Immunovirological response to combined antiretroviral therapy and drug resistance patterns in children: 1- and 2-year outcomes in rural Uganda

Background Children living with HIV continue to be in urgent need of combined antiretroviral therapy (ART). Strategies to scale up and improve pediatric HIV care in resource-poor regions, especially in sub-Saharan Africa, require further research from these settings. We describe treatment outcomes in children treated in rural Uganda after 1 and 2 years of ART start. Methods Cross-sectional assessment of all children treated with ART for 12 (M12) and 24 (M24) months was performed. CD4 counts, HIV RNA levels, antiretroviral resistance patterns, and non-nucleoside reverse transcriptase inhibitor (NNRTI) plasma concentrations were determined. Patient adherence and antiretroviral-related toxicity were assessed. Results Cohort probabilities of retention in care were 0.86 at both M12 and M24. At survey, 71 (83%, M12) and 32 (78%, M24) children remained on therapy, and 84% participated in the survey. At ART start, 39 (45%) were female; median age was 5 years. Median initial CD4 percent was 11% [IQR 9-15] in children < 5 years old (n = 12); CD4 count was 151 cells/mm3 [IQR 38-188] in those ≥ 5 years old (n = 26). At M12, median CD4 gains were 11% [IQR 10-14] in patients < 5 years old, and 206 cells/mm3 [IQR 98-348] in ≥ 5 years old. At M24, median CD4 gains were 11% [IQR 5-17] and 132 cells/mm3 [IQR 87-443], respectively. Viral suppression (< 400 copies/mL) was achieved in 59% (M12) and 33% (M24) of children. Antiretroviral resistance was found in 25% (M12) and 62% (M24) of children. Overall, 29% of patients had subtherapeutic NNRTI plasma concentrations. Conclusions After one year of therapy, satisfactory survival and immunological responses were observed, but nearly 1 in 4 children developed viral resistance and/or subtherapeutic plasma antiretroviral drug levels. Regular weight-adjustment dosing and strategies to reinforce and maintain ART adherence are essential to maximize duration of first-line therapy in children in resource-limited countries.


Background
Despite the progress made in the last few years, in 2009 only 356,400 of the estimated 2.5 million children currently infected with HIV were receiving combined antiretroviral therapy (ART) in low-and middle-income countries [1]. Ninety percent of children infected with HIV were living in sub-Saharan Africa and the ART coverage in this area was 26% [2]. In the absence of therapy, more than 50% of HIV-infected children die before the age of 2 years [3]. Therefore, there is an urgent need to increase access to treatment and care and to monitor the effectiveness of therapy.
Comprehensive evaluations of the effectiveness of generic fixed-drug ART in children come mainly from studies conducted in high-income countries [4][5][6][7][8] and, to a lesser extent, from urban areas of resource-poor settings [9,10]. Assessments conducted in rural, resource-limited settings, where programs face important challenges to provide pediatric HIV care, are needed to adapt and improve existing treatment strategies.
In this study, we evaluated pediatric care provision in an HIV/AIDS program in rural, northwestern Uganda. Specifically, we describe survival and program retention, and, among children alive and on ART, patient adherence to therapy and presence of antiretroviral (ARV)related toxicity. Also presented are plasma non-nucleoside reverse transcriptase inhibitor (NNRTI) concentrations and CD4 and virological responses, including resistance patterns, observed in children with detectable HIV viral load.

Arua Hospital AIDS Care Project
The Arua Hospital AIDS Care Project in Arua, Uganda has provided free treatment and care to patients living with HIV/AIDS since July 2002. Eligibility criteria for ART, patient management, and treatment are based on World Health Organization (WHO) guidelines for scaling up ART in resource-poor settings [11,12].
Nurses or clinical officers monitored the children every 2 to 6 months to determine clinical stage and diagnosed and treated ARV-related toxicity or intercurrent diseases. CD4 testing was performed at initiation and every 6 months thereafter (every year after 2005). No routine viral load monitoring was performed. Adherence counseling focused on parent/caregiver education. No specialized child psychological support or specific training for pediatric clinical management was offered at that time.

