Is Intrapartum Intravenous Zidovudine for Prevention of Mother-to-Child HIV-1 Transmission Still Useful in the Combination Antiretroviral Therapy Era?
Clinical Infectious Diseases Sept 15 2013
Nelly Briand,1,2 Josiane Warszawski,1,2,3 Laurent Mandelbrot,1,4,5 Catherine Dollfus,6 Emmanuelle Pannier,7 Ludovic Cravello,8 Rose Nguyen,9 Isabelle Matheron,10 Norbert Winer,11 Roland Tubiana,12,13 Christine Rouzioux,14,15 Albert Faye,4,16 and Stephane Blanche;14,17 for the ANRS-EPF CO1-CO11 Study Groupa
1INSERM, Centre for Research in Epidemiology and Population Health, U1018, Equipe VIH et IST, 2Universite Paris-Sud, UMRS 1018, and 3AP-HP, Hopital Bicetre, Service d'Epidemiologie et Sante Publique, Le Kremlin-Bicetre; 4Universite Paris Diderot, Paris; 5AP-HP, Hopital Louis Mourier, HUPNVS, Service de Gynecologie et d'obstetrique, Colombes; 6AP-HP, Hopital Trousseau, Service d'Hematologie et d'oncologie pediatrique, and 7AP-HP, Hopital Cochin-Port Royal, Paris; 8Hopital de la Conception, Service de Gynecologie et d'obstetrique, Marseille; 9Centre Hospitalier Sud Francilien, Service de Gynecologie et d'obstetrique, Evry Corbeil; 10Centre Hospitalier Intercommunal, Service de Gynecologie et d'obstetrique, Villeneuve Saint Georges; 11CHU Nantes, Service de Gynecologie et d'obstetrique, Nantes; 12AP-HP, Hopital Pitie Salpetriere, Service de maladies infectieuses, 13INSERM, U943, 14EA 3620, Universite Paris Descartes 5, 15AP-HP, Hopital Necker, Service de Virologie, 16AP-HP, Hopital Robert Debre, Service de Pediatrie Generale, and 17AP-HP, Hopital Necker, Unite d'Immunologie Hematologie Pediatrique, Paris, France
"The main finding of our study is that intrapartum intravenous ZDV was not significantly associated with a lower risk of transmission for the vast majority of women with well-controlled viral load at delivery, whereas it remains, in the cART era, an effective measure to reduce transmission in cases of maternal virological failure. According to French guidelines before 2005, ZDV or dual NRTI could be prescribed as PMTCT prophylaxis to women with no indication of ART for themselves. But our results remained unchanged when restricted to women who received cART during pregnancy......
.......Interestingly, a missed opportunity of intravenous ZDV could be compensated by an intensified postnatal prophylaxis. In cases of virological failure (viral load at delivery ≥1000 copies/mL) when the mother did not receive intravenous ZDV, the MTCT rate was 10.2% in infants who received postnatal monotherapy exclusively and 4.8% in infants who received any postnatal intensification (9.8% and 6.1%, respectively, in the subgroup of term births). The postnatal intensified prophylaxes used were dual nucleoside therapy (ZDV plus lamivudine) or PI-based cART."
"Our findings support recent evolutions in the French and US guidelines, which no longer recommend systematic intravenous ZDV during labor and delivery in women with a low plasma viral load, taking into account obstetrical risk factors. Intravenous ZDV does remain an effective prophylactic measure to prevent MTCT in women with a high or unknown HIV load near delivery."
Background. Intrapartum intravenous zidovudine (ZDV) prophylaxis is a long-standing component of prevention of mother-to-child transmission (MTCT) of human immunodeficiency virus (HIV) in high-resource countries. In some recent guidelines, intravenous ZDV is no longer systematically recommended for mothers receiving combination antiretroviral therapy (cART) with low viral load. We evaluated the impact of intravenous ZDV according to viral load and obstetrical conditions.
Methods. All HIV-1-infected women delivering between 1 January 1997 and 31 December 2010 in the French Perinatal Cohort (ANRS-EPF) were analyzed if they received ART during pregnancy and did not breastfeed. We identified maternal and obstetrical characteristics related to lack of intravenous ZDV and compared its association with MTCT rate and other infant parameters, according to various risk factors.
