BACKGROUND: Plasmodium blood-stage infections can be identified by assaying for protein products expressed by the parasites. While the binary result of an antigen test is sufficient for a clinical result, greater nuance can be gathered for malaria infection status based on quantitative and sensitive detection of Plasmodium antigens and machine learning analytical approaches. METHODS: Three independent malaria studies performed in Angola and Haiti enrolled persons at health facilities and collected a blood sample. Presence and parasite density of P. falciparum infection was determined by microscopy for a study in Angola in 2015 (n = 193), by qRT-PCR for a 2016 study in Angola (n = 208), and by qPCR for a 2012-2013 Haiti study (n = 425). All samples also had bead-based detection and quantification of three Plasmodium antigens: pAldolase, pLDH, and HRP2. Decision trees and principal component analysis (PCA) were conducted in attempt to categorize P. falciparum parasitemia density status based on continuous antigen concentrations. RESULTS: Conditional inference trees were trained using the known P. falciparum infection status and corresponding antigen concentrations, and PCR infection status was predicted with accuracies ranging from 73-96%, while level of parasite density was predicted with accuracies ranging from 59-72%. Multiple decision nodes were created for both pAldolase and HRP2 antigens. For all datasets, dichotomous infectious status was more accurately predicted when compared to categorization of different levels of parasite densities. PCA was able to account for a high level of variance (>80%), and distinct clustering was found in both dichotomous and categorical infection status. CONCLUSIONS: This pilot study offers a proof-of-principle of the utility of machine learning approaches to assess P. falciparum infection status based on continuous concentrations of multiple Plasmodium antigens.

Routine assessment of the efficacy of artemisinin-based combination therapies (ACTs) is critical for the early detection of antimalarial resistance. We evaluated the efficacy of ACTs recommended for treatment of uncomplicated malaria in five sites in Democratic Republic of the Congo (DRC): artemether-lumefantrine (AL), artesunate-amodiaquine (ASAQ), and dihydroartemisinin-piperaquine (DP). Children aged 6-59 months with confirmed Plasmodium falciparum malaria were treated with one of the three ACTs and monitored. The primary endpoints were uncorrected and polymerase chain reaction (PCR)-corrected 28-day (AL and ASAQ) or 42-day (DP) cumulative efficacy. Molecular markers of resistance were investigated. Across the sites, uncorrected efficacy estimates ranged from 63% to 88% for AL, 73% to 100% for ASAQ, and 56% to 91% for DP. PCR-corrected efficacy estimates ranged from 86% to 98% for AL, 91% to 100% for ASAQ, and 84% to 100% for DP. No pfk13 mutations previously found to be associated with ACT resistance were observed. Statistically significant associations were found between certain pfmdr1 and pfcrt genotypes and treatment outcome. There is evidence of efficacy below the 90% cutoff recommended by WHO to consider a change in first-line treatment recommendations of two ACTs in one site not far from a monitoring site in Angola that has shown similar reduced efficacy for AL. Confirmation of these findings in future therapeutic efficacy monitoring in DRC is warranted.

The spread of drug resistance to antimalarial treatments poses a serious public health risk globally. To combat this risk, molecular surveillance of drug resistance is imperative. We report the prevalence of mutations in the Plasmodium falciparum kelch 13 propeller domain associated with partial artemisinin resistance, which we determined by using Sanger sequencing samples from patients enrolled in therapeutic efficacy studies from 9 sub-Saharan countries during 2014-2018. Of the 2,865 samples successfully sequenced before treatment (day of enrollment) and on the day of treatment failure, 29 (1.0%) samples contained 11 unique nonsynonymous mutations and 83 (2.9%) samples contained 27 unique synonymous mutations. Two samples from Kenya contained the S522C mutation, which has been associated with delayed parasite clearance; however, no samples contained validated or candidate artemisinin-resistance mutations.

The ability to identify mixed-species infections and track the origin of Plasmodium parasites can further enhance the development of treatment and prevention recommendations as well as outbreak investigations. Here, we explore the utility of using the full Plasmodium mitochondrial genome to classify Plasmodium species, detect mixed infections, and infer the geographical origin of imported P. falciparum parasites to the United States (U.S.). Using the recently developed standardized, high-throughput Malaria Resistance Surveillance (MaRS) protocol, the full Plasmodium mitochondrial genomes of 265 malaria cases imported to the U.S. from 2014-2017 were sequenced and analyzed. P. falciparum infections were found in 94.7% (251/265) of samples. Five percent (14/265) of samples were identified as mixed- Plasmodium species or non-P. falciparum, including P. vivax, P. malariae, P. ovale curtisi, and P. ovale wallikeri. P. falciparum mitochondrial haplotypes analysis revealed greater than eighteen percent of samples to have at least two P. falciparum mitochondrial genome haplotypes, indicating either heteroplasmy or multi-clonal infections. Maximum-likelihood phylogenies of 912 P. falciparum mitochondrial genomes with known country origin were used to infer the geographical origin of thirteen samples from persons with unknown travel histories as: Africa (country unspecified) (n = 10), Ghana (n = 1), Southeast Asia (n = 1), and the Philippines (n = 1). We demonstrate the utility and current limitations of using the Plasmodium mitochondrial genome to classify samples with mixed-infections and infer the geographical origin of imported P. falciparum malaria cases to the U.S. with unknown travel history.

High throughput sequencing (HTS) generates large amounts of high quality sequence data for microbial genomics. The value of HTS for microbial forensics is the speed at which evidence can be collected and the power to characterize microbial-related evidence to solve biocrimes and bioterrorist events. As HTS technologies continue to improve, they provide increasingly powerful sets of tools to support the entire field of microbial forensics. Accurate, credible results allow analysis and interpretation, significantly influencing the course and/or focus of an investigation, and can impact the response of the government to an attack having individual, political, economic or military consequences. Interpretation of the results of microbial forensic analyses relies on understanding the performance and limitations of HTS methods, including analytical processes, assays and data interpretation. The utility of HTS must be defined carefully within established operating conditions and tolerances. Validation is essential in the development and implementation of microbial forensics methods used for formulating investigative leads attribution. HTS strategies vary, requiring guiding principles for HTS system validation. Three initial aspects of HTS, irrespective of chemistry, instrumentation or software are: 1) sample preparation, 2) sequencing, and 3) data analysis. Criteria that should be considered for HTS validation for microbial forensics are presented here. Validation should be defined in terms of specific application and the criteria described here comprise a foundation for investigators to establish, validate and implement HTS as a tool in microbial forensics, enhancing public safety and national security.