The silent threat: asymptomatic parasitemia and malaria transmission

Abstract

Scale-up of malaria control interventions has resulted in a substantial decline in global malaria morbidity and mortality. Despite this achievement, there is evidence that current interventions alone will not lead to malaria elimination in most malaria-endemic areas and additional strategies need to be considered. Use of antimalarial drugs to target the reservoir of malaria infection is an option to reduce the transmission of malaria between humans and mosquito vectors. However, a large proportion of human malaria infections are asymptomatic, requiring treatment that is not triggered by care-seeking for clinical illness. This article reviews the evidence that asymptomatic malaria infection plays an important role in malaria transmission and that interventions to target this parasite reservoir may be needed to achieve malaria elimination in both low- and high-transmission areas.

Keywords::

• antimalarial
• asymptomatic
• elimination
• infection
• interventions
• malaria
• parasite

Figure 1. The proportion of malaria infections that are asymptomatic compared with population prevalence of malaria across 16 sites in 14 countries.

Error bars correspond to the 95% CI. Hollow and filled shapes indicate diagnosis by microscopy and PCR, respectively. Diamonds, triangles and squares indicate surveys conducted in Asia, Latin America and Africa, respectively.

Data taken from [35,36,51,52,94,114,120,124,126–132].

Figure 2. Studies reporting on the proportion of the asymptomatic population with malaria parasitemia as diagnosed by PCR, by estimated malaria transmission intensity (n = 39).

The top and bottom of the error bars indicate the maximum and minimum malaria prevalence, respectively; the dark line in the box indicates the median prevalence; and the top and bottom of the box indicate the 75th and 25th percentile, respectively.

Data taken from [51–53,77,91,95,98–126].

Figure 3. Proportion of malaria infections detected by PCR that are asymptomatic by age across three studies in five sites.

The error bars correspond to the 95% CI for the proportion.

Data taken from [90,110,131].

There has been an extraordinary scale-up of malaria control interventions over the last decade with significant increases in household ownership and individual use of insecticide-treated bednets (ITNs), unprecedented growth in the number of houses protected by indoor residual spraying (IRS) and important improvements in access to effective treatment for clinical malaria. As a result, there has been an estimated 17% decline in the number of malaria cases and a 26% decrease in the malaria-specific child mortality rate globally between 2000 and 2011 [1]. In this optimistic climate, the African Union, at its third session of the Conference of Ministers of Health (Johannesburg, South Africa) in April 2007, advocated for eventual elimination of malaria from the continent [2]; this was followed by a call in October of the same year from Bill and Melinda Gates [201], with support from the WHO [3], for global malaria eradication. As of 2011, 36 of the 99 countries remaining with malaria transmission are pursuing elimination [4].

Despite the recent progress on reducing malaria morbidity and mortality, there is empirical and theoretical evidence that the current suite of interventions will not be sufficient to eliminate malaria from many areas in sub-Saharan Africa with historically high levels of malaria transmission. For example, in western Kenya, despite more than 10 years with high coverage of ITNs, parasite prevalence in children <5 years of age had declined from 83% in 1992 to only 41% by slide microscopy in 2009 [5,6]. Similar observations have been noted in areas formerly considered as high-transmission areas, such as Zambia [7] and Uganda [8]. A mathematical simulation of Plasmodium falciparum transmission in Africa suggested that only in areas with the lowest baseline level of transmission (<three infective bites per person per year) could malaria be eliminated through a combination of ITNs, IRS and case management with an artemisinin-based combination therapy (ACT) [9].

The situation in the Americas and Asia is somewhat different from Africa due to the higher proportion of malaria cases caused by Plasmodium vivax, whose dormant liver stage poses an extra challenge to elimination. Additionally, many of the vectors in these areas are exophagic and, therefore, are less likely to be affected by the use of ITNs or IRS. It is recognized that the same malaria interventions successfully used in sub-Saharan Africa may not work as well in P. vivax-endemic areas, and the technical feasibility of eliminating P. vivax from areas where it is currently endemic is not yet known [10].

As a result of the empirical and modeling data suggesting that current interventions will need to be supplemented by additional strategies for malaria to be eliminated in most of sub-Saharan Africa, and the recognition that there is not yet a feasible technical strategy to eliminate P. vivax from Latin America and Asia, there is increased attention being paid to the human parasite reservoir [9,11]. The human parasite reservoir consists of all malaria infections in people in a given area, including symptomatic and asymptomatic infections, and both the sexual and asexual stages of the parasite (the special case of dormant liver stages of P. vivax and Plasmodium ovale will not be addressed in this review).

Improvements in case management through the use of more effective antimalarials have helped in reducing malaria transmission in some areas [12], but low utilization of healthcare, suboptimal performance of healthcare workers, problems with patient adherence to treatment regimens and drug stock-outs limit the ability of case management strategies to significantly decrease the fraction of the parasite reservoir harbored in symptomatic individuals [13]. The expansion of community case management may help to address problems of access, but health system issues are likely to remain a challenge in most areas. Additionally, case management, by definition, only addresses symptomatic individuals, and there has been increasing recognition that a substantial proportion of the parasite reservoir may be found in persons who do not show symptoms and therefore do not seek care [11]. In part, the use of molecular assays such as PCR to detect parasite DNA has improved the sensitivity of diagnostics to find subpatent (i.e., below the detection limits for microscopy) infections that are more likely to be asymptomatic, and this has contributed to the understanding of the extent of the asymptomatic parasite reservoir [14].

As the magnitude of the asymptomatic parasite reservoir has been revealed through increasing use of more sensitive molecular diagnostic methods, new strategies to target individuals with silent infections are being developed and evaluated. The objective of this review is to examine the data that support the hypothesis that targeting the asymptomatic parasite reservoir will contribute substantially to reductions in malaria transmission and eventual malaria elimination.

How does Plasmodium infection cause acute clinical symptoms?

