Background: There are little data on changes in insulin sensitivity during the first few years of life following in utero human immunodeficiency virus (HIV) and antiretroviral (ARV) exposure.

Methods: The Tshilo Dikotla study enrolled pregnant persons with HIV (PWH) (receiving tenofovir/emtricitabine or lamivudine plus dolutegravir or efavirenz) and pregnant individuals without HIV, as well as their liveborn children. Newborns were randomized to receive either zidovudine (AZT) or nevirapine (NVP) postnatal prophylaxis. Homeostasis Model Assessment for Insulin Resistance (HOMA-IR) was assessed at birth and 1, 18, 24, and 36 months of life. We fit linear mixed-effects models to evaluate the association between in utero HIV/ARV exposure and average HOMA-IR from birth through 36 months of life, adjusting for confounders.

Results: A total of 419 children were included (287 with in utero HIV/ARV exposure and uninfected [CHEU] and 132 without in utero HIV/ARV exposure [CHUU]). CHEU were born to older women (29.6 vs 25.3 years of age) with higher gravidity (3 vs 1). HOMA-IR was persistently higher in CHEU versus CHUU in adjusted analyses (mean difference of 0.07 in log10 HOMA-IR, P = .02) from birth through 36 months of life. Among CHEU, no differences in HOMA-IR were observed from birth through 36 months by in utero ARV exposure status or between AZT and NVP infant prophylaxis arms.

Conclusions: In utero HIV/ARV exposure was associated with lower insulin sensitivity throughout the first 36 months of life, indicating persistent early life metabolic disturbances which may raise concern for poorer metabolic health later in life.

Keywords: homeostatic model assessment; in utero HIV exposure; insulin sensitivity; metabolic; perinatal HIV exposure.