Background: Exposure to some antiretrovirals (ARVs), in particular thymidine analogue nucleosides, has been associated with a higher risk of diabetes. However, this risk may vary according to ARVs and combinations. We estimated the risk of new onset diabetes in a large HIV-infected adult cohort in Thailand and studied the impact of four different nucleos(t)ide reverse transcriptase inhibitors (NRTIs) containing first-line regimens.
Methods: We selected all HIV-1 infected, antiretroviral therapy (ART)-naïve adults, enrolled in the multicenter PHPT cohort in Thailand (NCT00433030) between January 1, 2000 and December 31, 2011, with no history of diabetes, who received exclusively and for at least 2 years tenofovir disoproxil fumarate (TDF), zidovudine (ZDV), stavudine (d4T) or didanosine + stavudine (ddI+d4T) as part of their first-line regimen. Diabetes was defined as confirmed either fasting plasma glucose ≥126 mg/dL or random glucose ≥200 mg/dL. Incidence of diabetes was estimated as the number of new cases divided by the total number of person-years. Cox proportional hazards models were used to compare the risk of diabetes between regimens.
Results: A total of 520 HIV-infected patients, 329 (63%) female, participated in this analysis. At ART initiation, median age was 34.1 years (interquartile range 29.5-40.1), body mass index (BMI) 20.7 kg/m2 (18.9-22.9), CD4 count 139 cells/mm3 (74-208) and HIV RNA load 4.8 log10 copies/mL (4.2-5.2). 329 (63%) patients received TDF, 28% ZDV, 7% d4T and 2% ddI+d4T, usually in addition to lamivudine or emtricitabine. Over 3,318 person-years, 13 patients met the criteria for diabetes. Incidence of new onset diabetes was 3.9 per 1,000 person-years (95% confidence interval [CI] 2.3-6.7). Upon multivariate analysis, adjusting for gender, age, BMI, hepatitis B surface antigen, hepatitis C antibody and CD4 cell count at ART initiation, the adjusted hazard ratios for new onset diabetes were 6.8 (95% CI 1.6-30.0) in patients on ZDV, 14.7 (1.3-167.8) on d4T, and 91.3 (12.4-674.2) on ddI+d4T compared to those on TDF containing regimens (reference) (p <0.001).
Conclusions: Overall, the incidence rate of diabetes in this lean and predominantly young, female population was relatively low. However, first-line use of ZDV, d4T and d4T+ddI resulted in increased diabetes incidence. In most ART programs, d4T as well as ddI have now been phased out but our study shows that ZDV as first-line regimen was associated with a higher risk of diabetes than TDF.