher mortality in that group. The relative tolerance for hyperglycaemia, which
will be discussed in the next section, might contribute to decreased mortality rates,
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
157
but all proposed explanations are still only speculative. To further clarify the relation
between diabetic status and mortality, a meta-analysis making distinction between
subgroups of patients is needed.
III. Glucose regulation
A. Is hyperglycaemia deleterious to diabetic ICU patients? In patients without diabetes marked critical illness associated hyperglycaemia
is undisputedly related to morbidity and mortality 1;33. Critical illness-induced
hyperglycaemia is caused by inflammatory and neuro-endocrine derangements in these
patients leading to high hepatic glucose output and insulin resistance 34. In patients with
diabetes hyperglycaemia is already a pre-existing situation, although critical illness may
of course further derange blood glucose values by the above mentioned mechanisms.
Several studies examined the relation between hyperglycaemia and outcome in critically
ill patients with diabetes. In diabetic cardiac surgery patients hyperglycaemia above 11.1
mmol/l seems independently associated with increased wound infection rates 22;35 as
well as mortality and length of hospital stay 35 (to convert from mmol/l to mg/dl divide
by 0.0555). Contrary, Reyes et al. 36 did not find an association between hyperglycaemia
and complication rates in this group of patients, but in their study glucose levels were
very well regulated (mean [SD] postoperative BG 7.6 [2.5] mmol/l).
In mixed ICU populations the relationship between hyperglycaemia and mortality in
diabetic patients is less clear. Three studies containing large cohorts of diabetic patients
do report a relation between hyperglycaemia and mortality. Rady et al. 37 retrospectively
analyzed 1,083 ICU patients with diabetes and found that patients with median glucose
levels above 11.1 mmol/l showed increased mortality compared with patients with median
glucose levels between 4.4 and 11.1 mmol/l. In the latter group median glucose was
not related to mortality rates. Graham et al. 29 performed a retrospective analysis of
36,414 diabetic patients from the Mayo Clinic (Rochester, MN, USA). They found increased
hospital mortality rates in patients with peak glucose levels above 9.1 mmol/l compared
with patients with peak glucose levels between 7.2 and 9.1 mmol/l. Interestingly, patients
with peak glucose levels lower than 7.2 mmol/l were also found to have higher mortality
rates. However, no analyses were performed adjusting for possible confounders. Falciglia
et al. 38 did adjust for an important confounder: severity of disease. In a retrospective
cohort of 78,142 diabetes patients subdivided into groups with increasing mean glucose
levels, increasing hyperglycaemia was significantly associated with higher hospital
mortality compared to normoglycaemia (3.9-6.1 mmol/l).
158
In contrast with these findings, Egi et al. 28 could not confirm the deleterious effect
of hyperglycaemia in diabetic patients. In a cohort of 728 patients with diabetes, no
significant difference in ICU- and hospital mortality was found when analyzing four
equally sized groups of patients with increasing mean glucose levels, the glucose of
the lowest group ranging from 8.1 mmol/l to below. Also mean glucose values were
comparable between diabetic survivors and non-survivors (mean [SD] glucose 9.5 [2.9]
and 9.6 [2.8] mmol/l, respectively) 39. Two other studies including 574 (21.2%) 40 and 188
(22.7%) 26 diabetic patients investigated the effect of admission glucose on mortality but
no effect of hyperglycaemia higher than 11.1 mmol/l was seen. It might be possible that
the degree of pre-existing hyperglycaemia, expressed as HbA1c, alters the association
between acute glycaemia and mortality. A recent study including 415 patients with
diabetes shows that in patients with preadmission HbA1c levels above 7.0%, the higher
the glucose levels during admission, the lower the hospital mortality, in contrast to
patients with HbA1c levels under 7.0%, where higher glucose levels during admission
translate into higher mortality rates 41.
Without exception it has been found that at any given mean glucose level in the
hyperglycaemic range the mortality of patients with diabetes is lower than the mortality
of non-diabetic patients. The cut-off values where this effect occurs vary however. Rady
showed already an increased mortality rate and Falciglia an increased adjusted odds-ratio
for mortality in non-diabetic patients compared to diabetic patients in the subgroup
with a median 37 or mean 38 glucose between 6.2 and 8.0 mmol/l. This was not confirmed
in other studies which report a significant difference in mortality in favour of patients
with diabetes only for a mean glucose level of 8.0 mmol/l and above 39 or a peak glucose
level of 9.1 mmol/l and above 29. Another study investigating only admission glucose
values above 11.1 mmol/l shows also lower mortality rates for patients with diabetes 26.
The different effects of hyperglycaemia in ICU patients with and without diabetes suggest
that acute hyperglycaemia in critical illness and chronic hyperglycaemia in diabetes are
two distinct pathophysiological entities. Adaptation to hyperglycaemia might be a key
mechanism. Acute hyperglycaemia and inflammation induce e.g. oxidative stress which
causes endothelial damage 42. It is possible that patients with diabetes are already adapted
to these insults and therefore better tolerate episodes of hyperglycaemia compared with
non-diabetic patients, whose cellular adaptation mechanisms are not yet activated.
Attractive as it may be, this hypothesis is not substantiated any further in the literature.
B. Do diabetic patients benefit from intensive insulin therapy?Apart from whether there is any association between hyperglycaemia and mortality in
diabetic ICU patients, the relevant clinical question is whether they would benefit from
glucose lowering therapy. Several large clinical studies addressed this question but none
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
159
of them was specifically designed to look at the effect of intensive insulin therapy (IIT)
in diabetes patients. The results presented here are therefore derived from sub-analyses
except for three trials investigating intensive insulin therapy in diabetes patients during
cardiac surgery. Characteristics of the trials are presented in Table 1.
In 2001 van den Berghe et al. awoke the intensive care community publishing the results
of the so-called first Leuven study 3. They reported that IIT, with an achieved mean [SD]
morning blood glucose of 5.7 [1.1] mmol/l, in the surgical ICU significantly reduced
ICU and hospital mortality compared with conventional treatment (mean [SD] 8.5
[1.8] mmol/l), with an absolute mortality reduction from 8.0 to 4.6%. This was mainly
attributed to patients with an ICU stay of more than 5 days (mortality reduction from
20.2 to 10.6%). 204 of the 1548 patients included in this study had a history of diabetes.
Subanalysis of this group showed somewhat less survival benefit of IIT in diabetic patients
with an ICU stay longer than 5 days (16.0 to 9.5%), although it remained a significant
effect. In the second Leuven study 4, performed in a medical ICU, the reduction in
mortality with IIT was restricted to patients with an ICU stay of more than 3 days (52.5
to 43.0%, P = 0.009). For the subgroup of patients with diabetes (n = 203) IIT showed no
survival benefit overall and also not in those admitted to the ICU for more than 3 days.
Other randomised controlled trials (RCT’s) investigating the effect of intensive insulin
therapy on mortality with sub-analyses for patients with diabetes were performed
at mixed surgical and medical ICU’s, without making distinction between surgical
and medical patients 2;43-46. In all these studies IIT did not show survival benefit over
conventional glucose control in patients with diabetes. Also when analyzing pooled data
from both Leuven trials no benefit of IIT was seen in the diabetic patients 46. The achieved
glucose levels in the IIT group of the pooled analysis ranged from 5.8 to 6.5 mmol/l and in
the conventional group from 8.0 to 9.5 mmol/l. The NICE-SUGAR trial 2 published in 2009
included the largest population of diabetes patients (n = 1211) and achieved mean (SD)
glucose values of 6.4 (1.0) and 8.0 (1.3) mmol/l in the total population. In this subgroup
the benefit tended towards the conventional treatment, just like the overall outcome.
Pooling these data for the purpose of this manuscript, using Review manager version 5
(The Cochrane Collaboration, Oxford, UK), shows no benefit of intensive or conventional
treatment regarding mortality with little heterogeneity (I2 = 17%; Figure 1). The negative
results in the mixed populations are supported by analyses comparing mortality rates
before and after the implementation of an IIT protocol showing no significant decrease
in mortality with IIT in patients with diabetes 47;48.
The use of IIT in cardiac surgery patients however shows more positive results. Lazar et
al. 49 showed in 141 diabetic patients who underwent cardiac surgery that those receiving
IIT during surgery had a survival advantage over the initial two years after surgery and
160
Tab
le 1
Mai
n c
har
acte
rist
ics
of
inte
rven
tio
n s
tud
ies
on
in
ten
sive
in
suli
n t
her
apy
in d
iab
etic
IC
U p
atie
nts
Inte
nsi
veC
on
ven
tio
nal
Mo
rtal
ity
(%)
Stu
dy
Typ
ePo
pu
lati
onn
DM
ach
ieve
d B
G95
% C
I/SD
ach
ieve
d B
G95
% C
I/SD
IIT
Cty
pe
van
den
Ber
ghe
2001
3R
CT
Surg
ical
204
5.7
1.1
8.5
1.8
4.0
5.8
ICU
Surg
ical
>5d
ICU
46n
an
an
an
a9.
5*16
.0IC
U
van
den
Ber
ghe
2006
4R
CT
Med
ical
203
5.7
na
8.5
na
39.6
35.0
Hos
pit
al
Med
ical
≥3d
ICU
117
na
na
na
na
47.4
49.2
Hos
pit
al
van
den
Ber
ghe
2006
46R
CT
Mix
ed40
75.
81.
38.
41.
813
.013
.5IC
U
Mix
ed40
75.
81.
38.
41.
823
.222
.0H
osp
ital
Bru
nkh
orst
200
8 44
RC
TM
ixed
163
6.2
6.1-
6.3
8.4
8.2-
8.6
25.0
31.9
28-d
ay
Ara
bi 2
008
43
RC
TM
ixed
208
6.4
1.0
9.5
1.9
12.9
20.3
ICU
De
La R
osa
2008
45R
CT
Mix
ed61
6.5
5.6-
7.8
8.2
6.8-
10.0
31.0
37.5
28-d
ay
Fin
fer
2009
2R
CT
Mix
ed1,
211
6.4
1.0
8.0
1.3
31.7
27.7
90-d
ay
Kri
nsl
ey 2
006
47Pr
e-p
ost
Mix
ed53
27.
7n
a10
.4n
a19
.222
.6H
osp
ital
Kri
nsl
ey 2
009
48Pr
e-p
ost
Mix
ed94
27.
16.
2-8.
310
.28.
1-12
.626
.529
.4H
osp
ital
Laza
r 20
04 49
R
CT
Car
dia
c su
rger
y14
17.
50.
214
.80.
31.
4*10
.02-
year
Furn
ary
2003
50Pr
e-p
ost
Car
dia
c su
rger
y3,
554
9.8
1.7
11.9
2.3
2.5*
5.3
Hos
pit
al
Sum
mar
y of
th
e m
ain
ch
arac
teri
stic
s of
th
e in
terv
enti
on s
tud
ies
inve
stig
atin
g th
e ef
fect
of
inte
nsi
ve in
suli
n t
her
apy
(IIT
) in
cri
tica
lly
ill p
atie
nts
wit
h d
iabe
tes.
Glu
cose
va
lues
ach
ieve
d a
re o
f th
e to
tal
pop
ula
tion
an
d d
epic
ted
in
mm
ol/l
wit
h 9
5% c
onfi
den
ce i
nte
rval
or
stan
dar
d d
evia
tion
. Ou
tcom
e is
giv
en a
s m
orta
lity
per
cen
tage
s.
*P =
<0.
05 f
avou
rin
g II
T. S
ee f
or m
eta-
anal
ysis
Fig
ure
1. B
G, b
lood
glu
cose
; C
, con
ven
tion
al t
reat
men
t; D
M, d
iabe
tes
mel
litu
s; I
CU
, in
ten
sive
car
e u
nit
; p
re-p
ost,
era
of
con
ven
tion
al t
reat
men
t (p
re) c
omp
ared
wit
h a
n e
ra o
f in
ten
sive
tre
atm
ent
(pos
t); R
CT,
ran
dom
ised
con
trol
led
tri
al.
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
161
Figure 1; Meta-analysis of RCT’s on intensive insulin therapy in mixed medical/surgical diabetic patients in the ICU Figure 1 legend: Meta-analysis of randomised controlled trials comparing intensive insulin therapy (IIT) with conventional treatment in mixed medical/surgical patients with diabetes admitted at the intensive care unit.
also shorter postoperative length of stay, decreased episodes of recurrent ischemia and
less wound infections. This RCT was supported by a pre-post analysis showing that IIT
was protective for hospital mortality 50 and decreased length of stay but interestingly
not post-operative infection rates 51 after cardiac surgery. It has to be noted however
that the achieved blood glucose levels in the studies including cardiac surgery patients
lie above those conducted in the ICU. The achieved mean glucose in the intervention
group ranged between 7.5 and 9.8 mmol/l, which is roughly comparable with the mean
glucose of the conventionally treated ICU groups, and in the conventionally treated
group between 11.9 and 14.8 mmol/l, which are glucose levels associated with increased
mortality and morbidity in the observational studies 22;35. Therefore it is unknown
whether more intensive glycaemic control aiming at glucose levels below 7.0 mmol/l
is beneficial compared to moderate glycaemic control in cardiac surgery patients. A
study performed by Gandhi et al. in 2006 52 was designed to assess this question and
they achieved glucose levels after surgery of 6.3 (SD 1.6) mmol/l in the IIT group and 8.7
(2.3) mmol/l in the conventionally treated group. Unfortunately, the number of diabetic
patients was too small (n = 37) and the overall mortality rate too low (1%) to perform
subgroup analyses, though all 4 deaths occurred in the IIT group.
In summary, these data show that survival in mixed surgical/medical diabetic ICU
patients treated with strict glycaemic control (mean glucose 5.8-6.5 mmol/l) is not
different from patients treated with moderate glycaemic control (mean glucose 8.0-9.5
mmol/l). In diabetic cardiac surgery patients, moderate glycaemic control (mean glucose
7.5-9.8 mmol/l) has shown better results than loose glycaemic control (mean glucose
11.5-14.8 mmol/l). This implies that moderate glycaemic control aiming at glucose levels
between 7.5 and 10.0 mmol/l is perhaps the best treatment for all critically ill diabetic
patients, also because the lower the target, the more hypoglycaemic events occur.
162
IV. Hypoglycaemia
An important side-effect of insulin treatment is the occurrence of hypoglycaemia.
All intervention studies of intensive insulin therapy report a substantial increase in
hypoglycaemia incidence with intensive insulin therapy compared to less intensive
therapy. Besides the use of insulin, also the presence of diabetes is a risk factor for the
occurrence of severe hypoglycaemia, defined by cut-off levels of 3.3 mmol/l 53, 2.5 mmol/l 54 as well as 2.2 mmol/l 55, independent of insulin dose at the time of the event 54. Patients
with a prior diagnosis of diabetes may have an impaired counterregulatory response
hampering adequate reaction to overdosed exogenous insulin, likely explaining these
findings.
In mixed populations, without distinction between patients with or without diabetes,
the occurrence of severe hypoglycaemia seems to be associated with mortality
independently from severity of disease 55;56. Moreover, there is evidence that not only
severe hypoglycaemia using the diabetes outpatient definition of 2.2 mmol/l but also
glucose levels already lower than 4.7 mmol/l are harmful in critically ill patients 56.
Unfortunately, such association studies have not been performed separately in diabetic
patients. One study comparing nadir glucose values between diabetic survivors and
non-survivors demonstrated significantly lower values in non-survivors (4.9 [2.6] and 5.7
[2.7] mmol/l, P = 0.02) 39.
When comparing mortality rates of patients with and without diabetes in the
lower glycaemic range a notable phenomenon is seen. Mortality rates at any given
hyperglycaemic glucose level are higher in non-diabetic patients compared with diabetic
patients as shown previously, but in the lower glucose range the opposite seems to
occur. Graham showed hospital mortality rates to be significantly higher in diabetic ICU
patients with peak glucose values below 7.2 mmol/l compared with non-diabetic patients
in the same glucose range (P = 0.004) 29. Krinsley reported similar findings for diabetic
versus non-diabetic patients with mean glucose values under 6.6 mmol/l 47. Both studies
show unadjusted results only. The finding that diabetic patients show higher absolute
mortality rates when having a lower mean glucose during admission was confirmed by
Egi et al. 39 for mean glucose values between 4.4 and 6.1 mmol/l. But when adjusting also
for severity of disease, the significant effect of diabetes on ICU and hospital mortality
in this glucose range disappeared, although a trend remained visible; OR (95% CI) 0.33
(0.10-1.16, P = 0.08) and 0.45 (0.18-1.14, P = 0.09) for non-diabetic patients versus diabetic
patients regarding ICU and hospital mortality, respectively. In the latter study no absolute
mortality difference between the two groups was seen looking at mean glucose values
lower than 4.4 mmol/l.
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
163
These findings show that diabetic patients are not only at risk for severe hypoglycaemia
but also suggest a relative intolerance for normal and hypoglycaemic glucose values
compared with patients without diabetes, although firm evidence is lacking.
V. Glucose variability
Glycaemic variability is a consequence of severe illness and associated with intensive
insulin therapy. Also in diabetic outpatients, shortage of endogenous insulin production,
presence of insulin resistance and/or diminished counter-regulatory responses cause
instability of plasma glucose levels. Whether there is a negative effect of glucose
variability over and above pure hyperglycaemia seems dependent on the patient
population. In various adult and pediatric critically ill populations glucose variability
is strongly associated with mortality independent of the overall glycaemic status 48;57-60,
but the effect of short-term glucose fluctuations in patients with diabetes outside the
hospital remains subject of debate 61. However, no intervention study specifically aiming
at lowering glucose variability in diabetic patients at the ICU has been performed yet,
although emerging data in diabetes outpatients suggest that lowering glucose variability
does not result in improved outcome 62. It is therefore interesting to investigate whether
glucose variability is deleterious in diabetic critically ill patients.
Diabetic postoperative cardiac surgery patients 63 as well as diabetic patients in mixed ICU
populations 48;64 show larger glycaemic variability than non-diabetic patients. Only two
studies looked at the effect of glucose variability in critically ill diabetic patients. Egi et al.
did not find increasing mortality in quartiles of increasing glucose variability (assessed as
standard deviation) and ICU or hospital survival in 728 diabetic patients in a mixed ICU 28, except for an univariate comparison of mortality rates between the lowest and highest
glucose variability quartile, that is a standard deviation lower than 1.7 mmol/l versus 2.5-3.5
mmol/l, respectively (P = 0.002). In the non-diabetic population in this study the standard
deviation was an independent and strong predictor for mortality. Krinsley showed no
association between glucose variability (assessed as coefficient of variation) and mortality
in multivariate analysis of 942 diabetic predominantly medical ICU patients 48, which is in
contrast with their earlier findings in a population with only 23.8% patients with diabetes 58. There was however a marked increase in mortality across increasing coefficient of
variation strata in the subgroup of patients with the lowest mean glucose values (3.9-5.5
mmol/l). This is possibly due to the occurrence of hypoglycaemia, increasing both glucose
variability and mortality, but no analysis was presented adjusting for hypoglycaemia.
In conclusion, it is evident that patients with diabetes have higher glucose variability
during ICU stay but high glucose variability seems to be less harmful than in non-diabetic
164
patients. These results are in line with the observation that hyperglycaemia is more
detrimental in non-diabetic patients. It has to be noted though that no intervention
studies looking at the effect of specifically lowering glucose variability have been
performed in critically ill diabetic as well as in non-diabetic patients.
VI. The role of continuous glucose monitoring in the ICU
At this time glucose control is practiced by means of frequent point-of-care measurements.
Given the critical role of hypoglycaemia and glucose variability, the lack of information
in between those measurements may be of importance. Continuous glucose monitoring
(CGM) could be a useful tool in ICU glucose regulation by decreasing severe hypoglycaemia
frequency 65 and possibly increasing time in target range. However, accuracy results of
subcutaneous CGM systems are inconsistent and seem dependent on the population and
type of sensor used 66-69, but recent results in cardiac surgery and medical ICU patients
are promising 68;69. In addition to subcutaneous glucose monitoring, also intra-vascular
glucose monitoring devices are being developed with good results regarding accuracy 70. Future studies should investigate the benefit of these systems.