Malaria sporozoites of the genus Plasmodium (comprising four species that are transmitted between humans by Anopheles mosquitoes: P. falciparum, P. vivax, P. ovale and Plasmodium malariae) are injected into humans through the bite of an infective female Anopheles mosquito. The malaria sporozoites pass rapidly to the liver, within 30 min of infection, where they mature into schizonts within hepatocytes over 5–16 days, depending on the parasite species. The mature schizonts then rupture the cell and enter the bloodstream as merozoites. Merozoites infect red blood cells and over 2–3 days develop into eryrthrocytic schizonts, eventually destroying the erythrocyte, and releasing more merozoites and cellular debris into the bloodstream. The immediate impact of this rupture is a significant decline in red cell mass, and cellular debris and cytokines released by the host lead to the development of acute clinical symptoms, including fever, rigors and myalgias [15].

How are malaria infections transmitted?

Gametocytes, the transmissible stage of the Plasmodium parasite, are produced by a small fraction of merozoites that differentiate into gametocytes upon entering the red blood cell, although gametocytes of P. vivax, P. ovale and P. malariae can also arise from emerging liver stage merozoites [16]. Whereas these three species produce gametocytes within the same time frame as asexual parasitemia, P. falciparum gametocyte production is delayed in comparison [17]. As a consequence, a higher proportion of P. vivax malaria patients are found with gametocytes shortly after developing symptoms than in patients with P. falciparum [18], and P. vivax infections can be transmitted before becoming symptomatic. Gametocytes are ingested by female Anopheles mosquitoes in a blood meal and they undergo sexual reproduction in the mosquito midgut, eventually forming an ookinete that crosses the midgut wall and develops into an oocyst. The oocyst produces sporozoites that migrate into the mosquito salivary gland, waiting to be injected into a susceptible host at the time of the anopheline’s next blood meal. The extrinsic incubation period of the Plasmodium parasite within the mosquito is temperature dependent and ranges from 10 to 14 days for P. falciparum [19] and 11 to 22 days for P. vivax [20].

Why are some malaria infections asymptomatic?

Partial immunity to malaria infections can lead to a reduction in the acute clinical symptoms of disease. Antidisease immunity is acquired from exposure to malaria infection and develops more quickly with frequent exposure; most children in areas with moderate-to-high levels of malaria transmission gain protection from severe disease by a very young age, usually by 2–5 years of age, followed by a decrease in the rate of symptomatic illness in early adolescence [21]. By contrast, antiparasite immunity increases with maturation of the immune system and appears to be somewhat independent of exposure frequency in areas of moderate-to-high transmission. By adulthood, parasite densities following infection often remain at very low levels, frequently undetectable by microscopy, and most infected adults do not exhibit clinical symptoms. However, asymptomatic parasitemia can occur at any age.

Development of partial immunity to clinical malaria is antigen-specific, mediated by exposure to different genetically distinct parasite subpopulations, termed clones. Cross-protection can be conferred by antigenically similar clones [22]; in areas of high transmission, residents are frequently exposed to a diversity of clones, resulting in rapid development of antidisease immunity and asymptomatic infections [23]. In areas of low transmission, however, there is a lower rate of multiclonal infections and antidisease immunity develops more slowly [24].

Immunity to P. vivax is acquired more rapidly than to P. falciparum. In Papua, New Guinea, children aged 5–13 years were 21-times more likely to experience fever after infection with P. falciparum than P. vivax [25]. These findings were mirrored in children aged 1–4 years of age who showed an increasing ability with age to keep parasite densities of P. vivax, but not P. falciparum, below the pyrogenic threshold [26].

How is asymptomatic parasitemia defined?

There is no standard definition of asymptomatic parasitemia, and previous studies have used a range of definitions and diagnostic criteria. Most definitions involve the detection of asexual or sexual parasites and an absence of any acute clinical symptoms of malaria (usually fever) during a specified time frame. Whereas it is not always explicit, asymptomatic parasitemia refers to bloodstream infections and does not include dormant liver stages.

A number of studies have used parasite density thresholds to define cases of clinical malaria, that is, only cases of fever with parasite densities above a predetermined cutoff were considered to be symptomatic malaria cases [27]. This results in a de facto classification of febrile infections with low density parasitemia as asymptomatic, since their fever was not statistically attributed to malaria. While the use of parasite density threshold cutoffs provides a more specific end point for studies of vaccines or clinical treatments [28], its converse should not be used to define asymptomatic parasitemia for burden or impact assessments.

Laishram et al. provide a list of diagnostic criteria that have been used in various studies to define asymptomatic malaria cases (Table 1) [29]. The duration of time used to define an infection as asymptomatic, both prior to and after the diagnosis, varies between studies; many have defined asymptomatic infection as no measured fever at the time of the survey, whereas others have required as many as 60 days of follow-up without clinical symptoms [30]. In some cases, fever is the only clinical symptom considered whereas other studies include a variety of additional nonspecific disease symptoms. When follow-up periods are not monitored for the development of symptoms, there may be a concern that parasitemia detected during the malaria incubation period may be misclassified as an asymptomatic infection, but given the short duration of the time between the completion of the prepatent period and the end of the incubation period (~1 day for P. falciparum and 4 days for P. vivax), this is likely an infrequent occurrence. In some studies, cases have been excluded if they reported prior use of antimalarial medication to avoid infections in the process of being cleared.

Gametocytes do not cause disease symptoms, and asymptomatic individuals may or may not have detectable gametocytes. The relationship between asexual parasite density and gametocyte density is not straightforward; studies have found both positive and negative associations [31]. Gametocytes may linger in peripheral blood up to several weeks after an asexual parasite infection has been cleared (whether by natural immunity or by drugs); one trial demonstrated that gametocytes persist an average of 55 days after treatment with a non-ACT and 13.4 days after treatment with an ACT [32].

What factors are associated with asymptomatic malaria infection?

Immunity is the factor that most strongly determines whether a malaria infection produces symptoms. An individual’s level of immunity to infection is determined by past exposure history and age. Increased immunity leads to improved control over parasite multiplication and decreased parasite density, which in turn lessens the severity of symptoms. In a national survey in Mozambique, children <10 years of age with low-density P. falciparum infections (1–499 parasites [p]/µl) had a prevalence of fever of 7.2%, compared with 42.1% among children whose asexual parasite densities were ≥50,000 p/µl [33]. In Brazil, parasite density was compared between symptomatic (age 12–78 years) and asymptomatic (age 4–56 years, with no fever or malaria symptoms for 7 days prior to blood collection) individuals infected with P. vivax and P. falciparum; lower asexual parasite densities were found among the asymptomatic individuals for both parasite species, although the difference was much greater for P. vivax [34].

