Circadian modulation of glucose and insulin responses to ......Circadian modulation of glucose and insulin responses to meals: relationship to cortisol rhythm EVE VAN CAUTER, E. TIMOTHY

Circadian modulation of glucose and insulin responses to meals: relationship to cortisol rhythm

EVE VAN CAUTER, E. TIMOTHY SHAPIRO, HARTMUT TILLIL, AND KENNETH S. POLONSKY Department of Medicine, University of Chicago, Chicago, Illinois 60637; and Institute of Interdisciplinary Research, Universitk Libre de Bruxelles, B-1070 Brussels, Belgium

Van Cauter, Eve, E. Timothy Shapiro, Hartmut Tillil, and Kenneth S. Polonsky. Circadian modulation of glucose and insulin responses to meals: relationship to cortisol rhythm. Am. J. Physiol. 262 (Endocrinol. Metab. 25): E467-E475, 1992.-To determine whether glucose and insulin responses to a mixed meal are influenced by time of day irrespective of duration of prior fast, eight normal subjects (4 males, 4 females) were studied on two separate occasions, involving ingestion of identical meals at either 6- or 12-h intervals. The 24-h profiles of plasma glucose, serum insulin, and plasma C-peptide were obtained at 200min intervals. Plasma cortisol levels were measured on each sample to evaluate possible relationship between diurnal variations in metabolic responses and circadian rhythm of cortisol. Rates of secretion of insulin and cortisol were mathematically derived from peripheral concentrations by deconvolution using two-compartment models for clearance kinetics. Postmeal responses of glucose, insulin, and insulin secretion rate were evaluated by calculating maximum postmeal increment, total area under curve, area under curve for 2 h after meal ingestion, and total duration of response. Postmeal cortisol responses were quantified by increment in plasma level and amount secreted in postmeal pulse. For glucose responses, irrespective of duration of prior fast, all four parameters char- acterizing the response were significantly greater in the evening than in the morning, with total area under curve and 2-h area under curve being approximately twofold larger in the evening than in the morning. Time of day did not significantly influence maximum postmeal increment in insulin secretion rate or duration of insulin secretory response, but total and 2-h areas under curve were 25-50% greater in the evening than in the morning. Meal ingestion was followed by a significant pulse of cortisol secretion in 37 of 40 cases. Magnitude of morning-to- evening increase in insulin response to meals was correlated with magnitude of morning-to-evening decrease in cortisol level. We conclude that glucose and insulin responses to mixed meals in normal adults of both sexes are profoundly modulated by circadian rhythmicity and that this diurnal variation in meal tolerance may be mediated at least partially by circulating cortisol concentrations. In the evening, failure of insulin secretion to increase in proportion to changes in postmeal glucose responses suggests that, in addition to circadian changes in insulin sensitivity, there may be a circadian variation in re- sponsiveness of &cells to glucose. circadian rhythmicity; food intake; glucose tolerance; insulin secretion

THE ROLE OF CIRCADIAN RHYTHMICITY in modulating human endocrine function is well established for systems directly dependent on the hypothalamopituitary axes. It is often assumed that the function of more peripheral endocrine systems, such as the pancreas, is not significantly influenced by central mechanisms responsible for circadian timing. However, in normal humans, a series of studies have suggested that the set point of glucose regulation may be under circadian control. Indeed, glu-

cose tolerance to an oral glucose load or to a single intravenous injection of glucose is decreased in the afternoon compared with the morning (8). With the use of experimental designs involving constant glucose infusion for 24-30 h, we have shown that glucose tolerance con- tinues to deteriorate as the evening progresses and reaches a minimum around midsleep (19,22). In a recent study, we have further demonstrated that the diurnal variation in glucose levels during constant glucose infusion is paralleled by a similar variation of insulin secretion, which is inversely related to the circadian rhythm of cortisol concentrations (21). These observations obtained during constant glucose infusion suggest that there may be a systematic circadian variation in the glucose and insulin responses to a mixed meal, with larger responses in the evening than in the morning, and that this effect of time of day on meal responses could be partially mediated by cortisol. The existence of such a consistent effect of time of day on meal tolerance could be of importance in the design of meal schedules, which would optimize glucose control for subjects with impaired glucose tolerance and in the development of strategies to cope with conditions of abnormal circadian rhythmicity (i.e., “jet lag” or shift work), which are frequently associated with gastrointestinal disorders.

In all previous studies designed to examine the possible effects of time of day on meal tolerance, the response to the morning meal, which was presented after a lo- to 12- h fast, was compared with the response to the evening meal, which was presented after a 4- to 5-h fast. Thus, when morning vs. evening differences in meal responses were observed, it could not be determined whether they reflected differences in the duration of prior fast or true circadian modulation. Reduced carbohydrate tolerance to mixed meals absorbed later in the day has been shown in some, but not all, studies involving identical meals given at various times of day after varying durations of prior fast (1, 10,12,13,17). The presence and magnitude of a diurnal variation seemed to be dependent on meal size and composition (1,12,17). Two studies by the same group have indicated that the effects of time of day may be more prominent in women than in men (1, 12). It is noteworthy that, in all these previous studies, the “evening” meal was actually given in the late afternoon, between 1630 and 1800 h, at a time when, according to our 24-h studies during constant glucose infusion (19, 22), glucose tolerance has only begun to deteriorate. Thus larger and more consistent effects of time of day may occur when the evening meal is consumed later, as is often the case under ordinary conditions of daily living.