VII. Summary
As a consequence of the increasing incidence of diabetes throughout the world, the
number of diabetic patients admitted at the ICU is growing. This needs attention since
their treatment is in some aspects different from patients without diabetes. Nearly all
studies show that diabetic patients suffer from more complications and have longer
ICU and hospital length of stay. Mortality rates however are not simply higher. Diabetic
cardiac surgery patients do have a decreased survival as compared to non-diabetic
patients but data from medical ICU populations do not show a difference in mortality
between diabetic and non-diabetic patients. In mixed populations there is even evidence
that diabetic patients are relatively protected, however the data is not conclusive as there
are also studies showing equal or increased mortality rates. A meta-analysis on this topic
is needed. Hyperglycaemia is common in diabetic critically ill patients. There is discussion
about the association between hyperglycaemia and mortality in this patient group but
severe hyperglycaemia, above 11.1 mmol/l, is considered harmful and it seems useful
to lower these high values with insulin therapy. Considering the currently available
data on implementing insulin therapy, moderate glycaemic control is equally effective
in reducing mortality compared with strict glycaemic control, the latter increasing
hypoglycaemia substantially. This is of concern since diabetic patients are more prone
to develop severe hypoglycaemia which is associated with mortality and it is suggested
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
165
that they are already intolerant for glucose values considered normoglycaemic, although
conclusive evidence is lacking. Therefore we recommend to treat critically ill diabetic
patients with moderately intensive insulin therapy aiming at blood glucose levels
between 7.5 and 10.0 mmol/l, in that way avoiding extreme hyperglycaemia as well as
normo- and hypoglycaemia. Currently there are insufficient arguments to specifically
lower glucose variability, but intervention trials on this topic are awaited.
Our conclusions regarding the glucose target for ICU admitted diabetic patients support
the recommendations of the American Association of Clinical Endocrinologists and
the American Diabetes Association 71, that for critically ill patients in general insulin
treatment should be initiated at a threshold of 10.0 mmol/l and a glucose range of 7.8
to 10.0 mmol/l should be maintained. This consensus statement does not distinguish
between patients with or without previously diagnosed diabetes. We think that it is
important to add the presence of diabetes to the clinical situations that increase the
risk for hypo- and hyperglycaemia.
Practice Points
- Diabetic patients are at high risk for developing complications in the ICU
- Glucose control needs attention also in diabetic patients
- We recommend to maintain glucose levels between 7.5 and 10.0 mmol/l in diabetic
patients
- Hypoglycaemia and extreme hyperglycaemia should be vigorously avoided since these
are associated with mortality
- Currently there are insufficient arguments to specifically lower glucose variability
Research Agenda
- A meta-analysis has to be performed to objectify the influence of diabetes on mortality
risk in (subgroups of) critically ill patients
- Larger trials are needed to assess the effect of intensive insulin therapy in different
critically ill diabetic populations
- The threshold level below which glucose values are harmful in critically ill diabetic
patients needs to be determined
- Trials are needed to assess the effect of specifically lowering glucose variability
166
References1. Krinsley JS (2003) Association between hyperglycemia and increased hospital mortality in a heterogeneous
population of critically ill patients. Mayo Clin Proc 78: 1471-14782. Finfer S, Chittock DR, Su SY, et al (2009) Intensive versus conventional glucose control in critically ill patients.
N Engl J Med 360: 1283-12973. Van den Berghe G, Wouters P, Weekers F, et al (2001) Intensive insulin therapy in the critically ill patients.
N Engl J Med 345: 1359-13674. Van den Berghe G, Wilmer A, Hermans G, et al (2006) Intensive insulin therapy in the medical ICU. N Engl
J Med 354: 449-4615. Braithwaite SS (2010) Through the eyes of the A1C: a call to re-examine stress hyperglycemia. Crit Care Med
38: 717-7196. Alexiewicz JM, Kumar D, Smogorzewski M, Klin M, Massry SG (1995) Polymorphonuclear leukocytes in
non-insulin-dependent diabetes mellitus: abnormalities in metabolism and function. Ann Intern Med 123: 919-924
7. Marhoffer W, Stein M, Maeser E, Federlin K (1992) Impairment of polymorphonuclear leukocyte function and metabolic control of diabetes. Diabetes Care 15: 256-260
8. Lemkes BA, Hermanides J, DeVries JH, Holleman F, Meijers JC, Hoekstra JB (2010) Hyperglycemia: a prothrombotic factor? J Thromb Haemost 8: 1663-1669
9. Brown JR, Edwards FH, O’Connor GT, Ross CS, Furnary AP (2006) The diabetic disadvantage: historical outcomes measures in diabetic patients undergoing cardiac surgery -- the pre-intravenous insulin era. Semin Thorac Cardiovasc Surg 18: 281-288
10. Bucerius J, Gummert JF, Walther T, et al (2003) Impact of diabetes mellitus on cardiac surgery outcome. Thorac Cardiovasc Surg 51: 11-16
11. Ryan T, Mc Carthy JF, Rady MY, et al (1997) Early bloodstream infection after cardiopulmonary bypass: frequency rate, risk factors, and implications. Crit Care Med 25: 2009-2014
12. Ahmad R, Cherry RA, Lendel I, et al (2007) Increased hospital morbidity among trauma patients with diabetes mellitus compared with age- and injury severity score-matched control subjects. Arch Surg 142: 613-618
13. Bochicchio GV, Joshi M, Bochicchio K, Shih D, Meyer W, Scalea TM (2006) Incidence and impact of risk factors in critically ill trauma patients. World J Surg 30: 114-118
14. Laupland KB, Gregson DB, Zygun DA, Doig CJ, Mortis G, Church DL (2004) Severe bloodstream infections: a population-based assessment. Crit Care Med 32: 992-997
15. Alvarez-Lerma F, Nolla-Salas J, Leon C, et al (2003) Candiduria in critically ill patients admitted to intensive care medical units. Intensive Care Med 29: 1069-1076
16. Michalia M, Kompoti M, Koutsikou A, et al (2009) Diabetes mellitus is an independent risk factor for ICU-acquired bloodstream infections. Intensive Care Med 35: 448-454
17. Slynkova K, Mannino DM, Martin GS, Morehead RS, Doherty DE (2006) The role of body mass index and diabetes in the development of acute organ failure and subsequent mortality in an observational cohort. Crit Care 10: R137
18. Gong MN, Thompson BT, Williams P, Pothier L, Boyce PD, Christiani DC (2005) Clinical predictors of and mortality in acute respiratory distress syndrome: potential role of red cell transfusion. Crit Care Med 33: 1191-1198
19. Moss M, Guidot DM, Steinberg KP, et al (2000) Diabetic patients have a decreased incidence of acute respiratory distress syndrome. Crit Care Med 28: 2187-2192
20. Vardakas KZ, Siempos II, Falagas ME (2007) Diabetes mellitus as a risk factor for nosocomial pneumonia and associated mortality. Diabet Med 24: 1168-1171
21. Tissot E, Limat S, Cornette C, Capellier G (2001) Risk factors for catheter-associated bacteriuria in a medical intensive care unit. Eur J Clin Microbiol Infect Dis 20: 260-262
22. Guvener M, Pasaoglu I, Demircin M, Oc M (2002) Perioperative hyperglycemia is a strong correlate of postoperative infection in type II diabetic patients after coronary artery bypass grafting. Endocr J 49: 531-537
23. Leavitt BJ, Sheppard L, Maloney C, et al (2004) Effect of diabetes and associated conditions on long-term survival after coronary artery bypass graft surgery. Circulation 110: II41-II44
24. Heuser MD, Case LD, Ettinger WH (1992) Mortality in intensive care patients with respiratory disease. Is age important? Arch Intern Med 152: 1683-1688
25. Ligtenberg JJ, Meijering S, Stienstra Y, et al (2006) Mean glucose level is not an independent risk factor for mortality in mixed ICU patients. Intensive Care Med 32: 435-438
26. Stegenga ME, Vincent JL, Vail GM, et al (2010) Diabetes does not alter mortality or hemostatic and
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
167
inflammatory responses in patients with severe sepsis. Crit Care Med 38: 539-54527. Vincent JL, Preiser JC, Sprung CL, Moreno R, Sakr Y (2010) Insulin-treated diabetes is not associated with
increased mortality in critically ill patients. Crit Care 14: R1228. Egi M, Bellomo R, Stachowski E, French CJ, Hart G (2006) Variability of blood glucose concentration and
short-term mortality in critically ill patients. Anesthesiology 105: 244-25229. Graham BB, Keniston A, Gajic O, Trillo Alvarez CA, Medvedev S, Douglas IS (2010) Diabetes mellitus does not
adversely affect outcomes from a critical illness. Crit Care Med 38: 16-2430. Douglas SL, Daly BJ, Kelley CG, O’Toole E, Montenegro H (2007) Chronically critically ill patients: health-
related quality of life and resource use after a disease management intervention. Am J Crit Care 16: 447-45731. Douglas SL, Daly BJ, O’Toole EE, Kelley CG, Montenegro H (2009) Age differences in survival outcomes and
resource use for chronically critically ill patients. J Crit Care 24: 302-31032. Finney SJ, Zekveld C, Elia A, Evans TW (2003) Glucose control and mortality in critically ill patients. JAMA
290: 2041-204733. Siegelaar SE, Hermanides J, Oudemans-van Straaten HM, et al (2010) Mean glucose during ICU admission
is related to mortality by a U-shaped curve in surgical and medical patients: a retrospective cohort study. Crit Care 14: R224
34. Dungan KM, Braithwaite SS, Preiser JC (2009) Stress hyperglycaemia. Lancet 373: 1798-180735. Furnary AP, Wu Y (2006) Clinical effects of hyperglycemia in the cardiac surgery population: the Portland
Diabetic Project. Endocr Pract 12 Suppl 3: 22-2636. Reyes M, Jensen L, Stewart J, Kidd WT (2008) Glycemic control in diabetic and non-diabetic cardiac surgical
patients and length of hospital stay. Dynamics 19: 18-2437. Rady MY, Johnson DJ, Patel BM, Larson JS, Helmers RA (2005) Influence of individual characteristics on
outcome of glycemic control in intensive care unit patients with or without diabetes mellitus. Mayo Clin Proc 80: 1558-1567
38. Falciglia M, Freyberg RW, Almenoff PL, D’Alessio DA, Render ML (2009) Hyperglycemia-related mortality in critically ill patients varies with admission diagnosis. Crit Care Med 37: 3001-3009
39. Egi M, Bellomo R, Stachowski E, et al (2008) Blood glucose concentration and outcome of critical illness: the impact of diabetes. Crit Care Med 36: 2249-2255
40. Whitcomb BW, Pradhan EK, Pittas AG, Roghmann MC, Perencevich EN (2005) Impact of admission hyperglycemia on hospital mortality in various intensive care unit populations. Crit Care Med 33: 2772-2777
41. Egi M, Bellomo R, Stachowski E, et al (2011) The interaction of chronic and acute glycemia with mortality in critically ill patients with diabetes. Crit Care Med 39: 1-7
42. Brownlee M (2001) Biochemistry and molecular cell biology of diabetic complications. Nature 414: 813-82043. Arabi YM, Dabbagh OC, Tamim HM, et al (2008) Intensive versus conventional insulin therapy: a randomized
controlled trial in medical and surgical critically ill patients. Crit Care Med 36: 3190-319744. Brunkhorst FM, Engel C, Bloos F, et al (2008) Intensive insulin therapy and pentastarch resuscitation in
severe sepsis. N Engl J Med 358: 125-13945. De La Rosa GC, Donado JH, Restrepo AH, et al (2008) Strict glycaemic control in patients hospitalised in a
mixed medical and surgical intensive care unit: a randomised clinical trial. Crit Care 12: R12046. Van den Berghe G, Wilmer A, Milants I, et al (2006) Intensive insulin therapy in mixed medical/surgical
intensive care units: benefit versus harm. Diabetes 55: 3151-315947. Krinsley JS (2006) Glycemic control, diabetic status, and mortality in a heterogeneous population of critically
ill patients before and during the era of intensive glycemic management: six and one-half years experience at a university-affiliated community hospital. Semin Thorac Cardiovasc Surg 18: 317-325
48. Krinsley JS (2009) Glycemic variability and mortality in critically ill patients: the impact of diabetes. J Diabetes Sci Technol 3: 1292-1301
49. Lazar HL, Chipkin SR, Fitzgerald CA, Bao Y, Cabral H, Apstein CS (2004) Tight glycemic control in diabetic coronary artery bypass graft patients improves perioperative outcomes and decreases recurrent ischemic events. Circulation 109: 1497-1502
50. Furnary AP, Gao G, Grunkemeier GL, et al (2003) Continuous insulin infusion reduces mortality in patients with diabetes undergoing coronary artery bypass grafting. J Thorac Cardiovasc Surg 125: 1007-1021
51. Kee CA, Tomalty JA, Cline J, Novick RJ, Stitt L (2006) Change in practice patterns in the management of diabetic cardiac surgery patients. Can J Cardiovasc Nurs 16: 20-27
52. Gandhi GY, Nuttall GA, Abel MD, et al (2007) Intensive intraoperative insulin therapy versus conventional glucose management during cardiac surgery: a randomized trial. Ann Intern Med 146: 233-243
168
53. Durao MS, Marra AR, Moura DF, et al (2010) Tight glucose control versus intermediate glucose control: a quasi-experimental study. Anaesth Intensive Care 38: 467-473
54. Vriesendorp TM, van SS, Devries JH, et al (2006) Predisposing factors for hypoglycemia in the intensive care unit. Crit Care Med 34: 96-101
55. Krinsley JS, Grover A (2007) Severe hypoglycemia in critically ill patients: risk factors and outcomes. Crit Care Med 35: 2262-2267
56. Hermanides J, Bosman RJ, Vriesendorp TM, et al (2010) Hypoglycemia is associated with intensive care unit mortality. Crit Care Med 38: 1430-1434
57. Hermanides J, Vriesendorp TM, Bosman RJ, Zandstra DF, Hoekstra JB, Devries JH (2010) Glucose variability is associated with intensive care unit mortality. Crit Care Med 38: 838-842
58. Krinsley JS (2008) Glycemic variability: a strong independent predictor of mortality in critically ill patients. Crit Care Med 36: 3008-3013
59. Dossett LA, Cao H, Mowery NT, Dortch MJ, Morris JM, Jr., May AK (2008) Blood glucose variability is associated with mortality in the surgical intensive care unit. Am Surg 74: 679-685
60. Hirshberg E, Larsen G, Van DH (2008) Alterations in glucose homeostasis in the pediatric intensive care unit: Hyperglycemia and glucose variability are associated with increased mortality and morbidity. Pediatr Crit Care Med 9: 361-366
61. Siegelaar SE, Holleman F, Hoekstra JB, DeVries JH (2010) Glucose variability; does it matter? Endocr Rev 31: 171-182
62. Siegelaar SE, Kerr L, Jacober SJ, DeVries JH (2011) A decrease in glucose variability does not reduce cardiovascular event rates in type 2 diabetes patients after acute myocardial infarction: a reanalysis of the HEART2D study. Diabetes Care 34:
63. Masla M, Gottschalk A, Durieux ME, Groves DS (2010) HbA1c and Diabetes Predict Perioperative Hyperglycemia and Glycemic Variability in On-Pump Coronary Artery Bypass Graft Patients. J Cardiothorac Vasc Anesth epub
64. Al-Dorzi HM, Tamim HM, Arabi YM (2010) Glycaemic fluctuation predicts mortality in critically ill patients. Anaesth Intensive Care 38: 695-702
65. Holzinger U, Warszawska J, Kitzberger R, et al (2010) Real-time continuous glucose monitoring in critically ill patients: a prospective randomized trial. Diabetes Care 33: 467-472
66. Logtenberg SJ, Kleefstra N, Snellen FT, et al (2009) Pre- and postoperative accuracy and safety of a real-time continuous glucose monitoring system in cardiac surgical patients: a randomized pilot study. Diabetes Technol Ther 11: 31-37
67. Rabiee A, Andreasik RN, Abu-Hamdah R, et al (2009) Numerical and clinical accuracy of a continuous glucose monitoring system during intravenous insulin therapy in the surgical and burn intensive care units. J Diabetes Sci Technol 3: 951-959
68. Siegelaar SE, Barwari T, Hermanides J, Stooker W, van der Voort PHJ, Devries JH (2011) Accuracy and reliability of continuous glucose monitoring at the ICU; a head to head comparison of two subcutaneous glucose sensors in cardiac surgery patients. Diabetes Care 34: e31
69. Brunner R, Kitzberger R, Miehsler W, Herkner H, Madl C, Holzinger U (2011) Accuracy and reliability of a subcutaneous continuous glucose-monitoring system in critically ill patients*. Crit Care Med
70. Skjaervold NK, Solligard E, Hjelme DR, Aadahl P (2011) Continuous measurement of blood glucose: validation of a new intravascular sensor. Anesthesiology 114: 120-125
71. Moghissi ES, Korytkowski MT, DiNardo M, et al (2009) American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Diabetes Care 32: 1119-1131
Special considerations for the diabetic patient in the ICU
Ch
apte
r 11
169
Chapter 12
The effect of diabetes on mortality in critically ill patients; a systematic review and meta-analysis
Sarah E. Siegelaar, Maartje Hickmann, Joost B.L. Hoekstra,
J. Hans DeVries and Frits Holleman
Submitted for publication
172
Abstract
Objective: Critically ill patients with diabetes are at increased risk for development
of complications but the impact of diabetes on mortality is unclear. We conducted a
systematic review and meta-analysis to determine the effect of diabetes on short-term
mortality in critically ill patients, making a distinction between different ICU types.
Data Sources: We performed an electronic search of MEDLINE and EMBASE for
observational as well as intervention studies that reported on mortality of adult ICU
patients from May 2005 to May 2010.
Study Selection: Two reviewers independently screened the 3,220 publications
obtained for information regarding ICU, hospital or 30-day mortality of patients
with and without diabetes. We included 141 studies containing 12,489,574 patients,
including 2,705,624 deaths (21.7%). Of these patients at least 2,327,178 (18.6%) had
diabetes.
Data Extraction: The number of deaths among patients with and without diabetes
and/or mortality risk associated with diabetes was extracted. When only crude
survival data were provided, odds ratios (OR) and 95% confidence intervals (CI) were
calculated.
Data Synthesis: We used inverse variance with OR’s as the effect measure. A random
effects model was used because of anticipated heterogeneity. Overall no association
between diabetes and mortality risk was found. Analysis for ICU type showed a
disadvantage for patients with diabetes for all mortality definitions when admitted
at the surgical ICU (ICU mortality: OR [CI] 1.48 [1.04-2.11]; hospital mortality: 1.59
[1.28-1.97]; 30-day mortality: 1.62 [1.13-2.34]). In medical and mixed ICU’s no effect of
diabetes was seen. Sensitivity analysis showed that the disadvantage in the diabetic
surgical population was attributable to cardiac surgery (1.77 [1.45-2.16], P <0.00001)
and not to general surgery patients (1.21 [0.96-1.53], P = 0.11).
Conclusions: This meta-analysis showed that diabetes was not associated with
increased mortality risk in any ICU population except for those who underwent
cardiac surgery.