Results. Intravenous ZDV was used in 95.2% of the 11 538 deliveries. Older age, multiparity, and preterm and vaginal delivery were associated with lack of intravenous ZDV (n = 554). In women who delivered with viral load ≥1000 copies/mL, the overall MTCT rate was higher without than with intravenous ZDV (7.5% vs 2.9%; P = .01); however, there was no such difference when the neonate received postnatal intensification therapy. Among them, 77% of women who had viral load <400 copies/mL, there was no difference in MTCT rate (0% without intravenous ZDV vs 0.6% with intravenous ZDV; P = .17). Intravenous ZDV was not associated with increased short-term hematological toxicity or lactate level.
Conclusions. Intravenous ZDV remains an effective tool to reduce transmission in cases of virological failure, even in cART-treated women. However, for the vast majority of women with low viral loads at delivery, in the absence of obstetrical risk factors, systematic intravenous ZDV appears to be unnecessary.
A remarkable decrease in mother-to-child transmission (MTCT) of human immunodeficiency virus type 1 (HIV-1) has been obtained in industrialized countries, with the widespread use of antiretroviral therapy (ART) reducing transmission rates to about 1% [1-6]. The relative contributions of ART during the pregnancy, intrapartum prophylaxis, and postexposure prophylaxis are somewhat difficult to determine, as all 3 components are usually implemented. The timing of MTCT, in studies of performed before ART was available, was in the majority of cases during labor and delivery . Investigators designing the US-French pivotal trial ACTG076-ANRS024  thus proposed administering intrapartum intravenous zidovudine (ZDV) during labor or cesarean delivery, as a continuity between the pre- and postnatal prophylaxis, as well as an "intensification" of the ZDV monotherapy during this crucial phase .
Intravenous ZDV was then adopted by all guidelines in high-resource countries. With the advent of combination antiretroviral therapy (cART), the benefit of intravenous ZDV has been questioned. In a previous report from the French Perinatal Cohort (ANRS-EPF CO1-CO11) , performed on data collected between 1997 and 2004, we observed that intravenous ZDV was not associated with transmission rate in women who received any ART during pregnancy and had an HIV RNA <400 copies/mL near delivery. In contrast, it was strongly associated with a lower risk of transmission in women whose delivery HIV RNA was >10 000 copies/mL. Following this report, recent guidelines suggested that intravenous ZDV could be avoided for HIV-infected women receiving cART with optimal control of HIV-1 RNA near delivery, not defined in the French guidelines in 2010  and defined as <400 copies/mL in the US guidelines updated in July 2012 . We aim here to evaluate the contribution of intravenous ZDV in the cART era.
The ANRS French Perinatal Cohort
Since 1985, the ANRS French Perinatal Cohort (ANRS CO1 and CO11 EPF) has prospectively enrolled HIV-infected women giving birth at 90 centers throughout France. The methodology has been described elsewhere . Informed consent was obtained from all mothers. The EPF study was approved by the institutional review board of Cochin Hospital and the French computer database watchdog (Commission Nationale de l'Informatique et des Libertes).
No specific recommendation for HIV treatment and obstetric care was made for women included in the cohort. French national guidelines for prevention of MTCT (PMTCT) are published and updated regularly . Antenatal prophylaxis was based on ZDV monotherapy since 1994 and on dual nucleoside analog therapy since 1997, with elective cesarean delivery according to risk/benefit evaluation. Combination of antiretroviral therapies, initially recommended to pregnant women who had indications for themselves, was extended to all women during pregnancy in 2004. Elective cesarean delivery for the purpose of PMTCT was restricted since 2002 to women not treated by cART whose viral load was >400 copies/mL near delivery. Up to July 2010, intrapartum intravenous ZDV and neonatal prophylaxis were systematically recommended.
HIV-1-infected women delivering between 1 January 1997 and 31 December 2010 in the French national ANRS Perinatal Cohort were included in this analysis if they received ART during pregnancy and did not breastfeed; women with multiple pregnancies were excluded (n = 11 538) (Figure 1).
