Two-Dose Intrapartum/Newborn Nevirapine and Standard Antiretroviral Therapy to Reduce Perinatal HIV Transmission:  A Randomized Trial FREE

Transmission of human immunodeficiency virus (HIV) from mother to child can occur in utero and during or after delivery.¹ Most transmission occurs late in pregnancy or at the time of delivery.² The results of the Pediatric AIDS [acquired immunodeficiency syndrome] Clinical Trials Group (PACTG) 076, published in 1994, offered proof of the concept that a zidovudine regimen given antepartum, intrapartum, and to the newborn for 6 weeks could significantly reduce HIV transmission.³ Implementation of this regimen in the developed world has reduced transmission rates of perinatal HIV to between 5% and 8%.^4- 6 About two thirds of the remaining transmission occurs intrapartum,^3- 4 and therefore has the potential to be further reduced by additional intrapartum interventions.

Nevirapine is a potent nonnucleoside reverse transcriptase inhibitor that rapidly reduces viral load, is well absorbed orally, and readily crosses the placenta. The safety and pharmacokinetics of nevirapine administered to pregnant women during labor and to their newborns were previously evaluated.^7- 8 Administration of a 200-mg oral dose to women during active labor with a 2-mg/kg oral dose to their newborns between 48 and 72 hours after birth maintained newborn nevirapine concentrations above 100 ng/mL (10 times the in vitro inhibitory concentration) through the first 7 days of life.

Recommendations have been developed, which reflect studies occurring subsequent to the PACTG 076, regarding the potential benefit of combination therapy in reducing perinatal transmission as well as for maternal disease during pregnancy (http://hivatis.org/trtgdlns.html). During the time the PACTG 316 trial was being conducted, investigators showed that the 2-dose maternal/newborn nevirapine regimen produced a decrease in maternal to newborn HIV transmission compared with an ultrashort course of zidovudine in predominantly breastfeeding women in Uganda who were not receiving antiretroviral therapy (ART) (13.1% vs 25.1%, respectively, at 14-16 weeks).⁹ It remained unknown if this simple nevirapine regimen would provide additional benefit in the interruption of mother to newborn transmission of HIV in nonbreastfeeding women receiving standard ART. To test the hypothesis that a 2-dose intrapartum/newborn nevirapine regimen in addition to standard ART could further reduce perinatal HIV transmission, the following randomized, placebo-controlled trial was conducted.

The PACTG 316 study was a phase 3, international, multicenter, randomized, double-blind study of intrapartum and newborn nevirapine regimens in addition to standard ART (Figure 1). In addition to US PACTG sites, international collaborating sites were added to obtain sufficient numbers of subjects to address the primary study question, and subjects in Europe, Brazil, and the Bahamas were enrolled. In Europe, 2 study groups participated (Agence Nationale de Recherches sur le SIDA [ANRS] and the European Collaborative Study [ECS]).^10- 11

The study protocol and informed consent were reviewed by the institutional review board at each study site. After written informed consent was obtained, HIV-infected pregnant women were randomized to either nevirapine (200 mg of oral nevirapine at onset of labor and 2 mg/kg of oral nevirapine for newborns between 48 and 72 hours after birth) or the corresponding placebo doses. Study drug remained at the hospital where delivery was planned and was directly administered by medical staff when delivery was determined to be likely within 24 hours. An additional dose of nevirapine or placebo could be dispensed to the mother if labor continued more than 48 hours after the initial dose or to the newborn immediately after delivery if the mother did not receive study drug or received her dose less than a hour prior to delivery.

In addition to study medication, participating women were receiving standard ART, as determined by their health care clinician, which could include any licensed antiretroviral agent except a nonnucleoside reverse transcriptase inhibitor. Women were encouraged to receive the PACTG 076 zidovudine regimen as a minimum therapy.³

Eligible subjects were HIV-infected women who were 13 years or older (or of legal age to give consent) and at least 28 weeks' gestation. Following closure of another study (PACTG 185), which had enrolled patients at 20 weeks' or more gestation,⁴ PACTG 316 lowered the gestational age to 20 weeks. Gestational age was estimated using either reliable menstrual history that corresponded with uterine size, or physical examination, and/or sonogram performed at less than 20 weeks gestation. Women were excluded from the study if enrolled in other perinatal treatment trials; previously received nonnucleoside reverse transcriptase inhibitors; hypersensitive to benzodiazepines; or had elevated serum alanine aminotransferase (>10 times upper limit of normal). Women who intended to breastfeed and those pregnancies identified with a fetal anomaly incompatible with life were also excluded.