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
173
Introduction
The proportion of patients with diabetes admitted to the ICU is growing as a result of
the worldwide increase in type 2 diabetes. In cardiac surgery patients the growth is even
more pronounced since patients with diabetes-associated, often complex multivessel
macrovascular complications are preferably treated with surgery rather than by
percutaneous intervention 1;2. It is known that diabetic patients admitted to the ICU are
more prone to develop complications 3-5, at least in part due to hampered immune cell
function associated with the disease 6;7. One would expect an increased mortality rate for
diabetic ICU patients but the literature is at this point conflicting, reporting increased,
equal or even decreased mortality rates compared to patients without diabetes. Also
there might be a difference in outcome between various ICU populations, for example
cardiac surgery and medical patients.
To better understand the role of diabetes in critical illness, we conducted a systematic
review and meta-analysis regarding short-term mortality, including observational as
well as intervention studies that reported ICU, hospital or 30-day mortality rates of ICU
admitted patients with diabetes.
Materials and methods
This study was conducted according to the recommendations of the Meta-analysis of
Observational Studies in Epidemiology (MOOSE) group 8.
Data sources and search strategyTogether with the clinical librarian at our institution, an electronic search of MEDLINE
and EMBASE from May 1st 2005 to May 1st 2010 was performed for observational as well as
intervention studies that reported on mortality of adult ICU patients. The five-year limit
was chosen because we expected that insulin treatment regimens and other therapies
would be comparable among studies in this rapidly evolving field. Text terms and medical
subheading (MeSH) terms for ‘intensive care unit’, ‘critical care’ and ‘mortality’ were
combined. Since a preliminary search showed that diabetes was not always the primary
interest of included studies and to prevent publication bias, ‘diabetes mellitus’ as a
search term was not used to narrow the search. To avoid possible treatment induced
bias, randomised controlled trials (RCT) comparing intensive insulin therapy regimes
were excluded 9. We limited our search to research performed in humans, use of English
language and adult populations. Unpublished studies were not identified.
174
Study selection Two reviewers (SES and MH) independently screened the records. Agreement on final
inclusion was reached by consensus. Inclusion and exclusion criteria were defined a priori.
A study was included when it reported crude ICU, hospital and/or 30-day mortality of ICU
admitted patients with and without diabetes and/or univariate or multivariate analysis
of mortality risk of diabetic patients represented as odds ratio (OR), hazard ratio (HR) or
relative risk (RR). Studies reporting 28-day or 30-day mortality rates were combined. We
excluded studies where diabetes was the reason for ICU admittance, i.e. ketoacidosis,
as well as studies concerning patients with gestational diabetes and ICU readmissions.
When it was not possible to obtain the full manuscript from our institutional library
or the internet, the corresponding author was contacted if the population included in
the study was larger than 500 patients.
Data extractionFrom the included publications the following data were extracted: first author, year
of publication, country where the work was performed, study design, type of ICU,
population specification, reported mortality type, definition of diabetes and the
number of patients with and without diabetes. Subsequently the number of deaths
among the patients with and without diabetes and/or the mortality risk associated
with the presence of diabetes was collected. We contacted the corresponding author
for additional information when the publication reported only a P-value for mortality
risk or when the exact number or proportion of patients with diabetes was not
reported.
Study qualityNo individual assessment of study quality was performed. We did not expect bias in
outcome reporting since death is a robust endpoint and it is unlikely that patients were
lost to follow-up as the primary outcome was counted in the hospital.
Data synthesis and statistical methodsThe meta-analysis was performed using Review manager version 5 (The Cochrane
Collaboration, Oxford, UK). Analyses were performed separately for ICU, hospital and
30-day mortality as outcome variables. Data were synthesised using inverse variance
with odds ratios (OR) as the effect measure. In the primary analysis unadjusted results
were used where possible. When only crude mortality data were provided, the OR, 95%
confidence interval (CI) and standard error (SE) were calculated. An OR >1 suggested that
diabetes was associated with an increased risk of death. We stratified the analyses by
ICU type: trauma, surgical, medical and mixed ICU’s. Data were pooled using a random
effects model because heterogeneity between studies was anticipated. Heterogeneity
was assessed using the I2 statistic.
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
175
We performed sensitivity analyses in order to explore the robustness of the data and the
influence of the following factors on effect size by repeating the analysis 1) using a fixed
instead of a random model, 2) making distinction between cardiac and other types of
surgery and 3) for studies reporting unadjusted as well as adjusted outcomes to assess
the possible influence of confounders.
Results
Figure 1 summarises the study identification and selection process. After removal of
duplicates in the EMBASE and MEDLINE search, 3,220 potentially interesting records were
identified. Of these, we excluded 556 publications by review of the abstract. After review
of 2,664 full articles, 2,520 were excluded: 1,814 because no data regarding diabetes was
available at all, 579 because mortality data was not available, 35 publications reported
only on long-term mortality and 2 duplicate publications were identified. We were not
able to obtain the full texts of 90 potentially relevant records, of which 14 reported a
sample size of more than 500 participants. Three records were excluded by data extraction
as only a P-value for mortality was reported and after contacting the authors the raw
data could not be retrieved 10-12. Finally, 141 studies could be included in the analysis of
ICU mortality (n = 50 13-62), hospital mortality (n = 74 26;63-135) and 30-day mortality (n = 20 65;128;136-153). Three studies showed results of two outcome types; one reporting both ICU
and hospital mortality 26 and two reporting both hospital and 30-day mortality 65;128. Only
a few studies specified the type of diabetes of the patients included (n = 26). In most of
these studies diabetes was defined by use of glucose lowering drugs. Table 1 contains
the characteristics of the 141 included studies.
ICU mortality The analysis of ICU mortality contained 52,908 patients, including 7,576 deaths (Figure
2). Of those patients, at least 8,852 were diagnosed with diabetes (16.7%). Of two studies
the number of patients with diabetes could not be retrieved 59;62. Overall, pooling of the
data showed no survival advantage for either group (OR 1.03, CI 0.87-1.22, P = 0.74, I2 =
64%). Analysis per ICU type showed that in the surgical ICU patients without diabetes
did have a survival benefit over the patients with diabetes (OR 1.48, CI 1.04-2.11, P = 0.03,
I2 = 0%). No differences were observed in the trauma, medical and mixed ICU’s.
Hospital mortality The largest cohort could be analysed for hospital mortality: 12,403,355 patients (mortality
rate 21.7%) of whom 2,313,466 (18.7%) with diabetes (Figure 3). Pooling of all data did show
a small mortality increase for the patients with diabetes (OR 1.08, CI 1.00-1.15, p=0.04,
I2=70%). The disadvantage for patients with diabetes was attributable to the effect in
176
the surgical (OR 1.59, CI 1.28-1.97, P <0.0001, I2 = 56%) and trauma (OR 1.23, CI 1.12-1.36,
P <0.0001, I2 = 0%) population after stratifying to ICU type. For the medical and mixed
ICU’s no difference in outcome between the groups was seen.
30-day mortalityA total of 4,860 (25.5%) patients with diabetes and 14,180 patients without diabetes
were included in this analysis (Figure 4). For one study, including 14,271 patients, the
proportion of patients with and without diabetes could not be retrieved 152. Overall there
was no mortality difference between the patients with and without diabetes (OR 1.19,
CI 0.96-1.47, P = 0.10, I2 = 65%). Stratifying the data for ICU type showed a difference in
mortality favouring the patients without diabetes in the surgical ICU (OR 1.62, CI 1.13-
2.34, P = 0.009, I2 = 54%). Outcome did not differ between groups in medical and mixed
ICU’s. No separate analysis of trauma patients could be performed since none of the
studies reported 30-day mortality outcome in a trauma ICU.
Sensitivity analysesFor ICU and 30-day mortality the effect pointed to the same direction when using a
fixed instead of a random model; no benefit for patients with or without diabetes in
the trauma, medical and mixed population and a significant disadvantage for surgical
patients with diabetes. Regarding hospital mortality, the effect in the mixed and medical
ICU shifted towards an advantage for diabetes patients (medical: 0.89 [0.85-0.92], P
<0.00001; mixed: 0.90 [0.88-0.92], P <0.00001). This effect was attributable to the weight
of two very large studies. In the medical population the study of Martin et al. 101 had an
overall weight of 17.5% and in the mixed population the study of Graham et al. 26 weighed
69.9%. Performing the fixed analysis without these two large studies resulted in the same
conclusions as the random analysis; that is no effect.
For all outcomes we observed a mortality benefit for non-diabetic subjects in the
surgical ICU. This population consisted of cardiac as well as general surgery patients,
quite distinct regarding etiology of the underlying disease as compared to other ICU
populations. We performed a sensitivity analysis investigating whether the effect of
diabetes would be different for cardiac surgery or general surgery patients. Since the
effect was the same for ICU- hospital as well as 30-day mortality outcome and to increase
power, we performed the sensitivity analysis combining all surgical studies regardless
of mortality definition. This analysis showed that the increased mortality in patients
with diabetes in the surgical ICU was mainly attributable to the studies including
cardiac surgery patients (OR 1.77, CI 1.45-2.16, P <0.00001, I2 = 49%). Pooling of the general
surgery study data showed a trend towards higher mortality in diabetic patients (OR
1.21, CI 0.96-1.53, P = 0.11, I2 = 30%).
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
177
Most studies reported only crude mortality data or unadjusted regression analyses. To
assess the possible influence of confounders we first looked at the data of five studies
reporting unadjusted as well as adjusted results for hospital mortality 26;65;75;95;121. Among
other parameters, these studies adjusted for severity of disease. Overall, no difference
in effect size was observed when pooling the unadjusted results (OR 1.06 [0.83-1.35]) vs.
the adjusted results (OR 1.02 [0.79-1.34]). Introduction of adjusted instead of unadjusted
data in the analysis of hospital mortality did not change the outcome except for a shift
towards non-significance in the analysis of the trauma patients (1.02 [0.93-1.13], P = 0.45,
I2 = 0%).
Figure 1 Study selection
178
Figure 2 ICU mortality Forest plot showing odds ratio (OR) and 95% confidence intervals (CI) of ICU mortality risk comparing patients with and without diabetes. When ‘0’ is displayed as number of diabetic or non-diabetic patients, the information was not available.
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
179
180
Figure 3 Hospital mortality Forest plot showing odds ratio (OR) and 95% confidence intervals (CI) of hospital mortality risk comparing patients with and without diabetes.
Figure 4 30-day mortality Forest plot showing odds ratio (OR) and 95% confidence intervals (CI) of 30-day mortality risk comparing patients with and without diabetes. When ‘0’ is displayed as number of diabetic or non-diabetic patients, the information was not available.
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
181
Discussion
This large meta-analysis shows no significant difference in mortality between critically
ill patients with and without diabetes, except for a survival advantage for patients
without diabetes admitted to the ICU after cardiac surgery. Sensitivity analyses show
good robustness of the data. If anything, using a fixed model shifts the outcome from
equal survival towards a small benefit for diabetic patients in the medical and mixed
cohorts for hospital mortality which can be attributed to the inclusion of two very large
cohorts 26;101.
It did not come as a surprise to find a negative impact of diabetes on survival in cardiac
surgery patients. It is known that the diabetic cardiac surgery population is different
from the non-diabetic population. As a result of diabetes they are more likely to have
3-vessel coronary artery disease, left main coronary artery stenosis and left ventricular
systolic dysfunction 154;155, all associated with worse outcome.
On the other hand, it is remarkable to find no increased mortality for patients with
diabetes in the medical and mixed populations. Patients with diabetes are known
to have a higher chance of developing complications such as sepsis and acute organ
failure when admitted to the ICU 3-5. These are associated with mortality, at least in
the non-diabetic population. The increased infection risk is the result of immune cell
dysfunction associated with the disease 6;7. Apparently diabetes is only a risk factor for
the development of complications but once acquired, mortality risk is equal or perhaps
even lower.
A possible explanation for the relative protection of critically ill patients with diabetes
could lie in the different effects of stress-induced hyperglycaemia in the two groups.
Hyperglycaemia is common in critically ill patients and not only in those with diabetes.
It is associated with mortality 156 but it has been shown that patients with diabetes are
less affected by high glucose levels compared to their non-diabetic companions 26;77;157;158.
It might be that the relative protection from stress-induced hyperglycaemia counteracts
the increased mortality risk due to an increased amount of complications. Further studies
are needed to unravel the exact effects of hyperglycaemia in patients with and without
diabetes.
Due to the nature of the included studies we mainly show unadjusted results. It might
be possible that the baseline characteristics between patients with and without diabetes
were different, influencing the outcome. If there were a difference, it would seem likely
that patients with diabetes were more severely ill at admission. Adjusting for severity
of illness would have decreased the OR’s, as was shown in the sensitivity analysis,
182
thus indicating only greater advantage for patients with diabetes. This effect is seen
in the hospital mortality outcome of the trauma population. It might be possible that
adjustment for severity of disease decreases the negative effect of diabetes now seen
in cardiac surgery patients. However, our results represent the mortality risk of the
average patient with and without diabetes, irrespective of the differences between the
populations associated with diabetes.
There are a few limitations associated with this analysis. First, the published data do
not allow differentiating between type 1 and type 2 diabetes or between insulin treated
and non-insulin treated diabetes, and it may be possible that there are differences in
outcomes between these groups. Second, we could not retrieve mortality data of three
studies. It is unlikely that including these studies could have shifted the results to a
disadvantage for patients with diabetes, since the P-values for mortality related to the
presence of diabetes in these studies were not significant. Third, of 90 potentially relevant
studies the full manuscript could not be retrieved. These studies have included together
48,263 patients, which would have contributed only 0.39% to the total population, so
little influence on the outcome is expected.
Conclusions
We show in this meta-analysis that patients with diabetes who are admitted at the
medical, mixed and trauma ICU have similar chances of survival compared to patients
without diabetes. Diabetes only significantly increases mortality risk in patients
admitted after surgery, more specifically after cardiac surgery, a population with distinct
characteristics of the underlying disease. Further studies are needed to unravel the
pathophysiological mechanisms by which patients with diabetes seem to be protected
in non-surgical settings, despite encountering higher complication rates.
AcknowledgementsWe acknowledge the contributions of Heleen C. Dyserinck, Clinical Librarian, Academic
Medical Centre, Amsterdam, for her help performing the search and Rob J. Scholten,
department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical
Centre, Amsterdam, for his excellent support in the data extraction and synthesis. None
of the authors report a conflict of interest relevant to the content of the manuscript. No
funding was involved in preparation of the manuscript.
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
183
Tab
le 1
Ch
arac
teri
stic
s o
f th
e 14
1 st
ud
ies
incl
ud
ed i
n t
he
met
a-an
alys
is
Au
tho
rYe
arC
ou
ntr
yIC
U t
ype
Pop
ula
tio
nD
esig
nM
ort
alit
yN
pat
ien
ts%
DM
Dia
bet
es t
ype
Ail
awad
i20
08U
SASu
rgic
alM
itra
l va
lve
surg
ery
+ >7
0yR
etro
spec
tive
30d
315
21.9
n.s
.
Alm
og20
06Is
rael
Med
ical
APA
CH
E >
12Pr
osp
ecti
ve30
d78
42.3
n.s
.
An
2007
Ch
ina
Surg
ical
Car
dia
c su
rger
yR
etro
spec
tive
Hos
pit
al24
20.8
n.s
.
An
dri
kos
2009
Gre
ece
Mix
edA
KI
Pros
pec
tive
ICU
170
31.2
n.s
.
An
gstw
urm
2005
Ger
man
yM
edic
alSe
psi
sR
etro
spec
tive
Hos
pit
al30
824
.4n
.s.
Au
burt
in20
06Fr
ance
Med
ical
Pneu
moc
occa
l m
enin
giti
sPr
osp
ecti
veIC
U15
612
.2n
.s.
Bab
uin
2008
USA
Mix
edn
.s.
Ret
rosp
ecti
veH
osp
ital
, 30d
929
26.8
n.s
.
Bar
lage
2009
Ger
man
yM
ixed
Sep
sis
Pros
pec
tive
ICU
151
11.9
IDD
M +
NID
DM
Bit
to20
10It
aly
Trau
ma
Ch
est
trau
ma
Pros
pec
tive
Hos
pit
al24
20.8
n.s
.
Boc
hic
chio
2006
USA
Trau
ma
n.s
.Pr
osp
ecti
veH
osp
ital
1,16
78.
1ID
DM
+ N
IDD
M
Bod
i20
05Sp
ain
Med
ical
CA
PPr
osp
ecti
veIC
U52
922
.9n
.s.
Boy
er20
09Fr
ance
Med
ical
Nec
roti
zin
g so
ft t
issu
e in
fect
ion
Ret
rosp
ecti
veH
osp
ital
106
25.5
n.s
.
Bri
tt20
06U
SATr
aum
aB
urn
R
etro
spec
tive
ICU
200
7.0
n.s
.
Cag
atay
2007
Turk
eyM
ixed
Noc
osom
ial
bact
erae
mia
Pros
pec
tive
Hos
pit
al17
613
.1n
.s.
Ch
an20
05Si
nga
por
eM
ixed
Mel
ioid
osis
Ret
rosp
ecti
veIC
U27
59.3
n.s
.
Ch
en20
06Ta
iwan
Med
ical
Cir
rhos
isPr
osp
ecti
veH
osp
ital
102
28.4
n.s
.
Ch
en20
08Ta
iwan
Mix
edPy
ogen
ic l
iver
abs
cess
Ret
rosp
ecti
veIC
U72
51.4
n.s
.
Ch
en20
09Ta
iwan
Med
ical
Sep
sis
Pros
pec
tive
Hos
pit
al12
128
.1n
.s.
Ch
eng
2007
Ch
ina
Surg
ical
Sep
sis
Pros
pec
tive
Hos
pit
al31
812
.9n
.s.
Ch
ien
2005
Taiw
anM
edic
alSe
psi
sPr
osp
ecti
ve30
d63
36.5
n.s
.
Con
nel
ly20
07A
ust
rali
aSu
rgic
alA
orti
c/m
itra
l va
lve
surg
ery
Ret
rosp
ecti
ve30
d11
312
.4T
ype
1 +
2
Con
nel
ly20
09U
SAM
ixed
HA
PR
etro
spec
tive
Hos
pit
al21
924
.7n
.s.
Coo
ke20
08U
SAM
ixed
ALI
Ret
rosp
ecti
veH
osp
ital
1,11
321
.3n
.s.
Cu
thbe
rtso
n20
09U
KSu
rgic
alC
ard
iac
surg
ery
Pros
pec
tive
30d
1,01
017
.5n
.s.
184
Au
tho
rYe
arC
ou
ntr
yIC
U t
ype
Pop
ula
tio
nD
esig
nM
ort
alit
yN
pat
ien
ts%
DM
Dia
bet
es t
ype
Dam
as20
08B
elgi
um
Mix
edIn
fect
ion
+ O
FPr
osp
ecti
veH
osp
ital
1,19
116
.7n
.s.
Deg
oric
ija
2006
Cro
atia
Med
ical
Sep
sis
Pro-
/re
tros
pec
tive
ICU
314
44.6
n.s
.
Die
z20
09G
erm
any
Surg
ical
Valv
e ±
CA
BG
+ r
enal
fai
lure
Ret
rosp
ecti
veH
osp
ital
916
30.8
n.s
.
Dja
van
i20
09Sw
eden
Surg
ical
Ru
ptu
red
AA
APr
osp
ecti
ve30
d44
27.3
n.s
.
Du
nse
r20
09W
orld
wid
eM
edic
alSe
pti
c sh
ock
Ret
rosp
ecti
ve30
d29
013
.4n
.s.
Egi
2006
Au
stra
lia
Mix
edn
.s.
Ret
rosp
ecti
veIC
U4,
946
14.7
n.s
.
Egi
2008
Au
stra
lia
Mix
edn
.s.
Ret
rosp
ecti
veH
osp
ital
4,94
614
.7ID
DM
+ N
IDD
M
Enge
l20
09U
SASu
rgic
alC
AB
GR
etro
spec
tive
Hos
pit
al10
,590
33.3
n.s
.
Erbe
s20
06G
erm
any
Med
ical
Act
ive
pu
lmon
ary
TBC
Ret
rosp
ecti
veIC
U58
13.8
n.s
.
