In Central China, the epidemiology of P. vivax malaria has changed significantly over the past couple of decades, with overall declining incidence interrupted by epidemic expansions and changing dynamics of imported cases. These epidemiological changes may impact the transmission dynamics of the local P. vivax population. Indeed, we recently demonstrated unstable, epidemic P. vivax transmission in the pre-elimination setting of Sabah, Malaysia, which we hypothesized to be largely affected by the changing epidemiological dynamics in the rapidly shrinking parasite population . Previous studies have used genotyping of variable surface antigens to describe the diversity of P. vivax populations in Central China in the mid to late 2000s for the purposes of assessing a potential vaccine candidate [41–43]. However, assessments of P. vivax population structure and transmission dynamics, requiring neutral genetic markers, have not yet been described in this region. The current analysis focused on the population structure and transmission dynamics of P. vivax isolates in Central China using neutral genetic markers on parasites collected over the past few years following the most recent resurgence, presenting evidence of moderately unstable, largely clonal transmission in a parasite population with high levels of diversity. This data set should provide a useful baseline against which to compare patterns of P. vivax diversity and structure in Central China in later years in order to assess the impact of ongoing transmission intervention efforts in the region, as recently demonstrated in Sri Lanka .
Parasite isolates were sourced from two neighbouring provinces in Central China (Anhui and Jiangsu). Anhui exhibits the highest incidence of malaria in Central China (API 5.0 - 0.28 between 2007 and 2010), significantly greater than Jiangsu (API 0.05 - 0.09 between 2007 and 2010). The latter has not experienced any indigenous malaria cases in 2012 and 2013. Only moderate genetic differentiation was observed between the two provinces and this was apparent even after adjusting for the extensive marker diversity (F’
= 0.1). However, the number of isolates from Jiangsu was small, reducing the power of the analysis. Nonetheless, in accordance with the F’
results, limited evidence for differentiation between Anhui and Jiangsu was observed with STRUCTURE analysis, with any sub-populations rather being shared amongst the two provinces. The level of differentiation observed with the F’
analysis was lower than levels observed in intra-country comparisons in Sabah (F’
= 0.5 - 0.6)  and Colombia (F’
= 0.4 - 0.7) , where P. vivax endemicity is comparably low. Rather, the differentiation is more comparable to levels observed between sites in Papua New Guinea (PNG) (F’
= 0.14 - 0.16) . The mechanisms responsible for these observations in Central China are likely to be different from those in PNG, where transmission intensity is considerably higher, relapse dynamics differ, and human movement between populations may be limited. In contrast to the marked geographical boundaries in PNG, the majority of the Chinese isolates were sourced from neighbouring prefectures with limited geographical boundaries. This is not unexpected as hundreds and thousands of people travel between the two provinces every day. The recent epidemic expansion in Central China may have added to the increased panmixis between the provinces affected. In addition, the long dormancy duration (often exceeding eight months) of the temperate P. vivax strains endemic to Central China may have contributed to the persistence and spread of infections both temporally and geographically. Indeed, the majority of the multi-locus haplotypes observed three or more times in the data set (10/11), were observed in two or three different years of collection. Hence the current evidence suggests that the recently malaria-free province of Jiangsu is at moderately high risk of re-introduced malaria from Anhui.
The current analysis of seven STR markers demonstrated polyclonal infections to be infrequent in both Anhui (4%) and Jiangsu (12%). These levels of polyclonal infection are generally lower than those observed in tropical and sub-tropical endemic populations across the globe [24, 26, 27, 45–49], possibly reflecting the different relapse dynamics and relatively short window for local transmission in the more temperate setting of Central China. However, caution is required in comparison between different study sites owing to differences in markers and/or potential differences in allele calling. Previous analysis of P. vivax surface antigens on isolates from Anhui Province demonstrated a low frequency of polyclonal infections in 2004 (6%) , and moderate frequency (15%) between 2006 and 2008 . Across both Anhui and Jiangsu, a trend of increasing frequency of polyclonal infections was observed between 2008 and 2010 (2 - 19%). The results across the studies agree on a generally low frequency of polyclonal infections in Central China, compatible with largely clonal transmission dynamics over the past 10 years. However, the factor(s) responsible for the subtle variation between Anhui and Jiangsu, and temporally, remain unclear. It might be speculated that infections introduced from regions with more “tropical” relapse dynamics are responsible for some of the variation. Indeed, the high relatedness (maximum of one multi-allelic locus) between the clones in the polyclonal infections detected may support a relapse versus super-infection dynamic. Clonal transmission dynamics during expansions may have also contributed to the observed trends. However, firm conclusions can not be made owing to the limited sample size. Further investigations with larger sample size are required to confirm the trend and to elucidate the relative impact of relapse and imported/introduced cases on polyclonal infection dynamics in Central China.
