There are two important aspects in this patient cohort, characteristic of ART cohorts in resource-limited settings: a) all patients received the same first-line antiretroviral regimen and b) patients, on average, were in advanced stages of HIV/AIDS when starting their first antiretroviral regimen . When antiretroviral therapy was first introduced in Peru, uniform criteria were established by the MINSA to ensure the allocation of resources and medication to those most in need. This first cohort of patients at the largest children’s hospital in Peru suddenly became eligible for therapy at a time when many had already progressed to disease stages beyond the eligibility threshold.
The effect of delayed access to ART in this first cohort becomes evident in comparison to a recent study observing the transmission of resistant virus in a much younger cohort of neonates and children with timely access to pMTCT and ART in Peru, revealing a predominance of NNRTI mutations, whereas mutations conferring high-level resistance to ARV were still found to be rare . This observation is unlikely an effect of age. Even though our cohort started treatment after the disease had progressed significantly, age by itself was not associated with an advanced clinical stage at enrollment. To the contrary, young age (thus earlier treatment initiation) seemed to favor virologic failure. This may also be due to a survivor effect, i.e. slower progression in those patients who had already survived the first years after MTCT.
Chances of virologic failure were high in this first pediatric cohort gaining access to ART in Peru in 2002/3, with ~44% showing virologic failure after the first year of ART, ~53% after two years. The majority of children were in poor health, as evidenced by malnutrition 93% of children below the reference weight for the respective age group ) and a high prevalence of opportunistic infections. Of note, 43% showed AIDS-defining conditions and 17% co-infections with active tuberculosis. Immunologically, 67% of the children had already r.eached the immunologic CDC category 3 (corresponding to an adult CD4 levels of < 200 cells/μL) prior to gaining access to structured ART.
Immunologic classification at baseline was very predictive for virologic failure. In agreement with studies in industrialized countries [36, 37], these findings indicate that the percentage of CD4 cells in children with HIV/AIDS (i.e. the immunologic category) could be used to guide treatment initiation. In fact, the immunologic classification may be more valuable for the decision of ART initiation than relying on DNA-PCR results alone .
Despite relatively high rates of virologic failure in this cohort, both immunological and clinical conditions improved during ART, in particular throughout the first and second years of treatment. Thereafter little additional improvement was achieved. Overall, from the time of initiation of ART up until the time of the cohort assessment, 57% had showed marked improvement with respect to their clinical status (as measured by CDC category/visit), whereas 35% were unchanged clinically, and only 8% showed disease progression. With respect to the immunologic CDC-scores, 76% had improved, 22% had experienced no change, and 2% showed a decline in CD4 counts.
For improved visualization of the overall development of treatment cohorts during ART, we summarized the clinical and immunological response to therapy in an innovative fashion using a Clinical Course Integrated Display (CCID) with 3-by-4 tables based on the revised CDC clinical and immunological categories . Here, we applied the CDC scores as a flexible tool to examine the cohort on a yearly basis, allowing for CDC scores to improve or deteriorate, according to the CD4 counts and reported clinical symptoms. Using this simple system in cross-sectional analyses and surveillance programs, rates of disease progression (Figure 3) may be computed for different cohorts allowing the comparison of treatment strategies in terms of their clinical and immunologic effects in a given population. This system may be applicable to similar cohort studies in developed and developing countries alike, especially in conjunction with cross-sectional analyses of antiretroviral drug resistance.
Previous exposure to (often incomplete) ART was significantly associated with virologic failure, indicating that short courses of unsupervised ART prior to the initiation of coordinated long-term treatment programs may be counterproductive as they may lead to the rapid development of drug resistance. Archived drug resistance mutations, acquired during previous exposures to antiretrovirals and still present in the PBMC compartment may be detected reliably by DNA OLA.
Exposure of the newborn to post-natal pMTCT with AZT did not increase the likelihood of subsequent virologic failure, neither did maternal exposure to pMTCT with AZT +/− NVP. There are three possible explanations why pMTCT did not affect subsequent treatment success:
The pMTCT did not lead to a transmission/selection and “archivation” of drug resistance,
Although drug resistance against the pMTCT regimen (i.e. AZT +/− NVP) developed and was archived, it did not impede the success of subsequent triple-drug ART consisting of AZT + 3TC + NFV.
Drug resistance did not persist until the initiation of ART.
In fact, in only one child (P028T) we detected archived drug resistance by DNA OLA (mutation 215Y; AZT resistance) at the time of cross-sectional resistance testing. This child (P028T) did not encounter virologic failure (hinting towards scenario b).
Drug resistance in the context of pMTCT may emerge- or be transmitted - by two possible mechanisms:
Drug resistant virus is selected in the mother and passed on to the child (e.g. during birth or breastfeeding).
The newborn is infected with susceptible virus and subsequently selects drug resistant virus, e.g. during ARV exposure.
Ad (i): When a single dose of antivital medication for maternal pMTCT is administered at the onset of labor, it is rather unlikely that drug resistant virus is passed on to the child. Although the pMTCT regimen may induce a selective pressure on the maternal virus, there is hardly enough time for this virus to be selected to sufficient numbers to be transmitted during birth, see also . However, drug resistant virus may, with some probability, be transmitted during subsequent breastfeeding .
