Plasmodium malariae Prevalence and csp Gene Diversity, Kenya, 2014 and 2015

Abstract

In Africa, control programs that target primarily Plasmodium falciparum are inadequate for eliminating malaria. To learn more about prevalence and genetic variability of P. malariae in Africa, we examined blood samples from 663 asymptomatic and 245 symptomatic persons from western Kenya during June–August of 2014 and 2015. P. malariae accounted for 5.3% (35/663) of asymptomatic infections and 3.3% (8/245) of clinical cases. Among asymptomatic persons, 71% (32/45) of P. malariae infections detected by PCR were undetected by microscopy. The low sensitivity of microscopy probably results from the significantly lower parasitemia of P. malariae. Analyses of P. malariae circumsporozoite protein gene sequences revealed high genetic diversity among P. malariae in Africa, but no clear differentiation among geographic populations was observed. Our findings suggest that P. malariae should be included in the malaria elimination strategy in Africa and highlight the need for sensitive and field-applicable methods to identify P. malariae in malaria-endemic areas.

Over the past decade, malaria control strategies in Africa have reduced the number of malaria cases and deaths. Nevertheless, non–Plasmodium falciparum malaria still presents a major challenge for malaria elimination (1,2). Global malaria elimination programs focus primarily on P. falciparum. Recent research efforts and control programs have drawn resources to P. vivax malaria. By contrast, P. malariae and P. ovale receive little attention, and malaria caused by these organisms is among the most neglected tropical diseases (3). In those rural areas of Africa where malaria is most common, affordable diagnostic tools are rapid diagnostic tests and microscopy, but they are not effective for detecting these 2 species, mainly because parasitemia with these species is low (4–6). As a result, P. malariae and P. ovale infections are often underestimated, and epidemiologic information, such as distribution and prevalence of these species in malaria-endemic areas, is lacking. This knowledge is essential for implementation of specific strategies for monitoring and eliminating all types of malaria where it is endemic to Africa.

Although P. malariae infection is often asymptomatic and rarely leads to severe clinical illness or death, this species causes a low-grade chronic infection that persists for decades and is associated with nephropathy and anemia (7–9). The persistence, as well as submicroscopic features of P. malariae, have contributed to intermittent outbreaks of malaria in the Colombian Amazon region (10). In addition, P. malariae can cause irreversible stage 5 kidney failure (11). The prevalence of this species may increase the risk for kidney injuries and impair renal function, particularly in children with no immunity against P. malariae. Ample evidence shows peak prevalence for severe and uncomplicated clinical P. falciparum malaria among infants and children in sub-Saharan Africa (12–14). Contrary to this age pattern, patients with P. malariae infections in Papua, Indonesia, were older (median 22 years of age) than those with non–P. malariae infections (e.g., P. vivax; median 10 years of age) (9). Knowledge of the age patterns of patients with P. malariae infection is critical for understanding its epidemiology and developing effective preventative strategies.

Compared with the distribution of P. falciparum and P. vivax, the distribution of P. malariae is relatively sparse and variable. P. malariae is endemic to West Africa (3), South America (15), Asia (16,17), and the western Pacific region (18,19). Knowledge of genetic variation among isolates from these geographic areas is still lacking. One study indicated a remarkably low level of sequence diversity at the msp1 locus in P. malariae from Brazil (20). Similarly, the lack of variation at the dhfr and dhps loci has been shown for P. malariae from Asia and the western Pacific region (21,22). These findings suggested that antimalarial drugs might be imposing selective pressure on the genetic diversity of P. malariae. The circumsporozoite protein (csp) gene, which is known to be critical for plasmodia sporozoite motility and hepatocyte invasion (23), has been shown to be variable in length and is a sequence of the tandemly repeated peptide units in P. falciparum (24,25), P. vivax (26,27), and P. malariae isolates from Central Africa (28). The vast antigenic variation observed in P. falciparum as a result of immune selection pressure can influence the capacity of mosquito transmission and the effectiveness of malaria vaccine (29). In this study, we sought to determine the prevalence of infection and age distribution of persons with asymptomatic and symptomatic P. malariae infection in western Kenya, the genetic affinity between P. malariae isolates from East Africa and other regions, and the level of csp gene diversity among P. malariae and the significance of this diversity.

