Plasmodium malariae and Plasmodium ovale – the ‘bashful’ malaria parasites

Abstract

Although Plasmodium malariae was first described as an infectious disease of humans by Golgi in 1886 and Plasmodium ovale identified by Stevens in 1922, there are still large gaps in our knowledge of the importance of these infections as causes of malaria in different parts of the world. They have traditionally been thought of as mild illnesses that are caused by rare and, in case of P. ovale, short-lived parasites. However, recent advances in sensitive PCR diagnosis are causing a re-evaluation of this assumption. Low-level infection seems to be common across malaria-endemic areas, often as complex mixed infections. The potential interactions of P. malariae and P. ovale with Plasmodium falciparum and Plasmodium vivax might explain some basic questions of malaria epidemiology, and understanding these interactions could have an important influence on the deployment of interventions such as malaria vaccines.

Geographical distribution

Although distribution of Plasmodium malariae infection is reported as being patchy, it has been observed in all major malaria-endemic regions of the world [1]. P. malariae infections are most common in sub-Saharan Africa and the southwest Pacific, where age-specific prevalence in mass blood surveys have exceeded 15–30% [2–8]. By contrast, when P. malariae has been detected in malaria-endemic regions of Asia [9–12], the Middle East [13], South America [14] and Central America [15], it is observed as an infrequent infection, with blood-smear light microscopy (LM) prevalence rarely exceeding 1–2%. Much higher levels of infection were, however, found in montagnard refugees from the Cambodian–Vietnamese border [16]. In South America, P. malariae is thought to be a zoonotic infection because the genetically identical Plasmodium brasilianum infects new-world monkeys [17] and both monkeys and humans in endemic areas show high levels of seropositivity to P. malariae and P. brasilianum antigens [18].

Plasmodium ovale was thought to have a much more limited distribution, with endemic transmission traditionally described as being limited to areas of tropical Africa, New Guinea, the eastern parts of Indonesia and the Philippines [19,20]. Infections with P. ovale, however, have also been reported in the Middle East [13], the Indian subcontinent [21] and different parts of Southeast Asia [11,22,23]. In West Africa (and to a lesser extent Central Africa), age specific LM prevalence of >10% have been observed [3,6]. However, in most places where P. ovale is observed, it is relatively uncommon and its prevalence (as detected by LM) rarely exceeds 3–5% [22,24–26].

Indepth descriptions of the epidemiology of both infections based upon LM data are, thus, restricted almost exclusively to highly endemic areas in Africa and the Southwest Pacific. Detailed epidemiological studies from South America and Asia are lacking.

Variation in parasite prevalence

In West Africa, P. malariae prevalence has been reported to peak at ages similar to those of P. falciparum (i.e. in children under ten years of age) [3,6,8]. In children under four years, in addition to adults, most LM-diagnosed infections were observed to be of low density (<500 parasites per µl) and, at least in children, usually of long duration [6]. Coinfections of P. malariae and P. falciparum were reported more commonly than would be expected by chance [3,27,28]. Overall, P. malariae has been characterized to exhibit opposing seasonal fluctuation with P. falciparum, with P. malariae prevalence [8,26] and/or parasite densities [5] increasing in the dry season. A high incidence of new P. malariae infections in young, nonimmune children reported during the rainy seasons [8] led Molineaux and Grammiccia to suggest that seasonal fluctuation in the Plasmodium species infection formula resulted from suppression of P. malariae by frequent P. falciparum infection during the rainy season rather than variation among factors that could lead to differences in parasite transmission [8].

Studies from Papua New Guinea (PNG) indicate that LM peak prevalence of P. malariae infection might be similar to those observed in West Africa (i.e. 15–30%); in PNG, however, P. malariae infection is observed predominantly in older children (seven to nine years) and adolescents (ten to 16 years) [4,25,29]. In addition, P. malariae infection prevalence peaked in noticeably later age groups than did P. falciparum infections [30]. Outside Africa there has been no report of opposing seasonal fluctuation in the prevalence of P. malariae and P. falciparum infections. Moreover, mixed infections are either randomly distributed [31,32] or there might be a short age of LM-detected mixed infections [28]. Apart from differences in a wide range of ecological factors (such as climate and Anopheline vector species), these contrasting observations in P. malariae epidemiology could be linked to the virtual absence of Plasmodium vivax infections in many African populations because of the lack of the Duffy blood-group-antigen, which P. vivax requires to invade red blood cells [33]. Interestingly, Smith et al. reported that in Wosera, PNG, P. vivax infections might be associated with an antagonistic effect on P. malariae infections [34].

