Pregnancy Rates and Birth Outcomes Among Women on Efavirenz-Containing Highly Active Antiretroviral Therapy in Botswana

Abstract

Background: 

Millions of HIV-infected women in developing countries are in need of safe and highly effective antiretroviral therapy. Pregnancy rates are usually high in developing countries, and efavirenz (EFV) use in women of childbearing age is of concern because of its potential teratogenicity.

Methods: 

As part of a prospective study comparing 6 initial highly active antiretroviral therapy (HAART) regimens, 3 of which contained EFV, pregnancy and birth outcomes were evaluated among female participants enrolled in a randomized clinical trial in Botswana. Before enrollment, all female participants indicated a willingness to avoid pregnancy for the 3-year duration of the study. Monthly urine pregnancy testing and regular contraceptive education and counseling were given to all women on study.

Results: 

Four hundred fifty-one (69.4%) of 650 enrolled study participants were female and experienced 71 pregnancies, for a rate of 7.9 per 100 person-years during the study. The mean time from HAART initiation to time of first pregnancy was 385 days. The median birth weight of babies was 2950 g (interquartile range: 2700-3250 g); the gender of babies (24 female and 15 male) and occurrence of early pregnancy loss (42%) and stillbirths (3%) did not differ between EFV- and non-EFV-exposed pregnancies (P = 0.7). First-trimester EFV exposure occurred in 38 (53.5%) of the 71 pregnancies; 22 (57.9%) of these 38 pregnancies resulted in live births. One infant (4.5%) of the 22 EFV-exposed live births had a congenital abnormality with right limb shortening that was assessed to be unrelated to EFV exposure.

Conclusions: 

The restoration of health and longevity in many HAART-treated women is often accompanied by childbearing, as evidenced by the large fraction of women in our cohort who became pregnant despite their initial statements of intent to avoid pregnancy. Of 22 first-trimester EFV-exposed live births, 1 neonate was found to have a major congenital abnormality; however, this defect was unrelated to EFV exposure. The small sample size is insufficient to estimate accurately the underlying risk of congenital malformation after exposure to EFV in early pregnancy, underscoring the importance of reporting to the existing international Antiretroviral Pregnancy Registry. In addition to accessing safe and effective HAART regimens, HIV-infected women require access to comprehensive family planning services, including contraception and procreation counseling.

Millions of HIV-infected women in developing countries are in need of safe and highly effective antiretroviral therapy (HAART). In sub-Saharan Africa, nonnucleoside reverse transcriptase inhibitor (NNRTI)-based HAART regimens are the preferred first-line treatment, largely because of factors such as cost, the potential for drug-drug interactions, and pill burden considerations. When used with 2 nucleoside reverse transcriptase inhibitors (NRTIs) as initial treatment, the NNRTIs efavirenz (EFV) and nevirapine (NVP) seem to be equipotent in terms of virologic activity.¹ Although EFV does possess a more favorable side effect profile and a lower pill burden than NVP, these advantages need to be carefully weighed against its potential teratogenicity, especially in resource-poor settings with limited contraceptive and family planning services.

Rates of congenital malformations after HAART-exposed pregnancies seem to be comparable to rates of congenital abnormalities that have been reported in the general (HIV-1-negative) population. In the latest update, the international Antiretroviral Pregnancy Registry has reported a rate of 2.6% (95% confidence interval [CI]: 2.1% to 3.0%) birth defects per 100 live births among pregnancies exposed to HAART at any time during pregnancy (132 birth defects in 5169 HAART-exposed pregnancies). Similar prevalence rates were observed in 2 large European studies.^2,3 Also, no excess of congenital abnormality has been reported after EFV exposure in early pregnancies. Among 223 early EFV-exposed pregnancies, the Antiretroviral Pregnancy Registry has observed only 5 birth defects, none of which was a neural tube defect (2.2%, 95% CI: 0.7% to 5.1%).⁴ Four retrospective human case reports of central nervous system malformations in neonates born to women exposed to EFV^5,6 have recently led the manufacturer to modify the “pregnancy risk” warning label on EFV from category C (“risk cannot be ruled out”) to category D (“positive evidence of risk”), however.⁷ The earlier category C pregnancy warning for EFV was based primarily on preclinical primate studies.⁸

