Impact of an Additional Pregnancy on the Natural History of Insulin Sensitivity, β-Cell Function, and Glycemia in Parous Women

OBJECTIVE: It has been postulated that pregnancies may contribute to the lifetime risk of type 2 diabetes mellitus (T2DM) in women, although direct evidence is lacking. In this context, we hypothesized that pregnancy may impact the long-term natural history of the pathophysiologic determinants of T2DM: insulin sensitivity and β-cell function. We sought to evaluate 10-year trajectories of these features in parous women with and without an additional pregnancy. RESEARCH DESIGN AND METHODS: Metabolic characterization was done in 303 women at 1 year postpartum and on three or more occasions over the decade thereafter. Each assessment included an oral glucose tolerance test, enabling serial evaluation of insulin sensitivity/resistance (Matsuda index; HOMA-insulin resistance [IR]), β-cell function (Insulin Secretion-Sensitivity Index-2; insulinogenic index/HOMA-IR), and glycemia. Linear mixed-effect models adjusted for diabetes risk factors were constructed to evaluate the natural time trend of these measures over 10 years in those with (n = 78) and without (n = 225) an additional pregnancy. RESULTS: At baseline, women who subsequently had an additional pregnancy were younger (P < 0.001) and more likely to be primiparous (P < 0.001) than their peers. Notably, they had higher covariate-adjusted insulin sensitivity (Matsuda index) at baseline (estimated adjusted average difference 1.64; 95% CI 0.22-2.86) and at 1.5 years' follow-up (1.28; -0.15 to 2.50), but this difference disappeared by 3.5 years and 5.5 years. The 10-year trajectory of the Matsuda index differed between the groups (time-group interaction P = 0.02), with no differences in trajectories of β-cell function and glycemia. CONCLUSIONS: Pregnancy is associated with worsening of the long-term trajectory of insulin sensitivity, revealing both a pathophysiologic basis for enhanced diabetes risk and a target for risk modification.