Some asymptomatic infections may be residual or recrudescent parasitemia remaining after treatment for a clinical episode. In a low-transmission setting in Sudan, a higher proportion of people who experienced a clinical episode of malaria during the transmission season from September to December had an asymptomatic infection detected by PCR in the following January than did those without a prior clinical episode (35 vs 8%, respectively), despite having been clinically cured of their symptomatic infection by treatment with chloroquine [35]. It is possible that drug-resistant parasites could persist at a low level after treatment, with densities controlled by immunity developed during the initial infection [36]. However, the parasites in these infections were not genotyped to determine whether they were the same clones present during the clinical episode or new infections.

Coinfections may affect the development of symptoms by altering immune system function. HIV-1 infection has been shown to increase rates of malarial fevers in a dose–response fashion, with declining CD4 T-cell counts [37]. The association between soil-transmitted helminths and development of clinical symptoms among children with malaria parasitemia is less clear: hookworm infection appears to increase malaria symptoms, whereas infection with Ascaris lumbricoides appears to decrease symptoms [38]. Coinfection of malaria and schistosomiasis is a frequent occurrence, but the effects on malaria immunity and transmission are complex. Some studies suggest that schistosomiasis coinfection favors development of antimalarial immunity [39,40], whereas others have found lower levels of protective malaria antibodies in Schistosoma haematobium carriers [41]. A recent prospective study demonstrated that children coinfected with malaria and schistosomiasis were more likely to have detectable gametocytes and higher gametocyte densities than children infected only with malaria, potentially facilitating more intense malaria transmission [42].

The method used to diagnose infection will determine the number of low density, asymptomatic infections that are identified. The detection limits of microscopy are typically estimated at 4–20 p/µl in a reference laboratory, but are more realistically 50–100 p/µl under field conditions [43]. A few rapid diagnostic tests (RDTs) have been shown to have greater than 90% sensitivity and specificity for P. falciparum at parasite densities ≥200 p/µl [43–45] but may often fail to detect lower density infections. PCR is considered to be the gold standard for detection of parasitemia with a limit of detection of 0.02 p/µl for the most sensitive procedures [46].

A review of studies reporting both PCR and microscopy results for P. falciparum detection found that, on average, microscopy underestimates the prevalence of Plasmodium infection as detected by PCR by 50.8%, and this difference is greatest in low-transmission settings [14]. RDTs, with detection thresholds higher than microscopy, would also be expected to underestimate the proportion of the population that has an infection, most of which will be asymptomatic. As a result, interpretation of estimates of asymptomatic infections should consider the diagnostic method used. Despite the increased sensitivity of PCR over microscopy and RDTs, there is still concern that it may not detect all malaria cases with very low parasite densities [47].

Does the prevalence of asymptomatic parasitemia vary by Plasmodium species?

All malaria species can infect without causing symptoms, but because P. falciparum and P. vivax are more prevalent globally, more asymptomatic infections of these two species occur than the other two human malaria species. Asymptomatic infections with P. ovale are rare but have been noted [48,49], whereas P. malariae is known to persist in the bloodstream for decades without causing any, or only mild, symptoms [48].

Individuals with P. vivax develop immunity more quickly than with P. falciparum and consequently are able to control parasite densities to a greater degree, but it is not clear whether this translates into a different proportion of infections that are asymptomatic by species. Few comparisons of the prevalence of symptoms among individuals infected with P. falciparum and P. vivax have been conducted in areas where the two species are sympatric, and results are contradictory. Using microscopy to diagnose infections, a larger proportion of P. falciparum infections (37.5%) compared with P. vivax infections (18.5%) in Brazil were asymptomatic (presenting with none of the 13 malarial symptoms) [50]. In comparison, a greater proportion of P. vivax infections (97.1%) in the Solomon Islands compared with P. falciparum infections (82.2%) were asymptomatic (axillary temperature: <38°C) [51]. Few studies have reported both the number of species-specific malaria cases and species-specific symptom rates, and diagnosed all infections by PCR; in Cambodia, 92% of the P. falciparum and 83% of the P. vivax infections were asymptomatic [52], whereas in Brazil, these proportions were 78 and 93%, respectively [53]. While it appears that the majority of prevalent infections of both species are asymptomatic in cross-sectional surveys, there are no data to suggest that either species is associated with a larger reservoir of asymptomatic infections.

Do asymptomatic blood-stage infections eventually become symptomatic?

Studies have suggested that a proportion of people with asymptomatic blood-stage malaria infection may become symptomatic, although reinfections or other causes of fever cannot always be ruled out. In Brazil, 93 asymptomatic (presenting with none of 13 malarial symptoms) infections in 33 persons aged 5 years and older were followed for 2 months after their infection was identified and ten (10.7%) became symptomatic during that period [50]. However, the background transmission rate would suggest that eight of those cases could have been the result of new infections. In a cohort study of asymptomatic (defined as no self-reported fever for the previous 24 h and axillary temperature <37.5°C) parasitemic Tanzanian children aged 1–5 years who were followed over 31 days, 55.9% (19 out of 34) of the children eventually developed fever, which was associated with spikes in parasite density [23]. A prospective study of the risk of developing clinical malaria in a high-transmission area of Tanzania followed 265 parasitemic but asymptomatic (no fever during previous 4 weeks or during 1 week after recruitment) residents aged 1–84 years of age over 40 weeks, and observed 21 (7.9%) cases of fever in conjunction with a parasite density >400 p/µl [54].

Subpatent asymptomatic infections may be less likely to become symptomatic than infections detectable by microscopy. In Uganda, 25 out of 63 (39%) children aged 6 months to 5 years with subpatent infections who were asymptomatic (no malaria treatment in previous 2 weeks or fever in previous 48 h) developed symptoms within 20 weeks of observation compared with 43 out of 53 (82%) children with patent infections [55]. These findings suggest that low-density malaria infections may persist for long periods without causing symptoms if not treated.

It is not clear what triggers the appearance of symptoms in individuals who have been parasitemic but remained asymptomatic for a period of time. It has been suggested that reinfection with new clones (to which the individual had not previously been exposed) could trigger an increase in parasite density and the development of symptoms [56], but other studies have demonstrated that symptoms can appear without any change in clonal diversity [55].

