750 pregnant women were screened for inclusion into the trial () starting in February 2003, of whom 181 (24%) did not meet the entry criteria, the majority (144) of this group because their viral load was below 2,000 copies/ml. 162 women were not randomised during labour for the following reasons: they either delivered before randomisation or not at the study site (89), had elective caesarean section (32), or received nontrial NVP (19). 407 women were randomised from March 2003 to November 2005, 406 of whom took trial medication. Their median age was 27 y (interquartile range [IQR] 24–30), median CD4 count was 318 cells/mm3 (IQR 212–425), and median screening HIV RNA viral load was 27,100 copies/ml (IQR 8,130–74,200) (). There were 376 mothers (92.4% of randomised) with follow-up information at 6 wk; no participants were omitted from the 6-wk analysis other than those lost to follow-up or with no postpartum genotypic results. Four NVP-only, four NVP/CBV4, and six NVP/CBV7 did not return for any follow up visits; their viral load and CD4 counts were similar to the women that did return. There was no sequencing data on another 17 women, all of whom were in the CBV-containing arms; 14 had viral loads <400 copies/ml and the samples provided by the remainder were either insufficient or unable to be amplified. There were 413 live births and two stillbirths. Among the live births, there were eight sets of twins. One intrauterine death was detected after randomisation but before receipt of treatment, leading to the mother not being treated. Two infants were unable to take study medication because of birth asphyxia (both in the NVP/CBV4 arm). The 411 infants who received medication had a median weight of 3.0 kg (IQR 2.9–3.4 kg) and a median length of 50 cm (IQR 48–52 cm); 48.9% were female (). HIV-1 clade C was detected in 98.3% (399/406) of pregnant women, and 12 (3%; 12/393; 95% confidence interval [CI] 2%–5%) had an NNRTI resistance-associated mutation immediately prior to receipt of NVP; one woman in the NVP-only arm, eight in the NVP/CBV4 arm, and three in the NVP/CBV7 arm. Of these women, 11 had one NNRTI resistance-associated mutation (two A98G, two V101E, two K103N, one V106IM, one 108I, two V179D, one M230L) and one had V106M, Y181C, and G190A. Sequencing for resistant mutations immediately prior to receipt of NVP did not yield a result in 14 mothers (3.4% of mothers tested), their median viral load was 539 copies/ml (IQR<400–17,300 copies/ml).
Baseline characteristics of randomised HIV-infected women and those of their infants at birth, by study arm.
Selection of Maternal NNRTI Resistance Mutations
The NVP-only, NVP/CBV4, and NVP/CBV7 regimens were administered to 75, 163, and 168 women, respectively, of whom 71 (95%), 154 (95%), and 151 (90%) had follow-up visits and sequence results at 2 and 6 wk postnatally. Only eight of all the resistance mutations looked for were able to be sequenced in the postnatal specimens (); the remaining were either present at proportions below the limit of detection or were not present at all. New NNRTI resistance-associated mutations were detected at 2 wk, at 6 wk, or at both visits in 59.2% (42/71; 95% CI 46.8%–70.7%), 9.7% (15/154; 95% CI 5.1%–14.4%), and 7.3% (11/151; CI 3.7%–12.7%) of these women, respectively.
New resistance mutations by study arm with estimated efficacy of preventing NNRTI resistance mutations in women who received either the 4 or 7 d of AZT plus 3TC (CBV4/7) treatment arms compared to those exposed to sdNVP alone.