Fan
g20
08Ta
iwan
Med
ical
Cir
rhos
is +
AR
FPr
osp
ecti
veH
osp
ital
111
27.9
n.s
.
Fort
alez
a20
09B
razi
lM
ixed
n.s
.R
etro
spec
tive
ICU
122
32.0
n.s
.
Frei
re20
05U
SAM
edic
aln
.s.
Pros
pec
tive
Hos
pit
al1,
023
11.0
n.s
.; D
KA
exc
lud
ed
Frie
dri
ch20
06C
anad
aM
ixed
n.s
.Pr
osp
ecti
veH
osp
ital
182
21.4
His
tory
or
dis
char
ged
wit
h
med
s
Frio
ud
2010
Swit
zerl
and
Surg
ical
Car
dia
c su
rger
yR
etro
spec
tive
Hos
pit
al64
229
.1H
isto
ry o
r A
PG >
11.1
m
mol
/l
Gal
e20
07U
SATr
aum
an
.s.
Ret
rosp
ecti
veH
osp
ital
103
12.6
n.s
.
Gan
esh
2007
UK
Surg
ical
Lun
g tr
ansp
lan
tati
onR
etro
spec
tive
30d
681
6.3
n.s
.
Gam
ach
o-M
2005
Spai
nM
ixed
VAP
Pros
pec
tive
Hos
pit
al81
25.9
n.s
.
Gam
ach
o-M
2007
Spai
nM
ixed
VAP
Ret
rosp
ecti
veH
osp
ital
183
25.1
Req
uir
ing
trea
tmen
t
Gam
ach
o-M
2008
Spai
nM
ixed
Cat
het
er r
elat
ed B
SIPr
osp
ecti
veH
osp
ital
6619
.7n
.s.
Geo
rges
2009
Fran
ceM
edic
alA
cute
bac
teri
al m
enin
giti
sR
etro
spec
tive
ICU
8222
.0n
.s.
Gon
g20
06U
SAM
edic
alA
RD
SPr
osp
ecti
veIC
U21
115
.6n
.s.
Gra
ham
2010
USA
Mix
edn
.s.
Ret
rosp
ecti
veH
osp
ital
1,50
9,89
022
.6IC
D-9
; no
DK
A,
HO
NC
Pros
pec
tive
ICU
, hos
pit
al36
,414
16.3
n.s
.; n
o D
KA
, HO
NC
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
185
Gri
nia
stos
2008
Gre
ece
Surg
ical
Ch
olec
yste
ctom
yPr
osp
ecti
veH
osp
ital
2420
.8n
.s.
Hay
s20
06U
SAM
edic
alIs
chem
ic s
trok
eR
etro
spec
tive
Hos
pit
al23
522
.6n
.s.
Hen
ckae
rts
2009
Bel
giu
mM
edic
aln
.s.
Ret
rosp
ecti
veH
osp
ital
774
17.1
Insu
lin
, OA
D o
r d
iet
Hol
ley
2009
Wor
ldw
ide
Mix
edC
and
idem
iaR
etro
spec
tive
Hos
pit
al18
923
.3n
.s.
Hsu
2009
Taiw
anM
edic
alSe
pti
c m
enin
giti
sR
etro
spec
tive
Hos
pit
al40
57.5
n.s
.
Irib
arre
n-D
2009
Spai
nM
ixed
Mix
edPr
osp
ecti
veH
osp
ital
377
11.9
n.s
.
Jam
al20
09K
uw
ait
Mix
edA
. bau
man
ii i
nfe
ctio
nPr
osp
ecti
veH
osp
ital
2429
.2n
.s.
Jen
q20
07Ta
iwan
Med
ical
Cir
rhos
isPr
osp
ecti
veH
osp
ital
134
25.4
n.s
.
Kao
2006
USA
Trau
ma
n.s
.R
etro
spec
tive
Hos
pit
al34
3,25
02.
7ID
DM
+ N
IDD
M
Kar
imi
2008
Iran
Surg
ical
CA
BG
Cro
ss-
sect
ion
al30
d8,
890
33.6
n.s
.
Kem
arre
c20
08Fr
ance
Surg
ical
Abd
omin
al s
epsi
sPr
osp
ecti
veIC
U27
33.3
n.s
.
Ken
nea
lly
2007
USA
Mix
edC
. dif
fici
le i
nfe
ctio
nR
etro
spec
tive
30d
278
28.1
n.s
.
Kes
2007
Cro
atia
Med
ical
Isch
emic
str
oke
Ret
rosp
ecti
ve30
d63
034
.9n
.s.
Kri
nsl
ey20
07U
SAM
ixed
n.s
.R
etro
spec
tive
Hos
pit
al5,
365
20.7
n.s
.
Labe
lle
2008
USA
Med
ical
Can
did
a B
SIR
etro
spec
tive
ICU
111
34.2
n.s
.
Lan
2006
Taiw
anM
edic
alIs
chem
ic s
trok
eR
etro
spec
tive
ICU
233
32.2
n.s
.
Lee
2007
Kor
eaM
edic
alC
AP
Ret
rosp
ecti
veH
osp
ital
8527
.1n
.s.
Lem
ant
2008
Fran
ceM
edic
alC
hik
un
guya
in
fect
ion
Pros
pec
tive
ICU
3333
.3n
.s.
Lero
y20
09Fr
ance
Mix
edIn
vasi
ve C
and
ida
infe
ctio
nPr
osp
ecti
veIC
U26
810
.8T
ype
1
Liao
2009
Taiw
anM
ixed
AR
DS
Pros
pec
tive
Hos
pit
al17
225
.0n
.s.
Lin
(a)
2009
Taiw
anM
edic
aln
.s.
Pros
pec
tive
Hos
pit
al20
138
.3n
.s.
Lin
(b)
2009
Taiw
anM
edic
alPu
lmon
ary
TBC
+ R
FR
etro
spec
tive
Hos
pit
al59
22.0
n.s
.
Lore
nte
2009
Spai
nM
edic
alSe
psi
sPr
osp
ecti
veIC
U19
225
.5n
.s.
Lun
del
in20
10Sp
ain
Mix
edn
.s.
Pros
pec
tive
ICU
3828
.9In
suli
n, O
AD
or
die
t
Mar
kogi
ann
akis
2009
Gre
ece
Surg
ical
Infe
ctio
nPr
osp
ecti
veIC
U12
315
.4n
.s.
Mar
tin
2006
USA
Med
ical
Sep
sis
Ret
rosp
ecti
veH
osp
ital
10,4
22,3
0118
.6n
.s.
Mar
tin
-L20
10Eu
rop
eM
edic
alC
AP
+ in
tuba
tion
Pros
pec
tive
ICU
137
24.1
n.s
.
186
Au
tho
rYe
arC
ou
ntr
yIC
U t
ype
Pop
ula
tio
nD
esig
nM
ort
alit
yN
pat
ien
ts%
DM
Dia
bet
es t
ype
Mas
cia
2008
Euro
pe
Mix
edN
euro
logi
cal
Ret
rosp
ecti
veIC
U27
36.
2n
.s.
Mit
chel
l20
06A
ust
rali
a/N
ZL
Mix
edn
.s.
Pros
pec
tive
Hos
pit
al93
317
.5In
cl. i
nsu
lin
tr
eatm
ent
Mow
ery
(a)
2009
USA
Surg
ical
Gen
eral
su
rger
yPr
osp
ecti
veH
osp
ital
925
10.7
Typ
e 1
+ 2
Mow
ery
(b)
2009
USA
Trau
ma
Bra
in i
nju
ryPr
osp
ecti
veH
osp
ital
106
16.3
IDD
M +
NID
DM
Mu
rph
y20
09U
SAM
edic
alA
LI +
sep
tic
shoc
kR
etro
spec
tive
Hos
pit
al21
232
.1n
.s.
Mu
rth
y20
07U
SASu
rgic
alC
ard
iac
surg
ery
Ret
rosp
ecti
veIC
U68
614
.6In
suli
n, O
AD
or
die
t
Mu
sci
2009
Ger
man
ySu
rgic
alE
nd
ocar
dit
ic s
urg
ery
Ret
rosp
ecti
ve30
d22
122
.2n
.s.
Nai
r20
09U
SASu
rgic
alLi
ver
tran
spla
nta
tion
Pros
pec
tive
Hos
pit
al19
328
.5H
isto
ry, i
nsu
lin
/O
AD
, HbA
1c >
7%
Nas
raw
ay20
06U
SASu
rgic
al>4
day
s IC
UR
etro
spec
tive
ICU
393
20.1
n.s
.
Nse
ir20
06Fr
ance
Med
ical
CO
PD +
>48
h M
VPr
osp
ecti
veIC
U65
917
.8n
.s.
On
g20
09U
SATr
aum
a>3
0day
s IC
UR
etro
spec
tive
ICU
205
20.0
n.s
.
Pate
l (a
)20
09U
SAM
edic
alC
and
ida
sep
tic
shoc
kR
etro
spec
tive
ICU
3135
.5n
.s.
Pate
l (b
)20
09U
SASu
rgic
alC
ard
iac
surg
ery
+ w
oun
d i
nf.
Ret
rosp
ecti
ve30
d12
456
.5In
suli
n o
r O
AD
Pau
l 20
10Is
rael
Mix
edIn
fect
ion
Pros
pec
tive
30d
495
30.1
n.s
.
Pier
acci
2008
USA
Surg
ical
Surg
ery
Pros
pec
tive
Hos
pit
al94
64.
3n
.s.
Prat
ikak
i20
09G
reec
eM
ixed
Blo
odst
ream
in
fect
ion
Pros
pec
tive
ICU
148
18.9
n.s
.
Qu
ach
2009
Can
ada
Mix
edn
.s.
Ret
rosp
ecti
veH
osp
ital
3,77
812
.4IC
D-1
0 co
de
Rad
y20
05U
SAM
ixed
n.s
.R
etro
spec
tive
Hos
pit
al52
219
.7n
.s.
Ram
mae
rt20
09Fr
ance
Med
ical
CO
PD +
MV
Pros
pec
tive
ICU
116
19.8
n.s
.
Ran
ucc
i20
08It
aly
Surg
ical
Car
dia
c su
rger
yR
etro
spec
tive
Hos
pit
al4,
546
17.8
Insu
lin
or
OA
D
Rei
nta
m20
08Es
ton
iaM
ixed
n.s
.Pr
osp
ecti
veIC
U19
611
.2n
.s.
Rel
los
2006
Gre
ece
Mix
ed>9
0yPr
osp
ecti
veH
osp
ital
6025
.0n
.s.
Rh
odes
2006
UK
Mix
edn
.s.
Pros
pec
tive
ICU
5213
.5n
.s.
Ria
chy
2008
Leba
non
Med
ical
Seve
re s
trok
ePr
o-/
retr
osp
ecti
veIC
U62
27.4
n.s
.
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
187
Riv
ero-
A20
08Eu
rop
eM
edic
alA
tria
l fi
bril
lati
onR
etro
spec
tive
Hos
pit
al10
,701
26.9
n.s
.
Rod
rigu
ez20
09Sp
ain
Med
ical
Bac
teri
al C
AP
Ret
rosp
ecti
veIC
U18
423
.9T
ype
2
Ros
amel
2005
Fran
ceSu
rgic
alEn
doc
ard
itic
su
rger
yR
etro
spec
tive
Hos
pit
al98
9.2
n.s
.
Ros
e20
09U
KSu
rgic
alM
assi
ve t
ran
sfu
sion
Ret
rosp
ecti
veH
osp
ital
204
8.8
n.s
.
Ru
iz-B
2005
Spai
nM
edic
alA
MI
pos
t th
rom
boly
sis
Ret
rosp
ecti
veIC
U15
123
.8n
.s.
Sail
ham
er20
09U
SAM
edic
alC
. dif
fici
le c
olit
isR
etro
spec
tive
Hos
pit
al19
929
.6n
.s.
Sakr
2006
Euro
pe
Mix
edn
.s.
Pros
pec
tive
ICU
1,05
87.
7n
.s.
Sakr
2008
Euro
pe
Mix
edn
.s.
Pros
pec
tive
Hos
pit
al1,
729
7.0
n.s
.
San
tin
i20
10C
roat
iaM
ixed
Sep
sis
Ret
rosp
ecti
veIC
U15
229
.6n
.s.
Saye
d20
09Eg
ypt/
Ger
man
ySu
rgic
alC
ard
iac
surg
ery
Ret
rosp
ecti
veH
osp
ital
120
47.5
RPG
≥8.
3, F
PG≥7
m
mol
/l o
r m
eds
Sen
thu
ran
2008
Au
stra
lia
Med
ical
ESR
FR
etro
spec
tive
ICU
7044
.3n
.s.
Shah
2007
USA
Med
ical
n.s
.Pr
osp
ecti
veH
osp
ital
179
27.4
n.s
.
Silv
a20
06B
razi
lM
ixed
AR
FR
etro
spec
tive
Hos
pit
al12
89.
4n
.s.
Slei
man
2008
Ital
yM
edic
aln
.s.
Ret
rosp
ecti
veH
osp
ital
1,15
528
.8H
isto
ry o
r m
edic
atio
n
Smit
h20
07U
SASu
rgic
alTr
aum
a/su
rger
yPr
osp
ecti
veH
osp
ital
807
16.5
IDD
M +
NID
DM
Steg
enga
2010
Wor
ldw
ide
Med
ical
Sep
sis
Ret
rosp
ecti
ve30
-day
830
22.7
n.s
.
Svir
cevi
c20
09N
eth
erla
nd
sSu
rgic
alC
ard
iac
surg
ery
Ret
rosp
ecti
veH
osp
ital
7,98
919
.8n
.s.
Swob
oda
2008
Ger
man
ySu
rgic
alSe
psi
s/se
pti
c sh
ock
Ret
rosp
ecti
veIC
U70
25.7
n.s
.
Thie
lman
n20
06G
erm
any
Surg
ical
Car
dia
c su
rger
yPr
osp
ecti
veH
osp
ital
254
31.1
n.s
.
Tsai
2008
Taiw
anSu
rgic
alE
CM
O +
acu
te d
ialy
sis
Pros
pec
tive
Hos
pit
al10
417
.3n
.s.
Tsen
g (a
)20
09Ta
iwan
Med
ical
Pan
crea
titi
s +
RF
Ret
rosp
ecti
veIC
U60
48.3
n.s
.
Tsen
g (b
)20
09U
SAM
edic
alPn
eum
onia
Ret
rosp
ecti
veH
osp
ital
406
26.6
n.s
.
Vale
nte
2008
Ital
yM
edic
alST
EM
IPr
osp
ecti
veIC
U31
9.7
n.s
.
Van
dij
ck20
08B
elgi
um
Mix
edB
SIR
etro
spec
tive
Hos
pit
al15
217
.7n
.s.
Var
ela
2005
Spai
nM
ixed
MO
FPr
osp
ecti
veIC
U24
16.7
n.s
.
Vasi
le20
09U
SAM
edic
alG
I bl
eed
ing
Ret
rosp
ecti
veH
osp
ital
, 30d
754
24.3
n.s
.
188
Au
tho
rYe
arC
ou
ntr
yIC
U t
ype
Pop
ula
tio
nD
esig
nM
ort
alit
yN
pat
ien
ts%
DM
Dia
bet
es t
ype
Vin
cen
t20
08Eu
rop
eM
ixed
Blo
od t
ran
sfu
sion
Pros
pec
tive
30d
3,14
77.
2ID
DM
Vin
cen
t20
09W
orld
wid
eM
ixed
Infe
ctio
nPr
osp
ecti
veH
osp
ital
13,7
969.
7ID
DM
Wan
g20
09Ta
iwan
Med
ical
Sep
sis/
sep
tic
shoc
kPr
osp
ecti
veIC
U86
30.2
n.s
.
Was
ir20
06In
dia
Surg
ical
Car
dia
c su
rger
yPr
osp
ecti
veH
osp
ital
1,00
033
.0n
.s.
Wh
itco
mb
2005
USA
Mix
edn
.s.
Ret
rosp
ecti
veIC
U2,
713
21.2
n.s
.
Wil
son
2005
Au
stra
lia
Med
ical
CA
PR
etro
spec
tive
ICU
9614
.6n
.s.
Wu
2008
Ch
ina
Trau
ma
AR
DS
Ret
rosp
ecti
veIC
U44
0u
nk
n.s
.
Yava
s20
09Tu
rkey
Surg
ical
Car
dia
c su
rger
y +
trac
heo
stom
yR
etro
spec
tive
Hos
pit
al20
543
.9n
.s.
Ylip
alos
aari
2006
Fin
lan
dM
ixed
Infe
ctio
n +
LO
S >4
8hPr
osp
ecti
veH
osp
ital
335
20.9
n.s
.
Yosh
imot
o20
05Ja
pan
Med
ical
CA
PR
etro
spec
tive
Hos
pit
al72
12.5
n.s
.
Yu20
09C
hin
aM
edic
alSe
pti
c sh
ock
Pros
pec
tive
ICU
4032
.5n
.s.
Zan
gril
lo20
06It
aly
Surg
ical
Car
dia
c su
rger
yPr
osp
ecti
veH
osp
ital
6,42
311
.4In
suli
n o
r O
AD
Zar
agoz
a20
09Sp
ain
Mix
edC
and
ida
infe
ctio
nPr
osp
ecti
veIC
U53
22.6
n.s
.
Zar
e20
07U
SASu
rgic
alN
on-e
mer
gen
t m
ajor
su
rger
yR
etro
spec
tive
30d
14,2
71u
nk
Insu
lin
or
OA
D
Zar
rill
i20
07It
aly
Mix
edA
. bau
man
ii i
nfe
ctio
nPr
osp
ecti
veH
osp
ital
7436
.5n
.a.
Zh
ao20
08C
hin
aTr
aum
aA
RD
SR
etro
spec
tive
ICU
247
un
kn
.s.
Zil
berb
erg
2009
USA
Mix
edC
. dif
fici
le i
nfe
ctio
nR
etro
spec
tive
30d
148
33.8
n.s
.
AA
A, a
cute
abd
omin
al a
neu
rysm
; A
KI,
acu
te k
idn
ey i
nju
ry;
ALI
, acu
te l
un
g in
jury
; A
MI,
acu
te m
yoca
rdia
l in
farc
tion
; A
PAC
HE
, acu
te p
hys
iolo
gy a
nd
ch
ron
ic h
ealt
h
eval
uat
ion
(sco
re);
APG
, ad
mis
sion
pla
sma
glu
cose
; A
RD
S, a
cute
res
pir
ator
y d
istr
ess
syn
dro
me;
AR
F, a
cute
ren
al f
ailu
re;
BSI
, blo
odst
ream
in
fect
ion
; C
AB
G, c
oron
ary
arte
ry b
ypas
s gr
afti
ng;
CA
P, c
omm
un
ity
acqu
ired
pn
eum
onia
; DK
A, d
iabe
tic
keto
acid
osis
; DM
, dia
bete
s m
elli
tus;
EC
MO
, ext
ra c
orp
oral
mem
bran
e ox
ygen
atio
n; E
SRF,
en
d s
tage
ren
al f
ailu
re; F
PG, f
asti
ng
pla
sma
glu
cose
; GI,
gas
tro-
inte
stin
al; H
AP,
hos
pit
al a
cqu
ired
pn
eum
onia
; HO
NC
, hyp
ergl
ycae
mic
hyp
eros
mol
ar n
on-k
etot
ic c
oma;
ID
DM
, in
suli
n d
epen
den
t d
iab
etes
mel
litu
s; I
CD
-9/1
0, i
nte
rnat
ion
al c
lass
ifica
tion
of
dis
ease
s, 9
th o
r 10
th r
evis
ion
; IC
U,
inte
nsi
ve c
are
un
it;
LOS,
len
gth
of
stay
; M
V,
mec
han
ical
ven
tila
tion
; (M
)OF,
(mu
lti)
org
an f
ailu
re;
NID
DM
, non
in
suli
n d
epen
den
t d
iabe
tes
mel
litu
s; n
.s.,
not
sp
ecifi
ed;
NZ
L, N
ew Z
eala
nd
; O
AD
, ora
l an
tid
iabe
tic
dru
gs; R
PG, r
and
om p
lasm
a gl
uco
se; R
F, r
esp
irat
ory
fail
ure
; STE
MI,
ST-
segm
ent
elev
ated
myo
card
ial i
nfa
rcti
on; T
BC
, tu
berc
ulo
sis;
UK
, Un
ited
Kin
gdom
; un
k, u
nkn
own
; U
SA, U
nit
ed S
tate
s of
Am
eric
a; V
AP,
ven
tila
tor
asso
ciat
ed p
neu
mon
ia
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
189
Au
tho
rYe
arC
ou
ntr
yIC
U t
ype
Pop
ula
tio
nD
esig
nM
ort
alit
yN
pat
ien
ts%
DM
Dia
bet
es t
ype
Vin
cen
t20
08Eu
rop
eM
ixed
Blo
od t
ran
sfu
sion
Pros
pec
tive
30d
3,14
77.