How long do asymptomatic infections last?

As described above, some portion of asymptomatic infections may become symptomatic and receive appropriate antimalarial treatment. However, many asymptomatic infections can persist for significant periods of time. Reports in the literature of soldiers returning home from malaria-endemic areas have demonstrated asymptomatic infections lasting up to 13 months for P. falciparum [57]. A statistical modeling approach combined with highly sensitive molecular techniques using an all-age cohort in Ghana – an area of high P. falciparum transmission – estimated that untreated asymptomatic infections had a mean duration of 194 days (95% CI: 191–196) [58]. In areas of highly seasonal transmission, low-density asymptomatic infections are believed to persist over the course of the dry season and to reseed transmission when mosquito populations increase along with wetter conditions [59,60]. P. malariae has been found in at least one case to remain asymptomatic for decades [61].

Does the prevalence or density of gametocytes differ between symptomatic & asymptomatic infections?

Several studies have shown that the presence of P. falciparum gametocytes is positively associated with an absence of symptoms and low asexual parasite densities. While some antimalarial treatments such as sulfadoxine–pyrimethamine (SP) may stimulate the production of gametocytes [62], this association has been detected even when non-gametocyte-stimulating treatments or no treatments have been used, and may arise from the natural concurrence between resolving infections and the delay before gametocytes are produced in P. falciparum infections. On the western border of Thailand, being afebrile and having low P. falciparum asexual parasite densities were independently associated with increased prevalence of gametocytemia [63]. Researchers in The Gambia evaluating children with P. falciparum infections at enrollment for antimalarial trials found that being afebrile at the time of examination increased the risk of gametocytemia by 67%, and having lower asexual parasite density (<100,000 p/µl) increased the risk fivefold [64]. Similarly, in children with P. falciparum infections in Nigeria, being afebrile (axillary temperature 37.5°C) was associated with a 61% increase in gametocytemia, and having asexual parasite densities <5000 p/µl more than doubled the odds of being gametocytemic [65]. None of these studies used PCR to detect gametocytes, suggesting that the prevalence of gametocytemia may have been underestimated; however, it is unlikely that this would have changed the association described between asymptomatic infections and the presence of gametocytes.

Whereas a greater proportion of asymptomatic infections may have gametocytes, it is not clear whether gametocyte density will be higher. As gametocytes in P. falciparum infections arise from asexual parasites (i.e., merozoites), there could be a positive correlation between the density of asexual and sexual parasite infections, leading to an association between low density, asymptomatic infection and lower gametocyte densities. However, it is also possible that inflammatory and nonspecific immune factors related to symptomatic infections may negatively affect gametocyte production, potentially leading to higher gametocyte densities among asymptomatic infections. In Brazil, asymptomatic P. vivax gametocyte carriers had lower gametocyte densities as measured by PCR than symptomatic individuals [66]. However, in Kenya, asymptomatic P. falciparum gametocyte carriers had higher gametocyte densities, but lower asexual parasite densities than symptomatic individuals [67]. These inconsistencies could be attributed to differences in the biology of P. falciparum and P. vivax. No clear relationship between asymptomatic infection and gametocyte density has been established.

How much do asymptomatic infections contribute to transmission?

Gametocyte density appears to be a critical factor in determining whether a mosquito develops infection from an infective blood meal. An analysis of 930 transmission experiments showed a largely log–linear positive relationship between gametocyte density and the prevalence of infection in mosquitoes [68]. This suggests that if asymptomatic infections are associated with lower gametocyte densities, asymptomatic infections would be less likely to result in mosquito infections. In a Brazilian study aimed at demonstrating the viability and contribution of asymptomatic infections to malaria transmission, 1.2% of the mosquitoes feeding on asymptomatic persons (n = 15) versus 22% of the mosquitoes feeding on symptomatic persons (n = 17) developed oocysts in their midguts [30]. Different feeding techniques in the two populations may have affected the comparison but it is likely that the lower gametocyte density among the asymptomatic individuals contributed to the observed differences in mosquito infection.

Despite increasing probability of infection with increasing gametocyte density, numerous studies have demonstrated that mosquitoes can become infected by blood from individuals with gametocyte densities as low as five gametocytes (g)/µl, and theoretically as low as 1 g/µl [69]. To look at the relative transmissibility of infections when gametocytes were detectable by microscopy, by PCR or were not detectable, researchers in western Kenya used venous blood from children with and without gametocytes detected by microscopy to feed mosquitoes through membrane feeders [70]. Blood from children with subpatent gametocytes infected half as many mosquitoes as those with patent gametocytemia, but due to the frequency of subpatent gametocytemia in the sample of children, the end result was that both groups contributed equally to the total number of infected mosquitoes. In this study, children with gametocytemia that was undetectable even by PCR were still found to contribute to almost 10% of the overall number of infected mosquitoes, demonstrating that gametocytes below detection thresholds can still result in malaria transmission. P. vivax and other Plasmodium species are more efficient at transmitting earlier in the infection and at lower densities than P. falciparum, and therefore a greater proportion of individuals infected with these species can transmit without detectable gametocytemia [69].

There may be factors associated with the presence of symptoms that alters infectivity to mosquitoes. In western Kenya, a significantly smaller proportion of mosquitoes that fed on blood from symptomatic individuals (0.6%) developed oocysts than those that fed on asymptomatic persons (12%) [67]. Although symptomatic individuals were found to have higher asexual but lower gametocyte densities than asymptomatic individuals, the authors concluded that the increased oocyst development in mosquitoes that fed on asymptomatic individuals was not due solely to the higher gametocyte densities, but also to an increased infectivity of these gametocytes. This increased ‘quality’ of gametocytes has been postulated by others [29] and could be due to a variety of factors, including the stage of development of the gametocytes (for P. falciparum, symptoms tend to occur earlier in infection when gametocytes may not have reached a level of maturity to be optimally infective), an inherent property of the parasite strain, a direct effect of symptomatology (i.e., the febrile response may affect the infectivity of gametocytes), or to a more specific host immune response, such as transmission-reducing antibodies [17].

Are individuals with asymptomatic infections more likely to get bitten by malaria vectors?

