Background: HIV-infected patients have a lower bone mineral density (BMD) and a higher incidence of fractures compared to the general population of same age and sex. To assess the impact of exposure to each ARV on the risk of osteoporotic fractures, we conducted a nested case-control study.
Methods: Cases were individuals enrolled while ART-naïve, with a first prospectively reported fracture between 01/2000 and 12/2010. Controls were randomly selected after matching on sex, age (±3 years), diagnosis period (< 1997/≥1997) and clinical centre. The risk of fracture was analysed using conditional logistic regression models. Exposure to each ARV was measured either by cumulative duration of exposure (model1), or by exposure yes/no (model2). In a 3rd model, the exposure variable was chosen for each ARV according to the lowest Akaike criterion value. All exposure variables and potential confounders were included in the multivariable models.
Results: Among the 861 reviewed fractures, 261 were osteoporotic fractures, and 254 were matched to at least one control (376 controls). The median year of fracture diagnosis was 2007(IQR 2004-2009), 67% of cases were men, 71% diagnosed with HIV-infection before 1997, median age was 49 years, CD4 436[293-592], nadir CD4 196[82-287], 31% at AIDS-stage, 65% with VL< 50cp/mL, and 49% exposed to tenofovir, 82% to PIs and 37% to efavirenz.
After accounting for transmission group, AIDS-stage, geographic origin, BMI, current smoking, alcohol consumption, exposure to systemic glucocorticoids, and period of enrolment, no association was found between the risk of fracture and exposure to tenofovir (OR for cumulative exposure: 1.03[0.86-1.24], similar results for exposure yes/no), or to NRTIs, or to PIs (exposure to PI overall: OR 1.01[0.92-1.11] or to each PI). Cumulative exposure to efavirenz was associated with a lower risk of fracture in models 1 and 3 (respective OR 0.81[0.69-0.95] and 0.82[0.70-0.96] per year of exposure), while exposure to efavirenz (yes/no) was not (OR 0.84[0.51-1.40]). Sensitivity analyses do not favour the causal nature of the association with exposure to efavirenz.
Conclusions: There was no evidence of excess risk of fracture following exposure to tenofovir or to PIs. This is an important result in the debate about TAF versus generic tenofovir.