Drug resistance among HIV-infected pregnant women receiving antiretrovirals for prophylaxis

Objective: To quantify primary resistance mutations (PRMs) among HIV-1-infected women receiving antiretroviral therapy (ART) for prevention of mother-to-child transmission (MTCT).

Methods: Peripheral blood mononuclear cell samples from HIV-1-infected women enrolled in a prospective cohort study in Argentina, the Bahamas, Brazil, and Mexico (NISDI Perinatal Study) were assayed for PRMs. Eligible women were those enrolled by March 2005 and diagnosed with HIV-1 infection during the current pregnancy, and who received ART for MTCT prophylaxis and were followed for 6-12 weeks postpartum.

Results: Of 819 women, 198 met the eligibility criteria. At enrollment, 98% were asymptomatic, 62% had plasma viral load < 1000 copies/ml, 53% had CD4+ cell count ≥ 500 cells/μl, and 78% were ART-exposed (mean duration, 8.0 weeks; 95% confidence interval, 7.1-8.9). The most complex ART regimen during pregnancy was usually (81%) a three-drug regimen [two nucleoside reverse transcriptase inhibitors (NRTIs) + one protease inhibitor or two NRTIs + one non-nucleoside reverse transcriptase inhibitor). PRMs were observed in samples from 19 (16%) of 118 women that were amplifiable at one or both time points [11/76 (14%) at enrollment; 14/97 (14%) at 6-12 weeks]. The occurrence of PRMs was not associated with clinical, immunological, or virological disease stage at either time point, whether ART-naive versus exposed at enrollment, or the most complex or number of antiretroviral drug regimens received during pregnancy (P > 0.1). Of 55 women with amplifiable samples at both time points, PRMs were detected in 11 samples (20%).

Conclusions: PRMs occurred among 16.1% of relatively healthy HIV-1-infected mothers from Latin American and Caribbean countries receiving MTCT prophylaxis.

Introduction

Both a higher maternal plasma HIV-1 RNA concentration (viral load) and the antiretroviral drug regimen received during pregnancy are independent predictors of the risk of mother-to-child transmission (MTCT) of HIV-1, suggesting the mechanisms of protection of the infant from HIV-1 infection are both through lowering maternal viral load ^[1,2] and through provision of pre-exposure prophylaxis to the infant. Antiretroviral drug prophylaxis of MTCT of HIV-1 has proven efficacy ^[3-5] and effectiveness ^[6]. Consequently, antiretroviral drug prophylaxis has become a routine part of the management of HIV-1-infected women in many countries ^[7,8]. However, antiretroviral drug resistance could limit the clinical effectiveness of antiretroviral drug prophylaxis, during the index pregnancy and during future pregnancies, as well as of future antiretroviral therapy (ART). The primary objective of this analysis was to quantify resistance mutations among pregnant women diagnosed with HIV-1 infection during the current pregnancy, receiving antiretroviral drugs for prevention of MTCT of HIV-1, and enrolled in the National Institute of Child Health and Human Development (NICHD) International Site Development Initiative (NISDI) Perinatal Protocol at multiple sites in Latin America and the Caribbean.

Material and methods

The NISDI Perinatal Protocol is a prospective cohort study conducted in Latin American and Caribbean countries ^[9]. Enrollment began in September 2002, and is ongoing. Maternal study visits are conducted during pregnancy, at labor/delivery, at hospital discharge following delivery, and following delivery (at 6-12 weeks and 6 months postpartum). During each of these study visits, a medical history is obtained, a physical examination is conducted, and laboratory samples are obtained (except at the labor/delivery and the 6-month postpartum visits). Maternal clinical disease staging is performed at each study visit ^[10]. Women were classified as having received antiretroviral prophylaxis if they initiated one or more antiretroviral drugs during pregnancy and discontinued these drugs at or before the 6-12 week postpartum visit. Otherwise, women were classified as having received antiretroviral drugs for treatment. The study population was restricted to HIV-1-infected women enrolled in the study as of 1 March 2005, who were diagnosed with HIV-1 infection during the current pregnancy, who received antiretroviral drug(s) during pregnancy for perinatal transmission prophylaxis, and who were followed up at least until the 6-12-week postpartum visit. Signed informed consent was obtained for all women prior to enrollment into the study. The protocol was approved by the ethical review board at each clinical site where the women were enrolled, as well as by institutional review boards at the sponsoring institution (NICHD) and at the data management center (Westat).

All available peripheral blood mononuclear cell (PBMC) samples collected at baseline and/or at 6-12 weeks postpartum for women who were eligible for inclusion in these analyses were assayed, and the presence of primary resistance mutations (PRMs) was determined. Viral DNA was isolated from PBMC samples using the QIAamp DNA blood kit (Qiagen, Hilden, Germany). A nested polymerase chain reaction (PCR) was performed to obtain an amplified product from the HIV-1 pol gene. Briefly, primers 5′CP1 (gaa ggg cac aca gcc aga aat tgc agg g) and RT3.1 (gct cct act atg ggt tct ttc tct aac tgg) were used for the first round of PCR. Subsequently, 10 μl of the first round PCR was used for a second round PCR with the following primers:

1F: cag acc aga gcc aac agc ccc

6B: cat tgt tta act ttt ggg cc

A35: att ggt tgc act tta aat ttt ccc att agc cct att

NE1: cga cct gac agt tac tgt atg tct tca atc acc

RT3208F: aac atc aga aag aac ctc cat

RT3798R: caa act ccc act cag gaa tcc a

For samples for which no PCR product was obtained, an ethanol precipitation of DNA was performed.