Like P. malariae, P. ovale infections in West African populations tend to be most common in children under ten years of age [3,6,35]. The low numbers of LM detectable infections in most PNG and Southeast Asian surveys precludes a firm assessment of age distribution of infections in these regions. Indepth studies of P. ovale in Senegal reported that, in all age groups, most P. ovale infections were patent (i.e. detectable by LM) for fewer than two weeks and maximum parasitemia rarely reached levels that were sufficient to introduce clinical attacks (i.e. <800 parasites per µl) [6,35]. However, the incidence of infections was high, with 40% of people of all ages experiencing at least one P. ovale infection during the four-month study period [6,35]. The rarity of reported P. ovale infections in cross-sectional surveys might be heavily influenced by the low species-specific parasitemia and short duration of patent infections.

Clinical malariae and ovale malaria

Although P. malariae and P. ovale infections are common and pyrogenic thresholds likely to be lower than in P. falciparum [36], the incidences of clinical P. malariae and P. ovale episodes are low [6,25] and might account for only 1–2% of fever episodes [36]. In areas with marked variation in seasonal climate, P. malariae might contribute to 50% of malaria episodes in the low-transmission season [37]. Because differentiation between P. malariae and P. falciparum is difficult by thick-film microscopy, the incidence of clinical P. malariae could, however, often be underestimated.

Despite a recent report of chloroquine (CQ) resistant P. malariae in Indonesia [38], P. malariae and P. ovale remain highly sensitive to CQ and other common antimalarials [39] and might, thus, be effectively killed even by residual drug levels [40]. In addition, given their slower development in the mosquito [41], P. malariae and to a lesser extent P. ovale might be susceptible to interventions that are aimed at reducing transmission. In a study performed in Burkina Faso, insecticide-treated bednets were associated with a greater reduction of P. malariae prevalence than of P. falciparum [42]. Similarly, in PNG, P. malariae prevalence was strongly reduced by both indoor residual dichloro-diphenyl-trichloroethane (DDT) spraying and mass drug administration and, unlike P. falciparum, P. malariae failed to recover to precontrol levels after DDT spraying was stopped [43,44].

Most of what is known with regard to the epidemiology of P. malariae and P. ovale has been shaped by malariometric surveys based upon LM diagnosis, and this has certainly influenced the studies described here so far. Unfortunately, differentiation of P. malariae from P. falciparum and P. ovale from P. vivax by LM can be challenging [45]. Low parasitemias, which are commonly observed for P. malariae and P. ovale, require evaluation of thick-films; however, this technique results in the loss of many distinguishing features of infected red blood cells. Therefore, in endemic regions where P. falciparum and/or P. vivax predominate, P. malariae and P. ovale are frequently overlooked. For more accurate diagnosis and estimates of the burden of P. malariae and P. ovale infections, more sensitive diagnostic methods are needed.

Improved diagnosis of P. malariae and P. ovale infections by PCR

Although acridine orange microscopy substantially improves detection levels over Giemsa thick-film LM [11,46], PCR-based diagnostic methods targeting the small subunit rRNA (SSU rRNA) gene [47] have transformed perspectives on malaria epidemiology [48]. These assays enable detection of the four Plasmodium parasite species that infect humans at densities ~100 times lower than the limit of LM detection [47] and have evolved from simple species-specific amplification strategies to multiplex approaches that incorporate semiquantitative assessments [31,49–56]. We have greatly extended the potential use of PCR diagnosis for a wide range of epidemiological studies through our recent efforts to develop a ligase detection reaction fluorescent microsphere assay (LDR–FMA) [52,57] to evaluate all four malaria parasites of humans in a single-well multiplex format.

