| |
ART Interruption in Pregnancy Increases MTCT
|
| |
| |
"in our cohort, temporary ART discontinuation early in pregnancy is more frequent in women receiving HAART, probably because of the fear of drug toxicity in the embryo. However, temporary discontinuation led to a 10-fold increase in the rate of MTCT, overcoming all other risk factors, except for the independent factor of high plasma viral RNA load at delivery.. When viral load increased by 1 log10 copies/mL, the risk of MTCT was, indeed, more than doubled."..."counseling on the temporary discontinuation of ART in the first trimester should consider both the need for maternal health [7] and the increased risk of HIV-1 transmission to the offspring"...."Overall, the rate of MTCT in the whole cohort was 1.3% (95% CI, 0.7%-2.3%). The rate of MTCT among children born to mothers who had ART interrupted in the first trimester was 4.9% (95% CI, 1.9%-13.2%), and the rate of MTCT among children born to mothers who had ART interrupted in the third trimester was 18.2% (95% CI, 4.5%-72.7%)."
Is the Interruption of Antiretroviral Treatment During Pregnancy an Additional Major Risk Factor for Mother-to-Child Transmission of HIV Type 1?
Clinical Infectious Diseases 2009 Epub March 24 2009
Luisa Galli,1 Donella Puliti,2 Elena Chiappini,1Clara Gabiano,3 Gabriele Ferraris,4Federica Mignone,3Alessandra Vigano,5Carlo Giaquinto, 7Orazio Genovese,8 Gianfranco Anzidei, 9Raffaele Badolato,11 Wilma Buffolano,12 Anna Maccabruni,13 Filippo Salvini,6 Monica Cellini,14 Maurizio Ruggeri,15 Mariano Manzionna,16 Stefania Bernardi,10 Pierangelo Tovo,3 and Maurizio de Martino,1 for the Italian Register for HIV Infection in Children a
1Department of Pediatrics, University of Florence, and 2Clinical and Descriptive Epidemiology Unit, Centre for the Study and Prevention of Tumours, Research Institute of the Tuscany Region, Florence, 3Department of Pediatrics, University of Turin, Turin, 4Division of Neonatology, Mangiagalli Hospital, 5Division of Pediatrics, University of Milan, Sacco Hospital, and 6Division of Pediatrics, University of Milan, S. Paolo Hospital, Milan, 7Department of Pediatrics, University of Padua, Padua, 8Department of Pediatrics, Gemelli Hospital, 9Division of Pediatrics, Spallanzani Hospital, and 10AIDS Unit, Department of Pediatrics, Bambino Ges Children's Hospital, Rome, 11Department of Pediatrics, University of Brescia, Brescia, 12Department of Pediatrics, Federico II University, Naples, 13Department of Infectious Disease, University of Pavia, Pavia, 14Department of Mother and Child, University of Modena, Modena, 15Division of Pediatrics, Hospital of Bergamo, Bergamo, and 16Neonatology Unit, Department of Pediatrics, University of Bari, Bari, Italy
ABSTRACT
Background. There is currently an experts' agreement discouraging interruption of antiretroviral treatment (ART) during the first trimester of pregnancy in women infected with human immunodeficiency virus type 1 (HIV-1). However, this recommendation is poorly supported by data. We evaluated the effects of discontinuing ART during pregnancy on the rate of mother-to-child transmission.
Methods. Logistic regression models were performed in a prospective cohort of 937 children who were perinatally exposed to HIV-1 to estimate adjusted odds ratios for confounding factors on mother-to-child transmission, including maternal interruption of ART.
Results. Among 937 pregnant women infected with HIV-1, ART was interrupted in 81 (8.6%) in the first trimester and in 11 (1.2%) in the third trimester. In the first trimester, the median time at suspension of ART was 6 weeks (interquartile range [IQR], 5-6 weeks) and the time without treatment was 8 weeks (IQR, 7-11 weeks). In the third trimester, the median time at suspension of ART was 32 weeks (IQR, 23-36 weeks) and the time without treatment was 6 weeks (IQR, 2-9 weeks). The plasma viral load was similar in women who had treatment interrupted in the first trimester and in those who did not have treatment interrupted.
