Background
Although the global burden of
Plasmodium vivax malaria dropped from 24.5 million cases in 2000 to 14.3 million cases in 2017 (a 42% reduction [
1]), the control of
P. vivax malaria remains neglected worldwide [
2]. The reasons for this neglect include its lower prevalence and mortality compared with that of
P. falciparum and the uneven distribution of
P. vivax malaria in the world. Nearly 80–90% of
P. vivax cases reported are from Asia, the Middle East, the Western Pacific, and the remainder occurring in Central and South America, and the poor understanding of the contribution of
P. vivax to morbidity and mortality. The distinctive feature of
P. vivax is relapse which can occur long after an initial attack [
3], which is caused by dormant liver-stage parasites (hypnozoites). This creates more resilience in
P. vivax transmission than that of
P. falciparum in situations adverse to the transmission of the parasites. Hence, in the last mile, elimination of malaria in one country or region is frequently limited by residual
P. vivax.
P. vivax was identified as a separate malaria species by Grassi and Feletti [
4]. As far back as 1897, it was established that the incubation period of
P. vivax, defined as the time between infection with the parasite by an
Anopheles mosquito bite and the onset of the first clinical symptoms, could be much longer than that of
P. falciparum. This incubation period was estimated to be 10 to 20 days [
5]. Malaria therapy, whereby fever was induced in patients with neurosyphilis by infecting them with malaria parasites between the 1920s and the 1950s, gave a clearer understanding of relapse in
P. vivax [
6]. The
vivax strains from Northern Europe and Russia often generated ineffective acute febrile illness, while strains from “Madagascar” successfully produced a short incubation infection with high fever [
7]. Soldiers from the Korean war and patients from World War II who had received a large number of inoculations often experienced recurrences of malaria [
8]. Korteweg recorded the mean monthly number of malaria cases in the village of Wormerveer, the Netherlands, from 1902 to 1923, and reported on long-latency
P. vivax [
9‐
11]. The demonstration that relapse is due to dormant hypnozoites in the liver dates back to 1948 by Shortt and Garnham [
12]. The effects of recurrence of
P. vivax via childhood and adult life are only rarely directly fatal. However, they can have major harmful consequences on individual growth and development, and on economic development at various levels, namely individual, family, community, and national [
2].
In the 1940s, many strains were isolated from different geographical areas, and a systematic study was conducted on the length of the
P. vivax incubation period and relapse patterns [
3]. The effect of inoculum on the probability and pattern of relapse has been studied extensively in humans, in classic studies in chimpanzees [
13], and in the Rhesus monkeys model with
P. cynomolgi [
14]. The World Health Organization (WHO) recommended classifying
P. vivax into three types according to the length of incubation period and pattern of relapse. Type I (tropical strains) describes species with short incubation periods (10–20 days) and a series of relapses at short intervals (less than 6 months) after primary infection. Type II (Elizabeth strain) with two phenotypes describes infections of a short incubation period followed by a long dormant phase until a series of short interval relapses or a long-incubation period followed by a few relapses at short intervals after primary infection. Type III (temperate strain) describes types of long-incubation period (more than 6 months) with only a few short intervals of relapsing after primary infection [
3].
In 1963, China reported the existence of long-incubation period
P. vivax (types II or III) [
15], and this was confirmed by human parasite inoculation experiments conducted between 1970s and 1980s [
16]. Tracing back to the Second World War, long incubation
P. vivax had already been observed among Japanese soldiers. In the 1980s, a large number of studies were conducted on the incubation period and relapse patterns, including both neonate cohort studies and human inoculation experiments [
17‐
22]. From the late 1980s to the end of 1990s, there were limited reports or studies on malaria in China as malaria incidence was less than 1/10,000 in most areas of the country but from 2000 onward, malaria rebounded in the Huanghuai Plain, with small outbreaks and local epidemics in Anhui, Henan, and Jiangsu provinces [
16,
23,
24]. New cases of malaria with no infection history in the previous year were often observed in spring, the non-transmission season. These cases could have played an essential role in the persistence of the infection.
Understanding the incubation periods and relapse patterns, and their geographical distributions, is critical to understanding how malaria control tools achieved their effects in the control stage before elimination, and to understanding the potential transmission risks in historical epidemic areas of
P. vivax after malaria elimination in China. This epidemiological understanding is also relevant for predicting the risk of renewed transmission seeded by introductions. Exploring the types of incubation periods and relapses of
P. vivax is of significance for the continuous improvement of malaria control strategies in the regions with similar settings. According to He Bin’s review, there were 71 counties (cities) in 16 provinces with both short- and long-incubation period
P. vivax in China in 1986. These provinces cover a vast area between 29° 30′ and 39° 52′ north latitude, spanning from South Temperate Zone to North Tropical Zone. The proportion of
P. vivax with short- versus long-incubation periods varied [
25].
