Background
Deforestation and standard malaria control efforts including early, appropriate case management and distribution of insecticide-treated bed nets have reduced malaria prevalence to historically low levels in much of Western Cambodia. Unfortunately these control measures have failed to contain the emergence of anti-malarial drug resistant strains of
Plasmodium falciparum in an expanding geographical area [
1,
2]. In areas with very low malaria transmission the large majority of malaria infections were thought to be symptomatic and hence accessible to passive detection and curative treatment [
3]. Yet malaria has historically been difficult to eliminate even when most symptomatic patients received highly effective anti-malarial treatments. Symptomatic infections as a sole source of transmission cannot explain the virtual disappearance of malaria cases each year during the cool dry season and prompt return with the onset of rains. A significant sub-patent reservoir of
P. falciparum carriers does explain both the epidemiology of malaria in these areas of seasonal malaria and why the current control and containment activities fail to contain resistant malaria [
4,
5]. To understand and eliminate such a reservoir it is not only important to understand the prevalence at any one point in time but also the duration of individual infections.
Apparently healthy people who migrate to regions without malaria transmission and later give blood donations can remain infected asymptomatically with
P. falciparum for up to 13 years [
6]. Information on the distribution of persistent infections usually comes from cohort studies. Observational studies of untreated malaria patients were not uncommon during the last century. Lowe followed 16 people with untreated malaria in India 1934 [
7]. Hill and co-workers followed 22 children with falciparum, malariae and/or vivax infections at weekly intervals during 1937 and 1938 in Aguas de Moura, Portugal who were only treated if symptomatic [
8]. Earle et al. [
9] followed 76 mostly untreated children in weekly intervals in Puerto Rico in 1939. McGregor and co-workers [
10] studied falciparum infected, untreated children in The Gambia during the 1950s. Bruce-Chwatt [
11] reported a cohort study of a group of West African adults in 1963. Bruce et al. [
12] reported a study conducted in 1992 in which 70 people from a single village in Papua New Guinea (PNG) were sampled for up to 61 days. The advent of PCR based molecular diagnostics on low volume capillary blood spots allowed the documentation of persistent low-density
P. falciparum infections in Sudan over more extended periods [
8]. In 1997, a cohort of 43 recently malaria-infected Sudanese, aged from 9 to 53, agreed to donate fortnightly blood samples for the next 9 months. Of the 43 individuals, 16 (37 %) were found to maintain chronic
P. falciparum infections for the follow-up period of 9 months. Together these studies demonstrated that individual patients can carry
P. falciparum infections for weeks to years.
These historical field studies in endemic countries are not only limited by the detection threshold of light microscopy (the exception being the study in Sudan) but also by uncertainty regarding re-infections. The risk of reinfection is much reduced in very low or no transmission areas. A substantial contribution to the current understanding of the natural history of
Plasmodium infections comes from the malaria therapy of neurosyphilis. In well-documented treatments patients with neurosyphilis were infected with
P. falciparum or
P. vivax in South Carolina and Georgia, USA during the period 1940–1963 [
13,
14].
Plasmodium falciparum infections persisted (by microscopy) for a mean of 222 days the longest being 480 days [
15]. Malaria naïve patients who received sporozoite-induced vivax infections and no anti-malarial therapy were found to have waves of parasitaemia during the follow-up period of 108 days [
16]. In contrast to
P. falciparum infections the dynamics of vivax infections are complicated by the liver reservoir of
P. vivax hypnozoites, which cause periodic relapses and so contribute to the chronicity of parasitaemia [
14].
The development of high volume ultrasensitive qPCR (uPCR) allows as few as 22 parasites/mL to be measured accurately compared to ~1000 parasites/mL using conventional low volume PCR methods [
17]. Using uPCR for the detection of parasitaemia, the entire population of three Cambodian villages in Pailin province were followed over a year in order to describe the reservoir of sub-patent
Plasmodium infection.
Methods
Study site and population
Pailin is an agricultural province adjacent to the Thailand border in western Cambodia. The villages are farming communities, which grow cash crops and fruit. Nearby forests are used as a source of plants and fruit, small game, bamboo and wood. Containment efforts in Cambodia have resulted in a marked decline in the incidence of clinical malaria over the last decade. Between 2004 and 2013 a 145-fold reduction in
P. falciparum cases and a 4.8-fold reduction in
P. vivax cases was observed [
18]. Malaria control in Pailin has been based on case management by village health workers (VHW) or village malaria workers (VMW) and the distribution of long-lasting insecticide-treated bed nets (LLIN). There has been substantial replacement of forest by agriculture and rubber plantations, which could have contributed to the reduction in malaria transmission. Historically
P. falciparum has been the dominant
Plasmodium species causing malaria but more recently
P. vivax infection has become predominant [
18]. Malaria transmission is low and seasonal with entomological inoculation rates below one infectious bite/person/year [
19,
20].