Evaluations that included safety assessments were conducted for women throughout follow-up to 6 weeks postpartum and for infants throughout follow-up to age 6 months. Safety assessments consisted of obtaining medical history, physical examination, and laboratory testing. Women underwent clinical and laboratory evaluation at enrollment, during labor, and between 48 and 72 hours postpartum and 4 and 6 weeks postpartum. Evaluations included history and physical examination, complete blood count, CD4 cell count, blood chemistry levels, and HIV RNA quantification.

Newborns had a complete birth history recorded. Additional study visits occurred between 48 and 72 hours after birth; between days 6 and 9; between weeks 4 and 6; and at months 3 and 6. History and physical examination were obtained at each visit through the end of the 6-month follow-up. The safety laboratory tests to obtain complete blood count and alanine aminotransferase were performed after birth, between days 6 and 9 of life, and again between weeks 4 and 6. If laboratory results were significantly abnormal, testing was repeated until the abnormalities resolved. A HIV DNA polymerase chain reaction (PCR) assay was performed at birth, between weeks 4 and 6, at month 3, and at month 6. If any DNA PCR assay was positive, the infant returned to the study site for another DNA PCR assay and a peripheral blood mononuclear cell culture. Women or infants discontinued from the study prior to receiving study medication continued to be followed up for a prespecified duration.

The HIV DNA PCR assays were performed in laboratories certified by the ACTG or another regional quality-assurance program. At all PACTG sites, and at most international sites, DNA PCR assay was performed using Amplicor Detect kits (Roche Diagnostic Systems, Branchburg, NJ) as outlined in the PACTG virology manual.¹² Although the commercial Amplicor Detect assay is sensitive for HIV clade B subtypes, sensitivity is reduced for non–clade B. Therefore, the international sites used the DNA PCR method with optimal sensitivity in their patient populations. The HIV peripheral blood mononuclear cell culture was required for infants with HIV-positive DNA PCR assay results. The HIV culture was performed using fresh cells with standard techniques.¹³

The HIV RNA quantitation was performed in laboratories certified by the ACTG¹⁴ or other regional quality-assurance program. Plasma was separated and frozen at −70°C within 30 hours of collection. Analysis was done using either reverse transcriptase PCR (Amplicor HIV Monitor, Roche Diagnostic Systems), nucleic acid sequence-based amplification (Nuclisens, Organon Teknika Corp, Durham, NC) or branched-chain DNA signal amplification (Quantiplex HIV RNA, Bayer Diagnostics, East Walpole, Mass) per manufacturers' instructions.

An infant was considered HIV-infected if specimens obtained at 2 different time points tested positive for the virus (with HIV DNA PCR assay and/or culture). Infants were considered uninfected if they had 2 HIV-negative DNA PCR assays, one at or after age 1 month and the other after age 2 months. If infection status could not be determined based on protocol-specified virological studies, a subgroup of the study team, who were blinded to study treatment assignment, reviewed clinical and virological data to determine HIV status. Infection status was considered "indeterminate" if an infant had a positive assay, but did not meet the definition of infected, or did not have a DNA PCR assay beyond age 1 month. Twin births were assessed as a single HIV transmission if either infant was infected with HIV, and as a single nonoccurrence of transmission if both twins were uninfected. Twins were considered indeterminate if they did not meet either of the above criteria.

The study was designed to achieve an 80% power to detect a reduction in HIV transmission from 5% in the control group to 2.5% in the nevirapine group using a 1-sided .05-level test. Target accrual was 2009 patients to obtain 1808 cases assessable for the primary analysis. Randomization was stratified according to use of ART during pregnancy (none, monotherapy only, or multiagent therapy) and CD4 cell count at randomization (<200, 200-399, or ≥400 cells/mm³).