2ID
DM
Vin
cen
t20
09W
orld
wid
eM
ixed
Infe
ctio
nPr
osp
ecti
veH
osp
ital
13,7
969.
7ID
DM
Wan
g20
09Ta
iwan
Med
ical
Sep
sis/
sep
tic
shoc
kPr
osp
ecti
veIC
U86
30.2
n.s
.
Was
ir20
06In
dia
Surg
ical
Car
dia
c su
rger
yPr
osp
ecti
veH
osp
ital
1,00
033
.0n
.s.
Wh
itco
mb
2005
USA
Mix
edn
.s.
Ret
rosp
ecti
veIC
U2,
713
21.2
n.s
.
Wil
son
2005
Au
stra
lia
Med
ical
CA
PR
etro
spec
tive
ICU
9614
.6n
.s.
Wu
2008
Ch
ina
Trau
ma
AR
DS
Ret
rosp
ecti
veIC
U44
0u
nk
n.s
.
Yava
s20
09Tu
rkey
Surg
ical
Car
dia
c su
rger
y +
trac
heo
stom
yR
etro
spec
tive
Hos
pit
al20
543
.9n
.s.
Ylip
alos
aari
2006
Fin
lan
dM
ixed
Infe
ctio
n +
LO
S >4
8hPr
osp
ecti
veH
osp
ital
335
20.9
n.s
.
Yosh
imot
o20
05Ja
pan
Med
ical
CA
PR
etro
spec
tive
Hos
pit
al72
12.5
n.s
.
Yu20
09C
hin
aM
edic
alSe
pti
c sh
ock
Pros
pec
tive
ICU
4032
.5n
.s.
Zan
gril
lo20
06It
aly
Surg
ical
Car
dia
c su
rger
yPr
osp
ecti
veH
osp
ital
6,42
311
.4In
suli
n o
r O
AD
Zar
agoz
a20
09Sp
ain
Mix
edC
and
ida
infe
ctio
nPr
osp
ecti
veIC
U53
22.6
n.s
.
Zar
e20
07U
SASu
rgic
alN
on-e
mer
gen
t m
ajor
su
rger
yR
etro
spec
tive
30d
14,2
71u
nk
Insu
lin
or
OA
D
Zar
rill
i20
07It
aly
Mix
edA
. bau
man
ii i
nfe
ctio
nPr
osp
ecti
veH
osp
ital
7436
.5n
.a.
Zh
ao20
08C
hin
aTr
aum
aA
RD
SR
etro
spec
tive
ICU
247
un
kn
.s.
Zil
berb
erg
2009
USA
Mix
edC
. dif
fici
le i
nfe
ctio
nR
etro
spec
tive
30d
148
33.8
n.s
.
AA
A, a
cute
abd
omin
al a
neu
rysm
; A
KI,
acu
te k
idn
ey i
nju
ry;
ALI
, acu
te l
un
g in
jury
; A
MI,
acu
te m
yoca
rdia
l in
farc
tion
; A
PAC
HE
, acu
te p
hys
iolo
gy a
nd
ch
ron
ic h
ealt
h
eval
uat
ion
(sco
re);
APG
, ad
mis
sion
pla
sma
glu
cose
; A
RD
S, a
cute
res
pir
ator
y d
istr
ess
syn
dro
me;
AR
F, a
cute
ren
al f
ailu
re;
BSI
, blo
odst
ream
in
fect
ion
; C
AB
G, c
oron
ary
arte
ry b
ypas
s gr
afti
ng;
CA
P, c
omm
un
ity
acqu
ired
pn
eum
onia
; DK
A, d
iabe
tic
keto
acid
osis
; DM
, dia
bete
s m
elli
tus;
EC
MO
, ext
ra c
orp
oral
mem
bran
e ox
ygen
atio
n; E
SRF,
en
d s
tage
ren
al f
ailu
re; F
PG, f
asti
ng
pla
sma
glu
cose
; GI,
gas
tro-
inte
stin
al; H
AP,
hos
pit
al a
cqu
ired
pn
eum
onia
; HO
NC
, hyp
ergl
ycae
mic
hyp
eros
mol
ar n
on-k
etot
ic c
oma;
ID
DM
, in
suli
n d
epen
den
t d
iab
etes
mel
litu
s; I
CD
-9/1
0, i
nte
rnat
ion
al c
lass
ifica
tion
of
dis
ease
s, 9
th o
r 10
th r
evis
ion
; IC
U,
inte
nsi
ve c
are
un
it;
LOS,
len
gth
of
stay
; M
V,
mec
han
ical
ven
tila
tion
; (M
)OF,
(mu
lti)
org
an f
ailu
re;
NID
DM
, non
in
suli
n d
epen
den
t d
iabe
tes
mel
litu
s; n
.s.,
not
sp
ecifi
ed;
NZ
L, N
ew Z
eala
nd
; O
AD
, ora
l an
tid
iabe
tic
dru
gs; R
PG, r
and
om p
lasm
a gl
uco
se; R
F, r
esp
irat
ory
fail
ure
; STE
MI,
ST-
segm
ent
elev
ated
myo
card
ial i
nfa
rcti
on; T
BC
, tu
berc
ulo
sis;
UK
, Un
ited
Kin
gdom
; un
k, u
nkn
own
; U
SA, U
nit
ed S
tate
s of
Am
eric
a; V
AP,
ven
tila
tor
asso
ciat
ed p
neu
mon
ia
References1. Abizaid A, Costa MA, Centemero M, et al (2001) Clinical and economic impact of diabetes mellitus on
percutaneous and surgical treatment of multivessel coronary disease patients: insights from the Arterial Revascularization Therapy Study (ARTS) trial. Circulation 104: 533-538
2. Banning AP, Westaby S, Morice MC, et al (2010) Diabetic and nondiabetic patients with left main and/or 3-vessel coronary artery disease: comparison of outcomes with cardiac surgery and paclitaxel-eluting stents. J Am Coll Cardiol 55: 1067-1075
3. Laupland KB, Gregson DB, Zygun DA, Doig CJ, Mortis G, Church DL (2004) Severe bloodstream infections: a population-based assessment. Crit Care Med 32: 992-997
4. Michalia M, Kompoti M, Koutsikou A, et al (2009) Diabetes mellitus is an independent risk factor for ICU-acquired bloodstream infections. Intensive Care Med 35: 448-454
5. Slynkova K, Mannino DM, Martin GS, Morehead RS, Doherty DE (2006) The role of body mass index and diabetes in the development of acute organ failure and subsequent mortality in an observational cohort. Crit Care 10: R137
6. Alexiewicz JM, Kumar D, Smogorzewski M, Klin M, Massry SG (1995) Polymorphonuclear leukocytes in non-insulin-dependent diabetes mellitus: abnormalities in metabolism and function. Ann Intern Med 123: 919-924
7. Marhoffer W, Stein M, Maeser E, Federlin K (1992) Impairment of polymorphonuclear leukocyte function and metabolic control of diabetes. Diabetes Care 15: 256-260
8. Stroup DF, Berlin JA, Morton SC, et al (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 283: 2008-2012
9. Siegelaar SE, Hoekstra JB, DeVries JH (2011) Special considerations for the diabetic patient in the ICU; targets for treatment and risks of hypoglycaemia. Best practice and research: clinical endocrinology and metabolism, in press
10. Kao HK, Chang KP, Ching WC, Tsao CK, Cheng MH, Wei FC (2010) Postoperative morbidity and mortality of head and neck cancers in patients with liver cirrhosis undergoing surgical resection followed by microsurgical free tissue transfer. Ann Surg Oncol 17: 536-543
11. Ligtenberg JJ, Meijering S, Stienstra Y, et al (2006) Mean glucose level is not an independent risk factor for mortality in mixed ICU patients. Intensive Care Med 32: 435-438
12. Wahl WL, Taddonio M, Maggio PM, Arbabi S, Hemmila MR (2008) Mean glucose values predict trauma patient mortality. J Trauma 65: 42-47
13. Andrikos E, Tseke P, Balafa O, et al (2009) Epidemiology of acute renal failure in ICUs: a multi-center prospective study. Blood Purif 28: 239-244
14. Auburtin M, Wolff M, Charpentier J, et al (2006) Detrimental role of delayed antibiotic administration and penicillin-nonsusceptible strains in adult intensive care unit patients with pneumococcal meningitis: the PNEUMOREA prospective multicenter study. Crit Care Med 34: 2758-2765
15. Barlage S, Gnewuch C, Liebisch G, et al (2009) Changes in HDL-associated apolipoproteins relate to mortality in human sepsis and correlate to monocyte and platelet activation. Intensive Care Med 35: 1877-1885
16. Bodi M, Rodriguez A, Sole-Violan J, et al (2005) Antibiotic prescription for community-acquired pneumonia in the intensive care unit: impact of adherence to Infectious Diseases Society of America guidelines on survival. Clin Infect Dis 41: 1709-1716
17. Britt RC, Devine A, Swallen KC, et al (2006) Corticosteroid use in the intensive care unit: at what cost? Arch Surg 141: 145-149
18. Chan KP, Low JG, Raghuram J, Fook-Chong SM, Kurup A (2005) Clinical characteristics and outcome of severe melioidosis requiring intensive care. Chest 128: 3674-3678
19. Chen W, Chen CH, Chiu KL, et al (2008) Clinical outcome and prognostic factors of patients with pyogenic liver abscess requiring intensive care. Crit Care Med 36: 1184-1188
20. Degoricija V, Sharma M, Legac A, Gradiser M, Sefer S, Vucicevic Z (2006) Survival analysis of 314 episodes of sepsis in medical intensive care unit in university hospital: impact of intensive care unit performance and antimicrobial therapy. Croat Med J 47: 385-397
21. Egi M, Bellomo R, Stachowski E, French CJ, Hart G (2006) Variability of blood glucose concentration and short-term mortality in critically ill patients. Anesthesiology 105: 244-252
22. Erbes R, Oettel K, Raffenberg M, Mauch H, Schmidt-Ioanas M, Lode H (2006) Characteristics and outcome of patients with active pulmonary tuberculosis requiring intensive care. Eur Respir J 27: 1223-1228
23. Fortaleza CR, Melo EC, Fortaleza CM (2009) Nasopharyngeal colonization with methicillin-resistant
190
staphylococcus aureus and mortality among patients in an intensive care unit. Rev Lat Am Enfermagem 17: 677-682
24. Georges H, Chiche A, Alfandari S, Devos P, Boussekey N, Leroy O (2009) Adult community-acquired bacterial meningitis requiring ICU admission: epidemiological data, prognosis factors and adherence to IDSA guidelines. Eur J Clin Microbiol Infect Dis 28: 1317-1325
25. Gong MN, Thompson BT, Williams PL, et al (2006) Interleukin-10 polymorphism in position -1082 and acute respiratory distress syndrome. Eur Respir J 27: 674-681
26. Graham BB, Keniston A, Gajic O, Trillo Alvarez CA, Medvedev S, Douglas IS (2010) Diabetes mellitus does not adversely affect outcomes from a critical illness. Crit Care Med 38: 16-24
27. Kermarrec N, Marmuse JP, Faivre J, et al (2008) High mortality rate for patients requiring intensive care after surgical revision following bariatric surgery. Obes Surg 18: 171-178
28. Labelle AJ, Micek ST, Roubinian N, Kollef MH (2008) Treatment-related risk factors for hospital mortality in Candida bloodstream infections. Crit Care Med 36: 2967-2972
29. Lan MY, Wu SJ, Chang YY, Chen WH, Lai SL, Liu JS (2006) Neurologic and non-neurologic predictors of mortality in ischemic stroke patients admitted to the intensive care unit. J Formos Med Assoc 105: 653-658
30. Lemant J, Boisson V, Winer A, et al (2008) Serious acute chikungunya virus infection requiring intensive care during the Reunion Island outbreak in 2005-2006. Crit Care Med 36: 2536-2541
31. Leroy O, Gangneux JP, Montravers P, et al (2009) Epidemiology, management, and risk factors for death of invasive Candida infections in critical care: a multicenter, prospective, observational study in France (2005-2006). Crit Care Med 37: 1612-1618
32. Lorente L, Martin MM, Labarta L, et al (2009) Matrix metalloproteinase-9, -10, and tissue inhibitor of matrix metalloproteinases-1 blood levels as biomarkers of severity and mortality in sepsis. Crit Care 13: R158
33. Lundelin K, Vigil L, Bua S, Gomez-Mestre I, Honrubia T, Varela M (2010) Differences in complexity of glycemic profile in survivors and nonsurvivors in an intensive care unit: a pilot study. Crit Care Med 38: 849-854
34. Markogiannakis H, Pachylaki N, Samara E, et al (2009) Infections in a surgical intensive care unit of a university hospital in Greece. Int J Infect Dis 13: 145-153
35. Martin-Loeches I, Lisboa T, Rodriguez A, et al (2010) Combination antibiotic therapy with macrolides improves survival in intubated patients with community-acquired pneumonia. Intensive Care Medicine 36: 612-620
36. Mascia L, Sakr Y, Pasero D, Payen D, Reinhart K, Vincent J-L (2008) Extracranial complications in patients with acute brain injury: A post-hoc analysis of the SOAP study. Intensive Care Medicine 34: 720-727
37. Murthy SC, Arroliga AC, Walts PA, et al (2007) Ventilatory dependency after cardiovascular surgery. J Thorac Cardiovasc Surg 134: 484-490
38. Nasraway SA, Jr., Albert M, Donnelly AM, Ruthazer R, Shikora SA, Saltzman E (2006) Morbid obesity is an independent determinant of death among surgical critically ill patients. Crit Care Med 34: 964-970
39. Nseir S, Di PC, Cavestri B, et al (2006) Multiple-drug-resistant bacteria in patients with severe acute exacerbation of chronic obstructive pulmonary disease: Prevalence, risk factors, and outcome. Crit Care Med 34: 2959-2966
40. Ong AW, Omert LA, Vido D, et al (2009) Characteristics and outcomes of trauma patients with ICU lengths of stay 30 days and greater: a seven-year retrospective study. Crit Care 13: R154
41. Patel GP, Simon D, Scheetz M, Crank CW, Lodise T, Patel N (2009) The effect of time to antifungal therapy on mortality in Candidemia associated septic shock. Am J Ther 16: 508-511
42. Pratikaki M, Platsouka E, Sotiropoulou C, et al (2009) Risk factors for and influence of bloodstream infections on mortality: a 1-year prospective study in a Greek intensive-care unit. Epidemiol Infect 137: 727-735
43. Rammaert B, Verdier N, Cavestri B, Nseir S (2009) Procalcitonin as a prognostic factor in severe acute exacerbation of chronic obstructive pulmonary disease. Respirology 14: 969-974
44. Reintam A, Parm P, Kitus R, Kern H, Starkopf J (2008) Primary and secondary intra-abdominal hypertension--different impact on ICU outcome. Intensive Care Med 34: 1624-1631
45. Rhodes A, Wort SJ, Thomas H, Collinson P, Bennett ED (2006) Plasma DNA concentration as a predictor of mortality and sepsis in critically ill patients. Crit Care 10: R60
46. Riachy M, Sfeir F, Sleilaty G, et al (2008) Prediction of the survival and functional ability of severe stroke patients after ICU therapeutic intervention. BMC Neurol 8: 24
47. Rodriguez A, Lisboa T, Blot S, et al (2009) Mortality in ICU patients with bacterial community-acquired pneumonia: when antibiotics are not enough. Intensive Care Med 35: 430-438
48. Ruiz-Bailen M, Brea-Salvago JF, de Hoyos EA, et al (2005) Post-thrombolysis intracerebral hemorrhage: data from the Spanish Register ARIAM. Crit Care Med 33: 1829-1838
49. Sakr Y, Reinhart K, Vincent JL, et al (2006) Does dopamine administration in shock influence outcome? Results of the Sepsis Occurrence in Acutely Ill Patients (SOAP) Study. Crit Care Med 34: 589-597
50. Santini M, Kutlesa M, Pangercic A, Barsic B (2010) The importance of pathogens in sepsis: Staphylococcus
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
191
aureus story. Scand J Infect Dis 42: 172-17651. Senthuran S, Bandeshe H, Ranganathan D, Boots R (2008) Outcomes for dialysis patients with end-stage
renal failure admitted to an intensive care unit or high dependency unit. Med J Aust 188: 292-29552. Swoboda S, Ober M, Hainer C, et al (2008) Tigecycline for the treatment of patients with severe sepsis or
septic shock: a drug use evaluation in a surgical intensive care unit. J Antimicrob Chemother 61: 729-73353. Tseng CC, Fang WF, Chung YH, Wang YH, Douglas IS, Lin MC (2009) Clinical outcomes in patients with ICU-
related pancreatitis. World J Gastroenterol 15: 4938-494454. Valente S, Lazzeri C, Saletti E, Chiostri M, Gensini GF (2008) Primary percutaneous coronary intervention in
comatose survivors of cardiac arrest with ST-elevation acute myocardial infarction: a single-center experience in Florence. J Cardiovasc Med 9: 1083-1087
55. Varela M, Calvo M, Chana M, Gomez-Mestre I, Asensio R, Galdos P (2005) Clinical implications of temperature curve complexity in critically ill patients. Crit Care Med 33: 2764-2771
56. Wang JL, Chin CS, Chang MC, et al (2009) Key process indicators of mortality in the implementation of protocol-driven therapy for severe sepsis. J Formos Med Assoc 108: 778-787
57. Whitcomb BW, Pradhan EK, Pittas AG, Roghmann MC, Perencevich EN (2005) Impact of admission hyperglycemia on hospital mortality in various intensive care unit populations. Crit Care Med 33: 2772-2777
58. Wilson PA, Ferguson J (2005) Severe community-acquired pneumonia: an Australian perspective. Intern Med J 35: 699-705
59. Wu J, Sheng L, Ma Y, et al (2008) The analysis of risk factors of impacting mortality rate in severe multiple trauma patients with posttraumatic acute respiratory distress syndrome. Am J Emerg Med 26: 419-424
60. Yu TJ, Liu YC, Yu CC, Tseng JC, Hua CC, Wu HP (2009) Comparing hydrocortisone and methylprednisolone in patients with septic shock. Adv Ther 26: 728-735
61. Zaragoza R, Peman J, Quindos G, et al (2009) Clinical significance of the detection of Candida albicans germ tube-specific antibodies in critically ill patients. Clin Microbiol Infect 15: 592-595
62. Zhao XG, Wu JS, He XD, et al (2008) Risk factors of mortality in road traffic injury patients with acute respiratory distress syndrome. Chin Med J 121: 968-972
63. An Y, Xiao YB, Zhong QJ (2007) Open-heart surgery in patients with liver cirrhosis. Eur J Cardiothorac Surg 31: 1094-1098
64. Angstwurm MW, Gaertner R, Schopohl J (2005) Outcome in elderly patients with severe infection is influenced by sex hormones but not gender. Crit Care Med 33: 2786-2793
65. Babuin L, Vasile VC, Rio Perez JA, et al (2008) Elevated cardiac troponin is an independent risk factor for short- and long-term mortality in medical intensive care unit patients. Crit Care Med 36: 759-765
66. Bitto A, Barone M, David A, et al (2010) High mobility group box-1 expression correlates with poor outcome in lung injury patients. Pharmacol Res 61: 116-120
67. Bochicchio GV, Joshi M, Bochicchio K, Nehman S, Tracy JK, Scalea TM (2006) Impact of obesity in the critically ill trauma patient: a prospective study. J Am Coll Surg 203: 533-538
68. Boyer A, Vargas F, Coste F, et al (2009) Influence of surgical treatment timing on mortality from necrotizing soft tissue infections requiring intensive care management. Intensive Care Med 35: 847-853
69. Cagatay AA, Ozcan PE, Gulec L, et al (2007) Risk factors for mortality of nosocomial bacteraemia in intensive care units. Med Princ Pract 16: 187-192
70. Chen YC, Tian YC, Liu NJ, et al (2006) Prospective cohort study comparing sequential organ failure assessment and acute physiology, age, chronic health evaluation III scoring systems for hospital mortality prediction in critically ill cirrhotic patients. Int J Clin Pract 60: 160-166
71. Chen YC, Jenq CC, Tian YC, et al (2009) Rifle classification for predicting in-hospital mortality in critically ill sepsis patients. Shock 31: 139-145
72. Cheng B, Xie G, Yao S, et al (2007) Epidemiology of severe sepsis in critically ill surgical patients in ten university hospitals in China. Critical Care Medicine 35: 2538-2546
73. Connelly SM, Trinh JV, Johnson MD, et al (2009) Mortality and time to extubation in severe hospital-acquired pneumonia. Am J Infect Control 37: 143-149
74. Cooke CR, Kahn JM, Caldwell E, et al (2008) Predictors of hospital mortality in a population-based cohort of patients with acute lung injury. Crit Care Med 36: 1412-1420
