Geographical variation in Plasmodium vivax relapse

Abstract

Background: Plasmodium vivax has the widest geographic distribution of the human malaria parasites and nearly 2.5 billion people live at risk of infection. The control of P. vivax in individuals and populations is complicated by its ability to relapse weeks to months after initial infection. Strains of P. vivax from different geographical areas are thought to exhibit varied relapse timings. In tropical regions strains relapse quickly (three to six weeks), whereas those in temperate regions do so more slowly (six to twelve months), but no comprehensive assessment of evidence has been conducted. Here observed patterns of relapse periodicity are used to generate predictions of relapse incidence within geographic regions representative of varying parasite transmission.

Methods: A global review of reports of P. vivax relapse in patients not treated with a radical cure was conducted. Records of time to first P. vivax relapse were positioned by geographic origin relative to expert opinion regions of relapse behaviour and epidemiological zones. Mixed-effects meta-analysis was conducted to determine which geographic classification best described the data, such that a description of the pattern of relapse periodicity within each region could be described. Model outputs of incidence and mean time to relapse were mapped to illustrate the global variation in relapse.

Results: Differences in relapse periodicity were best described by a historical geographic classification system used to describe malaria transmission zones based on areas sharing zoological and ecological features. Maps of incidence and time to relapse showed high relapse frequency to be predominant in tropical regions and prolonged relapse in temperate areas.

Conclusions: The results indicate that relapse periodicity varies systematically by geographic region and are categorized by nine global regions characterized by similar malaria transmission dynamics. This indicates that relapse may be an adaptation evolved to exploit seasonal changes in vector survival and therefore optimize transmission. Geographic patterns in P. vivax relapse are important to clinicians treating individual infections, epidemiologists trying to infer P. vivax burden, and public health officials trying to control and eliminate the disease in human populations.

Figure 1

Pathways to infection of blood and clinical attacks in Plasmodium vivax malaria.

Figure 2

Proposed distributions of Plasmodium vivax relapse latency phenotypes. Panel A, adapted from White [47], shows the historical distribution of frequent relapsing and long-latency relapse strains. The geographic limits were modified using a historical malaria endemicity map from Lysenko and Shemashko [50]. Tropical frequent relapsing strains are in pink and long-latency strains in grey. Much of Africa is shown with grey hatching because the influence of Duffy negativity and its effect on vivax transmission in this part of the world is not yet understood. Purple areas are thought to have both long-latency and frequent relapsing strains. Panel B shows the Old and New World classification system based on the analysis and findings from Lover and Coker [51]. Tropical zones (red and pink) harbour strains that relapse more quickly than those in temperate zones (light and dark blue). New World tropical strains (pink) relapse more slowly than Old World tropical strains (red) and Old World temperate strains (dark blue) relapse slower than New World temperate strains (light blue). The dotted lines indicate the ±23.5° latitude lines to delineate temperate and tropical areas. Old World refers to Africa, Eurasia and the Pacific and New World to the Americas and Caribbean regions.

Figure 3

Zoogeographical zones of malaria transmission. Zones defined by Macdonald (1957) based on regions with similar ecological and epidemiological characteristics [52].

Figure 4

Schematic overview of the literature review procedure and results to obtain individual records of relapse and follow-up.

Figure 5

Revised zoogeographical zones and observed time to first relapse. Panel A illustrates the revised zoogeographical zones used to describe the time to first relapse. Panel B shows the median observed time to relapse in each study used to obtain individual data. The size of each point varies by sample size and the time to first relapse is shown on a spectrum of red (less than one month) to dark blue (>12 months). Violin plots in Panel C show the observed time to first relapse in individuals from each zone in Panels A and B. The coloured areas correspond to each zone and to a smoothed approximation of the frequency distribution (a kernel density plot) of the time relapse within each geographic region. The black central bar represents the interquartile range and the white circles indicate the median values. Note that the maximum value for zone 2 extends beyond the plot.

Figure 6

Survival curves for the modified Macdonald system. Shown are the Kaplan-Meier estimates (solid lines) with 95% confidence intervals (dotted lines). For each curve, all individual-level data from the respective zone have been pooled. The curves terminate at the longest follow-up day in each zone. The ticks indicate censoring events, i.e., losses to follow-up or the completion of a study without relapse. Zone 1: North America, zone 2: Central America, zone 3: South America, zone 5: Mediterranean and North Africa, zone 7: sub-Saharan Africa, zone 8: Monsoon Asia, zone 10: Southeast Asia, zone 11: northern Europe and Asia, zone 12: Melanesia.

Figure 7

Modelled relapse incidence and mean time to relapse. Panel A illustrates the relapse incidence per 100,000 person days on a spectrum of blue to red, with red being the highest incidence of relapse. Zone 8 is hatched to indicate that the predication is particularly uncertain. Panel B illustrates the predicted mean time to relapse on a spectrum from blue to red, with red being most frequent relapse. The numbers of the zones correspond to those shown in Figure 5A.