The primary end point was incidence of HIV transmission from mother to infant. Because the study was blinded and study treatment commenced at labor and delivery, the independent data and safety monitoring board endorsed an unbiased maximally powerful primary analysis that included only women who were dispensed the study drug during labor.

The primary treatment comparison was based on a Fisher exact test, stratifying by participating group.¹⁵ Fisher exact tests were also used in all univariate analyses, including comparison of infection rates between treatments within subgroups of patients, and comparison of infection rates for different levels of a potential risk factor. P<.05 was considered significant. Logistic regression models were used to assess the association between multiple risk factors and transmission. Maternal and neonatal clinical and laboratory toxic effects were monitored and reported according to the National Institutes of Health Division of AIDS standardized grading system (grades extended from mild [grade 1] to serious [grade 4]). Data analysis was performed by the Statistical and Data Management Center of the Pediatric AIDS Clinical Trials Group at Harvard School of Public Health.

Administrative and safety data were reviewed at least once per year by an independent data and safety monitoring board composed of members from the Division of AIDS in the National Institutes of Health. Four interim analyses and 1 final efficacy analysis of the primary treatment comparison were planned. At the third interim analysis in June 2000, the transmission rate in both groups was significantly less than the 5% assumed for the study design. Because a substantial increase in sample size would have been required to meet the original study objectives, the data and safety monitoring board recommended discontinuing new patient enrollment, continuing blinded follow-up for those who had delivered, and removing women from the study who had not yet delivered. Therefore, study enrollment was discontinued June 8, 2000. Sample size calculations and interim stopping boundaries were based on the binomial data design module of the program EaSt (Cytel Software Corp, Cambridge, Mass). Statistical analyses were performed using SAS software (version 6.12, SAS Institute Inc, Cary, NC).

Between May 13, 1997, and June 8, 2000, a total of 1506 women were enrolled. The PACTG accrued patients during the entire period and enrolled 1052 women from 67 sites distributed throughout mainland United States and Puerto Rico. International groups began enrolling at the following times: ANRS, December 1998 (287 subjects at 21 sites); ECS, February 1999 (118 subjects at 14 sites); the Bahamas, September 1999 (18 subjects at 1 site); and Brazil, December 1999 (31 subjects at 2 sites).

Of the 1506 women enrolled, 1270 received study drug and delivered a newborn (Figure 1). These 1270 deliveries resulted in 1245 singleton births and 25 sets of twins (total of 1295 newborns). Study drug was not administered for 236 women primarily per data and safety monitoring board study closure recommendation, precipitous delivery, or logistical problems (Figure 1). The number of women who did not receive intrapartum study drug had a similar distribution between the nevirapine (n = 112) and placebo arms (n = 124) and between PACTG and international groups. Ninety-eight percent of women (1245/1270) had a final study visit at least 4 weeks postpartum and 98% of infants (1222/1248) had a final study visit at least 5 months after birth.

The study drug was dispensed at the hospital. Among the 1270 women, 92% received the study drug at least 1 hour prior to delivery and 98% of their 1295 newborns received study medication. Median time between the last study dose and delivery was 5.3 hours (18 women received 2 doses). Eleven newborns received a second dose of nevirapine. Proportions of patients receiving study drug were similar between the 2 treatment arms (C.K.C., unpublished data, December 2001).

As shown in Table 1, women who received intrapartum nevirapine (n = 642) were similar to those who received placebo (n = 628) with respect to baseline and delivery characteristics. Only 12% of women enrolled had a CD4 cell count below 200 cells/mm³ at baseline. Eleven percent had HIV RNA above 10 000 copies/mL at the time of delivery, with over half below 400 copies/mL. Fifty percent of women from PACTG had a delivery HIV RNA below 400 copies/mL compared with 62% from ANRS, 62% from ECS, 47% from Brazil, and 13% from the Bahamas.