Newborns P019T, P020T and P028T were not breastfed and their mothers received a single dose of AZT +/− NVP shortly before birth. However, these newborns received 6 mg/day (P019T, P020T) or 28 mg/day (P028T) of post-natal AZT. As explained above, postnatal AZT administered to P028T may explain the archiving of AZT resistance in the child’s PBMC DNA (case ii). However, this did not lead to subsequent therapeutic failure (case b).
The mothers of newborns P002T, P003T, P027T and P046T were breastfeeding. They received extended AZT for periods shorter than the actual duration of breastfeeding. None of these children (P002T, P003T, P027T and P046T) showed evidence of archived AZT resistance based on DNA-OLA at the time of cross-sectional assessment. These children could have either been infected with susceptible virus during labour, or during breastfeeding (after cessation of extended maternal AZT), or else resistance may not have persisted until treatment initiation or until the DNA OLA was performed.
A potential weakness of a cross-sectional study design is that clinical and laboratory data from the beginning of ART up until the date of cross-sectional analysis had to be extracted from medical records and parent/patient interviews. Adherence data using the PACTG questionnaire are always self-reported. This study design does not allow for detailed cause-effect analyses, prospective surveillance and follow-up visits, or the assessment of mortality data. The cross-sectional analysis however does reflect the real-world effectiveness of a medical intervention in a low-resource setting, which often includes patients who would not typically be able to participate in controlled clinical trials. The focus of this study was the assessment of the usefulness of cross-sectional resistance testing using the DNA versus RNA OLA.
The DNA OLA may be particularly useful for the purposes of population-based surveillance in low resource settings where genotyping tests may not be readily available. The DNA-OLA was very indicative for the presence of resistance (high specificity, low false positive rate), but less indicative for the absence of resistance (low sensitivity, high false negative rate) in comparison to the RNA OLA. To the contrary, the RNA-OLA was more useful to determine the absence rather than the presence of drug resistance. Therefore, DNA-OLA can be used to rule-in resistance, whereas RNA-OLA may be used to rule-out resistance.
The detection of the resistance mutations M184V, N88D and L90M by DNA-OLA was highly sensitive for virologic failure in this cohort treated with lamivudine-azidothymidine-nelfinavir as first-line therapy. The analysis of archived HIV-DNA resistance in PBMC provided useful results in most patients, even if virologic failure was not (yet) evident. The DNA-OLA may detect resistance mutations that have been acquired during previous exposure to erratic short-term ART, still present in the lymphocyte compartment. This may occur in low-resource settings before antivirals become universally available, when patients and their families are restricted to temporary access to limited, often insufficient amounts of antiviral medications. Turnover rates within the lymphocyte compartment may however be too low for the early detection of antiretroviral drug resistance during therapy (i.e. in time before viral failure becomes apparent).
A possible strategy for the improvement of ART in resource-poor settings (where genotyping is often not available) could be to use the DNA-OLA as a baseline screening tool before starting therapy. This could be combined with the use of RNA-OLA in those patients experiencing virologic failure. Notably, a positive RNA OLA at positions M184V, N88D or L80M was highly sensitive for virologic failure (sensitivity: 0.93, 1.0 and 1.0, respectively). Therefore, drug resistance monitoring at key residues using RNA OLA in patients experiencing virologic failure may be particularly useful as an economical indicator of drug resistance and could suggest a treatment change.
Success rates could likely be improved even further if treatment was initiated at higher CD4 counts, in line with recent revisions of the treatment guidelines in industrialized countries (initiation of treatment at an adult CD4 count of 350 cell/μL) [36, 37]. This is in agreement with recent reports from other cohorts in Latin America. A recent cross-sectional analysis and evaluation of clinical outcomes of ART in Latin America showed that nearly half of the patients were so-called “late testers/presenters”. Evaluations of outcomes with ART in Latin American children revealed a higher incidence of opportunistic infections when compared to US cohorts (such as PACTG 129C) [36, 37].
While consensus RNA genotyping (if available) will likely remain the mainstay of individualized resistance testing during ongoing antiretroviral therapy, the applicability of the OLA in population-based surveillance remains to be fully assessed in larger cohorts, including cost-effectiveness analyses and assessments of the personnel and training required for either method. At the time of the study, genotyping was not available. In recent years, capacities for monitoring drug resistance have been expanded at the Peruvian INS including sequencing facilities and an e-health driven, web-based laboratory information system [40, 41]. The national ART program was expanded in 2004 to include larger parts of the population living with HIV/AIDS, including infants in earlier stages of HIV infection [41–43].
Our data emphasize the need for timely antiretroviral treatment initiation and early HIV testing to contribute to this aim [5, 12, 35, 44]. For children undergoing therapy, regular follow-up visits with viral load and resistance testing and concrete measures to monitor and improve adherence (using PDA’s, cellphone reminders and other e-health features) may be a key to success of ART in Latin America and beyond [45–52].