Malaria in HIV-infected children receiving HIV protease-inhibitor-compared with non-nucleoside reverse transcriptase inhibitor-based antiretroviral therapy, IMPAACT P1068s, substudy to P1060

Background: HIV and malaria geographically overlap. HIV protease inhibitors kill malaria parasites in vitro and in vivo, but further evaluation in clinical studies is needed. Methods: Thirty-one children from Malawi aged 4±62 months were followed every 3 months and at intercurrent illness visits for ≤47 months (September 2009-December 2011). We compared malaria parasite carriage by blood smear microscopy (BS) and confirmed clinical malaria incidence (CCM, or positive BS with malaria symptoms) in children initiated on HIV antiretroviral therapy (ART) with zidovudine, lamivudine, and either nevirapine (NVP), a non-nucleoside reverse transcriptase inhibitor, or lopinavir-ritonavir (LPV-rtv), a protease inhibitor. Results: We found an association between increased time to recurrent positive BS, but not CCM, when anti-malarial treatment and LPV-rtv based ART were used concurrently and when accounting for a LPV-rtv and antimalarial treatment interaction (adjusted HR 0.39; 95% CI (0.17,0.89); p = 0.03). Conclusions: LPV-rtv in combination with malaria treatment was associated with lower risk of recurrent positive BS, but not CCM, in HIV-infected children. Larger, randomized studies are needed to confirm these findings which may permit ART optimization for malaria-endemic settings. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. P1068s was substudy to the International Maternal Pediatric Adolescent AIDS Clinical Trials Network (IMPAACT) study P1060, details of which have been reported elsewhere [10, 11]. Briefly, P1060 was an HIV treatment study conducted at 6 countries in sub-Saharan Africa and India that enrolled HIV-infected children aged 2 months-36 months who qualified for treatment according to World Health Organization (WHO) criteria. Children were randomized to receive zidovudine and lamivudine combined with either NVP or LPV-rtv, and were followed for < = 47 months (median follow-up 32 months), and stratified by exposure to NVP at birth and by age (<1 year vs. > = 1 year) in the context of P1068s. ART failure was defined as permanent discontinuation of the treatment regimen for any reason, including the need for treatment of tuberculosis during the course of the study [10, 11]. The malaria substudy described herein, P1068s, was conducted at three sites with endemic malaria transmission according to published data at the time, which included Kampala, Uganda; Lusaka, Zambia; and Lilongwe, Malawi [12–14]. Analysis was performed only on data from the Malawi site, however, because of low blood smear positivity rates at the other sites: among the 74 subjects enrolled at the Zambia and Uganda sites, only 8 subjects had a positive blood smear over the course of follow-up for a total of 19 positive BS over the combined 4 year accrual and follow-up period, although one child in Uganda died from unconfirmed but suspected severe malaria [11]. All participants from the parent HIV treatment study, P1060 were eligible for inclusion in this malaria substudy. Although not randomized for the primary objectives of the P1068s substudy described herein, it was originally intended the randomization of P1060 would be used for P1068s; however, due to later initiation of this substudy, a number of subjects had switched from the regimen to which they had been originally randomized. Thus, most data collected under P1068s that links treatment to outcome is observational and most analysis was conducted using an “as-treated” approach. Study visits were conducted 2 and 4 weeks after the initiation of treatment, every 4 weeks until week 16, at week 24, and every 12 weeks thereafter, and patients were encouraged to come in for intercurrent illness. Follow up for P1068s paralleled P1060 (Table 1). History and physical exams were completed at all scheduled and intercurrent illness visits, and children were seen by study physicians at all visits. Heel or finger stick or venous blood were used to prepare thick blood smear using 2% Giemsa for 30 minutes, in addition to dried blood spots (DBS, with 50 µL/spot) collected on Whatman 903 paper (Florham Park, NJ) and handled as previously described [15] for confirmatory PCR (see Supporting Information, S1 File [16, 17]). CD4 cell count, CD4%, and HIV viral load were measured in real time during the P1060 trial [11]. Subjects were diagnosed with malaria on site and treated as per WHO recommendations, which for this study included artemether-lumefantrine for uncomplicated malaria and quinine or artesunate for severe malaria [18]. In Malawi, trimethoprim-sulfamethoxazole prophylaxis was administered as per WHO guidelines [19, 20], and all patients received prophylaxis. At the beginning of the study, the infant feeding policy was exclusive breastfeeding recommended for the first 6 months, which changed to the first 12 months of life as per WHO, in 2010 [21], but this was not adopted by Malawi until July 2011. All children on study were given an insecticide-treated bednet at the beginning of the study and lived within 30 km of the study site. Clinical illness was managed based on the Integrated Management of Childhood