Western Cambodia has been the epicentre for the emergence of
P. falciparum strains resistant against a range of anti-malarial drugs including chloroquine, sulfadoxine/pyrimethamine, artemisinins and piperaquine [
21‐
26]. The first-line treatment for falciparum malaria through December 2013, the first 6 months of the study, was atovaquone/proguanil (Malarone
©) which temporarily replaced artesunate–mefloquine, and was itself replaced in January 2014 with dihydroartemisinin/piperaquine (DHA/piperaquine) [
27]. Vivax malaria is treated with a schizontocidal drug which is usually an artemisinin combination therapy (ACT). Radical cure for vivax malaria with primaquine after G6PD testing is recommended but as G6PD tests are unavailable this is seldom used. In the Pailin area the primary health care providers for febrile illness are the VMWs who are supervised by the local government health centre. The VMWs stock rapid diagnostic tests (RDTs) and ACT. Primary healthcare from VMWs is intended to be available 24 h a day and is free. Patients with a diagnosis other than malaria are referred to or go directly to a local health centre, which is approximately 6 km from the study villages and serves other villages in the surrounding area. Those who require hospitalization travel to the Pailin Referral Hospital, which is roughly 30 km from the study site. Malaria treatment by the private sector is prohibited in Pailin but pharmacies and drug sellers do stock anti-malarial drugs. In practice patients who believe that a malaria diagnosis is unlikely bypass the VMW.
In 2013, the Cambodian National Centre for Parasitology, Entomology and Malaria Control (CNM) and Mahidol-Oxford Tropical Medicine Research Unit (MORU) formed a research team based in Pailin Referral Hospital to investigate the prevalence of subclinical parasitaemia. Three study villages were selected for participation in the study on the basis of high relative incidence of clinical falciparum malaria in the village malaria worker records.
Study procedures
In each village, a study committee was formed consisting of village leaders, VMWs, and volunteers. The committee assisted the study team in organizing the study and in engaging and mobilising the community. All households were geo-referenced by GPS and given a unique household number. In a census in April 2013 all residents were recorded, assigned a unique identification number and linked to a household number. In June 2013, the first of five cross-sectional surveys (M0) was conducted, followed by further surveys in October 2013 (M3), January 2014 (M6), April 2014 (M9) and June 2014 (M11). Before each survey, the population residing in the village at the time, including temporary residents and migrants were invited to participate. People moving into the villages between surveys for more than 2 weeks were included in the study.
Individual informed consent was obtained from adults and from parents or guardians of children aged less than 16 years. Information on demographics and household relationships were collected with a brief history of recent illness and travel. The tympanic temperature, weight, and height of all participants were measured. A venous blood sample (3 mL) from all individuals aged ≥5 years was collected in EDTA tubes, or 500 µL from children aged ≥6 months to 5 years. Participants with fever ≥37.5 °C were tested for malaria by rapid diagnostic test (Healgen malaria
P. falciparum/Pan one-step RDT, Zhejiang Orient Biotech, China), and if positive were treated according to national guidelines [
27].
Blood samples were stored on wet ice in the field and then transported within 9 h to a local laboratory in Pailin. Blood was centrifuged at 1500
g for 10 min (Heraeus labfuge 400) to separate plasma and buffy coat from packed red blood cells (pRBC). 500 µL pRBC samples for uPCR analysis were then frozen and stored at −80 °C. Batches of frozen samples were transported monthly on dry ice to the Molecular Tropical Medicine laboratory in Bangkok, Thailand for uPCR analysis as described previously [
17].
Participants aged 16 years and older who were found to be parasitaemic by uPCR were informed of their status and invited to join a study cohort. Cohort members were tested monthly for parasitaemia. The first survey (M1) was in August 2013. Subsequent cohort surveys were conducted in September 2013 (M2), November 2013 (M4), December 2013 (M5), February 2014 (M7), March 2014 (M8) and May 2014 (M10). After the study closed, the local health centre provided free treatment according to national guidelines to all participants with persistent parasitaemia.
