Women attending antenatal clinics in Lilongwe are tested for HIV infection at their first antenatal visit through a program to reduce perinatal HIV transmission administered by the University of North Carolina (UNC Project), together with the Malawi Ministry of Health. Women found to be HIV-1 infected who have a CD4 cell count ≥ 250 cells/mm3
are offered single-dose nevirapine14
. Women with < 250 cells/mm3
are started on combination antiretroviral therapy (ART) 15
The Breastfeeding, Antiretrovirals, Nutrition (BAN) Study (www.thebanstudy.org
), conducted in Lilongwe by UNC Project and sponsored by the U.S. Centers for Disease Control and Prevention (CDC), is a randomized clinical trial to assess the benefit and safety of antiretroviral medications given to infants or their mothers during breastfeeding (ClinicalTrials.gov Identifier: NCT00164736) 16
. From April, 2004 to September, 2008, HIV-1 infected women who met eligibility criteria were offered participation in the BAN study as an alternative to the standard regimen. All women and infants in the BAN study received single-dose nevirapine at labor and an additional 7 days of twice daily zidovudine+lamivudine. Mother-infant pairs were randomized to receive 28 weeks of maternal or infant daily antiretroviral prophylaxis or no ART during the breastfeeding period.
Women were eligible for the BAN study if they met the following criteria at enrollment: HIV-1 infected, ≤30 weeks gestation, ≥18 years of age (≥14 years of age if married), hemoglobin >7 g/dL, CD4 lymphocyte count ≥ 200 cells/mm3 (later raised to 250 cells/mm3), no prior history of antiretroviral use, normal liver function tests (alanine aminotransferase <2.5 times the upper limit of normal), and no serious complications of pregnancy.
Based on sample size calculations, we selected 132 BAN study participants for the current study from women who were randomized to the arm in which neither mother nor child received antiretrovirals after the 7 days of zidovudine+lamivudine (referred to hereafter as the nevirapine/zidovudine+lamivudine group). The first 132 consecutive women with available specimens were included, and women with HIV-infected infants were over-sampled. The comparison group in the current analysis (the nevirapine group) consisted of 110 consecutively presenting HIV-infected pregnant women receiving single-dose nevirapine through the UNC Project public program at a Lilongwe clinic not participating in the BAN study. Eligibility for the nevirapine group was similar to that in the BAN Study in terms of age, hemoglobin level, CD4 count, history of ARV use and liver function. Study visits relevant for this analysis occurred approximately one week after the first antenatal clinic visit, 2 weeks after the first antenatal visit, at delivery, and at 2 and 6 weeks postpartum. All women gave written informed consent for study participation.
All mothers’ antenatal plasma specimens were tested for HIV viral load, hemoglobin level, CD4 lymphocyte count, and liver function. Drug concentrations were measured at delivery to assess compliance with drug regimens. At 2 and 6 weeks postpartum, all plasma and breastmilk specimens were tested for nevirapine, zidovudine and lamivudine concentrations and resistance mutations using population sequencing and real time PCR. Mutations were classified based on information in the Stanford University Drug Resistance Database (http://hivdb.stanford.edu/index.html
). If any resistance mutation was detected postpartum, an antenatal plasma specimen was tested for preexisting resistance mutations. All plasma from HIV-infected infants was tested for nevirapine, zidovudine and lamivudine resistance mutations using population sequencing and real time PCR.
