ABSTRACT: Factors that influence viral response when antiretroviral therapy (ART) is initiated in neonates are not well characterized. We assessed if there is consistency in predictive factors when operationalizing viral response using different methods. Data were collected from a clinical study in South Africa that started ART in neonates within 14 days of birth (2013-2018). Among 61 infants followed for ≥48 weeks after ART initiation, viral response through 72 weeks was defined by three methods: (1) clinical endpoints (virologic success, rebound, and failure); (2) time to viral suppression, i.e., any viral load (VL: copies/mL) <400, <50, or target not detected (TND) using time-to-event methods; and (3) latent class growth analysis (LCGA) to empirically estimate discrete groups with shared patterns of VL trajectories over time. We investigated the following factors: age at ART initiation, sex, birthweight, preterm birth, mode of delivery, breastfeeding, pre-treatment VL and CD4, maternal ART during pregnancy, and maternal VL and CD4 count. ART was initiated 0-48 h of birth among 57.4% of the infants, 48 h-7 days in 29.5% and 8-14 days in 13.1%. By Method 1, infants were categorized into 'success' (54.1%), 'rebound' (21.3%), and 'failure' (24.6%) for viral response. For Method 2, median time to achieving a VL <400, <50, or TND was 58, 123, and 331 days, respectively. For Method 3, infants were categorized into three trajectories: 'rapid decline' (29.5%), 'slow decline' (47.5%), and 'persistently high' (23.0%). All methods found that higher pre-treatment VL, particularly >100,000, was associated with less favorable viral outcomes. No exposure to maternal ART was associated with a better viral response, while a higher maternal VL was associated with less favorable viral response and higher maternal CD4 was associated with better viral response across all three methods. The LCGA method found that infants who initiated ART 8-14 days had less favorable viral response than those who initiated ART earlier. The other two methods trended in a similar direction. Across three methods to operationalize viral response in the context of early infant treatment, findings of factors associated with viral response were largely consistent, including infant pre-treatment VL, maternal VL, and maternal CD4 count.