The present study was therefore undertaken to com- 0193-1849/92 $2.00 Copyright 0 1992 the American Physiological Society E46’7

by 10.220.33.4 on June 26, 2017http://ajpendo.physiology.org/

Dow

nloaded from

http://ajpendo.physiology.org/

E468 CIRCADIAN VARIATION IN MEAL RESPONSES

pare the responses to identical meals given at 0800 and at 2000 h while controlling for the duration of the prior fast. To delineate the size, duration, and dynamics of the meal responses, peripheral levels of glucose, insulin, and C-peptide were determined at 20-min intervals over a 24-h span, and insulin secretion rates were derived from the C-peptide levels by deconvolution. Cortisol levels were measured on all samples to evaluate the relationship between diurnal variations in metabolic responses to meals and the circadian periodicity of cortisol secretion.

SUBJECTS AND METHODS

Subjects

Eight normal subjects, four men and four women aged 22- 35 yr, were studied. All were of normal weight [body mass index 21.5 $- 2.2 (SD) k g / m2, range 18.8-24.5 kg/m2], and none had a personal or family history of diabetes. Fasting glucose and insulin levels averaged 4.94 t 0.16 (SD) mmol/l and 45.9 t 4.3 pmol/l, respectively. Shift workers or subjects who had expe- rienced a transmeridian flight <6 wk before the study were excluded. The protocol was approved by the Institutional Re- view Board of the University of Chicago, and all subjects gave written informed consent.

Protocol

Each subject participated in three different studies. The order of the studies was randomized.

Twenty-four-hour study with meals at 6-h intervals. After a 6-h fast, the subjects were admitted in the Clinical Research Center at 1800 h and, starting 2 h later, ingested five identical 5000kcal meals presented at 6-h intervals over a 36-h period. Thus meals were served at 2000 h, 0200 h (nocturnal meal), 0800 h (morning meal), 1400 h (lunch meal), and 2000 h (evening meal). Meal composition was 43% carbohydrate, 39% protein, and 18% fat. All meals were entirely consumed within 30 min. The subjects slept in darkness from 2300 to 0600 h but were awakened at 0200 h to consume the nocturnal meal. The hand with the sampling catheter was kept in a heating blanket to ensure arterialization of the venous blood sample. Arterial- ized blood samples were taken at 20-min intervals for 24 h starting at 0600 h, i.e., 2 h before the morning meal.

Twenty-four-hour study with meals at 12-h intervals. After a 6-h fast, the subjects were admitted at 1800 h and ingested three identical meals at 12-h intervals over a 36-h period. Thus the meals were served at 2000 h, 0800 h (morning meal), and 2000 h (evening meal). The subjects were awakened for 45 min at 0200 h to control for the sleep loss involved in the study with meals at 6-h intervals. In all other respects the protocol was identical to that in the study with meals at 6-h intervals.

Determination of individual C-peptide kinetics. To obtain accurate estimates of insulin secretion, the kinetics of C-peptide were determined in each subject as previously described in detail (15). Briefly, endogenous C-peptide secretion was sup- pressed by means of a primed intravenous somatostatin infusion (500 pg/h; Bachem Fine Chemicals, Torrance, CA), and a 1509pg bolus injection of biosynthetic human C-peptide (Eli Lilly, Indianapolis, IN) was administered. Plasma C-peptide levels were measured at 1- to 3-min intervals over the first 20 min, at 5-min intervals over the following 40 min, and at IO- to 20-min intervals over the following 2 h. For each individual, the rate constants describing the distribution and metabolism of C-peptide were derived from the analysis of the decay curve using a two-compartment mathematical model (15). Mean t SE values for the short and long half-lives of C-peptide were

4.8 t 0.2 and 33.6 & 1.9 min, respectively, mean volume of distribution was 4,090 & 180 ml, and the fraction of decay associated with the short half-life was 0.78 t 0.01.

Assays

Serum insulin was assayed by a double-antibody technique (11) with an intra-assay coefficient of variation averaging 6% throughout the range of measured concentrations. Plasma C- peptide levels were determined using a previously described radioimmunoassay (6) with an average intra-assay coefficient of variation of 4%. Plasma cortisol levels were measured by radioimmunoassay (Diagnostic Products, Los Angeles, CA) with an average intra-assay coefficient of variation of 5%. All samples from the same subject were analyzed in duplicate in the same assay. Plasma glucose concentrations were measured in duplicate with a glucose analyzer (Yellow Springs Instru- ments, Yellow Springs, OH) with a coefficient of variation of <3%.

Data Analysis

Derivation of insulin secretory rates. The rates of production of C-peptide, and, by inference, the insulin secretory rates, were mathematically derived from the plasma C-peptide concentrations using a two-compartment model for C-peptide distribution and metabolism, with individual parameter values derived from the decay curve observed after the bolus injection of C- peptide. This calculation, commonly described as “deconvolution,” involves the integration of equations described by Eaton et al. (5). We have demonstrated that this procedure provides accurate estimates of insulin secretion, even under non-steady- state conditions (15).

