Hormonal and Metabolic Factors Associated with Variations in Insulin Sensitivity in Human Pregnancy

Diabetes Care Publish Ahead of Print, published online October 30, 2009

Hormonal and metabolic factors associated with variations in insulin sensitivity in human pregnancy

Harold David McIntyre FRACP1, Allan Mang Zing Chang PhD1, Leonie Kaye Callaway PhD2 , David Michael Cowley FRCPA1, Alan Richard Dyer PhD3, Tatjana Radaelli MD4,5, Kristen Anne Farrell MS 5, Larraine Phyllis Huston-Presley MS5, Saeid Baradaran Amini PhD, JD 5, John Patrick Kirwan PhD 5, Patrick Michael Catalano MD 5, for the HAPO Study Cooperative Research Group*

1 The University of Queensland and Mater Health Services, South Brisbane, Australia; 2 The University of Queensland and Royal Brisbane and Women’s Hospital, Herston, Australia; 3 Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois; 4 Department of Mother, Child and Neonate “L. Mangiagalli”, IRCCS Ospedale Maggiore Policlinico Mangiagalli and Regina Elena, 20122, Milan, Italy; and 5 Department of Reproductive Biology, Case Western Reserve University at Metro Health Medical Center, Cleveland, Ohio. * Members of the HAPO Study Cooperative Research Group are listed in the Appendix of: HAPO Study Cooperative Research Group. Hyperglycemia and Adverse Pregnancy Outcomes. N Engl J Med. 2008;358:1991-2002.

. Corresponding Author / Reprint requests: Professor Harold David McIntyre [email protected]

Additional information for this article can be found in an online appendix at http://care.diabetesjournals.org

Submitted 30 June 2009 and accepted 16 October 2009.

This is an uncopyedited electronic version of an article accepted for publication in Diabetes Care. The American Diabetes Association, publisher of Diabetes Care, is not responsible for any errors or omissions in this version of the manuscript or any version derived from it by third parties. The definitive publisher- authenticated version will be available in a future issue of Diabetes Care in print and online at http://care.diabetesjournals.org.

Objective: To determine maternal hormonal and metabolic factors associated with insulin sensitivity in human pregnancy

Research Design and Methods: Prospective observational cross sectional study of one hundred and eighty normal pregnant women, using samples collected at the time of blinded oral glucose tolerance testing between 24 and 32 weeks gestation as an ancillary to the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study. Setting - two public university teaching hospitals – Cleveland, Ohio, USA and Brisbane, Australia. Methods of assessment - Fasting maternal serum cholesterol, triglycerides, free fatty acids, insulin, leptin, tumour necrosis factor alpha, placental growth hormone (PGH), insulin like growth factors (IGFs) 1 and 2 and insulin like growth factor binding proteins (IGFBPs) 1 and 3 were assayed. Correlation and multiple regression analyses were used to determine factors associated with maternal insulin sensitivity estimated using both OGTT derived (IS OGTT) and fasting (IS HOMA) insulin and glucose concentrations.

Results: Insulin sensitivity correlated (r = x, y for IS OGTT , IS HOMA respectively) with fasting maternal serum leptin (-0.44, -0.52 ), IGFBP1 ( 0.42, 0.39) and triglycerides (-0.31, -0.27 ). These factors were significantly associated with insulin sensitivity in multiple regression 2 analyses (adjusted R 0.44 for IS OGTT and IS HOMA). These variables explained more than 40% of the variance in estimates of insulin sensitivity.

