|Home | About | Journals | Submit | Contact Us | Français|
Among 30 human immunodeficiency virus type 1 (HIV-1)–infected women who received single-dose nevirapine (NVP), 17 (57%) had NVP-resistant HIV-1 detected in breast milk. NVP resistance in breast milk persisted for at least 8 months postpartum and was apparently transmitted to at least 1 infant. NVP resistance was detected less often in women who also received zidovudine.
Participants who donated milk were a subset of a previously described cohort of women and infants in Beira, Mozambique [7, 9]. The study was approved by the institutional review boards of Seattle Children's Hospital (Seattle, WA) and the Mozambique Ministry of Health (Maputo, Mozambique). All women had CD4 counts ≥ 350 cells/μL and World Health Organization (WHO) clinical stage <4, and they were prescribed either single-dose nevirapine (sdNVP; 200 mg given once by mouth) during labor or sdNVP combined with zidovudine (ZDV; 300 mg given by mouth 2 times a day) starting as early as 32 weeks gestation and continued for 1 week after delivery. Participants were excluded if they took >1 dose of NVP, took any antiretroviral (ARV) other than sdNVP or ZDV, or had NVP resistance detected before delivery. Blood and breast milk specimens were collected from women at 2, 4, 6, and 8 weeks and 3, 4, 5, 6, and 8 months postpartum, and specimens were used to assess immunodeficiency virus type 1 (HIV-1) mutations associated with NVP resistance. Infants' blood was tested by HIV-1 DNA polymerase chain reaction (PCR) at birth; at 2, 4, 6, and 8 weeks; and at 18 months of age, or earlier if clinically indicated [7, 10].
Manually expressed breast milk was centrifuged at 1000 x g for 15 minutes, the lipid and skim milk were removed, and the cell pellet was frozen at –70°C . Milk pellet and whole-blood DNA was extracted using the Gentra Puregene Blood Kit (QIAGEN), and the HIV-1 DNA concentration was determined in peripheral blood mononuclear cells (PBMC) and breast milk cells by real-time PCR of subtype C gag using primers (forward, 5′-CAAGCAGCCATGCAAATGTT-3'; reverse, 5'-TGCTATGTCACTTCCCCTTGGTTCTCT-3') and a probe (5'-FAM-AAAGATACCATCAATGAGGAGGCTGCAGAA-TAMRA-3') and conditions as previously described [9, 11]. Four HIV-1 pol mutations that confer NVP resistance (K103N, V106M, Y181C, and G190A) in subtype C were evaluated with an oligonucleotide ligation assay (OLA; http://depts.washington.edu/idimmweb/faculty/frenkel/OLAmanual1305april04.pdf), which can detect as few as 2%–5% of each mutant in the individual's HIV-1 population [7, 9].
Proportions were compared using a 2-sided Fisher's exact test. Odds ratios (ORs) for detecting NVP resistance in samples were calculated with adjustment for within-participant correlation by logistic regression with robust standard errors. Analyses were performed using Stata/SE 9.2 for Macintosh (Stata Corporation).
Of the 64 eligible pregnant women who received sdNVP, 30 (46.9%) had breast milk specimens of sufficient volume and HIV-1 copy numbers to be analyzed; 27 also had PBMC available for testing. Nine (30.0%) of the women took ZDV for a median of 33 days (range, 13–79). A total of 96 unique breast milk cell specimens, representing 70 time points at which left and/or right breast milk was available, and 91 PBMC specimens were analyzed (Table 1). The median number of HIV-1 genomes in specimens assayed for NVP resistance by OLA was 130.0 for PBMC (interquartile range, 101.2–312.0) and 59.8 for breast milk cells (interquartile range, 21.5–127.3).
Seventeen (56.7%) of the 30 participants had NVP resistance detected in at least 1 breast milk specimen, and 15 of 27 (55.6%) had any NVP resistance detected in PBMC (Table 1). Four participants (77, 90, 99, and 134) did not have a blood sample available for resistance testing before delivery; NVP-resistant HIV-1 was detected during follow-up in the breast milk of 2 of these women at least once, and in the PBMC of 3. Of 23 participants with breast milk and PBMC specimens collected at the same time, 6 (26.1%) had discordant results, ie, NVP resistance was detected in either breast milk (n = 3) or PBMC (n = 3), but not both (Table 1). There was a trend toward less frequent detection of NVP resistance in breast milk of participants who took ZDV in addition to sdNVP compared with those who received sdNVP only (3 of 9 vs 14 of 21 subjects; P = .123). On a unique specimen basis, the odds of NVP resistance were 4-fold greater in breast milk from participants who took sdNVP alone compared with ZDV plus sdNVP, after adjustment for within-individual correlation (OR, 4.05; 95% confidence interval [CI], 0.85 and 19.2; P = .078). In this substudy, as in the larger cohort from which these subjects were taken , NVP resistance was detected in PBMC more often in participants who took sdNVP alone than in those who took ZDV plus sdNVP (15 of 18 vs 3 of 9 participants, respectively; P = .026), as well as in unique PBMC specimens from those participants (OR, 6.25; 95% CI, 1.33 and 33.3; P = .020). None of the milk or PBMC specimens tested was 100% NVP-resistant mutants; ie, all HIV-1 populations were mixes of wild-type and mutant HIV-1.