Estimation of endogenous insulin clearance. In most situa- tions in which insulin clearance is measured, the endogenous secretion rate of the peptide is not known, and the metabolic clearance rate is derived as the ratio of the exogenous infusion rate of the peptide and its steady-state plasma concentration. In the present study, endogenous insulin secretion was estimated from plasma C-peptide concentrations as described above, and the clearance of endogenously secreted insulin during each meal was calculated as the ratio of the area under the curve (AUC) of insulin secretory rate and the AUC of simultaneously measured peripheral serum insulin concentration as previously described (18).

Postmeal responses in plasma glucose and insulin secretion. The basal level in each individual study was taken to be the morning premeal level. For each meal ingested during the sampling period, the premeal levels of plasma glucose, serum insulin, and insulin secretion were calculated as the mean of the levels during the hour preceding meal presentation. The maximum increments of plasma glucose, serum insulin, and insulin secretory rate were defined as the difference between the maximum level attained after the meal and the premeal level. Because caloric intake was only 1,000 or 1,500 kcal during the sampling period, premeal levels of plasma glucose, serum insulin, and insulin secretion were generally lower in the evening than in the morning. The duration of the response was calculated as the time interval between meal presentation and return to premeal level or basal level, whichever came first. The postmeal responses were estimated as the total AUC above the premeal level. In addition, to evaluate differences in the short-term response to the meal, the AUC for the first 2 h after the meal were similarly calculated.

Analysis of pulses of serum insulin, plasma glucose, plasma cortisol, and insulin secretion. To identify significant pulses of serum insulin, plasma glucose, plasma cortisol, and insulin secretion, each individual 24-h profile was analyzed using Ultra,


Dow

nloaded from


CIRCADIAN VARIATION IN MEAL RESPONSES E469

a computerized pulse-detection algorithm previously described and validated (20). The general principle of this algorithm is the elimination of all peaks for which either the increment (difference between level at the peak and level at the preceding trough) or the decrement (difference between level at the peak and level at the following trough) do not exceed a certain threshold related to measurement error. For serum insulin, plasma glucose, and plasma cortisol, this threshold was set at twice the intra-assay coefficient of variation, a level which, in view of our previous estimations of the frequency and magnitude of pulses of these blood constituents (14, 19, 22), is expected to optimally balance false-positive and false-negative errors (20). For insulin secretory rates, a more conservative threshold of three times the intra-assay coefficient of variation of C-peptide determinations was used to take into account the amplification of noise involved in the deconvolution procedure.

Postmeal responses in plasma cortisol. For each meal, the premeal plasma cortisol level was calculated as the mean of the levels measured during the hour preceding meal presentation. Cortisol secretory rates were mathematically derived from the plasma cortisol concentrations using a two-compartment model for cortisol distribution and metabolism (3). Thus the same deconvolution procedure as that used to estimate insulin secretion was applied to the estimation of cortisol secretory rates, except that, instead of using individually derived kinetic parameters, the parameter values for the disappearance kinetics of cortisol were obtained from previously published studies (3). The volume of distribution, short half-life, and fraction of decay associated with the short half-life were taken to be 4.1% of body weight, 5 min, and 0.80, respectively, for all subjects. For each subject, the long half-life was adjusted in the previously reported physiological range of 65-90 min (3) by an iterative process designed to minimize the number of negative secretory rates. On average, the long half-life was 72.5 & 5.3 min, and the number of significantly negative secretory rates was 0.25 t 0.20 across all 16 studies, i.e., 0.4% of all calculated secretory rates. The error associated with each secretory rate, which results from measurement error on plasma levels, was derived following the theory of error propagation, and the process of pulse identification was repeated for the profile of cortisol secretory rates using a threshold of two times the calculated error. The profiles of cortisol secretory rates generally included more pulses than the profile of plasma concentrations, because the deconvolution procedure revealed that a single large pulse of plasma concentration may reflect two successive pulses of secretion. A postmeal cortisol response was considered to be present if a peak of cortisol secretion rate occurred within 60 min (i.e., 3 sampling intervals) after meal presentation, and the magnitude of the response was estimated by integrating the secretory rates over the duration of the postmeal pulse. The magnitude of the postmeal cortisol response was also charac- terized by the increment of plasma concentration above premeal level.

Statistical tests. Unless otherwise indicated, all group results were expressed as means 2 SE. Nonlinear least-square analysis of the C-peptide decay curves was performed using the BMDP 3R program (BMDP Statistical Software, Los Angeles, CA). The effects of timing of meal presentation and duration of interval between meals were analyzed simultaneously by two- factor analysis of variance for repeated measures. Within each study, group means were compared using the Wilcoxon test for paired data. Correlations were estimated using the Spearman coefficient of correlation.

RESULTS

Figures 1 and 2 depict the profiles of plasma cortisol, plasma glucose, serum insulin, and calculated insulin

7,

300,

400, M

I M c

300,

24-HOUR CLOCK TIME

J I I I I I 1

06 10 14 16 22

24-HOUR CLOCK TIM!

02 06

Fig. 1. Twenty-four-hour profiles of plasma cortisol, glucose, serum insulin, and insulin secretory rates (ISR) in subject 1M2 during study with meals at 12-h intervals (left) and during study with meals at 6-h intervals (right). Large upward arrows, times of meal presentation; downward arrows, significant pulses of cortisol, glucose, insulin, and ISR. M, meal.

secretory rates observed in two representative subjects during the studies with meals presented at 6- and 12-h intervals. A clear increase in the levels of glucose, insulin, and insulin secretion was observed after each meal. A significant pulse in plasma cortisol was observed after 35 of the total of 40 meals presented during the 2 studies. Pulses in peripheral concentrations of glucose, cortisol, and insulin, as well as insulin secretory rates, were evident in response to meals as well as during the periods separating the meals.