There is no direct evidence that individuals with asymptomatic infections are more likely to be bitten than symptomatic individuals, but gametocyte carriers have been shown in one study to be more attractive to mosquitoes than both uninfected individuals and individuals with only asexual parasites. Comparing the same children before and after they were cleared of P. falciparum gametocytes to control for inherent differences in individual attractiveness, researchers in Kenya were able to show that Anopheles gambiae were more attracted to children when they were gametocytemic than when they were uninfected or when they harbored only asexual parasites [71]. As asymptomatic individuals infected with P. falciparum may be more likely to be gametocytemic than symptomatic individuals, it is possible that asymptomatic persons have an increased likelihood of being bitten, but there is no clear evidence yet to support this hypothesis, nor is there any information on P. vivax infections.

How does transmission intensity affect the prevalence of asymptomatic parasitemia?

It has long been held as conventional wisdom that in lower transmission settings, the proportion of infected individuals who are asymptomatic will be less than in high-transmission settings because the population level of immunity is decreased [72]. Figure 1 (updated from data presented by Macauley [73] to include more recent surveys that use PCR) shows the proportion of malaria infections (all species) that are asymptomatic across a wide range of transmission, prepared from 16 different sites in 14 countries. Only studies that measured the point prevalence of all malaria infections using a population-based cross-sectional survey design and included the number of infections that were asymptomatic were included, although the symptoms and the reference time period used in the definition of asymptomatic differed between studies. The data suggest that, while there is an increase in the proportion of the parasite reservoir that is asymptomatic as transmission increases, the majority (>60%) of prevalent infections are asymptomatic even at low transmission. Studies using PCR had the highest rates of asymptomatic parasitemia, but even some studies based on microscopy found greater than 80% of the cases of parasitemia to be asymptomatic. Many of the studies did not report information allowing calculation of the proportion of asymptomatic infections by species, but seven studies that reported on P. falciparum found 73–98% of infections to be asymptomatic, and the five studies reporting on P. vivax found 64–100% of infections to be asymptomatic.

The data presented by this graph challenge conventional wisdom and understanding of how immunity develops, but the malaria cases reported in these studies were identified through active surveillance and are estimates of point prevalence. Prevalence is a product of incidence and duration of infection; as symptomatic infections are more likely to receive treatment, and the duration of treated infections is significantly shorter than untreated infections, it follows that symptomatic infections are less likely than asymptomatic infections to be present at any one point in time. Similarly, once appropriate treatment is provided, symptoms may resolve more quickly than infection. The few studies that have attempted to capture all new infections over a period of time, both symptomatic and asymptomatic, have found a lower proportion of asymptomatic infections using cumulative incidence than point prevalence. Employing a combination of passive and active case detection over a 14-month period to identify all new malaria infections in an area of Brazil, 326 episodes of malaria were identified, of which 96 (29.4%) were asymptomatic at the time of diagnosis [50] – a proportion much lower than has been found in any cross-sectional survey.

Without a strong association between transmission intensity and the proportion of infections that are asymptomatic, the population prevalence of asymptomatic malaria infection mirrors the overall transmission level. Figure 2 demonstrates the relationship between the prevalence of infection, as detected by PCR, among asymptomatic individuals and transmission intensity (as measured by the overall prevalence of infection) in 34 different studies in 38 sites in Latin America (n = 9), Asia (n = 10) and sub-Saharan Africa (n = 19). In this analysis, studies using a population-based cross-sectional survey design, in which the asymptomatic population could be identified and malaria was confirmed using PCR, were included. Age groups varied, but there is a positive correlation between transmission intensity and the prevalence of malaria infection among asymptomatic individuals (Table 2).

How does age affect the proportion of malaria infections that are asymptomatic?

In areas of high malaria transmission, the proportion of the population that is parasitemic tends to decrease with age [31,74], although this relationship may be somewhat muted when molecular assays are utilized [75]. Conversely, the proportion of malaria infections that are asymptomatic generally increases with age, presumably due to acquired immunity and maturation of the immune system. After manipulating age categories to develop consistent groupings between studies, Figure 3 presents the proportion of malaria infections by age (0–4, 5–14 and 15+ years) across three studies in five sites that used PCR for diagnosis. There is a consistent positive trend in the proportions of infections that are asymptomatic with age across the five sites, but confidence intervals are wide. The lack of more significant differences by age could be a result of using prevalent rather than incident infections, or the age groupings may be too wide to detect differences between age categories.

What public health interventions target asymptomatic parasitemia?

There are a number of drug-mediated strategies that implicitly or explicitly target the asymptomatic parasite reservoir. By definition, these strategies do not rely on any symptomatic trigger, and thus do not include health facility treatment, community case management, or fever surveys. Essentially, there are three approaches that will target drugs to the asymptomatic parasite reservoir: providing the population in a given geographic area with antimalarials irrespective of their infection or symptom status (mass drug administration [MDA]); universal screening of the population with a malaria diagnostic test, followed by treatment of those infected (mass screen and treat [MSaT], also referred to as aggressive-active case detection) [73]; and repeated treatment of high-risk groups with an antimalarial not guided by diagnostic testing (intermittent preventive therapy). Whereas the primary objective of the first two approaches is a reduction of malaria transmission, intermittent preventive therapy strategies were conceived to reduce the burden of malaria and associated adverse consequences in high-risk groups, although theoretically, there may be an impact on transmission if the high-risk group comprises a significant proportion of the infective reservoir. These approaches can be modified by targeting smaller geographic areas (such as hotspots of transmission) or combined with other interventions, such as follow-up of passively detected cases, to fit different transmission settings and objectives [76]. Additionally, the frequency of MDA and MSaT can be adjusted in light of the underlying transmission intensity, seasonality and type of antimalarial used.

What theoretical data suggest that these approaches can reduce malaria transmission?

Several modeling attempts have demonstrated that targeting the asymptomatic parasite reservoir with MSaT or MDA could result in reductions in malaria transmission, even though some of the effects were modest [9,77,78]. Griffin et al. found that absolute reductions of 5–12% in parasite prevalence could be achieved in low-to-moderate transmission settings (entomologic inoculation rates of 7–94 infective bites per person per year) with 80% coverage of annual single rounds of MSaT using RDTs, in conjunction with 80% ITN coverage over 15 years of sustained program; the addition of IRS would significantly improve the final outcome [9]. In areas with high malaria transmission (entomologic inoculation rates of 630 and 703 infective bites per person per year), the modeled impact of MSaT alone was marginal, with the ultimate outcomes depending greatly on the initial transmission level as well as the frequency and duration of the intervention and the level of vector control established in conjunction.

