Longitudinal changes in glucose metabolism during pregnancy in obese women with normal glucose tolerance and gestational diabetes mellitus

Longitudinal changes in glucose metabolism during pregnancy in obese women with normal glucose tolerance and gestational diabetes mellitus Patrick M. Catalano, MD, Larraine Huston, MS, Saeid B. Amini, PhD, MBA, and Satish C. Kalhan, MD Cleveland, Ohio OBJECTIVE: This study prospectively evaluated the longitudinal changes in insulin sensitivity, insulin response, and endogenous (primarily hepatic) glucose production and suppression during insulin infusion in women with normal glucose tolerance (control) and gestational diabetes mellitus before and during a planned pregnancy. STUDY DESIGN: Eight control subjects and 7 subjects in whom gestational diabetes mellitus developed were evaluated with an oral glucose tolerance test, an intravenous glucose tolerance test, and hyperinsulinemic-euglycemic clamp with infusion of [6,6 2H2]glucose before conception and at 12 to 14 and 34 to 36 weeks’ gestation. Insulin response was estimated as the area under the curve during the intravenous glucose tolerance test. Basal endogenous glucose production was estimated from isotope tracer dilution during steady state with [6,6 2H2]glucose and suppression during insulin infusion. Insulin sensitivity to glucose was defined as the glucose infusion rate required to maintain euglycemia during steady-state insulin infusion. Body composition was estimated with hydrodensitometry. Data were analyzed with 2-way analysis of variance with repeated measures for 2 groups. RESULTS: There were increases in first-phase (P = .006) and second-phase (P = .0001) insulin responses in both groups with advancing gestation, but the increase in second-phase response was significantly greater (P = .02) in the gestational diabetes mellitus group than in the control group. Basal glucose production increased significantly (P = .0001) with advancing gestation, and there was resistance to suppression during insulin infusion in both groups (P = .0001). There was less suppression of endogenous glucose production however, in the gestational diabetes mellitus group than in the control group (P = .01). Insulin sensitivity decreased with advancing gestation in both groups (P = .0001), and there was lower insulin sensitivity in the gestational diabetes mellitus group than in the control group (P = .04). Significant decreases in insulin sensitivity with time (P = .0001) and between groups (P = .03) remained when the data were adjusted for differences in insulin concentration or residual hepatic glucose production. CONCLUSION: Obese women in whom gestational diabetes mellitus develops have a significant increase in insulin response but decreases in insulin sensitivity and suppression of hepatic glucose production during insulin infusion with advancing gestation with respect to a matched control group. These metabolic abnormalities in glucose metabolism are the hallmarks of type 2 diabetes, for which these women are at increased risk in later life. (Am J Obstet Gynecol 1999;180:903-16.)

Key words: Euglycemic clamp, gestational diabetes mellitus, glucose metabolism, insulin sensitivity, pregnancy

Gestational diabetes mellitus and obesity are common metabolic abnormalities encountered in routine obstetric practice. Gestational diabetes mellitus affects between From the Departments of Reproductive Biology and Pediatrics, Case Western Reserve University at MetroHealth Medical Center, and Rainbow Babies and Children’s Hospital. Supported by National Institutes of Health grant HD-22965, grant P-50HD-11089, and General Clinical Research Center grant M01-RR-080. Presented at the Seventeenth Annual Meeting of The American Gynecological and Obstetrical Society, Hot Springs, Virginia, September 3-5, 1998. Reprint requests: Patrick M. Catalano, MD, Department of Reproductive Biology, Case Western Reserve University at MetroHealth Medical Center, 2500 MetroHealth Dr, Cleveland, OH 44109. Copyright © 1999 by Mosby, Inc. 0002-9378/99 $8.00 + 0 6/6/97075

3% and 5% of all pregnant women and obesity, defined as weight >15% of desirable weight, affects 32 million women, or a third of the adult population in the United States.1 There is a significant increase in obstetric morbidity associated with both these conditions, such as increased risk of preeclampsia, cesarean delivery, fetal macrosomia, and long-term risk of diabetes and hypertension. Evaluation of the longitudinal changes in carbohydrate metabolism in obese women with normal glucose tolerance and gestational diabetes mellitus should improve our understanding of carbohydrate metabolism in these women and may provide a basis for therapeutic strategies to prevent obstetric complications. In addition 903

904 Catalano et al

April 1999 Am J Obstet Gynecol

Table I. Pregravid maternal characteristics of subjects with gestational diabetes mellitus and control subjects Control group (n = 8) Age (y) Weight (kg) Body mass index (kg/m2) Fat-free mass (kg) Fat mass (kg) Body fat (%) Parity Time to conception (mo)

Gestational diabetes mellitus Statistical group (n = 7) significance

31.6 ± 11.4 71.4 ± 14.9 28.2 ± 5.0

29.9 ± 17.1 78.2 ± 19.6 31.0 ± 7.6

P = .38 P = .64 P = .40

49.4 ± 6.1 22.0 ± 9.4 30.0 ± 5.3 1.0 ± 0.3 2.1 ± 1.9

51.7 ± 7.5 26.5 ± 12.5 32.6 ± 6.6 1.0 ± 0.3 2.5 ± 1.7

P = .55 P = .47 P = .41 P = .99 P = .75

Data are mean ± SD.

alterations in maternal carbohydrate metabolism during pregnancy may also serve as a paradigm for the development of type 2 diabetes mellitus, for which these women are at significantly increased risk in later life.2 Previous studies have reported either no difference3 or lower insulin sensitivity4 and lower insulin response in women with gestational diabetes mellitus4 than in women with normal glucose tolerance. However, many of these studies were either cross-sectional in nature or relied on postpartum measurements to assess pregravid maternal metabolic status. In addition matching for maternal pregravid weight or body mass index was not accounted for in any of these reports. We previously reported on the longitudinal changes in carbohydrate metabolism in lean women with normal glucose tolerance and gestational diabetes mellitus.5 We found that in lean women in whom gestational diabetes mellitus developed there was a decrease in insulin sensitivity primarily before conception and in early gestation, in contrast to a decrease in insulin response and suppression of endogenous (primarily hepatic) glucose production in later gestation with respect to a matched control group. Therefore the primary purpose of this study was to prospectively evaluate the longitudinal changes in peripheral insulin sensitivity, insulin response, and endogenous glucose production and suppression during insulin infusion in obese women with normal glucose tolerance and gestational diabetes mellitus before and during a planned pregnancy. We hypothesized that women in whom gestational diabetes mellitus develops have decreased insulin sensitivity, insulin response, and suppression of hepatic glucose production with respect to the changes in glucose metabolism that occur during the course of normal gestation in a matched control group. Material and methods Subjects. This study was conducted in the General Clinical Research Center at the University of Vermont (n