Conclusions: Maternal hormonal and metabolic factors related to the placenta, adipose tissue and the growth hormone axis are associated with the variation in insulin sensitivity seen during normal human pregnancy.

he development of insulin Hyperglycemia and Adverse Pregnancy resistance in pregnancy has Outcome (HAPO) study (8) and subjects T been recognized for many consented in writing both to the main HAPO years, but the causal mechanisms remain study and to this ancillary study. Subjects and unclear. Ryan and colleagues first investigators were blinded to the results of the demonstrated a 40% decrease in insulin oral glucose tolerance test (OGTT), so as not sensitivity (1) in women with gestational to affect the outcome of the primary HAPO diabetes as compared with a control group at project. term. Later, Catalano and colleagues Clinical methods. One hundred confirmed these results describing and eighty (180) women, including 80 from longitudinal changes in insulin sensitivity and Cleveland, enrolled in HAPO were recruited insulin response in women with normal into this ancillary study. Their characteristics glucose tolerance and gestational diabetes are shown in Table 1. A 75g OGTT was before and during pregnancy (2). Despite a performed after 8-10 hours overnight fasting general tendency to attribute whole body in all subjects between 24 and 28 weeks (as insulin resistance in pregnancy to “placental close as possible to the 28th week) of gestation hormones” (3), the precise contribution of according to standardized procedures. The various hormonal factors remains poorly OGTT consisted of fasting (0 min), 60 and defined. Human placental lactogen (hPL) was 120 minute glucose measures and fasting and an early candidate though findings have been 60 minute C-peptide determinations. variable (4). Kirwan et al have suggested an Ancillary study patients had estimations of important role for tumor necrosis factor alpha serum insulin at 0, 60 and 120 mins. As part (TNF∝) (5) whilst placental growth hormone of the HAPO protocol, subjects were (PGH) has been shown to induce insulin unblinded if their fasting plasma glucose level resistance in a mouse model (6) and to (PG) exceeded 105 mg/dL (5.8 mmol/L), if correlate with maternal glycemia in patients the 2-hour OGTT PG exceeded 200 mg/dL with diabetes (7). Our study was designed to (11.1 mmol/L) or if any recorded value was further explore the maternal metabolic and less than 45 mg / dL (2.5 mmol/L). This hormonal correlates of insulin resistance in a study includes only women whose OGTT healthy pregnant population. results were within HAPO limits. Three We hypothesized that factors in women would have been classified as having addition to placental hormones were gestational diabetes by the NDDG criteria and associated with insulin resistance during eight by the Carpenter Coustan criteria. (9) normal pregnancy. However, since all glucose results were blinded, we have not excluded these women MATERIALS AND METHODS from this report. Other hormonal and The protocol was approved by the metabolic factors were measured in the Hospital Institutional Review Board (IRB) fasting state. and the Scientific Review Committee of the Laboratory methods. Glucose General Clinical Research Center (GCRC) at assays in HAPO employed the glucose Metro Health, Cleveland, Ohio, USA and by oxidase method and were carefully the Human Research Ethics Committee of standardized across all HAPO centers under Mater Health Services, South Brisbane, the supervision of the central laboratory in Australia. Both of these centers participated Belfast. The other biochemical and hormonal in the international, multi centre assays for this ancillary study were performed

