A new analysis of D:A:D presented at the Eighth International Congress on Drug Therapy in HIV Infection in Glasgow last week looked at the risks associated with developing new-onset diabetes. It found that three nucleoside (NRTI) drugs were significantly associated with developing diabetes (d4T, ddI and AZT); that nevirapine appeared to have a protective role; and that, rather surprisingly, no protease inhibitor appeared to be related to diabetes development despite this drug class’s association with metabolic abnormalities.
Indeed having ritonavir in the regimen appeared to be mildly protective, though this could be an artefact since taking ritonavir-boosted protease inhibitors could be a surrogate marker for regimens featuring more up-to-date NRTIs which cause less diabetes.
The D:A:D (Data Collection on Adverse Effects of Antiretroviral Drugs) Study is the largest cohort study following a group of patients taking anti-HIV drugs and charting the risk of certain side effects. It has already produced significant data on the relationship between antiretroviral exposure and cardiovascular events. It is now producing significant data on the development of diabetes.
D:A:D contains 33,389 patients from 11 patient groups in Europe, the USA and Australia and with six years of longitudinal data it can now begin to look at other side effects as well as strokes and heart attacks, and can begin to distinguish between the effects of individual drugs and drug classes. It has already begun to do so for cardiovascular events – see this report.
In this analysis diabetes was defined as having a fasting blood glucose level of more than 7.0 mmols per litre, measured on at least two consecutive occasions. If such data was missing, taking anti-diabetes therapy was regarded as a ‘possible’ diagnosis.
Importantly, D:A:D did not measure insulin resistance, the precursor of diabetes in which the body becomes less able to process sugar: this can be defined in different ways one definition is a fasting blood glucose level of more than 6.1 mmol/l. Some commentators wondered if the surprising lack of a diabetes ‘signal’ from protease inhibitors was due to the study not picking up on milder cases of insulin resistance.
What the figures tell us is that 952 (2.65%, or one in 35) of the D:A:D Study patients had diabetes when they entered the study. During the study’s six years another 745 patients were diagnosed with diabetes. This is equivalent to an incidence of 0.572% per patient/year or one diabetes diagnosis for every 175 patients in a year. This implies a doubling of diabetes prevalence every 4.6 years and the risk did appear to rise significantly over the course of the study: the investigators calculate that there is an increased risk of diabetes of 6% for every year spent on antiretroviral therapy, although there appeared to be something of a leveling-off after four years on therapy.
Stavudine (d4T, Zerit) was by some way the drug most strongly associated with developing diabetes with an increased risk of 19% for every year spent on d4T, and this risk did not appear to level off. The other two nucleosides associated with diabetes were ddI (didanosine, Videx) and AZT (zidovudine, used most frequently in Combivir) though they carried the same relative risk as ARV therapy in general (6% a year).
Nevirapine (Viramune) appeared to have a protective effect, with a reduced risk of 11% for every year spent on the drug and, at reported above, taking ritonavir was associated with a small protective effect too (6% less risk per year).
All these risks were statistically significant and they were also significantly associated with other metabolic changes: the risk of developing diabetes was significantly associated with rises in cholesterol, triglycerides and the development of lipoatrophy (fat wasting) and with falls in ‘good’ HDL-cholesterol (which may explain the nevirapine effect, as this drug seems to raise HDL-cholesterol relative to total cholesterol).
Being obese, older, male, or black were all associated with diabetes too. Smoking was not, indeed it was slightly negatively associated, possibly due to nicotine’s appetite-suppressant effects.
The investigators said that it remained to be found out whether d4T directly caused impaired glucose intolerance and diabetes or whether the development of lipoatrophy led to diabetes as a consequence. It was possible that a small but direct link between protease inhibitors and diabetes (which has been seen in other studies) was masked by a larger but indirect link between d4T, lipoatrophy and diabetes.
De Wit S et al. Relationship between use of stavudine and diabetes mellitus. Eighth International Congress on Drug Therapy in HIV Infection, Glasgow. Abstract PL9.5. 2006.