= 4) and MetroHealth Medical Center, Case Western Reserve University (n = 11). The study was approved by the hospitals’ institutional review boards and informed consent was obtained from each subject before the study. Fifteen otherwise healthy obese women without a history of diabetes mellitus were recruited to participate in the study. We defined obesity as pregravid body fat >25%. None of the study subjects had diabetes mellitus before conception according to oral glucose tolerance.6 Seven of the women were at high risk for development of gestational diabetes mellitus because of gestational diabetes mellitus in a previous pregnancy (n = 4), a strong family history of type 2 diabetes in a first-degree relative (n = 2), or an impaired glucose tolerance before conception plus a first-degree relative with type 2 diabetes (n = 1). Other than gestational diabetes mellitus, all these pregnancies were uncomplicated. Eight other women were recruited, none of whom had a history of abnormal glucose tolerance either before or during a previous pregnancy. The father of 1 woman had a history of type 2 diabetes but she had shown a normal glucose tolerance in a previous pregnancy. None of the 15 subjects were breast-feeding, using hormonal contraception, taking other medications that might affect glucose metabolism, or using tobacco. All subjects were planning to conceive as soon as the baseline pregravid studies were completed. The 8 subjects with normal glucose tolerance in a previous pregnancy, no family history of type 2 diabetes, or both all had normal oral glucose tolerance before6 and throughout pregnancy.7 These 8 subjects comprised our control group. The 7 subjects at high risk for gestational diabetes mellitus because of family history, previous history of gestational diabetes mellitus, or both all had ≥2 abnormal values on an oral glucose tolerance test (GTT) during pregnancy7 and comprised our gestational diabetes mellitus group. The 2 groups were recruited to avoid significant differences in other factors that might affect glucose metabolism (Table I). Experimental protocol. A series of experiments was initiated before conception. Twelve of the 15 subjects were evaluated in the follicular phase of the menstrual cycle and 3 were evaluated in the luteal phase. The series of experiments was repeated in early gestation (12-14 weeks) and late gestation (34-36 weeks). Each subject was instructed in a standard dietary regimen 2 weeks before each study period. This dietary regimen was designed to standardize nutritional intake for each subject, to maintain weight before conception, and to allow appropriate increases in weight during pregnancy. The regimen was identical to that used in the treatment of gestational diabetes mellitus at our institution. At the time of diagnosis of gestational diabetes mellitus all subjects were asked to perform self-monitoring of blood glucose levels. Target fasting and preprandial glucose values were <100 mg/dL and target postprandial values were <120 mg/dL. One subject with gestational diabetes mellitus eventually re-

Volume 180, Number 4 Am J Obstet Gynecol

quired insulin therapy because euglycemia could not be achieved with dietary control alone. Physical activity was encouraged throughout gestation and was estimated by means of the Minnesota Leisure Time Physical Activity Questionnaire.7a The following sets of studies were performed on 3 separate days: (1) oral GTT, routine laboratory studies, and islet cell antibody assay; (2) hydrodensitometry and intravenous GTT; and (3) the hyperinsulinemic-euglycemic clamp with infusion of the stable isotope [6,6 2H2]glucose to estimate basal endogenous glucose production and response to insulin infusion. Oral GTT. Before conception all subjects were administered a 75-g oral GTT, as defined by the National Diabetes Data Group.6 During pregnancy all subjects were given a 100-g oral GTT. Normal glucose tolerance during gestation was defined according to the criteria of Carpenter and Coustan.7 The venous plasma glucose concentrations were measured with the glucose oxidase method with a Yellow Springs (Yellow Springs Instruments, Yellow Springs, Ohio) glucose analyzer. Intravenous GTT. For subjects <120% ideal body weight we infused 0.5 g/kg glucose as a bolus during 3 minutes, and samples of glucose and insulin were obtained at baseline (0 minutes) and at 1, 3, 5, 10, 15, 30, 45, and 60 minutes after the completion of the infusion. For those subjects whose weight was >120% ideal body weight, the glucose bolus was 19 g/m2 body surface area. The dissociation constant (Kt value), or fractional rate of disappearance of glucose from the circulation, was estimated. By means of the trapezoidal rule the first-phase insulin secretory capacity was calculated as the area under the insulin curve from 0 to 5 minutes and the second-phase insulin secretory capacity was calculated as the area under the curve from 5 to 60 minutes. Insulin concentrations were measured with a double-antibody method of Jaffe and Berman.8 Insulin was bound to an anti-insulin primary antibody (guinea pig, ICN 65-101) and the complex was precipitated by addition of anti–guinea pig immunoglobulin G (Antibodies, Inc, Davis, Calif) and a precipitating agent. Percentage binding was compared against insulin standards and concentrations were determined in microunits per milliliter. The assay is linear, from 0 to 100 µU/mL, and the minimal measure of sensitivity is 2.5 µU/mL. The intra-assay coefficient of variation is 6% and the interassay coefficient of variation is 8%. Body composition analysis. Each subject’s body composition was estimated by underwater weighing with adjustment for residual lung volume by helium or nitrogen dilution. Percentage of body fat was calculated according to the equations of Keys and Brozek9 to estimate fat-free mass. Hyperinsulinemic-euglycemic clamp and infusion of [6,6 2H ]glucose. The hyperinsulinemic-euglycemic clamp was 2 performed after an 11-hour fast as described by DeFronzo et al10 to estimate peripheral insulin sensitivity. At 6 AM an

Catalano et al 905

intravenous catheter was placed in an antecubital vein for infusion of glucose, insulin, and [6,6 2H2]glucose (Cambridge Isotope Laboratories, Andover, Mass). Another catheter was placed retrograde in the dorsal vein of the contralateral hand for blood withdrawal and warmed to 70°C with a warming box. A primed constant infusion of [6,6 2H2]glucose was infused at 0.133 mL/min at a concentration intended to achieve approximately 1.0 molar percentage excess for all subjects. The infusion of [6,6 2H ]glucose was continued for 2 hours and plasma sam2 ples were obtained at 90, 100, 110, and 120 minutes after priming to estimate basal glucose turnover and basal insulin concentrations. Basal endogenous glucose production was estimated according to the steady-state equations of Steele.11 The rate of appearance of glucose was estimated according to the following equation: Rate of appearance of glucose = F × (100 – 1)/E

in which F is the constant isotope infusion rate and E is the enrichment of the plasma glucose. We assumed that under steady-state conditions the rate of appearance of glucose was equal to the rate of disposal. At the completion of the 2-hour infusion we began insulin infusion to commence the euglycemic clamp. The first 5 subjects (n = 3 control subjects and n= 2 subjects with gestational diabetes mellitus) to complete the protocol had a primed constant infusion of Humulin (Lilly, Indianapolis, Ind) insulin at 40 mU · m–2 min–1 to achieve a plasma insulin concentration of approximately 100 µU/mL. Plasma glucose level was maintained at 90 mg/dL for 120 minutes by sampling the plasma glucose level every 5 minutes and varying the infusion of 20% glucose according to a computer program. Plasma was collected for measurement of insulin concentrations every 10 minutes during the last 40 minutes of the procedure. The insulin concentration achieved during the clamp was defined as the mean of the values obtained during the last 40 minutes of the procedure. The amount of glucose infused was calculated for each 10-minute interval and averaged for the last 40 minutes of the 2-hour infusion. This value was used to estimate glucose disposal as glucose uptake in peripheral tissues under steady-state conditions. The next 10 subjects to complete the protocol (n = 5 control subjects and n = 5 subjects with gestational diabetes mellitus) completed a 2-stage hyperinsulinemic-euglycemic clamp in which insulin was infused at 20 mU · m–2 · min–1 for 2 hours followed immediately by infusion of 40 mU · m–2 · min–1 insulin for 2 hours. The lower-dose insulin infusion was aimed at achieving plasma insulin concentrations of approximately 50 µU/mL. Plasma glucose concentrations were maintained at 90 mg/dL and the plasma insulin and glucose infusion rates were calculated as described previously during the last 40 minutes of the insulin infusion rate of 20 mU · m–2 · min–1. The primary purpose of the 2-dose insulin infusion clamp was to determine whether the slope of