3 at either the GCRC Cleveland (insulin; leptin; approximate a normal distribution. Natural free fatty acids (FFA); TNF ∝ and insulin logarithms have been used in further analyses. like growth factors one and two [IGF1, We employed linear product moment IGF2]) or Mater Health Services Brisbane (Pearson’s) correlations followed by multiple labs (PGH, insulin like growth factor binding linear regression analysis to explore the proteins one and three [IGFBP1, IGFBP3], relationships between variables. Dependent cholesterol and triglycerides) in one or two variables were ISOGTT and ISHOMA. batches, with one shipment of samples in each Independent variables included all measured direction. Samples with hemolysis were maternal biochemical parameters, maternal excluded prior to testing. All assays were pre pregnancy BMI, BMI at the OGTT, age performed in duplicate. Assay CVs are and centre (Cleveland vs. Brisbane). Results shown in Supplementary Table 1A in the reported include standardized regression online appendix which is available at coefficients (β) with 95 % confidence http://care.diabetesjournals.org. intervals and partial correlation coefficients. Insulin samples were centrifuged at STATA (StataCorp TX, USA) and Statistica 4ºC and stored at -70ºC. Insulin was assayed (StatSoft, OK, USA) were used for statistical using a double-antibody radioimmunoassay analyses. Significance was accepted at the (Linco, St. Charles, MO) as previously 5% level on two tailed testing. described. (2) Leptin, FFA, TNF∝, IGF1, IGF2, PGH, IGFBP1 and IGFBP3 were RESULTS assayed using previously described methods The characteristics (Mean (SD)) of the (5; 10; 11) 180 women who participated in this study are Based on previous work by the outlined in Table 1. The median and Cleveland group (12) the insulin sensitivity interquartile ranges for the biochemical and index calculated from the OGTT according to hormonal variables are also shown in Table 1. the equation first described by Matsuda and Only age at delivery differed between the DeFronzo (ISOGTT) formed our primary Cleveland and Brisbane participants. Non measure of insulin sensitivity. Specifically, Hispanic Whites were the predominant ethnic insulin sensitivity was calculated as follows: group (80%), Hispanics 3%, Asians 9% and ISOGTT = 10,000/√(FPG)*(FPI)*(G*I); where Other Ethnicities 8%. The subjects’ mean pre FPG and FPI are fasting plasma glucose (mg / pregnancy BMI was in the overweight range. dL) and insulin (µU / mL) respectively, while Mean gestation at the time of OGTT was very G and I are mean glucose and mean insulin of close to the HAPO goal of 28 weeks. all samples from 0 to 120 minutes. The Pearson correlation coefficients We also calculated the simpler HOMA between maternal biochemical variables, measure based on fasting samples only as first estimates of insulin resistance and maternal described by Matthews and colleagues BMI (pre pregnancy and at the OGTT visit) (ISHOMA) (12) In this case, insulin sensitivity are shown in Table 2. BMI, IGFBP1, is calculated as: ISHOMA = 405 / (FPG*FPI) triglycerides and leptin correlated Statistical methods The significantly with the estimates of maternal distributions of all variables were tested using insulin sensitivity. analysis of skewness and kurtosis. Maternal Subsequently, multiple regression characteristics (Table 1 A) and the dependant analyses were performed. Results are variables ISOGTT and ISHOMA were normally reported for ISOGTT in Table 3. Maternal BMI distributed, but all other biochemical calculated at the OGTT visit, whilst variables required log transformation to significant in simple correlations (Pearson’s r

4 = -0.47, -0.48 for BMI vs. ISOGTT , ISHOMA biochemical variables related to the placenta respectively), became not statistically (leptin) adipose tissue (leptin and significant after adjusting for the other triglycerides) and the growth hormone axis variables in the model. Models incorporating (IGFBP1). The placenta is a major source of pre pregnancy BMI rather than BMI at the leptin in pregnancy and also the source of OGTT showed essentially the same findings. high concentrations of PGH, which As can be seen from Table 3, the model upregulates the growth hormone / IGF axis incorporating all biochemical variables during pregnancy (11; 13) . Although leptin is accounted for 44% of the observed variance produced both in placenta and adipose tissue, in ISOGTT . Multiple regression findings for several lines of evidence suggest that the 2 ISHOMA were virtually identical (Multiple R major changes in leptin during pregnancy 0.48; Adjusted R2 0.44; p < 0.0001) and are relate to placental leptin production. (14) not shown separately. Leptin, IGFBP1 and Firstly, maternal leptin decreases abruptly triglycerides were significantly related to both following delivery of the placenta. Secondly, insulin sensitivity estimates. These findings there is no correlation between change in were not altered by exclusion of those maternal BMI and leptin. Thirdly, the participants who would have been classified pregnancy related increase in maternal leptin as suffering from gestational diabetes by the predates increased fat mass in pregnancy (14). NDDG or Carpenter Coustan criteria. In a longitudinal study, the Cleveland group To determine whether maternal (15) has also demonstrated a close correlation overweight / obesity influenced the factors between serum leptin and fat oxidation during associated with insulin sensitivity, we early and late pregnancy but not in the non repeated the regression analyses with pregnant state. This provides a further participants characterized by pre pregnancy mechanism by which leptin may influence BMI < or > 25 kg / m2. Because the maternal insulin sensitivity. Recent evidence relationship between BMI and fat mass may demonstrates that maternal obesity also vary across ethnic groups, we also repeated influences both placental and circulating all analyses using only those participants from monocyte / macrophage populations and the dominant ethnic group (non Hispanic inflammatory markers (16), suggesting that whites -NHW). Both the BMI stratified and adipose and placental tissue contributions to “NHW only” analyses gave very similar the overall maternal metabolic and results to those presented for the whole cohort inflammatory milieu are interlinked. and the data are not presented separately. The Interestingly, the inverse relationship other ethnic subgroups were considered too of maternal BMI to insulin sensitivity, well small for separate analysis. Differences in the recognized in many studies, was no longer relationship between BMI, adiposity and statistically significant in our model when the leptin concentrations between ethnic groups, panel of ten biochemical and hormonal may explain in part why leptin and not BMI variables were included. Although these has a stronger correlation with estimates of parameters are, in themselves, significantly insulin resistance. correlated with maternal BMI (Table 2), they appeared more strongly related to insulin DISCUSSION sensitivity in the multiple regression analyses. The current study demonstrates that a An etiologic role has been proposed for substantial proportion of the variance in reduction in IGFBP1 as a link between maternal insulin sensitivity in pregnancy is maternal obesity and increased birthweight, associated with variations in maternal through increased bioactive IGF1 in maternal

