Evaluating the performance of HIV testing in infants exposed to antiretroviral prophylaxis within South Africa’s prevention of mother to child transmission programme

Abstract

More than 280 000 HIV-exposed infants are born each year in South Africa (SA), all of whom require polymerase chain reaction (PCR) testing during early infancy. Failure to diagnose HIV early and initiate combination antiretroviral therapy (cART) is associated with rapid disease progression and mortality among infants with perinatal HIV-infection. Importantly, early infant diagnosis (EID) occurs within the context of efforts to prevent mother-to-child transmission (PMTCT) of HIV, including maternal cART and infant prophylaxis - with some infants exposed to virologically suppressive levels of antiretrovirals (ARVs) at time of testing. Although the decreasing mother-to-child transmission rate in SA, currently <5%, is expected to result in an increased rate of false-positive results there is also evidence to suggest infant exposure to ARVs is associated with reduced sensitivity of PCR tests. The aim of this study was to evaluate the performance of infant HIV testing within the context of SA’s PMTCT programme. Analysis of routine laboratory data from the National Health Laboratory Service’s (NHLS) Corporate Data Warehouse (CDW) demonstrated that each year, for the period 2010–2015, more than 17 000 specimens (>6%) failed to yield a laboratory verified positive or negative HIV PCR result. Indeterminate results accounted for nearly 3 000 of these ‘missed diagnostic opportunities’ each year. Since 2013, a standard operating procedure has been used within the NHLS defining HIV-detected results with a cycle threshold (Ct) >33.0 as indeterminate. Hence, indeterminate results are valid instrument-positive results that are interpreted as being inconclusive (i.e. not clearly positive but not negative either). Whereas the proportion of indeterminate results remained constant between 2013–2015, at approximately 17% of instrument-positive results per annum, this increased subsequent to the introduction of routine birth testing in June 2015. In 2017, more than a third of instrument-positive results among infants aged <7 days were verified as indeterminate. Importantly, among a cohort of birth-tested infants more than half of those with an indeterminate result were confirmed HIV-infected on follow-up testing. The high proportion of indeterminate results at birth can partly be accounted for by low-level viraemia - HIV-infected infants aged <1 month have been found to have markedly lower pre-treatment viral load compared with older infants (P <0.001). Furthermore, age-adjusted viraemia has significantly declined since 2010 (P <0.001), likely attributable to PMTCT practices. As EID Ct and plasma viral load are inversely correlated, with every one cycle increase in EID Ct associated with a 0.3 log10 RNA decrease in viral load (95% CI: -0.3–-0.2), it is clear that low level viraemia among HIV-infected infants is contributing to the high level of indeterminate results. Unless current NHLS verification practices are revised, increasing diagnostic uncertainty within EID services can be expected. By employing irreproducibility of instrument-positive specimens instead of Ct, to define indeterminacy, our findings suggest that the volume of indeterminate results can be reduced without exacerbating the false-positive rate. Findings from this study have informed World Health Organization recommendations and SA guidelines for EID and will improve accuracy of results and reduce delay in diagnosis and linkage to care.