906 Catalano et al

April 1999 Am J Obstet Gynecol

the change in insulin sensitivity (ratio of change in glucose infusion rate to change in insulin concentration) from the low-dose to the high-dose insulin infusion changed significantly during gestation or was significantly different between control subjects and subjects with gestational diabetes mellitus. A decrease in insulin receptors results in a shift to the right in the dose-response curve.12 A secondary purpose of the 2-level insulin infusion was to determine whether lower doses of insulin resulted in less suppression of hepatic glucose production in women with normal glucose tolerance in pregnancy as compared with pregravid measurements, thereby providing evidence for decreased hepatic insulin sensitivity in normal pregnancy. Endogenous glucose production during the insulin clamp was estimated by discontinuing the constant infusion of [6,6 2H2]glucose 15 minutes after the start of the clamp and by the addition of a known amount of labeled glucose to the 20% glucose infusion. The rationale for adding the tracer to the 20% glucose infusion was to avoid rapid changes in enrichment of plasma glucose during the clamp procedure. The infused 20% glucose enriched with [6,6 2H2]glucose aids in maintaining the molar percentage excess at a steady-state level, as described by Finegood et al.13 The following equation was used to correct turnover rate calculations: P = F × (100/E – 1)

where P is the turnover rate in milligrams per kilogram per minute, E is the isotope enrichment, and F is the infusion rate in milligrams per kilogram per minute. The stable isotopes were prepared for analysis by mass spectroscopy by isolating glucose from the plasma through ion-exchange chromatography. A pentacetate derivative of glucose was prepared as described by Tserng and Kalhan.14 Standard glucose solutions of known deuterium enrichment were run to calibrate for instrument variation. Plasma enrichment was determined with a gas chromatograph mass spectrometer (model 5985B; Hewlett-Packard Company, Palo Alto, Calif). Insulin sensitivity was estimated as the glucose infusion rate required to maintain plasma glucose constant at 90 mg/dL during the clamp. Because there was not complete suppression of endogenous glucose production during insulin infusion and plasma insulin concentrations varied during the clamp across time and between groups, we estimated the insulin sensitivity index to account for these variations. We defined the insulin sensitivity index as the glucose infusion rate plus any residual endogenous glucose production divided by the mean insulin concentration achieved during the clamp; that is, the ratio of total glucose uptake to the mean insulin concentration. Metabolic clearance rate of insulin. The metabolic clearance rate of insulin was calculated as the infusion rate of insulin under steady-state conditions divided by

the steady-state insulin concentration multiplied by fasting or basal insulin corrected for suppression of C peptide during the clamp: MCRI = I/SSI – FI × (SSC peptide/FC peptide)

where MCRI is the metabolic clearance rate of insulin, I is insulin infusion rate during the clamp, SSI is the insulin concentration during steady-state infusion during the clamp, FI is the fasting or basal insulin concentration, SSC peptide is the steady-state C peptide concentration during the clamp, and FC peptide is the fasting C peptide concentration. Basal or fasting insulin level was corrected for the steady-state and fasting C peptide ratio because infused insulin was unable to completely suppress the basal insulin production and because of the inability of our insulin assay to distinguish between infused from endogenous insulin. C peptide was measured with a commercially available double-antibody radioimmunoassay method (Diagnostics Products Corp, Los Angeles, Calif). The assay is designed for quantitative determination of C peptide concentrations in serum, plasma, and urine and is linear from 0 to 25 ng/mL and sensitive to 0.22 ng/mL. The intra-assay coefficient of variation is 4% and the interassay coefficient of variation is 6%. Islet cell antibodies were measured with a specific monoclonal antibody at the Joslin Diabetes Center.15 Statistical analysis. On the basis of our previous studies noting a 21% difference in pregravid insulin sensitivity between obese control subjects and subjects with gestational diabetes mellitus, we performed a power analysis and determined that 7 control subjects and 7 subjects with gestational diabetes mellitus were needed to achieve an 80% power to find a significant change in insulin sensitivity at the P = .05 level across time. Similar numbers of subjects were needed to achieve an 80% power to show a significant change in hepatic glucose production at the P = .05 level across time. Each of the studies in this protocol was repeated with all subjects during early and late pregnancy, with the exception of a hyperinsulinemic-euglycemic clamp that was not repeated in 1 control subject during late gestation. The clamp data for this subject were estimated from 2 independent variables, pregravid and early pregnancy clamp data, and regression analysis from the other control subjects. Data were expressed as mean ± SD. Statistical analysis for differences in pregravid maternal characteristics were performed with the unpaired Student t test, the Mann-Whitney U test, and χ2 analysis. The longitudinal changes in insulin response, basal endogenous glucose production, and insulin-stimulated glucose uptake during the clamp and the metabolic clearance rate of insulin were analyzed with 2-way analysis of variance with repeated measures for 2 groups. These analyses allowed us to evaluate the longitudinal changes in a given patient across time, differences between groups, and time-group interactions. Differences

Catalano et al 907

Volume 180, Number 4 Am J Obstet Gynecol

Fig 1. Longitudinal changes in plasma glucose response to oral GTT (mean ± SD). GDM, Gestational diabetes mellitus. Pregravid challenge was 75 g; early and late pregnancy challenges were 100 g. Individual longitudinal changes with time averaged P = .001; difference between groups was P = .0006.

between groups at specific times were analyzed with profile analysis. Areas under the curve were compared with 2-sample t tests or the Mann-Whitney U test. Statistical analyses were performed with the Statview II statistical package (Abacus Concepts, Berkeley, Calif) and the Statistical Analysis System (SAS Institute Inc, Cary, NC). P < .05 was considered significant.

Fig 2. Longitudinal changes in plasma insulin response to oral GTT (mean ± SD). GDM, Gestational diabetes mellitus. Individual longitudinal changes with time averaged P = .002; difference between groups was P = .08.

Results No subjects had diabetes mellitus at the pregravid testing. All subjects had normal oral GTT results before conception, except for 1 subject with gestational diabetes mellitus who had an impaired oral GTT result. All subjects had normal results of routine renal, thyroid, and liver chemistry studies as well as negative findings of islet

908 Catalano et al

Fig 3. A, Longitudinal changes in first-phase insulin response during intravenous GTT (mean ± SD). B, Longitudinal changes in second-phase insulin response during intravenous GTT (mean ± SD). GDM, Gestational diabetes mellitus; Pt, individual longitudinal changes with time; Pg, difference between groups.

cell antibodies during each period of study. There were no significant differences in age, weight, body composition, body mass index (Weight/Height2), parity, or months to conception between the 2 study groups (Table I). We defined obesity as having >25% body fat, and all subjects had >25% body fat before conception. Gestational age was confirmed in all cases with obstetric ultrasonography at 12 to 14 weeks’ gestation. There were no significant differences in the caloric intake in study subjects across time or between groups. In addition there were no significant differences in the percentages of protein, fats, and carbohydrates in their diets across time or between groups, as demonstrated by a 3day dietary record. Although physical activity decreased

April 1999 Am J Obstet Gynecol

Fig 4. A, Longitudinal changes in basal hepatic glucose production expressed in milligrams per minute (mean ± SD). B, Longitudinal changes in basal hepatic glucose production expressed in milligrams per kilogram fat-free mass per minute (mean ± SD). GDM, Gestational diabetes mellitus; Pt, individual longitudinal changes with time.