Only 4 women refused standard ART or prophylaxis. Of the remaining 1266 women, 291 started therapy before pregnancy and 975 started treatment at a median gestational age of 22 weeks. Twenty-three percent of the women received zidovudine alone; 28%, zidovudine plus lamivudine; 8%, other combinations without a protease inhibitor (PI); and 41%, combinations including a PI. The most commonly used PI was nelfinavir (60%), followed by indinavir (21%), saquinavir (6%), and ritonavir (1%). Twelve percent received treatment with 2 PIs. The proportion of women from PACTG sites who received PIs increased during the study period from 32% during the first year of recruitment to 45% during the second year and 53% during the third year. Thirty percent of women from ANRS and 39% from ECS received PIs. Seventeen percent of PACTG patients received zidovudine alone compared with 30% of patients from ANRS, 36% from ECS, 42% from Brazil, and 100% from the Bahamas.

Thirty-four percent of women delivered by elective cesarean section (before onset of labor and before rupture of membranes). Twenty-four percent of PACTG patients delivered by elective cesarean section compared with 58% of patients from ANRS, 81% from ECS, 26% from Brazil, and 0% from the Bahamas. Nineteen percent of infants were preterm (<37 weeks' gestation); 1.5%, very preterm (<32 weeks' gestation); 12%, low birth weight (<2500 g); and 3%, very low birth weight (<2000 g). The frequency of preterm delivery (<37 weeks) was the same (19%) for all women regardless of PI treatment (P>.99, Fisher exact test).

Infant HIV infection status was determined for 1248 of 1270 deliveries based on the protocol-specified definition (1200 cases) or on sufficient data to allow classification by the study virological review subgroup (48 cases). Twenty-two deliveries (11 in each treatment group) were classified as indeterminate. All newborns classified as indeterminate had either no DNA PCR results or negative results, and no samples available after age 4 weeks. No newborns in this group had any positive DNA PCR assay results. Detection of HIV infection occurred in 19 newborns, 9 (1.4%) of 631 deliveries in the nevirapine arm and 10 (1.6%) of 617 deliveries in the placebo arm (Table 2). The 95% confidence interval for difference in transmission rate (−0.2%) between nevirapine and placebo arms ranged from –1.5% in favor of nevirapine to 1.2% in favor of placebo (P = .82, Fisher exact test). Ten (53%) of the 19 transmissions occurred in utero as defined by a positive DNA PCR assay within 72 hours of birth; 1 infected infant did not have a birth sample obtained; and the remaining 8 (42%) had negative samples at birth with subsequent positive results and would be considered to be infected intrapartum¹⁶ (Table 2).

Rates of transmission according to randomized treatment assignment based on study group, maternal characteristics (including type of antepartum ART), and infant characteristics are shown in Table 3. Differences according to randomized treatment assignment were not statistically significant for any subgroup, even for women with baseline CD4 cell counts below 200 cells/mm³ (P = .37), delivery HIV RNA above 10 000 copies/mL (P = .40), or antenatal zidovudine monotherapy (P = .69). All P values were calculated using the Fisher exact test.

Overall, women with low baseline CD4 cell counts (<200 cells/mm³) had a higher risk of HIV transmission (3.4%) than those with CD4 cell counts of 400 cells/mm³ or higher (0.8%) (P = .03, Fisher exact test). Transmission risk was also higher among women with HIV RNA levels of 400 copies/mL or higher (2.9%) compared with those with HIV RNA below 400 copies/mL (0.3%) (P<.001, Fisher exact test). In univariate analyses, risk for HIV transmission did not differ significantly according to type of antepartum ART or mode of delivery.

Treatment arm, study group (PACTG compared with other groups), type of antenatal ART, mode of delivery, and maternal HIV RNA level were included in multiple logistic regression models. Maternal viral load at delivery (85 were missing data) was the only significant predictor of HIV transmission (C.K.C., unpublished data, December 2001).

