HIV Articles  
Women exposed to single-dose nevirapine in successive pregnancies: effectiveness and nonnucleoside reverse transcriptase inhibitor resistance
27 April 2009 - Volume 27 - Issue 7 - p 809-816
doi: 10.1097/QAD.0b013e328323ad49
McIntyre, James A; Martinson, Neil A; Morris, Lynn; Johnson, Jeffrey; Gray, Glenda E; Pillay, Visva; Ledwaba, Johanna; Dhlamini, Puleng; Cohen, Sarah; Puren, Adrian; Steyn, Jan; Heneine, Walid aJohns Hopkins University School of Medicine, Baltimore, Maryland, USA bPerinatal HIV Research Unit, University of the Witwatersrand, South Africa cNational Institute for Communicable Diseases, Johannesburg, South Africa dDivision of HIV/AIDS Prevention, Centers for Disease Control and Prevention, Atlanta, Georgia, USA.
Objective: To assess the impact of prior exposure to single-dose nevirapine (sdNVP) on mother-to-child transmission and genotypic resistance in HIV-infected women.
Design: Prospective study of 120 women exposed to the HIVNET 012 sdNVP regimen in two successive pregnancies and 240 antiretroviral (ARV)-naïve, multiparous women who received sdNVP for the first time.
Results: One hundred and eight of 120 and 193 of 240 women returned for a postpartum visit by 6 weeks. HIV-1 was detected in 11.1% (95% confidence interval = 5.9-18.6) of the infants of women previously exposed to sdNVP and 4.2% (95% confidence interval = 1.3-7.0) of those exposed for the first time (P = 0.028). Rates of maternal HIV-1 genotypic resistance at 6 weeks postdelivery were 37.5% and 46.4%, respectively (P = 0.119). Sensitive mutation-specific real-time PCR testing found three of 12 previously exposed women who transmitted HIV-1 to their infants had either K103N or Y181C at baseline compared with one of eight ARV-naïve, transmitting women who had Y181C. None of 40 randomly selected nontransmitting women from either group had detectable NVP resistance mutations prior to sdNVP exposure.
Conclusion: This study shows that effectiveness of sdNVP may be compromised by prior exposure to sdNVP, although the increase in transmission rate after prior exposure could not be explained by the detection of NVP resistance mutations prior to re-exposure as measured both by standard genotyping and highly sensitive allele-specific PCR assays. Furthermore, transmission rates of women with prior exposure were not higher than those reported elsewhere.
Single-dose nevirapine (sdNVP), administered by itself to pregnant mothers at the onset of labour and to newborns soon after birth, reduces peripartum mother-to-child transmission (MTCT) of HIV-1 [1,2]. It is inexpensive, safe and is used extensively in resource-poor settings. Demonstrations of enhanced potency when used in combination with antenatal and neonatal zidovudine (ZDV) [3] and ZDV with lamivudine (3TC) [4] ensure the continued use of sdNVP in these settings despite conventional genotyping assays revealing that 19-69% of women will develop resistance mutations [5-7], which have been shown to persist in 14% of mothers for upto 6 months after exposure [8]. More sensitive assays suggest that the majority of women develop resistant mutations after sdNVP [9-11]. Furthermore, the selection of nonnucleoside reverse transcriptase inhibitor (NNRTI) resistance mutations detectable by conventional methods following exposure to sdNVP appears to have a negative impact on virological response to subsequent NNRTI-containing antiretroviral (ARV) therapy if started within 6 months of exposure [12,13].
Whether sdNVP retains its effectiveness to prevent MTCT of HIV-1 when used in second labour has become an increasingly important issue. First, although birth rates in several sub-Saharan African countries have fallen [14], many high HIV prevalence countries have total fertility rates in excess of four births per woman [15] and fertility preferences of HIV-infected women are similar to those of uninfected women [16,17], although their fertility is reduced compared with uninfected women [18,19]. Second, a large number of HIV-infected women in developing countries are now receiving sdNVP [20] either alone or with AZT. It is likely, therefore, that an increasing number of HIV-infected pregnant women will have a second exposure to sdNVP, possibly when they have resistant virus. The objective of this study was to compare mother-to-child 6-week HIV-1 transmission rates and drug resistance rates in women who received sdNVP in a second labour (NVP-2) with multiparous, HIV-infected, ARV-naïve women who received their first exposure to sdNVP (NVP-1).