significantly (P = .01) with advancing gestation, there was no significant difference between groups. The results of the oral GTTs are shown in Figs 1 and 2. Fig 1 depicts the changes in plasma glucose levels in the 2 study groups across time. As expected, there was a significant increase in the areas under the glucose curves across time in both groups (P = .0001) and the area under the curve was greater in the gestational diabetes mellitus group than in the control group (P = .0006). Fig 2 depicts the changes in plasma insulin response during the oral GTT in the 2 study groups across time. There was a significant increase in the areas under the insulin curves across time in both groups (P = .002) but only a

Volume 180, Number 4 Am J Obstet Gynecol

trend toward a greater insulin response in the gestational diabetes mellitus group than in the control group (P = .08) because of the wide variation in insulin responses. This may also reflect a β error associated with small sample size. There was no significant (P = .45) difference in the Kt value across time or between control (pregravid Kt of 1.29 ± 0.59, early pregnancy Kt of 1.38 ± 0.46, and late pregnancy Kt of 1.16 ± 0.18) and gestational diabetes mellitus (pregravid Kt of 1.20 ± 0.57, early pregnancy Kt of 1.15 ± 0.22, and late pregnancy Kt of 1.06 ± 0.39) groups. The results of the first- and second-phase insulin responses are shown in Fig 3, A and B. There was a significant increase in first-phase insulin response across time (P = .006) in both the control (pregravid insulin response of 285 ± 163 µU/mL, early pregnancy insulin response of 397 ± 145 µU/mL, and late pregnancy insulin response of 447 ± 198 µU/mL) and gestational diabetes mellitus (pregravid insulin response of 299 ± 168 µU/mL, early pregnancy insulin response of 525 ± 231 µU/mL, and late pregnancy insulin response of 479 ± 331 µU/mL) groups, but there was no significant (P = .53) difference in first-phase insulin response between groups. Second-phase insulin response (Fig 3, B) was significantly (P = .0001) increased across time in both control (pregravid insulin response of 1478 ± 710 µU/mL, early pregnancy insulin response of 2130 ± 728 µU/mL, and late pregnancy insulin response of 3420 ± 1030 µU/mL) and gestational diabetes mellitus (pregravid insulin response of 3336 ± 2478 µU/mL, early pregnancy insulin response of 4288 ± 2408 µU/mL, and late pregnancy insulin response of 5399 ± 2400 µU/mL). The increase in second-phase insulin response was significantly (P = .02) greater in women with gestational diabetes mellitus than in the control group. The results of the basal endogenous glucose production changes across time are shown in Fig 4, A and B. There was a significant (P = .0001) 30% increase in total basal hepatic glucose production by late gestation in both control (pregravid production of 135 ± 20 mg/min, early pregnancy production of 140 ± 14 mg/min, and late pregnancy production of 175 ± 23 mg/min) and gestational diabetes mellitus (pregravid production of 132 ± 8 mg/min, early pregnancy production of 144 ± 15 mg/min, and late pregnancy production of 173 ± 20 mg/min) groups but there was no significant (P = .95) difference between groups (Fig 4, A). When basal hepatic glucose production was expressed in milligrams per kilogram fat-free mass per minute, there remained a significant (P = .0001) 16% increase in basal hepatic glucose production by late gestation in both control (pregravid production of 2.74 ± 0.29 mg · kg–1 fat-free mass · min–1, early pregnancy production of 2.82 ± 0.26 mg · kg–1 fatfree mass · min–1, and late pregnancy production of 3.12 ± 0.37 mg · kg–1 fat-free mass · min–1) and gestational di-

Catalano et al 909

Fig 5. Longitudinal changes in basal or fasting insulin concentration in milliunits per milliliter (mean ± SD) GDM, Gestational diatetes mellitus; Pt, individual longitudinal changes with time; Pg, difference between groups; Ptg, time-group interaction.

abetes mellitus (pregravid production of 2.59 ± 0.36 mg · kg–1 fat-free mass · min–1, early pregnancy production of 2.79 ± 0.37 mg · kg–1 fat-free mass · min–1, and late pregnancy production of 3.05 ± 0.38 mg · kg–1 fat-free mass · min–1) groups, and again there were no significant (P = .48) differences between groups (Fig 4, B). Moreover these changes in basal hepatic glucose production occurred despite a significant (P = .0006) increase in fasting or basal insulin concentrations across time in both control (pregravid insulin concentration of 8.8 ± 3.6 µU/mL, early pregnancy insulin concentration of 7.4 ± 5.6 µU/mL, and late pregnancy insulin concentration of 10.2 ± 5.5 µU/mL) and gestational diabetes mellitus (pregravid insulin concentration of 14.9 ± 10.2 µU/mL, early pregnancy insulin concentration of 16.0 ± 10.7 µU/mL, and late pregnancy insulin concentration of 29.1 ± 11.8 µU/mL) groups (Fig 5). In addition, in the gestational diabetes mellitus group there was a significant (P = .009) increase in fasting insulin concentration as well as a significant (P = .006) time-group interaction in comparison with the control group. The results of the glucose infusion rates during the hyperinsulinemic-euglycemic clamp are shown in Fig 6, A and B. There was a significant (P = .0001) 40% to 50% decrease in the glucose infusion rate required to maintain euglycemia in both groups during insulin infusion of 20 mU · m–2 · min–1 during gestation; control (pregravid glucose infusion rate of 5.01 ± 1.62 mg · kg–1 · min–1, early pregnancy glucose infusion rate of 4.84 ± 1.70 mg · kg–1 · min–1, and late pregnancy glucose infusion rate of 2.43 ± 0.36 mg · kg–1 · min–1) and gestational diabetes mellitus (pregravid glucose infusion rate of 2.72 ± 1.56 mg · kg–1 · min–1, early pregnancy glucose infusion rate of 2.76 ± 1.10

910 Catalano et al

April 1999 Am J Obstet Gynecol

Fig 7. Longitudinal changes in percentage suppression of hepatic glucose production during 40 mU · m–2 · min–1 insulin infusion (mean ± SD). GDM, Gestational diabetes mellitus; Pt, individual longitudinal changes with time; Pt, difference between groups.

Fig 6. A, Longitudinal changes in glucose infusion rate during 20 mU · m–2 · min–1 insulin infusion (mean ± SD). B, Longitudinal changes in glucose infusion rate during 40 mU · m–2 · min–1 insulin infusion (mean ± SD).GDM, Gestational diabetes mellitus; Pt, individual longitudinal changes with time; Pg, difference between groups.