Illness Guidelines [22]. Malaria transmission in Malawi is perennial and holoendemic, with seasonal increases after the rains from November to April. Malaria mapping analysis in Malawi showed lack of significant changes in transmission between 2000–2010, with a population-adjusted P. falciparum rate (PAPfr2-10) in 2010 of 32%; this encompasses part of the period in which this study was conducted [23]. This substudy was approved by site-specific institutional review boards (IRBs), including the New York University School of Medicine (NYU) IRB (January 17, 2008) and the Malawian Ministry of Health and Population National Health Sciences Research Committee (June 8, 2009) and the NIH/NIAID IRB through a reliance agreement with NYU IRB (May 18, 2012). Each child’s parent or legal guardian provided written informed consent. The study was first opened to accrual on August 19, 2009 with the first patient enrolled September 25, 2009. A blood smear was deemed negative if no parasites were seen in > = 200 high-powered fields. All blood smears in the study, with the exception of one P. malariae sample from Lilongwe, were reported as P. falciparum. The diagnosis and management of malaria was based readings of blood smears on site, but reported results herein are from reads performed at NYU, with a primary and secondary reads by two microscopists, with discrepant results resolved and a random subset of smears reviewed by a third microscopist. Quality control of blood smear reading was maintained by 30 hours of training which covered both slide preparation, cover slipping, microscope care and maintenance and identification and quantification of malaria parasites, both at NYU and at study sites. Over the time period of the study the NYU microscopists were proficiency tested four times using archived smears provided by Hydas World Health (http://hydasworldhealth.org/), whose stock smears are read by WHO Level 1-Certified microscopists [24]. Each examination consisted of 15 cover-slipped, Giemsa stained blood smears containing thick and thin blood smears from malaria infected and uninfected blood. Blood smears included either single or mixed infections of Plasmodium falciparum, vivax, ovale or malariae. For the total 60 test slides, the readers’ overall sensitivity was 100%, specificity 92%, and density determination was within +/-25% of the accepted value in 96% of the slides. During the final 2 years of the study, the sensitivity and specificity were both tested at 100%. Separately, DNA was isolated from DBS and parasite genomic DNA was detected using real time from DBS collected for confirmation in parallel to smears. The primary end points were the time to first or recurrent events of and the rate of positive malaria blood smear (BS) or confirmed clinical malaria (CCM), defined as a positive BS with diagnosed malaria symptoms, as per WHO guidelines [18]. Only episodes of CCM greater than 14 days apart were considered, as those were assumed to represent new infections rather than recrudescence [9]. Thirty-one of 288 subjects from the P1060 parent study enrolled in P1068s, and 8 of these 31 children changed regimens based on the primary study before or after enrollment. For this reason we undertook an as-treated analysis, attempting to link observed treatment and observed outcomes. Per-protocol analysis, censoring subjects when they switched regimens, and as-randomized analysis (as randomized for the P1060 parent study) analysis are also provided in Supporting Information (S2 File and S1 and S2 Tables). During this study, another clinical trial suggested that a drug interaction between LPV-rtv and lumefantrine (the second component of artemether-lumefantrine, used for treatment of uncomplicated malaria) reduced clinical malaria burden in children on LPV-rtv ART in an area of high transmission intensity [8]. For this reason, the relationship between time to recurrent episodes of malaria (positive BS or CCM) and ART and potential synergy between HIV ART and malaria treatment is the primary scientific question evaluated in this report. This analysis was conducted by fitting a recurrent event (count process) Cox model [25]. Malaria treatment is considered a time-varying indicator in the models, until the occurrence of a malaria event, censoring by ART change or censoring by the end of follow up. Both malaria treatment and ART regimen were treated as time-varying covariates and models were adjusted for average CD4% over prior period, sex, age at enrollment, months since start of enrollment into P1060 to start of P1068s and an indicator for malnutrition status (as defined by mid-upper arm circumference [26]) collected at each study visit. In addition to evaluation of recurrent malaria events, comparisons of the rate of incident positive BS and CCM per time at risk by HIV treatment type was analyzed using negative binomial models including an offset for time-on-trial and adjusted for potential confounders. HIV treatment was accounted for using the following three time categories of LPV-rtv ART use: LPV-rtv ART assigned at P1060 baseline; LPV-rtv ART duration in months; or majority of time on LPV-rtv ART Reported p-values are not corrected for the number of analyses conducted. Statistical analysis was performed with R software, version 3.1.3.