This preliminary study shows that the effectiveness of sdNVP in preventing peripartum HIV-1 transmission a second time may be reduced when compared with multiparous women receiving sdNVP for the first time. We were unable to show, however, that HIV-1 NNRTI resistance was the cause for higher transmission rates in the NVP-2 group as both population level sequencing and sensitive AS-PCR did not demonstrate higher rates of resistance prior to sdNVP re-exposure. The proportion of HIV-infected infants from the NVP-1 group was very low (4.2%) compared with that of women receiving sdNVP in two consecutive pregnancies (11.1%).
Our findings are at odds, however, with two previous reports describing effectiveness of sdNVP used a second time, including a report of our own. Data from Kampala, Uganda suggest that transmission following a second exposure was not significantly higher in women attending the MTCT program the first time [26]. Using two analyses, similar rates of HIV-1 transmission in NVP exposed and unexposed women are reported; in their analysis that most closely resembles our study, 6-week transmission rates were 17..9 (95% CI = 8.9-34.0) and 18.7% (95% CI = 11.1-30.6; P = 0.92), respectively. Our study comparing transmission rates of first and second pregnancies of women exposed to sdNVP in both pregnancies, which included all women in the NVP-2 group from this study and another 41 women from Abidjan, suggested that transmission rates did not increase in the second pregnancy compared with the first despite progression of HIV-1 and despite prior exposure to sdNVP [21]. Moreover, in our study, HIV-1 transmission in the NVP-2 group for their second pregnancy (11.1%) is similar to transmission rates reported from previous studies of women exposed to sdNVP for the first time [1,27-29] and to a report of operational effectiveness of sdNVP from South Africa [30], but is lower than rates seen among women who receive no intervention to prevent MTCT [31-34]. Thus, although the transmission rates are statistically significantly higher in the NVP-2 group when compared with the NVP-1 group, the NVP-2 rate is not unusually high for intra- and peripartum transmission in settings in which sdNVP is used. We are, however, unable to explain the low transmission rate (4.2%) seen in the multiparous ARV-naive women attending MTCT clinics in Soweto for the first time as it is considerably lower than any reported previously.
A potential explanation for transmission rate differences may be differential ingestion of NVP [35] or other ARVs between the NVP-2 and NVP-1 women and infants, but this seems unlikely as similar proportions of women self-reported taking their NVP doses and written evidence of infant dosing was found in similar proportions of infants from both groups but we did not validate maternal self-report of taking sdNVP with maternal and/or infant NVP assays. Furthermore, ARVs other than NVP were not widely available to pregnant women in the public sector at the time of the study and viral loads of the NVP-1 women in their 6-week-postpartum visit were not significantly lower than those of the NVP-2 women. Additionally, rates of genotypic resistance in NVP-1 women at 6 weeks, although higher, were not statistically significantly different, suggesting similar exposure to NVP. The slightly higher percentage may be linked to our finding that NVP-1 women were more likely to develop multiple mutations compared with single mutations found in previously exposed women. However, the study sample size was not powered to detect small differences. Another explanation may be differential infant mortality rates. When all infant deaths that occurred prior to HIV-testing are assumed to be HIV-infected, transmission differences between the two groups are less marked.
Despite over 30 000 pregnant women being HIV-tested annually in Soweto, it was difficult to identify and recruit NVP-2 women, possibly due to stigma [36], and we have no data on numbers of potentially eligible NVP-2 women attending during the study period; the women we recruited, therefore, may not be representative of all NVP-2 women and unmeasured differences between the two groups (NVP-2 and NVP-1) may be residual confounders; for example, differential rates of losses to follow-up and differences in interdelivery duration. The latter are not unexpected as multiparous women attending the PMTCT program for a first time had their previous child prior to the establishment of the PMTC program. Our data cannot discriminate between intrapartum and peripartum transmission as we did not subject neonates to HIV testing soon after birth. Although we assayed both proviral DNA and circulating RNA, we did not assess the presence of resistance mutations in resting T cells [37,38], which may hold an archive of mutant viruses that could yield further insights into differences in transmission rates. However, the short exposure to NVP and data from Botswana showing no impact on ARV response in women who received ARV therapy more than 6 months after sdNVP [12] suggest that, even if archived, resistant variants may not be clinically significant. Indeed, in a separate study [39], we were unable to show high rates of NNRTI resistance in HIV-1 isolated from resting CD4 T cells of women exposed to sd-NVP. In the current study, selection of maternal and infant resistance mutations detected by population sequencing after sdNVP exposure or re-exposure was not statistically different between the two groups and was similar to that reported previously from Soweto [8] and elsewhere [6,7,40].
