Persistence of nevirapine-resistant HIV-1 in women after single-dose nevirapine therapy for prevention of maternal-to-fetal HIV-1 transmission

NVP Resistance After sdNVP.

All of the women studied were infected with subtype C HIV-1, as determined by standard genotype analysis. The women were categorized into three groups on the basis of detection of NVP-resistance mutations by standard genotyping (Tables 1–3, Fig. 1). Group 1 (n = 6) had NVP-resistance mutations detected at 2 mo and 6 mo, but not at 12 mo, after sdNVP. Group 2 (n = 9) had NVP resistance mutations detected at 2 mo, but not at 6 mo. Group 3 (n = 7) had no resistance mutations detected at 2 mo after sdNVP, which was the only sampling time point.

The percentage of NVP-resistant variants K103N (AAA to AAT or AAC) and Y181C (TAT to TGT) was determined by allele-specific RT-PCR. Predose baseline samples had negligible levels of mutations (≤0.1%) at both codons, similar to the assay background level. Results from analysis of all patient samples are shown in Fig. 1 and Tables 1–3.

103N Mutants.

Consistent with standard genotype results, all 2-mo samples from groups 1 and 2 had 103N mutations detectable by allele-specific RT-PCR, with the AAC codon ranging from ≈3.2% to 55.8% of the virus population, and the less-frequent AAT codon ranging from background to 38.2% (Tables 1 and 2). In group 1, samples from 6 mo after sdNVP showed a decrease in the mutant populations at codon 103 (AAC) to ≈12% of the virus population. At 12 mo after sdNVP, five of six women still had detectable levels of AAC mutation (frequencies of 1.4–15.9%), but the AAT mutation had declined to undetectable levels in three of these women. In group 2, most women (six of eight) had 103N AAC variants detected 6 mo after sdNVP (2.0–10.1% of the virus population), despite negative standard genotypes. AAT mutations were detected in only one of nine women, at a frequency of 0.7%. At 12 mo after sdNVP, only three of nine women still had detectable levels of AAC mutation, with frequencies ranging from 0.7% to 21.6%. In group 3, three of seven women had the AAC mutation detected 2 mo after sdNVP (0.2–15.3% of the virus population), despite negative standard genotypes.

181C Mutants.

Analysis of codon 181 revealed that, 2 mo after sdNVP, five of six patients in group 1 and eight of nine in group 2 had minor populations of Y181C ranging from 1.1% to 19.6% of the virus populations. This mutant population was detected in four samples from group 1 and six samples from group 2 by standard genotype. In group 3, most women (five of six) had detectable levels of 181C at 2 mo, ranging from 0.3% to 2.9%, despite negative standard genotypes. At 6 mo, 181C was still detectable in two of six group 1 women (with a frequency of 0.4%) and two of nine group 2 women with frequencies of 0.3% and 0.7%, respectively. At 12 mo, the Y181C mutant population had declined to below the limit of detection in all patients from group 1 and in all but two patients from group 2. Overall, more than half of the patients in groups 1 and 2 (9 of 15) had 103N and/or 181C mutations 1 yr after sdNVP.

Average Frequency of 103N and 181C Mutations Over Time.

To better understand the variability in selection and decay of different mutants among the women studied, we plotted the average frequency of 103N and 181C variants over time (Fig. 2). The average frequency of the 103N AAC mutation in group 1 patient samples increased to 28% 2 mo after therapy and then decreased to 5% by 12 mo (Fig. 2A). Patient samples in group 2 showed a similar pattern of decline at somewhat lower average levels, with the average AAC mutant population increasing to ≈22% 2 mo after treatment and then declining over the next 10 mo. Importantly, 1 yr after treatment, the 103N AAC mutation persisted in these two patient groups at an average of 3–5% of the viral population. Two months after treatment, patient samples in group 3 had an average AAC mutation frequency of 3% (later samples were not available for analysis owing to the design of the parent study).

The average frequency of the AAT 103N mutation followed a similar pattern (Fig. 2B). Two months after sdNVP therapy, the average frequency of this mutation increased to 18% in group 1 patient samples and to 4% in group 2, declining to 1% and 0.2% respectively, at 1 yr. Even though the frequency of this mutation was low relative to AAC, the AAT mutation in both groups persisted above the baseline frequency of 0.003% for more than a year after treatment. The AAT mutation was not detected in group 3 patient samples at 2 mo.

The average frequency of the 181C TGT mutation in group 1 patient samples increased to 4% 2 mo after sdNVP and then decreased over the next 10 mo to 0.1% (Fig. 2C). Group 2 showed a similar pattern; however, the average frequency of the 181C mutation was 2-fold greater (8%) at 2 mo after treatment, and then declined to 0.4%. In samples from women in group 3 at 2 mo after sdNVP treatment, the average frequency of the 181C mutation was 1%.