In a series of studies from PNG that have combined LM and molecular diagnostics [29,31,32], we have found that molecular methods consistently detected significantly increased prevalence of all four malaria species of humans, with the largest increases in P. malariae and P. ovale. In these studies from three different PNG populations, the prevalence of P. falciparum and P. vivax increased between 1.6 and 3.0-fold and 2.1 and 3.4-fold, respectively, and a 2.6 to 10.9-fold increase in P. malariae prevalence was observed (LM: 1.1–14.4%, PCR: 10.0–37.0%) (Table 1). Even larger increases in prevalence were observed for P. ovale. Detection of P. ovale by LM ranged from 0 to 0.3%, whereas LDR–FMA detected P. ovale infection in 5.2 to 15.6% of the study subjects. Overall, the majority of P. malariae and P. ovale diagnoses have been observed in the context of mixed infections with P. falciparum and/or P. vivax, including several quadruple infections [29,31,32]. Therefore, it is likely that P. malariae, P. ovale and mixed species infections are substantially more common in PNG than previously thought. The direct comparison between LM and LDR–FMA diagnosis of P. malariae infection for 1182 individual samples showed that, although we observed an overall concordance between diagnostic methods of 89.1%, there was a wide variation between LM-based parasitemia and the semiquantitative LDR–FMA fluorescent signal [29]. These findings indicate that there are difficulties with LM in detecting parasites across wide-ranging levels of blood-stage infection. Similar observations were made for P. falciparum and P. vivax diagnoses [29].

Table 1.

Studies comparing LM and PCR diagnosis of Plasmodium malariae and Plasmodium ovale infections

Region and country	Population	P. malariae LM (%)	PCR (%)	Mixeda (%)	P. ovale LM (%)	PCR (%)	Mixeda (%)	Refs
Pacific
PNG (Drikikir)	CSb, all ages	14.4	37.0	91.8	0.0	15.6	100.0	[31]
PNG (Liksul)	CS, all ages	1.1	10.0	75.8	0.0	5.2	88.2	[32]
PNG (Wosera)	CS, all ages	2.0	21.8	72.9	0.1	6.2	80.0	[32]
PNG (Wosera)	CS, all ages	4.0	12.4	69.3	0.3	5.5	77.3	[29]
Africa
Cameron	CS, pregnant women	1.1	7.6	100.0	0.0	2.5	100.0	[77]
Equatorial Guinea	CS, under six years old	18.7	39.2	92.9	2.8	9.3	60.0	[59]
Guinea–Bissau	CS, all ages	0.0	23.3	97.7	0.0	6.9	100.0	[58]
Americas
Brazil	CS, all ages	1.2	11.9	69.5	–c	–	–	[14]
Brazil	Malaria patients	0.0	9.4	88.9	–	–	–	[63]
Asia
Thai–Myanmar border	Patients	2.2	24.3	99.2	0.4	3.8	100.0	[46]

^aProportion of P. malariae or P. ovale infections diagnosed by PCR that are coinfections with P. falciparum and/or P. vivax.

^bCross-sectional population survey.

^c– indicates not assayed.

Comparisons between infection prevalence and age can now be optimized because of the increased ability of molecular diagnostic assays to evaluate samples from large numbers of study participants with greater specificity and sensitivity. Results presented in Figure 1 show a shift of the burden of infections into older age groups for PCR detectable infections compared with diagnosis by LM [29]. In all species except P. vivax, in which infections were most commonly found in children seven to nine years old (42.6%), prevalence of infections peaked in the adolescent age group (ten to 19 years), which had an overall infection rate of 69.9%, P. falciparum = 48.6%, P. malariae = 20.9%, P. ovale = 9.3% and mixed infections = 25.7%. Further studies in other populations will be needed to confirm this observation.

Figure 1.

Age-specific prevalence of Plasmodium infections. (a) Plasmodium falciparum (red) and Plasmodium vivax (blue) infections. (b) Plasmodium malariae (brown) and Plasmodium ovale (green) infections. (c) Mixed infections: concurrent infection with more than one Plasmodium species. Infections were detected by light microscopy (broken line) and LDR–FMA assay (unbroken line) in 1182 paired samples from Wosera, PNG [29].

