Background: Although experts have recommended testing for pretreatment drug resistance (PDR) before antiretroviral therapy (ART) initiation, there is little evidence to support its implementation. We aimed to establish whether an inexpensive point mutation assay can improve virological suppression by identifying PDR to guide drug selection for ART in a lower-middle income country. Methods: Investigators did an open-label, randomised controlled trial at three HIV treatment sites in Kenya: two in Nairobi and one in rural Maseno. Individuals (aged ≥2 years) were eligible to participate if they were confirmed HIV-seropositive, qualified for first-line ART, planned to reside in the area for more than 1 year, and provided informed consent. We randomly assigned participants (1:1) to either PDR testing by oligonucleotide ligation assay (OLA) to guide selection of ART or to standard of care, which did not include OLA testing. The OLA-guided therapy group had pre-ART peripheral blood mononuclear cells evaluated for drug resistance to non-nucleoside reverse transcriptase inhibitor (NNRTI) at codons Lys103Asn, Tyr181Cys, Gly190Ala, and to lamivudine at Met184Val, and when at least one drug-resistant codon was detected in a participant's pre-ART specimen, clinicians were directed to prescribe protease inhibitor-based second-line ART. Those without detected resistance and those who were randomised to standard of care received NNRTI-based first-line ART. The primary outcome was plasma HIV-1 RNA of at least 400 copies per mL at 4, 8, or 12 months after ART initiation, which defined virological failure, assessed in all participants who received treatment (data were censored for those lost-to-follow-up or who died). The study has been completed and is registered with ClinicalTrials.gov, NCT01898754. Findings: We screened 1198 participants between May 28, 2013, and Nov 4, 2014, of whom 991 (83%) were enrolled (492 received OLA and 495 received standard of care; four did not begin treatment). 93 participants (prevalence 9·4%) had PDR (95% CI 7·7–11·4). 34 (8·5%) of 400 participants in the OLA group had virological failure at month 12 of ART (95% CI 6·0–11·7) compared with 39 (9·7%) of 402 (7·0–13·0) in the standard-of-care group (log-rank p=0·26). Among participants with PDR, virological failure was lower in the OLA-guided therapy group than in the standard-of-care group: five (14%) of 35 compared with 13 (50%) of 26; p=0·0020). Among those prescribed NNRTI-based ART, participants given efavirenz were less likely to have virological failure than were those receiving nevirapine (odds ratio 0·37, 95% CI 0·22–0·62; p<0·0001). The OLA-guided therapy group had 39 serious non-lethal adverse events and 34 deaths. The standard-of-care group had 34 severe adverse events and 43 deaths, differences that were not significant. Adverse events judged to potentially be due to ART were few and similar between groups, with 17 (16%) in the OLA-guided therapy group and 16 (16%) in the standard-of-care group (p=0·90). Interpretation: Our finding that OLA testing for PDR reduced virological failure in only those with specific PDR mutations suggests that PDR poses less of a risk for virological failure than that predicted by past prevalence estimates, and that the value of PDR testing to reduce virological failure should be assessed for antiretroviral treatment regimens. Funding: US National Institutes of Health.
