Predictors of longitudinal change in bone mineral density in a cohort of HIV-positive and negative patients.

OBJECTIVE Although low bone mineral density (BMD) is prevalent in HIV, changes in BMD over time remain unclear. We aimed to compare rates of, and factors associated with, BMD change between HIV-positive and HIV-negative patients. METHODS In a prospective, 3-year cohort, HIV-positive and HIV-negative patients provided annual demographic and clinical data, fasting bloods, and dual x-ray absorptiometry. Using longitudinal mixed models we compared and determined predictors of rate of change in BMD. RESULTS Of 384 study participants (45.8% HIV positive), 120 contributed two and 264 contributed three BMD measurements. Those with HIV were younger [median interquartile range 39 (34-46) vs. 43 (35-50) years; P = 0.04], more often men (61 vs. 46%; P = 0.003), and less likely Caucasian (61 vs. 82%; P < 0.001). Although BMD was lower in those with HIV, BMD declined in both groups, with nonsignificant between-group difference in rate of BMD change over time. Within the HIV group, starting antiretroviral therapy (ART) within 3 months of enrolment was associated with greater BMD decline at all anatomical sites (all P < 0.001). Age more than 30 years, Caucasian ethnicity, and not being on ART during follow-up were associated with greater decline and higher parathyroid hormone associated with a smaller decline in BMD at the femoral neck. We found no association between BMD change and exposure to tenofovir disoproxil fumarate or protease inhibitors. CONCLUSION We observed no difference in rate of BMD decline regardless of HIV status and in HIV-positive patient, having started ART within the previous 3 months was the only factor associated with greater BMD decline at all three sites.