Background:  The clinical role of assessing antiretroviral resistance in minor variants of HIV-1 remains to be defined. We reanalyzed a case-cohort substudy of AIDS Clinical Trials Group (ACTG) 5095, a randomized trial comparing the efficacy of efavirenz (EFV) + 2 or 3 NRTI, to determine the effect of minor NNRTI-resistant populations on response among subjects without major resistance to EFV-based regimens.

Methods:  The case-cohort sample included a randomly sampled cohort (n=220, 57 virologic failures, 163 non-failures) of the 765 EFV-randomized subjects, plus the remaining 136 unsampled subjects with virologic failure (n = 356: 193 virologic failures, 163 non-virologic failures). Bulk sequencing of HIV-1 protease and RT (TRUGENE) was performed on stored baseline plasma samples. In subjects without NNRTI resistance by bulk sequencing, presence of minority K103N or Y181C populations was assayed by allele-specific polymerase chain reaction (ASPCR). Prevalence of minority NNRTI-resistant variants (i.e., mutations detected by ASPCR but not by bulk sequencing) in the random cohort was compared with Fisher's exact test. Weighted Cox proportional hazards models estimated the relative risk of virologic failure in the presence (yes/no) of baseline minority Y181C by ASPCR for the case-cohort sample. Analyses were adjusted for recent adherence. An interaction was evaluated.

Results:  The detection threshold of ASPCR was calculated at:  K103N (AAT) = 0.001%, K103N (AAC) = 0.003%; Y181C = 0.04%. In the random cohort, median baseline plasma HIV-1 RNA (viral load) and CD4 count were 4.77 log₁₀ copies/mL and 206 cells/mm³, respectively. Of 345 subjects in the case-cohort sample with follow-up ≥16 weeks, 19 had baseline NNRTI resistance by bulk sequencing; 307 of the remaining 326 had at least 1 basaeline ASPCR result. Minority Y181C was detected more frequently than minority K103N (125 [41%] vs 42 [14%], respectively); 29 samples (9.5%) were positive for both mutations. In the random cohort, 25 of 43 (58%) with virologic failure vs 41 of 142 (29%) non-virologic failure had detectable Y181C (p = 0.001), whereas 5 of 43 (12%) virologic failure vs 14 of 142 (10%) non-virologic failure had detectable K103N (p = 0.8). Weighted Cox analyses showed a significantly increased risk of virologic failure for adherent and non-adherent subjects with detectable Y181C at baseline (HR 3.57, 95%CI 1.98 to 6.42, p <0.001; HR 6.09, 95%CI 2.83 to 13.1, p <0.001, respectively), compared to adherent subjects and no NNRTI mutations by either assay (interaction p = 0.08).

Conclusions: In antiretroviral-naive subjects without evidence of NNRTI resistance by bulk sequencing, presence of minority Y181C mutants detected by ASPCR more than tripled the risk of virologic failure of EFV-based regimens.