Table 4 shows the frequency of HIV transmission according to maternal viral load at delivery, type of antepartum ART, and mode of delivery. Although maternal viral load is an important risk factor for perinatal transmission, the relative merit of specific maternal ART regimens cannot be ascertained due to confounding. For example, women who received zidovudine monotherapy were more likely to undergo elective cesarean section (43% vs 30% receiving other treatment regimens) and have an HIV RNA level of 400 copies/mL or more (62% vs 43% receiving other treatment regimens).

Of the 236 women excluded from the primary analysis because they did not receive study drug, 152 delivered during the study and infection status was determined for 139 (71 in the nevirapine arm and 68 in the placebo arm). Transmission of HIV occurred in 2 deliveries (1.4%), both assigned to the placebo group.

All posttreatment infant and maternal rash events of grade 2 or higher were reported. Nonrash toxic events were reported for events of grade 3 or higher. Overall, severe toxicity was rare in both women and infants and did not differ significantly between nevirapine and placebo groups (Table 5). For all maternal and infant toxic events, there was no statistically significant difference identified between the nevirapine and the placebo group. In all cases, the comparisons were by Fisher exact test.

Maternal rashes of grade 2 or higher were uncommon, occurring in only 15 women (7 in the nevirapine and 8 in the placebo group). Of these, 10 had defined alternative etiologies for the skin changes, including surgical site infection and contact dermatitis. Seventy-eight women (6.1%) had grade 3 or higher nonrash toxicity. The most frequent events reported were anemia (2.5%) and delivery/postpartum complications (1.1%). Grade 3 or higher hepatic toxicity (elevated aminotransaminases or bilirubin) was reported in 10 women (5 in each treatment group). Underlying liver disease was identified in 4 of these women (2 in each treatment group: cholestasis, hepatitis B, and 2 cases of hepatitis C). Three women died during the postpartum follow-up period (2 in the nevirapine arm and 1 in the placebo arm); all had received study treatment. Deaths were attributed to ischemic cardiomyopathy, sickle cell disease, and disseminated histoplasmosis. All were considered unrelated to study treatment.

Toxicity evaluation included all newborns with exposure to study drug (n = 1413) through transplacental passage of maternal drug and/or through receipt of study drug directly to the newborn (Table 5). Grade 2 or higher rash was reported in 76 newborns (5.4%) (31 in the nevirapine arm and 45 in the placebo arm). Most rashes were attributed to typical newborn findings (16, diaper dermatitis; 15, seborrhea; 11, fungal skin lesions; 5, atopic dermatitis; and 1, impetigo). Of 4 infants with grade 3 rashes, 2 occurred between birth and age 3 days and 2 at age 4 days or older. No rash was attributed to nevirapine, which was determined by assessing factors such as timing and whether there was another obvious etiology, such as Candida dermatitis. No grade 4 rashes were reported.

Nonrash toxicity of grade 3 or higher was reported in 430 newborns (30.4%). Most events were judged to be unrelated to study drug, which was determined by assessing factors such as timing and whether there was another obvious etiology such as cytomegalovirus hepatitis (see below). The most frequent events included anemia (13.1%) and neutropenia (12.6%). More than 90% of events resolved during follow-up. As physiological hyperbilirubinemia is a common event in newborns, elevated bilirubin without elevated alanine aminotransferase was not considered a hepatic toxicity. Three newborns (1 in the nevirapine arm and 2 in the placebo arm) had grade 3 or higher hepatic events; all events occurred after study day 4. All 3 cases were attributed to cytomegalovirus hepatitis.

Eight infants died, 3 in the nevirapine and 5 in the placebo group. At the time of death, 4 infants were uninfected, 3 were classified as indeterminate, and 1 was HIV-positive. Deaths occurred between birth and age 24 weeks and were attributed to sudden infant death syndrome (3 cases), bacterial infection (2 cases), gastric hemorrhage (1 case after open-heart surgery for congenital heart disease), trauma (1 case), and cause of death undetermined (1 case).