The purpose of the study is to (1) understand the P. vivax incubation and relapse patterns based on previous published and unpublished malaria parasite inoculation experiments conducted using volunteer participants in different regions of People’s Republic of China between 1979 and 1988 and (2) to use a within-host relapse model based on the P. vivax temperate phenotype found to correspond to the inoculation data, to explain the success of the radical cure strategy for malaria elimination adopted by China since the 1980s, especially in combating major epidemics.
Discussion
In China, both short- and long-incubation period phenotypes of
P. vivax were observed in various studies, covering geographies ranging from the south temperate to north tropical zones [
25]. We found, in China, the short- and long-incubation periods of
P. vivax were 10–19 days and 228–443 days (8.0–14.7 months), respectively. The malaria transmission season in central China generally runs from early June to early November. However, before China’s malaria control and elimination program was launched, a large number of malaria cases were observed during the non-transmission season (from February to April), which could be well explained by the first relapse episode of long-incubation cases. Previously, Yang and colleagues performed a descriptive analysis using historical monthly incidence data from the Henan province (central China) and observed clear double peaks of malaria episodes during the 1960s and 1970s further supporting our findings for both short- and long-period incubations [
34].
Our incubation period estimates are similar to those found by Nishiura et al. [
35] from 416 people living in Korea between 2000 and 2003, with a mean long-incubation period of 48.2 weeks (95% CI: 46.8–49.5), equivalent to 337 days [
35] (demonstrated by the maximum likelihood method used to fit the frequency distribution of long- and short-incubation periods throughout the year). Swellengrebel et al. using data from the Netherlands also demonstrated two peaks of malaria cases representing delayed primary illness in late spring and early primary illness plus relapse in summer and autumn [
10,
36]. Similar findings were also observed by Zhou Wentao [
37], namely, that the distribution of cases with long- and short-incubation periods formed peaks in summer and autumn, as well as in the late spring of the following year, with an average 9.5 months (285 days) [
37] in Changde City, Hunan Province of China using 7 years of data. Using cluster analysis, we also found that the long-incubation period is around 294 days after sporozoite inoculation.
If the period between infection and the last relapse plus 2 weeks self-healing course [
18] is regarded as the natural life span of
P. vivax in humans, then the longest
P. vivax infection period based on various studies and observations in China is between 342 and 447 (geometric mean 366.6 + 14) days. This length is far less than previously believed at 3 years or longer in untreated
P. vivax [
19]. Consistent with the results here, in a South Korean study, a total of 79 eligible cases were selected to estimate the length of the long-incubation period, concluding that it was 337 days (95% CI: 328–347) [
35]. Therefore, three successive years without indigenous malaria cases reported suffices to allow certification of malaria elimination. This can be applied to China, which postponed certification due to the COVID-19 pandemic and announced malaria elimination on 30 June 2021.
A plausible explanation for the difference between short- and long-incubation is that these patterns are a result of two populations of sporozoites of
P. vivax within each inoculation. Short-incubation-sporozoites (SISs), called
Tachyspororozoites [
38], develop and proliferate rapidly, invade red blood cells, and then develop into exoerythrocytic schizonts. Long-incubation-sporozoites (LISs), called
Bradysporozoites [
38], do not immediately develop and proliferate but initially remain dormant in the liver and develop into hypnozoites. After varying periods of dormancy, hypnozoites develop and proliferate in batches and enter red blood cells causing multiple relapses [
39]. In other words, long incubation infections could be considered as asymptomatic sub-patent infections who eventually became patent. In temperate settings, LISs are present in great excess with a much smaller proportion of sporozoites characterized by SISs [
40]. Garnham [
3] already recognized Korean
P. vivax strains as a limiting case of almost exclusively LISs.
The assumption of two kinds of sporozoites can explain the conclusion that the duration of incubation, or the inter relapse interval(s) (for relapses with genetically homologous parasites), was determined by inoculum size [
41]. When the number of inoculated sporozoites is small, there is an increased likelihood that there will not be a malarial episode within the short incubation period. Primary illness occurs by nine or 10 months, or even longer, which our research supports. The more sporozoites which are inoculated, the more likely there will be short incubation periods and with a higher number of subsequent relapses. Experiments with quantitative inoculation of sporozoites by B. Yang in the Hunan province of China also confirmed this finding [
19,
42]. Different numbers of sporozoites were immediately injected into 21 volunteers by endothelial method after vitro extraction. The 16 volunteers inoculated with more than 10,000 sporozoites all showed short incubation periods of 15 ± .4.5 days. Five volunteers with 100 sporozoites all had long incubation periods, 312 ± 40.7 days. Ten volunteers with 1000 sporozoites inoculation presented both short (14.4 ± 0.8 days) and long (282 ± 29.9 days) periods. This study indicated that the ratio between SISs and LISs was around 1:1000, which is similar to the findings of Garnham and colleagues applying the North Korean strain in a chimpanzee model [
3].