75. Damas P, Ledoux D, Nys M, et al (2008) Intensive care unit acquired infection and organ failure. Intensive Care Med 34: 856-864
76. Diez C, Mohr P, Kuss O, Osten B, Silber RE, Hofmann HS (2009) Impact of preoperative renal dysfunction on in-hospital mortality after solitary valve and combined valve and coronary procedures. Ann Thorac Surg 87: 731-736
77. Egi M, Bellomo R, Stachowski E, et al (2008) Blood glucose concentration and outcome of critical illness: the impact of diabetes. Crit Care Med 36: 2249-2255
192
78. Engel AM, McDonough S, Smith JM (2009) Does an obese body mass index affect hospital outcomes after coronary artery bypass graft surgery? Ann Thorac Surg 88: 1793-1800
79. Fang JT, Tsai MH, Tian YC, et al (2008) Outcome predictors and new score of critically ill cirrhotic patients with acute renal failure. Nephrol Dial Transplant 23: 1961-1969
80. Freire AX, Bridges L, Umpierrez GE, Kuhl D, Kitabchi AE (2005) Admission hyperglycemia and other risk factors as predictors of hospital mortality in a medical ICU population. Chest 128: 3109-3116
81. Friedrich JO, Wilson G, Chant C (2006) Long-term outcomes and clinical predictors of hospital mortality in very long stay intensive care unit patients: a cohort study. Crit Care 10: R59
82. Frioud A, Comte-Perret S, Nguyen S, Berger MM, Ruchat P, Ruiz J (2010) Blood glucose level on postoperative day 1 is predictive of adverse outcomes after cardiovascular surgery. Diabetes Metab 36: 36-42
83. Gale SC, Sicoutris C, Reilly PM, Schwab CW, Gracias VH (2007) Poor glycemic control is associated with increased mortality in critically ill trauma patients. Am Surg 73: 454-460
84. Garnacho-Montero J, Ortiz-Leyba C, Fernandez-Hinojosa E, et al (2005) Acinetobacter baumannii ventilator-associated pneumonia: Epidemiological and clinical findings. Intensive Care Medicine 31: 649-655
85. Garnacho-Montero J, Sa-Borges M, Sole-Violan J, et al (2007) Optimal management therapy for Pseudomonas aeruginosa ventilator-associated pneumonia: an observational, multicenter study comparing monotherapy with combination antibiotic therapy. Crit Care Med 35: 1888-1895
86. Garnacho-Montero J, Aldabo-Pallas T, Palomar-Martinez M, et al (2008) Risk factors and prognosis of catheter-related bloodstream infection in critically ill patients: a multicenter study. Intensive Care Med 34: 2185-2193
87. Griniatsos J, Petrou A, Pappas P, et al (2008) Percutaneous cholecystostomy without interval cholecystectomy as definitive treatment of acute cholecystitis in elderly and critically ill patients. South Med J 101: 586-590
88. Hays A, Diringer MN (2006) Elevated troponin levels are associated with higher mortality following intracerebral hemorrhage. Neurology 66: 1330-1334
89. Henckaerts L, Nielsen KR, Steffensen R, et al (2009) Polymorphisms in innate immunity genes predispose to bacteremia and death in the medical intensive care unit. Crit Care Med 37: 192-193
90. Holley A, Dulhunty J, Blot S, et al (2009) Temporal trends, risk factors and outcomes in albicans and non-albicans candidaemia: an international epidemiological study in four multidisciplinary intensive care units. Int J Antimicrob Agents 33: 554-557
91. Hsu CL, Chang CH, Wong KN, Chen KY, Yu CJ, Yang PC (2009) Management of severe community-acquired septic meningitis in adults: from emergency department to intensive care unit. J Formos Med Assoc 108: 112-118
92. Iribarren-Diarasarri S, Aizpuru-Barandiaran F, Munoz-Martinez T, et al (2009) Health-related quality of life as a prognostic factor of survival in critically ill patients. Intensive Care Med 35: 833-839
93. Jamal W, Salama M, Dehrab N, Al HG, Shahin M, Rotimi VO (2009) Role of tigecycline in the control of a carbapenem-resistant Acinetobacter baumannii outbreak in an intensive care unit. J Hosp Infect 72: 234-242
94. Jenq CC, Tsai MH, Tian YC, et al (2007) RIFLE classification can predict short-term prognosis in critically ill cirrhotic patients. Intensive Care Med 33: 1921-1930
95. Kao LS, Todd SR, Moore FA (2006) The impact of diabetes on outcome in traumatically injured patients: an analysis of the National Trauma Data Bank. Am J Surg 192: 710-714
96. Krinsley JS, Grover A (2007) Severe hypoglycemia in critically ill patients: risk factors and outcomes. Crit Care Med 35: 2262-2267
97. Lee JH, Ryu YJ, Chun EM, Chang JH (2007) Outcomes and prognostic factors for severe community-acquired pneumonia that requires mechanical ventilation. Korean J Intern Med 22: 157-163
98. Liao KM, Chen CW, Hsiue TR, Lin WC (2009) Timing of acute respiratory distress syndrome onset is related to patient outcome. J Formos Med Assoc 108: 694-703
99. Lin JL, Lin-Tan DT, Chu PH, et al (2009) Cadmium excretion predicting hospital mortality and illness severity of critically ill medical patients. Crit Care Med 37: 957-962
100. Lin SM, Wang TY, Liu WT, et al (2009) Predictive factors for mortality among non-HIV-infected patients with pulmonary tuberculosis and respiratory failure. Int J Tuberc Lung Dis 13: 335-340
101. Martin GS, Mannino DM, Moss M (2006) The effect of age on the development and outcome of adult sepsis. Crit Care Med 34: 15-21
102. Mitchell I, Finfer S, Bellomo R, Higlett T (2006) Management of blood glucose in the critically ill in Australia and New Zealand: a practice survey and inception cohort study. Intensive Care Med 32: 867-874
103. Mowery NT, Dortch MJ, Dossett LA, et al (2009) Insulin resistance despite tight glucose control is associated with mortality in critically ill surgical patients. J Intensive Care Med 24: 242-251
104. Mowery NT, Gunter OL, Guillamondegui O, et al (2009) Stress insulin resistance is a marker for mortality in traumatic brain injury. J Trauma 66: 145-151
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
193
105. Murphy CV, Schramm GE, Doherty JA, et al (2009) The importance of fluid management in acute lung injury secondary to septic shock. Chest 136: 102-109
106. Nair S, Vanatta JM, Arteh J, Eason JD (2009) Effects of obesity, diabetes, and prior abdominal surgery on resource utilization in liver transplantation: a single-center study. Liver Transpl 15: 1519-1524
107. Pieracci FM, Hydo L, Pomp A, Eachempati SR, Shou J, Barie PS (2008) The relationship between body mass index and postoperative mortality from critical illness. Obes Surg 18: 501-507
108. Quach S, Hennessy DA, Faris P, Fong A, Quan H, Doig C (2009) A comparison between the APACHE II and Charlson Index Score for predicting hospital mortality in critically ill patients. BMC Health Serv Res 9: 129
109. Rady MY, Johnson DJ, Patel B, Larson J, Helmers R (2005) Cortisol levels and corticosteroid administration fail to predict mortality in critical illness: the confounding effects of organ dysfunction and sex. Arch Surg 140: 661-668
110. Ranucci M, Pazzaglia A, Bianchini C, Bozzetti G, Isgro G (2008) Body size, gender, and transfusions as determinants of outcome after coronary operations. Ann Thorac Surg 85: 481-486
111. Rellos K, Falagas ME, Vardakas KZ, Sermaides G, Michalopoulos A (2006) Outcome of critically ill oldest-old patients (aged 90 and older) admitted to the intensive care unit. J Am Geriatr Soc 54: 110-114
112. Rivero-Ayerza M, Scholte Op RW, Lenzen M, et al (2008) New-onset atrial fibrillation is an independent predictor of in-hospital mortality in hospitalized heart failure patients: results of the EuroHeart Failure Survey. Eur Heart J 29: 1618-1624
113. Rosamel P, Cervantes M, Tristan A, et al (2005) Active infectious endocarditis: postoperative outcome. J Cardiothorac Vasc Anesth 19: 435-439
114. Rose AH, Kotze A, Doolan D, Norfolk DR, Bellamy MC (2009) Massive transfusion--evaluation of current clinical practice and outcome in two large teaching hospital trusts in Northern England. Vox Sang 97: 247-253
115. Sailhamer EA, Carson K, Chang Y, et al (2009) Fulminant Clostridium difficile colitis: patterns of care and predictors of mortality. Arch Surg 144: 433-439
116. Sakr Y, Vincent JL, Ruokonen E, et al (2008) Sepsis and organ system failure are major determinants of post-intensive care unit mortality. J Crit Care 23: 475-483
117. Sayed S, Fischer S, Karck M, Hassouna A, Haverich A (2009) Effect of different preoperative patient characteristics on coronary surgery outcome: a comparative study between a developing and a developed country. J Card Surg 24: 275-280
118. Shah KB, Nolan MM, Rao K, et al (2007) The characteristics and prognostic importance of NT-ProBNP concentrations in critically ill patients. Am J Med 120: 1071-1077
119. Silva Junior GB, Daher EF, Mota RM, Menezes FA (2006) Risk factors for death among critically ill patients with acute renal failure. Sao Paulo Med J 124: 257-263
120. Sleiman I, Morandi A, Sabatini T, et al (2008) Hyperglycemia as a predictor of in-hospital mortality in elderly patients without diabetes mellitus admitted to a sub-intensive care unit. J Am Geriatr Soc 56: 1106-1110
121. Smith RL, Chong TW, Hedrick TL, et al (2007) Does body mass index affect infection-related outcomes in the intensive care unit? Surg Infect 8: 581-588
122. Svircevic V, Nierich AP, Moons KG, Brandon Bravo Bruinsma GJ, Kalkman CJ, van DD (2009) Fast-track anesthesia and cardiac surgery: a retrospective cohort study of 7989 patients. Anesth Analg 108: 727-733
123. Thielmann M, Massoudy P, Neuhauser M, et al (2006) Prognostic value of preoperative cardiac troponin I in patients undergoing emergency coronary artery bypass surgery with non-ST-elevation or ST-elevation acute coronary syndromes. Circulation 114: I448-I453
124. Tsai CW, Lin YF, Wu VC, et al (2008) SAPS 3 at dialysis commencement is predictive of hospital mortality in patients supported by extracorporeal membrane oxygenation and acute dialysis. Eur J Cardiothorac Surg 34: 1158-1164
125. Tseng CC, Fang WF, Huang KT, et al (2009) Risk factors for mortality in patients with nosocomial Stenotrophomonas maltophilia pneumonia. Infect Control Hosp Epidemiol 30: 1193-1202
126. Vandijck DM, Oeyen SG, Buyle EM, Claus BO, Blot SI, Decruyenaere JM (2008) Hyperglycaemia upon onset of ICU-acquired bloodstream infection is associated with adverse outcome in a mixed ICU population. Anaesth Intensive Care 36: 25-29
127. Vardakas KZ, Michalopoulos A, Kiriakidou KG, Siampli EP, Samonis G, Falagas ME (2009) Candidaemia: incidence, risk factors, characteristics and outcomes in immunocompetent critically ill patients. Clin Microbiol Infect 15: 289-292
128. Vasile VC, Babuin L, Rio Perez JA, et al (2009) Long-term prognostic significance of elevated cardiac troponin levels in critically ill patients with acute gastrointestinal bleeding. Crit Care Med 37: 140-147
129. Vincent JL, Rello J, Marshall J, et al (2009) International study of the prevalence and outcomes of infection in intensive care units. JAMA 302: 2323-2329
194
130. Wasir H, Mehta Y, Pawar M, et al (2006) Predictors of operative mortality following primary coronary artery bypass surgery. Indian Heart J 58: 144-148
131. Yavas S, Yagar S, Mavioglu L, et al (2009) Tracheostomy: how and when should it be done in cardiovascular surgery ICU? J Card Surg 24: 11-18
132. Ylipalosaari P, Ala-Kokko TI, Laurila J, Ohtonen P, Syrjala H (2006) Intensive care acquired infection is an independent risk factor for hospital mortality: a prospective cohort study. Crit Care 10: R66
133. Yoshimoto A, Nakamura H, Fujimura M, Nakao S (2005) Severe community-acquired pneumonia in an intensive care unit: risk factors for mortality. Intern Med 44: 710-716
134. Zangrillo A, Landoni G, Fumagalli L, et al (2006) Methicillin-resistant Staphylococcus species in a cardiac surgical intensive care unit: a 5-year experience. J Cardiothorac Vasc Anesth 20: 31-37
135. Zarrilli R, Casillo R, Di PA, et al (2007) Molecular epidemiology of a clonal outbreak of multidrug-resistant Acinetobacter baumannii in a university hospital in Italy. Clin Microbiol Infect 13: 481-489
136. Ailawadi G, Swenson BR, Girotti ME, et al (2008) Is mitral valve repair superior to replacement in elderly patients? Ann Thorac Surg 86: 77-85
137. Almog Y, Novack V, Megralishvili R, et al (2006) Plasma level of N terminal pro-brain natriuretic peptide as a prognostic marker in critically ill patients. Anesth Analg 102: 1809-1815
138. Chien JY, Jerng JS, Yu CJ, Yang PC (2005) Low serum level of high-density lipoprotein cholesterol is a poor prognostic factor for severe sepsis. Crit Care Med 33: 1688-1693
139. Connelly KA, Creati L, Lyon W, et al (2007) Early and late results of combined mitral-aortic valve surgery. Heart Lung Circ 16: 410-415
140. Cuthbertson BH, Croal BL, Rae D, et al (2009) N-terminal pro-B-type natriuretic peptide levels and early outcome after cardiac surgery: a prospective cohort study. Br J Anaesth 103: 647-653
141. Djavani K, Wanhainen A, Valtysson J, Bjorck M (2009) Colonic ischaemia and intra-abdominal hypertension following open repair of ruptured abdominal aortic aneurysm. Br J Surg 96: 621-627
142. Dunser MW, Ruokonen E, Pettila V, et al (2009) Association of arterial blood pressure and vasopressor load with septic shock mortality: a post hoc analysis of a multicenter trial. Crit Care 13: R181
143. Ganesh JS, Rogers CA, Banner NR, Bonser RS (2007) Does the method of lung preservation influence outcome after transplantation? An analysis of 681 consecutive procedures. J Thorac Cardiovasc Surg 134: 1313-1321
144. Karimi A, Ahmadi H, Davoodi S, et al (2008) Factors affecting postoperative morbidity and mortality in isolated coronary artery bypass graft surgery. Surg Today 38: 890-898
145. Kenneally C, Rosini JM, Skrupky LP, et al (2007) Analysis of 30-day mortality for clostridium difficile-associated disease in the ICU setting. Chest 132: 418-424
146. Kes VB, Solter VV, Supanc V, Demarin V (2007) Impact of hyperglycemia on ischemic stroke mortality in diabetic and non-diabetic patients. Ann Saudi Med 27: 352-355
147. Musci M, Weng Y, Hubler M, et al (2009) Predictors of early mortality in patients with active infective native or prosthetic aortic root endocarditis undergoing homograft aortic root replacement. Clin Res Cardiol 98: 443-450
148. Patel NV, Woznick AR, Welsh KS, Bendick PJ, Boura JA, Mucci SJ (2009) Predictors of mortality after muscle flap advancement for deep sternal wound infections. Plast Reconstr Surg 123: 132-138
149. Paul M, Bishara J, Levcovich A, et al (2010) Effectiveness and safety of colistin: Prospective comparative cohort study. J Antimicrob Chemother 65: 1019-1027
150. Stegenga ME, Vincent JL, Vail GM, et al (2010) Diabetes does not alter mortality or hemostatic and inflammatory responses in patients with severe sepsis. Crit Care Med 38: 539-545
151. Vincent JL, Sakr Y, Sprung C, Harboe S, Damas P (2008) Are blood transfusions associated with greater mortality rates? Results of the Sepsis Occurrence in Acutely Ill Patients study. Anesthesiology 108: 31-39
152. Zare MM, Itani KM, Schifftner TL, Henderson WG, Khuri SF (2007) Mortality after nonemergent major surgery performed on Friday versus Monday through Wednesday. Ann Surg 246: 866-874
153. Zilberberg MD, Shorr AF, Micek ST, Doherty JA, Kollef MH (2009) Clostridium difficile-associated disease and mortality among the elderly critically ill. Crit Care Med 37: 2583-2589
154. Waller BF, Palumbo PJ, Lie JT, Roberts WC (1980) Status of the coronary arteries at necropsy in diabetes mellitus with onset after age 30 years. Analysis of 229 diabetic patients with and without clinical evidence of coronary heart disease and comparison to 183 control subjects. Am J Med 69: 498-506
155. Woodfield SL, Lundergan CF, Reiner JS, et al (1996) Angiographic findings and outcome in diabetic patients treated with thrombolytic therapy for acute myocardial infarction: the GUSTO-I experience. J Am Coll Cardiol 28: 1661-1669
156. Krinsley JS (2003) Association between hyperglycemia and increased hospital mortality in a heterogeneous population of critically ill patients. Mayo Clin Proc 78: 1471-1478
157. Falciglia M, Freyberg RW, Almenoff PL, D’Alessio DA, Render ML (2009) Hyperglycemia-related mortality
The effect of diabetes on mortality in critically ill patients
Ch
apte
r 12
195
in critically ill patients varies with admission diagnosis. Crit Care Med 37: 3001-3009158. Rady MY, Johnson DJ, Patel BM, Larson JS, Helmers RA (2005) Influence of individual characteristics on
outcome of glycemic control in intensive care unit patients with or without diabetes mellitus. Mayo Clin Proc 80: 1558-1567
Chapter 13
Summary and future considerations
Samenvatting en toekomstperspectief
Authors’ affiliations
List of publications
Dankwoord
Curriculum Vitae
198
Summary and future considerations
This thesis is about the effects and treatment of glucose peaks in chronic and acute
hyperglycaemia. It addresses the question whether it is beneficial to curb these glucose
peaks: must all what goes up come down? In Part I, the consequences of glucose
variability in diabetes were studied with respect to oxidative stress, which is associated
with endothelial damage, and two chronic complications of diabetes, neuropathy and
cardiovascular events. Part II of this thesis discusses hyperglycaemia in critical illness.
