Plasmodium falciparum genomic surveillance reveals spatial and temporal developments, affiliation of genetic and bodily distance, and family clustering

0

As malaria control progresses towards elimination, genomic data has proven to be essential in assessing control and surveillance24. In this study we combine parasite genetic diversity indices, individual global position system (GPS) information at the neighborhood and household level, and a hierarchical Bayesian regression model to understand parasite diversity and connectivity over time and space in Thiès, Senegal. The main findings of this study were household clustering of genetic types, association with genetic distance and physical distance, as well as parasite sharing between participants from either the same household or different households which were geographically proximal. This study provides additional data from a well characterized low transmission setting at very focal spatial scales and with precise mapping of malaria in households and neighborhoods, specifically in daaras. This fine spatial and temporal scale is not always possible with large cross-sectional datasets; and interestingly, in past analyses from cross-sectional studies in Thiès, this spatial clustering of identical genotypes was not observed25.

Overall, the low level of Plasmodium parasite genetic diversity and the high frequency of monogenomic infection observed over years are generalizable and consistent with previous observations from cross-sectional sampling over time in Thiès14,25 and Dielmo and Ndiop in Senegal16. Average expected heterozygosity ((H_{e})), a common measure of parasite genetic diversity, represents the probability of being infected by two parasites with different alleles at a given locus. The value of (H_{e}) in this population was found to be 0.371 (95% CI (0.341, 0.401)). Coupled with the genotypic richness (0.425), relatively much lower than has been described in Malawi26 and in regions of declining transmission on the Thai-Burma border22, these measures emphasize the low genetic diversity in this population. In these localities 24-SNP molecular barcoding revels a predominance of monogenic infection and a significant percentage of shared genomic haplotypes in the population. These observations have been hypothesized to be the result of a significant reduction in malaria transmission due to the efficiency of malaria interventions post-2008.

Consistent with previous studies in Thiès, we also observed the existence of haplotypes persisting over several years14,25. Incorporating focal GIS data permitted us to monitor the genotype frequencies in different households nested within neighborhoods within the same year, and across seasons. For example, haplotype 796 was observed in the same neighborhood (Diakhao) in 3 years (2014, 2015, and 2017) in three different households (OB, OD, and MS). Similarly, haplotype 759 was observed in two different neighborhoods, in 3 successive years (2014-2017) and in three different households (OB, OD, and MS). When observing identical genotypes in households, there are two possibilities: 1) continued local transmission of a single parasite genotype that maintains genomic identity through selfing, or 2) the same infected mosquito biting multiple individuals within the household or neighborhood. The spatial and temporal nature of the infections can help distinguish which hypothesis is more likely; but additionally, this is an area where IBD has added value. Clonal propagation of identical parasites in households over short temporal scales (days) would favor infection of multiple individuals by the same infected mosquito, yielding identical parasites by both IBS and IBD. Continued focal transmission of identical parasites in households observed over longer temporal scales (weeks to months to years) would favor local household transmission of inbred parasite lineages, and here we may expect more heterogeneity in IBD even in parasites that are IBS due to limited outcrossing in the mosquito. We observe both scenarios in our study. An increase in genetic diversity and limited clonal propagation would imply imported parasites followed by outcrossing or co-transmission, with outcrossing resulting in genetically diverse monogenomic infections and co-transmission resulting in an increase in polygenomic infections. This study has also demonstrated that household transmission of the same genotype is frequent in Thiès, and that cross-neighborhood and cross-year transmission of the same genotype is also common, again implying a relative lack of outcrossing in the overall population27.

As the overall degree of genetic variation can vary depending on the season of transmission, the intensity of transmission, and also the degree to which new strains are introduced into the population28, genomic surveillance by longitudinal sampling can provide valuable insights into changing malaria ecology and transmission dynamics. Interestingly, but perhaps not unexpectedly, the genetic similarity of parasites identical by barcode (IBS) breaks down a bit when assessing genetic similarity at the whole genome level through IBD. IBD analysis shows that for haplotypes persisting for multiple years, identical parasites which are clonal by IBS most often are identical at the whole genome level, but can share from 70 to 100% of their whole genome, with those more distantly related in time sharing less overall genomic identity. We observed this notably in two instances, one of a haplotype spanning a two year gap and identified in a more distant household (Haplotype 759; 2015—Household MS and 2017—Household GB), and the second in a haplotype spanning multiple years (2014–2015–2016), and multiple households, but yet IBD was only decreased in the isolate from the highly diverse household MS. This finding implies that some of these barcode clonal parasites experienced some degree of out-crossing, and this was more likely over time and in households with a large and diverse pool of parasites serving as a potential recombination reservoir.