mg · kg–1 · min–1, and late pregnancy glucose infusion rate of 1.58 ± 0.71 mg · kg–1 · min–1). The insulin stimulated glucose uptake on the basis of the glucose infusion rate required to maintain euglycemia was significantly (P = .04) less in the gestational diabetes mellitus as compared with the control group, indicating decreased insulin sensitivity in the gestational diabetes mellitus group, Fig 6, A. During the insulin infusion rate of 40 mU · m–2 · min–1, there was again a significant (P = .0001) 50% decrease in the glucose infusion rate required to maintain euglycemia in both groups; control (pregravid 7.76 ± 2.36, E 7.30 ± 2.33, late pregnancy 3.86 ± 0.76 mg/kg/min) and gestational diabetes mellitus (pregravid 5.43 ± 2.35, E 5.15 ± 1.77, late

pregnancy 2.63 ± 0.88 mg/kg/min). The insulin-stimulated glucose uptake derived from the glucose infusion rate required to maintain euglycemia was again significantly (P = .045) less in the gestational diabetes mellitus group than in the control group (Fig 6, B). The results were essentially the same when the glucose infusion rate was adjusted for milligrams per kilogram fat-free mass per minute, except that the differences between control and gestational diabetes mellitus groups were of only borderline significance for the insulin infusion rates of 20 and 40 mU · m–2 · min–1 (P = .066 and P = .051, respectively, data not shown). The slopes of the dose-response curves fitted to individual subjects’ data points for insulin sensitivity from the 20 to 40 mU · m–2 · min–1 insulin infusion were calculated in 10 subjects who had both levels of insulin infusion. There was no significant change in the slope of the dose-response curve with advancing gestation. However, the slopes of the dose-response curve for insulin sensitivity were significantly less (P = .03) in the gestational diabetes mellitus group (pregravid slope of 0.098 ± 0.055, early pregnancy slope of 0.107 ± 0.047, and late pregnancy slope of 0.063 ± 0.029) than in the control group (pregravid slope of 0.130 ± 0.043, early pregnancy slope of 0.155 ± 0.39, and late pregnancy slope of 0.145 ± 0.063). There was no significant difference in the suppression of endogenous glucose production (approximately 90%) across time or between control and gestational diabetes mellitus groups during insulin infusion of 20 mU · m–2 · min–1. During the insulin infusion of 40 mU · m–2 · min–1 (Fig 7), however, there was a significant decrease in suppression in endogenous glucose production across time (P = .0001) and less suppression in the gestational dia-

Volume 180, Number 4 Am J Obstet Gynecol

betes mellitus group (pregravid suppression of 96% ± 5%, early pregnancy suppression of 94% ± 6%, and late pregnancy suppression of 85% ± 6%) than in the control group (pregravid suppression of 97% ± 5%, early pregnancy suppression of 98% ± 3%, and late pregnancy suppression of 93% ± 2%; P = .01). There remained a significant decrease in suppression of endogenous glucose production across time (P = .02) and less suppression in the gestational diabetes mellitus group than in the control group (P = .02) when the data were expressed in milligrams per kilogram fat-free mass per minute. There were no significant differences in the insulin concentrations achieved during the clamp across time or between groups during insulin infusion of 20 mU · m–2 · min–1. In contrast, there was a significant (P = .0001) decrease across time with advancing gestation in insulin concentrations achieved with the clamp during the insulin infusions of 40 mU · m–2 · min–1. Additionally, the gestational diabetes mellitus group had significantly (P = .02) greater insulin concentrations (pregravid insulin concentration of 123 ± 32 µU/mL, early pregnancy insulin concentration of 103 ± 24 µU/mL, and late pregnancy insulin concentration of 105 ± 31 µU/mL) than those of the control group (pregravid insulin concentration of 94 ± 15 µU/mL, early pregnancy insulin concentration of 80 ± 15 µU/mL, and late pregnancy insulin concentration of 71 ± 14 µU/mL). In an effort to explain these differences in insulin concentrations achieved during the clamp despite identical insulin infusion rates, we estimated the metabolic clearance rate of insulin during the clamp. There was a significant increase in the metabolic clearance rate of insulin with advancing gestation at insulin infusion rates of both 20 (P = .01) and 40 (P = .0001) mU · m–2 · min–1 in both groups. There was also a trend (P = .052) for the metabolic clearance rate of insulin during the insulin infusion of 40 mU · m–2 · min–1 to be greater in the control group (pregravid clearance of 453 ± 73 mL · m–2 · min–1, early pregnancy clearance of 553 ± 98 mL · m–2 · min–1, and late pregnancy clearance of 657 ± 131 mL · m–2 · min–1) than in the gestational diabetes mellitus group (pregravid clearance of 362 ± 89 mL · m–2 · min–1, early pregnancy clearance of 455 ± 101 mL · m–2 · min–1, and late pregnancy clearance of 534 ± 145 mL · m–2 · min–1). Furthermore, when evaluated before pregnancy there was a significantly greater metabolic clearance of insulin in the control group than in the gestational diabetes mellitus group at insulin infusion rates of both 20 (P = .03) and 40 (P = .046) mU · m–2 · min–1. Because there were significant differences in suppression of hepatic glucose production and insulin concentrations achieved during the hyperinsulinemic-euglycemic clamp at insulin infusions of both 20 and 40 mU · m–2 · min–1, we calculated the insulin sensitivity index for both control and gestational diabetes mellitus groups (Fig 8, A and B). There was a significant (P = .0002) decrease in the

Catalano et al 911

Fig 8. A, Longitudinal changes in insulin sensitivity index during 20 mU · m–2 · min–1 insulin infusion (mean ± SD). B, Longitudinal changes in insulin sensitivity index during 40 mU · m–2 · min–1 insulin infusion (mean ± SD). GDM, Gestational diabetes mellitus; Pt, individual longitudinal changes with time; Pg, difference between groups.

insulin sensitivity index across time in both groups during insulin infusion of 20 mU · m–2 · min–1. In addition there was a significant (P = .02) decrease in the insulin sensitivity index in the gestational diabetes mellitus group with respect to the control group (Fig 8, A). During insulin infusion of 40 mU · m–2 · min–1 there was a significant (P = .0001) decrease in the insulin sensitivity index in both groups across time and again a decrease in insulin sensitivity (P = .03) in the gestational diabetes mellitus group with respect to the control group (Fig 8, B). Of interest, even though there was a decrease in mean insulin sensitivity index in all subjects with all insulin infusion rates from the pregravid period through late gestation, there was actual-

912 Catalano et al

ly a mean 14% increase in the insulin sensitivity index from the pregravid period to early gestation in all subjects with all insulin infusion rates. Comment The longitudinal changes in carbohydrate metabolism were evaluated in obese women with normal glucose metabolism and compared with the changes across time seen in a matched group of subjects in whom gestational diabetes mellitus developed during pregnancy. We recruited subjects for the gestational diabetes mellitus group who had no diabetes mellitus before conception but, either because of a strong family history for type 2 diabetes mellitus or gestational diabetes mellitus in a previous pregnancy, were at high risk for development of gestational diabetes mellitus in the index pregnancy. These women had no evidence of type 1 diabetes and all had negative islet cell antibody assay results.15 There was a significant 60% increase in first-phase insulin response to intravenous glucose infusion and a 130% increase in second-phase insulin response with advancing gestation in obese women with normal glucose tolerance. In contrast, there was a 200% to 250% increase in insulin response to the same glucose stimulus in lean women with normal glucose tolerance with advancing gestation.16 We speculate that the lack of a robust insulin response observed in the obese women during pregnancy may represent relative beta cell dysfunction in obese women because of chronic decreased insulin sensitivity and compensatory basal hyperinsulinemia relative to a lean control group. The results of our study with respect to obese women with normal glucose tolerance agree with previously published reports describing a decrease in maternal glucose insulin sensitivity with advancing gestation. Studies by Sivan et al,17 Buchanan et al,3 and Ryan et al4 have all described significant (40%-80%) decreases in insulin sensitivity with advancing gestation. However, these studies used either cross-sectional study designs or assumed that postpartum estimates of insulin sensitivity were equivalent to pregravid measurements. Of interest, the 47% decrease in insulin sensitivity we observed in obese subjects with normal glucose tolerance with advancing gestation was similar to the 56% decrease in insulin sensitivity observed in lean subjects in an investigation with a similar study design and methodology.16 The significant increases in basal endogenous glucose production (30% when expressed in milligrams per minute and 14% when expressed in milligrams per kilogram fat-free mass per minute) with advancing gestation in obese women with normal glucose tolerance are consistent with the increases we observed in lean women with normal glucose tolerance.18 Our results are also in agreement with the data of Cowett et al19 and Kalhan et al,20 who estimated endogenous glucose production rates in

