Protease inhibitor use and the incidence of diabetes mellitus in a large cohort of HIV-infected women

Newly diagnosed diabetes mellitus (DM), exacerbations of preexisting DM, hyperlipidemia, and fat redistribution syndromes have been described among protease inhibitor (PI) users.

In the current study, researchers examined the incidence of self-reported DM among participants of the prospective multicentre Women’s Interagency HIV Study from 1994-1998, a period spanning PI introduction. The study assessed the epidemiologic relationships among diabetes, weight change, stage of illness, virologic response to therapy, and the use of specific antiretroviral medications among participants with no prior history of DM.

The HIV-infected and uninfected women (n= 1,435 and 350, respectively) were similar in median age (37 years vs. 36 years), race/ethnicity (55% vs. 52% African American, 26% vs. 30% Latina), and median follow-up time (2.91 years). The HIV-positive women had a lower baseline median body mass index than uninfected women (25.5 kg/m2 vs. 26.4kg/m2) and were less likely to be obese, defined as BMI greater than 30kg/m2 (23% vs. 33%). Among HIV-positive women, 27% had baseline CD4 counts below 200 cells/microliter and 34% had HIV RNA levels below 4,000 copies/mL. Median detectable baseline HIV RNA was 4.75 log10.

HIV-infected women were divided into three groups. PI users reported at least one PI, at least one nucleoside reverse transcriptase inhibitor (NRTI), and/or at least one non-nucleoside reverse transcriptase inhibitor (NNRTI). Reverse transcriptase inhibitor (RTI) users reported at least one NRTI and/or at least one NNRTI but never reported use of any PI. Women in the No ART group never reported using any PI, NRTI, or NNRTI.

There were 69 new cases of self-reported DM among 1,785 women who contributed 4,578 person-years (PY), an overall incidence of 1.5 per 100 PY. Per 100 PY, DM incidence was 1.4 among the HIV-uninfected women and 1.2, 1.2, and 2.8 in the RTI, No ART, and PI groups, respectively. The relative risk (RR) of reporting DM was highest among the morbidly obese (RR= 5.2) and the obese (RR= 2.8). Adjusted for BMI, women on PI therapy had a greater risk of reporting DM than women on RTI therapy (RR=2.2).

Incident DM and first PI use were reported at the same study visit for nine of 20 (45%) of the diabetic participants in the PI group. The remaining participants reported incident DM one (n= 3), two (n= 6), or three (n= 2) study visits after the first report of PI use. PI therapy was used continuously in 15 of the 20 diabetic participants. In the remaining five, PI use and incident DM were reported at the same visit, despite earlier interruptions in PI therapy.

Virologic response to therapy was determined for 86% of the RTI and PI groups. In the RTI group, virologic response occurred in 25% of diabetic participants and 28% of nondiabetic participants. In the PI group, virologic response occurred in 53% of diabetic participants and 52% of nondiabetic participants.

Frequency of zidovudine (AZT), stavudine (d4T), zalcitabine (ddC), didanosine (ddI), or lamivudine (3TC) use was the same for both diabetic and nondiabetic participants. There was no significant association between NNRTI use and DM. The frequency of individual PI use was also the same for both diabetic and nondiabetic participants. Among the 69 incident DM cases, only one reported use of megestrol.

Adjusting for known risk factors for diabetes and other HIV-related confounders, a significantly increased risk of DM in all multivariate models was associated with PI use, older age, and larger BMI category. Changes in CD4 cell count and viral load were not independent predictors of DM in HIV-positive women.

“This study found that PI use is an independent risk factor for self-reported DM, with a threefold increase in risk, a result consistent with other studies … In view of the clinical benefits of PI therapy, concern about diabetes per se should not dissuade patients or practitioners from using this potent class of antiretrovirals. Routine screening for diabetes, particularly among older and heavier patients using PI therapy, is clearly warranted,” the authors concluded.


Justman JE, Benning L, Danoff A et al. Protease inhibitor use and the incidence of diabetes mellitus in a large cohort of HIV-infected women. J Acquir Immune Defic Syndr 2003 Mar 1;32(3):298-302

Source: CDC summaries


These results should not be surprising, because the inhibition of glucose transporter 4 by indinavir has been demonstrated in vivo and in vitro. Another mechanism may be the inhibition of insulin secretion from the pancreatic islet cell. In HIV-negative individuals, as shown by Noor et al (AIDS 2002), indinavir rapidly induces insulin resistance. Interestingly RTIs induce reduced insulin sensitivity in about 25% of the patients compared to 50% on PI (Walli et al. Eur. J Med Res. 2002).

Links to other websites are current at date of posting but not maintained.