Laboratory methods
A detailed description, evaluation and validation of the high-volume uPCR methodology has been reported previously [
17]. Briefly, the DNA template for detection and quantification of
Plasmodium by PCR is purified from thawed pRBCs. The presence of malaria parasites and an estimate of the parasite numbers (genomes) in each sample are assessed by an absolute quantitative real-time PCR method using primers targeting the gene for 18S rRNA. A Quanti-Tect Multiplex PCR No ROX
® Kit (QIAGEN, Germany) was used for this purpose with the PCR reaction mixture and the cycling conditions as per manufacturer’s instructions. The probes used in the assay have been validated and are highly specific for
Plasmodium species [
28]. The lower limit of detection of this method is 22 parasites/mL of whole blood [
29]. For samples containing parasite DNA by uPCR analysis,
Plasmodium species detection was attempted using nested PCR protocols specific to
P. falciparum (microsatellite marker Pk2),
P. vivax (microsatellite marker 3.502) and
P. malariae (18 s rRNA) as described previously [
28,
30,
31]. Samples for which
Plasmodium species could not be determined were reported as
Plasmodium species.
To detect polymorphisms associated with reduced susceptibility to artemisinin derivatives the open-reading frame of the PF3D7_1343700 kelch propeller domain was amplified using a nested PCR protocol [
1,
32]. Purified PCR products were sequenced at Macrogen, Republic of Korea and analysed using BioEdit version 7.1.3.0. using the 3D7 kelch13 sequence as reference (Accession: XM_001350122.1). The definition of single nucleotide polymorphisms (SNPs) was based on analytical approaches described previously [
1,
33].
Data management and statistical analyses
Data were collected on smartphones using ODK software [
34] and then managed by importing into OpenClinica [
35].
Definitions
a.
At each quarterly survey, the numbers of residents and participants in the study villages changed as residents moved away, refused to participate, travelled, died, joined the village, were born or returned from travel. Reasons for non-participation were categorized as: moved away for more than 1 month, short travel (defined here as less than 1 month), refusal, unable, not known, or ineligible. Because the categories short travel, unable to attend and refusal can overlap, they were combined.
b.
The ineligible category included the severely ill, children less than 6 months of age or participants who did not consent to a blood draw.
c.
A participant was defined as an individual who participated in the survey and agreed to give a blood sample.
d.
Coverage was estimated as the percentage of residents who provided a blood sample for uPCR analysis (numerator) divided by the number of invited residents (denominator). The invited residents represent the de facto population in the village at the time of the respective survey.
e.
Cohort participants were those aged 16 years and older who were found to be parasitaemic at M0 and followed monthly during the study period.
f.
Carriers were a subgroup of the cohort defined for the purposes of this study as being parasitaemic during four or more surveys. The definition was derived from the frequency distribution. Carriers represent the top quartile of the frequency distribution, 75 % of the cohort had fewer than four episodes.
To determine independent predictors for parasitaemia within the adult cohort, a logistic regression model was developed wherein the order of observations for each subject was considered by month (M0–11). Each member of the cohort was fitted as a random effect to adjust for any dependence of repeated events. Risk factors investigated included, village, sex, occupation (farmer or not), and age (continuous). Time-varying risk factors (which were asked at each survey) included self-reported history of fever, self-reported history of malaria, travel (0/1), and bed net use (0/1). Potential interactions between covariates were also explored. To ensure good model fit, the quadrature approximation used in the random-effects estimators was checked. A p value <0.05 was considered statistically significant. All analyses were performed using Stata, version 13 (StataCorp, College Station, TX, USA).
Additional data sources
Daily rainfall and maximum/minimum temperature data were collected from the Pailin Meteorological Office. Information on clinical (symptomatic) malaria episodes were collected by VMW, VHW and primary health centres for 2013–14.
Ethics approval
The study was approved by the Cambodian National Ethics Committee for Health Research (0029 NECHR, dated 4th March 2013) and the Oxford Tropical Research Ethics Committee (1015-13, dated 29th April 2013).
Discussion
The availability of a new diagnostic tool, highly sensitive PCR (uPCR) provides a new perspective on the epidemiology of malaria, allowing characterization of the subclinical and submicroscopic
Plasmodium reservoir [
17]. The limit of parasite detection in this study is approximately 50 times lower than with the extensively employed capillary blood filter paper based PCR methods [
36]. This is sufficient to identify the majority of infected individuals [
37]. Approximately four times more asymptomatic infected individuals were identified in these studies with this method than with conventional microscopy or rapid diagnostic tests [
36].