Study Population
We retrospectively analyzed two observational cohorts of children treated with ART for 12 ± 2 months (M12 cohort) or 24 ± 2 months (M24 cohort). Children alive still followed on treatment were eligible to participate in the cross-sectional survey, conducted between November 2005 and May 2006.

Study Procedures
We used a standardized questionnaire to collect sociodemographic information, WHO clinical stage data, weight, height, ART information (history of ARV use before enrollment, date of therapy start and regimen, current regimen), reported clinical ARV intolerances (asthenia, lipodystrophy, or gastrointestinal, cutaneous, and neurological symptoms), presence of ARV-related morphological disorders, and adherence to treatment. Two indicators of adherence as reported by parents/ caregivers were used: i) percentage of pills taken in the last 4 days (number of pills taken divided by the total number of pills prescribed during the previous 4 days); and ii) percentage of adherence during the last 30 days using a 6-point visual analogue scale (VAS; 0 meaning no medication taken, and 6 all medication taken). For both indicators, poor adherence was defined using a threshold of < 95% [11], moderate adherence as 95-99%, and good adherence as 100%.
Hemoglobin level, platelet count, neutrophil fraction, plasma creatinine level, and transaminase level were measured in the children. Proteinuria and glucosuria was also determined using a urine dipstick technique. Severity of laboratory-based ARV toxicity was graded according to WHO guidelines [14]. CD4 counts were measured using either semi-automated (Cyflow counter, Partec, Münster, Germany) or manual (Dynabead, Dynal Biotech SA, Compiègne, France) techniques. HIV RNA testing was quantified with the automated TaqMan realtime reverse transcription-PCR (RT-PCR) assay (limit of detection 400 copies/mL) [15].
Samples of children with virological failure (viral load ≥ 1,000 copies/mL) were tested for genotypic resistance. Resistance mutation determination was based on the International AIDS Society Resistance Testing-USA panel and resistant virus defined according to the French ANRS resistance algorithm (http://www.hivfrenchresistance.org) [16,17]. High performance liquid chromatography (HPLC) [18,19] was used to determine NVP or EFV plasma concentrations in blood samples collected 12 hours after the last dose intake.
Signed informed consent was sought from all parents or legal caregivers. The study protocol was approved by the Uganda National Council for Science and Technology, the Ugandan AIDS Research Committee, and the Saint-Germain-en-Laye Hospital Consultative Ethics Committee, France.

Statistical Analysis
Weight-for-height nutritional Z-scores (WHZ) were calculated using WHO growth reference data for children and adolescents [20][21][22]. Normal NNRTI therapeutic ranges (4,000-8,000 ng/mL for NVP and 1,000-4,000 ng/mL for EFV) [23], were used to define three categories of plasma concentrations: low (below the lower therapeutic limit), normal (within the therapeutic range), and high (above the higher therapeutic limit). Immunological failure was defined according to WHO guidelines [24]: CD4 values < 15% in children of < 36 months, < 10% in the 36-59 month group, and < 100 cells/mm 3 in those aged ≥ 5 years.
Children who missed their last appointment for 2 months or more were considered lost-to-follow-up (LFU). Probabilities of survival and care retention were estimated using Kaplan-Meier methods. Patient followup was right-censored at the date of the study visit, death, or last visit for children LFU or transferred to another program. All study data were double-entered and analyses performed in Stata 9.0 (Stata Corp., College Station, TX, USA).

Results
Eighty-seven children had initiated ART 12 ± 2 months (M12 cohort), and 41 children 24 ± 2 months (M24 cohort), before the study inclusion period ( Figure 1). Overall, 6 children died, 11 were LFU, and 9 transferred to another program. A total of 102 children (80% of M12 and 79% of M24) were still alive and receiving ART. Of these, 59 (M12) and 27 (M24) children, respectively, participated in the cross-sectional survey (11 patients could not be found and 5 were screened too late).
At therapy start, two children (2.3%) were ARTexperienced (M12 cohort). Two others had received prevention of mother-to-child HIV infection prophylaxis, one patient received short-course AZT together with single-dose NVP (M12 cohort), and one patient received single-dose NVP (M24 cohort). Seventy (81.4%) children were started on AZT-3TC-NVP. A greater proportion of children had received ARV pediatric formulation in the M12 than in the M24 cohort (72.9% vs. 14.8%).

