Retrospective analysis finds higher prevalence of birth defects associated with efavirenz than previously reported

Polly Clayden, HIV i-Base

A study of children enrolled in PACTG protocols 219 and 219C – a multisite US cohort of children born to HIV-positive women set up to study the long term effects of in utero antiretroviral exposure – reports a higher prevalence of birth defects than found in other paediatric cohorts.

Preliminary data from this paper published in August 2010 edition of The Pediatric Infectious Disease Journal authored by Susan B Brogly and colleagues was previously shown as a poster at CROI this year, which we reported in the April 2010 edition of HTB. [1, 2 , 3]

Protocol 219 followed HIV-infected and uninfected children from May 1993 to August 2000. Children were eligible if their mothers were enrolled in a PACTG trial in pregnancy. In September 2000, protocol 219C was introduced, amending 219 to remove the eligibility criterion mandating enrolment in another PACTG trial. This study evaluated children enrolled in 219 and 219C before one year of age.

Birth defect data were recorded at study visits. Protocol 219 did not include a direct question about birth defects; 219C included this question.

The primary determinant was first trimester exposure. The investigators looked at overall antiretroviral exposure, classes of antiretrovirals and specific antiretrovirals to which at least one child with a birth defect had first trimester exposure.

The reference group was children unexposed to the particular antiretroviral (or class of drug) during the first trimester, which included antiretroviral unexposed children, those only exposed in labour, those unexposed to the particular drug but exposed to other antiretrovirals and children only exposed beyond the first trimester.

Clinicians were blinded to antiretroviral exposure and the outcome was presence of a birth defect within the first year of life.

The investigators used logistic regression models to estimate the association between first trimester in utero antiretroviral exposure and the most common categories of birth defects.

They found, a total of 117 children with a least one defect out of the study population of 2202 children. This gave an overall defect prevalence of 5.3% (95% CI, 4.4-6.3) and 4.7% (95% CI, 3.8-5.6) if just the 103 cases of major defects were included.

Prevalence was 4.8% (95% CI, 3.8- 5.6) in children unexposed in the first trimester and 5.8% (95% CI, 4.2-7.8) in exposed children.

They found that the majority of defects occurred in the heart and musculoskeletal system. They observed a higher prevalence in children whose mothers had participated in a PACTG study during pregnancy and an increase as with increased maternal age. Rates were also higher in children enrolled in protocol 219 (whether or not enrolled in 219C) than in 219C alone.

They reported a higher defect rate (5/32, 15.6%) among children exposed to efavirenz in the first trimester compared to unexposed children, AOR 4.31 (95%CI, 1.56-11.86). The defects included laryngomalacia (n=1), meningomyelocele with Arnold-Chiari Malformation Type II (n=1), hypospadias (n=1), club foot (n=1), hypertonicity of extremities (n=1) and cleft palate (n=1).

They also found an association in children exposed to lopinavir/ritonavir but when they adjusted for first trimester folate antagonist exposure, year of birth and perinatal study participation this did not persist (p=0.07) whereas the association with efavirenz did.

There was a protective effect of AZT first trimester exposure on musculosketal defects AOR 0.24 (95% CI, 0.08-0.69), but an association with heart defects AOR 2.04 (95% CI 1.03-4.05).


It is difficult to know whether this study advances or confuses the field. Although based on 2202 children, only a third (763) were exposed to any antiretroviral during the first trimester compared with almost 5000 enrolled in the Antiretroviral Pregnancy Registry (APR) to date. Consequently, with the exceptions of AZT, 3TC and nelfinavir, few subjects had been exposed to individual compounds, which accounts for the wide range of odds ratios (0.39 for ddI to 3.52 for efavirenz). A similar phenomenon has been observed over the years in the APR with new compounds that is followed by a gradual movement towards the mean as the denominator increases.

The findings also generally differ from the APR, in which only ddI (4.5%, 95% CI 2.6–7.1%) has attracted attention due to a small but persistent increase in risk of birth defects, while PIs in general, and lopinavir/ritonavir (1.7%, 95%CI 0.8–3.1%) in particular, have generally been found to be associated with no increase in risk.

As might be anticipated, folate antagonist exposure during the first trimester was associated with an increased prevalence of birth defects although data were largely incomplete and the observation did not reach statistical significance. The observed confounding of this risk with lopinavir/ritonavir exposure is a timely reminder not to forget the obvious.

It is difficult to know what to make of the apparent protective effect of AZT exposure in regard to musculoskeletal abnormalities and increased risk of cardiac defects. The authors rightly conclude that further study is needed to rule out confounding and to examine for associations between antiretroviral exposure and specific birth defects. Prospective reporting to the APR remains the best option and a significant improvement in UK reporting to this international drug safety registry (currently standing at 3.3% of all reports) is long overdue.


  1. Brogly et al. Birth defects among children born to human immunodeficiency virus-infected women. Pediatric AIDS Clinical Trials Protocols 219 and 219C. The Pediatric Infectious Disease Journal. Vol 29. Number 8. August 2010.
  2. Clayden P. Efavirenz use in pregnancy and birth outcomes. HIV Treatment Bulletin. April 2010.

Links to other websites are current at date of posting but not maintained.