One hundred twelve women were enrolled between December 23, 2002 and May 13, 2005. Twenty-eight women did not enter postpartum follow-up. Eighteen were ineligible for postpartum follow-up due to premature delivery such that no baseline (third trimester) viral load could be calculated (n = 10), fetal demise (n = 1), viral load ≥5,000 copies/mL near delivery (n = 4), or change in ARV therapy (n = 3). The other 10 did not enter postpartum follow-up for technical reasons: they were unable to come to clinic/missed postpartum entry visit time window (n = 7) or withdrew consent (n = 3).
Age, CD4 cell count, weight, gestational age, number of prior pregnancies, and history of prior AIDS-defining conditions at study entry did not differ significantly between those who did and did not enter postpartum follow-up (). However, a significantly larger proportion of women without postpartum follow-up reported a history of premature delivery (25% vs 4.8%; P = .01), and their median study entry plasma HIV-1 RNA level was higher (2.1 log10 vs 1.7 log10; P = .02). At delivery, the 84 women who were followed postpartum were on 29 different ARV regimens with the most common being zidovudine/lamivudine plus nelfinavir (n = 19), zidovudine/lamivudine plus nevirapine (n = 12), and zidovudine/lamivudine plus lopinavir/ritonavir (n = 8). Four or fewer women were on each of the other ARV regimens.
Of the 84 women who registered for postpartum follow-up, 12 withdrew from study follow-up prior to the primary endpoint at postpartum week 24 for reasons of loss to follow-up (4), subject decision (4), unable to attend clinic (2), no longer taking ARV (1), and moved (1). Of the 72 women with week 24 postpartum follow-up, viral load values were available on 63 and missing on 9 due to missed visits (8 women) or invalid assay result (1 woman). These 63 women comprise the dataset for our primary endpoint. Baseline characteristics are detailed for these 63 women as well as the subset of 44 women who remained on stable ARVs through week 24 postpartum ().
Eighteen women (18/63; 28.6%; 95% CI, 17.94–41.4) met criteria for viral load rebound (). In sensitivity analyses where missing viral load equals rebound, the estimated probability of rebound was 46.4% (95% CI, 35.5–57.7; 21 values imputed); using the last-value-carried forward, this probability estimate was 27.4% (95% CI, 18.2–38.2). The probability of viral load rebound did not differ significantly by whether women were on their initial HAART regimen, as 14 of 46 (30.4%; 95% CI, 17.7–45.8) women on initial HAART rebounded and 4 of 17 (23.5%; 95% CI, 7.0–49.9) of ARV-experienced women rebounded (P = .76).
Viral load and genotypic susceptibility score for women who developed rebound at 24 weeks postpartum
For the 63 women with postpartum week 24 viral load data in univariate logistic regression (exact method) analysis, only postpartum week 6 HIV-1 RNA and race/ethnicity were significantly associated with viral load rebound where non-black women had a higher odds of viral load rebound than black women (P = .03; estimated odds ratio [OR] 4.00; 95% CI, 1.13–19.02) (). Overall, 3 out of 23 black women, 9 out of 26 Hispanic women, and 6 out of 14 white/other race/ethnicity women met criteria for viral load rebound at 24 weeks postpartum. In a sensitivity analysis, when race/ethnicity was categorized as Hispanic versus non-Hispanic, this association was no longer significant (non-Hispanic vs Hispanic [reference]: OR 1.64; 95% CI, 0.54–5.0; P = .38). This observed association between race/ethnicity with virologic rebound may be confounded by the subjects’ clinical research site. While 13 of the 17 sites enrolled between 1 and 3 black women, 2 sites enrolled 73.1% (19) of the 26 Hispanic women. At these 2 sites, only 2 out of 19 Hispanic women with HIV-1 RNA data at week 24 met criteria for viral load rebound. Women not enrolled at these 2 sites were 5.85 times more likely to develop postpartum viral load rebound than those enrolled at these 2 sites (95% CI, 1.43–39.81, post hoc analysis). Additionally 16/23 (70%) of postpartum enrollees from these 2 sites remained on stable ARVs through week 24 versus 28/61 (46%) from other sites (P = .08, Fisher exact test). When both race/ethnicity (non-black vs black) and HIV RNA at postpartum week 6 (n = 58) were included in 1 model, estimates remained similar but race/ethnicity was no longer significant albeit still showing a trend for an association (non-black vs black [reference] race/ethnicity: OR 3.85; 95% CI, 0.99–19.66; P = .07; >50 vs ≤50 [reference] at week 6 postpartum visit: OR 4.07; 95% CI, 1.21–14.80; P = .03).
Logistic regression results for week 24 viral load rebound for women with week 24 viral load data
Although women who entered the study planned to continue ARV therapy after delivery and all women were encouraged to avoid ARV changes or discontinuations through 24 weeks postpartum, a greater number of women changed or discontinued ARV therapy than expected. Only 44 (69%) of the 63 women with a viral load measurement at 24 weeks postpartum remained on stable ARVs during this time period. Remaining on stable therapy was associated with being black non-Hispanic versus other (OR 4.44; 95% CI, 1.26–21.09; P = .02), older age (per 10 years older: OR 0.33; 95% CI, 0.12–0.84; P = .02), and better third trimester adherence (100% vs <100%: OR 6.85; 95% CI, 2.04–25.34; P < .01).