Consistent with our observations in PNG, when patients have been studied using PCR-based methods in Africa, Southeast Asia and Brazil, similar large increases in P. malariae and P. ovale prevalence has been detected (Table 1). In two African studies, PCR prevalence of P. malariae was 23.3% and 39.2%, and P. ovale prevalence was 6.9% and 9.3%. By contrast, LM detected no non-P. falciparum infections in the first study [58] and 18.7% P. malariae and 2.8% P. ovale in the second [59]. Findings consistent with these observations have been generated recently in a study of >1200 Kenyans. Preliminary comparisons between LM and LDR–FMA diagnostic methods indicate that, although LM detected no non-P. falciparum infections, LDR–FMA estimated prevalence of P. malariae to be 13.5% and of P. ovale to be 3.1% (C.H. King and P. Zimmerman, unpublished). PCR diagnosis detected similarly high rates of infections in a rural village in Nigeria (P. malariae = 26.1%, P. ovale = 14.8% [60]) and in pregnant women in Kenya (P. malariae = 24.5%, P. ovale = 22.5% [61]). However, in a study in southern Mozambique that typed infections by PCR, substantial spatial and seasonal variations in the prevalence of P. malariae were observed [62]. In the Thai–Myanmar border area, up to 25% of malaria patients were observed to carry subpatent P. malariae infections, whereas 4% showed evidence of subpatent P. ovale infections [46]. These observations were repeated in many other Southeast Asian settings, with an average of 16.6% and 3.5% of malaria patients harbouring P. malariae and P. ovale parasites, respectively [11]. This confirmed that both parasite species are common across much of the region. In the Brazilian Amazon, P. malariae prevalence of 11.9% in the general population [14] and 9.4% in patient samples were detected by PCR [63]. As in PNG, the majority of infections were found in the presence of P. falciparum and/or P. vivax (Table 1).

Although there are an increasing number of studies that use PCR diagnosis of P. malariae and P. ovale infections in human blood samples, the diagnosis of sporozoite infections in mosquitoes still relies almost exclusively on ELISA-based dectection of circumsporozoite protein (CSP) [64]. Unfortunately, recent analysis of the available CSP ELISA showed the test to have insufficient sensitivity to detect infections with low numbers of P. malariae sporozoites [65]. Beier et al. indicate that low sensitivity of the ELISA method is unlikely to be restricted to P. malariae but also apply to other malaria species, including P. falciparum [66]. Therefore, PCR diagnosis of mosquito salivary glands will be invaluable for accurate quantification of the prevalence of P. malariae and P. ovale sporozoite infections. A recent study in Guinea–Bissau [67] that used PCR on salivary glands and/or midgut oocysts from blood-fed resting mosquitoes found 1–5% and 0.4–20% to be infected with P. malariae and P. ovale, respectively, whereas another study from Mozambique found 5% of all positive mosquitoes to be infected with P. malariae [62].

One note of caution regarding current PCR-based reports on P. malariae and P. ovale prevalence is that the sequence variation in the SSU rRNA gene target of both species must be considered [68–70]. From studies in Southeast Asia, 27% of P. malariae and 36% of P. ovale infections were due to single variant strains in each species that have not been included in all PCR-based diagnostic assays [11]. To date, variants of this nature have not been observed in PNG. In addition, it should be noted that although the sensitivity of PCR is superior to LM even PCR will fail to detect extremely low-level infections.

On the best available evidence, it is probable that P. malariae is a common infection to all malaria-endemic areas, whereas P. ovale might be present at 2% to >10% prevalence in the general population of Africa and New Guinea and among malaria patients in Southeast Asia. Both large increases in prevalence and high complexity of infections indicate that PCR-based diagnosis has to be the standard in future studies of the epidemiology of P. malariae and P. ovale.

Why are P. malariae and P. ovale infections still important?

Given that both P. malariae and P. ovale are relatively mild infections and are easily curable with common antimalarials, why should we pay closer attention to this increased burden of infections with either species?

