Antiretroviral treatment outcome and molecular epidemiology of HIV-1 subtype C drug resistance in Ethiopia

Abstract: Over the last decade, antiretroviral therapy (ART) has been rolled out in Ethiopia at a large scale. However, the outcome and aspects of drug resistance have been poorly studied at the national level. Also, the epidemic has been reported to be dominated by HIV-1 subtype C (HIV-1C), but this has not been verified at a nationwide level. In this thesis, I aimed at studying the outcome of ART and molecular epidemiology of HIV-1C drug resistance in Ethiopia using the first countrywide HIV-1 cohort. Data and plasma samples were collected from 874 adult and adolescent patients (age ≥ 14 years; 60% females), recruited from seven university hospitals during 2009 – 2011. In Paper I, by on-treatment (OT) and intention-to-treat (ITT) analyses we determined the rate of treatment failure at month 6 and 12. Four multivariable logistic regression models were developed to identify baseline predictors of treatment outcome. OT analysis identified treatment failure in 8% and 7%, whereas with ITT analysis the figures were 23% and 34% at month 6 and 12, respectively. Hence, our study indicated early death and lost to follow up (LTFU) as the main risk factors for poor treatment outcome. In addition to the well-known baseline factors of ART outcomes, study site was identified as a strong predictor of failure, where regional hospitals had higher proportions of treatment failure and LTFU as compared to a national tertiary level hospital in the capital city. In Paper II, we assessed surveillance drug resistance mutations (sDRM) in major and minor viral populations by population-based Sanger sequencing (PBSS) and next-generation sequencing (NGS). The short-term impact of sDRM on the outcome of ART was also assessed. NGS detected sDRM associated with reverse transcriptase inhibitors (RTIs) and protease inhibitors (PIs) more frequently than PBSS as well as major integrase strand transfer inhibitors (INSTIs) DRMs in minor viral variants. The baseline RTI-DRMs were associated with treatment failure at month 6 and 12. In Paper III, by genotypic analysis of the V3-loop of env, we described the HIV-1C molecular epidemiology and co-receptor tropism trend. The epidemic has been still found to be monophylogenetically dominated by CCR5-tropic HIV-1C even in patients with advanced immunodeficiency, although a slight increasing temporal trend was observed in CXCR4- containing strains. In Paper IV, genotypic resistance testing (GRT) of the HIV-1 pol gene described acquired DRMs at month 6 and 12 and near-full length genome (NFLG) analysis assessed amino acid changes in the gag, pol, vif, vpr, tat, vpu, and nef genes in paired baseline and month 6 samples of virologic failures. Broad major RTI- DRMs were detected in most failure patients. K65R, at a high rate, was identified only in TDF treated patients. The NFLG assay described all target regions of interest for HIVDR. In conclusion, early death and LTFU are major contributors for ART failure in Ethiopia rather than detectable viremia; the universal test and treat strategy could possibly improve the treatment outcome. There is a geographical variation in ART outcome, which calls the need for provision of more support at regional hospitals. HIV-1C still dominates monophylogenetically the epidemic in Ethiopia. PBSS reports a broad DRM, but underestimates prevalence of the transmitted drug resistance (TDR). NGS identifies an even higher rate of DRMs including in the minor viral variants. Our NFLG assay covers all relevant target genes and is an attractive cost-efficient alternative for HIVDR surveillance.