The mean profiles of plasma cortisol, plasma glucose, serum insulin, and insulin secretion rates for the two studies are shown in Fig. 3. The estimated postmeal responses are represented by the shaded areas.

Postmeal Glucose Responses

Four measures of the postmeal glucose responses were used to evaluate the effect of time of day, including the maximum postmeal increment, the AUC during the en-


Dow

nloaded from



6.

d 4.

3.

24-tKWi CLOCK TIME 24-M CLOCK TIE

of day on the size of the response was not primarily due to a lengthening of the end tail of the response but to the maintenance of higher glucose levels during the first 2 h after the meal.

It is also evident from the data in Table 1 that the morning-to-evening differences tended to be greater when the meals were presented at 6-h intervals compared with the study where the meals were presented at 12-h

1 I I I 1 I 1 intervals. However, these differences failed to reach sta- M t

M M I tistical significance, and two-factor analysis of variance c

ia”.--,

indicated that the time of meal presentation was the only significant factor responsible for the differences between morning and evening responses.

t I The number of significant glucose pulses in the 6 h after meal presentation averaged 1.8 t 0.6 (SD). The frequency of glucose pulses was similar in both studies

M and did not appear influenced by time of day. I

06 10 14 16 22 02 06

Fig. 2. Twenty-four-hour profiles of plasma cortisol, glucose, serum insulin, and ISR in subject 1M3 during study with meals at 12-h intervals (left) and during study with meals at 6-h intervals (right). Large upward arrows, times of meal presentation; downward arrows, significant pulses of cortisol, glucose, insulin, and ISR.

tire duration of the response, the AUC during the first 2 h after meal presentation, and the duration of the response (Table 1). Irrespective of the interval between meals, the differences between morning and evening responses were statistically significant for each of these parameters. For each of the eight subjects, the total AUC and the 2-h AUC were larger in the evening than in the morning in both the study with meals at 12-h intervals and the study with meals at 6-h intervals. The total AUC were approximately twofold greater after meal ingestion in the evening than in the morning. This large evening increase in overall response was associated with a higher increment over the premeal level as well as to a prolon-

gation of the total duration of the response. However, 24-HUJH CLOCK TIME 24-tUM CLOCK TIME

when the glucose responses to the meal were evaluated Fig. 3. Mean 24-h profiles of plasma cortisol, glucose, serum insulin, over a shorter time interval by calculating the AUC for and ISR during study with meals at 12-h intervals (left) and during the first 2 h after the meal, the differences between study with meals at 6-h intervals (right). Arrows, times of meal pres-

morning and evening responses were similar in size to entation. Note absence of peak of plasma cortisol levels after 1400 h in

those found for the total AUC. Thus the effect of time absence of meal (top left) and well-defined peak associated with meal presentation at 1400 h (top right).


Dow

nloaded from



Table 1. Characteristics of responses to meal

Meals at 12-h Intervals Meals at 6-h Intervals

Morning Evening P Morning Evening P

Maximum increment, mmol/l Total AUC, mmol l 1-1o min 2-h AUC, mmol *l-l. min Duration of response, min

Maximum increment, pmol/l Total AUC, nmol .l-‘. min 2-h AUC, nmol l 1”. min Duration of response, min

Maximum increment, pmol/min Total AUC, nmol . min 2-h AUC, nmol l min Duration of response, min

Glucose response to meal

2.4k0.2 3.120.3 co.01 120.6k19.1 224.9t33.6 co.01 111.3k19.7 203.5k15.9 <O.Ol

93t16 155zk23 co.02

Serum insulin response to meal

281+23 238k29 co.03 21.53k2.57 25.79t2.53 NS 14.95t1.52 17.65k1.90 NS

220+30 263t33 NS

Insulin secretory response to meal

353*34 343t34 NS 28.25k4.25 35.62t3.63 co.01 21.89k3.05 26.67k3.03 co.01

283k17 263k34 NS

2.320.4 89.5k14.4 83.9t12.9

got15

258t34 17.86k2.02 13.29A1.24

223t38

336&82 21.5925.38 17.14k3.94

180t33

3.3kO.3 192.9t33.6 195.0t22.8

145&21

252t25 22.75k1.97 15.6Ok2.10

248k28

325k22 32.6823.07 25.06k2.42

240t27

co.01 co.02 co.01 co.02

NS NS NS NS

NS co.01 co.01

NS Values are means k SE. AUC, area under curve; NS, not significant.

Postmeal Responses in Insulin Secretion Rate

As for glucose, the duration of the interval between meals was not found significant in determining the characteristics of postmeal responses in insulin secretion. Maximal postmeal increments of insulin secretion rate were similar in the morning and in the evening in both studies (Table 1). In the study with meals at 12-h intervals, the total AUC for insulin secretion were -25% larger in the evening than in the morning. In the study with meals at 6-h intervals, the AUC were -50% larger in the evening than in the morning. As was the case for glucose, the enhancement of the morning-to-evening difference was related to smaller morning responses when meals were presented at 6-h, rather than 12-h, intervals. Even though the total duration of the response tended to be longer in the evening, analysis of the 2-h AUC indicated that the effect of time of day was primarily due to increased insulin secretion during the first 2 h after the meal.

