1. Safety of efavirenz in first- trimester of pregnancy Systematic review and meta-analysis of outcomes from observational cohorts Nathan Ford, MSF/UCT Lynne Mofenson, NIH Katarina Krazner, LSHTM Lisa Frigati, Red Cross Children Hospital Edward Mills, University of Ottawa Alexandra Calmy, Geneva University Hospital

2. Incidence of neural tube defects varies according to ethnic, geographic, and nutritional factors: 0.14% (UK) - 0.36% (South Africa) Risk data for EFV unclear: case reports contradicted by subsequent cohorts WHY DO THE STUDY?

3. RISK DATA I: ANIMAL STUDIES 20 pregnant cynomolgus monkeys dosed during pregnancy with EFV and compared with 20 controls. –3/20 EFV-exposed monkeys had neural-tube like defects But: only 2% of animal teratogens (30/1200) are teratogenic in humans

4. RISK DATA II: RETROSPECTIVE CASE REPORTS Italy, 2002: 3 NTDs, inc. 1 myelomeningocele Italy, 2002: 1 myelomeningocele USA 2005: 1 myelomeningocele But: unsystematic retrospective observations give no indication of risk –2 retrospective cases of NTDs reported for other ARVs

5. CONFLICTING RECOMMENDATIONS Conflicting guidelines - US/WHO : recommend against use in first trimester - EU/SA: recommend against use throughout pregnancy Conflicting drug labels - BMS: “Fetal harm can occur when administered to a pregnant women during the first trimester” - Aspen: “Use of Aspen efavirenz in pregnancy is not recommended”

6. Systematic review -Databases MEDLINE via PubMed, EMBASE, Cochrane CENTRAL, CINAHL, PsycInfo, LILACS, Current Controlled Trials (www.controlled-trials.com) and US National Institutes of Health -Conference abstracts All International AIDS Society conferences (up to Cape Town, July 2009) All Conferences on Retroviruses and Opportunitistic Infections (up to Montreal, February 2009) -Pregnancy registry Antiretroviral Pregnancy Registry (apregistry.com/) -Cohorts ART-LINC, ART-CC, ECS, IeDEA-SA, MSF, MTCT+, NISDI, PENTA, PHRU, RHRU Ford et al, AIDS 2010.24: 1461-70

7. Systematic review Primary endpoint -Birth defects of any kind Primary analysis - 1 st Trimester EFV vs 1 st Trimester non-EFV ART Secondary outcomes -Spontaneous abortions, termination of pregnancy, stillbirths, pre-term delivery Secondary analyses: -Sensitivity analyses: location, duration of exposure, publication status -EFV risk 1 st trimster vs 2 nd /3 rd trimster -Prevalence of secondary outcomes -GRADE assessment of quality of evidence

8. 90 full text papers reviewed 26 studies considered eligible for inclusion 1098 articles screened by title and abstract 1008 Papers excluded (not relevant) 10 studies excluded Data not disaggregated by trimester; outcome data not available unsystematic observational studies; duplicate reports 72 Papers excluded 19 reviews or commentaries 9 duplicate reports 45 Not relevant 8 additional articles identified through bibliographies 16 included in analysis Identification process for eligible studies

9. Results I: Study Characteristics 16 studies 11 prospective cohorts 5 retrospective cohorts 9 MLIC: SA, Botswana, Brazil, Ivory Coast, multisite) 7 HIC: Europe and US

10. StudySettingDescriptionSize pregnancy cohort Birth defect data MTCT-PlusMultisiteProspective cohort495No (other outcome) WestreichS AfricaProspective cohort81Yes APRMultisiteBirth registry11,867Yes BeraS AfricaProspective cohort851Yes TownsendUKPopulation based surveillance7,135Yes CoffieCote d’IvoireProspective cohort in trial168Yes LaherS AfricaRetrospective cohort117No (other outcome) MachadoBrazilProspective cohort696Yes Gonzales- Tome SpainProspective cohort619Yes RossouwS AfricaRetrospective cohort37Yes BussmanBotswanaProspective cohort71Yes FloridiaItalyNational surveillance study334Yes JoaoBrazilRetrospective cohort90Yes JeantilsFranceRetrospective cohort12Yes PatelEuropeProspective cohort1,973Yes BatallanFranceRetrospective cohort9Yes

12. Primary outcomes –Pooled RR for EFV vs non EFV: 0.85 (95% CI 0.60-1.21, p=0.47) Low heterogeneity (I 2 =0; 95% CI 0-56.3%, p=0.85) –Prevalence of birth defects: 2.9% (95% CI 2.1-4.0%) similar ranges reported in the general population in the US (2.7%) France (2-5%) and South Africa (2.5 - 8%) –Prevalence neural tube defect s (1/1301): 0.08% (95% CI 0.02- 0.43%) similar to ranges in general population in the UK (0.14%) and South Africa (0.36%)

14. Secondary outcomes RR 1 st vs 2 nd /3 rd trimester EFV =0.91 (95% CI = 0.46-1.79) Prevalence of stillbirth, spontaneous abortion and pre-term within range of general population Termination of pregnancy –5.3% (CI95 0.64-17.7%) to 33.7% (CI95 23.7- 44.9%). –Soweto: RR EFV vs non-EFV 5.73 (CI95 1.45- 22.75)

15. CONCLUSIONS: I No increased risk of overall birth defects among women exposed to EFV during 1 st trimester compared to exposure to non-EFV regimens – can exclude a 2-fold increase in overall birth defects Only 1 NTD reported for 1301 live births, giving a prevalence (0.08%) – Point prevalence within range of general population – Fragile data (upper CI95% = 0.43%)

16. CONCLUSIONS: II High rates of TOP point to need for careful counselling Birth registries should be supported –Need for standardized birth outcome data collection –Data exist but not collected/reported