Long-term outcomes of early initiated antiretroviral therapy in sub-Saharan children: A Cameroonian cohort study

Background: In most studies, the virological response is assessed during the rst two years of antiretroviral treatment initiated in HIV-infected infants. However, early initiation of antiretroviral therapy exposes infants to very long-lasting treatment. Moreover, maintaining viral suppression in children is dicult. We aimed to assess the virologic response and mortality in HIV-infected children after ve years of early initiated antiretroviral treatment (ART) and identify factors associated with virologic success in Cameroon. Methods: In this cohort study, we included 149 children who were still alive after two years of early ART. Virologic response was assessed after 5 years of treatment. The probability of maintaining virologic success between two and ve years of ART was estimated using Kaplan-Meier curve. Factors associated with a viral load < 400 copies/mL in children still alive at ve years of ART were studied using logistic regressions. Results: The viral load after ve years of early ART was suppressed in 66.8% (60.1 - 73.5) of the 144 children still alive and in care. Among the children with viral suppression after two years of ART, the probability of maintaining viral suppression after ve years of ART was 64.0% (54.0 - 74.0). The only factor associated with viral suppression after ve years of ART was achievement of conrmed virological success within the rst two years of ART (OR = 2.7 (1.1-6.8); p = 0.033). Conclusions: This study reported diculty in maintaining virologic success in sub-Saharan HIV-infected children between two and ve years of early initiated ART. It also highlighted the importance of initial viral suppression for achieving and maintaining virologic success in the long-term.

In most studies, the virologic response is assessed during the rst two years following ART initiation of HIV-infected infants (7)(8)(9)(10)12,(21)(22)(23)(24)(25)(26)(27)(28). However, early initiation of ART in vertically HIV-infected children exposes them to a very long-lasting treatment relative to that of adults. Moreover, maintaining viral suppression in children is very di cult (8, 29,30). Pediatric ART options are limited in Sub-Saharan Africa due to low availability of pediatric formulations (16, 22,31,32). Thus long-term success of early infant ART in Sub-Saharan Africa depends on the e cacy of rst-line regimens (33,34). There is a need to study longer-term e cacy of early initiated ART in HIV-infected children.
In the ANRS-12140 Pediacam cohort study, Cameroonian HIV-infected children were initiated on ART at a median age of four months (35,36). We aimed to determine the virologic response and mortality after ve years of antiretroviral treatment initiated during the rst year of life based on data from this cohort study and identify factors associated with virologic success in a sub-Saharan country (Cameroon).

Methods
The ANRS-12140 Pediacam Study ANRS-12140 Pediacam is a prospective cohort study of HIV-infected children included between November 2007 and October 2011 from three referral hospitals in Cameroon: the "Centre Mère et Enfant de la Fondation Chantal Biya" (CME-FCB) and "Centre Hospitalier d'Essos" (CHE), both in Yaounde, and the "Hôpital Laquintinie de Douala" (HLD) in Douala. Pediacam comprised two phases and has been described previously (37,38). Brie y, during the rst phase, infants born to HIV-infected mothers and those born to HIV-uninfected mothers were matched according to gender and site of recruitment during the rst week of life and followed until the fourteenth week. HIV-exposed infants were tested for HIV using Polymerase Chain Reaction (PCR) test at six weeks of age, according to the Cameroonian guidelines at the time of the study. Results of HIV test were available at 10 weeks with retesting for HIV-positive diagnoses. HIV-negative breastfed infants were retested six weeks after weaning. All HIV-infected infants and subsamples of uninfected HIV-exposed and HIV-unexposed infants were eligible for the second phase of follow-up, planned to continue until ve years of age. During the above inclusion period, direct inclusion into the phase 2 follow-up was proposed to the mothers of HIV-infected infants identi ed and diagnosed after the rst week but before seven months of life. Overall, 210 HIV-infected infants were included in this second phase. Antitretroviral therapy (ART) was systematically proposed as soon as the HIV status was con rmed, according to the Cameroonian guidelines at the time of the study: zidovudine (or abacavir or stavudine) + lamivudine, for all infants, combined with ritonavir-boosted lopinavir or nevirapine, depending on the previous history of nevirapine used for prevention of mother-to-child transmission (PMTCT). Viral load of HIV was measured by RT PCR (Biocentric, Bandol, France) using plasma specimen, with 300 copies/mL as lower limit of quanti cation from 2007 to 2011. Thereafter, new devices for measuring viral load were acquired with the threshold of 60 copies/mL as lower limit of quanti cation. Data concerning biological and clinical parameters collected at inclusion and every three months after ART initiation until the age of two years and every six months thereafter. At the same timepoints, a standardized questionnaire was administered to caregivers to collect data on family living conditions and adherence to ART. Children whose HIV viral load remained high (≥1000 copies/mL) for at least three consecutive medical visits were selected for HIV genotypic resistance test to ARVs performed with 1 mL of plasma using the French ANRS (French National Agency for Research on AIDS and Viral Hepatitis) protocol (39). The targeted regions were Protease and Reverse transcriptase of the HIV polymerase gene. The exams and transport fees were paid by the Pediacam project.