Pulses of insulin secretion after meal presentation were significantly more frequent than pulses of glucose, averaging 3.2 t 1.2 (SD) in the 6-h interval after the meal (i.e., -1 pulse every 2 in pulse frequency between of day.

h). There was no difference studies or according to time

in the evening; P < 0.10) and was highly significant in the study with meals at 6-h intervals (1.22 t 0.18 l/min in the morning vs. 1.81 t 0.31 l/min in the evening; P < 0.01).

The frequency of pulses of plasma insulin after meal presentation was similar to that observed for the insulin secretory rates and averaged 3.7 t 1.1 (SD) pulses in the 6-h interval after the meal. There was no difference in the frequency of plasma insulin pulses between studies or according to time of day.

Postmeal Cortisol Responses

As exemplified by the cortisol depicted in Figs. 1 and 2, postmeal rises in cortisol concentrations were usually clearly identifiable. However, in a few instances, deconvolution analysis was necessary to detect postmeal rises in secretion, which were not apparent in the profile of plasma concentrations. An example is shown in Fig. 4, which illustrates the profile of plasma concentrations and the calculated profile of secretory rates in subject A45 during the study with meals at 12-h intervals. The evening meal was presented at a time when plasma cortisol levels were declining after the occurrence of a large spontaneous pulse. No pulse of plasma concentration occurred after the meal. However, examination of

Postmeal Responses in Plasma Insulin

As shown in Table 1, the quantitative characteristics of the postmeal responses in plasma insulin showed trends similar to those observed for the insulin secretory rates (i.e., higher total and 2-h AUC in the evening than in the morning and prolonged duration of response), but these failed to reach statistical significance. Estimations of endogenous insulin clearance during the morning and evening meal suggested that the absence of circadian variation in postmeal insulin responses could be due to increased insulin clearance in the evening. Indeed, a tendency for increased insulin clearance in the evening was present in the study with meals at 12-h intervals (1.42 t 0.16 I/ min in the morning vs. 1.59 t 0.16 l/min

the profile of secretory rates reveals that, at the time of presentation of the evening meal, the secretory process that had caused the large pulse of plasma concentration was terminated and meal ingestion was followed by a smaller secretory pulse. Due to the prolonged disappearance kinetics of the hormone, this meal-related pulse failed to result in an elevation of plasma level. In total, a significant pulse of cortisol secretion was identified after 37 of the 40 meals presented. One of the exceptions is shown in Fig. 4. Indeed, in this individual study, the morning meal was presented when cortisol levels were peaking and only resulted in a small nonsignificant rise in cortisol secretion. The cortisol profiles presented in Figs. 1, 2, and 4 also demonstrate that numerous (i.e ‘., - 4 10) cortisol pulses occurred independently of meals and


Dow

nloaded from



F .#-I E ‘I= 90 - $

S

0 -

I I I I I I I

06 10 14 18 22 02 06

24-HOUR CLOCK TIME

Fig. 4. Profiles of plasma cortisol (top) and calculated profile of cortisol secretory rate (bottom) obtained in subject M5 during study with meals at 12-h intervals. Upward arrows, times of meal presentation; downward arrows, significant pulses of cortisol concentration and of cortisol secretory rates. Note morning meal was given at time when cortisol levels were already peaking and did not elicit significant postmeal rise. Evening meal was given at time of declining cortisol concentrations and resulted in shoulder, rather than peak, of cortisol concentrations. Deconvolution calculations revealed existence of secretory response to meal.

that, as may have been expected, some of these spontaneous pulses occurred between 1400 and 1500 h, even when no meal was served at 1400 h. However, this fortuitous association was only observed in three of the eight subjects (including M2 and M5, shown in Figs. 1 and 4, respectively) and thus failed to result in a well- defined midday peak of cortisol level in the mean profiles for the study with meals at 12-h intervals (Fig. 3, top left)

Table 2 gives the quantitative characteristics of the postmeal cortisol responses. Irrespective of the interval

between meals, the premeal level, the postmeal increase in plasma level, and the amount of cortisol secreted in response to the meal were all significantly lower in the evening than in the morning. In the morning, the amount of cortisol secreted in response to the meal was positively correlated with the magnitude of the postmeal responses in plasma glucose, serum insulin, and insulin secretion (Table 3). In the evening, there were no significant correlations between the magnitude of the cortisol response and the magnitude of the postmeal metabolic responses (Table 3).

To determine whether the effect of time of day on the magnitude of the postmeal glucose and insulin responses was related to the diurnal variation in cortisol concentrations, correlations between the morning-to-evening increases in postmeal responses in plasma glucose, serum insulin, and insulin secretion and the morning-to-evening decline of cortisol levels were calculated. On average, the postmeal responses in plasma glucose, serum insulin, and insulin secretion, as estimated by the 2-h AUC, increased by 166 $- 57, 20 t 9, and 49 sz ll%, respectively, from morning to evening. Premeal cortisol levels decreased from 0800 to 2000 h in 14 of 16 individual studies, and the relative decline averaged 33 t 16% across all studies. As shown in Fig. 5, the magnitude of the morning-to-evening increase in meal response was positively correlated with the magnitude of the morning-to- evening decline of premeal cortisol levels for serum insulin (P < 0.01) and insulin secretion (P < 0.05). The correlation was not significant for plasma glucose (P < 0.11). These results were not dependent on the two studies where cortisol levels were higher, rather than lower, at 2000 than at 0800 h and which were thus associated with “negative” decreases from morning to evening (as shown on the abscissa of Fig. 5). Indeed, when the cortisol decrease in these two studies was considered to be zero, the coefficients of correlation were 0.421 (P < O.ll), 0.614 (P < O.Ol), and 0.511 (P < 0.05) for plasma glucose, serum insulin, and insulin secretion, respectively.