Analyses of intravenous ZDV tolerance for infants were restricted to HIV-uninfected newborns, that is, after exclusion of infected children diagnosed either by HIV-1 polymerase chain reaction done on sites on 2 separate samples or with anti-HIV-1 antibodies detected by enzyme-linked immunosorbent assay and Western blot at 18 months of age.
Sociodemographic, obstetric, and HIV treatment data recorded included age, geographic origin (categorized as sub-Saharan African countries, metropolitan France, or other origins), gestational age at booking in the maternity ward, type of ART and time of treatment initiation and cessation, gestational age, and mode of delivery (vaginal, emergency cesarean delivery or elective cesarean delivery) . For this analysis, we used the CD4 cell counts and percentages and plasma HIV-1 RNA obtained closest to the time of delivery, not more than 7 days after delivery. Gestational age was determined from the date of the last menstrual period, corrected when appropriate by the date of conception as evaluated from the early ultrasound. Baseline characteristics (including pretherapeutic immunovirological status) were the first data collected during pregnancy, even if they were available prior to the booking visit in the maternity.
Maternal ART was categorized into 7 groups according to the last treatment used during the pregnancy, as follows: nucleoside reverse transcriptase inhibitor (NRTI) monotherapy; double NRTI therapy; 3 or more NRTIs; protease inhibitor (PI)-based cART (PI with or without NRTI , or PI plus nonnucleoside reverse transcriptase inhibitor [NNRTI] with or without NRTI); non-PI-based cART (NNRTI with or without NRTI); and other cART. Neonatal prophylaxis was categorized as NRTI monotherapy, or intensification with double NRTI therapy (usually ZDV plus lamivudine) or triple therapy.
In HIV-uninfected infants, hemoglobin (g/dL), percentage of neutrophil, and lactate concentrations (mmol/L) were measured at 1, 3, and 6 months, and then every 6 months until 24 months of age. Grades of severity were defined using established grades and thresholds .
Associations between maternal, obstetrical, and immunovirological characteristics and no intravenous ZDV was studied using χ2 or Fisher exact tests to compare percentages, and Student t tests or Wilcoxon rank-sum tests to compare continuous variables in univariate analysis. The propensity score of receiving intravenous ZDV was calculated after searching for predictive factors for treatment attribution, which were generated with a logistic regression [13-15].
The MTCT rate was compared between women who received intravenous ZDV and those who did not. The analysis was conducted separately according to the level of viral load at delivery (<400, 400-999, or ≥1000 copies/mL), and stratified on the mode of delivery, term at delivery (<37 or ≥37 gestational weeks), and type of postnatal prophylaxis. Independent association of MTCT with intravenous ZDV was assessed using multivariate logistic regression, adjusted for other noncollinear risk factors of MTCT. This analysis was performed only among women with viral loads ≥1000 copies/mL at or close to delivery because there was no case of transmission in women who presented with lower viral loads and did not receive intravenous ZDV.
Hematological parameters and lactate concentrations at 1, 3, and 6 months of age were also compared between the 2 groups of women but restricted to the HIV-uninfected infants. Multivariate linear regression was used to study the association adjusted for the propensity score of receiving intravenous ZDV using restricted cubic spline functions, postnatal prophylaxis, and birth level of parameters, except for lactates not measured at birth.
Analyses were conducted with SAS statistical software (version 9.3; SAS Institute, Cary, North Carolina) and the %RCS_Reg SAS macro . Statistical significance was set at P < .05.
Description of Study Population
Among the 11 538 women included between 1997 and 2010, 95.2% (n = 10 984) received intravenous ZDV, and 554 did not (Figure 1). The last ART regimen prescribed during pregnancy was PI-based cART in 60.9% of the women, non-PI-based cART in 7.8%, NRTI multitherapy in 2.6%, NRTI bitherapy in 17.5%, NRTI (ZDV) monotherapy in 10.2%, and integrase inhibitor or CCR5-based multitherapy in 0.9% of women (Table 1). At delivery, the median CD4 cell count was 461 cells/mL (interquartile range [IQR], 319-636 cells/mL), and 43.2% of women had >500 cells/mL. Maternal HIV-1 RNA at delivery was <400 copies/mL for 77.7% of women. Almost 79% of neonates received monotherapy as neonatal prophylaxis and 21% received an intensified prophylactic treatment.
