Background
Since 2000, huge efforts in malaria control have resulted in a 47 % decrease in malaria-associated mortality worldwide and in a 30 % decrease in incidence of malaria [
1]. Unfortunately, these achievements are threatened by the emergence and spread of
Plasmodium falciparum resistance to artemisinin and partner drugs. Current efforts to contain artemisinin resistance, consisting of a wide range of activities, such as long-lasting insecticide-treated net campaigns, implementation of accurate and widely-available malaria rapid diagnostic tests (RDT), banning of artemisinin monotherapies, and universal access to artemisinin-based combination therapy (ACT) [
2], have yet to show success. Elimination of artemisinin resistance through direct elimination of the
P. falciparum parasite may be the only present strategy [
3].
In low transmission settings, the human asymptomatic and/or microparasitaemic reservoir is an important challenge in the context of malaria and artemisinin resistance elimination, as this reservoir typically escapes routine surveillance, but can contribute to active transmission [
4,
5]. However, successful attempts to identify and quantify the asymptomatic reservoir have only rarely been documented.
Artemisinin resistance has been reported in Western Cambodia since 2008 [
6‐
9]. Artemisinin resistance is defined as delayed parasite clearance, which represents a partial resistance. Recently, a molecular marker of artemisinin resistance was identified. Mutations in the Kelch 13 (k13)-propeller domain (especially C580Y) were shown to be associated with delayed parasite clearance in vitro and in vivo [
10]. In 2013, Médecins Sans Frontières (MSF) in collaboration with the Ministry of Health initiated a programme for the elimination of
P. falciparum in Chey Saen district, Preah Vihear province, in the North of Cambodia, a region with documented artemisinin resistance. This programme included two successive prevalence surveys (relying on molecular methods for the identification of asymptomatic carriers and of individuals carrying resistant parasite strains), and support and scale-up of the passive case detection network (relying on a strong collaboration with the public sector—Health Centers, Health Posts, Village Malaria Workers, and Malaria Contact Persons identified in important plantations and settlements—and the registered private sector in the district). Considering the dearth of information on the performance of surveys and routine surveillance in low malaria transmission settings, the objective of this study was (i) to document the prevalence and incidence of
Plasmodium spp. and
P. falciparum at district and village level in Chey Saen district, as assessed through respectively a prevalence survey and a passive case detection system (the aim was to be able to target villages with higher transmission and indicators of resistance), and (ii) to document the molecular and clinical indicators for artemisinin resistance in cases identified through both approaches.
Methods
Study setting
The study was conducted across all villages of the district of Chey Saen, Preah Vihear province, located in the North of Cambodia, bordering Thailand and the Lao People’s Democratic Republic. The district consisted of 21 villages, with two sub-villages considered as independent from the main village, resulting in 23 geographical units included in the study. The district population consisted of 22,343 individuals in 4585 households, and village sizes ranged from 328 to 2016 inhabitants (67–397 households). The study concerned the entire district population, but did not cover workers from a major Chinese company in the district, as access to this population was not possible.
Study period
The prevalence study was conducted in September and October 2014. The study period for the passive case detection covered June 2014 to May 2015.
Study population and sample size
For the prevalence survey, random sampling was done from the entire district population (a population census was done prior to the survey); children less than 1 year were excluded. Sample sizes were calculated per village, with an assumed P. falciparum prevalence of 7 %, a confidence of 95 %, a precision of 5 %, and a correction for finite populations based on each village size. If the target sample size for a village was not reached, a second sampling was done. To minimize absenteeism, participants were given an incentive for their participation (krama, a piece of local fabric).
For the passive case detection (PCD) component, all individuals presenting with fever and/or other malaria symptoms (headache, fatigue, body pain, nausea, etc.), in the public and private registered sector were included. Unregistered private providers were not included in the surveillance system. Private providers were mapped through consultative meetings with the village chiefs, community elders, and representatives of the private sector.
Prevalence survey—methodology
For each participant of the prevalence survey, 3 ml of whole blood was drawn (0.5 ml for children less than 5 years of age) in EDTA tubes. Samples were centrifuged within 24 h to separate plasma and buffy coat, stored in a freezer (−20 °C), and transported to the Malaria Molecular laboratory at Mahidol University, Bangkok, Thailand, in cold chain. For each sample, high volume quantitative real-time PCR (qPCR) was performed, with a detection limit of 22 parasites per millilitre [
11]. Samples were lysed in protease-containing buffer, and DNA extraction was performed using the QIAamp Blood Midi Kit (Qiagen). For samples containing parasite DNA,
Plasmodium species detection was attempted using PCR protocols specific to detect the 18S rRNA gene of
Plasmodium spp. Samples for which
Plasmodium species could not be determined were reported as
Plasmodium species. In each screening run, controls assessing both the DNA extraction and the PCR steps were included (extraction/PCR process, cross contamination and reagents).
Plasmodium falciparum-positive samples were screened for mutations in the k13-propeller domain gene, a molecular marker associated with artemisinin resistance [
12]. Purified PCR products were sequenced at Macrogen, Republic of Korea (BioEdit v. 7.1.3.0., using the 3D7 kelch13 sequence as reference). The definition of single nucleotide polymorphisms (SNPs) was based on analytical approaches described previously [
13,
14].