April 1999 Am J Obstet Gynecol

pregnancy by means of various stable isotope methods. The increases in basal glucose production were observed despite the significant increases in basal insulin concentration. In addition the significant decrease in suppression of endogenous glucose production during infusion of 40 mU · m–2 · min–1 insulin supports our hypothesis for decreases in hepatic insulin sensitivity with advancing gestation, even in obese women with normal glucose tolerance. These data are consistent with the data of Sivan et al,17 who reported less suppression of hepatic glucose production in the third trimester than second-trimester or postpartum estimates in obese women with normal glucose tolerance. In contrast, there was no evidence for decreased endogenous insulin sensitivity in lean women with normal glucose tolerance at any time during gestation.18 Thus obesity in addition to the metabolic stress of pregnancy acts as a predisposing factor toward increased hepatic insulin resistance in these women. There were significant differences with advancing gestation between glucose metabolism in obese women with gestational diabetes mellitus and in the matched control group. Women with gestational diabetes mellitus had significantly greater second-phase insulin response than did the control group, but there were no significant differences in first-phase insulin response. These data are in contrast to data of Yen et al,21 Fisher et al,22 and Buchanan et al,3 all of whom showed a decrease in firstphase insulin response in women with gestational diabetes mellitus.5 We believe that these differences in results arose because the earlier studies were calculated only in later pregnancy or post partum, in contrast to our observations of the longitudinal changes in insulin response with advancing gestation. Although there was a decrease in first-phase insulin response in late gestation in women with gestational diabetes mellitus with respect to pregravid and early pregnancy measurements, according to a power analysis of our data an additional 66 subjects would need to be evaluated in each group to provide sufficient power (0.80) to detect a difference in first-phase insulin response in late pregnancy at α = .05 between groups. Furthermore, we speculate that hyperinsulinemia in late pregnancy in obese women with gestational diabetes mellitus may result in less use of lipids in late pregnancy and enhance the maintenance of increased body fat stores in these women. There was less glucose insulin sensitivity in the gestational diabetes mellitus group than in control subjects. This decrease in insulin sensitivity was significant even when adjusted for residual hepatic glucose production and variation in insulin concentrations achieved during the clamp (the insulin sensitivity index). These data are consistent with the data of Ryan et al4 but differ from the data of Fisher et al22 and Buchanan et al,3 who showed no significant differences in insulin sensitivity between women with gestational diabetes mellitus and a control

Volume 180, Number 4 Am J Obstet Gynecol

group. The differences in results may be explained in part by the methods used to estimate insulin sensitivity and by the evaluation of subjects only in late gestation, when differences in insulin sensitivity between groups are less pronounced. The finding of decreased insulin sensitivity in obese women with gestational diabetes mellitus with respect to a matched control group is consistent with our previous longitudinal studies in lean women with abnormal glucose tolerance.5 Again, the greatest differences between groups in insulin sensitivity were noted before conception and in early gestation, before the greatest decreases in insulin sensitivity occur in late pregnancy. It is only with the observation of the longitudinal changes in glucose metabolism during gestation that the relationship between groups of changes across time become readily apparent. The differences in insulin concentration achieved during the clamp studies are in part a function of the changes in insulin clearance across time and between groups. We previously reported a 24% increase in the metabolic clearance rate of insulin during pregnancy in lean women with normal and abnormal glucose tolerances.23 The data in that study were similar in direction to but of a lesser degree than the mean 45% increase in insulin clearance that we observed in our obese subjects. In addition the decrease in the metabolic clearance rate of insulin in obese women with gestational diabetes mellitus with respect to that in the control group may in part explain the increased insulin concentrations achieved during the clamp in the gestational diabetes mellitus group, as well as the overall hyperinsulinemia of the subjects with gestational diabetes mellitus, although the mechanism remains speculative. The increased metabolic clearance rate of insulin with advancing gestation may be a function of placental clearance of insulin. As described by Goodner and Freinkel,24 the placenta is rich in enzymes capable of degrading insulin. The mechanisms for the decrease in insulin sensitivity during pregnancy are not known with certainty. Investigators have suspected both receptor and postreceptor defects as possible factors. As described by Olefsky et al,12 the use of a multilevel insulin infusion may provide some insights into potential mechanisms. A decrease in insulin receptors may result in a shift to the right in the slope of the dose-response curve for insulin sensitivity without a change in maximal response. A pure postreceptor defect in insulin action would lead to proportionate reductions in glucose disposal at all insulin concentrations. If both receptor and postreceptor defects exist together, then a rightward shift in the dose response curve for insulin sensitivity plus a decrease in maximal insulin responsiveness might be observed. The use of 2 levels of insulin infusion allowed us to evaluate whether there was a shift in the slope of the dose-response curve for insulin sensitivity in both groups and with advancing

Catalano et al 913

gestation. Because our study design did not allow for maximal insulin infusion rates (1200 mU · m–2 · min–1) we could not assess for postreceptor defects. On the basis of the significant decreased slope of the dose-response curve for insulin sensitivity in gestational diabetes mellitus in comparison with the control group both before and during pregnancy, however, women with gestational diabetes mellitus may have decreased glucose insulin sensitivity because of a receptor defect in peripheral tissue, most probably skeletal muscle. These data, however, do not rule out the possibility of an additional postreceptor defect to account for decreased glucose insulin sensitivity in women with gestational diabetes mellitus or the decreases in glucose insulin sensitivity observed in both control and gestational diabetes mellitus groups with advancing gestation. We observed significant but not dissimilar increases in mean basal endogenous glucose production with advancing gestation in obese gestational diabetes mellitus subjects with respect to the control group. Our results are also in agreement with the data of Cowett et al19 and Kalhan et al,20 who showed, by means of various stable isotope methods, no significant differences in endogenous glucose production rates between women with gestational diabetes mellitus and control subjects. In addition we observed significantly less suppression of endogenous glucose production during insulin infusion in the obese gestational diabetes mellitus group (85%) than in the control group (93%) in late gestation. Although not quantitatively great, increased residual endogenous glucose production despite increased insulin concentrations may become more relevant in later life for these women. As described by DeFronzo,25 decreased suppression of hepatic glucose production constitutes a hallmark of type 2 diabetes, for which these women are at increased risk in later life. There were significant increases in insulin response and basal endogenous glucose production and decreases in glucose insulin sensitivity in obese control subjects with advancing gestation. In obese women in whom gestational diabetes mellitus developed there were significantly greater increases in insulin response, less suppression of endogenous glucose production during insulin infusion, and decreases in glucose insulin sensitivity in comparison with a matched control group. The primary difference between lean and obese women in whom gestational diabetes mellitus develops is the increased insulin response in obese gestational diabetes mellitus. We speculate that this hyperinsulinemia may play a role in the increased risks of persistent obesity, hypertension, and diabetes mellitus among these women. We thank Timothy Highman, MS, Noreen DragoRoman, BS, Lourdas Gruca, MS, Alicia Thomas, MS, RD, and Elaine Tyzbir, MS, for their expert technical assis-

914 Catalano et al

tance. In addition, we thank the nurses and staff of the General Clinical Research Center, without whose assistance these studies would not have been possible. REFERENCES