Characteristics at Survey
Mothers were the main caregivers for 50.8% of children in the M12 cohort but only for 37.0% in the M24 cohort ( Table 2). Twenty-one percent of all children were orphans.

Plasma NNRTI Concentrations
Plasma concentrations of NNRTIs were measured in all children: 81 receiving NVP-based and 5 EFV-based  (Table 2). Only children weighing < 30 kg were found to be underdosed for NVP (37.0% of those who were prescribed 100 mg, 18.8% of those on 200 mg, and none of those on 25 mg of NVP twice daily; Figure 2b). Those weighing 20-30 kg were more frequently underdosed for NVP than other children (51.9% compared to 21.3% for children of < 10 kg; P = 0.005). Furthermore, 16.9% (10/59) of M12 and 7.4% (2/27) of M24 children had high plasma concentrations of NNRTIs (NVP or EFV).

Virological Response and Drug Resistance
HIV RNA suppression (< 400 copies/mL) was observed in 59.3% (35/59) of M12 and 33.3% (9/27) of M24 children (Table 3). Treatment adherence was not associated with either viral suppression (overall Fisher's exact P = 0.42 for VAS and 1.00 for 4-day recall) or presence of low NNRTI plasma concentrations (P = 0.35). Similarly, no association was observed between weight and viral suppression (P = 0.17), and, among children with detectable HIV RNA, weight was not correlated with viral load measurements (overall Spearman's correlation coefficient 0.07, and 0.05 for M12 only).
Genotypic resistance testing was successful in all 34 specimens with viral load ≥ 1,000 copies/mL (Table 3). HIV-1 subtypes were A1 (n = 7), C (n = 1), and D (n = 4), and for 22 children the subtype was not specified. Wild-type viruses were found in 4 patients (3 at M12 and 1 at M24). The most prevalent mutations were M184V (nucleoside reverse transcriptase inhibitors [NRTIs]) and Y181C (NNRTIs). Twenty-nine (85.3%) children had at least 1 major mutation conferring drug resistance to both NRTIs and NNRTIs. Prevalence of