Eight of the 44 women on stable ARVs through week 24 postpartum (18.2%; 95% CI, 8.2–32.7) met criteria for viral load rebound (). Those with a postpartum week 6 HIV-1 RNA >50 copies/mL (vs ≤50 copies/mL) had significantly higher odds of rebound at week 24 postpartum (estimated OR 16.5; 95% CI, 1.64–861.1; P = .01; ). Six of 14 women with postpartum week 6 HIV-1 RNA value >50 copies/mL had viral load rebound at week 24 postpartum compared with 1 of 25 with postpartum week 6 HIV-1 RNA ≤50 copies/mL. As a group, only 28 of 44 (63.6%) women on stable therapy had a viral load of ≤50 copies/mL at 24 weeks postpartum. None of the other 18 covariates analyzed univariately showed a significant association with week 24 postpartum viral load rebound while on stable ARV (P > .08). Multivariable analyses were not performed due to the small event number.
Ten of the 19 (52.6%; 95% CI, 28.9–75.5) women who did not remain on a stable ARV therapy regimen after entry for postpartum follow-up had viral load rebound at week 24 postpartum (). All 10 rebounders had discontinued their ARV therapy: 4 per their request at 4.7, 13.3, 14.9, and 15.9 weeks postpartum, 3 per physician request at 14, 14.4, and 18 weeks postpartum, and 1 each for social issues, hepatotoxicity to zidovudine/lamivudine and nelfinavir, and nausea due to subsequent pregnancy. In general, the participant and physician decisions to stop medication seemed dependent on a stable CD4 and difficulty with adherence to the medication regimen and/or clinical follow-up.
We were able to successfully perform genotypic resistance testing on 15 of the 18 rebounders. Two women’s samples were not amplifiable (HIV-1 RNA 2,550 copies/mL and 4,162 copies/mL) and one woman’s sample was lost (). Of the 7 rebounders with genotypic resistance data on stable ARVs at the time of rebound, 3 had a genotypic susceptibility score of ≤1 compared with none of the 8 rebounders on unstable ARVs. In total, 8/15 rebounders had a genotypic susceptibility score of <3. The median time from discontinuing ARVs to the week 24 postpartum genotypic resistance assay sample date for the 8 rebounders on unstable ARVs was 9.7 weeks (range, 2.5 to 20.9).
For the secondary endpoint of cumulative viral load rebound, 27 of the 84 women who entered postpartum follow-up had viral load rebound between postpartum week 2 and 24 (estimated probability of remaining free of viral load rebound, 63.4%; 95% CI, 51.1–73.4). Nine women who had viral load rebound at 2, 6, or 12 weeks postpartum were not counted as viral load rebounders in the cross-sectional week 24 postpartum primary endpoint because 5 ended the study follow-up prematurely and 4 had a decline in viral load to <200 copies/mL (only 1 to <50 copies/mL) by week 24 postpartum. At week 6 postpartum, 10 of 74 (13.5%; 95% CI, 6.7–23.5) women with data had viral load rebound; at week 12 postpartum, 8 of 65 (12.3%; 95% CI, 5.5–22.8) with data had viral load rebound.
Among the 64 women with baseline and week 24 postpartum CD4 cell count data, the median (minimum, maximum) increase in CD4 cell count was +84 cells/mm3 (range, −304 to 915) and the median increase in CD4% was +1% (range, −13.5 to 25.5). For the 44 women who were on stable HAART at week 24 postpartum, the median changes in CD4 and CD4% were +103.5 (range, −304 to 915) cells/mm3 and +1.5 (range, −8 to 25.5), respectively. For the 19 women who were not on stable HAART at week 24 postpartum, the median changes in CD4 and CD4% were +49.5 (range, −276 to 330) cells/mm3 and +3 (range, −13.5 to 18), respectively.
In cross-sectional analysis, the proportions of women with data who reported “no missed doses in the last 4 days” and “never skip medications” were lower at weeks 6 and 12 postpartum compared with the third trimester (). The sample size reporting adherence measures decreased from 83 to 53 women at 24 weeks postpartum. There was no obvious difference in week 24 postpartum adherence reported by women who were on their initial HAART regimen versus those on a subsequent regimen during the pregnancy; 30/38 (79%) on their initial HAART regimen reported no missed doses within the past 4 days versus 12/15 (80%) on a subsequent regimen. Week 6 postpartum adherence, as measured by 100% adherence versus < 100% adherence or by never missed ARVs versus missed ARVs, was not significantly associated with week 24 postpartum viral load rebound (). We found no significant evidence of the median change in adherence being different from zero for the continuous outcome of change in 4-day recall adherence when considering change from third trimester to week 6 postpartum, week 24 postpartum, or change from week 6 to week 24 postpartum (signed rank test, P ≥ .12). We did find that women who reported that they had missed taking their ARVs at least once during the third trimester had significantly increased odds of missing their postpartum week 24 visit (ever vs never skip [reference]: estimated OR 2.92; 95% CI, 1.03–9.15; P = .04; 20 events).
Logistic regression results for week 24 viral load rebound for women with week 24 viral load data on stable ARV therapy at week 24
Infant HIV-1 infection status was collected retrospectively but could not be obtained for all children. For the 84 women who registered to postpartum follow-up, 73 infants were known to be HIV uninfected and 11 had unknown status.