First, considering the problems with LM diagnosis, the actual burden of illness could be markedly underestimated and PCR diagnosis studies of clinical cases are thus needed. Although considered mild, P. malariae can cause a chronic nephrotic syndrome that, once established, does not respond to treatment and carries a high rate of mortality [71]. In addition, P. malariae is known to cause chronic infections that can last for years [39] and might reoccur decades after initial exposure when people have long since left endemic regions [72]. The health burden of such chronic or reoccurring infections in an endemic context is not clear.

Second, some studies based on LM indicate that there is evidence for interactions between P. malariae, and possibly P. ovale, with P. falciparum and P. vivax infections. Although cross-sectional studies have reported positive associations between infections of P. falciparum and P. malariae and/or P. ovale, respectively [3,27,28,34], these associations are more likely to represent individual differences in exposure (i.e. children with a high risk of acquiring P. falciparum infections also have a high risk of acquiring other infections) [34] or susceptibility to infection [27], rather than true biological interactions between the parasite species.

Third, because P. malariae is often overlooked by LM, it is difficult to assess potential negative associations or ‘suppression’ by P. falciparum and/or P. vivax in mixed infections. The observed seasonal differences in Africa between P. falciparum and P. malariae prevalence rates and P malariae densities have been interpreted as a suppression of P. malariae during the periods of high P. falciparum transmission, at least in older individuals [5,8,27]. In addition, in PNG a comparatively later age of peak prevalence of P. malariae, seven to 16 years [4,25,29], is observed, compared with that in Africa, under ten years [3,6,8]. Does this observation indicate a negative interaction between P. vivax (which is most common in children under ten years) and P. malariae?

Most importantly, however, is the possibility that P. malariae infections might have a mitigating effect on both P. falciparum and non-P. falciparum illness and, therefore, that treatment of P malariae and/or P. ovale could increase the pathogenesis that is associated with other malaria species. In Nigeria, Black et al. [73] observed (by PCR) a significantly lower prevalence of coinfections with P. malariae in clinical P. falciparum cases (0%) compared with asymptomatic controls (27%). They hypothesized that chronic infections with P. malariae could contribute to a downregulation of the cytokine cascade. In another African study, α-thalassaemic pregnant women had a higher risk of harbouring mixed P. malariae and P. falciparum infections but a lower risk of febrile symptoms and signs of inflammation than women who were infected with P. falciparum alone [74]. This led the authors to propose that the increased susceptibility to P. malariae might partly be responsible for the mild courses of P. falciparum malaria that occur in α-thalassaemic pregnant women. Similarly, in a study in PNG, infections with P. malariae were associated with a subsequent decrease in overall health-centre attendance with presumptive malaria, with a stronger effect on non-P. falciparum rather than P. falciparum disease [34]. By contrast, observations from the Gambia showed that in children under seven years P. malariae episodes are most common in the dry season, when P. falciparum infections and illness are less common [37]. This indicates that suppression of P. malariae by P. falciparum might account for the low level of morbidity associated with P. malariae infections in African children. Understanding the potential for such complex interactions on the morbidity that is attributable to each of the malaria species of humans could be complicated by inaccurate diagnosis.

Little is known about the potential for interactions between P. ovale and other malaria infections. However, the fact that P. ovale has been found to be most prevalent in areas of West Africa, where P. vivax is almost absent because of the high prevalence of the Duffy blood-group-negative phenotype [33], might also indicate a negative interaction between these two species. Because P. ovale prevalence and parasitemia are consistently low, it is often the case that there are insufficient observations to enable meaningful statistical evaluations regarding interactions between P. ovale and the other malaria parasite species that infect humans.

Concluding remarks

With vaccines being developed against P. falciparum and P. vivax, it is important to determine the burden of infection and disease due to P. malariae and P. ovale and assess the potential for interaction between these malaria-parasite species. Should negative interactions be important, then reducing the burden of P. falciparum or P. vivax morbidity could increase the burden of morbidity that is attributable to P. malariae and P. ovale. The possibility for such an effect is highlighted by studies in Brazil and PNG that found high levels of P. malariae infections in isolated populations where either P. falciparum was absent [75] or P. vivax rare [76]. Further studies, using PCR diagnosis, into the burden of infection and illness with P. malariae and P. ovale in different parts of the world are, therefore, clearly warranted.