Scientific and ethical clearance was given by the institutional scientific and ethical review boards of the Kenya Medical Research Institute and the University of California Irvine. Written informed consent/assent for study participation was obtained from all consenting heads of households, parents/guardians (for minors <18 years of age), and each person who was willing to participate in the study.

Materials and Methods

Study Areas and Participants

During June–August of 2014 and 2015, blood samples were collected from persons in 4 villages at the Lake Victoria basin (elevation ≈1,000 m) of western Kenya (Figure 1). These villages represent parts of the Lake Victoria area previously shown by nested and quantitative PCR (qPCR) methods to have high, stable rates of malaria transmission and prevalence (10%–40%) among children 5–14 years of age (30,31).

Figure 1.

Location of sites in western Kenya for study of Plasmodium malariae prevalence and circumsporozoite protein gene diversity, Kenya, 2014 and 2015. Prevalence (logarithmic vertical scales) of P. falciparum monoinfections (PF), P. malariae monoinfections (PM), and P. falciparum and P. malariae co-infections (PF+PM) are shown for each study site.

Community samples were collected from nonfebrile schoolchildren in 7 public primary schools (70–100 children/school, 2 schools/village except Kombewa). An equal number of boys and girls 6–15 years of age were randomly selected from each school. To determine P. malariae prevalence in the adult population, we randomly selected 63 persons (32 male and 31 female) >15 years of age from 18 households in Kombewa. We examined a total of 663 samples from the communities, which provided an estimation of 4% margin of error in parasite prevalence with 0.05 type I error. At the time of sampling, none of these persons exhibited fever or malaria-related symptoms.

Clinical samples were collected from 113 male and 132 female patients, <1 to 76 years of age, in 3 district hospitals. This sample size provided an estimation of 6% margin of error in parasite prevalence with 0.05 type I error. These patients had fever or malaria-related signs or symptoms and were determined to be positive for Plasmodium spp. by microscopy at the time of sampling. Thick and thin blood smears were prepared for microscopic examination to determine the Plasmodium species, and ≈50 μL blood was blotted onto Whatman 3MM filter (Sigma Aldrich, St. Louis, MO, USA) papers. Filter papers were air dried and stored in zip-sealed plastic bags with silica gel absorbent at room temperature until DNA extraction.

Microscopy and PCR of Plasmodium spp.

We examined slides under microscopes at 100× magnification and counted the number of parasites per 200 leukocytes. A slide was considered negative when no parasites were observed after counting >100 microscopic fields. At the time of sample collection, all slides were read by 2 microscopists. If counts were discordant, the slides were examined by a third microscopist. The density of parasitemia was expressed as the number of asexual parasites per microliter of blood, assuming a leukocyte count of 8,000 cells/μL, according to World Health Organization guidelines.

We extracted parasite DNA from half of a dried blood spot by using the Saponin/Chelex method (32). The final extracted volume was 200 μL. For all samples, nested amplification of the 18S rRNA gene region of plasmodia (P. falciparum, P. vivax, P. malariae, and P. ovale) was used for parasite detection and species identification. As positive controls for all amplifications, we used DNA from P. falciparum isolates 7G8 (MR4-MRA-926) and HB3 (MR4-MRA-155), P. vivax Pakchong (MR4-MRA-342G) and Nicaragua (MR4-MRA-340G), P. malariae (MR4-MRA-179), and P. ovale (MR4-MRA-180). As negative controls, we used water and noninfected samples to ensure lack of contamination. Reaction was performed in a Bio-Rad MyCycler thermal cycler according to the published protocol (33) (details in Technical Appendix 1).

In addition, the amount of parasite DNA was estimated by using the SYBR Green (Thermo Scientific, Foster City, CA, USA) qPCR detection method with Plasmodium species–specific primers that targeted the 18S rRNA genes (34,35). Reactions were performed in a CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Foster City, CA, USA). To confirm specific amplifications of the target sequence, we performed melting curve analyses for each amplified sample. To measure reproducibility of the cycle threshold (C_t), we calculated the mean value and standard deviations from triplicates in 2 independent assays. The parasite gene copy number in a sample was quantified according to C_t by using the equation (30) GCN_sample = e^{(E × ΔCtsample)}, where GCN stands for gene copy number; ΔCt, the difference in C_t between the negative control and the sample; e, exponential function; and E, amplification efficiency (Technical Appendix 1).

