Sepsis : Update Biomarkers Prof. Jérôme Pugin Soins Intensifs Hôpitaux Universitaires de Genève
Intravenous E.coli endotoxin injection vs. saline to healthy human volunteers Transcriptomic analysis in blood leukocytes at times 0, 2, 4 6, 9, 24 hours after injection
Endotoxin NaCl
Blue : Downregulated Red : Upregulated
Calvano et al. Nature 2005
Red : pathways upregulated Blue : pathways downregulated • Mitochondrial respiratory chain complex I and III • Mitochondrial permeability • ATP synthase complex • Pyruvate dehydrogenase complex • Ribosomal proteins • Proteasome Based on 292 representative genes of the inflammatory pathway
gINF
IL-18
IL-1ra sTNFR2
sTNFR1 IL-10
IL-4
IL-12
Endothelin Nitrites
NO
iNOS
Lymphotoxin
IL-8 IL-6 IL-1ß
PCT
C3a
sVCAM-1
sE-selectin
sICAM-1 LBP
cortisol
Neopterin
CRP
HMGB-1
MIF
C5a
D-dimers
APC
TNF
ATIII
PGE2
PGI2
TxA2
LPS
LTB4
sCD14
ESM-1
albumin
HLA-DR
Sepsis
Why do I want a biomarker and what for? • To make a diagnosis?
• To make a decision? - antibiotic therapy? - ICU admission, stratification? - diagnostic procedure? - to start specific therapy (steroids, …)?
• To monitor?
• To predict outcome?
• To enroll patients into studies?
• To perform epidemiological studies?
• To understand the disease?
PCT, PSP?
PCT ? ? cortisol
PCT, CRP, lactate
IL-6, PCT, PSP
LPS, IL-6, PCT?
Genetics
Many
Adapted from Samraj et al. Shock 2013
• Excellent diagnostic yield
sensitive discriminative with other sepsis-like situations • Cross-validated • Easy & rapid to measure • Cheap • Impact on the care of the sepsis patient
The ideal biomarker
1. Sepsis is a continuum between colonization, infection and systemic infection with various levels of severity
2. Sepsis is a heterogeneous syndrome
3. Consequences (organ dysfunction) of sepsis vary
4. Clinical definition of sepsis is vague
5. Bacterial pathogens triggering sepsis are numerous
It is unlikely that a single biomarker will yield perfect diagnosis in the context of sepsis
• Single biomarker: simple, rapid, cheap, sensitive, usually not very specific • Single biomarker associated with clinical signs: probably better, introduces the
clinical context • Multiple biomarkers: technology? Less rapid, more expensive, increased
specificity, lack of sensitivity? • Combine bacterial and host markers?
Single vs. multiple biomarkers
gINF
IL-18
IL-1ra sTNFR2
sTNFR1 IL-10
IL-4
IL-12
Endothelin Nitrites
NO
iNOS
Lymphotoxin
IL-8 IL-6 IL-1ß
PCT
C3a
sVCAM-1
sE-selectin
sICAM-1 LBP
cortisol
Neopterin
CRP
HMGB-1
MIF
C5a
D-dimers
APC
TNF
ATIII
PGE2
PGI2
TxA2
LPS
LTB4
sCD14
ESM-1
albumin
HLA-DR
Sepsis
• Recognized by Assicot et al. as a marker of severe bacterial infection in children
(Lancet 1993)
• Serum PCT levels increase 4-8 hrs after experimental sepsis or LPS injection to
humans, and peak after 12-24 hrs
• Serum PCT levels are slightly elevated in severe parasite et fungal infections
• Serum PCT levels are low in any viral diseases
• “False positives”: 2 first days of life, heat stroke, multiple trauma, some surgical
patients, medullar thyroid cancer
Procalcitonin: FAQs
Müller et al., J Clin Endocrinol Metabol, 2000, 86: 396-404
Procalcitonin gene induction during experimental sepsis
.01
.1
1
10
100
1000
PC
T (
ng
/mL
)
Septic
shock
Severe
sepsis
Sepsis SIRS
Harbarth et al. Am J Respir Crit Care Med 2001
Diagnostic yield of plasma PCT in sepsis
1.1 ng/mL
Sensitivity 97%
Specificity 78%
PCT adds to a clinical prediction model for the diagnosis of sepsis
1 - specificity
sen
sit
ivit
y
1.00
0.75
0.50
0.25
0 .00
0.00 0.25 0.50 0.75 1.00
Clinical model with PCT
Clinical model without PCT
Harbarth et al. Am J Respir Crit Care Med 2001
Sepsis
Antibiotic therapy: who needs guidance?