Consensus antiretroviral (ARV) treatment guidelines vary somewhat in their NNRTI selection for first-line HAART. Botswana's national policy since the beginning of the public national ARV program in January 2002 has been to place all women with reproductive potential on zidovudine (AZT) and lamivudine (3TC) with NVP, reserving EFV combined with the 2 NRTIs for men and and for women without reproductive potential.⁹ Other national ARV guidelines such as those of Tanzania¹⁰ recommend stavudine (d4T) and 3TC plus NVP as first-line HAART for all qualifying HIV-infected adults.

In Botswana, within the context of a large randomized clinical trial, we have intensive longitudinal follow-up from 650 HAART-treated adults, 69.4% of whom were female. All women enrolled in the study received extensive education and counseling, pregnancy screening, and access to family planning services. We report our preliminary data on the pregnancy and birth outcomes of all female study participants, 38 of whom were briefly exposed to EFV during their first trimester of pregnancy.

METHODS

Study Population

The Adult Antiretroviral Treatment and Drug Resistance Study (also known as the “Tshepo Study”) is a randomized clinical trial that compares the efficacy, tolerability, development of drug resistance, and adherence rates of 6 different NNRTI-based first-line HAART regimens in urban Botswana. Because 3 of the 6 initial HAART regimens under study included EFV, all female participants were counseled extensively regarding the potential harmful effects of EFV on pregnancy; before study enrollment, all indicated a willingness not to become pregnant while on study. All female participants also received counseling before study enrollment and throughout study participation regarding the importance of dual protection by using condoms and a nonbarrier contraceptive method. During each monthly study visit, reproductive health issues were raised, including the following: the status of their current sexual relationships (if applicable), contraceptive use, and menstrual history. In addition, in accordance with the study protocol, all women with reproductive potential had rapid urine pregnancy tests (Pregwise OneStep hCG [Pregnancy] Rapidip Strip; Pharma Chem Pharmaceuticals, Pretoria, Republic of South Africa) performed during each scheduled monthly study visit and when clinically indicated. Participants in need of nonbarrier contraceptive methods (depomedroxyhydroprogesterone [DMPA] injections, oral contraceptive pills, and intrauterine contraceptive devices) were all initially referred to easily accessible local government health clinics, where contraceptive services are provided free of charge.¹¹ Beginning in May 2005, family planning services became available on-site at the study clinic. Ascertainment of congenital abnormalities was determined from examination of hospital neonatal records and confirmed by a specialist pediatrician when infants were examined at least once by 3 months of age. No morphologic assessments of cases involving fetal loss were available.

This study was approved by the Health Research and Development Committee of the Botswana Ministry of Health and the Harvard School of Public Health Human Subjects Committee.

Statistical Methods

Descriptive univariate analyses included means and medians for normally distributed continuous data, percentages for categoric data, and quartiles for ordinal or nonnormal continuous data. To evaluate for potential differences in continuous outcomes by EFV exposure, analyses were performed with the Statistical Product and Service Solutions software, version 14.0 (SPSS, Chicago, IL) using the Kruskal-Wallis 1-way ANOVA test. The χ² test was used to compare differences between the groups for categoric and ordinal outcomes. P values <0.05 were considered to be statistically significant.

RESULTS

Contraceptive Uptake

According to data collected during routine study visits on case report forms, effective and consistent nonbarrier contraceptive methods were used by 80 (19%) of our female study participants. Of these 80 female study participants consistently using nonbarrier contraceptives, 41 (51.3%) used oral contraceptive pills, 36 (45%) received DMPA injections every 3 months, and 3 (3.7%) opted for an intrauterine contraceptive device (IUCD).