CSP Sequencing and Phylogenetic Analyses

Four internal primers were designed specifically on the P. malariae csp gene region and used together with the published primers (28; Technical Appendix 1 Table) to unambiguously amplify the 3 segments, the N terminal, the central repeat, and the C-terminal regions of the csp gene. A total of 37 P. malariae isolates were amplified and sequenced. All resulted sequences were verified by comparing them with those in the GenBank database by using BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Sequences were translated into protein sequences and analyzed together with all csp protein sequences available in GenBank of P. malariae from East Africa (Kenya and Uganda), West Africa (Cameroon), Central Africa (Côte d’Ivoire), and South America (Venezuela) and of P. brasilianum from South America (Brazil and Venezuela). It is noteworthy that although P. malariae and P. brasilianum coexist in Brazil, no csp sequence for P. malariae is available. Because of the potential for alignment errors associated with gaps in the nucleotide sequences, we used translated amino acid sequences with unambiguous indels in phylogenetic analyses. Sequence diversity, including measures of evolutionary distances and average pairwise divergence, were estimated and compared among geographic regions (Technical Appendix 1).

Statistical Analyses

A 1-tailed t-test was used to test for the significance of differences in parasite gene copy number between P. malariae from symptomatic and asymptomatic patients and between P. malariae and P. falciparum in co-infected samples. In addition, we calculated the Pearson correlation coefficient (r²) for parasite gene copy number and age by using R (https://www.r-project.org/).

Results

P. malariae Prevalence and Patient Age Distribution

Among the 663 samples from asymptomatic persons, P. malariae was detected by PCR in 35 (5.3% prevalence). Among these, 29 were mixed infections (with P. falciparum) and 6 were P. malariae monoinfections (Figure 1; Table 1). P. malariae was found to be most prevalent in Kombewa (14.3%, 19/133 cases), followed by Kendu Bay (5.3%, 8/150 cases). Prevalence of P. falciparum prevalence was relatively high at these 2 sites (44% and 59%, respectively; Table 1). In Kombewa, 13 of 19 P. malariae cases were detected in younger persons (<15 years of age), which was significantly higher than the number of cases detected in older persons (6 cases, p = 0.04; Table 1). Although such a comparison between age groups cannot be made for the other sites, a similar pattern was observed for symptomatic patients.

Table 1. Prevalence of Plasmodium malariae and P. falciparum among asymptomatic persons in the community, Kenya, June–August 2014 and 2015* .

Site, patient age, y	No. tested	No. (%) infections
		Total	P. malariae	P. falciparum	Mixed†
Kombewa
<15	63	41 (65.1)	0	28 (44.4)	13 (20.6)
>15	70	35 (50)	2 (2.9)	29 (41.4)	4 (5.7)
Chulaimbo
<15	190	76 (40)	2 (1.1)	71 (37.4)	3 (1.6)
Kendu Bay
<15	150	97 (64.7)	0	89 (59.3)	8 (6)
Port Victoria
<15	190	57 (30)	2 (1.1)	54 (28.4)	1 (0.5)
Total	663	306 (46.2)	6 (0.9)	271 (40.9)	29 (4.4)

*According to nested PCR of the 18S rRNA gene.
†P. malariae and P. falciparum.

Among the 245 samples from symptomatic patients, 8 (3.3%) P. malariae cases were detected; 6 were mixed infections with P. falciparum and 2 were P. malariae monoinfections (Table 2). When the samples were stratified by patient age, all P. malariae infections in symptomatic persons were in infants or very young children of <5 years of age (8/135, 5.9% infection rate). Although P. falciparum infection was highest among patients >5 to <15 years of age, no P. malariae was detected in persons in this and older age groups despite smaller samples in these groups. No significant difference was detected between male and female patients.

Table 2. Prevalence of Plasmodium malariae and P. falciparum among symptomatic persons, Kenya, June–August 2014 and 2015* .

