Risk of Birth Defects and Perinatal Outcomes in HIV-Infected Women Exposed to Integrase Strand Inhibitors During Pregnancy

Jeanne Sibiude; Jérôme Le Chenadec; Laurent Mandelbrot; Catherine Dollfus; Sophie Matheron; Nathalie Lelong; Véronique Avettand-Fenoel; Maud Brossard; Pierre Frange; Véronique Reliquet; Josiane Warszawski; Roland Tubiana


AIDS. 2021;35(2):219-226. 

In This Article

Materials and Methods

In the large national prospective French Perinatal Cohort (EPF) (Agence Nationale de Recherche sur le Sida et les Hepatites [ANRS] CO1/CO11), we compared birth defects and other perinatal outcomes according to the timing of exposure to INSTI during the pregnancy. We also matched each mother-infant pair exposed to INSTI with a mother-infant pair exposed to darunavir/ritonavir receiving the same backbone of nucleoside reverse transcriptase inhibitors and matched for other characteristics detailed as follows.

The French Perinatal Cohort, Enquête Périnatale Française

EPF (ANRS CO1/CO11) is a national multicentre cohort, prospectively enrolling pregnant HIV-infected women delivering in 90 centres throughout France.[1] No specific recommendations are made for women included in the cohort, but clinicians are encouraged to follow the most recent French national guidelines. These recommendations include prenatal ultrasound at each trimester of pregnancy and paediatric clinical examinations at birth, 1, 3, 6, 12 and 18–24 months.[9] Standardized questionnaires were filled out by clinicians, after delivery for pregnancy, and at each visit for children. Variables collected are described below. The EPF coverage is estimated to be around 70% of HIV-infected women in metropolitan France. In each participating maternity, around 95% of pregnant women living with HIV are included, with informed consent. The study was approved by the Hôpital Cochin IRB and the French computer database watchdog commission (Comission Nationale Informatique et Libertés).

Study Population

Integrase Strand Transfer Inhibitors-exposed Mother-infant Pairs. All ART combinations administered during pregnancy were recorded with the dates when started and stopped. We included all pregnant women who received an INSTI-based ART during pregnancy between January 2008 (first prescription of an INSTI in the cohort) and December 2017 and for whom pregnancy outcome was available. Terminations of pregnancy (TOP) for foetal abnormalities and stillbirths were included in the analysis. We classified mother-infant pairs in three groups: (Group 1) ongoing INSTI-based ART at conception; (Group 2) not receiving any ART at conception, and initiating an INSTI-based treatment during pregnancy; and (Group 3) starting with another ART-combination and switching to an INSTI-based treatment during pregnancy, whatever the timing of initiation of the first ART.

Matched Mother-infant Pairs. Each INSTI-exposed mother infant pair was matched 1 : 1 with an INSTI-unexposed mother-infant pair according to the type of associated ART drugs in the combination (ART backbone), age (< 35 vs. ≥35 years), geographic origin, centre, year of delivery, gestational age at ART-initiation and number of foetuses (singleton vs. twins). INSTI-exposed women who did not receive protease inhibitor or NNRTI were matched to women who received darunavir/ritonavir, with the same other drugs, as this is currently the first-line recommended regimen and has not yet been associated with birth defects.[9] For example, a woman receiving a combination of raltegravir/tenofovir/emtricitabine was matched with a woman receiving a combination of darunavir/ritonavir/tenofovir/emtricitabine; a woman receiving both an INSTI and a protease inhibitor as in this regimen: raltegravir/darunavir/ritonavir/abacavir/emtricitabine was matched with a woman receiving darunavir/ritonavir/abacavir/emtricitabine. Women in INSTI-exposed group 3 were matched according to timing of first ART (trimester of pregnancy). In case of twins, the first-born twin was retained for analysis, and we also conducted a sensitivity analysis for birth defect considering the outcome if at least one of the twins was affected.


Birth Defects. All clinical events in infants were recorded at each visit (at birth, and at 1, 3, 6, 12 and 18–24 months). We first coded the birth defects with the International Classification of Diseases (ICD) 10 codes. We then used EUROCAT inclusion criteria and guidelines in order to assess the overall prevalence of abnormalities and to classify them in different organ systems.[13,14] When comparing the overall rates of birth defect, each child was only counted once, even if several defects were present. Clusters of defects were studied by comparing the rates of birth defects by organ system.

Perinatal Outcomes. Other perinatal outcomes studied included: stillbirths; preterm birth, defined as gestational age less than 37 weeks' gestation; small for gestational age (SGA) as birthweight less than third centile, length and head circumference at birth less than third centile according to French references.[15]

Maternal variables included age, geographic origin (France, sub-Saharan Africa, other), parity, timing of HIV diagnosis (number of years since diagnosis), viral load and CD4+ cell count at delivery. Neonatal variables used for analysis included sex, gestational age, birthweight and HIV infection status.

Statistical Analysis

Maternal characteristics and perinatal outcomes were compared between the three groups of exposure with chi-square tests. Associations between birth defect and group of exposure was then studied with univariate and multivariate logistic regression, adjusting on potential confounders, chosen among variables known to be associated to the rate of birth defects: age, ethnicity and multiple pregnancy.

In the matched analysis, the prevalence of birth defects and other perinatal outcomes was compared between INSTI exposed mother-infant pairs and matched unexposed pairs, according to specific INSTI drugs (raltegravir, dolutegravir or elvitegravir/cobicistat) and by grouping all INSTI-based ART together. For the comparison of outcomes according to matched pairs, Mc Nemar tests were used whenever the discordant pairs were at least 10, allowing the use of this test.

A two-sided P value less than 0.05 was taken as indicating statistical significance. Data were analysed using Stata 14.0 software (Stata Corp., College Station, Texas, USA).[16]