Overall, the rate of mother-to-child transmission in the whole cohort was 1.3% (95% confidence interval [CI], 0.7%-2.3%), whereas it was 4.9% (95% CI, 1.9%-13.2%) when ART was interrupted in the first trimester and 18.2% (95% CI, 4.5%-72.7%) when ART was interrupted in the third trimester. In the multiple logistic regression models, only interruption of ART during either the first or the third trimester, maternal mono- or double therapy, delivery by a mode other than elective cesarean delivery, and a viral load at delivery >4.78 log10 copies/mL were independently associated with an increased rate of mother-to-child transmission.
Conclusions. Discontinuing ART during pregnancy increases the rate of mother-to-child transmission of HIV-1, either when ART is stopped in the first trimester and subsequently restarted or when it is interrupted in the third trimester. This finding supports recommendations to continue ART in pregnant women who are already receiving treatment for their health.
After the results of the Pediatric AIDS Clinical Trials Group Protocol 076 Study Group [1], several prospective studies have confirmed the effectiveness of antiretroviral treatment (ART) during pregnancy in decreasing the rate of mother-to-child transmission (MTCT) of HIV-1 [2-5]. In developed countries, the synergic effect of ART in the mother and newborn, cesarean delivery, and avoidance of breast-feeding have resulted in a rate of MTCT of <2% [2-5]. In the HAART era, the rate of MTCT is about 1%, probably because of the strong effect of HAART on the maternal viral load [6]. Therefore, there is a consensus on the benefits of HAART for the mother's health and for prevention of MTCT, provided that drugs with a potential toxicity for the offspring are avoided [7]. Data from prospective registries [8] support lack of teratogenicity with exposure to antiretroviral drugs during pregnancy, but with the increasing use of new molecules, monitoring of birth defects is continually needed.. An increasing number of treated women are now identified as pregnant, and presently there is an experts' opinion discouraging a temporary discontinuation of ART during the first trimester [7]. However, this opinion is only partially supported by evidence. To date, insufficient data support an increase in the rate of MTCT as a consequence of ART discontinuation during pregnancy. We performed an analysis involving a large cohort of children prospectively enrolled in the Italian Register for HIV Infection in Children, to evaluate the effect of discontinuation of ART during pregnancy on the rate of MTCT.
Discussion
To our knowledge, this is the first study to reveal that discontinuing ART during pregnancy increases the rate of MTCT of HIV-1, either when ART is stopped in the first trimester and subsequently restarted, or when it is interrupted in the third trimester. The last finding was partly expected, because discontinuation of treatment could lead to a viremic rebound [12, 13] and because a higher viral load near the time of delivery increases the risk of MTCT, as we and others have clearly shown [14, 15]. The incremental risk of MTCT associated with treatment interruption is high among all treated women, even among HAART-treated women, maybe because the viremic rebound is stronger if the virological potency of therapy is higher. Therefore, our findings confirm the recommendation that pregnant women infected with HIV-1 who receive ART should continue treatment after the first trimester, particularly at the end of pregnancy [7]. More interestingly, we found that interruption of ART early in pregnancy increased the rate of MTCT both in the whole cohort and among children born to HAART-treated women. To our knowledge, no other large study has investigated this issue. Of note, in our cohort, temporary ART discontinuation early in pregnancy is more frequent in women receiving HAART, probably because of the fear of drug toxicity in the embryo. However, temporary discontinuation led to a 10-fold increase in the rate of MTCT, overcoming all other risk factors, except for the independent factor of high plasma viral RNA load at delivery. When viral load increased by 1 log10 copies/mL, the risk of MTCT was, indeed, more than doubled. A small proportion of HAART-treated women may have high plasma viral loads because of poor compliance or mutations associated with viral resistance [16]. Nevertheless, we found no difference in predelivery plasma viral load between mothers who temporarily discontinued therapy and mothers who did not. It might be speculated that a transient viral load increase, occurring close to the time of ART interruption, favors HIV-1 transmission.