Using the same basic underlying model, Okell et al. examined the impact of MDA and MSaT under different assumptions of coverage, drug choice, timing and screening sensitivity, and found that MDA could significantly improve chances for malaria elimination in small, concentrated pockets of transmission [78]. In areas of moderate transmission (baseline parasitemia prevalence: 10–30%), single rounds of MDA would not have a lasting effect, but repeated rounds could reduce malaria transmission substantially. The authors found that in these higher transmission settings, MSaT using microscopy to identify infections had less of an impact than MDA, but they noted that information on the infectivity of subpatent infections is limited.

What empirical data suggest that these approaches can reduce malaria transmission?

Italy was the first country to implement MDA on a large scale beginning in 1900, through both quinine prophylaxis and curative treatment [79]. This approach led to large reductions in malaria morbidity and mortality, but the country eliminated malaria only after an IRS program using DDT was added to the MDA with quinine after World War II. Since the 1930s, at least two dozen additional MDA interventions have been conducted [80]. Whereas many MDA interventions achieved temporary reductions in malaria morbidity and transmission, few had rigorous study designs and MDA was often combined with other interventions. Only one project on the small, isolated island of Aneityum in Vanuatu succeeded in permanently interrupting transmission through a combination of weekly chloroquine, SP and primaquine, as well as high coverage of ITNs [81]. In 1981, 70% of the Nicaragua population participated in a 3-day MDA with chloroquine and primaquine. The immediate impact of the MDA was substantial but the effect lasted only a few months [82].

The first large-scale community-randomized, placebo-controlled trial of MDA in Africa, which took place in The Gambia in 1999, found that one round of MDA with SP and single-dose artesunate in the dry season resulted in a brief reduction in malaria incidence that was not sustained over the 20-week observation period [83]. The only other recent trial of MDA in Africa attempted to interrupt malaria transmission in Tanzania using a treatment course of artesunate and SP plus primaquine to clear gametocytes [84]. However, despite achievement of high coverage, transmission intensity was too low overall to demonstrate any impact. A recent trial of MDA with one round of a full course of dihydroartemisinin–piperaquine and a low dose of primaquine every 10 days for 6 months in 17 villages of Cambodia found that parasite prevalence among both children and adults was dramatically reduced after 3 years from 52.3 to 2.6%, as measured by microscopy [85].

The Malaria Eradication Program of the 1950s and 1960s used screening of symptomatic individuals through mass fever surveys as an intervention, and in 1960, this approach was proposed to be extended to asymptomatic carriers [86]. MSaT of asymptomatic carriers was carried out on a large scale in Taiwan, India, Oman, Brazil, China and on one island of the Philippines, and led to large reductions in malaria and, in one case (Taiwan), elimination [73]. In the USSR, significant declines in malaria cases to near-elimination levels resulted after the implementation of mass surveys in the late 1950s focused on asymptomatic parasite carriers [87]. China was also able to achieve malaria elimination in a significant proportion of the country in the second half of the 20th century through rigorous active case detection focused on both symptomatic and asymptomatic individuals [88,89]. Despite their apparent success, these early experiences with MSaT – from the 1950s to the 1990s – typically did not have a control group and were often implemented in combination with other interventions.

One variation on MSaT is screening and treating only within known malaria hotspots, also termed focal or focused screen and treat (FSaT). Given the focalized nature of malaria, particularly in low-transmission settings, FSaT has the potential to more efficiently target screening and treatment resources to areas that could have a disproportionate impact on transmission reduction. The malaria control program of Zanzibar, conducted prior to the transmission season, has experimented with FSaT in hotspots and documented significant decreases in malaria cases in the intervention areas over time and compared with control areas [90]. FSaT using PCR was used in 20 villages of Cambodia in 2010, reaching 6931 (72.7%) residents over a 5-month period [52]. A total of 133 infections were detected and successfully treated, but the impact on transmission is not yet known.

Using passively detected, symptomatic cases to identify high-risk households or geographic areas has been referred to as reactive or targeted case detection. In response to a household with a positive case, other household members along with the population within some distance of the index household would be screened and treated [13]. Depending on how clustered or focalized malaria transmission is in the area, reactive case detection could more efficiently identify asymptomatic infections than MSaT. Using this approach in Zambia, members of RDT-positive households, passively detected clinical malaria cases had a prevalence of malaria parasitemia by nested PCR of 8.0% compared with 0.7% in households in which no member had sought treatment for clinical illness [91]. The study was not designed to evaluate the impact of the approach on malaria transmission. Both Zanzibar and selected areas of Zambia have implemented reactive case detection in a programmatic fashion, although results are not yet available on the impact of this approach on malaria prevalence and transmission.

Conclusion

Individuals who receive repeated malaria infections over time eventually achieve increased immune control with a resultant decrease in acute symptoms. Without clinical illness, these infections are silent and remain untreated, resulting in chronic carriage that can last for 6 months or longer. Asymptomatic infections may be associated with a greater probability of gametocyte carriage, although there is also likely to be a lower density of gametocytes in these individuals, and these effects may cancel themselves out with respect to altering the infectivity of asymptomatic infections. However, the proportion of mosquito infections that arise from asymptomatic infections is likely to be high due to a combination of the proportion of asymptomatic infected individuals in the population at any given point in time and the duration of their infections. Evidence summarized in this report suggests that even in areas of low transmission, the contribution of asymptomatic infections to transmission is likely to be substantial, and in areas with seasonal transmission, asymptomatic infections may serve as the source of infections for a new generation of mosquitoes emerging after the start of the rains.

