HIV-1 transmission and survival according to feeding options in infants born to HIV-infected women in Yaoundé, Cameroon

We aimed to evaluate HIV-1 vertical transmission and infant survival rates according to feeding options, in a typical RLS with generalized HIV epidemiology, for possible knowledge generalizability to PMTCT high-priority countries with similar programmatic features.

We conducted a retrospective cohort-study from April 2008 through December 2013 amongst infants born to HIV-infected women, and followed-up throughout the PMTCT cascade-care until 24-months at the Djoungolo Health District in Yaoundé-Cameroon.

Patients and their respective data were retrieved from the Djoungolo PMTCT cohort for the period 2008-2013. Characteristics of this perinatal cohort were extensively described elsewhere [11].

Briefly, PMTCT program was launched in 2008 at the Djoungolo Health District, a setting within the catchment area of the Essos Hospital Centre (EHC), classified as the area HIV reference treatment center. Routinely, pregnant women diagnosed positive for HIV in primary healthcare centers of Djoungolo were referred to EHC for HIV-related clinical assessment, CD4 cell count measurement, infant feeding counseling (IFC) and ART initiation following standard-of-care [11]. Routine antenatal care (ANC) and delivery were done in their respective primary clinics. After delivery, mothers were advised to register their babies at the EHC for ART provision, postnatal follow-up and monitoring of PMTCT-related outcomes. Postnatal care consisted of HIV-1 early infant diagnosis (EID) at 6 weeks of age, clinical visits, continuous counseling on and follow-up of feeding practices, administration of Cotrimoxazole and late serological testing. Routine immunizations could be done at EHC or at another recognized health facility within the Djoungolo Health District. Routine appointments were planned at week 6, month 9, month 12, and between months 15-24. EID, performed using either real time PCR or qualitative PCR, was done at week 6 and further confirmed after month 12, considering 6 weeks of post weaning period (i.e. for breastfeeding populations).

During the study period, changes in PMTCT recommended ARVs used for pregnant women were taken into account. Before 2010, women having a CD4 count above 350 cells/mm³ received zidovudine (AZT) from week 28 of pregnancy till birth. As from 2010, AZT was given from week 14, while women with CD4 count below 350 cells/mm³ or clinically classified at the WHO stage 3 or 4 were eligible for lifelong triple ART.

Counseling for infant feeding options was part of ANC and postnatal care. Specifically, IFC was done during each ANC visit, and HIV positive women were referred to the infant feeding counselors for in-depth counseling purposes. During IFC, the two recommended exclusive options were either exclusive breastfeeding (EBF) or exclusive replacement feeding (ERF), with emphasis to avoid mixed feeding (MF) practice.

EBF was defined as breastfeeding with no added supplements except vitamins; ERF was defined as the administration of commercial formula feeding using bottles; while MF was described as a combination of both breastfeeding and formula feeding practices. Mothers opting to practice ERF were trained on the appropriate use of feeding bottles in terms of preparation, conservation and hygiene of feeding bottles and teens.

After IFC, mothers were all capable to freely make a choice of the appropriate feeding option for their babies. Additionally, supportive counseling was done during follow-up visits. EBF mothers were advised to EBF for 6 months, followed by weaning during a short period (i.e. 1 week). After 2009, the weaning period was extended to 1 month.

Infants born alive, to an HIV-infected mother, were enrolled at their first postnatal clinic attendance, specifically in the frame of EID (i.e. around week 6). The reported feeding option of mothers was recorded and those with doubtful or missing feeding information were not included. Reported feeding options were then used to classify infants as EBF, ERF or MF. During follow-up, the conclusive HIV-free result was provided upon the criteria of one negative virological + negative serological testing ≥12 months assuming 6 weeks post weaning in case of breastfeeding. A confirmed HIV infection was based on two positive PCR at any moment of the follow-up irrespective of feeding option, or a positive HIV serology after 18 months. Children with indeterminate HIV serological result were excluded from the analysis.