When to stop the antibiotic therapy?
“We said 8 days” “Patient is stable” “Patient is transferred to the ward” “Patient develops a rash” “Renal function is deteriorating” “The attending is changing” “Cultures came back negative” …
% p
atients
without
antibio
tics
n=68 HR: 1.9 (1.2-3.1) p=0.009
Probability to have antibiotics stopped based on PCT algorythm
Time to antibiotic discontinuation (days)
PCT
controls
0 5 10 15 20
0.00
0.25
0.50
0.75
1.00
Nobre et al. Am J Respir Crit Care Med 2008
Control
(n=37)
PCT
(n=31) p value
28-day mortality 16.2% 16.1% 0.74
Clinical cure 83.8% 90.3% 0.33
Nosocomial infection 29.7% 22.6% 0.20
Infection replase, % 2.7% 3.2% 0.70
Shortening of antibiotic treatment duration does not affect outcome
All
patients
VAP abdominal
infection
UTI (+) Blood
cultures
N
CAP
20 14
9.9
6.1
10.6
5.6
9.4
7.3
10.8
8.1
14.5
7.4
12.8
9.8
0
2
4
6
8
10
12
14
16
Du
rati
on
of
treatm
en
t (d
ays)
314 307 101 79 66 75 18 24 53 55
PCT-guided
Control
Use of procalcitonin to shorten antibiotic exposure in ICU patients. the ProRata trial
Bouadma et al. Lancet 2010
Prob
ab
ilit
y o
f su
rviv
al,
%
Days after inclusion
Procalcitonin (n=311)
Control group (n=319)
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60
Bouadma et al. Lancet 2010
p = n.s.
Use of procalcitonin to shorten antibiotic exposure in ICU patients. the ProRata trial
PCT algorithm for patients with sepsis
Schuetz P BMC Med 2011
Antibiotic stewardship We recommend daily assessment for de-escalation of antimicrobial therapy in patients with sepsis and septic shock.
(Expert recommendation)
We suggest that measurement of procalcitonin levels can be used to support shortening the duration of antimicrobial therapy in sepsis patients.
(Weak recommendation; low quality of evidence)
Early sensitive, speciic and prognostic marker of sepsis
Point of care testing, few microliters of blood
Results in 5 min
Pancreatic stone protein
Abioscope®
Benninga et al.
Results of Abionic study in February 2020 296 patients included; Multicentric, CH (Geneva, Lausanne and Bern), France. Italy, UK ICU patients at risk to develop sepsis, sampled every 12 hours: will PSP anticipate diagnosis of inectiion ans sepsis ?
Diagnostic and prognostic yield of sepsis biomarkers
Eggimann et al. Biomark.Med. (2019) 13(02), 135–145
« Molecular biomarkers are taking front stage at present, but machine learning and other computational measures using bigdata sets are promising.» S. Opal and X. Wittebole Crit Care Clin 2020; 36:11-22
R.J.Piso KSO
2.3 MioCHF
5 ICUs from university hospitals + ETH Zürich
Hundreds of physiological, clinical, microbiological, laboratory variables per patient continuously feeding a shared data warehouse
Tagging patients with sepsis (ICU)
Machine learning Defining algorithms for sepsis prediction and prognosis
Personalized Swiss Sepsis Study
• Define the question before using (choosing) a sepsis biomarker : Diagnosis? Detection? Risk stratification? Antibiotic guidance? Prognosis?
• Procalcitonin remains the #1 biomarker for diagnosis and antibiotic guidance • Many have prognostic values… • Biomarkers frequently comfort the clinician impact on patient care? • Pancreatic stone protein seems to add prediction + prognosis + early • Digital prediction and prognosis based on algorithms (machine learning, big data)?
Conclusions
Thank you for your attention!