We also investigated whether a lower antiretroviral potency of the regimen administered after suspension could lead to higher risk of MTCT, but the proportions of nevirapine-based or PI-based regimens were similar before and after interruption. However, it is possible that virological failure occurred more frequently because of nevirapine-resistance mutations [17] in those women who discontinued and restarted the same nevirapine-based regimen. Viral load at delivery was similar in the women treated with a PI-based regimen and in those treated with a nonnucleoside reverse-transcriptase inhibitor-based regimen in both the whole cohort and in those who had therapy interrupted. In addition, we found a trend toward higher viral loads in women treated with a PI-based regimen than in those treated with an nonnucleoside reverse-transcriptase inhibitor-based regimen, as recently was reported by the European Collaborative Study [18]. It is possible that women with lower baseline viral loads (before or at the beginning of pregnancy) are more often treated with nonnucleoside reverse-transcriptase inhibitor-based regimens, or the pharmacokinetic peculiarities of ART in pregnancy may be involved [18].
Because our data are from a large pediatric cohort, a limitation in our analyses is that some detailed information on the mothers was lacking. First of all, maternal viral load at delivery was unknown for 30% of mother-child pairs. However, when mother-child pairs were compared according to whether maternal viral load was known or unknown, exactly the same MTCT rate was found. As an additional check, 3 different logistic regression models were performed, but the main role of ART interruption in pregnancy was constantly confirmed. Another critical point is the unavailability in our database of information on the reasons for ART discontinuation during pregnancy. Therapy could be discontinued for several reasons: the development of virological failure or resistance mutations, the occurrence of adverse effects, or the desire to avoid fetal exposure during the first trimester. On the other hand, ART interruption was not homogeneously distributed during the whole pregnancy, peaking at 6 and 32 weeks. We think that avoidance of fetal exposure during the first trimester and intolerance of treatment are the most probable reasons for ART discontinuation, and if so, it would be unlikely that women who had ART suspended were a priori at higher risk of MTCT because of virological failure or resistance mutations. Unfortunately, we had no data on maternal viral load during gestation. Moreover, we could not assess whether infected children born to mothers who had ART interrupted acquired the infection in utero or perinatally [19], because we do not systematically measure HIV-1 RNA or DNA load during the first 24-48 h of life.
We investigated all interactions between variables that we thought were biologically convincing. Other important factors might be compliance to ART and genotype resistance in the mothers, but these data were lacking in our database. Finally, the very low number of infected children in the whole cohort and, consequently, the wide 95% CIs found, are additional caveats. Additional studies, prospectively performed using large cohorts of mother-child pairs starting from the beginning of gestation, may confirm our findings. It may be surprising that we found no relationship between intrapartum and neonatal therapy and the rate of MTCT, but the numbers of mother-child pairs who were not receiving intrapartum and neonatal prophylaxis were too small to draw any conclusion. Moreover, we found no differences according to trimester at start of ART, even if a higher proportion of HAART-treated women had a detectable viral load at delivery when treatment was used during the third trimester, compared with the first or second trimester. This finding supports recently revised guidelines [7] that recommend that, in women who do not require ART for their own health, ART for MTCT prophylaxis may be started during the second trimester, preferably with a 3-drug regimen..
The main finding of our study is that no other factors but the discontinuation of ART and high maternal viral load at delivery are associated with an increased risk of MTCT. Thus, although concerns about potential toxicity in the fetus need to be completely clarified, counseling on the temporary discontinuation of ART in the first trimester should consider both the need for maternal health [7] and the increased risk of HIV-1 transmission to the offspring. We believe that our findings may be useful for physicians who care for women infected with HIV-1 and for specialists who determine guidelines.
Results
Mother-infant pairs.