Most mothers receiving CBV took more than the prescribed number of doses as they all were dispensed two additional days of study drug; in the NVP/CBV4 treatment arm women took a median of nine doses (IQR 8–11), their median maternal CBV adherence was 113% (IQR 100%–133%); and in the NVP/CBV7 treatment arm women took a median of 15 doses (IQR 14–17), their adherence was 107% (IQR 100%–114%). However, when women with CBV adherence of <75% or >125% are excluded, resistance rates remain unchanged compared to those including these women: 10.7% (11/103; CI 6%–18%) for NVP/CBV4 and 6.3% (9/144; CI 3%–11%) for NVP/CBV7. An as-treated analysis, restricted to women who received CBV, suggested no relationship between the number of doses of CBV taken and the selection of resistance (permutation test, p
On the basis of the 59.2% rate of resistance emergence in the NVP-only arm, the expected number of mothers with emergent resistance in the NVP/CBV4 arm and NVP/CBV7 arm was estimated at 92 and 91, respectively. The observed rates of emergence correspond to 80.5% and 87.8% efficacy in preventing emergence of resistance, respectively. Comparing the NVP arm to the combined CBV arms the estimated efficacy is 85.6%. Mutation-specific efficacy was over 90% for both leading mutations: K103N and Y181C.
The pattern of emergent mutations was altered by use of CBV (): 106M was the second leading mutation with the selection of relatively uncommon mutations at 101, and 108 which were not seen with NVP alone. Furthermore, after NVP-only, 60% (25/42; CI 43%–74%) of the mothers with mutations at 2 and/or 6 wk had three or more different NNRTI mutations. Whereas, 19% (5/26; 95% CI 7%–39%) of those with mutations after NVP/CBV4 or NVP/CBV7 had three or more different NNRTI mutations (Cochran-Armitage trend test, p
0.001). One woman in the NVP/CBV7 arm had a new mutation associated with 3TC resistance at RT codon 184, detected by population sequencing at week 6 but not at week 12. This was the only nucleoside analogue reverse-transcriptase resistance mutation detected.
After the 6-wk visit, follow-up of women with resistance mutations continued until NNRTI mutation-free virus was documented. The last study follow up visit was in January 2007. Fading of NNRTI resistance mutations appeared to occur more rapidly in women who received NVP/CBV7 than NVP/CBV4 (). At 12 wk postpartum, 62% of NVP-only women who had NNRTI resistance mutations at 6 wk retained their NNRTI resistance, whereas 34% of the women who received one of the NVP/CBV regimens and who had NNRTI resistance at 6 wk retained resistance at 12 wk (Fisher's exact test, p
0.04). At 24 and 48 wk the results are similar, but there were few women at these two time points.
Long-term follow-up of women with emergent NNRTI mutations detected 2–6 wk after exposure to sdNVP during labour.
Maternal Viral Load
The median maternal viral load in labour, prior to receipt of study medication, was 27,900 copies/ml, 23,950 copies/ml, and 30,100 copies/ml for the three treatment arms, NVP, NVP/CBV4, and NVP/CBV7, respectively. HIV RNA was modestly reduced by day 2 and reached an observed median nadir at day 14 with counts of 7,650 copies/ml for NVP-only, 421 copies/ml for NVP/CBV4, and <400 copies/ml for NVP/CBV7. However, by the 6-wk visit median maternal HIV RNA levels were similar to those at baseline: 33,600 copies/ml, 28,600 copies/ml, and 33,850 copies/ml for the NVP-only, NVP/CBV4, and NVP/CBV7 arms, respectively.
Maternal Baseline CD4 and Emergent NNRTI Resistance
One-fifth of the women recruited into this trial (21%, 85/406, 95% CI 17%–25%) had CD4 cell counts of ≤200 cells/µl. We compared emergent resistance rates by whether baseline CD4 was ≤200 or >200 cells/µl using the Fisher's exact, two-sided test: NVP arm, 82.4% (14/17) versus 48% (24/50), p
0.02; NVP/CBV4 arm, 25.8% (8/31) versus 5.5% (6/110), p
0.003; NVP/CBV7 arm, 10.3% (3/29) versus 5.4% (6/112), p
Efficacy in Preventing MTCT
The overall, combined intra-uterine and peri/postpartum HIV transmission rate by the 6-wk visit was 12.2% (49/403; CI 9.3%–15.7%). The majority of infants with HIV-1 infection were HIV-1 positive when first tested within a day of delivery (HIV DNA- and RNA-positive) and confirmed-infected at follow-up (43/403 infants; 8/76 NVP-only, 25/161 NVP/CBV4, 10/166 NVP/CBV7), suggesting in utero HIV acquisition. Three additional infants were HIV-1 infected when first tested at 2–6 wk of age, all in the NVP/CBV7 arm and three were HIV-negative initially, but positive at 6 wk.