We observed household clustering and genetic differences between parasites to increase with distance between individuals. During this study, a particular household (MS) served as an example of a malaria hotspot of transmission at the household level, both in number of cases as well as genetic diversity of the parasites. Having such different parasites in the same household could be the result of importation of diverse genotypes due to human or mosquito mobility23,29,30,31,32, followed by genetic recombination (outcrossing) within the Anopheles mosquito33, and the subsequent transmission of new genetic combinations17 resulting in a hotspot of local intense transmission17. The predominance of polyclonal infections in this household would also favor this hypothesis. A similar study of malaria incidence and prevalence has demonstrated the existence of malaria transmission hotspots at the village level in Senegal7,8. In such villages, human density, human behavior, infrequent malaria bed-net use, substandard housing construction, and a favorable ecological environment for mosquito proliferation (presence of mosquito breeding sites) have all been identified as risk factors for a household to be in a hotspot34. The added value of our approach is being able to identify hotspots of transmission, but also to determine the genotypic nature of these hotspots – adding further to implications for control measures. If hotspots are populated by similar genotypes, it is more likely that local transmission of selfing strains is occurring. If multiple diverse monogenomic genotypes are present, the hotspot could serve as a hub of human or mosquito imported infections. If polyclonal infections increase, it implies a combination of both: importation following by increased local transmission. Identification of the transmission clusters at the household level will play an important role for interrupting malaria transmission chains5,35. Identifying neighborhoods or households with high malaria transmission can assist malaria control programs with focal interventions to reinforce malaria prevention and control.

Because P. falciparum is a sexually recombining organism, precise mapping of phylogeny and transmission chains is not possible; however, the 24-SNP barcode has been shown to be a proxy for whole-genome that allows resolution especially of highly similar parasite types14. While the 24-SNP barcode does not provide as complete information about genetic relatedness (identity by descent) as whole-genome sequencing or large SNP arrays36, it has been estimated that the 24-SNP barcode can confidently detect parasites that share greater than 70% genome similarity (identity by state)14. While the pairwise genetic distance in the 24-SNP barcode is not linearly associated with whole-genome genetic distance, our finding of significant associations with physical distance is even more noteworthy. Our statistical model demonstrated that genetic variation between parasite pairs increases with physical distance. Here we used the number of SNP differences between paired individuals as genetic distance, or identity by state. Studies in The Gambia and Kenya have demonstrated that variation between parasite genotypes increases with geographical distance37,38. Such findings will help in understanding how the parasite population is structured in Thiès and the connectivity between parasites, despite some studies in Thiès having suggested a mixed parasite population with no hidden population structure27. In this study, sampling biases (number of limited samples) may not reflect the overall parasite population that is captured by passive case detection, and notably, we found no asymptomatic infections in any of the follow-up time points in the cohort.

All of the enrolled participants in our study live in “daara”s, religious boarding schools where “talibe” (resident student followers) live together in large numbers. One particular daara, arbitrarily termed “MS” had a very large proportion of cases, a diverse haplotypes and all re-infected participant were from that household. As a specific population, little is known regarding the malaria burden in this specific community of children, although it has been proposed that this population is considered more vulnerable and may have higher risks of parasitic infectious diseases, due to living conditions39 The high malaria burden in this community may be explained by the household size, close living and sleeping quarters, socio-demographic factors, and less consistent compliance with long-lasting insecticidal net (LLIN) use40. The Senegal national malaria program control (NMCP) has recently established a malaria case management at the level of daraas (PECADARA) to screen and treat students living in these boarding residences with the ultimate goal of preventing morbidity, mortality, and transmission in this demographic. Our findings of malaria burden, evidence of multiple infections of the same parasite in the same household, as well as some households with highly diverse infections, implying a “melting pot” for imported types and recombination23, all support the notion that extended malaria surveillance specifically in the daaras could be an important strategy to prevent continued malaria transmission chains in the community. Daaras also represent an attractive opportunity for intervention for NMCPs as there is the opportunity to systematically reach many children living in the same household.