Overall, 937 of 1016 mother-child pairs receiving ART entered the study; 79 (7.8%) of the exposed infants were excluded because they were lost to follow-up before ascertainment of infectious status. The characteristics of the 937 mothers and children are shown in table 1. Intrapartum and neonatal prophylaxis consisted of zidovudine in all but 8 cases in which nevirapine was given. Maternal plasma viral load was known in 630 (67.2%) of the 937 mother-child pairs. Of the 630 women with known viral load, the majority (453 [71..9%]) showed an undetectable plasma viral load (tested by an assay with a cutoff value of 50 copies/mL [n=340] and 400 copies/mL [n=113]). Characteristics of mother-child pairs were compared according to known or unknown maternal plasma viral load. Frequency of intrapartum ART administration was higher when maternal viral load was known (586 [93.0% ] of 630 women vs. 261 [85.0%] of 307 women; χ2 = 15.21.5; p<.001), and frequency of prepartum HAART administration was higher when maternal viral load was known (454 [72.1%] of 630 women vs. 197 [64.2%] of 307 women; χ2 = 6.07; p=.014); no difference was observed in the frequency of mode of elective cesarean delivery (580 [92.1%] of 630 women vs. 280 [91.2%] of 307 women; χ2 = 0.20; p=.653), the use of neonatal ART (614 [97.5%] of 630 infants vs. 301 [98.0%] of 307 infants; χ2 = 0.31; p=.579), and the infant's infectious status (8 [1.3%] of 630 cases vs. 4 [1.3%] of 307 cases; χ2 = 0.002; p=.966). The maternal viral load decreased as the complexity of the ART regimen increased. The mean plasma viral load was 2.6 log10 copies/mL (95% CI, 2.2-2.9 log10 copies/mL), 2.4 log10 copies/mL (95% CI, 2.2-2.6 log10 copies/mL), and 1.9 log10 copies/mL (95% CI, 1..8-2.0 log10 copies/mL) in women receiving monotherapy, double therapy, or HAART, respectively (analysis of variance, F=17.2; p<.001). The median gestational age at the beginning of treatment was 18 weeks (IQR, 13-35 weeks), 18 weeks (IQR, 13-25 weeks), and 13 weeks (IQR, 6-22 weeks) in women receiving monotherapy, double therapy, or HAART, respectively, thus decreasing with the increase in the ART complexity (χ2 = 36.8; ). The proportion of women with detectable viral load did not differ according to trimester at start of ART when monotherapy was given (n=50 ; 78.6% during the first trimester, 55.0% during the second trimester, and 56.3% during the third trimester; χ2 = 2.27; p=.321) In addition, the proportion of women with detectable viral load did not differ according to the trimester during which ART was initiated when double therapy was given (n=126 ; 38.1% during the first trimester, 37.3% during the second trimester, and 47.1% during the third trimester; χ2 = 0.56; p=.756). However, the proportion of women with detectable viral load was significantly higher when HAART (given to 454 women) was started during the third trimester (45.9%), compared with during the second (17.5%) or first trimester (17.8%; χ2 = 25.81; p<.001). Women treated with HAART who had a known plasma viral load had similar mean plasma viral loads when 281 women who were treated with a PI-based regimen (1.95 log10 copies/mL; 95% CI, 1.77-2.07 log10 copies/mL) were compared with 328 women treated with a nevirapine-based regimen (1.85 log10 copies/mL; 95% CI, 1.65-1.92 log10 copies/mL; Student's t test, 1.29; p=.197).
Discontinuation and subsequent restart of ART was reported in 81 (8.6%) of the 937 mother-child pairs in the first trimester of pregnancy (median time at suspension, 6 weeks [IQR, 5-6 weeks]; median time without treatment, 8 weeks [IQR, 7-11 weeks]) and in 11 (1.2%) of the mother-child pairs in the third trimester (median time at suspension, 32 weeks [IQR, 23-36 weeks]; median time without treatment, 6 weeks [IQR, 2-9 weeks]). Only 1 mother, who gave birth to an uninfected child, had treatment interrupted during both the first and the third trimesters. ART was more frequently interrupted at the beginning of pregnancy in women treated with HAART (74 [11.4%] of 651 women) than in women treated with double therapy (3 [1.6%] of 187 women) or with monotherapy (4 [4.0%] of 99 women; χ2 test, 20.50; p<.001).
We checked whether women switched therapy after interruption to a less or more active regimen, but the proportions of women treated with a nevirapine- or PI-based HAART regimen before or after treatment interruption were similar (table 1). Treatment was restarted with the same class of antiretroviral drugs in 47 (58.0%) of the 81 women, among whom 19 (23.5%) received the same nevirapine-based regimen that they received prior to the interruption.