5 serum (17; 18). Confirming these previous demonstrated increased TNF ∝ in peripheral findings, our study demonstrated a negative blood and placental mononuclear cells, correlation between maternal BMI and associated with insulin resistance, in obese IGFBP1. Reduced IGFBP1 in women with pregnant women, but no changes were noted higher BMI would be predicted to increase in maternal plasma TNF ∝ . In the current free maternal IGF1 and promote nutrient study, we did not find any association of transfer to the fetus and fetal growth. Indeed, TNF∝ with insulin sensitivity. A further IGFBP1 has been reported to be negatively recent study from Mastorakos et al (21) correlated with fetal lean body mass, though confirmed a relationship between leptin and not with fat mass (19), suggesting a specific insulin resistance, reported no relationship of effect on fetal body composition. IR with adiponectin and noted an association PGH showed a weak negative of insulin sensitivity with yet another correlation with insulin sensitivity in the adipocytokine, visfatin. The often divergent multiple regression analysis, but this failed to findings about relationships between reach statistical significance (p = 0.076). A adipocytokines, BMI and insulin sensitivity negative relationship of PGH with insulin are summarized in a recent review by Briana. sensitivity would be predicted from known (22) growth hormone actions in the non pregnant Do the correlations described in our state and with findings of decreased insulin study represent underlying causes of sensitivity related to elevated PGH in a variations in maternal insulin sensitivity in transgenic mouse model. (6) One previous pregnancy, or merely the consequences of study by Fuglsang et al (20) also found no such variations? A causal role seems possible correlation between PGH and fasting insulin for IGFBP1 as described above. Leptin has sensitivity estimated just prior to delivery. been noted to directly modulate insulin The effects of PGH thus appear (at best) sensitivity in vitro (23) and has been modest in normal human pregnancy. Other described as a predictor of gestational factors are clearly of greater importance. diabetes, independent of maternal BMI (24). Previous findings regarding the Pregnancy is a physiologic leptin resistant relationship of maternal hormones and state, in that increased maternal energy intake adipokines to insulin sensitivity have been and positive energy balance develop in late variable. Using the frequently sampled pregnancy despite increased leptin levels, intravenous glucose tolerance test (IVGTT) in which would be predicted to reduce appetite a small group of patients (n=38), McLachlan and energy intake in a fully leptin sensitive et al (13) reported that leptin correlated state (14). Leptin has also been reported to negatively with insulin sensitivity (IS), but reduce insulin secretion in both rodent and adiponectin, TNF ∝ and c reactive protein human islets in vitro. (25) However, the (CRP) proved unrelated to IS. In contrast, uniform hyperinsulinemia of normal one previous report of 15 pregnant women pregnancy despite high leptin concentrations using the insulin clamp (5) from our group again suggests leptin resistance at the level of noted TNF ∝ as a significant factor. the beta cells. Partitioning effects may also However, subjects in that study included be important for leptin, as it has been noted obese women with gestational diabetes who that placental leptin mRNA and protein also had significantly elevated plasma TNF∝ content is 3 – 5 fold higher in Type 1 diabetic during pregnancy. Partitioning of TNF ∝ pregnancies than in controls, despite may also be of importance in this regard. The comparable maternal serum leptin recent study from Challier et al (16) concentrations. (14)