Characteristics of Women Who Did Not Receive Intrapartum Intravenous ZDV
The percentage of women who did not receive intravenous ZDV decreased from 7.0% in 1997-1998 to about 4% since 2001 (Table 1).
Women who did not receive intravenous ZDV, compared with women who received intravenous ZDV, were older, more often multipara, initiated ART earlier during pregnancy, delivered more often preterm and/or vaginally, and had a higher HIV RNA level at delivery (Table 1).
After adjustment for these factors and for CD4 cell count at delivery, year of delivery, maternal origin, and last ART regimen, the absence of intravenous ZDV remained more frequent in women aged >35 years of age, multipara, delivering preterm and vaginally, and with high HIV RNA at delivery (Table 1).
Neonates whose mothers did not receive intravenous ZDV more often received postnatal intensification treatment than those whose mothers received intravenous ZDV (35% vs 20%, P < .001).
Association Between Intravenous ZDV and MTCT Transmission Rate
The overall MTCT rate was 0.9% (95/10 239) with intravenous ZDV and 1.8% (9/514) without intravenous ZDV (P = .06), but the pattern was different according to the level of viral load near delivery and to the type of postnatal prophylaxis (Table 2 and Figure 2).
In women who delivered with viral loads ≥1000 copies/mL, there was a higher risk of MTCT associated with the lack of intravenous ZDV, but only in neonates who received postnatal prophylaxis with ZDV monotherapy: 10.2% vs 2.5% in the presence of intravenous ZDV (P < .01). In neonates who received intensified postnatal prophylaxis, there was no difference in transmission rate (4.8% vs 4.1%, respectively; P = .83). In particular, for women who term delivered by elective cesarean delivery, the transmission rate remained higher without than with intravenous ZDV (9.5% vs 1.8%; P = .01). In multivariate analysis, the association between MTCT and lack of intravenous ZDV remained significant after adjustment for known risk factors of MTCT (adjusted odds ratio, 3.51 [95% confidence interval, 1.29-9.54]; P = .01; Table 3).
Among women who delivered with viral load <1000 copies/mL, there was no case of HIV transmission in those who did not receive intravenous ZDV 0% (0/369) vs 0.6% (47/8132) in the presence of intravenous ZDV, even in women with viral load <400 copies/mL (Table 2).
Similar results were observed among term births (Table 2), after including multiple pregnancies, or in the subgroups of women who received cART (data not shown).
Hematological Parameters and Lactate Concentrations in HIV-Uninfected Infants
Mean hemoglobin level at 1 month of age was lower in infants exposed to intravenous ZDV compared with unexposed infants, but this association was no longer significant after adjustment for propensity score of receiving intrapartum intravenous ZDV, type of neonatal prophylaxis, and value of hemoglobin at birth in multivariate linear regression (Table 4). The rates of grade 3/4 anemia and neutropenia did not differ significantly at 1 month between the 2 groups (4.4% and 3.9% in infants exposed to intravenous ZDV vs 5.3% and 3.3% in unexposed infants; P = .43 and P = .62, respectively) and at 3 months (2.8% and 3.7% of anemia and neutropenia in exposed infants vs 3.8% and 4.2% in unexposed infants; P = .27 and P = .63; data not shown). Moreover, exposed and unexposed children did not differ for lactate concentrations. Similar levels were observed for all these parameters at 6 months of age.
The main finding of our study is that intrapartum intravenous ZDV was not significantly associated with a lower risk of transmission for the vast majority of women with well-controlled viral load at delivery, whereas it remains, in the cART era, an effective measure to reduce transmission in cases of maternal virological failure. According to French guidelines before 2005, ZDV or dual NRTI could be prescribed as PMTCT prophylaxis to women with no indication of ART for themselves. But our results remained unchanged when restricted to women who received cART during pregnancy.