The PCR products obtained covered all the HIV protease codons (codons 1-99) and a portion of the viral reverse transcriptase codons (codons 1-400; 2252-3749 bp HXB2 numbering). All products were sequenced by dideoxynucleoside-based analysis using the Big Dye terminator kit (Applied Biosystems, Foster City, California, USA) and ABI Prism 3100/3100-Avant equipment (Applied Biosystems). The resulting nucleotide pol sequences were analyzed by Sequencher, Version 4.5 (Gene Codes Co., Ann Arbor, Michigan, USA) and the drug resistance profiles were interpreted using the Stanford University HIV Drug Resistance Database (http://hivdb6.stanford.edu).

The identification of PRMs was based on the work of the International AIDS Society-USA (IAS-USA) Drug Resistance Mutations Group ^[11]. The figures published by the IAS-USA are intended for use in identifying mutations associated with viral resistance to antiretroviral drugs and in making therapeutic decisions. Some amino acid substitutions, particularly minor mutations, represent polymorphisms that, in isolation, may not reflect drug-selective pressure or reduced drug susceptibility. Although minor mutations do not, by themselves, have a significant effect on drug susceptibility phenotype, major mutations do. Therefore, mutations designated by the IAS-USA as major were the focus of this investigation. In addition, the combination of mutation E44D with V118I and mutation T69D also were considered PRMs.

Characteristics of the study population are described overall and according to the timing of initiation of antiretroviral drugs (antiretroviral drug-naive at enrollment versus antiretroviral drug-exposed at enrollment). The association of timing of initiation of antiretroviral drugs with categorical variables was evaluated using the Fisher-Freeman-Halton ^[12] exact test. For variables measured on a continuous scale, the t-test was used to assess the statistical significance between means. Genotypic resistance was assessed at baseline and at the 6-12 week visit. The number and percentage of women with PRMs, identified singly or in combination, are reported overall and according to individual characteristics. As the number of PRMs detected was relatively small, no attempt was made to model the risk of PRMs as a function of individual characteristics.

Results

Of the 819 women enrolled in the NISDI Perinatal Protocol at clinical sites in Argentina, the Bahamas, Brazil, and Mexico as of 1 March 2005, a total of 198 women were eligible for inclusion in the analyses (Fig. 1). Among those who were antiretroviral drug-exposed at enrollment, the average duration of antiretroviral drug receipt (from date of initiation of antiretroviral drugs to the date of enrollment into the study) was 8.0 weeks [95% confidence interval (CI), 7.1-8.9 weeks]. Characteristics of the study population are shown in Table 1. In general, women who were antiretroviral drug-naive at enrollment were similar to those who were antiretroviral drug-exposed. As expected, antiretroviral drug-naive women had higher viral loads (53.5 versus 4.7% ≥ 10 000 copies/ml, respectively; P < 0.0001) and lower CD4+ cell counts (58.1 versus 43.6% ≤ 500 cells/μl, respectively; P = 0.02) than antiretroviral drug-exposed women.

Fig. 1

Table 1

DNA amplification could not be accomplished at either time point for 80 women (40%) (Fig. 2). For 71% (57/80) of these women, the plasma viral load at enrollment was below 1000 copies/ml. DNA amplification was unsuccessful with the enrollment specimen for an additional 42 women, bringing the total without amplification of the enrollment specimen to 122 (62% of the total study participants). At the 6-12 week postpartum visit, DNA amplification was not accomplished for an additional 21 women, bringing the total without DNA amplification at this time point to 101 (51%). Of the 76 women for whom resistance testing results were available at enrollment, 11 (14%) were found to have one or more PRMs. Three of these 11 women were antiretroviral drug-naive at enrollment whereas the remaining eight were antiretroviral drug-exposed at enrollment. Of the 97 women for whom DNA amplification was accomplished for the 6-12 week postpartum visit specimen, 14 (14%) were found to have one or more PRMs. Among the 118 women for whom DNA amplification could be performed at one or both visits (60% of total study participants), 19 (16%) were found to have PRMs. Among these women, there were five with PRMs at enrollment only, eight at 6-12 weeks postpartum only, and six with PRMs detected at both time points (Table 2).

Fig. 2

Table 2

There were no associations between maternal clinical classification, CD4+ cell count, and plasma viral load at enrollment or at 6-12 weeks postpartum and the presence of PRMs at 6-12 weeks postpartum. Similarly, there were no associations between maternal antiretroviral drug exposure (antiretroviral drug-naive or not at enrollment, most complex antiretroviral drug regimen during pregnancy, and number of antiretroviral drug regimens during pregnancy) and PRMs detected at the 6-12 week postpartum visit.