A well controlled large multisite cohort of HIV-infected pregnant women, enrolled at the time of their first attendance at the PMTCT program, with long-term follow-up could resolve whether there is a negative effect on transmission of previous exposure to sdNVP. This could resolve the apparently conflicting results our data show: lower transmission rates in the NVP-1 group compared with the NVP-2 group (although this difference could partially be explained by deaths in putatively HIV-infected infants prior to returning for HIV testing) compared with our previous work [21], which found no difference in transmission rates between first and second pregnancies when sdNVP is used in both pregnancies. A large longitudinal study could better compare transmission and HIV-1 resistance rates between first and second pregnancies than this study and would be able to assess whether our results are found in women receiving other NVP-containing PMTCT regimens. Our findings do not contradict the current WHO guidelines [37] for all HIV-infected pregnant women but suggest that HIV-infected pregnant women, who previously received the HIVNET012 regimen, should be offered more effective multi-drug PMTCT regimens.
A prospective cohort of pregnant women was enrolled at 32-38 weeks of gestation from 13 antenatal clinics in Soweto, South Africa where a prevention of mother to child transmission (PMTCT) program had been phased in, using the HIVNET 012 regimen [1,2], since 2001. Two groups of women were recruited: a convenience sample of 120 women, exposed to sdNVP in their previous pregnancy, returning to the PMTCT program a second time (NVP-2) and 240 ARV-naïve, multiparous, HIV-infected women attending the PMTCT program for the first time (NVP-1). For NVP-2 women, source documentation from their previous pregnancy had to be available; namely, a record of a positive HIV test result dating from their prior pregnancy and evidence that sdNVP had been dispensed to them then. Additionally, maternal self-reports both of taking a tablet of NVP at the onset of the previous labour and of not breastfeeding the previous infant were required for NVP-2 women because NVP-2 women were part of another study assessing longitudinal differences in in-utero and intrapartum vertical transmission rates between subsequent pregnancies exposed to sdNVP in the same women [21]. Two NVP-1 women were recruited from the same antenatal clinic within 10 days of each NVP-2 woman recruited. The South African ARV treatment program was initiated in Soweto in April 2004 [22] but, initially, ARV therapy was not available for pregnant women. Women eligible for ARV therapy were referred postpartum with a copy of their genotypic HIV resistance results. The study was approved by the University of the Witwatersrand's Ethics Committee. Resistance testing was performed on archived specimens that the US Centers for Disease Control and Prevention (CDC) Institutional Review Board determined did not involve research on identifiable human participants.
Study visits were at enrollment when women were pregnant and again at 6 weeks postpartum. Maternal CD4 T cell counts (FACScount, Becton Dickinson BioSciences, Immunocytochemistry Systems, San Jose, California, USA), HIV-1 plasma viral loads (Amplicor HIV-1 Monitor Test, v1.5; Roche Molecular Diagnostics, Basel, Switzerland) and a baseline resistance assay was performed on pregnant women and at 6 weeks postpartum when a second viral load and maternal resistance assay was performed. Infants were tested for HIV-1 infection at 6 weeks postpartum and, if HIV-infected, were tested for drug resistance. The presence of HIV-1 DNA in infant plasma samples was determined using the Amplicor HIV-1 DNA Test (Roche Molecular Diagnostics). The ViroSeq HIV-1 Genotyping System (Celera Diagnostics, Rockville Maryland, USA) was used to detect HIV-1 population level resistance mutations following the manufacturer's instructions; viral RNA was extracted from plasma using a MagNA Pure automated system with a MagNA Pure LC RNA isolation kit - high performance (Roche Diagnostics, Indianapolis, Indiana, USA). Sequences were submitted to the Stanford University HIV Drug Resistance Database ( for interpretation.
Samples that failed to amplify or failed to sequence using the ViroSeq assay were retested using an in-house genotyping assay that has been shown to perform well on HIV-1 subtype C samples. Direct comparison between the two assays using 42 plasma samples showed identical data and the in-house assay has been accredited by the South African National Accreditation Society and passed the three most recent Viral Quality Assessment panels. Neither assay was able to amplify 73 samples from either the enrolment or 6-week visit; of these, 60 had viral loads under 2000 copies/ml and were assumed for the purposes of the analysis not to have resistance. Of the 13 individuals with viral load more than 2000 copies/ml but were unable to be amplified, only three were from the 6-week time-point and were not included in the resistance analyses but were included in transmission rate calculations. Sequences with K103N, V106A/M, Y181C, Y188C or G190A are reported as high-level resistance, those with K101E as low-level resistance and A98G, V118I, V179D/E and K103R as polymorphisms. GenBank Accession numbers for reverse transcriptase and Pol sequences from this study are EF381747-EF382346 and EU152409-EU152472, respectively.
