Background:  Few studies have analyzed HIV diversity in children. It is not known whether diversity increases uniformly across the genome over time, whether it is associated with HIV subtype, or how it is impacted by ART. We analyzed env, gag, and pol diversity in 76 HIV⁺ Ugandan children age 6 months to 12.4 years who likely acquired HIV infection perinatally or through breast-feeding. Therefore, age is likely to reflect the duration of HIV infection. All children received ART in an observational study. Viral load was measured in the study, but ART was stopped only if there was clinical or immunologic decline.

Methods:  Plasma was analyzed from all children at baseline (prior to ART) and from 14 children who had a viral load >1000 copies/mL after 48 or 96 weeks of ART. Samples were analyzed using a high- resolution melting (HRM) diversity assay, which does not require sequencing and provides an HRM score that is significantly associated with genetic diversity in the region analyzed. HRM scores were generated for 2 regions in gag (GAG1, GAG2), 1 region in pol (POL), and 3 regions in env (ENV1, ENV2, ENV3, all in the gp41 coding region). HIV pol subtypes were determined for 71 children.

Results:  Baseline HRM scores were significantly higher in children >2 years of age than in younger children (p ≤0.001 for all 6 regions). In a multivariate model that included all 6 regions, higher HRM scores in POL and ENV3 were independently associated with older age (for POL, odds ratio [OR] 4.84, 95% confidence intervals [CI] 1.07 to 21.89, p = 0.04; for ENV3, OR 2.87, 95%CI 1.17 to 7.04, p = 0.02). HRM scores were not significantly associated with HIV subtype (A vs D). Among children who were clinically and immunologically stable on ART but were not virally suppressed, HRM scores in 4 regions were lower at 48 or 96 weeks on ART than baseline HRM scores (GAG1 p = 0.002; GAG2  p = 0.01; POL p = 0.01; ENV3 p = 0.003, Wilcoxon signed rank sum test). ENV1 and ENV2 HRM scores from children on non-suppressive ART were not significantly different from baseline HRM scores.

Conclusions:  Higher levels of HIV diversity were observed in children who were older at baseline. Different levels of diversity were observed in different regions of the HIV genome. Non-suppressive ART was associated with a reduction in viral diversity (genetic bottlenecking). However, bottlenecking was not observed in some regions of env.