The observed transmission rate of 1.5% in PACTG 316 was much lower than anticipated. Additionally, in the presence of standard therapy, most (53%) of the few remaining perinatal transmissions occurred in utero (as defined by a positive HIV DNA PCR assay at birth) and hence would be unaffected by an intervention targeted at the intrapartum period. The risk of transmission was the same for nevirapine and placebo arms and the tight confidence interval for difference in transmission risk indicates that any benefit of adding the intrapartum/newborn nevirapine regimen to standard ART is likely to be small.

Even among women with characteristics known to increase the risk of perinatal transmission (eg, delivery HIV RNA >10 000 copies/mL or CD4 cell count <200/mm³), transmission rates were below 5%. These low transmission rates may be due to the frequent use of antenatal treatment with highly active ART (HAART) and/or elective cesarean delivery in these high-risk populations.

Consistent with the low transmission rate observed in this study, several recent studies have demonstrated that antenatal treatment with ART regimens, including PIs in some women and/or combining ART with elective cesarean delivery, can decrease perinatal HIV transmission to less than 2%.^11,17- 20 Use of elective cesarean delivery and combination ART with PIs varied among groups involved in this study. Elective cesarean section was used frequently in European subjects. Elective cesarean section was performed less frequently in US subjects, while use of combination therapy including a PI was more frequent. Despite these differences, overall transmission rates did not vary by subject location (United States vs other). This suggests that the end result of these different standards of care related to the use of elective cesarean delivery and HAART is the same low transmission rate overall of 1.5% or less.

The 2-dose nevirapine regimen was safe and well tolerated by mothers and newborns, as has been reported in 2 other perinatal trials of the nevirapine regimen conducted in Africa in women who did not receive antenatal ART.^9,21- 22 Although there were grade 3 and 4 events, these were evenly distributed between the 2 treatment groups. This demonstrates that the events were related to the underlying diagnosis (perinatal events or HIV infection) or other medications received by the women and newborns. Furthermore, there were no additive adverse effects when nevirapine was combined with other ART.

While the risk of maternal and newborn toxicity with intrapartum/newborn nevirapine appears minimal, data from a Ugandan trial of the 2-dose nevirapine regimen (HIV Network for Prevention Trials [HIVNET] 012) indicate the potential for development of nevirapine-resistance mutations in women with active viral replication at the time they receive single-dose nevirapine.²³ In HIVNET 012, the resistance mutations detected during the early postpartum period were not apparent when the women were tested at later time points.²⁴ In a substudy of women participating in PACTG 316 who had active viral replication, new nevirapine-resistance mutations were demonstrated in 15% (95% confidence interval, 8%-23%), suggesting that risk of new nevirapine-resistance mutations may occur even in women receiving other ART.^25- 26 The long-term impact of these mutations on subsequent maternal treatment is not yet known.

The PACTG 316 does not address use of the intrapartum/newborn nevirapine regimen in women not receiving other ART, for whom the risk of perinatal HIV transmission is high (≥25%).³ Two African trials have shown that the 2-dose nevirapine regimen is effective in reducing transmission in women who have not received antenatal ART. In HIVNET 012, the 2-dose nevirapine regimen significantly reduced perinatal HIV transmission compared with an ultra-short intrapartum/newborn regimen of zidovudine in breastfeeding Ugandan women who did not receive antenatal treatment.⁹ An additional trial in South Africa demonstrated that the 2-dose nevirapine regimen is comparable with an effective intrapartum/newborn combined regimen of zidovudine and lamivudine in reducing perinatal HIV transmission in breastfeeding and formula-feeding women who had not received antepartum therapy.²¹ Although such a study has not been done in a developed nation, the intrapartum/newborn nevirapine regimen may still be a treatment option in developed countries for women who have received no ART prior to labor and delivery.

Thus, in situations in which antenatal ART has not been received and the risk of perinatal HIV transmission is therefore high, use of the 2-dose nevirapine regimen appears safe and offers significant benefit in reducing perinatal transmission. In developing countries, implementation of this regimen could potentially prevent HIV infections in many children each year, with the benefits of nevirapine clearly outweighing potential risks. However, in circumstances in which women are receiving prenatal care including antenatal ART and can safely undergo elective cesarean delivery, there is no demonstrable benefit of intrapartum/newborn nevirapine therapy.