Antiretroviral treatment and prevention of peripartum and postnatal HIV transmission in West Africa: Evaluation of a two-tiered approach

Background: Highly active antiretroviral treatment (HAART) has only been recently recommended for HIV-infected pregnant women requiring treatment for their own health in resource-limited settings. However, there are few documented experiences from African countries. We evaluated the short-term (4 wk) and long-term (12 mo) effectiveness of a two-tiered strategy of prevention of mother-to-child transmission of HIV (PMTCT) in Africa: women meeting the eligibility criteria of the World Health Organization (WHO) received HAART, and women with less advanced HIV disease received short-course antiretroviral (scARV) PMTCT regimens. Methods and Findings: The MTCT-Plus Initiative is a multi-country, family-centred HIV care and treatment program for pregnant and postpartum women and their families. Pregnant women enrolled in Abidjan, Côte d’Ivoire received either HAART for their own health or short-course antiretroviral (scARV) PMTCT regimens according to their clinical and immunological status. Plasma HIV-RNA viral load (VL) was measured to diagnose peripartum infection when infants were 4 wk of age, and HIV final status was documented either by rapid antibody testing when infants were aged ≥ 12 mo or by plasma VL earlier. The Kaplan-Meier method was used to estimate the rate of HIV transmission and HIV-free survival. Between August 2003 and June 2005, 107 women began HAART at a median of 30 wk of gestation, 102 of them with zidovudine (ZDV), lamivudine (3TC), and nevirapine (NVP) and they continued treatment postpartum; 143 other women received scARV for PMTCT, 103 of them with sc(ZDV+3TC) with single-dose NVP during labour. Most (75%) of the infants were breast-fed for a median of 5 mo. Overall, the rate of peripartum HIV transmission was 2.2% (95% confidence interval [CI] 0.3%-4.2%) and the cumulative rate at 12 mo was 5.7% (95% CI 2.5%-9.0%). The overall probability of infant death or infection with HIV was 4.3% (95% CI 1.7%-7.0%) at age week 4 wk and 11.7% (95% CI 7.5%-15.9%) at 12 mo. Conclusions: This two-tiered strategy appears to be safe and highly effective for short- and long-term PMTCT in resource-constrained settings. These results indicate a further benefit of access to HAART for pregnant women who need treatment for their own health. © 2007 Tonwe-Gold et al. This observational cohort included women enrolled in the MTCT-Plus Initiative, a multi-country, comprehensive HIV care and treatment program for pregnant and postpartum women and their families built on existing PMTCT services. It provides pregnant and postpartum women with holistic, family-centred HIV care including HAART to the woman, her partner, and her children [17]. Pregnant women identified as HIV-infected at two community-based antenatal clinics in two low-income urban districts of Abidjan were referred for enrolment into the MTCT-Plus program. The study population included all HIV-infected pregnant women and their live-born infants enrolled between August 2003 and April 2005 and followed until a final paediatric HIV status could be determined. Women were divided into two groups. The first cohort included the HIV-infected pregnant women who were eligible for HAART based on WHO criteria (at high risk of transmission); the second cohort included HIV-infected pregnant women who were not eligible for HAART and received sc-PMTCT (with low risk of transmission). The MTCT-Plus program was reviewed by the institutional review board (IRB) from Columbia University in 2000 (principal sponsor) and was not considered a research project but rather a demonstration program in the context of the ARV roll-out. Informed consent was not required by the Columbia IRB. Upon enrolment, all participants were screened for HAART eligibility, defined from August 2003 to December 2004 as: WHO clinical stage 4 irrespective of CD4 T cell count; WHO stage 2 or 3 and CD4 T cell count ≤ 350 cells/mm3; or CD4 T cell count < 200 cells/mm3. These criteria were revised by the MTCT-Plus Initiative in January 2005 to be in compliance with WHO 2005 PMTCT guidelines and to avoid the side effects reported with NVP especially in pregnant women with CD4 T cell count > 250 cells/mm3 [18]; from that time forward, women with WHO stage 2 and CD4 T cell count < 350 cells/mm3 were no longer considered eligible for HAART. Other criteria taken into account for deciding to initiate HAART were the absence of medical contraindications to HAART as well as major barriers to adherence. Pregnant women who met eligibility criteria initiated HAART as early as 24 wk of gestation according to the last menstruation date, usually with zidovudine (ZDV), lamivudine (3TC) and nevirapine (NVP). Treatment continued during labour and postnatally. Pregnant women who were not eligible for HAART received validated scARV prophylactic regimens, usually sc(ZDV+3TC) from 32 wk of gestation (until 3 d postpartum) and sdNVP in labour [9,19], or scZDV from 28 wk, or sdNVP alone, or both scZDV and sdNVP [4,7–9,20]. All infants received ZDV syrup for 7 d and sdNVP syrup on day 3 irrespective of the maternal drug regimen. Clinical classification using the WHO staging system