Management of children with HIV/AIDS is specially challenging. Although antiretroviral (ARV) treatment may lead to viremia control and clinical benefit comparable to adult patients, time for treatment initiation, therapy change strategies, along with other age-related issues, pose additional challenges, adding to higher risk of ARV failure among children.1 For children with vertically-acquired infection, acute infection occurs during the neonatal period, a time characterized by exceedingly high viremia levels that decline, albeit more slowly than in adults, over the next two or more years to reach a set point. This set point tends to be higher in children, as compared to adults, and is consistent with more rapid disease progression.2–4Although the goal to eliminate HIV maternal-to-child-transmission (MTCT) is a current priority for international agencies, about 400,000 children are infected annually, and pediatric AIDS remains a major public health issue in many areas. Moreover, for patients who receive adequate care and treatment, life expectancy is extended into adulthood. A better understanding of the pathogenesis of children with HIV/AIDS is instrumental, as the clinical evolution in this population might differ from patients that acquire HIV with a mature immune system.

Over the years, various clinical and laboratory markers of disease progression have been proposed for adults and children, although the determinants for disease progression, especially in children, are unclear. The CD4+ T-lymphocyte count (CD4) and viral load (VL) are the primary markers that have been used in different studies to initiate and monitor antiretroviral therapy (ART) in clinical practice.5–9 Additional parameters could improve clinical decision-making, especially those that can predict disease progression. Viral envelope characteristics have been associated with HIV pathogenesis since the beginning of the epidemic.10 To initiate cellular infection, HIV needs to use the CD4 receptor along with a chemokine coreceptor, mostly CCR5 or CXCR4. Based on the coreceptor usage, HIV-1 variants may be classified as R5-tropic, X4-tropic or dual-tropic (DM). While most primary infections involve viruses that use CCR5 as a coreceptor, CXCR4-using viruses are often identified in adults with advanced disease and its identification has been associated with rapid progression.11–14 Viral tropism, although a phenotypic characteristic, can be predicted with genotypic methods. These tests have greater accessibility, lower cost, and shorter assay time than phenotypic methods.15,16

The identification of surrogate markers that provide clues to the clinical outcome may be useful to subsidize clinical decision. The aim of this study was to evaluate HIV tropism genotypic prediction from the first available clinical sample and its association with disease progression in a clinical cohort of children and adolescents with HIV/AIDS in São Paulo, Brazil.

The main purpose of this study was to document clinical and laboratory evolution of this cohort to evaluate if tropism determination can be useful as a marker of disease progression. For this purpose, a longitudinal follow-up of HIV-1-infected children and adolescents was carried out for over 10 years, documenting known predictors of disease progression, such as CD4 counts and adherence to treatment. One major limitation of the study is the fact that tropism determination was not performed at study entry, but sometime during follow-up, with some cases determined at the sample available near last observation. Therefore, the study can only suggest potential associations, and cannot claim that the predicted tropism determined the outcome evaluated. Also, as in other longitudinal studies in clinical settings, different factors, such as assiduity of patients, parents or guardians, proper collection and storage of the samples, documentation of clinical and laboratory data throughout the follow-up impact data collection, and missing data compromise a robust support to our findings. Although clinical trials provide a stronger basis for the development of cohort studies, the ability to grasp real world dynamics makes clinical setting observations also relevant.

Brazil has the largest pediatric population with HIV/AIDS with public access to both monitoring tools, as well as ART. Little work, however, is available on the long-term evaluation of these patients, some already initiating follow-up in adult services.

Patients in this study were followed up at a single clinical site in São Paulo, the largest city in South America. The service studied is not a specific reference center for AIDS or for infected pregnant women and therefore receives sick children that end up finding their HIV diagnosis during the evaluation at entry. Most children observed without ART at entry were brought to the service due to some clinical condition and initiated ARV shortly after this initial observation. This might have caused an inclusion bias toward more severe cases. Another issue is the fact that the service cares for patients without insurance and therefore includes a significant proportion of children from lower income strata. However, the population of the study had some degree of homogeneity, such as access to free ARV, socio cultural background and the fact that all were treated by the same clinical team, diminishing the variability in clinical management. Because of limited numbers of infected children and of mothers with relevant data, we were unable to investigate effects of other factors possibly associated with disease progression in vertically infected children, such as maternal ART use and maternal VL.

Children in clinical services are followed based on clinical condition, routine laboratory workup, and two major HIV-related prognostic markers, CD4 and VL. Although clinical evolution will be influenced by different biological, socio cultural and access to health-related issues, it is conceivable that certain viral characteristics may influence this outcome. Viral tropism has been suggested to be an additional marker of disease severity in adults.13

Although viral tropism is more properly evaluated using phenotypic assays, genotypic methods allow for a reasonable prediction, and have been increasingly used. Moreover, it was considered sufficient by the European Consensus Group to define tropism for the usage of CCR5 antagonists.14 In our study, we did not try to evaluate the tropism assay itself, but rather correlate the genotypic prediction with the actual clinical and laboratory outcomes of these patients. To that end, we analyzed envelope sequences at the geno2pheno website, a commonly used resource that provides a FPR for the probability of misclassifying a sequence as X4. Therefore, the lower the FPR, the lower the chance of false X4 classifications. As different clinical cutoffs have been used, we evaluated our dataset with the three most commonly used cutoffs.