Long-short-long-incubation pattern rotation has been observed in many other studies [
25,
43]. Volunteers infected by one sporozoite-positive mosquito, or a small number of sporozoites, developed a long-incubation period phenotype. A batch of newly emerged
An. sinensis was then infected using these volunteers’ blood. Another group of volunteers were then infected with multiple sporozoite-positive mosquitoes or a large quantity of sporozoites from this batch. These volunteers had short incubation period phenotypes. The long- incubation period phenotype recurred when the previous steps in the experiment were repeated, and new healthy volunteers were bitten by one sporozoite-positive mosquito or inoculated with a small amount of sporozoites. If multiple sporozoites lead to relapses in the same human host, then the relapse pattern is classified by the period corresponding to the first one to relapse. The separation into two classes (Fig.
3) is a question of whether at least one
Tachyspororozoite happens to survive. A rotation study with a long-short-long-incubation pattern, which used the same malaria parasite strain as we did for our study, reenforces the theory that the short- or long-incubation phenotypes are determined by the amount of inoculated sporozoites instead of by the region where the parasite strain originated [
17,
25]. Using climatic zone as an indicator to classify malaria types may not be a proper way to make classifications, which was pointed out by Coatney in 1971. Coexistence of short- and long-relapse phenotypes of
P. vivax have been demonstrated through genotyping strains from Northeast India and Nepal [
44,
45]. However, in view of the clear differences in duration of incubation and frequency of relapse, it is worth developing effective anti-malaria strategies according to phenotypes instead of by climatic zone.
Although the sporozoite inoculation experiments attempted to mimic natural infections, many parameters could not be considered, for instance, immunity. The acquisition of blood stage immunity may also affect the length of incubation or the relapse period [
46]. In the case of malaria outbreaks in the non-immune or low-immunity populations, patients with fever episodes between days 3 and 6 showed high infectivity to mosquitoes—an infection rate of between 72 and 100%. The number of oocysts per positive mosquito varied from 27 to 316. With the increase in blood stage immunity, the mosquito infection rate drops sharply, especially if the number of oocysts within mosquitoes was reduced to 1–7, and this status lasted until transmission ended [
46]. It is therefore inferred that the ratio of the two phenotypes of incubation periods in the epidemic area depends also on the level of immunity in the population during the epidemic. In the early- and mid-stages of a malaria outbreak, due to the high infectious rate and infection intensity, a large number of sporozoites could be inoculated into a human host via mosquito bites, which would lead to a large number of short latency cases. With the increase in immunity among the general population in the late stage of a malaria endemic, infectivity of mosquitoes significantly declined. Long-latency malaria cases would be the dominant type owing to the small number of sporozoites inoculated, which is consistent with our findings. In our within-host model, although the quantitative relationship between the number of relapses and the amount of sporozoites is considered, the immunity effect was not considered and should be a focus in future studies. Furthermore, we lack evidence for correlation between the inoculum size and the sporozoite load in the salivary glands. Further investigation of this is warranted using both experimental approaches and within-host modeling.
Both the outcomes from the volunteer study and model simulation results provided evidence of the coexistence of short- and long-latency
P. vivax phenotypes with subsequent relapse, which was driven by inoculum size. As early as the 1960s, malaria control experts proposed a strategy to implement two rounds of radical cure mass drug administration (MDA). Through observation and field experiences, the strategies were specifically two rounds of radical hypnozoite clearance MDA deployed during transmission (summer and autumn) and non-transmission (late spring) seasons. Implementation of this strategy had a pronounced effect on controlling outbreaks during epidemics in the 1960s and 1970s. This strategy likely cleared both asymptomatic carriers and long-incubation patients, who would otherwise serve as the main contributors in the next transmission season. The treatment regimens were changing over time, beginning with pyrimethamine (100 mg) combining 5- or 8-day primaquine (150 or 180 mg, respectively) in the 1960s and early 1970s to a 3-day-chloroquine (1200 mg) combining 8-day primaquine (180 mg) administration in the late 1970s and onwards. In some regions in China, treatment was strengthened with pyrimethamine salt (5 mg per day) administered between mid-June and mid-September. However, large-scale MDA was not implemented when the incidence of malaria was below 5%. He Bin and colleagues proposed a substitute strategy to a village when incidence rates were less than 3%, namely only positive episodes receive malaria chemotherapy and during non-transmission seasons a 4-day chloroquine and primaquine treatment could be given. Further strategies were proposed for regions where
An. sinensis was the only vector, and when prevalence was less than 1%, this involved only targeting hypnozoite treatment to symptomatic patients as a substitute for the “two round radical cure MDA strategy.” This modified strategy was applied in two pilot areas in Huanghuai region in China and achieved satisfactory social and economic benefits [
24]. In China’s malaria elimination stage since 2010, the “1-3-7” surveillance response system together with a targeted high precision radical cure led to the achievement of malaria elimination [
47].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.