1. Manson JE, Willett WC, Stampfer MJ, et al. Body weight and mortality among women. N Engl J Med 1995;333:677-85. 2. O’Sullivan JB. Body weight and subsequent diabetes mellitus. JAMA 1982;248:949-52. 3. Buchanan TA, Metzger BE, Freinkel N, Bergman RN. Insulin sensitivity and β-cell responsiveness to glucose during late pregnancy in lean and moderately obese women with normal glucose tolerance or mild gestational diabetes. Am J Obstet Gynecol 1990;162:1008-14. 4. Ryan EA, O’Sullivan MJ, Skyler JS. Insulin action during pregnancy: studies with the euglycemic clamp technique. Diabetes 1985;34:380-9. 5. Catalano PM, Tybir ED, Wolfe RR, et al. Carbohydrate metabolism during pregnancy in control subjects and women with gestational diabetes. Am J Physiol 1993;264:E60-7. 6. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 1979;28:1039-57. 7. Carpenter MW, Coustan DR. Criteria for screening tests for gestational diabetes. Am J Obstet Gynecol 1982;144:768-73. 7a. Taylor HL, Jacobs DR, Schuker B, Knudsen J, Leon AS, Debacker G. A questionnaire for the assessment of leisuretime activities. J Chronic Dis 1978;31:741-55. 8. Jaffe BM, Behrman HR. Methods of hormone radioimmunoassay. New York: Academic Press; 1974. p. 613-42. 9. Keys A, Brozek J. Body fat in adult man. Physiol Rev 1953;33:245-325. 10. DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 1979;237:E214-23. 11. Steele R. Influences of glucose loading and of injected insulin on hepatic glucose output. Ann N Y Acad Sci 1959;82:420-30. 12. Olefsky JM, Kolterman OG, Scarlett JA. Insulin action and resistance in obesity and noninsulin-dependent type II diabetes mellitus. Am J Physiol 1982;243:E15-30. 13. Finegood DT, Bergman RN, Vranic M. Estimation of endogenous glucose production during hyperinsulinemic-euglycemic clamps. Diabetes 1987;36:914-24. 14. Tserng KY, Kalhan SC. Calculation of substrate turnover rate in stable isotope tracer studies. Am J Physiol 1983;245:E308-11. 15. Srikanta S, Rabizadeh A, Omar MA, Eisenbarth GS. Assay for islet cell antibodies: protein A-monoclonal antibody method. Diabetes 1985;34:300-5. 16. Catalano PM, Tyzbir ED, Roman NM, Amini SB, Sims EA. Longitudinal changes in insulin release and insulin resistance in nonobese pregnant women. Am J Obstet Gynecol 1991;165:1667-72. 17. Sivan E, Chen X, Homko CJ, Reece EA, Boden G. Longitudinal study of carbohydrate metabolism in healthy obese pregnant women. Diabetes Care 1997;20:1470-5. 18. Catalano PM, Wolfe RR, Tyzbir ED, Roman NM, Amini SB, Sims EA. Longitudinal changes in basal hepatic glucose production and suppression during insulin infusion in normal pregnant women. Am J Obstet Gynecol 1992;167:913-9. 19. Cowett RM, Susa JB, Kahn CB, Gilotti B, Oh W, Schwartz R. Glucose kinetics in non-diabetic and diabetic women during the third trimester of pregnancy. Am J Obstet Gynecol 1983;146:773-80. 20. Kalhan SC, Rossi K, Gruca L, Burkett E, O’Brien A. Glucose turnover and gluconeogenesis in human pregnancy. J Clin Invest 1997;100:1175-81. 21. Yen SC, Tsai CC, Vela P. Gestational diabetogenesis: quantitative analysis of glucose-insulin interrelationship between normal pregnancy and pregnancy with gestational diabetes. Am J Obstet Gynecol 1971;111:792-800. 22. Fisher PM, Sutherland HW, Brewsher PD. The insulin response to glucose infusion in gestational diabetes. Diabetologia 1980;19:10-4.

April 1999 Am J Obstet Gynecol

23. Catalano PM, Drago NM, Amini SB. Longitudinal changes in pancreatic b-cell function and metabolic clearance rate of insulin in pregnant women with normal and abnormal glucose tolerance. Diabetes Care 1998;21:403-8. 24. Goodner CJ, Freinkel N. Carbohydrate metabolism in pregnancy: the degradation of insulin by extracts of maternal and fetal structures in the fat. Endocrinology 1959;65:957-67. 25. DeFronzo RA. Lilly lecture 1987. The triumvirate: b-cell, muscle, liver: a collusion responsible for NIDDM. Diabetes 1988;37:667-87.

Discussion DR DONALD R. COUSTAN, Providence, Rhode Island. Patients with type 2 diabetes, the most common form of diabetes in most populations, manifest resistance to the effects of insulin, and it was tempting in the past to ascribe their diabetes solely to this problem. It has become increasingly clear, however, that insulin resistance alone cannot account for type 2 diabetes. Some individuals with marked insulin resistance have the capacity to increase their insulin production and release to maintain relatively normal circulating glucose levels. For example, there are individuals with the so-called “syndrome X,” which includes obesity, insulin resistance, and cardiovascular problems, who do not have diabetes. It thus appears that people with type 2 diabetes not only are insulin resistant but also are unable to mount a great enough increase in insulin secretion to maintain euglycemia. Questions persist as to which comes first, the insulin resistance or the insulin secretory defect. It would be desirable to study individuals early in the development of diabetes to determine the sequence of events that occurs. Gestational diabetes mellitus has long been regarded as a precursor of overt diabetes, particularly type 2 diabetes. The reasoning is that pregnancy is a time of decreased insulin sensitivity or, to put it another way, increased insulin resistance. Women with a propensity toward development of diabetes would be likely to have this occur transiently during pregnancy. To confuse matters, gestational diabetes mellitus is not a single entity but appears to affect a heterogeneous group of individuals, some of whom are obese and some of whom are lean, some of whom have strong family or ethnic risk of type 2 diabetes and some of whom may be on their way to development of type 1 diabetes. Dr Catalano has described a comprehensive longitudinal series of studies on a specific group of individuals, obese women with gestational diabetes mellitus. As a control group he appropriately chose obese women without gestational diabetes mellitus, so that the pathogenesis of gestational diabetes mellitus could be separated from those changes related only to the interplay between obesity and pregnancy. A unique feature of this report is the fact that both groups were studied before pregnancy, when their GTT results did not indicate diabetes, as well as throughout pregnancy as gestational diabetes mellitus developed. Other studies have looked at women with gestational diabetes mellitus after the diagnosis has been made, and those investigators assumed that findings in the postpartum state would be reflective of the state that would have been present before preg-