Our study has some limitations. Our cohort was completely male, although enrollment was open and encouraged for both male and females. The ages of participants were children and adolescents, sampling was limited to the high transmission season, and malaria infections were all symptomatic and detected by passive case detection. As the participants were enrolled in a longitudinal cohort and followed over time, we may have observed selection bias for more solitary individuals as those who intended to travel may have opted not to participate. As previously described, all of the samples from participants in this study came from residents of daaras, thus our results would be generalizable to other male residents of daaras; however not to the general population. Yet, this study is the first to provide a detailed genetic characterization of the parasite populations in daaras in Senegal and will provide valuable information to the Senegalese NMCP which is implementing specific interventions in daaras this year. Going forward, studies prospectively designed to specifically investigate malaria transmission dynamics and population genetics in daaras should intentionally include and enroll daaras with male children and an equal number of daaras with female children. Future studies could also apply the same methodologies, but in a population-based cross-sectional sampling approach, in both the high and low transmission periods, and outside the clinical setting, to capture the genetic complexity of both symptomatic and asymptomatic infections throughout the general population.

Additionally, our sample size of 70 participants with 74 infections over 3 years is relatively small. From previous cross-sectional studies in Thiès spanning detectable signals of declining and rebounding transmission intensity14, the mean number of samples (monogenomic and polygenomic) across years was approximately 170 and the mean number of monogenomic samples was approximately 125. Based on this data, we estimate that 100 monogenomic samples would be sufficient to detect subtle changes in transmission intensity over time. This sample size is ideal, but as malaria transmission declines in pre-elimination zones, it might prove difficult to achieve, highlighting the need for complimentary measures of transmission intensity such as serological markers of recent compared to historical past malaria exposure.

While the 24-SNP molecular barcode does have limitations in its ability to infer transmission levels and population connectivity, especially on highly local scales; as evidenced by this study and others14,23,41, the 24-SNP barcode can be a useful, and importantly field-deployable tool for rapid assessment of Plasmodium genomics. It can be useful in distinguishing polygenomic infections from monogenomic infections for measures of complexity of infection (COI)16, which increases with transmission intensity, even if it is unable to distinguish the identities of parasite genotypes within these complex infections. However, these simple genetic metrics still have value in the context of real-time genomic surveillance efforts and can provide useful and actionable data on transmission hotspots, probable importation or local transmission, as well as assessment of the impact of specific interventions aimed at decreasing malaria transmission. While whole genome sequencing and identity by descent provide a wealth of high-resolution genomic information to clarify population genetic connection and potentially transmission chains; at the moment, measures such approaches have not been actionable in real time. Taken together, our study provides important information in the micro-epidemiology of parasite population structure in space and time in daaras in Thiès. The study also provides evidence of the feasibility and power of including genomic analyses, with field-deployable methods performed on site, in making public health decisions.

In conclusion, Plasmodium spatial-temporal clustering at the household and neighborhood level were observed along with increasing genetic distance between parasites as a function of physical distance. The longitudinal study shows the importance of applying molecular surveillance along with spatial and temporal modeling to detect hotspots of malaria transmission at fine spatial scales.The value of genomic data is especially powerful when traditional epidemiologic measures of transmission are not available or are limited. Taken together, this work emphasizes the added value of combining traditional epidemiology data, including case investigation, household surveys, climate data, and travel history with genomic data and high-resolution temporal and spatial (GPS) data. Combined, they provide powerful insights into local transmission dynamics. These local patterns can have practical implications in providing data to NMCPs on ways to better target local interventions in a way to maximize impact. Identifying the degree to which sustained local transmission or continuous importation of cases from outside a community can influence the specific policy approach adopted, from a focus on specific household or neighborhood malaria prevention efforts to a focus on human mobility as the dominant driver of transmission. Such insights are facilitated by the rapid, real-time acquisition, analysis, and reporting of genetic data to malaria policy makers and represent an attractive model for integrating malaria genomics into decision-making strategy.

Leave A Reply

Your email address will not be published.