The study was conducted in an area of low seasonal malaria transmission at a time of strengthening malaria control and environmental change. The prevalence of
P. falciparum infections was already low and declined further while the prevalence of
P. vivax remained relatively stable. During the 1 year of follow up no obvious seasonality was detected. This disparity between the incidence of malaria and the prevalence of asymptomatic parasitaemia is also seen in areas of much higher transmission, such as the sub-Sahel, where there is also marked seasonality in the incidence of clinical malaria yet much less variation in asymptomatic carriage [
38,
39]. This illustrates the importance of asymptomatic malaria as the transmission reservoir infecting increasing numbers of anopheline vectors, as vectorial capacity rises abruptly at the beginning of the rainy season.
Ongoing molecular studies strongly suggest that the observed low densities of parasites are alive as they are expressing RNA. There is debate whether low-density parasitaemias are transmissible [
4]. Several studies from sub-Saharan Africa, Asia, South America, and Pacific Islands have demonstrated that submicroscopic falciparum as well as vivax infections can be transmitted to mosquitoes [
40‐
47]. These discussions usually revolve around the transmission potential at the time of sampling, but would be more epidemiologically relevant if they considered the transmission potential of an individual over time. An individual may have a low-density non-transmissible infection on 1 day, but later densities may rise and infect biting anopheline mosquitoes. A recent study conducted in Papua New Guinea detected
P. falciparum in 19 % and
P. vivax in 13 % of 2083 samples using conventional PCR. Gametocytes could be detected in 60 % of
P. falciparum-positive and 51 % of
P. vivax-positive samples by detection of pfs25 and pvs25mRNA transcripts [
48]. Gametocytaemia is probably present in all chronic infections but densities fluctuate out of phase with asexual parasitaemia waves.
A cohort of adults with Plasmodium parasitaemia at the time of the first survey was followed over a 12 months period in monthly intervals. In this cohort P. falciparum infections were short lived; three participants remained infected for 2–4 months. By contrast a third of the participants with P. vivax infections remained parasitaemic for more than half of the following surveys. Three different patterns of persistence were observed. First, the infection could be cleared within the first months of the study period and did not reoccur. Second, no parasitaemias were detected during 1 or 2 months, only to reappear during subsequent months. Third, in a minority of participants persistent parasitaemias could be detected during each survey. Changes in genotype profiles over time within individual participants are currently under investigation.
In participants who control their infections relatively well, as illustrated by three participants who remained parasitaemic and asymptomatic during 12 surveys, the parasite densities remain low and the oscillation amplitude was less than one hundred fold. Oscillating parasite densities have been described previously in vivax as well as falciparum malaria therapy for patients with neurosyphilis [
14,
49] and a more recent study conducted in an area of PNG with very high
P. falciparum, P. vivax and
P. malariae prevalence [
12]. In PNG 70 asymptomatic villagers were sampled every third day for a period of 61 days. Infections often lasted for several weeks in young children but generally for only a few days in adults. The periodicity of infections observed in PNG was attributed to sequestration of synchronously replicating
P. falciparum parasites as reported elsewhere [
50]. Periodic oscillation of
P. falciparum densities has also been reported in asymptomatic children in rural Tanzania sampled daily for 14 days [
51]. It seems more likely that parasite multiplication varies over time, controlled by mechanisms such as antibody responses to variant surface antigens [
52,
53].
The decline in
P. falciparum prevalence observed in this study could be short-lived. Half of the
P. falciparum strains had a mutation (C580Y) in the gene coding for the K-13 propeller which is associated with reduced susceptibility to artemisinin derivatives. Artemisinins used in combination with a partner drug are the global first line treatment for uncomplicated malaria. In the presence of artemisinin resistance
P. falciparum strains with reduced susceptibility of the partner drug emerge and spread rapidly [
22,
23,
54]. The complete replacement of wild type with C580Y mutant strains in one village suggests a need for urgent action to interrupt the transmission of the remaining
P. falciparum strains [
55,
56].
The study also shows a shift in malaria burden from P. falciparum to P. vivax which is increasingly responsible for malaria related morbidity and mortality in the region. The radical cure of patients with persistent vivax infections has a high priority not only for the benefit of the patient but to minimize the transmission of this infection. Radical cure of vivax infections requires use of 8-aminoquinolines such as primaquine. Fear of haemolysis in G6PD deficient patients and the absence of a practical, low-cost rapid diagnostic test to diagnose G6PD deficiency have so far obstructed the wider use of this class of drugs.
Authors’ contributions
Study design: RT, TP, NJW, AD. Molecular biology: MI. Data management: JC, Psi. Statistics: SJL. All authors have contributed to writing of the versions of the paper. All authors read and approved the final manuscript.