Background: Dyslipidemia is a risk factor of type 2 diabetes (T2D). Women with a history of gestational diabetes (GDM) have a greater T2D risk; however, whether lipids patterns contribute to risk of progression remains unclear.

Methods: We conducted a nested case-control study of incident T2D among women enrolled in the Nurses’ Health Study II reporting a history of GDM prior to the first blood collection (1995-1999) and followed-up through 2013. Plasma lipidomics containing 180 metabolites were measured in a subset of 175 incident cases and 176 age-matched controls by liquid chromatography-mass spectrometry. We constructed lipid networks using a weighted-correlation algorithm, and examined the association of individual lipids and sub-networks with T2D risk using logistic regression, adjusting for established T2D risk factors.

Results: We identified 39 lipids species significantly associated with T2D risk (FDR<0.05), and observed different association patterns (in which associations were differed by numbers of carbon and double-bonds) across various lipid classes. We constructed 5 co-regulating lipid sub-networks, one of which (featuring the interactions between diacylglycerol, triacylglycerol, and phosphatidylethanolamine with mid-length carbon and less double-bonds) was significantly associated with incident T2D risk (odds ratio for increasing quartiles was 1 [reference], 2.4, 3.4, and 4.4; P-trend=0.0008). The addition of significant lipid species to a base model (13 traditional risk factors and medications) to predict incident T2D increased the c-statistics from 0.83 to 0.90 (P=0.001) with a corresponding categorical net reclassification index of 0.27 (P<0.001) in cross-validation.

Conclusions: The dysregulation of certain lipid specie/class may contribute to a higher risk of incident T2D after a GDM pregnancy. Plasma lipid species and the network signatures may aid future clinical risk assessment and targeted interventions.


J. Li: None. L. Liang: None. F. Hu: None. C. Zhang: None. D. Tobias: None.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at