| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Other Original Articles |
Section of Endocrinology and Metabolic Medicine (E.K., N.J.L., A.P., B.A.M., V.A., H.M., S.R., M.I.M., D.G.J.), Imperial College Faculty of Medicine, St. Mary’s Hospital, London W2 1PG, United Kingdom; Department for Biological and Medical Systems (K.H.P.), Imperial College of Science, Technology, and Medicine, London SW7 2BY, United Kingdom; and School of Biomedical Sciences (D.C.F., I.A.M.), Nottingham University Medical School, Queen’s Medical Centre, Nottingham NG7 2UH, United Kingdom
Address all correspondence and requests for reprints to: Desmond G. Johnston, M.D., Section of Endocrinology and Metabolic Medicine, Imperial College Faculty of Medicine at St. Mary’s Hospital, Norfolk Place, London W2 1PG, United Kingdom. E-mail: . d.johnston{at}ic.ac.uk
Abstract
We assessed postprandial thermogenesis (PPT) for 3 h following a mixed meal in 29 normoglycemic European women with previous gestational diabetes (GDM), compared with 37 control women. Given the potential role of catecholamines and insulin in the regulation of PPT, we assessed insulin and catecholamine responses to the meal. There was no significant difference between the two groups in resting energy expenditure, PPT (although lower in the GDM group), or catecholamine levels. However, we observed a difference in the shape of the PPT curve between groups, and by applying a mathematical model, there was a consistent delay in PPT, insulin, and noradrenaline responses to the meal in the GDM group (T: fitted time constant, geometric mean (95% confidence interval), TPPT 58 (47–72) vs. 42 (37–48) min, P = 0.006; Tins 32 (28–37) vs. 22 (19–27) min, P = 0.002; TNA 30 (23–38) vs. 18 (14–23) min, P = 0.01, respectively). Fidgeting activity during the study was assessed by a novel technique and was lower in the GDM group, resting [427 (381–477) vs. 511 (466–560) kJ/min, P = 0.02] but not postprandially. These delayed PPT, insulin, and noradrenaline responses to the meal in post-GDM women represent early metabolic changes. The decrease in fidgeting activity while resting, observed in the post-GDM group, may have physiological significance for energy balance.
GESTATIONAL DIABETES (GDM) is carbohydrate intolerance first recognized during pregnancy (1). Although most women with GDM return to normal glucose tolerance following delivery, they remain at substantially increased risk of type 2 diabetes in later life (2). GDM may therefore be considered a forerunner of type 2 diabetes, and women with previous GDM provide a valuable model for detection of early metabolic abnormalities associated with the development of type 2 diabetes.
Postprandial thermogenesis (PPT) represents the additional energy expenditure, above resting levels, that follows ingestion of food. PPT is known to decrease during normal pregnancy (3), and there is evidence that this decrease is accentuated in women with previous GDM and persists into the postpartum period (4). Similar decreases in PPT have been reported in both lean and obese subjects with type 2 diabetes (5, 6, 7). It has been suggested that the reduced energy expenditure that results from decreased PPT might provide one of the mechanisms by which individuals are predisposed to obesity and type 2 diabetes (8). The ability to maintain a lower energy requirement, especially during pregnancy, may have offered a survival advantage during human prehistory (8).
Postprandial thermogenesis is considered to have two components. The obligatory component encompasses the fixed costs of digesting, absorbing, processing, and storing nutrients, and the facultative component is of variable magnitude. The mechanisms regulating this facultative component are unknown, although the sympathetic nervous system is thought to play an important role (9). In addition, there seems to be a strong positive correlation between PPT and insulin sensitivity (4, 10).
Previous studies of PPT in women during and after pregnancy have been conducted on small subject groups of mixed ethnicity (4). In this study, our aim was to extend these previous studies by undertaking more extensive estimation and characterization of thermogenic responses in considerably larger groups of women of exclusively European origin following delivery. Although in this study the absolute reduction in PPT in nonpregnant women with previous GDM was not statistically significant, compared with controls, we did observe marked differences in PPT dynamics with the women with post-GDM showing a delayed metabolic and thermogenic response to the mixed meal.
Subjects and Methods
Experimental subjects
Between July 1997 and June 1999, 29 European women with previous GDM were recruited retrospectively (via antenatal care databases) from several West London Hospitals, a median (interquartile range) of 24 (16–48) months after the index pregnancy. Although the clinical criteria used for the diagnosis of GDM in the different hospitals show minor variations, only women who fulfilled World Health Organization 1999 criteria for glucose intolerance during pregnancy (glucose >7.8 mmol/liter following 75 g glucose) were considered to have had GDM for the purposes of this study (11). Five of these women required insulin treatment during the index pregnancy.