Passive case detection—methodology
Case detection in the public—health facilities and village malaria workers (VMW)—and private sector was done by rapid test. The public sector used the First Response Malaria Ag. pLDH/HRP2 Combo Card Test (Premier Medical Corporation Limited, India till December 2014 and switched to SD BIOLINE Malaria Ag P.f/P.v (Standard Diagnostics, Korea) in 2015. The private sector usually relied on First Response Malaria or Malacheck (Standard Diagnostics, Korea). P. falciparum-positive cases (mono-infections and mixed infections) received dihydroartemisinin (DHA) and piperaquine (PPQ) (Eurartesim®, Sigma Tau, Italy) for 3 consecutive days (first dose under supervision, dosage according to national guidelines), and were tested for G6PD deficiency using the G6PD rapid test (G6PD Rapid Test CARESTART, Access Bio, USA). G6PD wild type patients were treated with a single dose of primaquine (0.25 mg/kg, REMEDICA, Cyprus). Patients were followed up 28 days (D28) after their initiation of treatment to establish treatment effectiveness, using the same qPCR technique as in the prevalence survey. D28 was chosen instead of D42 (traditional endpoint when using microscopy), to detect parasitological failures early enough and avoid transmission to the community of resistant parasites. Patients with D28 sustained parasitaemia were offered atovaquone-proguanil (Malarone®, GlaxoSmithKline, Belgium) as second-line treatment. G6PD screening, Primaquine treatment and D28 follow-up were implemented mid-December 2014.
A case investigation form (CIF) was completed by the MSF team for each confirmed malaria case (adapted from the WHO [
15]). The CIF included demographic information and other characteristics, a history of the current illness including diagnostic test results and treatment, travel history, and possible mode of transmission.
Testing, DHA–PPQ treatment, and training on case management were provided by the public and private sectors. G6PD testing, primaquine treatment, and D28 follow-up were implemented by MSF teams. Health promotion messages were provided both by the public and private sectors and MSF teams.
Statistical analysis
Prevalence calculations were conducted using sampling weights (W), calculated using village size (N) and number of respondents in each village (n) as follows: W = N/n. For prevalence at district level, the village was specified as sampling strata and a calibration adjustment was performed using a logit method [
16]. The population census done in June 2014 was used and calibration variables were sex and age category (0–4, 5–14, 15–34, 35–65, 65–Inf). Considering the criteria for eligibility, the sample design and the calibration adjustment, the study population at the district level was estimated as representative of the population living in Chey Saen district. Confidence interval calculations used the scaled Chi squared distribution for the log likelihood from a binomial distribution [
17]. District and village prevalence are presented with their 95 % confidence interval. Statistical analyses were conducted using R version 3.1.1 (R Development Core Team, 2014).
For PCD, the main outcomes were the monthly number of screened cases and confirmed P. falciparum (mono-infections and mixed infections) cases and the annual number of individuals treated per village, per month and per provider type. Incidences per village were calculated per year and per 1000 inhabitants using data from population census. Monthly incidences were calculated at district level.
Ethical considerations
Individual written consent was obtained before inclusion in the prevalence survey and for all components of the MSF programme not included in the national guideline. Ethical approval was granted by the Cambodian National Ethics Committee for Health Research (280 NECHR) and the Ethics Review Board of Médecins Sans Frontières (ID1401).
Discussion
The primary objectives of this study were to estimate the prevalence of
Plasmodium spp. and
P. falciparum at district and village level in Chey Saen district and to compare this with the incidence reported through the passive case detection system. At the same period of the year (September and October), the prevalence of
Plasmodium spp. was slightly elevated (from 2.61 to 3.73) compared to the findings of a survey conducted in 2013 [
18].
Despite the fact that the current survey relied on PCR testing of high blood volume, which is more sensitive than PCR from a filter paper dried blood spot (volume around 5 µl) as used in the 2013 survey, results showed a decrease at district level in the prevalence of
P. falciparum (with a slight overlap of confidence intervals).
Plasmodium falciparum prevalence remains low in Chey Saen district. Similar low prevalence and variation in the dynamic of transmission have also been observed, using this sensitive molecular method. For instance in Pailin [
18], the prevalence of
P. falciparum dropped from about 2 % in June 2013 to much lower than 1 % in June 2014. [
19]. In addition, MSF programme has been running in 2014, focusing on
P. falciparum cases management. The decrease in the ratio
P. falciparum/
P. vivax is typical in areas of pre-elimination when treatment programmes (with ACT) are correctly functioning for each malaria episode, without implementing a radical cure for patients with persistent
P. vivax infections.
The study had a number of limitations. The prevalence survey was designed to estimate prevalence at village level and to capture all parasite carriers from a representative sample, but a bias may have been introduced due to preferential absenteeism of the male young adult population (the most at risk sub-population). However, as characteristics of respondents in terms of gender and age category were close to district population, and age and gender-adjustment was performed, a sizeable underestimation of the true district prevalence was not anticipated. Likewise, while temporary workers living in the compound of private companies (i.e. plantations) were not reachable, which may have led to an underestimation of the prevalence, this specific group was probably limited in size.
Overall, passive case detection and prevalence survey showed very similar results to the extent that the two geographical clusters with high number of cases were identified by both approaches. Six out of the seven villages with higher incidence were also identified through the prevalence survey, suggesting that villages with high incidence also had a sizeable asymptomatic reservoir. However, the prevalence survey failed to identify
P. falciparum cases in 15 villages where cases were detected passively, and the village with highest malaria incidence had no positive cases in the prevalence survey. This may be the consequence of the survey design, as it was not powered for detection of low prevalence at village level as well as areas with specific and potentially limited in size at risk populations. Forest goers are also more likely to be out of the village during the survey. If parasite prevalence is indeed a better surveillance measure for elimination programs than numbers of symptomatic cases [
20], surveys have to enroll high numbers of participants per geographical unit studied. This may render the monitoring of elimination interventions highly resource-demanding in low prevalence settings. Passive and active case detection (focusing on at-risk populations) may thus remain an important tool to monitor malaria elimination.
Results can suggest that transmission in the villages is limited. Discrepancies between incidence and prevalence might be partly explained by the absence of infected people during the prevalence survey. Those who stay more permanently in the villages (hence more often included in the survey) would be less likely to be positive. In addition, prevalence of asymptomatic cases is either not detected in 15 villages, or very low in the seven others. Symptomatic patients may have been infected elsewhere (e.g. in the forest) and not through local transmission in the villages.
The distribution of positive P. falciparum cases per age group also suggests that infection occurs outside the villages: cases passively detected in the 0–4 age group represent 1.9 % of total cases (and 0 % in the prevalence survey), while children under five account for 11 % of the overall population. Young children are staying most of the time inside the villages, as they are indeed less susceptible to accompany adults outside. Finally the fact that 85.9 % of P. falciparum cases were men (and 54.3 % in the 15–34 age group), is also an indication of limited village transmission: women are well known to stay much more inside the village (despite not to the extent of young children). This possible transmission pattern outside of the villages should be confirmed with further studies as this has a direct impact on elimination strategies, particularly on vector control.
Amongst the 16 P. falciparum positive samples of this survey, only 11 could be screened for mutations (indicating artemisinin resistance), and six (55 %) presented the mutant-type k13 allele (C580Y) closely associated with artemisinin resistance. Asymptomatic individuals could thus be an additional reservoir for artemisinin resistance. Cases were found both in Eastern and Western part of the district, which means that k13 resistance-associated mutations are not confined to a limited area. This result was in accordance with the findings of the 2013 survey (seven P. falciparum samples out of 11–64 %-presented the mutant-type k13 allele), and suggests artemisinin resistance is a public health threat in Chey Saen district (realizing the absolute figures remain low). An additional reason for concern is the analysis of the treatment efficacy for symptomatic cases, with more than a third of all cases still positive with PCR 28 days after administration of first-line treatment. While poor treatment adherence and poor drug quality/storage conditions cannot be excluded in this study, recent drug efficacy studies using the standard WHO therapeutic efficacy study protocol for symptomatic cases have recently observed similar high treatment failure rates in two provinces bordering Preah Vihear province, with an adequate clinical and parasitological response at D42 in Siem Reap of 37.5 % and in Stung Treng of 66.7 % (Results from ‘TES and K13 Surveillance’, CNM, presentation at Malaria Elimination Partners Convening—May 2015).
Overall these results highlight the low effectiveness of DHA–PPQ as first-line treatment in many parts of Cambodia, and that resistance/treatment failure has been moving from the ‘classical’ Western Thai-Cambodian border area towards the Northern area. The findings of this study indicate that DHA–PPQ may no longer be effective for symptomatic cases with high parasitaemia. This may be due to resistance to artemisinin, as well as to the piperaquine partner drug [
21‐
24]. On the other hand, D28 on asymptomatic cases identified through the prevalence survey were all undetectable, suggesting that clearance of low parasitaemias in asymptomatic carriers was still achieved with this drug combination. To demonstrate however full efficacy on asymptomatic carriers, a longer follow up would be necessary in a larger sample size. This is important for the choice of anti-malarial treatment in mass drug administrations, aimed to target the asymptomatic reservoir.
These data illustrate the importance of using highly sensitive diagnostic tools to identify the asymptomatic carriers. They invite to take action against this reservoir, which contribute to maintaining transmission.
They also show big variations between villages, justifying different type of interventions (in addition to good traditional control programmes) depending on the levels of prevalence and incidence: active case detection aiming at profiling and targeting at risk people, reactive case detection (whose positivity rate is higher than passive case detection in the area), or targeted malaria treatment with mass drug administration proposed to certain groups.
Authors’ contributions
JMK, MDS, RVDB, AD, WE and CN conceived and designed the study; GF organized, facilitated and supervised the acquisition of data in Cambodia; GF performed the statistical analysis; MI carried out the laboratory analysis; All authors participated in the interpretation of the results; GF and JMK wrote the paper; MDS, RVDB and AD revised the manuscript for substantial intellectual content; All authors read and approved the final manuscript.