Background:  Long-term effects of abacavir/lamivudine (ABC/3TC) vs tenofovir/emtricitabine (TDF/FTC) combined with either efavirenz (EFV) or atazanavir /ritonavir (ATV/r) on bone mineral density and limb fat have not been compared.

Methods:  A5224s was a trial of HIV-infected, treatment-nave patients randomized to blinded ABC/3TC vs TDF/FTC with open-label EFV or ATV/r. Primary endpoints included dual energy x-ray absorptiometry (DEXA) -measured changes in lumbar spine and hip bone mineral density and the presence of lipoatrophy at week 96. Secondary endpoints included changes in limb fat (DEXA). Lipoatrophy was defined as ≥10% loss of limb fat from baseline. Primary analyses were intent-to-treat with as-treated being secondary. Statistical tests included linear regression and Fisher’s exact test.

Results:  We randomized 269 eligible subjects equally to ABC/3TC with EFV or ATV/r; or TDF/FTC with EFV or ATV/r. At baseline, 85% were male, 47% white, median HIV-1 RNA 4.6 log₁₀ copies/mL, age 38 years, weight 76 kg, CD4 233 cells/µL, and limb fat 7.3 kg. There was no significant evidence of an interaction between the NRTI and the EFV and ATV/r components for any of the bone mineral density or limb fat endpoints (P ≥0.52). At week 96 the mean percentage changes from baseline in lumbar spine and hip bone mineral density for ABC/3TC vs TDF/FTC were –1.3% vs –3% (P = 04) and –2.6% vs –3.9% (P = 0.025); and for EFV vs ATV/r were –1.7% vs –3.2% (P = 0.035) and –3.1% vs –3.4% (P = 0.59), respectively. Overall, 5.6% of subjects had ≥1 fracture (all traumatic), with no significant differences in fracture rates between the NRTI arms (P = 1.0), or between EFV vs ATV/r arms (P = 0.29). At week 96 for each NRTI, the proportion with protocol-defined lipoatrophy (one-sided upper 95%CI limit) were 18% (25%) for ABC/3TC and 15% (22%) for TDF/FTC; P = 0.70 between NRTI arms. As-treated analysis showed similar results. At week 96 mean changes in limb fat were higher in ABC/3TC vs TDF/FTC (+1.7 vs +1.1 kg; P = 0.12 in intent-to-treat and +2.1 vs +1.2 kg; P = 0.02 in as-treated), but lower in EFV vs ATV/r (+1.0 vs +1.9 kg; P = 0.008 in intent-to-treat and +1.0 vs +2.0 kg; P = 0.01 in as-treated). No statistically significant differences were seen in percentage change in limb fat between the NRTI arms by intent-to-treat or as-treated analyses

Conclusions:  Compared to ABC/3TC-, TDF/FTC-treated patients had significantly larger decline in lumbar spine and hip bone mineral density. ATV/r led to more significant losses in lumbar spine but not hip bone mineral density, than EFV. Both ABC/3TC- and TDF/FTC-based regimens on average increased limb fat at week 96, with similar proportion of patients with lipoatrophy.