Study population
We considered for this analysis the HIV-infected children included in phase 2 of the Pediacam study who were alive at two years of ART initiated no later than the age of one year ( Figure 1). Data were analyzed from the second to fth year of ART.

Endpoints
The main outcome variable for this study was virologic success (VS), de ned on the one hand as the achievement of viral load < 400 copies/mL at least once during the period from two to ve years after ART initiation, and one the other as the achievement of viral load < 400 copies/mL after ve years of ART initiation. This threshold had also been reported in previous studies (9,24,(40)(41)(42). We also studied mortality and the immune status, de ned by the percentage of CD4 lymphocytes, at ve years after ART initiation.
Virologic, immunological, and clinical data were analyzed during the visits scheduled every six months from 24 months after ART initiation: month 24 (M24), month 30 (M30), month 36 (M36), month 42 (M42), month 48 (M48), month 54 (M54), and month 60 (M60). A window of three months before and after these time points was accepted for data collection. The monitoring values used were the nearest measurements within these ± 3-month intervals. The children still alive who did not attend the scheduled visit were classi ed as virologic failure with respect to the corresponding visit.

Exposure variables
We mainly considered characteristics at M24 to identify factors associated with VS at M60: calendar period, ART regimen, site of ART delivery, age and clinical presentation of the child, attainment of con rmed VS (viral load (VL) < 1000 copies/mL) within the rst two years after ART initiation (43), virological status at M24 (viral load < or ≥1000 copies/mL), and immunological status (percentage of CD4 lymphocytes). We also considered variables related to living conditions at M24: people living in the household, access to running water, access to electricity, and availability of a functional refrigerator at home. Finally, we considered the reporting of caregiver adherence (recall of missed doses in the past three days) collected at M60. Children were classi ed as non-adherent if at least one dose was missed during the past three days (18,36).

Statistical analysis
The study population was initially divided into two groups by distinguishing their virological status at two years of ART: 1) children who had a viral load < 400 copies/mL at M24 (Group 1) and 2) children who had a viral load ≥ 400 copies/mL or whose viral load had not been measured at M24 (15 children of whom the last viral load measured was ≥ 400 copies/mL at M12) (Group 2).
The probability of maintaining VS from M24 to M60 for group 1 was estimated using the Kaplan-Meier model and the probability of achieving VS for group 2 was estimated using a competing risk regression curve (with death as the competing event).
The association of exposure variables with viral load < 400 copies/mL (versus a viral load ≥ 400 copies/mL) at M60 for children who were still alive was assessed in univariate and multivariate analysis using logistic regression. The site of ART delivery, virological status at M24, ART regimen at M24, and reporting of caregiver adherence (recall of missed doses in the past three days) at M60 were included in all multivariate models (a priori risk factors). Other exposure variables, with p ≤ 0.20 in univariate analysis, were also included in the initial multivariate model. The nal model was built using backwards elimination, comprising a priori risk factors and other exposure variables with p ≤ 0.20. All statistical analyses were performed using STATA © 13.0 (Stata Corp, College Station, TX).