Breast milk specimens were available through a median of 8 weeks postpartum (range, 2–32; Table 1). Of the 17 participants with NVP resistance detected in breast milk, 1 (5.9%; participant number 77) reverted to 100% wild-type virus at 4 weeks postpartum; the other 16 had mutant virus detected in the last specimen available (median 8 weeks; range, 4 weeks–8 months) for follow-up testing. Among the 4 participants with NVP-resistant HIV-1 detected in breast milk for whom a follow-up specimen was available at 3 months postpartum, none reverted to wild type, and both of the participants with breast milk samples tested at 8 months had detectable NVP-resistant HIV-1.
All 30 infants born to women in this cohort had 2 blood samples, taken from birth to 8 weeks of age, tested by HIV-1 DNA PCR. Eight (26.7%; 95% CI, 12.9% and 46.2%) infants acquired HIV-1 infection; 6 of these 8 (infant numbers 1, 57, 82, 134, 233, and 292) had a first positive test at birth or 2 weeks of life, consistent with in utero and peripartum transmission, respectively. Infant number 2 tested HIV-1 DNA negative at 8 weeks and positive at 8 months age, indicative of breast milk transmission. This infant had wild-type virus, although the G190A mutation was detected in the mother's breast milk at 4 weeks after delivery. Infant number 164 tested negative by HIV-1 DNA PCR at 0, 2, and 4 weeks of life but positive at 6 weeks of age, at which time 100% of the infant's virus had the G190A mutation, consistent with of breast milk transmission. The infant's mother had the G190A mutation detected in both breast milk and PBMC. Among the 22 infants who tested negative by HIV-1 DNA PCR using the last available specimen, 15 (68.2%) were last tested at 6–18 months of age; the remaining 7 (31.8%) infants were lost to follow-up and thus were not tested for HIV-1 after 8 weeks of age.
The findings of this study underscore the risks of drug-resistant HIV-1 associated with sdNVP to both the mother and to her child for the duration of breastfeeding, and the results suggest a benefit of ZDV use. First, NVP resistance mutations were detected in most mothers during the early months postpartum when most infants breastfeed, as previously reported [4, 5]. Second, NVP resistance was usually detectable through the last time point tested, supporting the prolonged persistence of NVP resistance mutants in maternal milk . Third, as others have observed , NVP-resistant mutants appeared to be transmitted to breastfed infants, which may undermine the efficacy of infant NVP prophylaxis. Fourth, although not definitive, the addition of ZDV to sdNVP appeared to reduced the risk of NVP resistance in breast milk, as we have shown in blood using the parent cohort , which is consistent with the higher genetic barrier to resistance with ZDV plus sdNVP compared with sdNVP alone.
Given the low concentration of HIV-1 in breast milk samples and that a smaller number of viral genomes were available for OLA, it is likely that the true frequency of NVP resistance mutations existing at low concentrations in breast milk was underestimated. It is unlikely that assaying only breast milk cells, rather than whole milk, underestimated the amount of NVP resistance, given that HIV-1 populations in breast milk cells and skim milk are similar , and NVP-resistant mutants after sdNVP persist longer in blood cells compared with plasma . The small sample size and limited follow-up preclude the precise determination of the temporal dynamics of NVP resistance in breast milk HIV-1. The relatively high rate of MTCT we observed likely reflects the small size of the cohort and the selection of women with relatively high levels of HIV DNA in breast milk.
To prevent MTCT, the WHO currently recommends that women receive ARV treatment, or ARV prophylaxis with ZDV, to which maternal sdNVP may be added. If the mother receives sdNVP, during the first week postpartum ZDV and lamivudine are recommended to reduce the selection of NVP-resistant viruses. At this time, a substantial number of resource-poor communities continue to use sdNVP, with or without ZDV. This study provides additional rationale to avoid sdNVP for prevention of MTCT, or if sdNVP is used, that it be combined with ZDV to minimize the risks of selecting NVP resistance in women and the risk of transmitting NVP-resistant HIV-1 to breastfeeding infants.
We thank the women and children who participated in this study, the Mozambican Ministry of Health staff, the study team, and the laboratory staff for valuable contributions to this work; and Dr Joseph Fitzgibbon at the National Institutes of Health for his guidance.
Financial support. This work was supported by National Institutes of Health Grants R01-AI058723 (to L. M. F.), T32-HD07233 (to S. G.), KL2-RR025015 (to S. G.), and T32-AI07140 (to M. A. M.).
Disclaimer. The funding agencies were not involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Potential conflicts of interest. All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.