6 In summary, our data demonstrate that but we would consider that clamp studies are variations in maternal insulin sensitivity in not feasible in a cohort of this size. Further, normal pregnancy relate in part to the we have established strong correlations with maternal adipocytokine and growth hormone / the clamp method in previous studies. insulin like growth factor axes. Our findings It is plausible, though not yet proven, are novel in that they extend the range of that these systems serve to regulate whole potential factors examined simultaneously in body insulin sensitivity in individual pregnant relation to maternal insulin sensitivity and women. An improved understanding of these include a much larger number of subjects than factors may potentially open up new avenues in most previous reports. We acknowledge of treatment in gestational diabetes and other that estimation of insulin sensitivity using the conditions associated with insulin resistance ISOGTT and ISHOMA is less precise than “gold in pregnancy, such as obesity and pre standard” measurement with an insulin clamp, eclampsia.

7 REFERENCES 1. Ryan EA, O'Sullivan MJ, Skyler JS: Insulin action during pregnancy. Studies with the euglycemic clamp technique. Diabetes 1985; 34:380-389 2. Catalano PM, Huston L, Amini SB, Kalhan SC: Longitudinal changes in glucose metabolism during pregnancy in obese women with normal glucose tolerance and gestational diabetes mellitus. Am J Obstet Gynecol 1999;180:903-916 3. Barbour LA, McCurdy CE, Hernandez TL, Kirwan JP, Catalano PM, Friedman JE: Cellular mechanisms for insulin resistance in normal pregnancy and gestational diabetes. Diabetes Care 2007; 30 Suppl 2:S112-119 4. Ryan EA, Enns L: Role of gestational hormones in the induction of insulin resistance. J Clin Endocrinol Metab 1988; 67:341-347 5. Kirwan JP, Hauguel-De Mouzon S, Lepercq J, Challier JC, Huston-Presley L, Friedman JE, Kalhan SC, Catalano PM: TNF-alpha is a predictor of insulin resistance in human pregnancy. Diabetes 2002; 51:2207-2213 6. Barbour LA, Shao J, Qiao L, Pulawa LK, Jensen DR, Bartke A, Garrity M, Draznin B, Friedman JE: Human placental growth hormone causes severe insulin resistance in transgenic mice. Am J Obstet Gynecol 2002; 186:512-517 7. McIntyre HD, Serek R, Crane DI, Veveris-Lowe T, Parry A, Johnson S, Leung KC, Ho KK, Bougoussa M, Hennen G, Igout A, Chan FY, Cowley D, Cotterill A, Barnard R: Placental growth hormone (GH), GH-binding protein, and insulin-like growth factor axis in normal, growth-retarded, and diabetic pregnancies: correlations with fetal growth. J Clin Endocrinol Metab 2000 ; 85:1143-1150 8. Metzger BE, Lowe LP, Dyer AR, Trimble ER, Chaovarindr U, Coustan DR, Hadden DR, McCance DR, Hod M, McIntyre HD, Oats JJ, Persson B, Rogers MS, Sacks DA: Hyperglycemia and adverse pregnancy outcomes. N Engl J Med 2008; 358:1991-2002 9. Carpenter MW, Coustan DR: Criteria for screening tests for gestational diabetes. Am J Obstet Gynecol 1982; 144:768-773 10. Clapp JF, 3rd, Schmidt S, Paranjape A, Lopez B: Maternal insulin-like growth factor-I levels (IGF-I) reflect placental mass and neonatal fat mass. Am J Obstet Gynecol 2004; 190:730-736 11. McIntyre HD, Serek R, Crane DI, Veveris-Lowe T, Parry A, Johnson S, Leung KC, Ho KKY, Bougoussa M, Hennen G, Igout A, Chan FY, Cowley D, Cotterill A, Barnard R: Placental growth hormone (GH), GH-binding protein, and insulin-like growth factor axis in normal, growth-retarded, and diabetic pregnancies: Correlations with fetal growth. Journal of Clinical Endocrinology and Metabolism 2000; 85:1143-1150 12. Catalano PM, Kirwan JP: Clinical utility and approaches for estimating insulin sensitivity in pregnancy. Semin Perinatol 2002; 26:181-189 13. McLachlan KA, O'Neal D, Jenkins A, Alford FP: Do adiponectin, TNFalpha, leptin and CRP relate to insulin resistance in pregnancy? Studies in women with and without gestational diabetes, during and after pregnancy. Diabetes Metab Res Rev 2006; 22:131-138 14. Bajoria R, Sooranna SR, Ward BS, Chatterjee R: Prospective function of placental leptin at maternal-fetal interface. Placenta 2002; 23:103-115 15. Okereke NC, Huston-Presley L, Amini SB, Kalhan S, Catalano PM: Longitudinal changes in energy expenditure and body composition in obese women with normal and impaired glucose tolerance. Am J Physiol Endocrinol Metab 2004; 287:E472-479