Lack of a standardized definition of asymptomatic infection may be limiting the ability to compare burden across space and time, and to measure the impact of control or elimination programs that target the asymptomatic parasite reservoir. Given that both P. falciparum and P. vivax have an approximately 48-h febrile cycle, it seems reasonable to define asymptomatic infections operationally as cases of malaria parasitemia of any density, preferably diagnosed using PCR, without current, measured fever or history of fever in the past 48 h that have not received appropriate antimalarial treatment in the past 3 days. Increasing the time without symptoms prior to diagnosis or adding an observation period afterwards (often not feasible) would potentially decrease the specificity of the definition by increasing the probability of encountering fever caused by another infection. It could be argued that including other symptoms besides fever might improve the specificity of the case definition, but there are no other symptoms commonly used to define clinical malaria, and use of fever alone greatly simplifies the definition of asymptomatic infections.

Strategies targeting asymptomatic infections are available, but rigorous research efforts to compare the relative effectiveness of different approaches at varying levels of malaria transmission are needed. The approaches that most efficiently and quickly find and eliminate asymptomatic parasite reservoirs may provide the endgame for malaria elimination in many areas.

Expert commentary

With the advent of molecular diagnostic techniques, there has been increasing recognition that a significant proportion of the infective reservoir across the spectrum of malaria transmission is comprised of asymptomatic infections. Ultimately, it is the gametocytes that are taken up in the blood meal of female Anopheles mosquitoes that go on to form sporozoites responsible for infecting other humans, but programmatic interventions to interrupt this transaction must focus on individuals with malaria infection at any stage, as asexual parasitemia left untreated will eventually produce gametocytes, and diagnostics for the sexual stage are limited.

Whereas our understanding of how partial immunity to malaria develops would suggest that malaria infections are more likely to be symptomatic in low-transmission settings, data summarized in this report suggest that at any one point in time, the majority of the parasite reservoir is likely to be asymptomatic. This finding is consistent across transmission settings and may be a result of symptomatic individuals receiving treatment that reduces both the length of time they are symptomatic as well as the time they are parasitemic, thereby removing symptomatic infections from the parasite reservoir more rapidly than asymptomatic infections. Regardless of why prevalent infections are more likely to be asymptomatic, interventions that are not linked to the presence of symptoms will be needed, even in low-transmission settings, to significantly reduce malaria transmission.

These data suggest several research questions that remain to be answered. At least one study has found that prevalent, asymptomatic infections are associated with earlier episodes that were clinically cured, albeit with an antimalarial medication (chloroquine) no longer recommended for sub-Saharan Africa [35]. The extent to which asymptomatic parasitemia is associated with partial treatment, treatment failures and residual populations of resistant parasites should be investigated further. The relative contributions of symptomatic and asymptomatic infections to transmission have been measured in very few studies; in particular, the role played by subpatent asymptomatic infections should be investigated further in both low- and high-transmission areas, as this proportion of the asymptomatic reservoir is reached by MDA but not by MSaT and could help determine which of these two strategies will be most efficacious in a given setting. It is essential to rigorously evaluate the cost–effectiveness of different approaches to target the parasite reservoir. Community-randomized controlled trials of different strategies in areas with varying levels of P. falciparum and P. vivax transmission should be initiated to inform program managers and elimination strategies of the best way forward. Care should be taken to monitor populations participating in MSaT, FSaT and MDA interventions for safety issues and development of drug resistance, as well as to identify any potential rebound in clinical malaria that could occur as a result of eliminating asymptomatic infections. Finally, integrated drug- and insecticide-mediated interventions should be evaluated in the field to identify optimal mixes that will result in malaria elimination.

Five-year view

Strategies targeting the asymptomatic parasite reservoir were used in the past, but results did not always match expectations. In particular, MDA has fallen out of favor because results were short lived and there was the potential for selection of drug-resistant strains. As a result, there is significant interest in evaluating MSaT approaches. MSaT has the additional advantage in that the location of malaria infections can be mapped easily without additional effort, allowing for more focused interventions to be layered on top. As we learn more about the potential for infections missed by MSaT to sustain transmission, MDA may prove to be a better option. Currently, there are a number of efforts underway to evaluate MSaT, MDA and FSaT in a variety of transmission intensities and with slightly different strategies for identification of cases and responses. Some evaluations will compare these approaches directly to case management, and much information on the role of asymptomatic parasitemia in malaria transmission will be gleaned from comparisons of strategies that target different segments of the parasite reservoir. Over the next 5 years, this research agenda is likely to expand significantly and contribute to new strategies to find and eliminate malaria infections.

Table 1. Examples of diagnostic criteria used for defining malaria patients as asymptomatic.

Study (year), region	Criteria used for identifying asymptomatic malaria cases	Study subjects, sample size (n)	Follow-up protocol and duration	Ref.
Africa
Abdel-Latif et al. (2003), Gabon	No clinical symptoms of malaria with a Plasmodium falciparum positive blood smear, asymptomatic for at least 5 days during follow-up	Children 6 months to 10 years of age (60)	Examined once daily for 7 days; thereafter, once every 2 days	[92]
Rottmann et al. (2006), Tanzania	Presence of P. falciparum on blood smear, axillary temperature of <37.5°C and no other symptoms or signs of malaria	Children 4–59 months of age (127)	No follow-up	[93]
South America
Alves et al. (2002), Brazilian Amazon	Individuals tested positive with microscopy and/or with PCR, and individuals tested negative with microscopy, which subsequently became positive using PCR	All age groups (172)	Follow-up on days 10 and 60	[94]
Cucunubá et al. (2008), Colombia	Presence of patent asexual parasite stages of P. falciparum, Plasmodium vivax or Plasmodium malariae or of mixed infections in blood, which persisted for at least 2 weeks without causing any symptoms, or as the detection of parasite DNA by PCR on day 0 in people who remained asymptomatic during the follow-up period	Individuals 2–78 years of age (21)	Follow-up on days 14 and 28	[95]
Asia
Boutlis et al. (2003), Papua	No fever history or treatment for malaria within the past week, no clinical evidence of malaria or other infection, no diarrhea and no current pregnancy but both P. falciparum- and P. vivax-positive individuals	Adults >16 years of age (105)	Supervised overnight at local health center. A third axillary temperature was recorded the following morning	[96]
de Mast et al. (2010), Indonesia	Presence of asexual P. falciparum or P. vivax parasitemia in the absence of fever (temperature ≤37.9°C) and clinical signs or symptoms that are suggestive for malaria or another infectious disease	Children 5–15 years of age (381)	No follow-up	[97]

A more comprehensive list of studies is given in the supplementary material of [29].