Site, patient age, y	No. tested	No. (%) infections
		Total	P. malariae	P. falciparum	Mixed†
Chulaimbo
<5	27	18 (66.7)	2 (7.4)	15 (55.6)	0
>5 to <15	4	3 (75)	0	3 (75)	0
>15	13	3 (23.1)	0	3 (23.1)	0
Kendu Bay
<5	44	38 (86.4)	0	35 (79.5)	3 (6.8)
>5 to <15	34	31 (91.2)	0	31 (91.2)	0
>15	24	23 (95.8)	0	23 (95.8)	0
Port Victoria
≤5	64	54 (84.4)	0	51 (79.7)	3 (4.7)
>5 to <15	22	20 (90.9)	0	20 (90.9)	0
>15	13	9 (69.2)	0	9 (69.2)	0
Total
<5	135	110 (81.5)	2 (1.5)	101 (74.8)	6 (4.4)
>5 to <15	60	54 (90)	0	54 (90)	0
>15	50	35 (70)	0	35 (70)	0

*According to nested PCR of the 18S rRNA gene.
†P. malariae and P. falciparum.

Comparisons of Diagnostic Approaches and Parasitemia

Compared with microscopy, nested PCR revealed a significantly higher number of P. malariae infections in the community (Table 3). All samples that were P. malariae positive by microscopy were identified as positive by PCR and qPCR. Across the study sites, nested PCR–based prevalence ranged from 0 to 12.2% (average 4.8%), >2-fold higher than by microscopy (0 to 3.8%, average 1.9%; Table 3). The discrepancy between the 2 methods was also reflected by the difference in P. falciparum prevalence; 10% more positive infections were detected by nested PCR than by microscopy. Nevertheless, such a discrepancy was not as substantial as that for P. malariae.

Table 3. Methods used to diagnose Plasmodium infections in asymptomatic populations, Kenya, June–August 2014 and 2015*.

Site, method	No. tested	No. (%) infections
		Total	P. falciparum	P. vivax	P. malariae	P. ovale	P. falciparum/malariae
Kombewa
Microscopy	133	54 (41.2)	49 (37.4)	0	0	0	5 (3.8)
PCR	133	70 (53.4)	54 (41.2)	0	0	0	16 (12.2)
Chulaimbo
Microscopy	190	46 (24.2)	42 (22.1)	0	1 (0.5)	0	3 (1.6)
PCR	190	76 (40.1)	71 (37.4)	0	2 (1.1)	0	3 (1.6)
Kendu Bay
Microscopy	150	78 (52)	75 (50)	0	0	0	3 (2)
PCR	150	97 (64.6)	89 (59.3)	0	0	0	8 (5.3)
Port Victoria
Microscopy	190	36 (18.5)	35 (18.5)	0	0	0	1 (0.5)
PCR	190	57 (30)	54 (28.4)	0	2 (1.1)	0	1 (0.5)
All sites
Microscopy	663	214 (32.4)	201 (30.4)	0	1 (0.2)	0	12 (1.8)
PCR	663	300 (45.4)	268 (40.5)	0	4 (0.6)	0	28 (4.2)

Although the number of P. malariae–positive clinical samples detected in this study was low, these samples indicated an overlapping range of parasite gene copy number (geometric mean 6.4 × 10¹/μL, range 4.3 × 10¹ to 1.2 × 10³/μL; Figure 2) with that of the samples from asymptomatic persons (geometric mean 4.8×10¹/μL, range 0.5 ×10¹ to 9.4 ×10²/μL) without differing significantly (p>0.05). Similar results were observed in the level of P. malariae parasitemia, for which samples from symptomatic and asymptomatic persons did not differ significantly (Figure 2). Parasite gene copy number and P. malariae parasitemia were significantly positively correlated with each other (r² = 0.77, p<0.01; Technical Appendix 1 Figure 1).

Figure 2.

Parasite gene copy numbers (per microliter) detected by SYBR Green (Thermo Scientific, Foster City, CA, USA) quantitative PCR and parasitemia (parasites per microliter) determined by microscopy of Plasmodium malariae samples from asymptomatic and symptomatic persons. Median, first quartile, and fourth quartile of the data are shown for each sample category (horizontal lines). No significant difference was observed between asymptomatic and symptomatic persons in terms of P. malariae parasite gene copy number and parasitemia. Squares represent samples with gene copy number measured by quantitative PCR; circles, samples with parasitemia estimated by microscopy; closed squares and circles, P. malariae samples from asymptomatic persons; open squares and circles, P. malariae samples from symptomatic patients. NS, not significant.