This study showed a relatively high proportion of X4-predicted variants, 45% if a more conservative (20%FPRclonal) cutoff is used. Some previous studies reported similar prevalence in chronically infected children.20,21 One study with HIV-1 subtype C infected children, observed a lower presence of X4,22 similar to a study in Uganda with different subtypes,23 whereas others found a higher X4 prevalence.24 The differences in study design, population, age of participants, tropism definition and viral subtype may all contribute to these discrepancies. Clade C cases are considered to be preferentially R5 isolates, and in the small subgroup of clade C infection in our study we also found only R5-predicted variants.

There was a significant correlation between a lower FPR and older age in our population. Other studies20,25 also found that correlation and that tendency is in accordance with most of the evidence in adults that suggests the emergence of X4 variants later, as the disease progresses.12,26 In adults, after a mostly R5 infection, about 50% of HIV infected patients evolve to X4 tropic variants later in disease.13

Suppression of viremia is the main goal of HAART. At last observation, only 30% (22/73) of the patients achieved aviremia. This figure is not unusual for children initiating therapy before the HAART era, as is the case of many of these patients. Response rates are better for children initiating HAART with new regimens (data not shown). From the data available upon entering follow-up, being older and having a R5 virus was associated with virologic suppression in multivariate analysis (Tables 2a and 2b). However, the association of R5 with viral suppression is weak and not sustained when variables obtained during follow-up are included. When follow-up information is considered, the association of aviremia with age becomes less striking and adherence is the only strong predictor of aviremia. It is well known that a good adherence to ART is the central determinant of viral suppression and long-term disease-free survival. This is a difficult goal, particularly in very young children who rely on parents or caregivers to administer drugs and in adolescents, who are more likely than younger children either to report poor adherence themselves or to have it reported by caregivers and/or health professionals.27

The association of age with viremia suppression may be due to an easier adaptation of older children to ARV and to the bimodal nature of HIV disease in children, as pointed out by Puthanakit et al.28 in their study of treatment initiation in older children. Therefore, children who do not present clinical disease early after birth might already represent a population of better clinical outcome. Similarly, early deaths in the first year may represent either more aggressive disease or cases that reached the service at an advance stage, in which clinical intervention was of limited impact.

We did not find an association between DRM and viral suppression. It was not the objective of this paper to evaluate this issue, and resistance was considered here as one of the possible variables observed during follow-up that should be used in adjusting the multivariate analysis. The resistance profile and viral diversity of pol gene in this population have been previously published.29,30 The lack of correlation of DRM with clinical severity or viremia is possibly due to the fact that DRM is generally absent in samples from patients with adequate adherence, aviremic cases, as well as from viremic, poorly controlled, low adherence patients, as only those submitted to sufficient drug exposure would develop genetic mutations. Nevertheless, the genotypic test was helpful in guiding salvage therapy in many patients (data not shown).

Contrary to viral suppression, CD4 levels, as well as a composite outcome combining both CD4 and clinical condition, showed a strong and independent association with viral tropism. Entry CD4 (CDC category) showed a tendency in predicting a less severe outcome (p=0.08). However, viral tropism, specially using the 10%FPRclonal cutoff shows an independent, strong association with AD (Tables 2a and 2b) when only information available at the study entry is considered. When follow-up information is included, ARV adherence and R5 viral tropism are both significantly associated to AD. We also found a strong, significant correlation between FPR and CD4 nadir values (Fig. 1). There is evidence that immunologic deterioration in HIV-1–infected children precedes the viral phenotypic switch from R5 to X4 variants31 and that the presence of X4 variants is linked with faster progression to AIDS, a rapid fall in CD4 count26 and a lower CD4 nadir. This association of viral tropism with CD4, but not with viremia control, has been observed in some studies with adult patients,13 and we and others32,33 have found an association of a lower geno2pheno FPR with CD4 depletion in adults.

Since some of our cases had insufficient follow-up time after tropism determination it is not possible to exclude the possibility that the predicted tropism was secondary to clinical deterioration. The use of tropism prediction information, in a clinical setting context, may assist pediatricians in planning a more aggressive intervention in cases with more severe prognosis, maximizing resources available for populations with increased risk of disease progression. Current Brazilian treatment guidelines indicate treatment for children with clinical disease, VL above 100,000copies/mL or CD4 below 25%. Although the efforts in improving adherence are a cornerstone of treatment success, information on viral tropism, along with other prognostic markers, could be useful to support clinical decision.

In conclusion, our study documented for the first time a strong predictive association, independent from other available markers, that a high FPR, predicting a CCR5 coreceptor usage, is protective in children and is related to a better immunological (CD4 level) and clinical outcome.

The authors declare no conflicts of interest.