Estimated Percentage of Women with NVP-Resistant Variants.

From our data, we estimated the proportion of women in the South African MTCT study who would be expected to have NVP-resistant variants 2, 6, and 12 mo after sdNVP. For this purpose, we multiplied the proportion of women represented in each group by the percentage of patients with mutations in each group at different time points and extrapolated the results to all 335 women enrolled in the study (Table 4). Using this calculation, we estimate that 83%, ≥34%, and ≥23% of the women in the trials would have NVP-resistant variants at 2, 6, and 12 mo, respectively.

Clinical Specimens.

Plasma samples were collected from 22 women participating in a prospective MTCT cohort at Chris Hani Baragwanath Hospital, Soweto, South Africa. Sample collection was approved by the Human Research Ethics Committee (Medical) of the University of the Witwatersrand, and all participants provided written informed consent for viral resistance testing. Baseline and follow-up plasma samples were selected randomly and analyzed from 22 of 335 women who had received sdNVP in the MTCT study. In the present study, all women had a blood draw scheduled at 6 wk postpartum. Thereafter, blood draws were performed only if the first sample was found to have resistant mutations by standard genotyping. Plasma samples were collected from the 22 women at baseline before exposure to sdNVP, and at a median of 2 mo (from 1.5 to 3.5 mo; n = 22), 6 mo (from 5 to 9.3 mo; n = 14), and 12 mo (from 11.7 to 14.1 mo; n = 15) thereafter.

Standard Genotype.

Standard genotype analysis was performed on all specimens at an accredited laboratory in South Africa, employing the HIV-1 ViroSeq genotyping system (Abbott/Celera, Abbott Park, IL), and data were analyzed by using both ViroSeq software and the Stanford database. One standard genotype was obtained for each time point, and sequence discrepancies were settled by reviewing the specific sequence chromatogram.

Viral RNA Extraction and Amplification.

RNA was extracted from virus pelleted from 250 to 1,000 μl of plasma (containing a minimum of 10,000 copies of HIV-1 RNA) as described (22, 29). First-round, real-time RT-PCR amplification of a 637-bp fragment of pol including reverse transcriptase codons 22–234 was performed by using HIV-1 subtype C-specific primers 1946F 5′-AAACAATGGCCATTGACAGAAGA-3′ and 2583R 5′-GTTCATACCCCATCCAAAGAAATG-3′, with PCR product monitored by SYBR green fluorescence as described (22). All primers used in the assay were subtype C-specific and specially designed to amplify subtype C virus found in South Africa; other primer sets may be required for non-C subtypes. The patient samples were run in triplicate. No-template controls and HIV RNA standards (3 × 10⁶ to 300 copies) were included on the same 96-well plate. To confirm amplification specificity, PCR products were subjected to thermal denaturation analysis of 70–80°C, with readings taken every 0.1°C for 3 sec. Samples that did not yield the correct melting temperature (T_m) for the amplified product were removed from the analysis.

Allele-Specific RT-PCR.

The products of the first amplification were diluted to ≈10⁷ total copies per reaction mixture, and a second round of PCR was performed with nondiscriminating primers (2090F 5′-AAG TGG AGA AAA TTA GTA GAT TTC AGG GA-3′ and 2222R 5′-ATC CCC CAC ATC TAG TAC TGT CAC TG-3′ for codon 103 or 2298F 5′-CAC CAG GGA TTA GAT ATC AAT ATA ATG TG-3′ and 2450R 5′-CTA CAT ACA AGT CAT CCA TAT ATT GA-3′ for codon 181), as well as with primers that amplified specific alleles (AAC 103N: 2218R 5′-CCC ACA TCT AGT ACT GTC ACT GAT TGG-3′ and AAT 103N: 5′-CCC ACA TCT AGT ACT GTC ACT GAT TCA-3′ or TGT 181C: 2450R 5′-CTA CAT ACA AGT CAT CCA TAT ATT GCC-3′) or wild-type sequences (AAA 103K: 2218R 5′-CCC ACA TCT AGT ACT GTC ACT GAT TCT-3′ or TAT 181Y: 2450R 5′-CTA CAT ACA AGT CAT CCA TAT ATT GCT-3′), by using SYBR green fluorescence to detect the amplified product, as described in ref. 22. To minimize the contribution of nonspecific amplification to the fluorescent signal, fluorescence was measured at a temperature just below the T_m of the correct product but above the T_m of smaller, nonspecific products. To confirm amplification specificity, the PCR products were subjected to a melting curve analysis as described above. All analyses were done in triplicate, and results are reported as percent mutant or wild type, which was determined by dividing the mutant or wild-type population detected by the total population detected. The values for mutant and wild-type populations for each sample were normalized to 100%. Results with a coefficient of variation 100% were repeated.