Pregnancy and Birth Outcomes

Four hundred fifty-one female participants and 199 male participants were enrolled in the clinical trial between December 22, 2002, and December 17, 2004. As of January 1, 2006, 71 pregnancies had occurred during 900 person-years of follow-up among women with reproductive potential, corresponding to an overall pregnancy rate of 7.9 per 100 person-years. The mean time from HAART initiation to time of first pregnancy was 385 days (interquartile range [IQR]: 232-562 days). Four women became pregnant twice. Four urine pregnancy test results were interpreted as false positive and were therefore excluded from the analysis. The median CD4⁺ count value at the time of pregnancy diagnosis was 348 cells/mm³ (IQR: 244-439 cells/mm³), reflecting a median gain of 206 cells/mm³ from the time of HAART initiation (Table 1). All infant polymerase chain reaction (PCR) results obtained up to the time of this report were negative for HIV-1 infection.

TABLE 1: Tshepo Study Participants Becoming Pregnant During Observation Period*

As of October 1, 2006, all 71 pregnancies have been completed and resulted in 39 live births and 32 pregnancy losses. The 39 pregnancies resulting in live births (24 female and 15 male babies) were all singleton, with a median birth weight of 2950 g (IQR: 2700-3250 g). Four (10.3%) of the 39 live births were delivered via cesarean section. Of the pregnancy losses, 26 (81.3%) were attributable to abortions of unknown cause, 4 (12.5%) were attributed to elective abortions (2 performed outside the country and 2 performed within the country for maternal medical indications), and 2 (6%) of the pregnancies ended in stillbirths.

Among 38 EFV-exposed pregnancies (all during the first trimester; Table 2), 22 (57.9%) resulted in live births and 16 (42.1%) resulted in abortions (2 elective and 14 of unknown cause). The median time of EFV exposure was 43 days (IQR: 31-60 days). All women were switched to non-EFV-containing HAART the same day that pregnancy was diagnosed and were given the option to switch back to EFV after completion of the pregnancy, provided that adequate contraception was practiced. Two (11.8%) minor congenital abnormalities, 1 infant with polydactyly and 1 with an umbilical hernia, were detected among 17 liveborn non-EFV-exposed infants. One (4.5%) major congenital anomaly was detected among 22 liveborn EFV-exposed infants. This infant has congenital shortening of the right lower extremity; radiographic imaging studies documented a shortened right femur with midshaft cortical thickening, absence of right femur head epiphysis, and dysplastic right fibular head.

TABLE 2: Pregnancy Outcome by HAART Exposure (N = 71)

DISCUSSION

Our study is one of the first to document a pregnancy rate among HAART-treated women in sub-Saharan Africa.¹² Our observed pregnancy rate of 7.9 per 100 person-years is similar to rates reported among HIV-infected women in the United States¹³ and Western Europe¹⁴ but lower than the pregnancy rate among the general Botswana population, which has recently been estimated to be approximately 11%.^15,16 Notably, at the time of study enrollment (eg, HAART initiation), all our female participants verbally agreed to defer childbearing for the duration of the 3-year study. As per study protocol, half of the women were randomized to EFV-based HAART. Because these participants received extensive counseling and education regarding the potential teratogenicity of EFV, even higher pregnancy rates might occur among HAART-treated women cared for in a nonresearch setting.