Study Population
Among the 210 HIV-infected children included in phase 2 of the Pediacam study, 149 who initiated ART no later than the age of one year were alive at two years and considered for this analysis (Figure 1). Of the 61 HIV-infected children who were not included in this study, 12 where deceased before ART, 5 were lost to follow-up or relocated before ART, 3 were placed on ART after age 12 months, 38 deceased within the rst two years of ART and 3 were lost to follow-up before 2 years of ART initiation. Among the 149 children (82 boys) followed-up and still alive at two years after ART initiation, with ART started during the rst year of life, three quarters (73.9%) were enrolled in the two Yaounde sites ( At two years after ART initiation, most of the children (77.2%) had achieved a con rmed VS at least once (43) and a viral load measurement was available for 134, of whom 67.2% were < 400 copies/mL («controlled viral load»). Among the 15 other children, the last available viral load was ≥ 400 copies/mL for 12.
We compared the characteristics at two years of treatment according to a viral load below or above 400 copies/mL by grouping together the children whose viral load was ≥ 400 copies/mL or not measured. The children with VS at two years of treatment were more often girls than boys (62.2% versus 37.8%, p = 0.017) and almost all (96.7%) had achieved con rmed virological success (CVS) at least once before two years of treatment, whereas this was true for only 45.5% of the other children (p < 0.001).

Mortality and retention in care between two and ve years after ART initiation
Among the 149 children, alive at two years of treatment, included in this study, a viral load measurement was available at M60 for 121 (84.0%) and 23 (15.4%) did not have viral load measurements at ve years of ART with no information concerning death (Table2). Of these, 12 (8.1%) were lost to follow-up before M24. Five (3.4%) deaths were recorded M24 and M60, corresponding to a death rate of 3.0% (95%CI: 0.2% -5.8%) estimated using the Kaplan Meier method. Among them, two occurred in children with a viral load ≥ 400 copies/mL and three in children whose viral load was unavailable at two years of treatment, whereas there were no deaths among children with VS at two years of treatment.

Immunological status at ve years of antiretroviral treatment
Among the children included in this study, almost three quarters had a high CD4 percentage level (≥25%) at 5 years of ART (Table 2).

Virologic control at ve years of antiretroviral treatment
Virologic control at ve years of ART was estimated at 66.8% (60.1 -73.5) among the 144 children still alive after ve years of treatment but 23 (16.0%) did not have an available viral load measurement at that deadline ( Table 2). Among the children who were virologically controlled (viral load < 400 copies/mL) at two years of treatment, the probability of maintaining virologic control at ve years, estimated using the Kaplan-Meier method, was 64.0% (54.0 -74.0), (Table 2 and gure 2). Of those whose viral load was uncontrolled or unknown at two years of ART, the probability of achieving a viral load < 400 copies/mL between two and ve years was 76.0% (63.0 -89.0).
Genotypic resistance test of HIV to antiretroviral drugs was performed for a sample of 25 children whose viral load remained high (≥1000 copies/mL) for at least three consecutive medical visits during the study. Of these, 15 (60.0%) were boys and 18 (72.0%) have initiated lopinavir-based ART. The distribution of sex and type of ART in this sample was comparable to the study population. Among these children, 5 (20.0%) rebounded to viral load ≥1,000 copies/mL during the study period for at least 3 consecutive visits before suppressing to viral load <400 copies/mL no later than ve years of ART. Overall, respectively 14 (56.0%), 9 (36.0%) and 2 (8.0%) of these children had no resistance, only lamivudine (3TC) resistance (not altering the effectiveness of current ART regimen) and resistance to non-3TC drugs (altering the effectiveness of current ART regimen). The resistance pattern did not differ by whether the child was suppressed or not at 5 years of ART, according to the Fisher exact test (p=0.440).

Factors associated with virological success at ve years of antiretroviral treatment
In univariate analysis (Table 3), VS at ve years of treatment (viral load < 400 copies/mL versus viral load ≥ 400 copies/mL or not measured) was associated with female gender (OR = 2.1 (1.0-4.4), p = 0.041) and achievement of con rmed virological success (at least two consecutive viral loads < 400 copies/mL during the rst two years of treatment) at least once within the rst two years of ART (OR = 3.0 (1.3-6.9), p = 0.010). There was a non-signi cant association between VS at M60 and controlled viral load (VL< 400 copies/mL) at two years of ART initiation (OR= 2.2 (0.9-4.8), p = 0.061), being followed at "Centre Hospitalier d'Essos" versus "Centre Mère et Enfant de Yaoundé" and "Hôpital Laquintinie de Douala" (OR=2.8 (1.0-7.6), p=0.056) and absence of missed doses during the last three days before the M60 visit (OR = 2.8 (0.9-8.7), p = 0.076). In multivariate analysis (Table 3), the only factors associated with VS at ve years of treatment was achievement of con rmed virological success at least once within the rst two years of ART (adjusted OR = 2.7 (1.1-6.8); p = 0.033) and the trend between VS and absence of missed doses during the last three days before the M60 visit remained (adjusted OR = 2.7 (0.9-10.4), p = 0.075).