Laboratory personnel were blinded to the NVP exposure status of the women and their infants until sequencing and infant DNA PCR results were finalized. Sensitive allele-specific PCR (AS-PCR) testing for the major NVP resistance mutations, K103N and Y181C, was performed on the pre-NVP (enrollment) HIV-1 RNA and proviral DNA as previously described [10] from all transmitting mothers and a random sample of 15% nontransmitting mothers from each NVP exposure group. All women who transmitted HIV-1 to their infants were excluded from the group to be randomly selected and the remaining women who did not transmit HIV-1 to their infants were then stratified into NVP-1 and NVP-2 groups. A sample of 15% of the women in each group was then randomly selected using a uniform distribution. AS-PCR was performed on viral RNA of infected infant samples taken 6 weeks postpartum and was expanded to include the G190A mutation. Viral RNA previously extracted for the genotyping assay was also used for the real-time PCR assay. DNA was extracted from buffy coat samples using a MagNA Pure LC DNA isolation kit I. Sensitive testing of DNA was performed in a similar manner to the RNA testing.
Statistical analysis
The present preliminary study enrolled 240 NVP-1 women and 120 NVP-2 women, with a predicted 90% return rate in both groups. It was able to detect a difference in transmission rate of 11% in the NVP-1 [23] and 25% (i.e. the estimated transmission rate in the absence of any PMTCT intervention) in the NVP-2 women with power of 80% and _ of 0.05. As the PMTCT program had been in operation for only 2 years prior to starting recruitment, we anticipated difficulty in recruiting NVP-2 women and enrolled two NVP-1 women for each NVP-2 woman [24,25]. Rates of HIV-infection in infants at the 6-week visit and maternal and their HIV-infected infants' resistance rates are reported with their 95% confidence intervals (CIs).. To compare rates, the χ2 test (Fisher's exact) was used and crude rate ratios with the NVP-2 rate as the numerator and the NVP-1 rate as the denominator are reported. Separate univariate and multivariate logistic regression models using generalized estimating equations (GEE) were programmed, with the HIV status of the infant and maternal resistance status as the outcomes, respectively.
Pregnant women were recruited from June 2003 through April 2005 and followed up until 6 weeks postpartum. Baseline characteristics of women exposed to sdNVP in a previous pregnancy (NVP-2) and those of ARV-naïve multiparous women (NVP-1) are shown in Table 1. For variables for which data were not always available, the median duration of rupture of membranes was 0.5 h [interquartile range (IQR) = 0.2-2.5] in 65 NVP-2 women and 1.2 h (IQR = 0.2-5.5) in 125 NVP-1 women (P = 0.105); the median time from maternal self-administration of sdNVP to delivery was 5 h (IQR = 2.1-8.0) in 94 NVP-2 women and 6 h (IQR = 2.5-11.75) in 175 NVP-1 women (P = 0.395). Ninety percent and 82.5% of NVP-2 and NVP-1 infants, respectively, had written evidence in the delivery record of receiving their sdNVP (P = 0.06) shortly after delivery. The median viral load of NVP-2 and NVP-1 women at their 6-week-postpartum visit was 21200 (IQR = 3190-64500) and 28300 copies/ml (IQR = 5370-73700), respectively (P = 0.461).
HIV-1 transmission rates
At 6 weeks postpartum, 106 (88%) women and 108 infants from the NVP-2 group and 193 (80%) women and 193 infants from the NVP-1 group returned and had a blood draw; maternal losses to follow-up were 12% and 20%, respectively (P = 0.06). HIV transmission rates at 6 weeks were 11.1% (12/108; 95% CI = 5.9-18.6) in the NVP-2 group and 4.2% (8/193; 95% CI = 1.8-8.0) in the NVP-1 group (P = 0.028); transmission rate ratio 2.7 (95% CI = 1.1-6.4). A further seven mothers and eight infants died prior to their follow-up visit, of which three and two, respectively, were from the NVP-2 group. If all infants who died prior to HIV testing are assumed to be HIV-infected, the transmission rate in the NVP-2 group is 12.7% (95% CI = 6.9-18.0) and in the NVP-1 group 7.0% (95% CI = 3.5-10.5; P = 0.090); transmission rate ratio 1.8 (95% CI = 0.8-4.0). Of four sets of twins, one infant of a NVP-2 mother was HIV-infected.