was performed by the physicians at enrolment into the MTCT-Plus program and at each subsequent visit. Women either used breast milk substitutes if feasible and affordable following WHO guidelines [21] or were encouraged to practice exclusive breast-feeding for a maximum of 6 mo and initiate early weaning from 4 mo onwards. Caesarean section was only performed for emergencies. Baseline adherence and psycho-social assessments were done before initiation of HAART, and followed by weekly visits for 8 wk using a standardized checklist of clinical symptoms to detect side effects and ensure adherence. A more detailed clinical evaluation, including symptom review, physical examination, and review of medication adherence was conducted monthly following MTCT-Plus clinical guidelines [22]. Adherence was assessed by physicians or nurses on the basis of a self report consisting of the number of pills taken by the women during the last 7 d preceding each schedule visit (all, most, few, none). Severe adverse events were classified according to international guidelines [23]. All blood samples of pregnant women who agreed to test were screened on site for HIV antibodies according to a validated algorithm described elsewhere [24]. CD4 T cell count was measured upon enrolment and at intervals of 6 mo. Hepatic and renal functions were measured prior to HAART initiation; liver function tests were repeated 2 wk after treatment initiation and monthly throughout pregnancy for women on HAART. Additional laboratory evaluation was restricted to patients with clinical indications. Laboratory tests were done at a central reference laboratory (CeDReS, Abidjan) certified by regional and international quality assurance programs. The CD4 T cell count and percentage were measured in the antenatal period at the enrolment visit and thereafter every 6 mo after the initiation of the treatment using a dual-platform flow-cytometry technique with an automated blood cell counter (MaxM, Beckman Coulter, http://www.beckmancoulter.com/) for absolute lymphocyte count and a flow cytometer (FACScan, Becton Dickinson, http://www.bdbiosciences.com/) for measuring the percentage of CD4 T cells (CD4%). Absolute CD4 T cell count was then calculated multiplying the CD4% by the total lymphocyte count. The laboratory quality control procedures included: (i) daily internal controls for both the automated blood cell counter (Coulter 5C cell control, Beckman Coulter) and the flow cytometer (BD multi-check control, Becton Dickinson); (ii) participation in two international assurance quality programs (UK-NEKAS and QASI). Plasma HIV-RNA viral load (VL) testing for the early diagnosis of paediatric HIV infection was performed using a quantitative real-time RT-PCR technique targeted at the HIV-1 LTR gene as previously validated [25]. The quantification limit of this method was 300 copies/ml with 200 μl of plasma. All infants were tested at 4 wk of age, and if positive, confirmed with repeat VL testing at 6 wk. Infants with two positive tests were classified as infected peripartum. Infants with a negative VL test at 4 wk of age were classified as uninfected peripartum and were subsequently tested at 12 mo by rapid HIV antibody testing. If the antibody test at 12 mo was positive, we performed a second test for confirmation at 15 mo or later. The HIV infection status of the children who stopped follow-up just after being tested HIV antibody-positive at 12 mo or who had ceased breast-feeding under the age of 2 mo was confirmed by VL testing. HIV antibody-negative or VL-negative children were classified as uninfected. Postnatal transmission was defined as a negative HIV PCR from a first sample at age 30–180 d followed by a positive VL or antibody test at 12 or 15 mo. Group comparisons according to ARV regimens used Student t-test or nonparametric Mann-Whitney test for quantitative variables, and Chi2 test or Fisher exact test for qualitative variables. The first-live born child was used for analysis in the case of multiple births. Two survival analyses were conducted using two different outcomes in children: HIV infection or HIV-free survival (which is defined as ending at the occurrence of HIV infection or death, whichever came first). Cumulative transmission risks of HIV and (HIV or death) were estimated using both Turnbull method and Kaplan-Meier probabilities. As results were similar (unpublished data), Kaplan-Meier estimates were retained to allow for comparisons between groups using the log-rank test. For survival analysis, time to infection was estimated up to the first positive test for peripartum cases or to the midpoint date between the last negative test and the first positive test for the postnatal cases. Results were expressed in percentages with their 95% confidence intervals (95% CIs). Determinants of HIV infection and (HIV or death) were explored using a Cox model with the following variables: maternal age, ARV regimen, WHO staging, CD4 T cell count, haemoglobinemia, infant feeding practice from birth, birth weight, and child gender. A stepwise descendant multivariate analysis included all variables with p < 0.25 in the univariate analysis. Statistical analyses were processed with Stata software, version 9.0 (Stata, http://www.stata.com/).