Among African treatment-naïve women with CD4 count <200 cells/mm3, initial ART with NVP+TDF/FTC showed equivalent efficacy compared with LPV/r+TDF/FTC in intention-to-treat analysis of the primary endpoint of VF or death (83% versus 80% of women were alive and had not experienced VF after a median follow-up of more than 2 y; HR 0.85, 95% CI 0.56–1.29, which was within the pre-specified range of equivalence from 0.5 to 2.0). Equivalence was not established in the as-treated analysis, with lower rates of VF/death in the NVP arm compared with the LPV/r arm (HR 0.71, 95% CI 0.45–1.13). However, significantly more women discontinued NVP than LPV/r (28% versus 9%, p<0.001) including 14% versus none due to AEs. Consequently, analysis of overall regimen failure (VF, death, or discontinuation due to AEs) revealed lower rates of these combined endpoints in the LPV/r arm. The proportions of women experiencing ≥grade 3 signs, symptoms, or laboratory abnormalities did not differ significantly between arms, but total cholesterol and triglyceride increases were significantly larger in the LPV/r arm. Therefore, the higher rate of permanent discontinuation of NVP was primarily driven by protocol-mandated thresholds for permanent discontinuation of NVP compared with LPV/r, rather than higher rates of severe or life-threatening toxicities in the NVP arm. Overall, these data support the WHO recommendation of NVP/TDF/FTC as an initial affordable and effective HIV treatment regimen in RLS, and provide reassurance regarding the efficacy of this regimen. However, these results also underscore the importance of early toxicity monitoring with NVP-based regimens. The treatment failure (due to VF or treatment discontinuation) observed in both arms also highlights the importance of access to effective second treatment options, as well as consideration of other effective, better-tolerated first-line regimens (including among women of reproductive potential).
The overall and relative efficacy and toxicity of NVP-based versus PI-based ART are relevant in RLS. Globally, NVP is the most frequently used antiretroviral in combination with two NRTIs, due primarily to its low cost, lack of teratogenicity, and heat stability; and regimens including TDF+NVP (with either 3TC or FTC) are recommended by the WHO and increasingly used 
. However, prior small studies of first-line ART composed of NVP+TDF/FTC demonstrated rather high rates of early VF on this regimen, raising concern about its use 
. If LPV/r were found to be significantly more (or more durably) potent or less toxic than NVP in first-line treatment, then it could potentially represent a cost-effective choice for initial regimen in RLS (as reported in a previous cost-effectiveness analysis comparing these regimens among women with prior sdNVP exposure 
), although this has yet to be modeled in ARV-naive patients. To our knowledge, this is the first direct comparison of the efficacy of NVP- and LPV/r-based ART, and one of only a small number of studies to compare the efficacy of ART using NVP versus any boosted PI 
. This is also one of the largest randomized treatment trials to be conducted in HIV-infected women, who may experience different toxicities or response compared with men, but who are often underrepresented in trials. In one previous study (“ARTEN” 
), 569 treatment-naive patients were randomized to initiate NVP or atazanavir/ritonavir, each in combination with Truvada. Similar proportions achieved virologic suppression with both treatments, although more patients stopped treatment due to AEs in the NVP (14%) than the atazanavir (4%) arm 
. These findings are qualitatively similar to the OCTANE Trial 2 results, but are in contrast to those of an observational European study, in which patients were more likely to stop first-line NVP because of treatment failure but less likely to do so because of toxicity or patient/provider choice, compared with EFV or LPV/r 
Genotype analysis in our study of samples at the time of failure revealed NNRTI- or NRTI-resistance mutations in nearly one-half of women in the NVP arm. This frequency of resistance is lower than that reported in other studies of failure of initial therapy 
, and may be related to more rapid regimen switch (in the context of careful HIV-1 RNA monitoring) or to premature discontinuation of therapy for suspected toxicity. In the LPV/r arm, major PI resistance mutations were not detected at failure (similar to findings from other studies of LPV/r 
), and NRTI-associated mutations were less frequent (13%) than in the NVP arm (31%). The occurrence of the K65R mutation in 28% of participants failing NVP+TDF/FTC is notable. These resistance findings have implications for selection of first-line and subsequent ART regimens to optimize long-term clinical outcomes and reduce spread of drug resistance.
Strengths of our study included its randomized nature and very high visit and data completeness. Limitations include the possibility that some participants may have had previous sdNVP exposure. However, the accuracy of our sdNVP exposure ascertainment is supported by a low rate (0.8%) of baseline NVP resistance, and by results from trial 1 (which demonstrated higher rates of VF/death with NVP versus LPV/r treatment among women with prior sdNVP exposure 
). Subgroup analyses need to be interpreted with caution given the number of subgroup analyses considered; the one difference (between women with higher versus lower CD4 counts) identified would not be significant with adjustment for multiple comparisons and so could plausibly be a chance finding. We could not compare long-term morbidity and mortality between regimens because of the study duration. This was an open-label trial, and the lack of blinding constitutes another potential limitation. Finally, patient outcomes in a closely monitored clinical trial setting, including ramifications of toxicity, may be better than those expected in a routine treatment setting in RLS.
We conclude that in antiretroviral-naïve women with CD4 count <200 cells/mm3, initial ART with LPV/r/TDF/FTC and with NVP+TDF/FTC is equivalent in achieving and maintaining virologic suppression and preventing mortality, but that treatment cessation due to toxicity concerns, and drug resistance at the time of VF, are higher with NVP-based ART. Our findings suggest that NVP (with careful early toxicity monitoring) remains an acceptable choice for first-line ART in RLS, until better tolerated and potentially more efficacious regimens become accessible.