There are likely several reasons for the relatively high pregnancy rate, which was independent of the NNRTI being taken (NVP or EFV), in our HAART-treated cohort. First, childbearing is often perceived as a visible and unmistakable expression of health. Although reports from developed countries indicate that pregnancy rates are lower in seropositive women when compared with their seronegative counterparts, HIV-infected women with high CD4⁺ cell counts not yet on HAART and those with robust immune reconstitution after HAART initiation^17,18 are more likely to become pregnant.¹⁹ Second, the advent of HAART has also drastically improved the quality and quantity of life for HIV-infected individuals.^20,21 It is reasonable to anticipate that increasing access to public HAART is likely to influence the reproductive attitudes and behavior of considerable numbers of HIV-infected women in the region. Third, and most significantly, despite extensive and consistent family planning and risk reduction counseling, uptake of dual contraceptive methods by female participants in our study was limited. Nationally, Botswana has a long-established and easily accessible national family planning service that is available at local government clinics.¹¹ Recently, however, in the context of widespread acceptance and use of barrier contraceptives (predominantly male condoms), the reported uptake of nonbarrier methods has dropped substantially.²² Contraceptive choice data from our HAART-treated cohort data are consistent with these unfortunate trends, in which study participants report high male condom use (∼100%), with few women reporting use of nonbarrier contraceptive methods, even in the setting of our clinical trial, where preconception counseling and pregnancy surveillance were stringent. Based on anecdotal reports from our study participants, the contraceptive advice that HIV-infected individuals often receive at family planning clinics is brief and often limited to advising them to abstain or use condoms. In response to these concerns and to our higher than expected pregnancy rates, our study team also made routine family planning services available to all study participants at the study clinic site beginning in May 2005.

One other issue that has surfaced during discussions with our study team is that women may feel more comfortable using condoms alone, fearing that concomitant use of a nonbarrier method may compromise their ability to negotiate the use of condoms. This is especially important in a setting such as ours, where condom use may only be viewed as being necessary to prevent HIV transmission to the index partner rather than being seen in a larger context (ie, family planning and reproductive counseling). Additional research is certainly planned to look more in depth at the individual social and behavioral factors influencing family planning and reproductive counseling decisions. These studies are to include the opinions and attitudes of men and women in an attempt to identify all potential barriers limiting the use of more efficient contraceptive methods.

To date, data from HAART-exposed women from Western Europe²³ and the United States⁴ have consistently documented congenital abnormality rates ranging from 2% to 4%, which are similar to those reported in the general (non-HIV-1-infected) population.⁴ Among the reported HAART regimens, however, only studies evaluating ARV drugs commonly used in pregnancy (AZT, 3TC, d4T, NVP, and nelfinavir) have been sufficiently powered to detect a doubling of the rate of overall birth defects.

Data on pregnancy outcome after EFV exposure are scarce, because EFV-containing HAART is generally avoided in pregnancy as a result of its potential teratogenic risk. The few prospective studies³ (including ours) have an insufficient sample size to estimate accurately the underlying risk of congenital malformation after exposure to EFV in early pregnancy, a fact that underscores the importance of prospective reporting to a registry, such as the existing Antiretroviral Pregnancy Registry.

Three neonates with meningomyeloceles and 1 case of a Dandy-Walker malformation in an aborted fetus have been documented after EFV exposure in early pregnancy. Within our HAART-treated cohort, the 1 infant with a major congenital abnormality (shortening of the right limb/femur) was exposed to EFV for 31 days, starting from the first day of the mother's last menstrual period. EFV was discontinued before the critical period of limb development;^23,24 thus, it is unlikely that this defect was related to EFV exposure. Our study team circulated the details of this case to international experts who helped to formulate “ARV medication relatedness” conclusions. Unfortunately, neither country-specific nor regional data on the background frequency of similar anomalies are available.