First, an optimal glucose target range for critically ill patients was proposed and a
new method to reach this target range, subcutaneous continuous glucose monitoring
(CGM), was tested for accuracy and reliability. Second, the implications of the presence
of diabetes in a critical care setting were investigated with respect to glucose regulation
and mortality.
Part IIn Chapter 2 we give an overview of the available methods to measure glucose variability
and review the evidence for its importance in addition to mean glucose. A large variation
in the number and duration of glucose peaks exists between patients with similar
haemoglobin A1c levels, but to date there is no “gold standard” for quantifying glucose
variability. The standard deviation from the mean seems the most extensively used and
mathematically best validated measure. In in vitro, animal and experimental human
studies, glucose peaks increase oxidative stress. However, the evidence for an independent
effect of glucose variability on oxidative stress and long-term diabetic complications in
type 1 and type 2 diabetes patients is marginal, and possibly limited to poorly regulated
type 2 diabetes patients on oral glucose lowering drugs. Contrary, in the critically ill
glucose variability is indisputably associated with mortality.
To understand the different findings of the effect of glucose variability in different
diabetic populations, we investigated in Chapter 3 at which glucose level the glucose-
dependent effects on vascular homeostasis first occur, and whether this is an on-off
phenomenon with a threshold or a continuous relationship. A stepwise glucose clamp
was performed in healthy humans, stabilizing plasma glucose levels for two hours at
6.0, 8.0, and 10.0 mmol/l successively. The effect of increasing glucose on oxidative
stress, coagulation and fibrinolysis, and the endothelial glycocalyx were investigated.
The results of this study reveal that changes to vascular homeostasis start already at
near normal glucose levels. The increase in oxidative stress was dose-dependent and
coagulation activation showed a threshold already at 6.0 mmol/l. Absence of a threshold
or a threshold at such a low level do not support a role for glucose variability in oxidative
stress and coagulation activation.
Summary and future considerations
Ch
apte
r 13
199
This conclusion was supported by the results described in Chapters 4 and 5. In Chapter
4, the relation between glucose variability, measured by continuous glucose monitoring,
and oxidative stress, measured by 24-hr excretion of 8-iso prostaglandin F2α, was assessed
in 24 well-regulated type 2 diabetes patients on oral glucose lowering drugs. No relevant
relationship was found between glucose variability and oxidative stress. In Chapter 5
a mealtime insulin approach was compared with a basal insulin regimen in a cross-
over study including 40 type 2 diabetes patients regarding glucose regulation and the
effects on oxidative stress. Addition of insulin to the patients’ medication significantly
lowered mean glucose as well as oxidative stress. However, again no relationship was
found between glucose variability and oxidative stress.
Oxidative stress is thought to induce vascular complications, but eventually it is an
indirect marker of disease, not a hard outcome. In Chapter 6 we investigated the effect of
glucose variability on the development of peripheral and autonomic neuropathy. For this
purpose data from the Diabetes Control and Complications Trial (DCCT) were reanalysed.
The DCCT was originally designed to assess the effect of intensive vs. conventional glucose
lowering treatment on the development of microvascular complications and included
1,441 type 1 diabetes patients. The development of neuropathy was strongly associated
with mean glucose but no additional effect of glucose variability was found.
Chapter 7 shows a reanalysis of The Hyperglycaemia and Its Effect After Acute Myocardial
Infarction on Cardiovascular Outcomes in Patients With Type 2 Diabetes Mellitus study
(HEART2D). This randomised controlled trial assessed the effect of a prandial insulin
regimen compared with a basal insulin regimen on future cardiovascular event rates
in type 2 diabetes patients after myocardial infarction, thereby specifically lowering
glucose variability. Overall glycaemic control was found to be similar in the two groups
but no differences in cardiovascular outcomes were seen despite eighteen percent lower
glucose variability in the prandial insulin group.
In conclusion, Part I of this thesis does not support a relationship between glucose
variability and oxidative stress or diabetic complications. Moreover, specifically lowering
glucose variability in type 2 diabetes patients did not result in a reduction in future
cardiovascular event rates. Therefore, there are currently insufficient arguments to
specifically lower glucose variability in patients with diabetes regarding complication risk,
and treatment should continue targeting mean glucose while avoiding hypoglycaemia
as much as possible.
200
Part IIMarked hyperglycaemia has to be avoided in critically ill patients but there is debate
on the optimal glucose target. In Chapter 8 we investigated the relationship between
mean glucose during intensive care unit (ICU) admission and mortality in surgical as
well as medical patients who were treated according to the most recent guidelines. In
both cohorts, mean glucose appeared to be related to ICU mortality by a U-shaped curve,
and a “safe-range” for mean glucose could be defined between approximately 7.0 and 9.0
mmol/l. These results are in line with the data from the NICE-SUGAR (Normoglycaemia
in Intensive Care Evaluation- Survival Using Glucose Algorithm Regulation) trial and
suggest that lowering glucose to normoglycaemia is perhaps doing more harm than good.
Continuous glucose monitoring (CGM) could be a useful tool to achieve more time in the
glucose target range. We report in Chapter 9 a head-to-head comparison investigating the
accuracy and reliability of the Guardian Real-Time (Medtronic Minimed) and the FreeStyle
Navigator (Abbott Diabetes Care) CGM system in 60 cardiac surgery patients admitted to
the ICU after surgery. The FreeStyle Navigator performed significantly better in accuracy
as well as reliability compared to the Guardian Real-Time. Remarkably, accuracy of both
systems was quite good compared to known data for outpatients. These data support the
use of the FreeStyle Navigator in cardiac surgery patients. Whether the use of CGM truly
increases the time spent in the target range and lowers the incidence of hypoglycaemia
should be subject of further study.
We hypothesised that the accuracy of the CGM systems could be influenced by the
microcirculation. In Chapter 10 the microcirculation and its effect on CGM accuracy was
investigated in the same 60 patients after cardiac surgery. Impairment in microcirculatory
parameters was found during the first hours of ICU admission during a median follow-
up of 23 hours, but this impairment was not related to CGM accuracy. A decrease in
peripheral temperature did decrease the accuracy of the two systems, and an increase in
age of the patient as well as an increase in severity of disease influenced the accuracy of
the FreeStyle Navigator in a negative way. Further studies need to assess the influence of
more profound microcirculatory changes on sensor accuracy in more severely ill patients.
Challenging it is when acute meets chronic hyperglycaemia: the critically ill patient
with diabetes. Chapter 11 gives an overview of the current literature on morbidity and
mortality in ICU patients with diabetes with special consideration to glucose regulation,
insulin treatment and hypoglycaemia. Diabetes is considered to be a risk factor for the
development of complications while admitted in the ICU but this does not seem to
translate directly into higher mortality. The relation between diabetes and mortality is
further investigated in Chapter 12. Hyperglycaemia is common in critically ill patients
with diabetes and associated with mortality when above 11.1 mmol/l, but there is
Summary and future considerations
Ch
apte
r 13
201
discussion about the detrimental effect of hyperglycaemia lower than 11.1 mmol/l. Also,
intensive insulin therapy seems not beneficial as compared with moderate glycaemic
control. Interestingly, at any given level of hyperglycaemia mortality rates are higher in
non-diabetic patients compared with diabetic patients, but in the lower glucose range
it is the other way round. Patients with diabetes are also vulnerable for developing
hypoglycaemic events which are strongly associated with mortality. We recommend to
maintain glucose levels between 7.5 and 10.0 mmol/l in critically ill patients with diabetes
while avoiding hypoglycaemia as well as extreme hyperglycaemia.
Finally, in Chapter 12 the results of a systematic review and meta-analysis are shown
assessing the effect of diabetes on mortality in different ICU types. In total, 141 studies
were included containing over 12.4 million patients, including 2.7 million (21.7%)
deaths and 2.3 million (18.6%) patients with diabetes. The meta-analysis showed that
patients with diabetes who are admitted at the medical, mixed and trauma ICU have
similar chances of survival compared to patients without diabetes. Diabetes significantly
increased mortality risk only in patients admitted after cardiac surgery, where it
distinctly influences the underlying coronary disease. Further studies are needed to
unravel the pathophysiological mechanisms by which patients with diabetes seem to
be protected in non-surgical settings, despite encountering higher complication rates.
From a clinical perspective, we may conclude from the findings described in Part II of this
thesis that the optimal glucose target range in critically ill patients lies above the range
considered normoglycaemic, irrespective of the diabetic status prior to admission. Marked
hyperglycaemia and hypoglycaemia should be avoided. Continuous glucose monitoring
shows good accuracy in cardiac surgery patients, and its accuracy seem independent
from microcirculatory parameters.
Future considerationsAs always, also this research raises new questions. A decrease in glucose variability seems
not to decrease cardiovascular event rates in type 2 diabetes patients after myocardial
infarction, but a randomised controlled trial specifically lowering glucose variability in
other patient groups has not been performed yet. The most promising results are to be
expected in the critically ill because epidemiological studies consistently show that in this
population glucose variability is associated with mortality. It will be a major challenge
to come up with an intervention in this population that lowers glucose variability while
leaving mean glucose unaffected, but it is the only way to investigate whether glucose
variability is causally related to mortality or only a manifestation of severe disease.
Continuous glucose monitoring might be a useful tool to increase time in target range
and decrease hypoglycaemia and perhaps also glucose variability in the critically ill. A
randomised controlled trial comparing the FreeStyle Navigator CGM system with point-
202
of-care glucose measurements assessing these questions is currently being performed in a
mixed population of critically ill patients and the results of this trial are avidly waited for.
In addition to subcutaneous glucose monitoring, also intra-vascular glucose monitoring
devices are being developed with promising results regarding accuracy. However, clinical
trails will have to demonstrate that the possible beneficial effects will counterbalance
the costs and possible complications of its invasiveness. Finally, an intriguing question is
why critically ill patients with diabetes are less affected by hyperglycaemia than patients
without previously diagnosed diabetes, while hypoglycaemia seems more harmful.
ConclusionThe central question of this thesis is whether it is necessary to curb all glucose peaks.
From the studies presented in this thesis we conclude that this is not always the case.
In diabetes it is important to lower mean glucose while avoiding hypoglycaemia, but we
found that lowering of glucose to normoglycaemia in critically ill patients seems actually
harmful, in patients with and without diabetes. In addition, our studies show that
glucose variability does not need separate treatment in diabetes, while it is associated
with mortality in critically ill patients without diabetes. Thus, not all what goes up
must come down.
Summary and future considerations
Ch
apte
r 13
203
204
Samenvatting
In een gezond lichaam wordt het glucose (suiker) gehalte in het bloed zeer strak
gereguleerd omdat zowel te hoge als te lage waarden schadelijk kunnen zijn. Als gevolg
hiervan overstijgt de glucoseconcentratie in het bloed bij gezonde mensen bijna nooit 7.8
mmol/l. Soms is dit evenwicht echter verstoord. Een chronisch verhoogd glucosegehalte
in het bloed (hyperglycemie) is karakteristiek voor de ziekte diabetes mellitus, maar
ook ernstig zieke patiënten zonder diabetes mellitus kunnen acute hyperglycemie
ontwikkelen. In beide gevallen is ernstige hyperglycemie schadelijk en geassocieerd
met het ontstaan van complicaties en zelfs overlijden.
Dit proefschrift gaat over de effecten en de behandeling van pieken in glucose die
optreden tijdens chronische en acute hyperglycemie. De centrale vraag is of het zinvol
is om deze glucosepieken altijd in te perken. In Deel I van dit proefschrift worden de
consequenties van schommelingen in glucose bij patiënten met diabetes besproken
in relatie tot oxidatieve stress, een proces welke geassocieerd is met schade aan de
binnenbekleding van de vaatwand, en twee chronische complicaties van diabetes,
schade aan de zenuwen en hart- en vaatziekten. In Deel II van dit proefschrift wordt
hyperglycemie besproken bij ernstig zieke patiënten die opgenomen zijn op de intensive
care. Allereerst stellen wij een veilig gebied voor glucoseregulatie voor. Ook testten wij
de nauwkeurigheid en betrouwbaarheid van het continu monitoren van glucose (CGM)
in het vetweefsel, een nieuwe methode die indirect het glucosegehalte in het bloed kan
monitoren. Tot slot hebben wij onderzocht wat de implicatie is van het hebben van de
ziekte diabetes tijdens een opname op de intensive care in relatie tot glucoseregulatie
en de kans op overlijden.
Deel IIn Hoofdstuk 2 wordt een overzicht gegeven van verschillende methoden om
schommelingen in glucose, ofwel glucose variabiliteit, te meten en bespreken we het
belang van glucose variabiliteit. Er bestaat namelijk een grote variatie in zowel het
aantal als de duur van glucosepieken bij patiënten met diabetes. Op dit moment is er
geen “gouden standaard” om glucose variabiliteit te kwantificeren. De standaarddeviatie
van het gemiddelde wordt het meest gebruikt in de literatuur en lijkt mathematisch
de best gevalideerde methode. In in vitro, dierexperimenteel en humane experimentele
studies, verhogen pieken in glucose oxidatieve stress. Bij patiënten met type 1 of type
2 diabetes mellitus is het bewijs voor een onafhankelijk effect van glucose variabiliteit
op het ontstaan van oxidatieve stress en complicaties echter marginaal, en mogelijk
beperkt tot een specifieke groep patiënten met type 2 diabetes die slecht ingesteld zijn
en behandeld worden met alleen orale glucoseverlagende middelen. In tegenstelling tot
patiënten met diabetes is glucose variabiliteit bij ernstig zieke patiënten opgenomen op
Ch
apte
r 13
205
Samenvatting
de intensive care (IC) onomstotelijk geassocieerd met overlijden.
In Hoofdstuk 3 onderzochten we bij welke glucoseconcentratie de effecten op
de vasculaire homeostase beginnen en of dit een aan/uit fenomeen is met een
drempelwaarde of een continue relatie. Hiervoor hebben we bij gezonde vrijwilligers de
glucoseconcentratie in het bloed stapsgewijs omhoog gebracht naar achtereenvolgens 6.0,
8.0 en 10.0 mmol/l. De effecten van glucose op oxidatieve stress, stolling en fibrinolyse,
en de endotheliale glycocalyx werden bestudeerd. De stijging in oxidatieve stress
bleek dosisafhankelijk en het activeren van het stollingssysteem gebeurde al bij een
drempelwaarde van 6.0 mmol/l glucose. De afwezigheid van een drempelwaarde of een
zeer lage drempelwaarde pleit tegen een rol voor glucose variabiliteit in het activeren
van oxidatieve stress of het stollingssysteem.
Deze conclusie wordt gesteund door de resultaten beschreven in Hoofdstuk 4 en 5. In
Hoofdstuk 4 onderzochten wij de relatie tussen glucose variabiliteit, gemeten met behulp
van CGM, en oxidatieve stress, bepaald door het meten van de 24-uurs uitscheiding van
de marker 8-iso prostaglandine F2α in de urine, bij 24 patiënten met type 2 diabetes
die goed gereguleerd waren en behandeld werden met alleen orale glucose verlagende
therapie. We vonden geen relevante relatie tussen glucose variabiliteit en oxidatieve
stress. In Hoofdstuk 5 werd het gebruik van drie keer daags een kortwerkend insuline
bij de maaltijd vergeleken met het gebruik van één keer daags een langwerkend insuline
wat betreft het effect op glucose regulatie en oxidatieve stress. In een cross-over design
werden 40 patiënten met type 2 diabetes geïncludeerd. Het toevoegen van insuline aan
de medicatie van de patiënten verlaagde de gemiddelde glucosewaarden en oxidatieve
stress significant. Echter, opnieuw werd geen relatie gevonden tussen glucose variabiliteit
en oxidatieve stress.
Het wordt aangenomen dat oxidatieve stress vooraf gaat aan vasculaire complicaties,
maar uiteindelijk is het een indirecte marker voor ziekte en geen harde uitkomstmaat.
In Hoofdstuk 6 onderzochten we het effect van glucose variabiliteit op het ontstaan
van perifere en autonome zenuwschade, neuropathie. Hiervoor hebben we data van de
Diabetes Control and Complications Trial (DCCT) opnieuw geanalyseerd. De DCCT was
oorspronkelijk opgezet om het effect van intensieve vs. conventionele glucoseverlagende
behandeling op het ontstaan van microvasculaire complicaties bij patiënten met type
1 diabetes te bekijken. Uit onze analyse bleek dat het ontstaan van neurologische
complicaties sterk geassocieerd was met de hoogte van de gemiddelde glucose, zoals
verwacht, maar we vonden geen effect van glucose variabiliteit.
206
In Hoofdstuk 7 wordt een secundaire analyse van The Hyperglycaemia and Its Effect
After Acute Myocardial Infarction on Cardiovascular Outcomes in Patients With Type 2
Diabetes Mellitus study (HEART2D) beschreven. Deze gerandomiseerde studie vergeleek
het effect van een kortwerkend insuline bij de maaltijden met een langwerkend insuline
op het ontstaan van toekomstige hart- en vaatziekten in patiënten met type 2 diabetes
die geïncludeerd werden na een acuut hartinfarct. De gemiddelde glucoseregulatie was
gelijk tussen de groepen, maar ondanks achttien procent minder glucose variabiliteit in
de maaltijdinsuline groep werden er geen verschillen in hart- en vaatziekten gevonden.
Concluderend ondersteunt Deel I van dit proefschrift een relatie tussen glucose
variabiliteit en oxidatieve stress of diabetische complicaties niet. Bovendien bleek
dat het specifiek verlagen van glucose variabiliteit bij patiënten met type 2 diabetes
niet resulteerde in een afname van het aantal hart- en vaatziekten in deze groep. Om
die redenen zijn er op dit moment onvoldoende argumenten om specifiek glucose
variabiliteit te verlagen bij patiënten met diabetes. De behandeling zal blijven bestaan
uit het verlagen van de gemiddelde glucose en het vermijden van te lage bloedglucose
concentraties, hypoglycemieën.
Deel IIHet is nodig om ernstige hyperglycemie te vermijden bij ernstig zieke patiënten. Er
is echter discussie over wat de streefwaarden voor glucose zouden moeten zijn. In
Hoofdstuk 8 onderzochten we de relatie tussen de gemiddelde bloedglucose concentratie
tijdens opname op de IC en de kans op overlijden in twee cohorten: patiënten opgenomen
met een internistische of een chirurgische reden. In beide cohorten bleek de gemiddelde
bloedglucose gerelateerd aan overlijden op basis van een U-vormige curve, met de
laagste kans op overlijden bij een gemiddelde glucose tijdens opname tussen 7.0 en
9.0 mmol/l. Deze resultaten zijn in overeenstemming met de resultaten van de NICE-
SUGAR (Normoglycaemia in Intensive Care Evaluation- Survival Using Glucose Algorithm
Regulation) studie, en suggereren dat het verlagen van de bloedglucose tot lagere
(normale) waarden wellicht meer kwaad dan goed doet in deze patiëntengroep.
