High level of HIV-1 drug resistance mutations in patients with unsuppressed viral loads in rural northern South Africa

BACKGROUND: Combination antiretroviral therapy (cART) has significantly reduced HIV morbidity and mortality in both developed and developing countries. However, the sustainability of cART may be compromised by the emergence of viral drug resistance mutations (DRM) and the cellular persistence of proviruses carrying these DRM. This is potentially a more serious problem in resource limited settings. METHODS: DRM were evaluated in individuals with unsuppressed viral loads after first or multiple lines of cART at two sites in rural Limpopo, South Africa. Seventy-two patients with viral loads of >1000 copies/ml were recruited between March 2014 and December 2015. Complete protease (PR) and partial Reverse Transcriptase (RT) sequences were amplified from both plasma RNA and paired proviral DNA from 35 of these subjects. Amplicons were directly sequenced to determine subtype and DRM using the Stanford HIV Drug Resistance Interpretation algorithm. RESULTS: Among the 72 samples, 69 could be PCR amplified from RNA and 35 from both RNA and DNA. Sixty-five (94.2%) viruses were subtype C, while one was subtype B (1.4%), one recombinant K/C, one recombinant C/B and one unclassified. Fifty-eight (84%) sequences carried at least one DRM, while 11 (15.9%) displayed no DRM. DRM prevalence according to drug class was: NRTI 60.8% NNRTI 65.2%, and PI 5.8%. The most common DRMs were; M184V (51.7%), K103N (50%), V106M (20.6%), D67N (13.3%), K65R (12%). The frequency of the DRM tracked well with the frequency of use of medications to which the mutations were predicted to confer resistance. Interestingly, a significant number of subjects showed predicted resistance to the newer NNRTIs, etravirine (33%) and rilpivirine (42%), both of which are not yet available in this setting. The proportion of DRM in RNA and DNA were mostly similar with the exception of the thymidine analogue mutations (TAMs) D67N, K70R, K219QE; and K103N which were slightly more prevalent in DNA than RNA. Subjects who had received cART for at least 5 years were more likely to harbour >2 DRM (p < 0.05) compared to those treated for a shorter period. DRM were more prevalent in this rural setting compared to a neighbouring urban setting. CONCLUSION: We found a very high prevalence of NRTI and NNRTI DRM in patients from rural Limpopo settings with different durations of treatment. The prevalence was significantly higher than those reported in urban settings in South Africa. The dominance of NNRTI based mutations late in treatment supports the use of PI based regimens for second line treatment in this setting. The slight dominance of TAMs in DNA from infected PBMCs compared to plasma virus requires further studies that should include cART subjects with suppressed virus. Such studies will improve our understanding of the pattern of drug resistance and dynamics of viral persistence in these rural settings.