Parasite gene copy number and parasitemia for P. falciparum were generally higher than those for P. malariae (Figure 3, panel A). Among the 35 mixed infections, 28 (80%) gene copy numbers were higher for P. falciparum than for P. malariae (Technical Appendix 1 Figure 2). Among these samples overall, the amount of P. falciparum DNA (geometric mean 1.6 × 10²/μL, range 1 × 10¹ to 5.5× 10³/μL) was significantly higher than the amount of P. malariae DNA (geometric mean 4.7 × 10¹/μL, range 0.4 × 10¹ to 1.1 × 10³/μL; p = 0.003), consistent with the difference in parasitemia according to microscopy (P. malariae geometric mean 3.2 × 10² parasites/μL vs. P. falciparum geometric mean 1.1 × 10³ parasites/μL; Technical Appendix 1 Figure 2).

Figure 3.

Plasmodium malariae and P. falciparum parasite gene copy numbers (per microliter) and parasitemia (parasites per microliter) in co-infected samples. Median, first quartile, and fourth quartile of the data are shown for each sample category (horizontal lines). Parasite gene copy number and parasitemia were lower in P. malariae–positive than in P. falciparum–positive samples. Squares represent samples with gene copy number measured by quantitative PCR; circles, samples with parasitemia estimated by microscopy; red, P. malariae samples; blue, P. falciparum samples.

When all P. malariae samples were pooled, the parasite gene copy number did not correlate significantly with patient age (r² = 0.07; Technical Appendix 1 Figure 3). Neither P. malariae prevalence rate nor parasite gene copy number differed significantly according to patient sex.

Genetic Relatedness and csp Divergence of P. malariae

The csp alignment comprised 530 aa, of which 34 (6.4%) were polymorphic among the studied parasites of different taxa (Technical Appendix 2). To avoid polymorphism caused by PCR error, we sequenced each isolate at least twice in both directions. Substantial length variation was observed in the central repeat region, where the number of NAAG_n (the repeat codon unit in which n denotes the number of repeats) in P. malariae ranged from 49 to 85 units. These tandem repeats could be rapidly evolving through a different mechanism and may influence genetic relationships among the samples. To examine such effect, we constructed phylogenetic trees with 2 sets of data: the entire sequence (530 aa) and partial sequences without the central repeat region (225 aa).

Maximum-likelihood analyses of the entire csp gene showed a clear distinction between isolates from South America and those from the other geographic regions (Figure 4, panel A). P. brasilianum and P. malariae from Venezuela formed a monophyletic group (bootstrap >95%) closely associated with P. brasilianum from Brazil. Sequences of P. malariae from Venezuela were almost identical to those of P. brasilianum from the same area. Closely related to the clade from South America was a large monophyletic group that contained P. malariae from East, Central, and West Africa and from China (bootstrap >90%). The isolates from these regions were divided into 2 subclades: I and II (Figure 4, panel A). Subclade I comprised a mix of P. malariae isolates from Kenya, Cameroon, and Côte d’Ivoire. Subclade II comprised a mix of P. malariae isolates from Kenya, Cameroon, Uganda, and China. Sequences without the central repeat region indicated consistently the distinctiveness between P. brasilianum from Brazil and P. malariae, but the P. malariae samples from different geographic regions were poorly resolved (Figure 4, panel B). The P. malariae isolate from China was nested within the African subclade, suggestive of an African origin (Figure 4, panel C). No clear microgeographic structure was detected, although sample size at the population level was small.

Figure 4.

