Tenofovir (TDF) Not Tied to Bone Fractures in French Case-Control Study

A case-control study has found no excess risk of fractures in HIV-positive people who have used tenofovir disoproxil fumarate (TDF, Viread) or protease inhibitors, according to a presentation at the 9th International AIDS Society Conference on HIV Science in Paris on July 26. The presenter, Dominique Costagliola, Ph.D., of the Sorbonne, said this was an important result that should inform the debate in high-income countries on the merits of TDF, which is available in many regions as a less-costly generic, versus the more recently approved prodrug tenofovir alafenamide (TAF, Vemlidy), which appears to be safer for bone density than TDF but is not available generically.

Costagliola provided a thorough review of the background to this issue. On average, people living with HIV have a lower bone mineral density (BMD) and a higher incidence of fractures than members of the general population of the same age and sex. Data from randomized control trials have shown that BMD decreases after starting antiretroviral therapy, in particular with regimens containing TDF. BMD also decreases when beginning protease inhibitors, but the decrease at the spine may be only an artifact linked to an increase in visceral fat, she said. Further, BMD increases after stopping antiretroviral therapy.

Whereas the associations between specific antiretrovirals and BMD are relatively clear, the picture is more complex when it comes to fractures. The available studies on the associations between specific antiretrovirals and fractures have important methodological differences, including whether they consider all factures or only 'low energy fractures' (fractures resulting from minimal trauma, such as a fall from standing height), as well as the range of confounders taken into account, Costagliola said.

Fractures are a rarer outcome than is a reduction in BMD. For TDF, four previous studies found no association, one found an increased risk, one found an increased risk that was not statistically significant and one found a decreased risk, Costagliola said. For protease inhibitors, three previous studies found no association, one found an increased risk that was not statistically significant and one found an association that the authors judged not to be causal, she said.

The aim of Costagliola's case-control study was "to assess the potential impact of exposure to each antiretroviral drug on the risk of low energy fracture[s] located at possible osteoporotic sites" in people living with HIV.

Cases and controls were recruited to the French Hospital Database on HIV between 2000 and 2010; they were antiretroviral-naive at the time of recruitment. Cases had sustained a low-energy fracture at a possible osteoporotic site (a site at which osteoporosis could be the cause). Osteoporotic sites were considered to be the vertebra, hip, wrist, humerus upper end, femur lower end and tibia upper end, as well as simultaneous fractures of three ribs. Controls were randomly selected from the database, matching on sex, age, year of diagnosis, clinical center and being under follow-up at the time of their case's follow-up.

Of 62,776 patients in the database, 1,026 reported a fracture. After excluding those with insufficient medical records and those whose fracture did not meet the researchers' criteria, there were 261 eligible cases. At least one control could be identified for 254 of these, resulting in 254 cases matched with 376 controls.

Two-thirds of cases and controls were men, with a median age of 49. They were well matched in terms of HIV parameters, with very similar viral load (less than 50 copies/mL), CD4 cell count (436 and 451 cells/mm3, respectively) and nadir CD4 (172 and 196 cells/mm3, respectively). However, cases were more likely than controls to have been diagnosed with HIV in the era prior to highly active antiretroviral therapy (71% and 67%, respectively), reflecting the long duration of infection of those affected by fractures. Cases were also more likely to be men who have sex with men (MSM; 34% and 20%, respectively), reflecting historical shifts in the French epidemic. (Diagnoses in the early years of France's HIV epidemic were concentrated in MSM, whereas more migrants from African countries have been diagnosed in recent years.)

In terms of fracture risk factors, cases were more likely to be underweight (13% versus 6%), to drink more than two glasses of alcohol a day (24% versus 13%), to have previously used systemic glucocorticoids (10% and 4%) and to have known osteoporosis (10% and 2%).

No association between use of TDF and fractures was found. This was true both in the first model, when exposure to the drug was measured by cumulative duration of exposure (adjusted odds ratio [AOR] 1.03, 95% confidence interval [CI] 0.86-1.24), and in the second model, when exposure was modeled as a dichotomous yes/no value (AOR 1.04, 95% CI 0.54-2.02).

Similarly, no association between use of protease inhibitors and fractures was found. This was true both for first-generation protease inhibitors grouped together and for all protease inhibitors grouped together, once again both in the first model, when exposure to the drug was measured by cumulative duration of exposure (AOR 1.01, 95% CI 0.92-1.11, data presented for all protease inhibitors), and in the second model, when exposure was modeled as a dichotomous yes/no value (AOR 0.98, 95% CI 0.73-1.31, data for all protease inhibitors). The only individual agent to show any association in adjusted models was atazanavir (Reyataz), in the first model only.

The analysis also highlighted a lower risk of fractures associated with cumulative use of efavirenz (Sustiva, Stocrin), in the first model only (AOR 0.81, 95% CI 0.69-0.95), but Costagliola suggested that this might not be causal.

Interpreting her results, Costagliola reminded her audience that the analysis focused only on low-energy fractures at possible osteoporotic sites, a restriction that not all previous researchers have followed. Nonetheless, her data are consistent with the fact that the majority of published studies have not found an association between fractures and exposure to either TDF or protease inhibitors. For TDF, she noted that its impact on BMD has generally been observed in the context of starting antiretroviral therapy, but there are few data on its impact in patients who have been treated for many years.

The abstract and slides for Costagliola's presentation are available online.