The results of the analyses of population diversity and linkage disequilibrium (LD) demonstrate evidence of moderate instability in P. vivax transmission in Central China. As observed in P. vivax populations from multiple endemic settings across the globe [24, 26, 27, 45–49], high levels of diversity were observed in both Anhui and Jiangsu (H
E = 0.8). The reservoirs sustaining this diversity in the face of aggressive containment efforts remain unclear. Sub-patent and asymptomatic infections may present a largely hidden reservoir, with undetected imported infections in particular enabling maintained diversity. Indeed, the propensity of P. vivax to relapse from its dormant liver stage weeks to months after the primary infection may greatly aid the introduction of imported strains. A recent study investigating temporal trends in the diversity of P. vivax isolates in Sri Lanka demonstrated that despite rapidly declining endemicity between 2003-2004 and 2006-2007, population diversity remained unexpectedly high in the latter years . The authors postulated that imported infections were likely to be a major source of the maintained population diversity in this setting. In tropical and subtropical regions with year round transmission, outcrossing may enhance the population diversity. Indeed, in the low endemic setting of Temotu Province, Solomon Islands, where a recent study demonstrated comparatively higher diversity in P. vivax (H
E = 0.85) relative to P. falciparum (H
E = 0.54), frequent relapse may have facilitated polyclonal infection and outcrossing in the P. vivax population, with moderate contribution to the observed diversity . However, as demonstrated in South Korea , the narrow window for transmission in temperate regions reduces the opportunity for recombination and, thus, is not likely to be a major source maintaining the diversity in Central China.
Despite the high population diversity, LD remained high in both Anhui and Jiangsu and in each year tested (I
S = 0.2 – 0.4), suggestive of low recombination, as might be anticipated with largely clonal transmission and limited opportunity for transmission. However, when each distinct haplotype was represented just once, the strength of LD (I
S) declined two- to four-fold, with the greatest decline observed in 2008 (shortly after the peak of the P. vivax resurgence in Central China). The strengthening of LD by the expansion of a few haplotypes in an otherwise moderately panmictic population is suggestive of occasional outbreaks/epidemic transmission. Indeed, multiple small outbreak clusters of identical haplotypes were observed in the neighbour-joining plot. In Sabah, similar outbreak dynamics were observed, but with the majority of the identical haplotypes observed within short temporal windows and moderately confined geographic space . In contrast, identical Chinese haplotypes were often observed across multiple years (Figure 4) and prefectures (Figure 3). Indeed, only a weak correlation was observed between the genetic and temporal distance between the Chinese isolates (Mantel r-test, r = 0.05). Variation in relapse dynamics between the tropical Sabahan isolates, and the temperate Chinese isolates may account in part for these differences.
In settings such as Central China, where the endemicity of local infections is declining rapidly but the relative proportion of imported cases rising, geographic markers enabling confirmation of imported versus local P. vivax cases are urgently needed. Using a selection of samples sourced from P. vivax cases acquired from Southern China and internationally, a provisional assessment was undertaken on the utility of the current marker panel to distinguish central Chinese from imported infections. As demonstrated with neighbour-joining analysis and principal component analysis, although none of the isolates from outside Central China shared identical haplotypes to any of the central isolates, evidence for the utility of the marker panel for detecting imported cases was limited. A recently developed data analysis platform (VivaxGEN) has been made available for enhanced comparison of STR-based P. vivax data between studies . This multi-centre approach enables data sharing between Central China and other P. vivax endemic regions across the globe, so that the utility of the APMEN marker panel as well as new markers can be investigated comprehensively. Indeed, as recently demonstrated in a global study of variation in the P. vivax mitochondrial genome, SNPs in this organelle may offer some assistance in resolving the geographic origin of isolates at least at a regional level . Furthermore, variants in the apicoplast genome may enable further geographic resolution as recently demonstrated in a global selection of P. falciparum isolates .