Volume 180, Number 4 Am J Obstet Gynecol

nancy. No such assumptions are necessary in the present study, and this is a significant accomplishment. Catalano et al found that total insulin sensitivity decreased throughout the course of pregnancy in both obese subjects with gestational diabetes and obese control subjects without diabetes mellitus. The group with gestational diabetes mellitus, however, manifested a greater fall in insulin sensitivity than did control subjects. Hepatic glucose production in the basal state increased throughout pregnancy in both the gestational diabetes mellitus group and control subjects, and insulin infusion was less able to suppress hepatic glucose production as pregnancy progressed. Subjects with gestational diabetes mellitus showed even greater hepatic resistance to suppression by insulin than did control subjects. Both subjects with gestational diabetes mellitus and control subjects increased the insulin response to an intravenous glucose load as pregnancy progressed, whether this response was measured during the first 5 minutes or during the interval from 5 minutes to 1 hour. Only the later insulin response increased significantly more in subjects with gestational diabetes mellitus than in control subjects. The studies presented by Dr Catalano are thoughtfully designed, are well executed, and have great importance for our understanding of both the pathophysiology of diabetes and the physiology of pregnancy. I have only a few questions. (1) Dr Catalano, although you found that obese subjects with gestational diabetes mellitus had an enhanced second-phase insulin response to an intravenous glucose challenge with respect to obese women with normal pregnancy, can we not still infer an insulin secretory defect in the gestational diabetes mellitus because of their higher glucose response? (2) Because you followed these patients longitudinally, it would be of interest to know at what point in pregnancy gestational diabetes mellitus appeared. (3) The disappearance of a significant difference in insulin sensitivity when results were expressed in terms of fat-free mass raises the question of whether the apparent increased insulin resistance in obese women with gestational diabetes mellitus is accounted for by obesity. DR STEVEN H. GABBE, Seattle, Washington. In your previous studies you have shown that women who are lean and have gestational diabetes mellitus do have a defect in the early or first phase of insulin release. What do you think that this implies in terms of pathogenesis and long-term prognosis for lean versus obese women with gestational diabetes mellitus. Also, did you attempt to characterize the obesity in your patients by doing waisthip measurements? DR JAMES M. ROBERTS, Pittsburgh, Pennsylvania. The only thing I am puzzled by is the conclusion. I wonder why, if the women are hyperinsulinemic, you would anticipate that the other tissues that might be involved in their storing fat are not also insulin resistant? DR CATALANO (Closing). Obese women with gestational diabetes had an enhanced second-phase insulin response to intravenous GTT when compared with the obese control patients. The question was, is there still some evidence of an insulin secretory defect in the

Catalano et al 915

women with gestational diabetes mellitus because of the higher glucose response? The answer is yes. When you compare the insulin response in these obese women with what Dr Gabbe referred to earlier in women who were lean, relatively speaking, the insulin response in these obese women both with normal glucose tolerance and with gestational diabetes was significantly less as a percentage increase with advancing gestation as compared with women who were lean. So, if you take lean women who were given the same glucose challenge, the insulin response is approximately 2-fold greater than what we observed in these obese subjects. There is something about obesity in and of itself that may relate to a defect in pancreatic beta cell dysfunction. Exactly what the mechanism may be is not known at this time. I would agree that obesity and decreased insulin sensitivity seem to be factors that relate to decreased beta cell response. The second question, then, is, because we evaluated these patients longitudinally, when in pregnancy did gestational diabetes develop? Gestational diabetes developed in all of these women in the late second or early third trimester of pregnancy. None of the subjects had abnormal results of GTT by the criteria of Carpenter and Coustan (reference 7 of article) in early pregnancy. These abnormal results developed in all of them in late pregnancy (ie, 34-36 weeks). These women also had gestational diabetes screening as part of their routine obstetric care; and to the best of my memory, all of them had abnormal results at that time. This finding was confirmed at the time of our 100gm oral GTT for the purposes of the study. Dr Coustan’s third question was, is the disappearance of a significant difference in insulin sensitivity between control and gestational diabetes mellitus groups when the results are expressed in terms of fat free mass accounted for by obesity alone? I think that there are certainly data to show that factors in pregnancy other than obesity per se are related to the changes in insulin sensitivity. For example, whatever happened during pregnancy to account for the 40% decrease in women with normal GTT results was also the same factor that accounted for decreased insulin sensitivity in women with development of gestational diabetes. They happened to be more insulin resistant before conception. Obviously, this is related to different factors, whether they be placental or other sources of hormonal antiinsulin effects. Additional factors noted by other investigators include free fatty acids, which are increased in women who are obese. Free fatty acid levels may be related to decreased insulin sensitivity, in addition to hormonal changes of pregnancy. Relative to Dr Coustan’s last question about using the 2 different glucose loads for our GTTs, our primary purpose of using a 75-gm oral GTT before pregnancy as compared with a 100-gm test during pregnancy was for definition. What we tried to do is classify patients according to accepted criteria of having normal glucose tolerance or having abnormal glucose tolerance, before and during pregnancy.

916 Catalano et al

Dr Gabbe asked a question that goes back to some of our original studies which looked at the changes in insulin response in lean women with development of gestational diabetes. They have a decreased first-phase insulin response as compared with that in lean control subjects. In contrast, obese women with development of gestational diabetes have an increased second-phase insulin response in comparison with that of a matched control group. I think this is probably the most significant finding in this study as compared with our previous study. The difference between lean women with development of gestational diabetes and obese women is not so much the insulin resistance but the insulin response. The changes in insulin resistance over time in both groups was essentially the same (ie, about a 40% to 50% decrease in insulin sensitivity). Really, where the significant difference seems to be, whether patients are lean or obese, is the pancreatic beta cell function (ie, the inability to mount an appropriate response can lead to the development of abnormal glucose tolerance defined by GTT). To get back to Dr Coustan’s point, the hyperinsulinemia we see in obese people may be related to the obesity and other factors. It is hyperinsulinemia that is the significant factor. The long-term significance of hyperinsulinemia may be that obese women with gestational diabetes have a significantly greater long-term risk of development of type 2 diabetes as compared with women who are relatively lean. Studies have shown between a 40% and 50% risk that type 2 diabetes would develop if a patient was obese and had gestational diabetes mellitus as compared with approximately a 5% to 10% risk if the patient was lean and had gestational diabetes mellitus. To the best of our ability we know that these women didn’t have subclinical type 1 diabetes because they were all negative for islet cell antibodies. Therefore in lean women with gestational diabetes the inability to mount a

April 1999 Am J Obstet Gynecol

first-phase response was enough to make the GTT result abnormal, defined as having gestational diabetes but not enough to significantly increase the long-term risk of diabetes in comparison with obese women with gestational diabetes. Once the stress of pregnancy was resolved, the insulin response was usually sufficient to maintain euglycemia. Last, we assessed distribution of body fat in our subjects using the waist/thigh ratio. There was no significant difference in the waist/thigh ratio in the control women (1.18 ± 0.12) as compared with the women who had gestational diabetes mellitus (1.29-0.15). Regarding Dr Roberts’ question about factors other than glucose relating to gestational diabetes, we do have data that other nutrients are affected to a differential degree. There are data to show that amino acids in pregnancy, in women with normal glucose tolerance, have decreased turnover with advancing gestation. Insulin resistance is not limited just to glucose; it also applies to other nutrients, in particular, amino acids, and our assumption is that pregnancy also affects free fatty acid insulin sensitivity. The effect of insulin on a specific nutrient’s sensitivity to insulin varies. For example, the ability of infused insulin to inhibit lipolysis is significantly greater than the ability of infused insulin to affect basal hepatic glucose or amino acid insulin sensitivity. The issue then is how hyperinsulinemia affects these women. We believe that in obese women hyperinsulinemia inhibits lipolysis more than glucose metabolism so that the nutrient utilized during pregnancy is primarily glucose. In these obese women less fat may be utilized in pregnancy because of hyperinsulinemia. Studies in Pima Indians and other groups support the hypothesis that hyperinsulinemia may, in effect, perpetuate obesity because fat is used for energy relative to other nutrients.