Among the 55 women with resistance data from both the enrollment and the 6-12 week postpartum visit, PRMs were detected in 11 samples (20.0%) and were not detected (at either time point) in 44 (80.0%). Six women (6/55; 10.9%) had PRMs identified at both visits, three (5.5%) had PRMs identified at enrollment only and two (3.6%) had PRMs detected only in the 6-12 week postpartum specimen. The K70R mutation was detected in five (9.1%) of enrolment samples and three (5.5%) of 6-12 week postpartum samples.

Discussion

Among a predominantly asymptomatic population of HIV-1-infected mothers from Latin American and Caribbean countries, with low plasma viral loads and relatively high CD4+ cell counts who were receiving antiretroviral drugs for MTCT prophylaxis, PRMs were detected in 19 of 118 samples (16.1%). PRMs were detected in 20% of samples from those women with resistance data from both the enrollment and the 6-12 week postpartum visit. As all of the women were diagnosed with HIV-1 during the index pregnancy, antiretroviral drugs were not initiated prior to pregnancy. Despite the fact that most (78%) had antiretroviral drug exposure prior to their enrollment into the NISDI Perinatal Study, the average length of exposure was only 8 weeks. The detection of PRMs at the enrolment visit suggests either transmitted resistance (three of these women were antiretroviral drug-naive at enrollment) or, less likely, the development of resistance during the relatively short period of exposure to antiretroviral drugs prior to enrollment. Over 80% of women received a three-drug combination of antiretroviral drugs during pregnancy, and the occurrence of PRMs was not associated with antiretroviral drug receipt during pregnancy (most complex antiretroviral drug regimen received during pregnancy, the most complex regimen received during pregnancy for ≥ 28 days, the number of antiretroviral drug regimens received during pregnancy).

The occurrence of PRMs to antiretroviral drugs in the context of receipt of prophylaxis for MTCT of HIV-1 has been described with variable frequency ^[13-15]. However, we did not exclude women from the study population on the basis of their viral loads or CD4+ cell counts, as was done in some studies ^[13,14]. Furthermore, our study population received various antiretroviral drug regimens during pregnancy, in contrast to earlier studies in which all women received combination regimens containing at least two ^[13,15], or three ^[14], antiretroviral drugs during pregnancy.

The overall proportion of PRMs detected in our study is similar to that described by Lyons et al ^[14], in which PRMs were detected in 13% of 39 samples obtained postpartum (none were detected prior to receipt of antiretroviral drugs). However, in this study ^[14], all five women with resistance mutations received nevirapine-containing regimens. In our study, 16% of women who received a nevirapine-containing regimen during pregnancy had PRMs, but none of these PRMs were mutations associated with resistance to nevirapine. Although the K70R mutation has been detected among individuals receiving treatment with zidovudine, the frequency with which this PRM was observed in this population was higher than expected. However, the clinical significance of this PRM is uncertain when it is observed without detection of other mutations such as M41L, D67N, T215Y or K219Q.

One of the strengths of this study is the large size of the study cohort. However, since most of the women received a three-drug antiretroviral regimen, viral loads were relatively low in this population, thus limiting the number of samples that could be amplified. Adherence is not assessed in the NISDI perinatal protocol. Poor adherence may have led to sub-therapeutic levels of antiretroviral drugs, thereby increasing the risk for drug resistance mutations.

The frequency of PRMs detected in this study population may be an underestimate of the true frequency. One reason for this is the time period between discontinuation of antiretroviral drugs after delivery and the 6-12 week postpartum visit; some women discontinued antiretroviral drugs immediately after delivery whereas others continued antiretroviral drugs until the 6-12 week postpartum visit. In addition, genotypic resistance testing has poor sensitivity for the detection of mutations present in less than 25% of the prevailing viral population ^[16]. Conversely, the estimated prevalence of resistance may be an overestimate since those women whose samples could not be amplified may be less likely to have resistance.

Our results indicate the occurrence of resistance while receiving antiretroviral drugs for prophylaxis of MTCT of HIV-1 occurs among women receiving combination antiretroviral drug regimens at a rate similar to that reported in other studies of HIV-1-infected individuals with newly diagnosed HIV-1 infection ^[17,18]. Although resistance mutations were not associated with MTCT of HIV-1, follow up of populations such as this for evaluation of responses to antiretroviral drug therapy in the future will be important. Furthermore, few data exist regarding receipt of antiretroviral drug prophylaxis during pregnancy and subsequent disease progression and responses to antiretroviral drugs utilized for treatment. One study of disease progression among HIV-1-infected women who received antiretroviral drugs during pregnancy suggests resistance is not associated with a poorer virologic response during subsequent HIV-1 treatment ^[19], but more studies are needed to assess whether resistance among HIV-1-infected pregnant women receiving antiretroviral drug prophylaxis have a higher risk of therapeutic failure when antiretroviral drug therapy is initiated subsequent to delivery.

Acknowledgements

Sponsorship: This study was supported by NICHD (Contract #N01-HD-3-3345) and the University of Buenos Aires (Grant # 01/M005).