8 16. Challier JC, Basu S, Bintein T, Minium J, Hotmire K, Catalano PM, Hauguel-de Mouzon S: Obesity in Pregnancy Stimulates Macrophage Accumulation and Inflammation in the Placenta. Placenta 2008; 29(3):274-81 17. Jansson N, Nilsfelt A, Gellerstedt M, Wennergren M, Rossander-Hulthen L, Powell TL, Jansson T: Maternal hormones linking maternal body mass index and dietary intake to birth weight. Am J Clin Nutr 2008; 87:1743-1749 18. Hills FA, English J, Chard T: Circulating levels of IGF-I and IGF-binding protein-1 throughout pregnancy: relation to birthweight and maternal weight. J Endocrinol 1996; 148:303- 309 19. Radaelli T, Uvena-Celebrezze J, Minium J, Huston-Presley L, Catalano P, Hauguel-de Mouzon S: Maternal interleukin-6: marker of fetal growth and adiposity. J Soc Gynecol Investig 2006; 13:53-57 20. Fuglsang J, Sandager P, Moller N, Fisker S, Frystyk J, Ovesen P: Peripartum maternal and foetal ghrelin, growth hormones, IGFs and insulin interrelations. Clin Endocrinol (Oxf), 2006; 64:502-509 21. Mastorakos G, Valsamakis G, Papatheodorou DC, Barlas I, Margeli A, Boutsiadis A, Kouskouni E, Vitoratos N, Papadimitriou A, Papassotiriou I, Creatsas G: The role of adipocytokines in insulin resistance in normal pregnancy: visfatin concentrations in early pregnancy predict insulin sensitivity. Clin Chem, 2007; 53:1477-1483 22. Briana DD, Malamitsi-Puchner A: Adipocytokines in Normal and Complicated Pregnancies. Reprod Sci, 2009; 16(10):921-37 23. Cohen B, Novick D, Rubinstein M: Modulation of Insulin Activities by Leptin. Science 1996; 274:1185-1188 24. Qiu C, Williams MA, Vadachkoria S, Frederick IO, Luthy DA: Increased maternal plasma leptin in early pregnancy and risk of gestational diabetes mellitus. Obstet Gynecol 2004; 103:519-525 25. Seufert J: Leptin effects on pancreatic beta-cell gene expression and function. Diabetes 2004; 53 Suppl 1:S152-158