To ensure that all women with post-GDM were normoglycemic at the time of study, they were invited to attend for a 75-g oral glucose tolerance test (OGTT). This was not possible in four subjects. One of these had undergone a normal OGTT 6 months previously and refused to repeat. The other three subjects also refused an OGTT but had normal fasting glucose on at least two occasions and normal hemoglobin A1c. Normal glucose homeostasis was defined according to World Health Organization 1999 criteria, based on all available results (11). Ten women reported a family history of type 2 diabetes (first- or second-degree relatives).
Over the same period, 37 control women were recruited from the same sources 26 (17–49) months after the index pregnancy. All these women had successfully passed the modified screening test of O’Sullivan et al. (12) during the most recent pregnancy and had maintained normal glucose levels throughout all previous pregnancies. In addition, all control subjects had a fasting glucose less than 6.1 mmol/liter (11) and normal hemoglobin A1c at the time of this study. None of these women had a first-degree family history of type 2 diabetes.
The two groups were matched for ethnicity, parity (median of two in both groups), and time since delivery. Parents’ and grandparents’ place of birth, language, and religion were documented to limit recruitment to individuals with exclusively European ancestry for at least three generations. No woman in either group was pregnant or breast-feeding at the time of the study. Measurements of thermogenesis were made in both groups during the follicular phase of their cycle, or in the pill-free interval for those who were on the oral contraceptive (three women with GDM and eight control women). Anthropometric measures included body mass index (BMI) and waist-hip ratio. Lean body mass (LBM) was measured by near-infrared light spectroscopy, which correlates well with bioelectrical impedance and is less influenced by obesity (13). The study protocol had been approved by local ethics committee for each recruitment hospital. Informed consent was obtained from all subjects studied.
Materials and methods
Assessment of PPT following a mixed meal. Metabolic rate was measured by continuous indirect calorimetry (Deltatrac metabolic monitor, Datex Instrumentarium, Helsinki, Finland), using standard methods (14). After an overnight fast (10–12 h), an intravenous cannula was inserted in an arterialized hand vein. The subjects were acclimatized to the thermoneutral environment for at least 30 min and resting energy expenditure (REE) measured for 30 min after steady state was reached. A mixed meal of Build Up (Nestlé Clinical Nutrition, Surrey, UK) calculated to provide 42 kJ/kg of LBM was given. The energy load was 19% fat, 28% protein, and 53% carbohydrate. The use of a mixed meal was previously validated as a physiological thermogenic stimulus (15). The metabolic rate was measured for the next 3 h while the subject remained resting. The total PPT (in kilojoules) was calculated as the increment in energy expenditure over baseline (i.e. postmeal energy expenditure minus REE).
Three women with a history of GDM and eight control women became restless in the last hour of the test, and reliable data for these subjects were available for only the first 2 h. In this study, in which we report 180-min postmeal data, this refers only to those subjects (26 GDM, 29 control) who completed the full 3-h postmeal period. However, conclusions reached from analysis of the 120-min data on the whole group of subjects were in all respects similar to those based on 180-min data in this smaller set.
Samples for estimation of glucose, insulin, nonesterified fatty acids (NEFA), and catecholamines were taken at 30-min intervals during the study (including two baseline samples at -30 and 0 min). Poststimulus measures of glucose, insulin, and catecholamines were calculated as the area under the curve (AUC) using the trapezoid rule. Fasting samples were also taken for analysis of plasma lipids.
Biochemical assays. Serum samples for specific insulin measurement were stored at -20 C and then analyzed in duplicate. Specific insulin levels were measured by ELISA (16), using an assay with sensitivity of 0.55 pmol/liter and interassay specificity not exceeding 8.6% at low and high values. The cross-reactivity of proinsulin with the intact insulin assay is negligible. Glucose, cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides were measured on an AU5200 analyzer (Olympus Corp. Opticals UK Ltd., Eastleigh, Hants, UK). HDL cholesterol levels were assayed following selective precipitation of non-HDL lipoproteins using dextran sulfate and magnesium. NEFA were measured by enzymatic methods using a centrifugal analyzer (COBAS-BIO/Roche, Welwyn Garden City, UK). Catecholamines were measured by HPLC, as previously described (17).