All blood plasma specimens collected at 2 and 6 weeks postpartum were evaluated for drug resistance mutations through population sequencing and real-time PCR, regardless of viral load level. Population sequencing cannot generally detect mutant virus at levels lower than 20%, while real-time PCR can detect mutant virus at levels as low as 0.5 to 1.0%17
. HIV-1 RNA was quantified from blood plasma using the Roche Amplicor Monitor v1.5 kit (Pleasanton, CA; LLQ = 400 copies/mL). HIV-1 RNA was quantified from 0.6 mL whole breastmilk pre-treated with 209ul Abbott RNA sample prep lysis buffer and 60ul Abbott Proteinase K (53°C incubation for 20 min) using the Abbott RealTime HIV-1 assay (Abbott Molecular, Des Plaines, IL; LLQ = 40 copies/ml). For population sequencing, the reverse transcriptase gene was amplified using an in-house RT-PCR assay from plasma or breastmilk viral RNA derived from Roche or Abbott RNA isolation, or after extraction using the QIAamp Viral RNA Mini Kit (Qiagen, Valencia, CA). Primers for cDNA synthesis and nested PCR were designed from the subtype C consensus (http://www.hiv.lanl.gov
) and the final PCR product corresponded to HXB2 nt 2564–3682. PCR products were sequenced using the ABI PRISM BigDye v1.1 Terminator Cycle Sequencing kit. Each product was sequenced with the second round PCR primers and two internal primers (5’-ATATATGGATGACTTGTATG-3’ and 5’-TTGTCTGGTGTGGTAAATCC-3’). After alignment using Sequencher 4.5, resistance mutations were determined as described above. A few blood plasma RNAs were sequenced using TruGene HIV-1 Genotyping Kit (Siemens Healthcare Diagnostics, Tarrytown, NY). Sample integrity was checked by aligning the sequences using CLC Sequence Viewer (CLC bio A/S) and Treemaker (http://www.hiv.lanl.gov/content/sequence/TREEMAKER/TreeMaker.html
For the sensitive real-time PCR resistance assays, HIV-1 genomic RNA from the blood plasma samples was extracted using Qiagen BioRobot M48 from 200 µL patient plasma. A region of the HIV-1 template that included nucleotide 58 to nucleotide 777 of RT was RT-PCR amplified as previously described 17,18
. This RT-PCR-amplified template was used in the real-time PCR testing for drug resistance mutations. For each mutation, we had established assay cut-offs (ΔCTs) equivalent to 0.5%–1% mutant virus, which are above the background reactivity observed when testing wild type virus, pre-nevirapine rollout samples from South Africa 17
. The RNA samples were tested for five mutations: RT K103N, Y181C, and T215F/Y (ΔCT 10 cycles), V106M and M184V (ΔCT 8.5 cycles). Sensitive testing for codon 215 Y and F mutations was performed to rule out evidence of primary resistance because population sequence-detectible polymorphisms were present at that position. To verify samples positive for drug resistance mutations by real-time PCR, we analyzed a 575 bp region of RT (nt. 133–708 of RT) amplified from the primary RT-PCR of the specimen and also the mutation-positive amplicon derived from the real-time PCR test. All templates underwent double-strand chain-termination sequencing (Big Dye kit v1.1, Applied Biosystems). The real-time PCR amplicon sequences were evaluated for insertions and deletions to verify that the virus sequence was intact. We also examined for the presence of other mutations linked on the mutation-targeted amplicons generated by the real-time PCRs 18
Drug concentrations in blood plasma and breastmilk were measured using validated high performance liquid chromatography (HPLC)/ultraviolet (UV) methods as previously described 19–21
. For blood plasma, the method was validated over the range of 10–10,000 ng/ml for all analytes. Intra- and interday precision was within 5.1% and 5.6%, respectively, and intra- and interday accuracy (average percent deviation from nominal) was between 97% and 105%. For breastmilk, the method was validated over the range of 20 to 20,000 ng/mL. Intra- and interday precision was within 3.6% and 7.5%, respectively, and intra- and interday accuracy was within 99.8% and 98.7%, respectively. Drug exposure was calculated as area under the concentration-time curve (AUC) from delivery to 6 weeks postpartum using non-compartmental methods and the linear/log trapezoidal rule in WinNonlin® Pro version 4.0.1 (Pharsight Corp, Mountain View, CA, USA).
We compared characteristics between treatment groups using Pearson’s chi-square and the Wilcoxon rank sum tests. Cumulative and type-specific nevirapine mutations assessed through population-sequencing and real-time PCR testing at 2 and 6 weeks postpartum in plasma and breastmilk were compared by treatment group. A relative risk and 95% confidence interval for nevirapine resistance comparing women receiving nevirapine/zidovudine+lamivudine to women receiving nevirapine alone was estimated in a modified multivariable Poisson regression model using robust error variances 22
. Drug resistance refers to combined results of population sequencing and/or real-time PCR testing, unless otherwise stated. Unamplifiable specimens and those with viral load levels too low to detect minority variants were excluded from analyses (2 weeks postpartum: n=54 in sdNVP/ZDV+3TC group and n=19 in sdNVP group (p=0.001); 6 weeks postpartum: n=15 in sdNVP/ZDV+3TC group and n=2 in sdNVP group (p=0.05). Statistical analyses were performed using SAS version 9.1.
The current study was approved by the Malawi National Health Science Research Council and the institutional review boards at the University of North Carolina at Chapel Hill and the U.S. Centers for Disease Control and Prevention (ClinicalTrials.gov Identifier NCT00164762). The institutions’ human experimentation guidelines were followed.