Primary endpoints were HIV transmission and survival rates at 24 months of infant age. Kaplan Meier survival curves were used to estimate cumulative probabilities of survival overtime according to feeding option, and log-rank test was used to compare the curves between different feeding options. Multivariate Cox proportional hazards regression was used to estimate unadjusted and adjusted hazard ratios, adjusting for HIV serological status and birth weight. Schoenfeld residuals test which was used in assessing the proportional hazard assumption [19] and Cox-Snell residual was used to determine the goodness-of-fit. All statistical tests were performed at 5% level of significance and with 95% confidence interval. Statistical analyses were done using the package survival incorporated in R software version 3.1.3.

Ethical clearance for the study was obtained from the National Ethics Committee for Research on Human Subjects in Cameroon (2013/02/027/L/CNER/SP) and the Institutional Review Board of EHC (2014/003/CE-CHE). All women participating with their infants in the IFC project provided a written informed consent. Data were processed using unique identifiers for purposes of privacy and confidentiality.

A total of 1086 eligible infants born to HIV positive mothers (median of 399 [IQR: 260-540] CD4 cells/mm³) were enrolled (Fig. 1). Regarding maternal ARV experience for PMTCT-interventions, 566 (52.12%) received triple ART, 411 (37.85%) received AZT and 109 (10.04%) had no ARV exposure. Regarding infant feeding options during the first 3 months of life, 663 (61.05%) were on ERF, 408 (37.57%) on EBF and 15 (1.38%) on MF (Table 1).

Overall rate of HIV-vertical transmission was 3.59% (39/1086), and all infected children were enrolled for HIV care. According to feeding options, HIV transmission varied considerably: EBF (2.72%); ERF (3.80%); MF (21.43%), p = 0.011; without any significance (p = 0.34) between EBF and ERF (Table 1). Infants experiencing MF therefore had 5.3 and 7.4 folds higher risks of HIV MTCT compared to ERF and to EBF respectively.

According to HIV status, mortality rate was higher among HIV-infected (10.25%) vs. uninfected (2.29%) infants (p = 0.016), thus indicating about 4 folds increment for infected children. HIV negative children survived more than the positive ones (p = 0.024); the log-rank test also shows significant differences in the Kaplan Meier cumulative probability curves (Fig. 2). Similarly, the 24-month survival cumulative probability was varied considerably: 89.3% [79.9%-99.8%] vs. 96.4% [94.8%-97.9%] in infected vs. uninfected infants respectively, p = 0.024.

Overall infant mortality in the study population was 2.58% (28), with varying trends according to feeding option: ERF (2.41%), EBF (2.45%), MF (13.33%), p = 0.032 (Table 2), indicating an increment of 5.4 and 5.5 folds in mortality for MF infants compared to EBF and ERF respectively. Similarly, the survival cumulative probability was 96% [94%-98%] in ERF, 96.4% [94.1%-98.8%] in EBF, and 86.67% [71.06%-100%] in MF, p = 0.04.

Using the Log-rank test of survival (Fig. 3), a significant difference was observed in the Kaplan Meier curves according to feeding options (p = 0.04). Of note, the MF curve falls within the area under the curve and is distant from both the ERF and EBF curves. Specifically, the cumulative probability survival curve of MF compared those of EBF and EBF revealed p-values of 0.011 and 0.025 respectively. However, there was no significant difference of Kaplan Meier survival curves between EBF and ERF (p = 0.48). The 24-months survival cumulative probability was 96% [94%-98%] with ERF, 96.4% [94.1%-98.8%] with EBF, and 86.67% [71.06%-100%] with MF.

Schoenfeld residuals test revealed a proportional hazard assumption with similar effects of EBF and ERF on survivals (unajusted HR 1.30 [0.59- 2.85]; adjusted HR 1.25 [0.56-2.79]). Of relevance, children in the MF arm appeared to survive less than those on ERF (unadjusted HR 0.22 [0.05-0.96]). However, after adjustment for children HIV status, the effect of MF disappeared (Adjusted HR 0.29 [0.063-1.34]), insinuating that HIV-status of children is an intermediate independent factor between infant feeding option and survival.