Analysis of postmeal data. The first step in analyzing the energy expenditure data was to determine a baseline for each subject by averaging all energy expenditure readings obtained before the start of the meal. Calculations of PPT were based on the excursion [e(t)] of the measured energy expenditure data from this baseline average. Measures of PPT in subject groups were compared by ANOVA using SPSS for Windows (version 8.0, SPSS, Inc., Chicago, IL).
Although there was no absolute difference in PPT between groups, we observed a difference in the shape of the PPT curve (see Results). This was further investigated by using the indicial response equation (18), commonly used in engineering systems to analyze the temporal response to a step change in conditions as the basis for a mathematical model of metabolic and thermogenic responses to the meal stimulus.
The initial attempt at model fitting employed the indicial response equation, e(t) = A exp(-t/T1) - B exp(-t/T2). The physical interpretation of this model is an exponential increase (with time constant T1) in response to the step change, together with an exponential decrease resulting from a homeostatic control process (with time constant T2). The relative influences of these competing responses depend upon the multiplicative constants A and B. In our model, A and B were constrained so that the integral under the model curve was equal to the sum of the measured data. The time constants, T1 and T2, were fitted to the data using a nonlinear least squares fitting routine written in Matlab using a Nelder-Mead simplex (direct search) method (The MathWorks, Natick, MA).
Because, in almost all subjects, the two time constants were found to be virtually identical, we present here the data derived from the simpler, single-time constant model. In this model, e(t) = C t exp(-t/T) where T is the single time constant and C is a scaling constant. This model assumes a family of curves for different time constants T, normalized so that the each AUC is unity. The peak value for each curve occurs at t = T. It should be noted that this model assumes that the thermogenesis returns to the premeal level asymptotically, which was not true for all subjects. This assumption could be relaxed but only at the expense of including some further assumptions about the variation of the basal state during the course of the experiment. Because the choice, within reasonable limits, of the asymptotic final level does not have much effect on the calculated time constant, we opted for the simpler model to describe the experimental variations in thermogenesis. The values of T and C are fitted to the data as described above.
For comparison with the time constants calculated for PPT (TPPT), the same model was also used to provide time constants for glucose, insulin (Tins) and noradrenaline (TNA). Parameter estimates could not be obtained on a small number of subjects (see Table 2
), in which missing values or outlying data meant that the model could not be fitted.
|
Homeostatic model assessment (HOMA) parameters. Fasting glucose and specific insulin measures (the average from two samples taken with a 30-min collection interval) were used to derive estimates of ß-cell function (%B) and insulin sensitivity (%S) using the HOMA algorithm (19). This is a model of the glucose/insulin feedback system, which derives estimates of %B and %S from paired measurements of fasting glucose and insulin and expresses these standardized to a young, lean reference population.
Statistical analysis. Data are presented as median (interquartile range) or, where log transformation was appropriate, as geometric mean (95% confidence interval).
Statistical analyses were performed in SPSS 8.0 for Windows (SPSS, Inc.) using unpaired t test, Pearson’s correlation, and ANOVA, as appropriate.
Power calculations. We aimed to detect whether the difference in PPT reported by Robinson et al. (4) could be replicated. We estimated the differences in the means in PPT to be approximately14 kJ (in post-GDM subjects vs. control women) and the SD to be 18 kJ in the Robinson et al. study (4). We estimated that our study has 86% power (at P < 0.05, two-sided) to detect a between-group difference in mean PPT equivalent to that reported in the study of Robinson et al. (4).
Results
Baseline data
Baseline characteristics of the subject groups are given in Table 1. The two subject groups did not differ significantly for age, BMI, or waist/hip ratio. However, because BMI was somewhat higher in the GDM group, BMI was included as an optional covariate for subsequent analyses. There was no difference in fasting total cholesterol or HDL cholesterol between the two groups. Fasting triglycerides were significantly higher in the GDM group, whether or not the difference in BMI was allowed for. The %S was lower in the post-GDM group, but this difference was not statistically significant. The %B was similar in the two groups.
|
Energy expenditure measures in the two groups are given in Table 2
and Fig. 1. There were no significant differences in REE between the two groups, whether assessed in absolute terms or after correction for LBM (REE/LBM). Total PPT was lower in the GDM group [137 (97–192) vs. 187 (165–211) kJ], a difference that was close to significance (P = 0.052, one sided). PPT corrected for LBM tended in the same direction [2.5 (2.0–3.1) vs. 3.0 (2.4–3.7) kJ/kg of LBM] but did not reach statistical significance (P = 0.45). There was no correlation between total PPT and fasting insulin (r = 0.04, P = 0.75), AUC for insulin (r = -0.004, P = 0.97), and %S (r = -0.02, P = 0.88) in the whole group of subjects.