Logistic regression models - with data grouped by clinic with HIV-infection status of the infant as the dependant variable - suggested that NVP-2 women were 3.2 times more likely than NVP-1 women to transmit HIV-1 to their infants after adjusting for transmission related variables (Table 2).. In univariate analysis, baseline laboratory parameters taken when the women were pregnant were also included; log10 viral load (copies/ml) and CD4 cell count (cells/μl) stratified into three groups. Additionally, parity at enrollment, whether or not delivery was by elective cesarean section, the presence of NNRTI viral resistance detected at the 6-week-postpartum visit, whether or not the mother self-reported ever breastfeeding her child by 6 weeks of age and maternal age in years were also assessed. Of these, we retained in the multivariate model NVP exposure group, log10 viral load and CD4 strata. In another model (data not shown), CD4 cell count divided by 100 was included as a continuous variable in both univariate and multivariate analyses with the outcome infant HIV infection at 6 weeks postpartum. Univariate and multivariate odds ratios (ORs) for each 100 cell/μl increase in CD4 cell count were 1.05 (95% CI = 0.95-1.16) and 1.02 (95% CI = 0.90-1.15), respectively. In that multivariate model, which also included exposure group and log10 viral load, ORs and 95% CIs were identical to those in the final model.
HIV-1 resistance mutations by standard genotyping
Population sequencing of enrollment plasma samples prior to sdNVP exposure showed all pregnant women had wild-type virus except two NVP-2 women and one NVP-1 woman with evidence of K103N and/or Y181C (Table 1). None of these three women transmitted HIV-1 to their infants. All but four mothers were infected with HIV-1 subtype C; three mothers were infected with subtype A and one was unclassified in the pol gene. None of the non-C infected women transmitted HIV to their infants. At 6 weeks postpartum, by standard genotyping, 37.5% (39/104; 95% CI = 28.2-47.5) of NVP-2 women and 46.4% (89/192; 95% CI = 39.1-53.7) of NVP-1 women had high-level resistance (P = 0.119); resistance rate ratio 0.8 (95% CI = 0.5-1.2). Samples of three women were not able to be amplified. NVP-2 mothers with resistance were more likely to have single mutations than NVP-1 women (P = 0.005), but they had similar rates of K103N (P = 0.268; Table 3). Seven women, six of whom were NVP-1 had the K101E mutation (which confers low-level resistance to NVP) at 6 weeks postpartum. Other mutations detected at low frequency such as A98G, V118I, V179D/E and K103R were also present in the baseline sample and were considered naturally occurring polymorphisms.
Half the HIV-infected infants (6/12 and 4/8 in the NVP-2 and NVP-1 groups, respectively) had detectable resistance mutations at their 6-week visit. When compared with their corresponding maternal sequence, one infant had the same mutation pattern, four shared mutations, whereas five had distinct mutation patterns. All sequences from mother-infant pairs clustered together significantly on a phylogenetic tree (data not shown).
A multivariate logistic regression model with the outcome as presence of maternal resistance at the 6-week-postpartum visit (defined by population sequencing), including both enrollment log10 viral load (OR = 2.2, 95% CI = 1.5-3.1) and enrollment CD4 cell count (cells/μl)/10 (OR = 0.98, 95% CI = 0.98-1.0), suggests that prior exposure to sdNVP (compared with no prior exposure to sdNVP) was not significantly associated with maternal resistance (OR = 0.7, 95% CI = 0.4-1.2).
HIV-1 resistance by sensitive allele-specific PCR
To determine whether drug resistance mutations below the level of detection of standard genotyping could be responsible for differing transmission rates in the two groups, we subjected the enrollment samples of all 20 transmitting mothers and those of 40 (15%) randomly selected nontransmitting mothers to sensitive AS-PCR. Prior to sdNVP re-exposure, three of the 12 NVP-2 mothers who transmitted had evidence of minority population mutants in either DNA or RNA by AS-PCR, (Table 4) and one of eight NVP-1 transmitting mothers was at the limit of detection for Y181C in her DNA (P = 0.494). None of the stratified random sample of 40 nontransmitting mothers (13 NVP-2 and 27 NVP-1 women) had RNA or DNA enrollment samples that were positive for K103N or Y181C using AS-PCR. AS-PCR found an additional two infants to have Y181C and/or K103N at 6 weeks postpartum in their RNA. Four infants also had evidence of minority G190A populations that were not detected on genotyping, making a total of 12/20 infected infants with resistance mutations.
  iconpaperstack view older Articles   Back to Top