Maximum-likelihood analyses of circumsporozoite protein gene (csp) sequences of Plasmodium malariae and distribution of the samples. A) Phylogenetic tree based on maximum-likelihood analyses of the entire csp amino acid sequences of P. malariae isolates from different geographic regions, shown by different colors. Asterisks denote clade with >90% bootstrap support. Sequences of P. malariae from Venezuela (red squares) were almost identical to those of P. brasilianum (red circles) from the same area. These samples were genetically closely related with P. brasilianum from Brazil (violet circles) but distant from P. malariae from East and West Africa. The samples from Africa were subdivided into 2 subclades, I and II. Subclade I comprised a mix of P. malariae isolates from Kenya (dark blue squares), Cameroon (yellow squares), and Côte d’ Ivoire (orange squares). Subclade II comprised a mix of P. malariae isolates from Kenya, Cameroon, Uganda (light blue squares), and China (green squares). B) Maximum-likelihood analyses of partial csp amino acid sequences without the central repeat region. P. brasilianum from Brazil was distant from P. malariae, but relationships among the P. malariae samples from different geographic areas were not well resolved. C) Locations of samples included in the analyses. Arrow indicates the possible African origin of P. malariae from China. Scale bars indicate length of phylogenetic tree. *Bootstrap value >90%.

Among the 3 geographic regions, the level of csp sequence divergence in P. malariae was higher in isolates from East Africa than from West Africa, as reflected by a higher number of polymorphic sites and a greater extent of csp length variation despite difference in sample size (Figure 5, panels A and B). These variations were located at the 3′ N terminal through the central repeat region, where the largest degree of mismatch was observed (Figure 5, panel B). To the contrary, the level of sequence polymorphism was lowest in isolates from South America (Figure 5, panel A), but the greatest range of difference in tandem repeat units where remarkable mismatch was observed was toward the end of the central region. Despite the small sample size, the number of tandem repeats was generally lower in P. brasilianum than P. malariae (Figure 5, panel C).

Figure 5.

Comparison of circumsporozoite protein (csp) gene sequence divergence among Plasmodium isolates from different geographic regions. A) Pairwise genetic distance plot of all amino acid positions of the csp gene. The matrix-normalized distances based on the standard point accepted mutation (Dayhoff–PAM) model that account for the probability of change from 1 amino acid to another were calculated. Samples were analyzed as a whole and partitioned by geographic regions as indicated by colors. B) Dot plot showing matching scores, a proxy of sequence similarity, between pairwise samples calculated based on the standard Dayhoff–PAM matrix. The greatest mismatch was detected at amino acid positions 110–310, representing the 3′ N terminal through the central repeat regions. C) Variation in the number of tandem repeats in the central region of the csp gene. The greatest length variation was observed in the isolates from South America despite the fact that both P. malariae (PM) and P. brasilianum (PB) were included. P. malariae from East Africa was more variable in the number of repeats than isolates from Central/West Africa, despite difference in sample size. Median, first quartile, and fourth quartile of the data are shown for each sample category (horizontal lines). Red represents samples from South America; yellow, Central/West Africa; blue, East Africa. Circles represent P. brasilianum; squares, P. malariae.

Discussion

In Kenya, areas along the shoreline of Lake Victoria and coastal regions are malaria hot spots, where intense and stable plasmodia transmission occurs throughout the year (31). For achieving the ultimate goal of eliminating malaria in Kenya, existing control programs that primarily target P. falciparum are inadequate. The use of rapid diagnostic tests or microscopy as first-line diagnostic methods can lead to gross underestimation of the actual prevalence of P. malariae (4–6). Our findings indicated that P. malariae accounted for ≈3% of clinical cases and ≈5% of asymptomatic infections in this malaria-endemic region. The prevalence of asymptomatic P. malariae infections was comparable to that recently reported for nearby islands of Lake Victoria (1.7%–3.96%) on the basis of PCR (36,37). These asymptomatic P. malariae infections are concerning because they are parasite reservoirs that can sustain long-term transmission. For instance, in the Colombian Amazon region, P. malariae was thought to account for <1% of all malaria infections (38,39); however, a recent study revealed that 43.6% (294/675) of clinical cases were caused by P. malariae (10) and suggested that these parasites have been circulating in the community undetected. Underestimation or lack of awareness of its occurrence could thus lead to increased transmission. The infectiousness of P. malariae for Anopheles mosquitoes in malaria-endemic areas remains unclear and merits further investigation.