9 Table 1: Maternal Characteristics and Biochemical Variables Variable (units) Mean Standard deviation Age at delivery (years) 29.1 5.5 Weight pre pregnancy (kg) 71.1 18.6 BMI pre pregnancy (kg / m2 ) 26.2 6.4 Gestation at OGTT (weeks) 27.9 1.6 Weight at OGTT (kg) 81.3 18.4 BMI at OGTT(kg / m2) 30.0 5.9 Variable (units) Median Interquartile range IGFBP1 (nmol/L) 6.18 3.30 - 5.69 IGFBP3 (nmol/L) 141.50 95.55 - 184.49 PGH (ng/mL) 10.11 5.94 - 12.48 Cholesterol (mmol/L) 6.18 5.52 - 6.82 Triglycerides (mmol/L) 2.02 1.56 - 2.28 Leptin (ng/mL) 35.92 21.01 - 45.43 FFA (mmol/L) 0.59 0.47 - 0.71 TNF∝ (pg/mL) 2.50 0.87 - 2.83 IGF1 (ng/mL) 164.32 25.14 – 276.60 IGF2 (ng/mL) 893.34 770.69 – 1050.41 Legend Table 1 Clinical characteristics of women who participated in the study. All of these variables were normally distributed. Median values and interquartile ranges for the biochemical and hormonal variables measured in the study. All variables were non – normally distributed and were transformed as natural logarithms for further analyses.

Table 2. Pearson correlation coefficients between insulin sensitivity estimates, BMI and biochemical variables IS OGTT IS HOMA BMI pre pregnancy BMI at OGTT visit BMI pre pregnancy -.415a -.410a 1.000 .940a BMI at OGTT visit -.470a -.484a .940a 1.000 IGFBP1 .421a .386a -.316a -.360a IGFBP3 .002 .043 -.0390 -.039 PGH -.041 -.005 -.198b -.223a Cholesterol -.047 -.076 -.123 -.096 Triglycerides -.311a -.269a .159b .106 Leptin -.437a -.519a .448a .550a Free fatty acids -.051 .006 .0462 .052 TNF α -.023 .0390 -.0128 -.039 IGF1 -.055 .027 .0434 .053 IGF2 -.104 -.033 .0550 .036

Legend Table 2. Pearson correlation coefficients between calculated maternal BMI (pre pregnancy and at the OGTT visit), biochemical and hormonal parameters measured in the study (transformed to natural logarithms) and estimates of a b insulin sensitivity (ISOGTT and ISHOMA). p < 0.01; p < 0.05 Table 3

10 Regression model: Dependant variable ISOGTT

β 95% CI (β) Partial p Independant Variable correlation Leptin -0.365 -0.535 to -0.195 -0.330 < 0.001 IGFBP1 0.319 0.180 to 0.458 0.349 < 0.001 Triglycerides -0.293 -0.432 to -0.155 -0.327 < 0.001 PGH -0.136 -0.286 to 0.015 -0.146 0.076 BMI at OGTT -0.142 -0.311 to 0.028 -0.135 0.100 Cholesterol 0.096 -0.045 to 0.238 0.110 0.181 Centre -0.160 -0.457 to 0.137 -0.088 0.288 Maternal age 0.063 -0.061 to 0.188 0.082 0.317 IGF1 -0.078 -0.359 to 0.203 -0.045 0.584 IGFBP3 -0.016 -0.206 to 0.174 -0.014 0.870 Free fatty acids -0.006 -0.141to 0.130 -0.007 0.935 TNFα 0.003 -0.143 to 0.149 0.003 0.967 IGF2 -0.001 -0.179 to 0.176 -0.001 0.989 Legend Table 3. Summary of multivariable regression of biochemical and other parameters associated with estimates of insulin sensitivity (ISOGTT and ISHOMA). Standardised correlation coefficients (β) and their 95% confidence intervals as well as partial correlations are shown for each variable. Overall multiple R2 0.49; adjusted R2 0.44; p < 0.0001