The timing of infants’ HIV-1 infection relative to nevirapine-selective pressure determined 1 of 3 distinct dynamic patterns of nevirapine-resistant HIV-1. Notably, all infants with in utero infections and available birth specimens initially had wild-type HIV-1. The first pattern, observed in those with established in utero infection defined by elevated viral loads at birth, was characterized by a high rate of selection followed by rapid decay of nevirapine-resistant HIV-1. The second pattern, in infants with low but escalating viral loads at birth, consistent with acute in utero infection, included lower rates of resistance, but resistant viruses persisted over time. The third pattern occurred in infants with peripartum infection and consisted of a substantial fraction acquiring nevirapine-resistant HIV-1, which persisted at high concentrations. As expected, use of infant zidovudine, maternal single-dose nevirapine, and other maternal antiretroviral agents influenced the rates of nevirapine-resistant virus detected among infants.
According to accepted models, a single [34
] or several [35
] HIV-1 variants may infect an infant, and within a few weeks of primary infection virus populates a large number of susceptible cells [36
]. Once infection is established, a high rate of viral replication and a lack of proofreading activity by HIV-1 polymerase generate mutations at all positions in the viral genome [21
]. Archiving of mutants results in individuals harboring all single-base changes that confer high-level resistance to nevirapine shortly after infection [21
] that can be rapidly selected by administration of nevirapine monotherapy [37
In keeping with these models, infants who already harbor large viral populations at the time single-dose nevirapine is administered, as in the case of established in utero infection, should rapidly select archived mutants that confer nevirapine resistance. Indeed, consistent with the expected selective forces [38
], multiple mutant variants emerged in nearly every infant with a high viral load at birth, peaked by 2–4 weeks of age with concentrations approaching 100%, and subsequently decayed. An exception to this pattern occurred in infants also administered zidovudine or whose mothers took combinations of antiretroviral drugs. Viral replication under multidrug therapy requires multiple mutations, and as suggested by models [39
], it appears unlikely that such multidrug-resistant viruses evolve through random mutations.
The infrequent selection of nevirapine-resistant HIV-1 in infants experiencing acute in utero infection, the second pattern, is consistent with infants harboring a small HIV-1 population at birth, typically with limited viral diversity [34
]. An absence of nevirapine-resistant HIV-1 in their viral quasispecies would preclude the rapid selection of mutants. Among the few infants with selection of nevirapine-resistant HIV-1, the rapid expansion of mutants to high concentrations that persisted over time corresponds to viruses populating a large number of susceptible cells during acute infection [36
The third pattern, acquisition of 100% nevirapine-resistant HIV-1 that persisted at high concentrations, was prominent in infants with peripartum infection. However, because more than half of peripartum-infected infants acquired wild-type viruses, transmissions in these infants may have occurred before the selection of maternal mutants or from mothers who did not take nevirapine. Infant zidovudine did not appear to affect whether an infected infant had nevirapine-resistant HIV-1, which would be consistent with acquisition of nevirapine-resistant virus. Too few mothers of peripartum-infected infants took zidovudine in combination with single-dose nevirapine to evaluate its effect on the selection of nevirapine-resistant HIV-1, which we anticipate would decrease the selection of mutants as observed in infants with established in utero infections.
The rapid decay of nevirapine-resistant HIV-1 among infants with established in utero infections parallels the decay in post-partum women [16
], whereas the persistence of mutants in acutely infected infants mirrors adults acquiring drug-resistant HIV-1 [22
]. During acute infection, HIV-1 rapidly populates a large population of vulnerable cells [36
], which comprise much of the long-lived HIV-1 reservoirs [19
]. Therefore, the persistence of nevirapine resistance is expected when these mutants predominate during acute infection.
The threshold concentrations above which drug-resistant HIV-1 populations have clinical consequences have not been defined. The rapid decay of nevirapine-resistant HIV-1 among the infants with established in utero infections in our study suggests that after an interval nevirapine ART may be effective in these infants, as suggested in studies of African women [16
] and infants [16
The absence of nevirapine-resistant HIV-1 in nearly half (42%) of the infants we studied using a highly sensitive assay is notable because nevirapine ART would most likely effectively treat these infants. However, only 73% of mothers received single-dose nevirapine, and because maternal single-dose nevirapine increased the rate of resistance over infant-only single-dose nevirapine, a greater maternal uptake should increase the rate of nevirapine-resistant HIV-1 among their infants. Alternatively, the rate of nevirapine resistance should decrease with greater use of zidovudine and ART, as recently recommended by the World Health Organization [42
]. Given the opposing effects of maternal single-dose nevirapine and infant zidovudine on the rate of nevirapine-resistant HIV-1 in infants, along with our data showing wide variability in the decay of mutants, a cautious approach is warranted in estimating the likelihood of nevirapine-resistant HIV-1 in particular infants. Until studies can define these parameters and evaluate the utility of susceptibility testing for guiding the selection of infant ART, the current recommendation to use protease inhibitor–based ART for treatment of single-dose nevirapine–exposed infants [43
] is the safest approach to treating infants’ HIV-1 infection.
This study has several limitations. First, although the overall study was large, the small number of infants in some subgroups and variable follow-up rates limited our power to compare infants exposed to different maternal and infant antiretroviral regimens. Variable follow-up rates may have additionally reduced our capture of transient nevirapine-resistant HIV-1 and may therefore have resulted in underestimations of resistance. Lastly, our ability to determine which antiretrovirals (other than single-dose nevirapine) were taken by mothers and infants was limited by the accuracy of mothers’ self-report. Although maternity clinic records were checked, data were sparse when participants did not follow up after delivery.
Despite these limitations, we found that after use of single-dose nevirapine for prophylaxis against HIV-1 infection, the detection and decay of nevirapine-resistant HIV-1 in infants follow predictable patterns linked to the timing of HIV-1 infection. The reduced rates of nevirapine-resistant mutants we observed among in utero infected infants exposed to additional antiretroviral agents identify an additional benefit from combination therapies and argue against nevirapine-only programs for the prevention of MTCT. The absence of nevirapine-resistant HIV-1 and the rapid decay of mutants in a substantial proportion of infants suggest that nevirapine ART has the potential to effectively treat many infants. However, the persistence of nevirapine-resistant HIV-1 at high concentrations in certain infants differs from the rapid decay observed in women [5
] and suggests that nevirapine-based ART is unlikely to ever be efficacious in this subgroup. Testing for drug resistance may identify infants whose viral replication could be suppressed by nevirapine ART; however, first the thresholds for clinically relevant nevirapine mutations must be defined.