Our observed rate of 26 (36.6%) unexplained early pregnancy losses from 71 total pregnancies is certainly higher than the rate one would expect from spontaneous abortions alone.²⁵ All early pregnancy losses in our study occurred within the first 3 months after the date of the last reported menstrual period. These early pregnancy losses may be attributable to multiple factors. First, underlying congenital malformations may have resulted in spontaneous abortion. Unfortunately, the lack of medical information on these early pregnancy losses precludes definitive conclusions regarding such a cause. Second, these early losses may be attributable to HAART alone; yet, we are unaware of any data linking individual ARV medications or classes of ARV medications to early pregnancy loss. Reports on increased rates of premature deliveries among HAART-treated women are inconsistent^26-29 and also would involve pathophysiologic mechanisms different from those leading to abortions. Third, although we cannot fully exclude that our high abortion rate is causally related to ARV treatment, we can note that a significant proportion of unexplained miscarriages in our study are most likely attributable to elective abortions. Based on feedback that research nurses obtained, a substantial proportion of female participants stated that their pregnancies were unwanted. This information was obtained at the time when their urine pregnancy test result first became positive-when the study nurses were reviewing the positive rapid urine pregnancy test results with them. Current Botswana statutes do not allow for the termination of pregnancy except in special circumstances, including cases posing a significant health risk to the mother and/or fetus with prior documentation by 2 medical physicians.^30,31 Of note, maternal HIV infection is not generally regarded as a basis to meet these requirements, and women may have obtained abortions outside of the existing government health care system to avoid social and legal consequences of elective pregnancy termination. In a larger context, however, we also realize that child wish is a strong desire for many women with reproductive potential and certainly may have been present but not verbally expressed to our study staff. A fraction of these pregnancies occurred among HAART-treated female study participants who had experienced an excellent clinical response and impressive CD4⁺ cell count gain and were eager to begin families and solidify existing relationships. The recently reported high incidence of induced abortion in Uganda³² further underscores the need to conduct more detailed behavioral studies to understand better the factors influencing female contraceptive and reproductive choices.

The manufacturer's warning that treatment with EFV should be avoided in the first trimester of pregnancy because of the risk of teratogenicity⁷ has led to the recommendation to avoid EFV use in women of reproductive potential unless effective contraception can be ensured.³³ With the rapid expansion of HAART programs in developing countries, EFV is likely to become a choice for women with reproductive potential in whom NVP is contraindicated (eg, prior NVP toxicity or among women initiating HAART with CD4⁺ counts >250 cells/mm³), particularly in settings where access to protease inhibitor drugs is limited. It is important, therefore, for HAART programs to develop parallel family planning services that are staffed by personnel uniformly trained to provide accessible and comprehensive counseling and education for HAART (including EFV)-treated women who become pregnant. The management of EFV-exposed pregnancies when identified during later stages of pregnancy (second or third trimester) is still done on a case-by-case basis. Routine screening for neural tube defects via ultrasound imaging and/or serum α-fetoprotein levels is of limited availability in the region; therefore, it is not routinely practiced.

CONCLUSIONS

The risk of birth defects after early EFV exposure precludes the widespread use of EFV for many women in need of convenient, efficacious, and safe HAART in resource-limited settings. Numerous HIV-infected and HAART-treated women are likely to receive EFV in the wake of the expansion of national ARV treatment programs worldwide. Evidenced-based data on the risk of EFV in pregnancy can only be strengthened if all care providers in the field prospectively report all HAART (and EFV) exposure outcomes to a pregnancy exposure registry, such as the international Antiretroviral Pregnancy Registry. Those involved in ARV treatment program monitoring and evaluation need to ensure that family planning and reproductive health counseling needs are met, especially as they relate to physical manpower and training requirements. To advise and educate burgeoning numbers of HAART-treated women and men in sub-Saharan Africa more effectively, the state of existing family planning services including contraception and procreation counseling warrants in-depth evaluation.

ACKNOWLEDGMENTS

The authors are indebted to the patients who participated in the Tshepo Study and thank the entire Tshepo Study team, who provided excellent care and assistance. They also thank the Botswana Ministry of Health for supporting the study and the Bristol-Myers Squibb's “Secure the Future” foundation for funding the Tshepo Study. The authors also acknowledge and thank Patricia Burns (Harvard School of Public Health AIDS Initiative, Boston, MA), Lynne Mofenson (National Institute of Child Health and Human Development [NICHD], Bethesda, MD), and Heather Watts (NICHD, Bethesda, MD) for their careful review, feedback, and editorial assistance in the preparation of this manuscript.