Sex Differences in Postmeal Responses

Previous reports (1,12) have indicated that the effects of time of day on glucose and insulin responses to meal may be sex dependent. In the present study, the 2-h AUC for glucose and insulin secretion were larger in the evening than in the morning in all 16 individual studies, irrespective of the sex of the subject. However, the magnitude of the effect of time of day had a tendency to be larger in women than in men. In men (n = 4), the evening glucose response, as estimated by the 2-h AUC, was on

Table 2. Postmeal co&sol responses

Meals at 12-h Intervals Meals at 6-h Intervals P P

Morning Evening Morning Midday Evening

Premeal plasma level, nmol/l Increment in concentration, nmol/l Amount secreted, pmol

364k66 168t33 <O.OOl 312*55 295t47 157t25 <O*OOl 177&52 113t36 co.02 270-r-69 188t58 166k47 eo.01

43.Ok12.1 18.4k3.8 <0.001 55.1t10.4 42.4k11.8 25.4k5.68 <O*OOl

Values are means k SE. Magnitude of postmeal cortisol response was estimated by integrating cortisol secretory rates over duration of postmeal pulse.


Dow

nloaded from



Table 3. Correlations between postmeal responses in plasma glucose, serum insulin and insulin secretion, and postmeal cortisol secretion

Spearman Coefficient of Correlation With Postmeal Cortisol Secretion

Morning Evening

2-h AUC for plasma r = 0.491, P < 0.06 r = 0.241, NS glucose

2-h AUC for serum r = 0.562,P < 0.03 r = -0.206, NS insulin

2-h AUC for insulin r = 0.553, P < 0.04 r = 0.268, NS secretion

AUC, area under curve; NS, not significant.

average 57 t 19% higher than the morning response in the study with meals at 12-h intervals and 96 t 43% higher than the morning response in the study with meals at 6-h intervals. The corresponding values for women (n = 4) were 167 t 57% (P < 0.06 compared with men) and 172 k 214% (P = 0.17 compared with men). Similarly, for insulin secretion, the postmeal responses in men in the evening were 15 t 6 and 40 t 22% higher than in the morning in the studies with meals at 12- and 6-h intervals, respectively. In contrast, in women, the corresponding values were 40 t 16% (P < 0.06 compared with men) and 101 & 12% (P < 0.02 compared with men). These sex differences in magnitude of the diurnal variation in meal response of glucose and insulin secretion were paralleled by a sex difference in the magnitude of the morning-to-evening cortisol decrease, which was significantly larger in women than in men. Indeed, in women, the relative morning-to-evening decrease averaged 69 t 6% whereas in men, it was only 27 t 9% (P < O.Ol), when the decrease was considered to be zero in the two studies where cortisol levels were higher at 2000 than at 0800 h.

DISCUSSION

The present study was undertaken to determine whether glucose and insulin responses to meals are influenced by circadian timing. The protocol was specifically designed to control for a number of potentially confound- ing variables that could influence the outcome and inter- pretation of the data. To avoid differences in plasma glucose and insulin, which could result from differences in the type of food, subjects were given meals that were identical in composition and caloric content but were presented at different times of the day. The period of fasting before each meal was either 12 or 6 h, respectively, and was kept constant irrespective of the time of day at which the meal was consumed. An equal number of male and female subjects were studied. To address the possibility that the deterioration in glucose tolerance after ingestion of a mixed meal may only reach a sufficient magnitude to be clearly detectable later in the evening, our subjects were given the evening meal at 2000 rather than in the late afternoon as was done in previous studies.

The results clearly demonstrate that the glucose response to a meal is substantially greater when the meal is eaten in the evening rather than the morning. This

GLUCOSE kO.412

-150 -100 0 50 100 % DECREASE IN PRE-MEAL CORTISOL LEVEL

SERUM INSULIN c0.618

100 1

-l@+----+I ’ I 1 -150 -100 0 50 100

% DECREASE IN PRE-MEAL CORTISOL LEVEL

INSULIN SECRETION k0.506

“1

-150 -100 0 50 100

% DECREASE IN PRE-MEAL CORTISOL LEVEL

Fig. 5. Relationship between morning-to-evening decrease in premeal cortisol level (x-axis) and morning-to-evening increase in postprandial response in plasma glucose, serum insulin, and insulin secretion, as estimated by the 2-h area under curve. In 2 individual studies, premeal cortisol levels were higher at 2000 h than at 0800 h, resulting in negative value for morning-to-evening decrease.

marked effect of time of day is robust and, in the present study, all measures of the postprandial glucose response were greater in the evening than in the morning, including the maximum increment, total area under the curve, the area under the curve during the 2 h after meal presentation, and the duration of the glucose response. The effect of time of day was statistically significant irrespective of whether there was a 6- or 12-h period of fasting before administration of the meal, although there was a tendency for the differences to be greater with the shorter duration of fasting, which is more comparable with normal meal schedules. Surprisingly, the large increase in glucose response was not associated with a commensurate increase in insulin concentrations or insulin secretion rates. Thus, although a trend toward greater postprandial responses in plasma insulin in the