Among women whose viral load at delivery was known, the mean plasma viral load was similar in those who had treatment interrupted in the first trimester (2.04 log10 copies/mL; 95% CI, 1.81-2.26 log10 copies/mL) and in those who did not have treatment interrupted (2.04 log10 copies/mL; 95% CI, 1.95-2.14 log10 copies/mL; Student's t test, -0.04; p=.967). Similarly, the mean CD4+ cell count at delivery was similar in women who had treatment interrupted (2.63 log10 cells/μL; 95% CI, 2.57-2.69 log10 cells/μL) and in those who did not have treatment interrupted (2.65 log10 cells/μL; 95% CI, 2.65-2.67 log10 cells/μL; Student's t test, -0.40; p=.689). Among HAART-treated women who had therapy interrupted and who had known plasma viral load, we found no difference in mean plasma viral load between 29 women treated with a PI-based regimen (2.06 log10 copies/mL; 95% CI, 1.70-2.42 log10 copies/mL) and 28 women treated with a nevirapine-based regimen (1.85 log10 copies/mL; 95% CI, 1.69-2.01 log10 copies/mL; Student's t test, 1.12; p=.268).
MTCT rate and confounding factors.
Overall, the rate of MTCT in the whole cohort was 1.3% (95% CI, 0.7%-2.3%). The rate of MTCT among children born to mothers who had ART interrupted in the first trimester was 4.9% (95% CI, 1.9%-13.2%), and the rate of MTCT among children born to mothers who had ART interrupted in the third trimester was 18.2% (95% CI, 4.5%-72.7%). Characteristics of the 12 mother-child pairs in which the children were infected are shown in table 2.
To evaluate the role of different risk factors for MTCT, univariate and multiple logistic regression models were performed. Maternal monotherapy or double therapy, interruption of ART during either the first or the third trimester, delivery by a mode other than elective cesarean delivery, and a viral load at delivery >4.78 log10 copies/mL were associated with an increased rate of MTCT (table 3). Plasma viral load >4.78 log10 copies/mL was associated with a 30-fold increased risk of transmission.. Other factors (maternal CD4+ cell count at delivery, trimester at the start of ART, child's sex, and intrapartum and neonatal ART) were not significantly associated with the rate of MTCT.
When we included in an additional model the 79 exposed infants lost to follow-up, with the assumption that all these children were uninfected, results were unchanged (adjusted OR for treatment interruption in the first trimester, 11.4; in the third trimester, 34.1). The probability of infection in children in this group was very low, because missing infected children are regained by periodical cross-matches with the Italian National AIDS Registry, to which all AIDS cases must be reported [11].
To check the interference of missing values for maternal plasma viral load, 3 logistic regression models were performed. Interruption of ART in the first trimester was associated with an adjusted OR of 10.33 (95% CI, 2.02-52.91; p=.005), and interruption of ART in the third trimester was associated with an adjusted OR of 46.86 (95% CI, 4.28-512.64; p=.002) in the whole cohort, with the inclusion of maternal viral load as a category with missing values. When only cases with known maternal viral load were included, the adjusted OR for interruption of ART in the first trimester was 11.45 (95% CI, 1.58-83.08; =.016) and in the third trimester was 24.33 (95% CI, 0.77-765.92; p=.070). When maternal viral load was excluded, the adjusted OR for interruption of ART in the first trimester was 8.32 (95% CI, 1.88-36.90; p=.005) and in the third trimester was 40.52 (95% CI, 4.24-387.64; p=.001) in the whole cohort.
Finally, a logistic regression model on a subgroup ( ) of mother-child pairs with both known plasma viral load and known CD4+ cell count was performed, including these factors as continuous variables. Treatment interruption in the first trimester was associated with an adjusted OR of 10.70 (95% CI, 1.48-77.18), and treatment interruption in the third trimester was associated with an adjusted OR of 30.38 (95% CI, 0.98-941.66). Plasma viral load was associated with an adjusted OR of 2.50 (95% CI, 1.27-4.91), and CD4+ cell count was associated with an adjusted OR of 1.87 (95% CI, 0.07-53.26). Thus, treatment interruption in the first trimester was associated with a 10-fold increased risk of MTCT, whereas a plasma viral load increase of 1 log10 copies/mL doubled the risk of MTCT.
