Background: In previous studies, protease inhibitors (PIs) have been associated with an increased risk of death and non-AIDS events, such as cardiovascular events and non-AIDS cancers. We investigated whether specific PIs and non-nucleoside transcriptase inhibitors (NNRTIs) are associated with increased non-AIDS mortality.
Methods: D:A:D study participants were followed from study enrolment until earliest of death, 1/2/2013 or last clinic visit. Exposure to specific PIs/NNRTIs was classified as recent (current use/use in last 6 months) or cumulative (/year). Poisson regression compared relative death rates (RR) for both types of exposure in two separate models. Follow-up among individuals dying from AIDS-related causes was censored on date of death.
Results: 3276 non-AIDS deaths occurred in 371,333 person years (PYRS) (incidence: 8.8/1000 PYRS; 95% CI; 8.5-9.1). In follow-up for which the current CD4 count was >500 cells/mm3, death rates were highest for those currently receiving indinavir (rate: 7.4; 5.4-9.5) and lowest for those currently receiving efavirenz (EFV 4.2; 3.6-4.9). Relative differences were similar across time-updated CD4 /viral load (VL) strata. After adjustment, there was no significant association between recent exposure to commonly used PIs and increased death rates. In contrast, recent exposure to EFV (RR: 0.86) was significantly associated with a decreased death rate (Figure). For cumulative exposure (RR/year longer), the commonly used PIs/NNRTIs lopinavir/ritonavir (LPV/r), atazanavir (ATV), saquinavir (SQV) and nevirapine (NVP) were significantly associated with small increases in death rates (Figure). Corresponding RR/10 years longer were; LPV/r (1.69; 1.36-2.09), ATV (1.69; 1.24; 2.30), SQV (1.50; 1.18- 1.89) and NVP (1.34; 1.14-1.58). Results were consistent across CD4/VL strata; when restricting analyses to those currently on antiretroviral drugs; excluding unknown causes of deaths and excluding intravenous drug users.
Conclusions: Our findings, based on a substantial sample size, suggest that cumulative exposure to some PI/NNRTIs is associated with a small but gradual increased risk of non-AIDS mortality. This effect is consistent among various types of PIs and of an extent comparable to earlier findings for non-AIDS events. Conversely, recent exposure to EFV was associated with a reduced risk of non-AIDS mortality. Choice of PIs/NNRTIs may affect long-term prognosis and although potential confounding cannot be ruled out, results argue for continued pharmacovigilance.