Dow

nloaded from



evening was apparent, the difference failed to reach statistical significance. Increased insulin clearance in the evening appeared to be partially responsible for the absence of circadian variation in magnitude of the plasma insulin response. Indeed, the areas under the curve for the insulin secretion rates derived by deconvolution of peripheral C-peptide were increased in the evening, but to a lesser extent than the postprandial glucose responses. The evening vs. morning differences in maximum increment and duration of the response were not significant. The failure of insulin secretion rates to increase in proportion to the observed increase in glucose raises the possibility that, in addition to diurnal changes in insulin sensitivity, there may be diurnal changes in the set point at which ,&cells respond to glucose, with reduced I respon siveness in the evening creased plasma glucose concentrations.

resulting in in-

Earlier studies had shown that meal ingestion may evoke short-term cortisol rises that are more pronounced when the meal is eaten around noon than in the evening (7, 16). The present results show that meal-related increases in cortisol levels also occur consistently in the morning and that the amount of cortisol secreted in response to a morning meal is positively correlated wi th the magnitude of the postprandial glucose and insulin responses. Furthermore, the magnitude of the morning- to-evening increase in postmeal insulin response was positively correlated with the magnitude of the morning- to-evening decline of cortisol levels. These findings are in agreement with our recent observation during constant glucose infusion of an inverse temporal relationship between the circadian variation of glucose levels and insulin secretion, on the one hand, and that of plasma cortisol, on the other hand (21). Under these experimental conditions, the amplitude of the diurnal variation in insulin secretion was significantly correlated with amplitude of the cortisol rhythm (21). Thus both studies suggest that the cortisol rhythm may mediate, at least partially, the diurnal variation in carbohydrate tolerance. However, thi .s concept is difficult to reconcile with current notions on cortisol counterregulatory mechanisms, which would predict that low cortisol levels in the evening would be associated with increased, rather than decreased, insulin sensitivity.

Other pathways by which the centrally generated circadian signal could modulate postprandial glucose and insulin responses include neural transmission and en-

small size (500 kcal) and were not of particularly high carbohydrate content (43%). With this type of meal, the glucose responses in the evening were approximately twofold larger than those observed in the morning. The careful studies of Service et al. (l7), who studied the effect of meal size, and of Nuttall et al. (12), who studied the effect of meal composition, indicate that the effect of time of day would have been even more pronounced if larger meals of higher carbohydrate content had been given. Taken together with the present studies suggest that, depending on meal position, the postprandial glucose response in the evening could be at least fivefold larger than in the morning.

Studies in human subjects living in isolation units

findings, size and

these com-

have shown correlated w

that the duration of intermeal intervals are ith the length of the “free-running” circadian

period (2), indicating a role for the circadian system in the timing of food ingestion in the absence of social constraints. Records of the spontaneous food intake of adults living in normal environments have revealed that the size and composition of meals also undergo diurnal changes, with increasing meal sizes and a shift from high carbohydrate to high fat and protein as the day progresses (4). The current analyses show that this intricate control of feeding behavior by circadian rhythmicity is matched by circadian variations in metabolic responses to meals. Further understanding of the mechanism link- ing circadian rhythmicity, feeding behavior, and metabolic responses to food could lead to the development of optimal dietary regimens for patients with impaired glucose tolerance and diabetes-as well as for individuals submitted to conditions of abnormal circadian rhythmicity, such as jet lag and shift work.

We thank C. Beebe for help in diet planning, Frank Borras for excellent technical assistance with the cortisol assay, Lan Anh Nguyen for preparing the illustrations, and the nursing staff of the General Clinical Research Center at the University of Chicago for skillful assistance in the completion of the studies.

This work was supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-31842, DK-20595, and DK-41814 and by Grant RR-00055 from the General Clinical Research Center at the University of Chicago.

Address for reprint requests: E. Van Cauter, Dept. of Medicine, Box 138, Univ. of Chicago, 5841 S. Maryland Ave., Chicago, IL 60637.

Received 28 November 1990; accepted in final form 8 November 1991.

REFERENCES

docrine factors outside of the corticotropic axis. Gas- 1. trointestinal hormones could be implicated, but, because the morning-to-evening differences in meal responses parallel the diurnal variations in glucose tolerance seen 2m after intravenous glucose administration (8, 19, 22), it appears unlikely that they are the primary cause of 3. circadian changes in meal tolerance. Diurnal variations in concentrations of counterregulatory hormones other than cortisol could conceivably be responsible for circa-

4 l

dian modulation of the set point of glucose regulation. However, peripheral levels of glucagon (19), growth hor- 5. mone (23), and catecholamines (9) do not undergo consistent diurnal variations during the waking portion of the 24-h cycle. 6.

