Virological and Immunological Long-Term Outcome of Human Immunodeficiency Virus-1 Infected Children Treated before One Year and after Two Years of Age in a Resource-Limited Setting of South Africa

Introduction : Benefits of early Highly Active AntiRetroviral Therapy (HAART) to reduce infant mortality and morbidity have been demonstrated in resource-limited and rich settings. However, immunovirological data collected in Sub-Saharan Africa are scarce. This study describes the long-term outcome of South African children who started HAART before one year of age (Early Starters Cohort or ESC) and compare their immunovirological outcomes to children who started their therapy after two years of age (Late Starters Cohort or LSC). Immunovirological results will be compared in order to evaluate the long-term non-inferiority of early treatment initiation. Methods : Fifty-five children were included in the ESC (mean follow-up period 7.9 years) and 96 children were included in the LSC (mean follow-up period 6.3 years). Children from the ESC and the LSC were subdivided into three subgroups according to CD4 + % at HAART initiation (<15%, between 15-24% and ≥ 25%). Results : All children included in the ESC achieved normal CD4 + % at least once during the entire follow-up period, contrary to the LSC where 89.6% children achieved normal CD4 + % ( p =0.014). Furthermore, mean CD4 + % became higher in the ESC six years after treatment initiation. Children with CD4 + % between 15 and 24% at HAART initiation reached higher CD4 + % in the ESC, three years after treatment initiation. The proportion of children who experienced virological failure (>1000 cp/ml) was comparable in both cohorts but persistent undetectable viral load (<50 cp/ml) after initial virological suppression was more frequent in the ESC ( p =0.008). Finally, the proportion of children with detectable viral loads (50 to 1000 cp/ml) at least once after initial virological suppression was higher in the LSC ( p =0.0022). Conclusion : HAART appeared highly effective in terms of immunovirological outcomes both in children treated before one and after two years of age. The results of this study demonstrate that early treated children more often achieved normal CD4 + %, tended to have higher mean CD4 + % and more sustained virological suppression. These results encourage the current international recommendations to initiate HAART as soon as possible in RLS.


Introduction
In 2010 the World Health Organization (WHO) recommended an immediate start of Highly Active AntiRetroviral Therapy (HAART) in all patients diagnosed with HIV before 2 years of age [1]. This recommendation was extended to children younger than 5 years of age in 2013 [2] and to all children and adults in 2015 [3]. Benefits of early treatment initiation on infant mortality [4,5], neurodevelopmental outcome [5,6], growth recovery [7], immunologic restoration and virological suppression [8,9], have been demonstrated in short term follow-up studies implemented in resource-limited settings (RLS). A systematic review including 5 African studies, showed a 12 months HIV RNA suppression rate of 70% and a CD4 + cell percentage (CD4 + %) increase of 13.7%, for children initiating HAART in RLS [10].
The aim of this study is to describe the long-term outcome in a real life setting of children who started HAART before 1 year of age (early starters cohort or ESC) and after 2 years of age (late starters cohort or LSC) in the same RLS of KwaZulu-Natal, South Africa (SA). Immunovirological results will be compared in order to know if early treatment initiation, which is a major advantage in a short-term point of view, could lead to a poorer long-term evolution in terms of CD4 + % and viral load (VL).

Method Study population and setting
This non-inferiority retrospective study was conducted in SA Virological and Immunological Long-Term Outcome of Human Immunodeficiency Virus-1 Infected Children Treated before One Year and after Two Years of Age in a Resource-Limited Setting of South Africa between September and December 2014 at Edendale Regional Hospital located in a peri-urban and RLS of KwaZulu-Natal. The overall prevalence of HIV is the highest of SA with 16.9% of people infected [18].
In total, hundred fifty-one children were included in the present study in the ESC and the LSC.
Children in the ESC were included in a previous study about feasibility and effectiveness of early HAART initiation in RLS, which took place between 2005 and 2008 [19]. First condition to be included in this previous study was to be younger than 1 year of age at HAART initiation (median 8.6 months, range 2.1-11.9 months). Other criteria to initiate HAART included positive HIV-1 DNA PCR, WHO stage 2-4 and/or CD4 + % <30% [19].
Children included in the LSC were invited to participate to the present study during a planned appointment in the HIV clinic. Inclusion criteria were to have initiated HAART after 2 years of age in the HIV clinic and to be vertically infected according to HIV status of the mothers. Age at HAART initiation had to be mentioned in the patient's file. These children were not included in the previous study between 2005 and 2008 because they were older than 1 year during this period or because they started HAART after 2008. They initiated HAART in function of their WHO clinical stage, CD4 + % or comorbidities following the successive South African guidelines between 2005 and 2012.
Children from both cohorts came from the same region with the same socio-economical environment. They were followed-up by the local staff between 2008 and 2014 and were treated following the national recommendations.

