Background:  Abnormal glucose metabolism in HIV-infected patients may be a side effect of antiretroviral use or associated with non-medication related factors. Most studies of glucose metabolism in HIV-infected patients have focused on men or have been limited by lack of an appropriate control population.

Methods:  We performed a 75-g oral glucose tolerance test in 125 HIV-infected and 90 at-risk HIV-uninfected women without a history of diabetes, and assessed the association of antiretroviral use and non-medication related factors with impaired glucose tolerance, diabetes mellitus, and insulin resistance (HOMA).

Results:  Median age was 45 years (range 35 to 70); 51% were black, 38% Hispanic, 10% white; 38% had a family history of diabetes mellitus and 13% reported giving birth to a baby >9 lbs; median body mass index was 28.8 kg/m² and mean waist-to-hip ratio was 0.89; 90% had ever smoked cigarettes (median 15.0 pack-years); 68% were current smokers; 41% had a history of injection drug use with no difference by HIV status. Among HIV-infected women, 25% were HAART-naive, 23% were on HAART but protease inhibitor (PI)-naïve, and 52% were on HAART with PI. Median duration of PI use was 43 months. Median CD4 count was 481 cells/mm³. The prevalence of diabetes (fasting glucose ≥126 mg/dL or 2-hour glucose ≥200 mg/dL) among all women was 6% (n = 14) and of IGT (2-hour glucose ≥140 and <200) was 11% (n = 23), with no difference by HIV status, HAART, or PI use. Mean log insulin resistance (HOMA) (μU/mL·mM) was lower among HAART-naïve HIV-infected women (0.40) compared with those on PI-HAART (0.45), non-PI HAART (0.48), or HIV-uninfected women (0.47), but this difference was not significant. In a logistic regression model, factors independently associated with an abnormal oral glucose tolerance test (impaired glucose tolerance or diabetes mellitus) included age ≥50years (ORadj 4.5, 95%CI 1.5, 13.4) and smoking (ORadj 1.7 per 10 pack-years, 95%CI 1.2, 2.4), after controlling for HIV, HAART use, PI use, race, family history of diabetes, and waist-to-hip ratio. In a linear regression model, factors independently associated with log insulin resistance (HOMA) among HIV-infected women included body mass index (p <0.0005), Hispanic race (p = 0.047), and non-PI HAART (p = 0.04), after controlling for PI use and CD4 count. Conclusions:  Impaired glucose tolernce and diabetes mellitus were detected by oral glucose tolerance tests in a substantial minority of women, and were associated with traditional diabetes risk factors rather than HIV infection, PI or HAART use. However, among HIV-infected women, non-PI HAART use was independently associated with greater insulin resistance.

Keywords: diabetes mellitus; insulin resistance; antiretroviral therapy