-- The meals given in the present study were of relatively

Ahmed, M., M. C. Gannon, and F. Q. Nuttal. Postprandial plasma glucose, insulin, glucagon and triglyceride responses to a standard diet in normal subjects. Diubetologia 12: 61-67, 1976. Aschoff, J., C. von Goetz, C. Wildgruber, and R. A. Wever. Meal timing in humans during isolation without times cues. J. Biol. Rhythms 1: 151-162,1986. Bradley, E. M., and C. Waterhouse. Effect of estrogen administration on extravascular cortisol. J. Clin. Endocrinol. Aletab. 26: 705-714,1966. De Castro, J. M. Circadian rhythms of the spontaneous meal pattern, macronutrient intake, and mood of humans. Physiol. Be- huv. 40: 437-446,1987. Eaton, R. P., R. C. Allen, D. S. Schade, K. M. Erickson, and J. Standefer. Prehepatic insulin production in man: kinetic analysis using peripheral connecting peptide behavior. J. Clin. Endocrinol. Metub. 51: 520-528, 1980. Faber, 0. K., C. Binder, J. Markussen, L. G. Heding, V. K. Naithani, H. Kuzuya, P. Blix, D. L. Horwitz, and A. H.


Dow

nloaded from


CIRCADIAN VARIATION IN MEAL RESPONSES E475 Rubenstein. Characterization of seven C-peptide antisera. Dia- betes 27, Suppl. 1: 170-177, 1978.

7. Follenius, M., G. Brandenberger, B. Hietter, M. Simeoni, and B. Reinhardt. Diurnal cortisol peaks and their relationship to meals. J. Clin. Endocrinol. Metab. 55: 757-761, 1982.

8. Jarrett, R, J. Rhythms in insulin and glucose. In: Endocrine Rhythms, edited by D. Krieger. New York: Raven, 1979, p. 247- 258.

9. Linsell, C. R., S. L. Lightman, P. E. Mullen, M. J. Brown, and R. C. Causon. Circadian rhythms of epinephrine and nor- epinephrine in man. J. Clin. Endocrinol. Metub. 60: 1210-1215, 1985.

10. Malherbe, C., M. de Gasparo, R. de Hertogh, and J. Hoet. Circadian variations of blood sugar and plasma insulin levels in man. Diabetologiu 5: 397-404,1969.

11. Morgan, C., and A. Lazarow. Immunoassay of insulin: two antibody systems: plasma insulin levels of normal, subdiabetic and diabetic rats. Diabetes 12: 115-126,1963.

12. Nuttall, F. Q., M. C. Gannon, J. L. Wald, and M. Ahmed. Plasma glucose and insulin profiles in normal subjects ingesting diets of varying carbohydrate, fat and protein content. J. Am. Colt. Nutr. 4: 437-450,1985.

13. Owens, D., K. Wragg, P. Briggs, S. Luzio, G. Kimber, and C. Davies. Comparison of the metabolic response to a glucose tolerance test and a standardized meal test and the response to serial test meals in normal healthy subjects. Diabetes Cure 2: 409- 413,1979.

14. Polonsky, K. S., B. D. Given, and E. Van Cauter. Twenty- four-hour profiles and pulsatile patterns of insulin secretion in normal and obese subjects. J. Clin. Invest. 81: 442-448,1988.

15. Polonsky, K. S., J. Licinio-Paixao, B. D. Given, W. Pugh, P. Rue, J. Galloway, T. Karrison, and B. Frank. Use of

biosynthetic human C-peptide in the measurement of insulin secretion rates in normal volunteers and type I diabetic patients. J. Clin. Invest. 77: 98-105, 1986.

16. Quigley, M. E., and S. S. C. Yen. A mid-day surge in cortisol levels. J. Clin. Endocrinol. Metab. 49: 945-947, 1979.

17. Service, F. J., L. D. Hall, R. E. Westland, P. C. O’Brien, V. L. W. Go, M. W. Haymond, and R. A. Rizza. Effects of size, time of day and sequence of meal ingestion on carbohydrate tolerance in normal subjects. Diabetologiu 25: 316-321,1983.

18. Shapiro, E. T., H. Tillil, M. A. Miller, B. H. Frank, 3. A. Galloway, A. H. Rubenstein, and K. S. Polonsky. Insulin secretion and clearance: comparison after oral and intravenous glucose. Diabetes 36: 1365-1371, 1987.

19. Shapiro, E. T., H. Tillil, K. S. Polonsky, V. S. Fang, A. H. Rubenstein, and E. Van Cauter. Oscillations in insulin secretion during constant glucose infusion in normal man: relationship to changes in plasma glucose. J. Clin. Endocrinol. Metab. 67: 307- 314,1988.

20. Van Cauter, E. Estimating false-positive and false-negative errors in analyses of hormonal pulsatility. Am. J. Physiol. 254 (Endocri- nol. Metub. 17): E786-E794,1988.

21. Van Cauter, E., J. D. Blackman, D. Roland, J. P. Spire, S. Refetoff, and K. S. Polonsky. Modulation of glucose regulation and insulin secretion by circadian rhythmicity and sleep. J. Clin. Invest. 88: 934-942, 1991.

22. Van Cauter, E., D. Desir, C. Decoster, F. F&y, and E. 0. Balasse. Nocturnal decrease of glucose tolerance during constant glucose infusion. J. Clin. Endocrinol. Metab. 69: 604-611, 1989.

23. Van Cauter, E., and S. Refetoff. Multifactorial control of the 24-hour secretory profiles of pituitary hormones. J. Endocrinol. Inuest. 8: 381-391, 1985.


Dow

nloaded from


Documents

Circadian modulation of glucose and insulin responses to ......Circadian modulation of glucose and insulin responses to meals: relationship to cortisol rhythm EVE VAN CAUTER, E. TIMOTHY