Table 2. Studies reporting on malaria prevalence in asymptomatic individuals diagnosed using PCR.

Study (year), country	Study design	Age group (years)	Patients tested (n)	Patients positive for malaria (%)	Patients positive for Plasmodium falciparum (%)	Patients positive for Plasmodium vivax (%)	Ref.
Low transmission
Fernando et al. (2009), Sri Lanka	Cross-sectional	All age groups	1854	0			[98]
Turki et al. (2012), Iran	Cross-sectional	4–60	500	0			[99]
Zoghi et al. (2012), Iran	Cohort of 500 patients in each of two study areas (three cross-sectional surveys)	2–60+	1000	0			[100]
Atkinson et al. (2012), Solomon Islands	Cross-sectional	All age groups	541	0.6	0	0.6	[101]
Kumudunayana et al. (2011), Sri Lanka	Cross-sectional	All age groups	140	1.4	0	1.4	[102]
Hoyer et al. (2012), Cambodia	Focused screening and treatment	All age groups	6310	1.8	0.9	1.0	[52]
Congpuong et al. (2012), Thailand	Cross-sectional	All age groups	475	1.9			[103]
Diallo et al. (2012), Senegal	Stratified urban sample	2–10	2231	2.2			[104]
Eisele et al. (2011), Haiti	Cross-sectional	All age groups	647	2.2	2.2	0	[105]
Barracks et al. (2010), Vanuatu	Cross-sectional	2–12	4230	2.5			[106]
Hsiang et al. (2010), Uganda	Cohort	1–10	472	3.2	3.2	0	[107]
Harris et al. (2010), Solomon Islands	Cross-sectional	All age groups	8107	3.6			[51]
da Silva et al. (2010), Brazil	Cross-sectional	5+	334	4.8	2.4	2.4	[108]
Kritsiriwuthinan and Ngrenngarmlert (2011), Thailand	Cross-sectional	Adults	241	6.2	3.3	2.5	[109]
Stresman et al. (2010), Zambia	Cohort	All age groups	176	7.4	7.4	0	[91]
Harris (2010), Solomon Islands	Cross-sectional	All age groups	9,491	8.2	4.5	2.8	[51]
Roper et al. (1998), Sudan	Serial cross-sectional studies	All age groups	77	10.4	10.4	0	[110]
Rodulfo et al. (2007), Venezuela	Cross-sectional	All age groups	133	10.5	3.8	6.0	[111]
Gahutu et al. (2011), Rwanda	Cross-sectional	0–5	529	13.2	13.2	0	[112]
Katsuragawa et al. (2010), Brazil	Cross-sectional	5+	324	13.9	1.2	12.7	[113]
Cucunubá et al. (2008), Colombia	Cross-sectional with follow-up on days 14 and 28	All age groups	212	14.6	4.2	9.4	[95]
Camargo et al. (1999), Brazil	Longitudinal	All age groups	169	14.8	0	14.8	[114]
Alves et al. (2002), Brazil	Cross-sectional	All age groups	147	15.0	4.8	9.5	[53]
Färnert et al. (2009), Kenya	Cross-sectional survey with follow-up	0–11	273	16.8	16.8	0	[115]
Moderate transmission
Ladeia-Andrade et al. (2009), Brazil	Five cross-sectional surveys (different villages)	All age groups	644	9.6	3.4	6.2	[116]
Vafa et al. (2008), Senegal	Cross-sectional survey as part of cohort study	2–10	372	13.7	13.7	0	[117]
Suarez-Mutis and Coura (2007), Brazil	Cross-sectional	All age groups	98	20.4	0	20.4	[118]
Baliraine et al. (2009), Kenya	Pooled data from 12 monthly cross-sectional surveys of a cohort	5–14	81	22.6	22.6	0	[119]
Baliraine et al. (2009), Kenya	Pooled data from 12 monthly cross-sectional surveys of a cohort	5–14	81	25.8	25.8	0	[119]
Baliraine et al. (2009), Kenya	Pooled data from 12 monthly cross-sectional surveys of a cohort	5–14	2611	33.3	33.3	0	[119]
Pinto et al. (2000), São Tome and Principe	Cross-sectional	All age groups	81	35.7	28.6	2.4	[120]
Kern et al. (2011), Kenya	Cross-sectional survey with follow-up	0–11	264	36.7	36.7	0	[77]
Crookston et al. (2010), Ghana	Cross-sectional	0–5	84	51.9	51.9	0	[121]
Liljander et al. (2011), Kenya	Cross-sectional	1–6	360	58.9	58.9	0	[122]
High transmission
Koukouikila-Koussounda et al. (2012), Congo	Cross-sectional survey of cohort of children	1–9	313	16.3	16.3	0	[123]
Bereczky et al. (2007), Tanzania	Cross-sectional part of cohort study	1–84	756	38.0	38.0	0	[124]
Dal-Bianco et al. (2007), Gabon	Cross-sectional	18–51	470	51.9	51.9	0	[125]
Pinto et al. (2000), São Tome and Principe	Cross-sectional	All age groups	563	53.3	51.6	2.9	[120]
Owusu-Agyei et al. (2002), Ghana	Cross-sectional	All age groups	297	82.5	82.5	0	[126]

Key issues

• • Advances in molecular diagnostic techniques have revealed a larger reservoir of asymptomatic human malaria infections than previously recognized.

• • Even in low-transmission settings, the majority of prevalent malaria infections are asymptomatic and may persist for weeks or months.

• • With renewed focus on global malaria eradication, strategies targeting the human parasite reservoir are needed to continue to drive down transmission and achieve elimination in malaria-endemic areas.

• • Strategies to target asymptomatic parasitemia include mass drug administration (MDA), mass screen and treat (MSaT), and focused screen and treat (FSaT) in defined ‘hotspots’ of malaria.

• • New research trials of MDA, MSaT and FSaT are underway in a number of countries. Findings from these studies will help inform elimination efforts in a variety of transmission settings.

• • Additional unanswered research questions on asymptomatic parasitemia include the extent to which asymptomatic parasitemia is associated with partial treatment, treatment failures and resistant parasites; the relative contribution of asymptomatic infections to malaria transmission; and the cost–effectiveness and safety of MSaT, FSaT and MDA interventions in different transmission settings.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.