Het continue monitoren van glucose (CGM) in het vetweefsel zou een nuttige methode
kunnen zijn om de bloedglucose beter te reguleren. In Hoofdstuk 9 vergelijken we de
nauwkeurigheid en de betrouwbaarheid van twee van deze apparaten: de Guardian Real-
Time (geproduceerd door Medtronic Minimed) en de FreeStyle Navigator (geproduceerd
door Abbott Diabetes Care). Voor deze studie zijn 60 patiënten geïncludeerd, opgenomen
op de IC na open hartchirurgie. De FreeStyle Navigator bleek nauwkeuriger en meer
betrouwbaar dan de Guardian Real-Time. Opmerkelijk was wel dat de nauwkeurigheid
van beide systemen behoorlijk goed was in vergelijking met resultaten die we kennen
van patiënten met diabetes buiten het ziekenhuis. Onze data ondersteunen het gebruik
Ch
apte
r 13
207
Samenvatting
van de FreeStyle Navigator voor patiënten die open hartchirurgie ondergaan. Echter,
toekomstig onderzoek moet uitwijzen of het gebruik van deze apparaten daadwerkelijk
zorgt voor een betere glucoseregulatie en het voorkomen van hypoglycemieën.
Onze hypothese was dat de nauwkeurigheid van de CGM systemen zou kunnen worden
beïnvloed door de microcirculatie, het stelsel van de allerkleinste vaten in het lichaam. In
Hoofdstuk 10 wordt de microcirculatie van dezelfde 60 patiënten na open hartchirurgie
beschreven en het effect op de nauwkeurigheid van de CGM systemen geanalyseerd.
Het bleek dat de microcirculatie verslechterd was gedurende de eerste uren na de
operatie, maar deze verslechtering had geen invloed op de nauwkeurigheid van de CGM
systemen. Een verlaging van de perifere temperatuur verslechterde de nauwkeurigheid
van beide systemen wel. Een hogere leeftijd en ernstiger ziekte beïnvloedden alleen
de nauwkeurigheid van de FreeStyle Navigator negatief. Vervolgstudies zullen moeten
uitwijzen wat de invloed is van grotere veranderingen in de microcirculatie op de
nauwkeurigheid van CGM systemen bij ernstiger zieke patiënten.
Het wordt een uitdaging wanneer acute en chronische hyperglycemie samenkomen:
de ernstig zieke patiënt met diabetes mellitus. Hoofdstuk 11 geeft een overzicht van
de huidige literatuur over de morbiditeit en mortaliteit van patiënten met diabetes die
opgenomen zijn op de IC. Er wordt specifiek aandacht besteed aan glucoseregulatie,
behandeling met insuline en hypoglycemie. Diabetes is een risicofactor voor het ontstaan
van complicaties tijdens opname op de IC, maar dit betekent niet meteen dat patiënten
met diabetes ook eerder overlijden (de relatie tussen diabetes en overlijden wordt verder
besproken in Hoofdstuk 12). Hyperglycemie komt vaak voor bij ernstig zieke patiënten
met diabetes. Dit leidt tot een verhoogde kans op overlijden als de glucoseconcentratie
boven de 11.1 mmol/l uitkomt, maar er is discussie over het schadelijke effect van
hyperglycemie lager dan 11.1 mmol/l. Zeer intensieve insulinetherapie lijkt niet beter
te zijn voor patiënten met diabetes dan minder intensieve insulinetherapie. Opvallend is
dat bij elke mate van hyperglycemie de mortaliteit van patiënten zonder diabetes hoger is
dan die van patiënten met diabetes, maar dat voor lagere glucosewaarden het omgekeerde
geldt. Patiënten met diabetes zijn kwetsbaar voor het ontwikkelen van hypoglycemie,
wat sterk geassocieerd is met overlijden. Wij raden aan om de glucoseconcentratie bij
ernstig zieke patiënten met diabetes tussen de 7.5 en 10.0 mmol/l te houden en het
ontstaan van hypoglycemieën en ernstige hyperglycemie te vermijden.
Tot slot wordt in Hoofdstuk 12 het resultaat van een meta-analyse beschreven waarin het
effect van het hebben van diabetes op overlijden op verschillende types IC onderzocht
is. In totaal werden 141 studies in deze analyse geïncludeerd, die meer dan 12.4
miljoen patiënten bevatte inclusief 2.7 miljoen (21.7%) doden en 2.3 miljoen (18.6%)
patiënten met diabetes. De analyse liet zien dat patiënten met diabetes dezelfde kans
208
op overlijden hebben als patiënten zonder diabetes wanneer zij opgenomen zijn in een
internistische, trauma of gemengde IC. De aanwezigheid van diabetes verhoogde alleen
de kans op overlijden bij patiënten die opgenomen waren na hartchirurgie, maar bij
deze patiëntengroep worden de onderliggende afwijkingen van de kransslagaderen
in grote mate negatief beïnvloed door diabetes. Meer onderzoek is nodig om de
pathofysiologische mechanismen te ontrafelen die een rol spelen bij de relatieve
bescherming van patiënten met diabetes in een niet-chirurgische setting, ondanks
een hoger aantal complicaties.
Vanuit een klinisch perspectief kunnen we concluderen uit Deel II van dit proefschrift
dat het optimale doel voor glucoseregulatie bij ernstig zieke patiënten met en zonder
diabetes boven het “normale” niveau ligt. Extreme hyperglycemie en hypoglycemie
moeten worden vermeden. Continue glucose monitoring bij patiënten die hartchirurgie
hebben ondergaan is behoorlijk nauwkeurig en deze nauwkeurigheid lijkt onafhankelijk
te zijn van de microcirculatie. De ziekte diabetes draagt niet bij aan een verhoogde kans
op overlijden, tenzij de patiënt is opgenomen voor hartchirurgie.
ToekomstperspectiefZoals altijd, roept ook dit onderzoek nieuwe vragen op. Het lijkt erop dat het verlagen
van glucose variabiliteit er niet voor zorgt dat het aantal nieuwe hart- en vaatziekten
vermindert bij patiënten met type 2 diabetes opgenomen na een hartinfarct, maar
een gerandomiseerd onderzoek met deze vraagstelling is nog niet verricht in andere
patiëntgroepen. Het meeste resultaat kan verwacht worden bij ernstig zieke patiënten
opgenomen op de IC, aangezien epidemiologische studies in deze patiëntengroep
consequent laten zien dat glucose variabiliteit is geassocieerd met overlijden. Het zal
wel een grote uitdaging worden om een interventie te bedenken die alleen glucose
variabiliteit verlaagt en het gemiddelde glucose ongemoeid laat, maar het is de enige
manier om uit te zoeken of hoge glucose variabiliteit een onafhankelijke oorzaak
is voor overlijden of alleen een manifestatie van ernstige ziekte. CGM is mogelijk
bruikbaar om de glucoseregulatie te verbeteren en het aantal hypoglycemieën en
glucose variabiliteit te verminderen bij ernstig zieke patiënten. Op dit moment wordt
een gerandomiseerd onderzoek uitgevoerd op een gemengde IC die deze vragen
probeert te beantwoorden; het gebruik van de FreeStyle Navigator wordt vergeleken met
minder frequente standaard glucosemetingen. We zijn dan ook zeer benieuwd naar de
resultaten van dit onderzoek. Naast CGM in het vetweefsel wordt er ook meetapparatuur
ontwikkeld die de glucoseconcentratie continu en direct in het bloedvat meet. Tot
nu toe zijn deze resultaten veelbelovend. Maar ook hier geldt dat klinische studies
zullen moeten uitwijzen of de potentieel gunstige effecten opwegen tegen de kosten
en mogelijke complicaties. Tot slot is een intrigerende vraag die beantwoord moet gaan
worden waarom ernstig zieke patiënten met diabetes minder schade ondervinden van
Ch
apte
r 13
209
Samenvatting
hyperglycemie dan patiënten zonder diabetes, terwijl hypoglycemie voor hun juist
schadelijker lijkt te zijn.
ConclusieDe centrale vraag in dit proefschrift is of het zinvol is om pieken in glucose altijd in te
perken. Op basis van de studies die gepresenteerd worden in dit proefschrift concluderen
we dat dit niet altijd het geval is. Voor patiënten met diabetes is het belangrijk om de
gemiddelde glucose te verlagen en het aantal hypoglycemieën te verminderen, maar we
hebben gezien dat bij ernstig zieke patiënten opgenomen op de IC het verlagen van de
glucoseconcentratie naar normale waarden juist schadelijk is, zowel bij patiënten met
als patiënten zonder diabetes. Daarnaast laten onze studies zien dat glucose variabiliteit
bij patiënten met diabetes niet apart behandeld hoeft te worden, maar bij ernstig zieke
patiënten zonder diabetes geassocieerd is met overlijden. Kortom, glucose pieken hoeven
niet altijd verlaagd te worden.
210
Authors’ affiliations
Alan S. Rigby
Academic Department of Cardiology, University of Hull and Hull-York Medical School,
Hull, United-Kingdom
Bregtje A. Lemkes
Department of Internal Medicine, Academic Medical Centre, Amsterdam
Durk F. Zandstra
Department of Intensive Care, Onze Lieve Vrouwe Gasthuis, Amsterdam
Eric S. Kilpatrick
Department of Clinical Biochemistry, Hull Royal Infirmary, Hull, United Kingdom
Frits Holleman
Department of Internal Medicine, Academic Medical Centre, Amsterdam
Heleen M. Oudemans- van Straaten
Department of Intensive Care, Onze Lieve Vrouwe Gasthuis, Amsterdam
Henk van Lenthe
Laboratory Genetic Metabolic Diseases, Academic Medical Centre, Amsterdam
Jeroen Hermanides
Department of Internal Medicine and Department of Anaesthesiology, Academic Medical
Centre, Amsterdam
J. Hans DeVries
Department of Internal Medicine, Academic Medical Centre, Amsterdam
Joost B.L. Hoekstra
Department of Internal Medicine, Academic Medical Centre, Amsterdam
Joost C. Meijers
Department of Experimental Vascular Medicine, Academic Medical Centre, Amsterdam
Ch
apte
r 13
211
Authors’ affilliations
Lisa Kerr
Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
Maartje Hickmann
Department of Internal Medicine, Academic Medical Centre, Amsterdam
Max Nieuwdorp
Department of Internal Medicine, Academic Medical Centre, Amsterdam
Peter H.J. van der Voort
Department of Intensive Care, Onze Lieve Vrouwe Gasthuis, Amsterdam
Robert J. Bosman
Department of Intensive Care, Onze Lieve Vrouwe Gasthuis, Amsterdam
Robin Mukherjee
Statistical Department, Pfizer Inc., New York, New York, USA
Scott J. Jacober
Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
Steven L. Atkin
Department of Diabetes, Hull-York Medical School, Hull, United Kingdom
Temo Barwari
Department of Internal Medicine, Academic Medical Centre, Amsterdam
Wim Kulik
Laboratory Genetic Metabolic Diseases, Academic Medical Centre, Amsterdam
212
List of publications
Siegelaar SE, Hoekstra JB, DeVries JH (2011) Special considerations for the diabetic patient
in the intensive care unit: targets for treatment and risks of hypoglycaemia. Best Pract
Res Clin Endocrinol Metab: in press
Siegelaar SE, Kerr L, Jacober SJ, DeVries JH (2011) A decrease in glucose variability does
not reduce cardiovascular event rates in type 2 diabetes patients after acute myocardial
infarction: a reanalysis of the HEART2D study. Diabetes Care 34(4):855-857
Siegelaar SE, Barwari T, Hermanides J, van der Voort PHJ, DeVries JH (2011) Accuracy and
reliability of continuous glucose monitoring in the intensive care unit; a head-to-head
comparison of two subcutaneous glucose sensors in cardiac surgery patients. Diabetes
Care 34(3): e31
Siegelaar SE, Barwari T, Kulik W, Hoekstra JB, DeVries JH (2011) No relevant relationship
between glucose variability and oxidative stress in well-regulated type 2 diabetes patients.
J Diabetes Sci Technol 5(1): 86-92
Siegelaar SE, Hermanides J, Oudemans- van Straaten HM, van der Voort PH, Bosman RJ,
Zandstra DF, DeVries JH (2010) Mean glucose during ICU admission is related to mortality
by a U-shaped curve in surgical and medical patients: a retrospective cohort study. Crit
Care 14(6): R224
Siegelaar SE, DeVries JH (2010) Strakke glucoseregualtie en de mogelijke rol voor continue
glucose monitoring op de intensive care. Intensive Care Capita Selecta: 15-22
Siegelaar SE, DeVries JH, Hoekstra JB (2010) Patients with diabetes in the intensive care
unit; not served by treatment, yet protected? Crit Care 14(2): 126 (commentary)
Siegelaar SE, Holleman F, Hoekstra JB, DeVries JH (2010) Glucose variability; does it
matter? Endocr Rev 31(2): 171-182
Siegelaar SE, Kilpatrick ES, Rigby AS, Atkin SL, Hoekstra JB, DeVries JH (2009) Glucose
variability does not contribute to the development of peripheral and autonomic
neuropathy in type 1 diabetes: data from the DCCT. Diabetologia 52(10): 2229-2232
Ch
apte
r 13
213
List of publications
Siegelaar SE, Kulik W, van Lenthe H, Mukherjee R, Hoekstra JB, DeVries JH (2009) A
randomized clinical trial comparing the effect of basal insulin and inhaled mealtime
insulin on glucose variability and oxidative stress. Diabetes Obes Metab 11(7): 709-714
DeVries JH, Siegelaar SE, Holleman F, Hoekstra JB (2008) Intensive insulin therapy in
patients with type 2 diabetes. Lancet 372(9640): 717
Siegelaar SE, Olff M, Bour LJ, Veelo D, Zwinderman AH, van Bruggen G, de Vries GJ,
Raabe S, Cupido C, Koelman JH, Tijssen MA (2006) The auditory startle response in post-
traumatic stress disorder. Exp Brain Res 174(1): 1-6
214
Dankwoord
Ha, het is af! Ik wil graag iedereen danken die bijgedragen heeft aan de totstandkoming
van dit proefschrift en een aantal in het bijzonder:
Allereerst alle deelnemers aan de verschillende studies. Zonder jullie bijdrage is klinisch
wetenschappelijk onderzoek niet mogelijk.
Mijn promotor, prof. dr. J.B.L. Hoekstra. Beste Joost, wat ben ik blij dat je precies op
het moment belde dat ik twijfelde om er überhaupt aan te beginnen. Je zag het zitten
dat ik het eerste jaar promotieonderzoek en topsport combineerde. Je vond het een
uitdaging. Je positivisme en enthousiasme zijn zeer bijzonder: toen de studie waar ik
op aangenomen was na twee weken toch niet doorging, bij alle manuscripten die ik
aan je voorlegde, of als er weer eens een hardloopwedstrijdje werd voorgesteld. Je vond
het allemaal supermooi. Heel veel dank voor alles, en het is zeker waar: kies eerst je
promotor, daarna je onderwerp.
Mijn co-promotor, dr. J.H. de Vries (alias J. Hans DeVries). Beste Hans, ik weet niet of ik
had verwacht dat je op een druilerige dag in een regenpak langs de Amstel zou komen
fietsen om naar roeien te kijken. Het typeert je betrokkenheid. Enorm veel dank voor je
snelheid, scherpte en (licht?) cynische humor. Stukken kwamen met razende snelheid
terug, en passages waar ik zelf niet helemaal zeker over was werden er feilloos uitgepikt.
Dit zorgde voor een mooie flow, fijn!
De overige leden van de promotiecommissie, prof. dr. Fliers, prof. dr. Romijn, prof. dr.
Smulders en prof. dr. Zandstra. Hartelijk dank voor het kritisch beoordelen van dit
proefschrift en de bereidheid om plaats te nemen in mijn promotiecommissie. Dear
prof. dr. Kilpatrick, it was a pleasure working with you. I am honoured you are willing
to serve on my doctorate committee.
Frits, met de ondergang van de SMILING studie verdween helaas ook onze directe
samenwerking een beetje naar een zijspoor. Toch heb ik veel geleerd van je inventiviteit
en duidelijke mening.
Gabor, zonder jou was ik hier niet terechtgekomen. Ik hoop dat de wijn lekker was!
Mijn collegae. Bregtje, roomie en partner in crime, onze (maandag) chitchat en
wetenschappelijke discussies ga ik missen daar in Noord-Holland! Sanne, Jeroen, Anne,
Els, Yoeri, Arianne, Airin, Wanda en Josefine, dank voor de gezellige tijd en altijd
scherpe (politiek geëngageerde) lunchdiscussies. Studenten, dank voor jullie hulp. In het
Ch
apte
r 13
215
Dankwoord
bijzonder: Maartje, het werd een megaproject. Maar je liet je niet afschrikken, waarvoor
dank. En natuurlijk Temo, altijd vrolijke ‘superstudent’, zonder jou waren het nog veel
langere dagen geworden in het OLVG. Dank voor je initiatief en humor!
Gootje, bij de Appie to Go is veel besproken. We gaan zeker tijd maken voor fietsen,
schaatsen, roeien en koffietjes in dit drukke bestaan. Superleuk dat je mijn paranimf wilt
zijn! Andere vriendjes, vriendinnetjes en negds. Het waren, zijn en blijven mooie tijden!
Lieve schoonfamilie, wat een warm bad. Fijn dat ik jullie er zomaar bijkreeg!
Anne en Henk, lievelings tante en oom, ik werd groot bij de Wijnvriend. Heel stoer vond
ik dat. Dank voor jullie betrokkenheid!
Olie, broet, ik weet zeker dat ze me met jou als paranimf naast me niet fysiek zullen
durven aan te vallen. Heel bijzonder was het om samen in Beijing te zijn. Ik kom zeker
naar Cal als je met je hoed gaat zwaaien! En Tiets, van klein sussie naar grote sus,
superleuk dat je samen met As lekker dichtbij bent komen wonen. Ik ben trots op jullie!
Lieve mam en pap, in één adem, jullie hebben er voor gezorgd dat ik ben wie ik ben.
Schouders eronder en doorgaan, van jullie geleerd. Ik ben trots op zulke lieve ouders!
Lieve Cockie, en Joop hierboven, jullie waren je tijd ver vooruit en hebben me enorm
geïnspireerd. Dankjewel.
Tot slot, lieve Jel, gelukkig vond ik het langlaufen na de eerste keer nog steeds leuk, anders
had je me misschien wel ingeruild. Je onvoorwaardelijke steun was en is fantastisch,
ook al was ik moe, op trainingskamp of op onchristelijke tijden in het ziekenhuis. Het
leven is een mooi ommetje, maar dan wel samen met jou. Luv joe!
216
Ch
apte
r 13
217
Curriculum Vitae
Curriculum Vitae
Sarah Elaine Siegelaar werd op 4 oktober 1981 als oudste van drie kinderen
geboren te Heemstede. Ze leerde de beginselen van het zelfdoen op de Haarlemse
Montessorischool. Met veel plezier doorliep ze daarna het Stedelijk Gymnasium te
Haarlem waar ze in 1999 eindexamen deed. Ze werd direct ingeloot voor de studie
geneeskunde aan de Universiteit van Amsterdam en tijdens het begin van haar
studie begon ze met roeien bij de ASR Nereus. Studentenroeien werd topsport en
na successen op de wereldkampioenschappen in 2003 en Olympische spelen in
2004, rondde ze haar co-schappen af en behaalde haar artsdiploma in 2007. Ze
beëindigde ze haar actieve roeicarrière in 2008 na het behalen van een zilveren
medaille in de vrouwenacht bij de Olympische spelen in Beijing. In oktober 2007
startte ze met haar promotieonderzoek op de afdeling Klinische Diabetologie
onder leiding van prof. dr. Joost Hoekstra. Tijdens haar promotieonderzoek gaf ze
klinisch onderwijs op de Hogeschool van Amsterdam en coachte ze verschillende
roeiploegen op Nereus. Per 1 april 2011 is ze begonnen met de opleiding tot
internist in het Westfriesgasthuis te Hoorn. Sarah is in oktober 2009 getrouwd
met Jelle Luijnenburg.