Study procedures
Parents or caregivers gave a written informed consent prior any procedure. Data were collected from the medical files (before 2010 records on paper and later on computer). The study was approved by the South African ethical committee «Umgungundlovu Health District Review Board (UHERB)» and the Belgian Ethical committee «Comité d'Ethique Hospitalo-Facultaire Saint Luc -UCL».
Virological failure was defined as a VL above 1000 copies/mL based on the annual blood check.
Viral suppression was obtained when VL was less than 50 copies/ mL.

Laboratory procedures
Blood samples were analyzed during routine consultation at the Edendale Hospital laboratory. CD4 + count was analyzed by flow cytometer FC 500 series ® from Beckman Coulter (Paris France) and expressed as % of leucocytes and absolute value. VL was measured by real time PCR with the "COBAS ® AmpliPrep/COBAS ® TaqMan ® HIV-1 Test, v2.0 (Roche Diagnostics)". The lower limit of quantification (LLOQ) was 20 copies/mL.

Current South African antiretroviral treatment guidelines
All infants and children under 3 years of age (or <10 kg) are treated with Abacavir (ABC), Lamivudine (3TC) and Lopinavir boosted ritonavir (LPV/r) when they initiate their treatment. Children over 3 years of age are treated with ABC, 3TC and Efavirenz (EFV) [20]. Before 2010, guidelines recommended to start treatment with Stavudine (d4T) instead of ABC [21].

Statistical analysis
Immunological and virological mean values were compared with the Student t test. Proportions between the ESC and the LSC were compared with the exact Fisher test. Differences in CD4 means were also tested with the non-parametric Wilcoxon test, which resulted in p-values similar to the ones obtained with the t-test.

Study population
From the 94 subjects described in the 2005-2008 study [19], 55 were included in the present research (23 females and 32 males). The remaining 36 patients were transferred to other centers and lost to follow-up (33/36) or died (3/36). Mean age at inclusion was 9.6 years (range 7.1-10.5 years) with a mean follow-up period of 7.9 years (range 6.8-9.8 years).
The control group (LSC) included 96 subjects (42 females and 54 males) who initiated HAART after 2 years of age (median 4.3 years range 2.0-7.6 years) between 2005 and 2012. Mean age was 10.6 years (range 6.6-14.8 years) and the mean follow-up period was 6.3 years (range 2.5-9.8 years).

Initial therapies
In the ESC, 98% of children were initiated on HAART with a regimen including one protease inhibitor (PI) and two nucleoside reverse transcriptase inhibitors (NRTIs) and 2% with one nonnucleoside reverse transcriptase inhibitor (NNRTI) and two NRTIs.
In the LSC, 12% of children were initiated on HAART with a regimen including one PI and two NRTIs (children between two and three years of age at treatment initiation) and 88% with one NNRTI and two NRTIs (children older than three years at treatment initiation) ( Table 1). Only 5 children were followed-up after 8 years in each cohort; therefore the results are not shown for any longer period of time.

Switch of therapy
A large proportion of children have changed treatment in both cohorts mainly due to revisions of the national recommendations. ESC and LSC replaced d4T by another NRTI (ABC, AZT or TDF) in 78% and 52% of patients respectively regardless of virological failure.
Switch to a second line of HAART due to virological failure was not significantly different between the ESC (11%) and the LSC (14%) (p = 0.624). Successive lines of treatment after virological failure are detailed in Table 2.
Among the ESC, one patient replaced AZT for d4T due to anemia and one patient changed from LPV/r to EFV (reason not specified). Among the LSC, one child changed from EFV to LPV/r due to drowsiness (data not shown).