Interestingly, a missed opportunity of intravenous ZDV could be compensated by an intensified postnatal prophylaxis. In cases of virological failure (viral load at delivery ≥1000 copies/mL) when the mother did not receive intravenous ZDV, the MTCT rate was 10.2% in infants who received postnatal monotherapy exclusively and 4.8% in infants who received any postnatal intensification (9.8% and 6.1%, respectively, in the subgroup of term births). The postnatal intensified prophylaxes used were dual nucleoside therapy (ZDV plus lamivudine) or PI-based cART.
The main strength of this study is the large number of patients included in the French nationwide prospective cohort. Its main weakness is the fact that the proportion of HIV-infected women who did not receive intravenous ZDV was relatively small (5%) and subject to treatment biases inherent to observational studies. Failure to receive intravenous ZDV was more frequent in older, multiparous women, in women with premature and vaginal deliveries, and in women who had higher HIV-1 RNA at delivery. It can be hypothesized that many of these women presented in advanced labor, too late to start intravenous ZDV .
Prematurity was shown to be a risk factor for MTCT of HIV [5, 18]. In the absence of ART or only with ZDV monotherapy, vaginal delivery was also shown to be associated with a higher rate of MTCT compared to elective cesarean delivery [19-27]. Nonetheless, even among women with viral loads ≥1000 copies/mL who had elective cesarean delivery at term, the lack of intravenous ZDV was associated with a significant 5-fold higher risk of MTCT.
When considering the women who delivered with well-controlled viral load (<400 copies/mL), no case of transmission occurred among those who did not receive intravenous ZDV. However, the 95% confidence interval upper limit of the transmission rate in women without intravenous ZDV was 1.1%, higher than the upper limit for the transmission rate among women with intravenous ZDV (0.6%; 95% confidence interval, 0.4% -0.8%). The only definitive proof would be provided by a randomized trial, but given the low transmission rate, this would require approximately 10 000 women to be able to test equivalence of risk, a figure clearly unrealistic. Moreover, despite the large size of our cohort, there were few infected infants; thus, the study lacked power to investigate the safety of withholding intravenous ZDV in case of obstetrical risk factors, such as premature rupture of membranes, preterm delivery, fever, or bleeding. While awaiting more data, it may appear reasonable to maintain intravenous ZDV even in women with low viral load in cases of complicated deliveries.
These results are consistent with our previous study, conducted between 1997 and 2004 in the French Perinatal cohort on a smaller sample size and with a lower proportion of women treated with cART during pregnancy . Extending the study to 2010 allowed us to evaluate, with an increased power, whether intravenous ZDV was still beneficial in the context of systematic cART recommended since 2004 for all pregnant women. In the current study, 72% of women received combination antiretroviral therapies, mostly PI-based, the proportion increasing from 38% in 1997-2004 to 86% in 2005-2010. A recent European study reported no significant association between intravenous ZDV and transmission rate (3.1% with vs 12.5% without; P = .23) , which was also the case when we performed the analysis on the overall study population. However, the authors did not report results separately for low and high viral load at delivery.
Toxicity outcomes were evaluated in the first 6 months of life for uninfected infants, according to the use of intravenous ZDV. The parameters studied were hemoglobin level, neutrophil percentage, and lactate concentrations because these were previously described to be influenced by ZDV exposure [29-32]. We used multivariate analysis adjusted for a propensity score of receiving intravenous ZDV. A propensity score is a conditional probability of being treated given a subject's observational characteristics, an approach designed to detect and remove the selection bias and estimation risk due to lack of randomization. Our results suggest that intrapartum intravenous ZDV did not significantly modify short-term hematological outcomes. As in the ANRS French Perinatal Cohort, the duration of intravenous ZDV was not recorded, so we could not take this parameter into account in our toxicity analyses. Hematological parameters did not differ in the subgroups of women who delivered by elective cesarean section, in which duration of intravenous ZDV, when administered, was likely to be more homogenous.
Our findings support recent evolutions in the French and US guidelines, which no longer recommend systematic intravenous ZDV during labor and delivery in women with a low plasma viral load, taking into account obstetrical risk factors. Intravenous ZDV does remain an effective prophylactic measure to prevent MTCT in women with a high or unknown HIV load near delivery.