We found that P. malariae infections were more common among infants and children than adults. A similar pattern has been found for Senegal, West Africa, where 91% (265/290 cases) of clinical P. malariae cases occurred in children <15 years of age and the mean incidence density was highest for those 5–9 years of age (3). These findings indicate that children are vulnerable to P. malariae infection and contrast with those reported for Papua, Indonesia, where P. malariae infection was higher among older (median 21 years of age) than younger persons (9). It is possible that our study sites in western Kenya, as well as in West Africa, are high-transmission areas where P. falciparum malaria prevalence can be ≈60% during the rainy season (30,40). Cumulative exposure to the parasites over time may enable gradual acquisition of immunity in adults. Nevertheless, our community samples were mostly obtained from schoolchildren 6–15 years of age. Underrepresentation of adult populations may underestimate the overall malaria prevalence in the study area. Although young children are more vulnerable to P. malariae infections, the level of P. malariae parasitemia does not seem to be associated with age. Chronic nephrotic syndromes attributed to P. malariae have been reported (41,42) and shown to be associated with significant illness from anemia in young children (8,9). However, the lack of hematologic data from our study participants limits further investigation.

Our data indicate that ≈50% of P. malariae–positive samples detected by PCR were undetected by microscopy. Such a low sensitivity of microscopy could be attributed to a significantly lower P. malariae than P. falciparum parasitemia, according to qPCRs. Because most P. malariae–positive samples had mixed infections, microscopists could have recorded only the dominant P. falciparum and overlooked the sparse P. malariae trophozoites. Also, the ring forms of P. falciparum and P. malariae are morphologically more similar to each other than to P. vivax and P. ovale (43). Misdiagnosis of parasite species by microscopy is possible (8).

In Africa, the standard treatment for P. malariae monoinfection is chloroquine, and for P. falciparum and mixed plasmodial infections it is artemisinin combination therapy (31). The combination treatment regime should cure P. malariae infections even in cases of misdiagnosis. However, P. malariae increases production of P. falciparum gametocytes in mixed infections, and these gametocytes can persist without proper antimalarial treatment or monitoring (44). Therefore, we highlight the need for sensitive methods to improve P. malariae diagnosis and provide accurate epidemiologic data for specific and effective management guidelines. Although PCR is a better diagnostic method, it uses a relatively small amount of blood from filter papers and could still underestimate P. malariae infections in samples with exceptionally low levels of parasitemia. More accurate prevalence data may be obtained from ultrasensitive PCR that targets multicopy regions of the parasite genome (45) or reverse transcription PCR of parasite RNA extracted from whole blood (46).

Sequences of the csp gene were shown to be highly polymorphic among P. malariae isolates from western Kenya. The most polymorphic region was in the central repeat region, where mutations and length differences were detected (24,28). Among the isolates from different geographic areas, P. malariae from East and Central/West Africa were genetically closely related and exhibited a comparable level of sequence variation. This variation could be attributed to positive selection, frequent recombination, and gene flow among the parasites, as follows. First, compared with msp1, dhfr, and dhps of P. malariae (20–22), the csp gene revealed a remarkably higher level of sequence diversity. It is possible that selection of csp genetic variants may confer immunogenic advantages to the pathogen during host invasion (28,47). Second, intense transmission and large vector populations in our study area might enhance frequent heterologous recombination of the parasite genome during reproduction in the mosquitoes and increase genetic diversity within populations (24,25). Third, recurrent gene flow between the parasite populations across countries, via human migration or dispersal of vector mosquitoes, promotes the spread of these genetic variants, leading to a lack of differentiation according to geographic region. Future study using other variable markers, such as microsatellites, on expanded population samples could validate our findings.

In summary, underestimation of the actual prevalence of asymptomatic infections hinders progress toward malaria elimination in Africa. The low parasitemia of P. malariae infections influences diagnostic sensitivity by microscopy. A more sensitive tool is needed to identify asymptomatic P. malariae and to improve control strategies, particularly among infants and children who are vulnerable to P. malariae infection.

Technical Appendix 1

Detailed description of PCR-based diagnostic assays and phylogenetic analyses of Plasmodium circumsporozoite protein gene sequences.

Technical Appendix 2

Additional information for Plasmodium malariae and P. brasilianum isolates.

Biography

Dr. Lo is a researcher focused on molecular epidemiology and evolution of pathogens. She is interested in exploring the effects of host–parasite interactions on parasite genomic and genetic structure.