Immunological outcome
Mean CD4 + % at HAART initiation was not significantly different and increased in the same proportions, in the ESC and the LSC during the first five years after treatment initiation (+18.3% vs. +16.2%). The largest increase occurred during the first year of therapy in cohorts (+13.8% vs. +11.9%). Six years after treatment initiation, mean CD4 + % stayed significantly higher in the ESC (Figure 1 and Table 3).
In a second analysis, results were subdivided in three subgroups in function of the CD4 + % at HAART initiation (as explained in the definitions paragraph). First subgroup is the LIS (47% of the ESC vs. 56% of the LSC), second is the IIS (37% of the ESC vs. 32% of the LSC) and third is the HIS (16% of the ESC vs. 12% of the LSC). The proportion of children included in each subgroup was not significantly different in both cohorts (p=0.56, 0.47 and 0.74, respectively). LIS from both cohorts had comparable mean CD4 + % at HAART initiation and during the entire follow-up period (p>0.05). Mean CD4 + % became higher than 25% in both cohorts after two years of treatment ( Figure 2 and Table 4). In the ESC, 26/26 children achieved normal CD4 + % at least once during the entire follow-up period and 47/54 (87%) children achieved normal CD4 + % in the LSC (p=0.089).
IIS from both cohorts had comparable mean CD4 + % at HAART initiation but in children from the ESC it became and stayed significantly higher than in the LSC, three years after HAART initiation. Mean CD4 + % became higher than 25% in both cohorts after two year of treatment ( Figure 2 and Table 4). In the ESC, 20/20 children achieved normal CD4 + % at least once during the entire follow-up period, and 27/30 (90%) children achieved normal CD4 + % in the LSC (p=0.27).
HIS from both cohorts had comparable mean CD4 + % at HAART initiation and during the entire follow-up period (p>0.05) except at six months of treatment (27.1% vs. 42.7%, respectively). Mean CD4 + % stayed higher than 25% in both cohorts during the entire follow-up period ( Figure 2 and Table 4).   The proportion of children who experienced virological failure at least once after the first year of HAART was comparable in both cohorts when the entire follow-up period was considered (p=0.48). However, when every year of follow-up was analyzed separately, proportion of

Virological outcome
Pre-HAART VL was significantly higher in the ESC but the difference in VL waned 6 months after treatment initiation (Table 5).
In the ESC, 17/55 (30.9%) children experienced virological failure at least once after treatment initiation, 25/55 (45.5%) never experienced virological failure and 13/55 (23.6%) had detectable viral loads between 50 and 1000 cp/ml at least one time during the entire follow-up period.      virological failure was higher in the LSC (p=0.038 and 0.014) at six and eight years of HAART, respectively ( Figure 3 and Table 5).
The proportion of children with persistent undetectable VL after initial virological suppression was higher in the ESC than in the LSC (p=0.008).
Finally, the proportion of children with detectable viral loads between 50 and 1000 cp/ml at least one time during the entire followup period was higher in the LSC (p=0.0022).

