HIV-1 Treatment Failure, Drug Resistance, and Clinical Outcomes in Perinatally Infected Children and Adolescents Failing First-Line Antiretroviral Therapy in Western Kenya

Abstract

Background: Long-term impact of drug resistance in perinatally infected children and adolescents living with HIV (CALWH) is poorly understood. We determined drug resistance and examined its long-term impact on failure and mortality in Kenyan CALWH failing first-line non-nucleoside reverse transcriptase inhibitor-based antiretroviral therapy (ART).

Setting: Academic Model Providing Access to Healthcare, western Kenya.

Methods: Participants were enrolled in 2010-2013 (timepoint 1) and a subsample re-enrolled after 4-7 years (timepoint 2). Viral load (VL) was performed on timepoint 1 samples, with genotyping of those with detectable VL. Primary endpoints were treatment failure (VL >1000 copies/mL) at and death before timepoint 2. Multinomial regression analysis was used to characterize resistance effect on death, failure, and loss-to-follow-up, adjusting for key variables.

Results: The initial cohort (n = 480) was 52% (n = 251) female, median age 8 years, median CD4% 31%, 79% (n = 379) on zidovudine/abacavir + lamivudine + efavirenz/nevirapine for median 2 years. Of these, 31% (n = 149) failed at timepoint 1. Genotypes at timepoint 1, available on n = 128, demonstrated 93% (n = 119) extensive resistance, affecting second line. Of 128, 22 failed at timepoint 2, 17 died, and 32 were lost to follow-up before timepoint 2. Having >5 resistance mutations at timepoint 1 was associated with higher mortality [relative risk ratio (RRR) = 8.7, confidence interval (CI) 2.1 to 36.3] and loss to follow-up (RRR = 3.2, CI 1.1 to 9.2). Switching to second line was associated with lower mortality (RRR <0.05, CI <0.05 to 0.1) and loss to follow-up (RRR = 0.1, CI <0.05 to 0.3).

Conclusion: Extensive resistance and limited switch to second line in perinatally infected Kenyan CALWH failing first-line ART were associated with long-term failure and mortality. Findings emphasize urgency for interventions to sustain effective, life-long ART in this vulnerable population.

Figure 1:. Flow diagram of study participants from CAMP (timepoint 1) to RESPECT (timepoint 2).

The Figure indicates which subgroups were used for analyses and testing in this study and their relevant results, including 480 timepoint 1 participants with available samples, for which VL testing was performed (top); 167 samples with detectable VL at timepoint 1, for which genotyping was attempted; 128 sample from which genotypes were obtained; 96 participants with available clinical information between timepoints 1 and 2 (gray cells); and 79 participants who were re-enrolled in timepoint 2 (bottom), for who VL testing was performed again. Switch to 2^nd-line ART and treatment failure status are shown for the groups for which these data were available. Abbreviations: C/mL, copies/milliliter; VL, viral load.

Figure 2:. Predicted Drug Resistance by HIV-1 Subtype to current and future antiretroviral regimens in 128 CAMP (timepoint 1) participants with genotypes.

The figure demonstrates detailed information (one row per participant) of one of five predicted levels of drug resistance according to Stanford Database scores, to current regimens (left panel) and future regimens (right panel). Resistance levels are color coded based on the legend on the left of the panels. HIV-1 subtypes are shown in gray. 3TC, lamivudine; ABC, abacavir; ARV, antiretroviral; AZT, zidovudine; D4T, stavudine; EFV, efavirenz; NVP, nevirapine; TDF, tenofovir; ETR, etravirine; RPV, rilpivirine.

Figure 3:. Associations between drug resistance at CAMP (timepoint 1) enrollment and outcomes at RESPECT (timepoint 2) re-enrollment.

Relative Risk Ratios (RRR) and 95% Confidence intervals from the 4-category outcome adjusted multinomial regression models, along with the adjusted p-value representing the false discovery rate (FDR), with an * placed for FDR<0.05. The comparison of the association between NVP and EFV resistance among the failures and suppressed participants (plot B) are done using Fisher Exact test because only four children did not have EFV and NVP resistance, none of whom were failing at RESPECT(timepoint 2) enrollment. Confidence intervals and point estimates beyond the plot limits are greater than 100 and the exact values are not shown. RRR are shown on the log scale for readability. Resistance is defined as intermediate or higher predicted resistance using Stanford susceptibility scores.