|
shows, the shape of the PPT curve differed markedly in the post-GDM women, compared with controls. At 30 min, averaged PPT rate was significantly lower in the post-GDM group [438 (287–668) vs. 740 (607–902) J/min, P = 0.04] (Fig. 1). Using the single-time constant model, the estimate of TPPT in post-GDM subjects was significantly longer than in controls [58 (47–72) vs. 42 (37–48) min, P = 0.006, Table 2], indicating that the peak PPT response occurred later in the GDM group. The fitted curves for the median values for postprandial energy expenditure are shown in Fig. 2.
|
). Metabolic data
Glucose and insulin levels during the study are shown in Fig. 3. There was no difference in fasting glucose between the two groups. The AUC for glucose and the 2-h glucose levels following the meal were higher in the GDM group (even though they were all normoglycemic, Fig. 3), and the difference remained significant when corrected for BMI [36.6 (35.2–38.1) vs. 33.0 (32.0–34.0) mmol/liter·h, P < 0.0001 and P < 0.0001; 6.0 (5.8–6.3) vs. 5.6 (5.4–5.8) mmol/liter, P = 0.008 and P = 0.02, respectively] (Table 2). There were no differences in fasting insulin or in the AUC for insulin between the two groups (Fig. 3 and Table 2). Noradrenaline levels before and following the meal are shown in Fig. 3. There were no differences in fasting or AUC catecholamine measures between the two groups (Tables 1 and 2). There was no difference in fasting NEFA (Table 1) or NEFA suppression at any time point following the meal between the two groups (data not shown).
|
). There was no significant difference in the time constant for glucose between the two groups. There was a significant correlation between TPPT and Tins (r = 0.29, P = 0.047) but not between TPPT and TNA (r = 0.19, P = 0.28).
Discussion
We studied a large group of European women with a history of GDM, a group known to be at increased risk of future type 2 diabetes (2, 20). These women were normoglycemic at the time of the study and were matched for ethnicity, age, parity, and time since delivery with a control population. There were no significant differences in REE between the two groups whether assessed in absolute terms or after correction for LBM. Although mean values of total PPT were lower in the GDM group, this difference did not quite attain statistical significance (P = 0.052 one-sided). However, we observed a marked difference in the shape of the PPT curve between the two groups, as quantified by the lower PPT rate at 30 min postprandially in the GDM group and the parameters derived from the mathematical model. Although the biological significance of the delayed PPT response is uncertain, one possibility is that this is an early metabolic manifestation that precedes an absolute decrease in PPT in these women with post-GDM.
Our observation of similar REE between women with post-GDM and control women is in agreement with previous observations (4). The lack of a significant difference in PPT between women with post-GDM and control women contrasts with our previous observations, which revealed a reduction in PPT not only during pregnancies complicated by GDM but also subsequently, when glucose homeostasis had normalized (4). These discrepant findings may reflect differences in the characteristics of the subjects enrolled in the two studies. The women in the previous study had more severe GDM: All had had at least two diabetic pregnancies and most had required insulin therapy during pregnancy (4), compared with only 5 of 29 in the current study. These clinical differences were supported by the metabolic findings. In the previous study, subjects and controls showed substantial differences in indices of %S, whereas fasting insulin and HOMA-%S did not differ between the two groups in the current report. This is likely to be the major factor, given that insulin action is an important modulator of the thermogenic response (4, 10). Finally, subjects in the earlier study were of mixed ethnicity, raising the possibility that ethnic differences in PPT contributed to the previous findings.
As well as the delay in PPT response, we also observed a delay in the insulin response to the meal in the GDM group. Individual patterns of PPT and insulin response to the meal were correlated: This may reflect a causal relationship, with the delayed metabolic response responsible for the delayed thermogenic response in subjects with GDM. This would be in keeping with the correlations observed between absolute measures of PPT and insulin action in normal subjects (10, 21) and women following GDM (4) and with evidence that PPT is generally reduced in insulin-resistant states, such as polycystic ovary syndrome (22) and obesity (5, 23, 24, 25). The causal relationship between insulin action and thermogenesis is further reinforced by the observation that, in obese individuals, experimental restoration of normal glucose disposal rates also normalizes this obesity-related reduction in glucose-induced thermogenesis (26).
