Birth Defects among Children Born to Human Immunodeficiency Virus-infected Women: Pediatric AIDS Clinical Trials Protocols 219 and 219C

Susan B. Brogly, PhD; Mark J. Abzug, MD; D. Heather Watts, MD; Coleen K. Cunningham, MD; Paige L. Williams, PhD; James Oleske, MD; Daniel Conway, MD; Rhoda S. Sperling, MD; Hans Spiegel, MD; Russell B. Van Dyke, MD


Pediatr Infect Dis J. 2010;28(8):721-727. 

In This Article


Of 5931 children in protocols 219 and 219C, 2202 enrolled by 1 year of age and constituted the study population. Following clinical review of birth defects according to MACDP guidelines, 117 children had at least 1 defect, 103 with at least 1 major defect, and 14 with 2 or more conditional defects but no major defect. Among these 117 children, 77 had 1 birth defect, 30 had 2 birth defects, 6 had 3 birth defects, and 4 had 4 birth defects. Overall defect prevalence was 5.3% (95% CI: 4.4, 6.3) including all 117 cases, and was 4.7% (95% CI: 3.8, 5.6) including 103 cases with major defects. Prevalence was 4.9% (95% CI: 2.6, 8.2) and 5.4% (95% CI: 4.4, 6.5) in HIV-infected and HIV-uninfected/indeterminate children (Table 1), respectively, and was 4.8% (95% CI: 3.7, 6.1) in first trimester unexposed children, and 5.8% (95% CI: 4.2, 7.8) in first trimester ARV exposed children (Table 2).

The majority of defects occurred in the heart and musculoskeletal system (Table, Supplemental Digital Content 1, Prevalence was significantly higher among children whose mother had participated in a PACTG study during pregnancy and increased with increasing maternal age (Table 1). Prevalence also was higher among males and children with first trimester folate antagonist exposure (ie, trimethoprim/sulfamethoxazole), although these differences were not statistically significant, and folate antagonist exposure was unavailable for over half of the children. There was no difference in defect prevalence by highest log10 median maternal HIV viral load (3.4 copies/mL [children with defects] vs. 3.5 copies/mL [children without defects]), or lowest median maternal CD4 count (360 cell/mL [children with defects] vs. 372 cells/mL [children without defects]) during pregnancy. Defect prevalence significantly differed by protocol: rates were 6.8% (95% CI: 5.2, 8.7) and 4.4 (95% CI: 3.3, 5.6) for children enrolled in protocol 219 (whether or not in 219C) and in 219C alone. Figure, Supplemental Digital Content 2,, shows the prevalence of birth defects by year of birth; 1992 and 2006 were excluded because of the small number of children born in these years. No overall difference in prevalence by year of birth was identified.

The unadjusted and adjusted estimates between first trimester in utero ARV exposure and birth defects are shown in Table 2. In unadjusted analyses, there was no significant association with overall first trimester ARV exposure or first trimester exposure to specific drug classes. However, significantly more children with birth defects were exposed to efavirenz in the first trimester. The mothers of all 5 cases were taking efavirenz at the time of conception and 3 stopped efavirenz around the time pregnancy would have been identified; the other 2 mothers stopped efavirenz in the second trimester. All mothers of the 5 efavirenz-exposed children with defects also were receiving lamivudine plus other ARV. The defects of these efavirenz exposed children included laryngomalacia (N = 1), meningomyelocele with Arnold-Chiari Malformation Type II (N = 1), hypospadias (N = 1), varus feet and hypertonicity of extremities (N = 1), and cleft palate (N = 1).

The rate of birth defects also was higher in children exposed to lopinavir/ritonavir in the first trimester than in children unexposed to lopinavir/ritonavir in the first trimester. The defects of the 6 lopinavir/ritonavir exposed children included hydronephrosis (N = 1), supernumerary nipple and umbilical hernia (N = 1), atrial septal defect (N = 1), pyloric stenosis (N = 2), and ventricular septal defect and hemangioma (N = 1). None of the children with defects were exposed to both efavirenz and lopinavir/ritonavir in the first trimester.

In models adjusted for first trimester folate antagonist exposure, year of birth, and perinatal study participation, the association with efavirenz persisted while the association with lopinavir/ritonavir was marginally significant (P = 0.07). To further explore possible confounding, we examined maternal and infant characteristics by perinatal protocol participation (data not shown). In models adjusted for year of birth, participation in a perinatal protocol was higher among infants with first trimester exposure to any ARV (Odd ratio [OR] = 1.47, 95% CI: 1.21, 1.79) and to any nucleoside analogue (OR = 1.48, 95% CI: 1.22, 1.80), and was lower among infants with first trimester exposure to any non-nucleoside analogue (OR = 0.57, 95% CI:0.38, 0.85). However, other characteristics generally were in the direction of a higher possible risk of defects in those who did not participate in a perinatal protocol (eg, more mothers <20 and >30 years of age, more maternal cocaine use, lower infant birth weights, more preterm births, and more HIV-infected infants) except for maternal alcohol use, which was higher among perinatal study participants.

We also examined associations between in utero ARV exposure and the most common categories of specific defects: musculoskeletal and heart (Table, Supplemental Digital Content 1, Because of the lower number of cases (N = 36 and 34, respectively),[2] these models were only adjusted for perinatal protocol participation and first trimester folate antagonist exposure. Protective effects of first trimester zidovudine exposure on musculoskeletal defects were detected in unadjusted (OR = 0.30, 95% CI: 0.10, 0.84) and adjusted models (OR = 0.24, 95% CI: 0.08, 0.69). Protective effects on musculoskeletal defects also were found with overall first trimester ARV exposure and any first trimester nucleoside analogue exposure in adjusted models. These latter findings appeared to be driven by zidovudine exposure; the frequency of exposure was similar for any ARV, for any nucleoside analogue, and for zidovudine. In contrast, significantly more children with heart defects—MACDP category of heart, other, which excludes conotruncal and obstructive defects (Table, Supplemental Digital Content 1, were exposed to zidovudine in the first trimester in unadjusted (OR = 2.11, 95% CI: 1.07, 4.16) and adjusted models (OR = 2.04, 95% CI: 1.03, 4.05). This association was marginally significant when conotruncal and obstructive defects were included (OR = 1.78, 95% CI: 0.93, 3.40, P = 0.08).

To examine possible selection bias, we assessed enrollment into 219 and 219C of children who participated in PACTG 076, 316 or IMPAACT P1025 by defect status and in utero ARV exposure. These latter 3 studies were examined because birth defect information was collected and reviewed in these studies by the 076, 316, and P1025 investigators. It should be noted that 74% of children in 219 and 219C who participated in a perinatal protocol were in one of these studies. Among children who participated in PACTG 076, 316 or IMPAACT P1025, more children with defects (31.2%) than without defects (24.8%) enrolled in protocols 219 and 219C (P = 0.054). However, the only important differences in enrollment by defect status and in utero ARV exposure were among children without defects: enrollment was higher among children unexposed to abacavir (17.0% exposed vs. 25.2% unexposed enrolled, P = 0.048), and exposed to saquinavir (44.4% exposed vs. 24.6% unexposed enrolled, P = 0.018). This differential enrollment among children without defects would increase and decrease estimated associations with abacavir and saquinavir exposure, respectively. No other evidence of selection bias was identified.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: