Preliminary review on the prevalence, proportion, geographical distribution, and characteristics of naturally acquired Plasmodium cynomolgi infection in mosquitoes, macaques, and humans: a systematic review and meta-analysis

Background Plasmodium cynomolgi is a simian malaria parasite that has been reported as a naturally acquired human infection. The present study aims to systematically review reports on naturally acquired P. cynomolgi in humans, mosquitoes, and macaques to provide relevant data for pre-emptive surveillance and preparation in the event of an outbreak of zoonotic malaria in Southeast Asia. Methods The protocol of the systematic review was registered at PROSPERO with approval ID CRD42020203046. Three databases (Web of Science, Scopus, and MEDLINE) were searched for studies reporting the prevalence of P. cynomolgi infections in Southeast Asian countries between 1946 and 2020. The pooled prevalence or pooled proportion of P. cynomolgi parasitemia in humans, mosquitoes, and macaques was estimated using a random-effects model. Differences in the clinical characteristics of P. cynomolgi infections were also estimated using a random-effects model and presented as pooled odds ratios (ORs) or mean differences (MDs) with 95% confidence intervals (CIs). Results Thirteen studies reporting on the prevalence of naturally acquired P. cynomolgi in humans (3 studies, 21 cases), mosquitoes (3 studies, 28 cases), and macaques (7 studies, 334 cases) were included. The results demonstrated that the pooled proportion of naturally acquired P. cynomolgi in humans was 1% (95% CI, 0.1%, I2, 0%), while the pooled proportion of P. cynomolgi infecting mosquitoes was 18% (95% CI, 10–26%, I2, 32.7%). The pooled prevalence of naturally acquired P. cynomolgi in macaques was 47% (95% CI, 27–67%, I2, 98.3%). Most of the cases of naturally acquired P. cynomolgi in humans were reported in Cambodia (62%) and Malaysia (38%), while cases of P. cynomolgi in macaques were reported in Malaysia (35.4%), Singapore (23.2%), Indonesia (17.3%), Philippines (8.5%), Laos (7.93%), and Cambodia (7.65%). Cases of P. cynomolgi in mosquitoes were reported in Vietnam (76.9%) and Malaysia (23.1%). Conclusions This study demonstrated the occurrence of naturally acquired P. cynomolgi infection in humans, mosquitoes, and macaques. Further studies of P. cynomolgi in asymptomatic human cases in areas where vectors and natural hosts are endemic are extensively needed if human infections with P. cynomolgi do become public health problems. Supplementary Information The online version contains supplementary material available at 10.1186/s12879-021-05941-y.

P. cynomolgi has accidentally and experimentally been reported as a cause of human malaria [3][4][5], but there was a case report showing this parasite as a naturally acquired human infection in a Malay woman from the east coast of Peninsular Malaysia who lives in an area where long-tailed macaques are present [6]. P. cynomolgi was first observed in 1907 in Macaca fascicularis collected in Java [7]. P. cynomolgi has morphological features similar to P. vivax, as shown by microscopy, including the asexual cycle (48 h), prepatent periods, and presence of hypnozoites, which can initiate relapses [6]. The identification of P. cynomolgi relies on the amplification of the 18S ribosomal RNA (rRNA) gene by nested polymerase chain reaction (PCR) sequencing. As P. cynomolgi is recorded as the most recently discovered simian malaria parasite infecting humans, its prevalence, proportion, geographical distribution, and characteristics remain unclear. The present study aimed to systematically review reports on naturally acquired P. cynomolgi in humans, mosquitoes, and macaques to provide relevant data for pre-emptive surveillance and preparation in the event of an outbreak of zoonotic malaria in Southeast Asia.

Protocol for reporting the systematic review and metaanalysis
The systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [8]. The protocol of the systematic review was registered at the PROSPERO International Prospective Register of Systematic Reviews with approval ID CRD42020203046.

Search strategy
Studies reporting the prevalence of P. cynomolgi parasitemia in humans, mosquitoes, and macaques in Southeast Asian countries between 1946 and 2020 were systematically searched in three databases: Web of Science, Scopus, and MEDLINE. The search term used to find the potentially relevant studies was "cynomolgi". The searches of the reference list in the included studies or the review articles and the searches of the additional source(s) such as Google Scholar were performed to maximize the number of the included studies and to prevent any missing studies during the searches of the main databases.

The eligibility criteria and study selection
The inclusion criterion for selection in the present study was that any included studies must be primary studies reporting the prevalence or incidence of P. cynomolgi parasitemia in humans, mosquitoes, and macaques. The exclusion criteria were (1) studies without data of interest; (2) studies including populations outside Southeast Asian countries where sufficient reports on P. cynomolgi could be retrieved; (3) studies without full text; and (4) studies published as case reports or case series, reviews, editorials, letters to the editor, in vitro/experimental studies, mosquito experiments, animal experiments, human experiments, or studies identifying P. cynomolgi in the same participants. Two authors (MK, FRM) independently screened the studies according to the eligibility criteria for potentially relevant studies. Any disagreement between the two authors was resolved through discussion or consultation with the third author (KUK) for the finalization of the study inclusion.

Data extraction
Data from the included studies were extracted for pilot standardization. The following data of each study were extracted: author name, year of publication, year of study, study site, types of participant or sample (humans, monkeys, or mosquitoes), age, gender ratio, types of PCR used for P. cynomolgi identification, target gene for PCR, number of Plasmodium species identified, and parasite density. Data extraction was performed by one author (MK) and cross-checked by two authors (FRM and GDM).

Quality of the included studies
The methodological quality of the included studies was assessed using the adapted version of the Newcastle-Ottawa Scale (NOS) [9] with a maximum of three scores (Is the Case Definition Adequate?, Representativeness of the Cases, and Ascertainment of Exposure). For the present study, studies achieving a NOS score of two or greater were considered high-quality studies.

Data synthesis
The pooled prevalence or the pooled proportion of P. cynomolgi parasitemia compared to all Plasmodium species or all cases enrolling humans, mosquitoes, and macaques was estimated using a random-effects model. In the case of a very low prevalence of P. cynomolgi infection compared to the number of participants enrolled, the pooled proportion instead of the pooled prevalence was used to estimate the proportion of P. cynomolgi parasitemia per Plasmodium species identified in the same participants. The difference in the geographical distribution was visualized by mapping the location(s) of the P. cynomolgi parasitemia provided by the included studies. Differences in the clinical characteristics of P. cynomolgi infections between the included studies were estimated using a random-effects model and demonstrated as pooled odds ratios (ORs) or mean differences (MDs) with 95% confidence intervals (CIs). The heterogeneity across the included studies was assessed using Cochrane's Q and I 2 (inconsistency) statistics.

Publication bias
Publication bias across the included studies was assessed using the funnel plot and Egger's test. In the absence of publication bias, the funnel plot should approximately resemble a symmetrical funnel, while an asymmetrical appearance indicates a small-study effect that will overestimate or underestimate the pooled effect [10].

Search results
The initial search yielded 3381 articles (Fig. 1). Out of those articles, 686 articles were removed as they were duplicates. After screening the titles and abstracts of the remaining 2695 articles, 2192 articles were excluded as they did not meet the inclusion criterion. Following a full-text review of 503 articles, a further 491 articles were excluded with reason: 216 were animal experiments, 193 were in vitro/experimental studies, 31 were mosquito experiments, 29 were review articles, 11 were human experiments, 5 were case reports of P. cynomolgi in humans, 2 had no infection of P. cynomolgi in humans, 1 was a case report not on P. cynomolgi in humans, 1 involved performing identification in the same participants, 1 was not performed in Southeast Asia, and 1 was not a full-text article. Therefore, 12 articles were included in the study. Another relevant article was retrieved from the searches of the reference lists of included articles or review articles. Finally, 13 studies were included in the present study. Out of the 13 studies selected, 7 studies [11][12][13][14][15][16][17] reporting on P. cynomolgi infection in macaques, 3 studies [18][19][20] reporting on P. cynomolgi infection in humans, and 3 studies [21][22][23] reporting on P. cynomolgi infection in mosquitoes were included in the final analysis.

Characteristics of the included studies
The characteristics of the included studies are shown in Table 1. Thirteen studies [11][12][13][14][15][16][17][18][19][20][21][22][23] reporting the prevalence or incidence of naturally acquired P. cynomolgi in humans (3 studies, 21 cases), mosquitoes (3 studies, 26 cases), and macaques (7 studies, 334 cases) were included for qualitative and quantitative syntheses. Details of P. cynomolgi mono and mixed infections in humans, Anopheles, and macaques are shown in Table 2. Among the 3 studies reporting on P. cynomolgi in humans between 2013 and 2017, one study was conducted in Northern Sabah of Malaysia [18], one in Pailin and Battambang provinces of Cambodia [19], and one in Sarawak of Malaysia [20]. Two studies [18,19] enrolled participants in the communities, while another study [20] enrolled malaria-positive cases for their studies. All studies used semi-nested or nested PCR with sequencing to identify Plasmodium species infection. All three studies [18][19][20] used semi-nested or nested PCR for amplification of the SSU rRNA gene. PCR amplification of the cytochrome C oxidase gene was used for DNA sequencing to confirm P. cynomolgi and P. knowlesi coinfections in one study [20]. Of the 21 cases of people who were infected with P. cynomolgi, 16 cases (76.2%) were male. The age of patients was 43 and 63 years in a study by Grignard et al. [18], median 28 years (range 7-64 years) in a study by Imwong et al. [19], and median 43 years (range 17-65 years) in a study by Raja et al. [20]. P. cynomolgi mono-infection was in 13 cases (61.9%), while P. cynomolgi mixed infection with other Plasmodium species was in 8 cases (38.1%). Mixed infections included P. cynomolgi with P. knowlesi (6 cases, 75%) and with P. vivax (2 cases, 25%). Only one study [18] reported on the occupation of patients and also observed monkeys. Naturally acquired P. cynomolgi in humans was reported in Cambodia (62%) and Malaysia (38%). Details of P. cynomolgi infections in humans are shown in Table 3.
Among the 7 studies reporting on 334 macaques infected with P. cynomolgi between 2004 and 2019, 4 studies (57.1%) [11,12,15,16] were conducted in Malaysia, 1 in the Philippines [13], and 1 in Singapore [14]. The study flow diagram. Twelve studies were from the database searches, and one study was from the searches of review articles Another study [17] was conducted in five countries, including the Philippines, Indonesia, Singapore, Cambodia, and Laos. Four studies [11,13,14,17] identified Plasmodium species in long-tailed macaques, while 3 studies [12,14,16] identified Plasmodium species in both longtailed and pig-tailed macaques. All studies used nested PCR for amplification of the 18S rRNA gene except a study by Muehlenbein et al. [16], which used nested PCR for amplification of the cytochrome b (cytb) gene. One study did not report the status of P. cynomolgi mono or mixed infection [16]. P. cynomolgi monoinfection occurred in 148 cases (50.3%), while P.

Quality of the included studies and publication bias
The methodological quality of the included studies was assessed using the adapted version of the NOS (Table  S1). Overall, all 13 studies achieving NOS scores of three were reviewed. Publication bias among the included studies could not be assessed using the funnel plot and Egger's test due to the low number of the included studies, as the analysis required a minimum of 10 studies for each group of participants enrolled [10].
Geographical distribution of P. cynomolgi infection in humans, mosquitoes, and macaques P. cynomolgi mono-infection cases were predominantly in humans and Anopheles vectors, while an almost equal number of cases were reported in macaques. The proportion of P. cynomolgi cases in humans, Anopheles, and macaques per Southeast Asian country is shown in Fig. 2.
The results showed that P. cynomolgi infection in humans was demonstrated in Malaysia and Cambodia. P. cynomolgi infection in Anopheles was demonstrated in Vietnam and Malaysia. P. cynomolgi infection in macaques was demonstrated in Malaysia, the Philippines, Indonesia, Cambodia, Laos, and Singapore. Human infection with P. cynomolgi was reported in Cambodia (n = 13) and Malaysia (n = 8). Anopheles infections with P. cynomolgi were reported in Vietnam (n = 20) and Malaysia (n = 8), while the highest numbers of macaques infected with P. cynomolgi were reported in Malaysia (n = 125) and Singapore (n = 66). The majority of the reports of infected macaques in Singapore (n = 40) did not identify whether the cases were mono or mixed infections. Studies on the presence of both vectors and natural hosts also reported human infection cases, as was the case in Malaysia (Fig. 3). P. cynomolgi infections in humans were reported in Cambodia and Malaysian Borneo, while P. cynomolgi infections in macaques were reported in the Philippines, Malaysia, Laos, Singapore, and Indonesia. P. cynomolgi infections in Anopheles mosquitoes were reported in southern Vietnam and Malaysian Borneo.

The proportion of P. cynomolgi infection in humans
The pooled prevalence of P. cynomolgi in humans was very low (< 0.1%). Because a low prevalence of P. cynomolgi in humans was reported in the included studies, the pooled proportion of P. cynomolgi compared to Plasmodium species infections in humans was estimated.

The proportion of P. cynomolgi parasitemia in mosquitoes
The pooled prevalence of P. cynomolgi in mosquitoes was very low at 0-1%. Because a low prevalence of P. cynomolgi in mosquitoes was reported in the included studies, the pooled proportion of P. cynomolgi compared to Plasmodium species infections in mosquitoes was estimated. The results demonstrated that the pooled proportion of P. cynomolgi infecting mosquitoes was 18%, with low heterogeneity (95% CI: 10-26%, I 2 : 32.7%) (Fig. 5). The highest proportion of P. cynomolgi infection in mosquitoes was demonstrated in northern Sabah (26, 95% CI: 13-46%) [22].

Streamlined publication years
To reduce a large variance in the pooled proportion of P. cynomolgi infection between the three hosts and to describe the actual prevalence of P. cynomolgi infection in the past and present day, the cut-off point for publication years was streamlined for the purpose of this study.  (Fig. 8).

Discussion
The successful transmission of zoonotic malaria is highly dependent on the bionomics and distribution of competent vectors and natural hosts. Humans and macaques live in and share overlapping spaces, particularly those who work in farming or agriculture near forests or tourists who come to visit areas where macaques are endemic. These close contacts and sharing of geographic distributions between humans and macaques and the presence of suitable vectors can lead to malarial disease transmission between host species. The increase in zoonotic malaria before the discovery of P. cynomolgi was primarily found in the occurrence of P. knowlesi malaria. P. knowlesi infection in humans is driven by multiple factors, including anthropogenic land-use factors leading to changes in the transmission pattern of the parasite between macaque reservoirs, vectors, and humans, as observed in Malaysian Borneo [24]. Deforestation for palm oil plantations or other clearing activities resulted in the loss of natural habitats for macaques, which led to their close contact with human settlements and increased incidence of P. knowlesi infection in humans [25]. The present study demonstrated a high prevalence and density of P. cynomolgi in macaques (47%), particularly in Indonesia [17] and Singapore [14,17]. Interestingly, half of the P. cynomolgi infections in macaques were mixed infections with other Plasmodium species, and among the mixed infections, triple and quadruple infections were the most common types of mixed infections observed. It should be noted that mono-infection cases cynomolgi infection in macaques was demonstrated in Malaysia, the Philippines, Indonesia, Cambodia, Laos, and Singapore. Note the higher proportion (n = 40) of undeclared speciation in Singapore compared to mono or mixed cases combined were predominantly in humans and Anopheles vectors, while an almost equal number of cases were reported in macaques. More interestingly, P. cynomolgi was likely to be a mixed infection with P. inui (90.4%), P. coatneyi (51.4%), and P. knowlesi (45.9%). The latter seems to support the similarity of transmission between P. cynomolgi and the more well-known zoonotic malaria caused by P. knowlesi. Zoonotic malaria caused by P. knowlesi infections was described and focused on in Sarawak, Malaysian Borneo, in 2004 [1], and later, it became a more common cause and was recognized as the fifth human malaria parasite throughout Southeast Asia [2]. Similar to P. knowlesi infection, naturally acquired P. cynomolgi infection in Malaysia was reported 10 years later in 2014 in endemic cases of people who lived in the same area with long-tailed macaques [6]. From that time to 2020, a tourist who traveled to Surat Thani Province, Thailand, was naturally infected with P. cynomolgi [26]. Then, naturally acquired P. cynomolgi infections in humans were detected in other parts of Southeast Asia, including Malaysia and Cambodia [18][19][20]. Moreover, when comparing the pooled proportion of P. cynomolgi infections in three different hosts in the studies conducted before 2013 and after 2013, the pooled proportion of P. cynomolgi infections in mosquitoes was higher in studies conducted after 2013. Nevertheless, the pooled proportion of P. cynomolgi infections in macaques was lower in studies conducted after 2013. The results of these streamlined publication years suggested that P. cynomolgi infections in humans might be due to the proximity of suitable vectors and monkeys, leading to conditions favorable for interspecies transmission, as demonstrated in cases of P. knowlesi infection [11]. Figure 3 shows that regions with the presence of both vectors and natural hosts also reported human infection cases, as was the case in Malaysia. However, human transmission in Cambodia seemed to have been caused by the presence of macaques in the country and its proximity to southern Vietnam, which reported the most cases of infected Anopheles vectors. Although cases in macaques have been reported in the Philippines, Laos, Singapore, and Indonesia, there is a curiosity as to why there is a lack of reports on infected Anopheles mosquitoes and humans. The results of the present study indicated that in areas where a high prevalence of P. cynomolgi was seen in macaques, sustained public information and advocacy in the affected areas is still necessary even if public advocacies have already been performed since 2014 [27][28][29][30].
Although a high prevalence of P. cynomolgi in macaques was demonstrated by the included studies, the prevalence of P. cynomolgi in mosquitoes was very low at 0-1%, and the proportion of P. cynomolgi compared to all Plasmodium species detected in mosquitoes was 18%. Therefore, the transmission of P. cynomolgi from macaques to mosquitoes should be limited by the bite Fig. 5 The estimated proportion of P. cynomolgi in vectors. ES: estimated proportion, overall: overall proportion, I 2 : level of heterogenicity, p: pvalue less than 0.05 is statistically significant Fig. 4 The estimated proportion of P. cynomolgi malaria in humans. ES: estimated proportion, overall: overall proportion, I 2 : level of heterogenicity, p: p-value less than 0.05 is statistically significant rates, the susceptibility of Anopheles vectors, and the parasite density of the primate host. The contrast between a high prevalence of P. cynomolgi in macaques and a lower parasitemia of Anopheles mosquitoes should be due to the limited susceptibility of Anopheles vectors to harbor a lower parasitemia of P. cynomolgi in a specific environmental condition, which can lead to disease transmission, as changes in host preference, biting behavior, and adaptation of the mosquitoes to habitat changes could affect the transmission of zoonotic malaria in Southeast Asia. Although the proportion of P. cynomolgi infecting Anopheles mosquitoes was low, the importance of transmission through the infection of human hosts by the bite of the infected Anopheles mosquitoes should be monitored. In addition, although the pooled proportion of naturally acquired P. cynomolgi in human hosts demonstrated by the included studies was 1%, the high density of infected mosquitoes, such as in Northern Sabah (26%) [22], could lead to a high prevalence of naturally acquired P. cynomolgi in humans, as demonstrated in Northern Sabah (4%) [18]. P. cynomolgi infection in humans was observed to be asymptomatic and submicroscopic [18,19]. This implied that the mono-infection of P. cynomolgi was very low in parasite density despite the mixed infection with P. vivax as demonstrated by Imwong et al. [19]. The morphological life cycle of relapse and the genetic similarities of P. cynomolgi and P. vivax could lead to undetected or undiagnosed P. cynomolgi cases in Southeast Asia. Most of the cases with suspected malaria admitted to the hospital were identified by the microscopic method, which has a low sensitivity and specificity to differentiate P. cynomolgi from P. vivax. In addition, the misidentification or missed detection of submicroscopic parasitemia by P. cynomolgi might lead to under-reported cases. Moreover, failure to treat P. cynomolgi malaria can lead to malarial recurrences, as observed in two individuals from Charkrya and Ou Treng, Cambodia [19]. Even in symptomatic infections of P. cynomolgi, such as in the mixed infection of P. cynomolgi and P. knowlesi, as reported in the study by Raja et al. [20], the parasite density ranged from 213 to 84,299 parasites per milliliter. Moreover, the most recent study with enrolled febrile Fig. 6 The estimated prevalence of P. cynomolgi malaria in macaques. ES: estimated prevalence, overall: overall prevalence, I 2 : level of heterogenicity, p: p-value less than 0.05 is statistically significant patients at malaria clinics or local hospitals in Thailand demonstrated that most of the patients with P. cynomolgi mixed infection with P. vivax, P. falciparum or both P. vivax and P. knowlesi had parasitemia less than 10,000 parasites/μL (< 0.2% parasitemia) as demonstrated by Putaporntip et al. [31]. The low proportion of P. cynomolgi detected in humans by three studies [18][19][20] might be due to the use of less-sensitive molecular protocols, as they used nested PCR amplifying the SSU rRNA gene for the detection of malaria parasites. PCR amplification of the malarial cytochrome oxidase gene is superior to PCR amplification of SSU rRNA, as cytochrome oxidase is a mitochondrial genome that is found at approximately 20-150 copies per parasite, whereas SSU rRNA is found at approximately 4-8 copies per parasite [32,33].
The natural infection of P. cynomolgi in humans is similar to the infection of the natural host of long-tailed or pig-tailed macaques since mixed infections of more than one species in a single host were observed in both humans and macaques. Nevertheless, the dominance of P. knowlesi over P. cynomolgi, as demonstrated by Raja et al. [20], might be due to the differences in the developmental cycles of the two species. P. cynomolgi has both an exoerythrocytic cycle in the liver and erythrocytic cycle in red blood cells. This means that the incubation period for P. knowlesi is shorter by approximately 9-12 days than that of P. cynomolgi, which is between 15 and 37 days [7]. Furthermore, the erythrocytic cycle of P. cynomolgi (48 h) is longer than that of P. knowlesi (24 h) [34]. Therefore, P. cynomolgi mixed infection with P. knowlesi might occur simultaneously in a single mosquito bite harboring two parasite species. The simultaneous infection of P. cynomolgi with P. knowlesi in a single mosquito was demonstrated in a study by Chua et al. [22].
The present study had limitations. First, the number of studies identifying P. cynomolgi in humans and mosquitoes was limited, resulting in an imprecise estimate of the pooled prevalence or pooled proportion and of the clinical characteristics of human hosts. Second, the Northern Sabah study [18] demonstrated a lower NOS Fig. 7 The estimated proportion of P. cynomolgi malaria in macaques. ES: estimated proportion, overall: overall proportion, I 2 : level of heterogenicity, p: p-value less than 0.05 is statistically significant scale (2 stars) as poor comparability against the other studies which used molecular detection methods but retained in the present analysis. Therefore, the pooled proportion of P. cynomolgi infection in humans requires careful interpretation. Further studies and continued surveillance of P. cynomolgi throughout Southeast Asia and neighbouring regions may be necessary through the use of sensitive molecular protocols to obtain accurate and relevant data for the detection and reporting of this emerging zoonotic malaria. Moreover, it is essential to further explore the biology of P. cynomolgi, the possibility of relapses/recurrences, and asymptomatic infections, which have a direct impact on the epidemiology of malaria.

Conclusion
In conclusion, P. cynomolgi could continue to be a public health concern in Southeast Asian countries, as the natural habitat of the natural hosts and vectors and the evolution of the P. cynomolgi parasite itself could drive the transmission of this neglected but emerging disease. In addition, the environmental and climatic changes experienced worldwide could affect the transmission dynamics of P. cynomolgi as an emerging cause of malaria in humans. Further studies using molecular and multidisciplinary approaches to search for P. cynomolgi infection in humans, vectors, and natural hosts are necessary if human infections with P. cynomolgi do become public health concerns.