Infants Developing NNRTI-Resistant Mutations
In the infants with intra-uterine HIV-1 infections and follow-up after exposure to antiretrovirals, NNRTI-resistant mutations were seen in seven of the eight HIV-infected infants treated with NVP-only, four of 25 of those who received NVP/CBV4, and none of the ten who received NVP/CBV7. The difference between NVP alone and the combined NVP/CBV arms is statistically significant (Fisher's exact, two-sided, p<0.001). No mother-infant pair developed identical NNRTI-resistant mutations.
Three maternal deaths occurred and all received NVP/CBV4. The deaths occurred at 4, 9, and 12 mo after delivery and all were caused by respiratory disease; tuberculosis was suspected in the deaths of two of the three women. Both of the participants who died of suspected TB had CD4 counts of <100 cells/mm3 at the time of screening, one woman had been treated for tuberculosis for 4 mo prior to her death.
During the antenatal screening period, before any study drug had been administered, one-third of the mothers suffered adverse events. The most common events during this period were urinary tract infections. Pregnancy-induced hypertension occurred in ten (2.5%) mothers, two of whom were diagnosed with pre-eclampsia. Following randomisation and delivery the most common adverse events were also infections: 4% (3/70), 6.7% (11/163), and 4.2% (7/168) for the NVP-only, NVP/CBV4, and NVP/CBV7 arms, respectively. Postpartum sepsis was diagnosed in 1.3% (1/70), 3.1% (5/163), and 2.4% (4/168) for the NVP-only, NVP/CBV4, and NVP/CBV7 arms, respectively. Suspected adverse drug reactions were reported in 2.5% (10/406) of mothers and were limited to the combination arms, 1.8% in NVP/CBV4 arm (3/163), and 4.2% in NVP/CBV7 arm (7/168); none were severe. Rashes were reported for six women, for one of them a relationship to NVP use was considered possible. The most common abnormal laboratory tests were below normal haemoglobin and above normal aspartate aminotransferase (AST) levels; there were 24, 55, and 56 women in the NVP, NVP/CBV4, and NVP/CBV7 arms, respectively, whose AST levels were elevated above normal. However, there were no hepatic adverse events reported. There were no significant trends associated with CBV use.
Serious adverse events were observed in 4.9% of mothers during the screening period and in 8.1% of mothers after randomisation. During the screening phase these events were primarily pregnancy and perinatal conditions (31/37), whereas infections were the majority after randomisation (21/33).
There were 13 infant deaths (3.2%; 95% CI 2%–5%); ten were HIV-infected, three in the NVP-only arm, five in the NVP/CBV4 arm, and two in the NVP/CBV7 arm. The causes of death in the HIV-infected infants were either gastroenteritis or respiratory infections; the oldest child to die in this group was 10 mo old. Three HIV-negative infants died, one from congenital pneumonia (NVP-only arm), one from complications of trisomy 18 (NVP/CBV4 arm), and one sudden infant death (NVP/CBV4 arm) without any preceding illness at 37 d after delivery. Suspected drug-related reactions were reported in 1.2% of infants (5/411). The events were hyperamylasaemia in the NVP-only arm (1/77), abdominal pain and vomiting in the NVP/CBV4 arm (2/164), and vomiting and jaundice in the NVP/CBV7 arm (2/170). The jaundice in the NVP/CBV7 arm was classified as serious. The most common abnormal laboratory tests were below-normal haemoglobin and above-normal bilirubin levels. There were no significant trends associated with CBV use.