Discussion
In 2012, only 34% of the HIV-infected children were treated with HAART in the world [2]. The 2013 and 2015 WHO guidelines successively revised and simplified their recommendations to expand treatment in children and increase accessibility to HAART in RLS [2,3]. The present study evaluated the long-term impact and application of these recommendations in the field.
Both cohorts initiated treatment with three antiretroviral drugs as recommended by the WHO and the South African guidelines [2,20]. Children included in the ESC and the LSC initiated HAART with 2 NRTI and 1 PI or 2 NRTI and 1 NNRTI according to age, which means that the guidelines for the first line therapies are clear and correctly applied.
Second line therapies were more heterogeneous in both cohorts probably because different recommended options are available in the WHO [2] (e.g. LPV/r replaced by EFV, d4T+3TC replaced by ABC/ TDF+3TC) and the South African guidelines [20] (e.g. LPV/r replaced by LPV/r+EFV or a 3 rd line, d4T+3TC replaced by ABC+AZT). ddI is no more recommended in both guidelines but is still used in RLS. Finally, expert advice is recommended after first line failure with a LPV/r based treatment. Experts are not always available and the choice of treatment depends more often on the habits of the local staff.
Immunological outcomes were favorable for both cohorts (ESC and LSC) and for the three subgroups (LIS, IIS, HIS), which underlines the importance of initiating HAART regardless of age. Children included in the ESC had at least comparable (or even better) immunological outcomes than children included in the LSC supporting the noninferiority of early treatment initiation, as recommended by the WHO and the South African guidelines.
Some favorable outcomes can be underscored. Before treatment, the ESC and the LSC both had a high proportion (47% vs. 56%) (p=0.56) of children with severe immunosuppression (<15% CD4 + %). This is consistent with a review from Koller et al. who reported 70% of severe immunosuppression in children <2 years of age who started treatment in low and middle-income countries in 2010 [22]. Surprisingly, patients included in the present study had comparable mean CD4 + % independently of age at HAART initiation. It was not the case in European and West African studies [14,23] who showed more severe immunosuppression when treatment was started later in life (-0.82% of mean CD4 + % per year of HAART initiation delay) [14]. This difference could be explained in our study by the exclusion of rapid progressors [24] who possibly died before 2 years (before HAART initiation) in the LSC.
The European-American PENPACT-1 study [31], analyzed combined effects of CD4 + % and age at baseline. They found a significant interaction between CD4 + % and age on the probability of ever recovering a normal CD4 + % within 4 years. With increasing age, baseline CD4 + % had a stronger effect on the capacity to recover a normal CD4 + %. Children included in our three subgroups achieved normal mean CD4 + % independently of their immunological status at baseline. There was no statistical difference in the number of children who achieved normal CD4 + % in the LIS and the IIS, probably due to the small effectives in these subgroups.
In terms of virological outcome, VL was higher in the ESC at treatment initiation. This could be explained by their younger age and was already described in untreated perinatally infected infants, who had the highest HIV-1 RNA levels at 2 months of age followed by a slow decline during the first two years of life. These high VL and slow decline reflect the lower efficiency of the immature immune system, which cannot contain the viral replication and has possibly a greater number of HIV-susceptible cells [32].
The proportion of children, who experienced virological failure at least once after initial suppression, was not significantly different in both cohorts; with 30.9% and 37.5% in the ESC and LSC respectively. A recent retrospective study implemented in another rural area of South Africa found similar rates (38%) of virological failure (median duration since beginning HAART of 31 months) [13].
During the entire follow-up period, the proportion of children who sustained undetectable viral loads was significantly higher in the ESC and the proportion of children who experienced low-level viremia was higher in the LSC. Knowing that persistent low-level viremia is a risk factor of viral drug resistance [33,34], maintenance of an undetectable viral load decreases the risk to develop viral drug resistances in the ESC.

Limitations
This research has some limitations. It is a retrospective study that only includes a limited number of children who survived from 2005 until now and a number of children were lost to follow-up. Data collected at baseline could not be representative of all the infants from this region. Furthermore over the years some data were missing in the patients' files, which decrease the power of the study. However this is an operational study representing the real life on the field. Children included in the ESC and the LSC had, for most of them, different HAART regimens (PI vs NNRTI) at treatment initiation. A PI-based regimen initiation was described to be an advantage in terms of virological outcome in comparison with a Nevirapine-based regimen [35]. Our conclusions could be different if PI or NNRTI-based regimen were used in both cohorts.

Conclusion
HAART appeared highly effective in terms of immunological and virological long-term outcomes both in children treated before one year of age and after two years of age. The long-term results of this study demonstrate that early HAART initiation was certainly not inferior to late treatment. Children from the ESC, treated with a PI-based regimen, more often achieved normal CD4 + % and sustained virological suppression than children from the LSC. These results encourage the current international recommendations to initiate HAART as soon as possible in RLS.