Methods
Data collection.The Italian Register for HIV Infection in Children is a nationwide multicenter study of children perinatally exposed to HIV-1 that was instituted in 1985. Data come from a network of 106 participating pediatric centers located throughout Italy, representing the overall population of exposed infants in Italy [3]. Informed consent at the local pediatric center and National Privacy Agency permission are obtained.. Both at-risk children identified at birth and infected children identified after birth are enrolled. In this study, only exposed children identified at birth who were born to treated mothers and who had known infectious status were considered [3]. Data with regard to mother-infant pairs were collected as described elsewhere [3, 9]. Detailed data on the type of regimen(s), gestational age at beginning and end of ART (for each single regimen if switches occurred during pregnancy), and ART administered intrapartum and/or to the newborn were included. Maternal data were obtained at each pediatric center from an infectious diseases report and/or a gynecologist's report, and they were recorded at the infant's first examination, at birth. From 2001, information on the maternal plasma viral RNA load and CD4+ cell counts at the time of delivery and information on the last antiretroviral regimen (the type and the date of initiation) prior to pregnancy were collected. Because this information was uniformly available only since 2002, only children born in 2002 or later were evaluated in the present study. Viral loads and CD4+ cell counts were evaluated as described elsewhere [3, 9]. Both baseline and follow-up information on the infant's infection status, HIV-1 antibodies, and viral markers (HIV-1 DNA or RNA detection) were collected [3].
Case definition.
Infection in children [3, 9, 10], elective cesarean delivery [2], and duration of ART during pregnancy [9] were defined as reported. To evaluate the effect of type of ART on the rate of MTCT, we defined monotherapy as zidovudine (according to the Pediatric AIDS Clinical Trials Group Protocol 076 Study Group protocol), double therapy as any regimen including 2 nucleoside reverse-transcriptase inhibitors (NRTI), and HAART as any combination regimen with >3 antiretroviral drugs (no mother received a triple NRTI regimen). Among HAART regimens, a nevirapine-based regimen was defined as any triple regimen including 2 NRTIs plus nevirapine, and a protease inhibitor (PI)-based regimen was defined as any triple (or more) regimen including NRTIs plus a PI. When >1 regimen was received during pregnancy, the more complex (i.e., HAART vs. double) was considered. Maternal interruption of ART was defined as the discontinuation of any ART for >15 days. Shorter periods of interruption, in fact, could not be reported by the mothers. Moreover, it is usually believed that plasma viral RNA load significantly decreases after 2 weeks of treatment, and perinatal guidelines recommend monitoring maternal viral load 2-6 weeks after initiating or changing ART [7]. The gestational age at ART initiation was defined as the week at the beginning of the first ART regimen for women who did not receive therapy during the periconceptional period.
Statistical analysis.
Analyses were performed on data reported through December 2005 for children followed up since birth who were born from 1 January 2002 through 31 December 2004. Children born from 1 January 2005 through 30 June 2005 were excluded from analysis because of the large percentage of exposed children with an undetermined HIV-1 status.
Data were expressed as median and interquartile range (IQR) or as mean and 95% CI. Differences in proportions were evaluated by the χ2 test. A nonparametric 3-sample test was performed to test the hypothesis of median equality. The analysis of variance was used to test for differences in maternal viral load at delivery. To evaluate the association of ART interruption with the rate of MTCT, logistic regression models were performed to estimate adjusted ORs for factors potentially influencing the rate of MTCT: infant's sex, gestational age (<37 weeks, 37 weeks, or unknown), plasma viral RNA load at delivery (<2.6, 2.6-4.00, 4.01-5.78, >5.78 log10 copies/mL, or unknown), CD4+ cell count at delivery (<200, 200-499, >500 cells/μL, or unknown), mode of delivery (vaginal delivery, elective cesarean delivery, cesarean delivery other than elective, or unknown), maternal ART in pregnancy (zidovudine monotherapy, double therapy, HAART, or unknown), maternal interruption of ART (yes or no), intrapartum ART (yes, no, or unknown), and neonatal ART (yes, no, or unknown). Because no child was breast-fed, the mode of feeding was not included in the regression analyses. To better evaluate interaction between quantitative variables potentially associated with the rate of MTCT, an additional logistic regression model with log10 plasma viral RNA load and log10 CD4+ cell count as continuous variables was performed for a subgroup of mother-child pairs with both variables known.
|
|
| |
| |
|
|
|