It seems likely, given that PPT is reduced by ß-adrenergic blockade, that catecholamine action also modulates PPT (27). Although we were able to detect no evidence of a decreased catecholamine response to the meal, as with insulin, the noradrenaline response was delayed in the GDM group. Again, causality cannot be inferred from these correlations, but, given evidence that the sympathetic nervous system is stimulated by insulin (28), the most plausible explanation is that the catecholamine response follows the insulin response.
It is important to note that the current study was restricted to women with proven normal glucose homeostasis. This avoids any confounding effects of hyperglycemia itself (or its treatment) on the thermogenic and metabolic responses we have studied (29) and therefore should help us detect primary etiological events. At the same time, however, this means that we have inevitably focused on a subset of women with previous GDM likely to have the least marked metabolic defects. Nevertheless, the study population still showed sufficient evidence of disordered carbohydrate and lipid metabolism to mark them out as a group at increased future risk of metabolic decompensation. Fasting triglyceride and postprandial glucose levels were both higher in the women with post-GDM, suggesting subtle defects in insulin secretion and/or action (30), which we could not detect using the HOMA algorithm. Our data suggest that a delayed thermogenic and metabolic response to the meal stimulus is an additional feature of these women and one that is detectable at an early stage in the etiology of abnormal glucose homeostasis.
PPT represents only one component of overall energy balance, and recent attention has been drawn to the possible role of individual differences in nonexercise activity thermogenesis (NEAT: comprising all physical activities other than volitional exercise including fidgeting, spontaneous muscle contractions, and the activities of daily living). A decrease in NEAT has been observed in normoglycemic individuals showing the most marked weight gain during a period of overfeeding (31). In this study, we used a novel method to provide an index of fidgeting based on the indirect calorimetry measurements to suggest that women with GDM had reduced fidgeting, at least in the resting period. Although further studies are required in this area, these data suggest that decreased fidgeting may contribute to differences in energy balance and thereby to susceptibility to weight gain and diabetes in GDM women.
In summary, this group of normoglycemic women with post-GDM display early metabolic defects as demonstrated by subtle abnormalities in carbohydrate and lipid metabolism. At the same time, they display a subtle defect in the thermogenic response to the meal stimulus, manifested by the delay in PPT response. This shift in thermogenic response is mirrored by, and correlated with, similar delays in insulin and noradrenaline responses to the meal. Although no definite conclusions regarding causality can be drawn from these correlative findings, one plausible explanation is that the delayed insulin response is responsible (possibly with sympathetic response as intermediary) for the delayed thermogenic response and that these delayed responses are early markers of metabolic defects, preceding the reductions in absolute indices seen in other subject groups, which are, presumably, further advanced in disease progression.
Acknowledgments
We are grateful to Dr. A. Dornhorst, Dr. S. Vijayaraghavan, Dr. A. Hattersley, Dr. C. Baynes, Dr. C. Johnston, Dr. M. de Swiet, Prof. P. J. Steer, Dr. Ana Grenfell, Dr. H. M. Mather, Zoë Penn, N. Mahadevan, Judy Harvey, Kate Kiely, Fiona Pullen, Molly Nanka-Bruce, D. Walker, J. Cardon, Dr. P. Watkins, Prof. S. Amiel, and Prof. L. Regan for help in patient recruitment and to Alec Sanderson (HICOM, Surrey, UK) and the British Diabetic Association for database design. Built Up was kindly provided by Nestlé Clinical Nutrition, Surrey.
Footnotes
This work was supported by a project grant from the United Kingdom Medical Research Council and the Joint Research Standing Committee at St. Mary’s Hospital.
Abbreviations: AUC, Area under the curve; BMI, body mass index; e(t), excursion; GDM, gestational diabetes mellitus; HDL, high-density lipoprotein; HOMA, homeostatic model assessment; LBM, lean body mass; NEFA, nonesterified fatty acids; OGTT, oral glucose tolerance test; %B, ß-cell function; %S, insulin sensitivity; PPT, postprandial thermogenesis; REE, resting energy expenditure; Tglu, time constant for glucose; Tins, time constant calculated for insulin; TNA, time constant calculated for noradrenaline; TPPT, time constant calculated for PPT.
Received August 10, 2001.
Accepted April 4, 2002.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