Discussion
The proportion of VS among children still alive and followed-up at ve years of treatment, initiated before the age of 12 months, was 66.8% (60.1 -73.5), comparable to the results reported for Italian children who initiated ART early (44), and adult populations (45). However, this result contrasts with a lower proportion of VS reported in a study conducted on children in Ivory Coast who initiated ART at a higher median age (36 months versus 4 months in our study) (46).
Among children alive at two years of treatment, 23 (15.4%) did not have viral load measurements at ve years and there was no information concerning death, despite active telephone recalls of non-returning families (47). The low mortality beyond the rst two years of antiretroviral treatment (3.3% (95%CI : 0.4 -6.2)) is coherent with the results of other studies conducted in children and adults, for which most of the deaths occurred during the rst year of ART (9,28,45,(48)(49)(50). The high mortality observed in the cohort before age 2 years of ART may re ect the rapid progression of the disease in infants enrolled in medical care after severe immunode ciency had set in (43).
The probability of maintaining VS between two and ve years of ART, estimated at 64.0% (54.0 -74.0), highlights the di culty of parents to administer drugs daily to their children (43). The only factor independently associated with a controlled viral load (< 400 copies/mL) at ve years of treatment was the existence of VS at two years of treatment, irrespective of living conditions. The association with VS at two years of ART may be related to factors that favor long-term family adherence to ART or favorable immunological or virological factors in these children (51).
One of the strengths of our study was the use of data from a prospective cohort of children, with scheduled determinations of viral load at six-month intervals, making it possible to use survival models. The current missed doses self-reported by the child caregiver may not be a good stool for detecting virological success at 5 years of ART since it does not re ect the longitudinal adherence evolvement. However, it has been demonstrated previously that the negative predictive value of current missed dose in detecting virological success (36) was as satisfactory as cumulative missed doses (around 75%). In fact, virological response seems to be mostly in uenced by adherence status within the last three months, especially when there is no drug resistance. The non-signi cant association between the absence of missed doses during the last three days before the M60 visit and virological success at this deadline would become signi cant if the study population size was larger. This hypothesis would con rm the di culty of parents to administer daily medications to their children, given the di culty in this study to maintain virologic suppression in children who had initiated ART early during infancy. We did not obtain exhaustive genotyping results for all the children in this study. However, the HIV-resistance patterns were similar, irrespective of virological status at ve years of treatment, among the 25 children who were tested for genotyping. However, it is not possible to draw conclusions on the effect of HIV resistance mutations on the virological response at ve years of ART early initiated during infancy because the study was not su ciently powered to address this issue.

Conclusions
This study reported di culty in maintaining virologic success in sub-Saharan HIV-infected children between two and ve years of early initiated ART. It also highlighted the importance of initial viral suppression for achieving and maintaining virologic success in the long-term. Further studies should focus on an exhaustive assessment of long-term HIV resistance pro les in children who initiated ART early during infancy.

Declarations Ethics approval and consent to participate
The ANRS-Pediacam study was granted ethical approval in Cameroon by the National Ethics Committee and in France by the Biomedical Research Committee of the Pasteur Institute of Paris. Parents signed written informed consent for participation in the study.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors have no con icts of interest relevant to this article to disclose.

Funding
No funding was provided for this study.
Authors' contributions FAN participated in the study design and conception, the recruitment and follow up of HIV-infected infants, the data analysis and the manuscript writing. MCT participated in the study design and conception, the data analysis, the manuscript writing and the study coordination. CIP participated in the study design and conception, the recruitment and follow up of HIV-infected infants and the manuscript writing. STN, JAN and GG participated in the recruitment and follow up of HIV-infected infants and the manuscript writing. AK participated in the study design and conception, manuscript writing and the study coordination. CLS participated in data analysis and the manuscript writing. AF and JW participated in the study design and conception, the data analysis, the manuscript writing and the study coordination. PAT-N and PM participated in the data analysis and the manuscript writing.
All authors approved the nal manuscript as submitted and agree to be accountable for all aspects of the work.    : Variables excluded from the multivariate analysis because of collinearity with virological status at 2 years after treatment initiation.