Discussion
In these two pediatric cohorts of children followed in rural Uganda, we found retention rates of 86% and sustained immunological responses 1 and 2 years after ART start. Nevertheless, virological responses were suboptimal, as only 59% of M12 and 33% of M24 children achieved HIV viral suppression, and resistant viral strains were found in 25% and 62% of children, respectively. As in other sub-Saharan African cohorts [25][26][27][28], all children followed in our program were enrolled and started therapy after the first year of life, median age at inclusion being 5 years old. Given that young, untreated children are at high risk of rapid disease progression and death before the age of 2 years [3,[29][30][31][32], the overall impact of HIV/AIDS programs in children living in these settings is likely to be suboptimal. Thus, delay in diagnosis of pediatric HIV infection and lack of availability of integrated HIV care and treatment in maternal and child health services at the time of the survey were likely to be partly responsible for the late start of therapy in our site.
Furthermore, as in other pediatric cohorts in resourcelimited settings [26,27,33,34], children in our program started therapy at advanced clinico-immunological stage and the majority of deaths and LFU occurred early in treatment. Delayed diagnosis of HIV infection and failure to diagnose and treat concurrent life-threatening infections such as tuberculosis, pneumonia, or sepsis [35] might be responsible for the early losses reported. Starting ART before 12 weeks of life has recently been associated with a 76% reduction in child mortality compared to deferred therapy until immunological or clinical progression is diagnosed [36]. Therefore, some of the early losses in our cohort could have been prevented if ART was started earlier. Despite this, we observed overall high and sustained survival and retention rates in children treated with ART up to 2 years after therapy start. These findings are consistent with reports from other programs from resource-limited settings [10,25,37].
Although direct comparison of immunological responses across studies is not possible because of its high dependence on patient age [38], we observed significant CD4 T-cell responses to ART independent of the virological status of the patient, and, as previously reported [27,28,38,39], greater increases were seen in the first than in the second year of therapy among children aged ≥5 years (gains of 206 vs. 132 cells/mm 3 ). The small sample size in our survey must be considered when interpreting these findings. The small numbers of patients did not allow investigating risk factors for virological failure. Estimates reported in other studies range from 49% in Abidjan, Côte d'Ivoire, to 80% in 63 children exposed to single-dose NVP in Durban, South Africa, at one year of therapy [9,27,[40][41][42]. In the urban studies in Côte d'Ivoire and South Africa, 49% virological suppression was reported at 2 years of ART [9,27]. Large differences in virological response rates have been observed between clinical studies among children, and the majority showed lower rates than in adults [38]. Possible explanations for such discrepancies could be differences in patient characteristics at ART initiation, regimens used, pediatric dose formulations, drug pharmacokinetics, and/or inadequate compliance to treatment.
As mentioned before, most children in our cohort started treatment at an advanced stage of disease. Presence of high HIV RNA at therapy start has been associated with slower decay rates of plasma viral levels and longer time to reach undetectable viral levels [42][43][44][45]. Increased risk of virological failure has also been reported in children with CD4 < 5% [40]. Our patients received NNRTI-based therapy, and most were given NVP. Higher rates of virological success have been shown in young patients treated with protease inhibitors than in those using NNRTI regimens in some [9] but not all studies [27,46]. Furthermore, virological response has also been shown to vary in clinical studies using the same medication [38].
Ensuring adequate administration of ART to children is difficult and requires continuous dose adjustment in Ensuring and maintaining good treatment adherence in children is also challenging, especially in resourcelimited settings, since many children are orphans or live in difficult social situations: 1 in 5 patients were orphans in our study, and 43% had a caregiver other than a parent. Furthermore, palatable syrups or appropriate pediatric tablet formulations are expensive and not broadly available. In our study, 13% of children were identified as poorly adherent to ART, but this is likely to be underestimated since it was based on parental or caregiver reports. Four children with virological failure had wild-type virus and had therefore not been taking any treatment for some time. Furthermore, the prevalence of resistance mutations was relatively high but similar to those reported by a previous study in Côte d'Ivoire [26]. A recent review of studies investigating ART adherence in children treated in low-and middle-income countries identified difficult familial situations and low socioeconomic status, absence of parental and/or child HIV status disclosure, complicated regimens, and drug-related adverse events as barriers to treatment adherence [47,48]. Age-adapted therapeutic education of children and extensive discussions and support to parents/caregivers need to be encouraged to improve ART uptake in this vulnerable group.
As reported by others [38,49], we did not observe severe adverse drug reactions at the time of the survey, apart from one patient with severe neutropenia. However, 3 of 7 children reported gastrointestinal symptoms, which might interfere with drug absorption and affect treatment adherence [50]. Peripheral neuropathy and morphological disorders were not uncommon (38% and 21% of children, respectively). One longitudinal and one cross-sectional study also reported relatively high rates of lipodystrophy in children treated with d4T or protease inhibitors (29% and 33% of children, respectively), especially in those who started therapy at an advanced stage of disease [51,52]. This syndrome is frequently associated with metabolic abnormalities such as hyperinsulinemia and dyslipidemia. However, its causal link to ART is to be established, and its long-term consequences for the treatment of children are unknown [52].

Conclusions
Important challenges need to be tackled to improve HIV pediatric care in resource-limited settings. These include increasing early HIV diagnosis through programs for the prevention of mother-to-child HIV transmission, training in pediatric clinical management and adherence counseling, and development of palatable and simplified pediatric drug formulations.  Authors' contributions LA designed and implemented the study, analyzed and managed data, interpreted results, and wrote the manuscript. GG, PA, and CO set up and implemented the study in the field and contributed to the interpretations of results. LP prepared the study's EpiData database and performed data management. CR and A-MT performed the virological and genotypic testing and contributed to interpretation of these results. SB and DMO helped to design the study, interpret the results, and draft the manuscript. MP-R managed the data, performed statistical analysis, interpreted results, and cowrote the manuscript. All authors read and approved the final manuscript.