Piecing Together the Puzzle of Severe Malaria

See allHide authors and affiliations

Science Translational Medicine  13 Nov 2013:
Vol. 5, Issue 211, pp. 211ps18
DOI: 10.1126/scitranslmed.3007432


Severe malaria manifests as several overlapping syndromes with high mortality. Interaction of parasites with endothelial protein C receptors and high parasite biomass have recently been identified as key determinants of severe disease. However, gaps in our understanding of severe malaria might hinder translation of these findings into new therapies.


To date, no effective adjunctive therapies exist for severe malaria (SM) (1). This implies that our understanding of its pathogenesis remains incomplete. Plasmodium falciparum is the main cause of SM in humans and accounts for at least 600,000 deaths per year, mainly in children in sub-Saharan Africa. Many manifestations of SM are described, and their frequency varies with age and prior exposure. However, most children with severe falciparum malaria can be identified by a combination of just three overlapping syndromes: impaired consciousness [cerebral malaria (CM)], severe anemia (SA), and acidosis/hyperlactatemia (often identified clinically by respiratory distress) (2). These syndromes, on which we focus this Perspective, differ in epidemiological, clinical, and biological characteristics (Fig. 1A). No single animal model satisfactorily reproduces all of the features of SM in humans (3), making it difficult to determine the pathogenesis (Fig. 1B).

Fig. 1. The puzzle of severe malaria.

(A) The overlapping syndromes of severe malaria (SM) are characterized by epidemiological, clinical, and biological differences. RR, relative risk. (B) Major gaps remain in the understanding of the pathogenesis of SM. The generation of a high asexual parasite biomass is necessary to cause SM, but the host and parasite factors that permit this are largely unknown. High parasite biomass initiates the pathological triumvirate of extensive pRBC sequestration, microvascular endothelial cell dysfunction, and inflammation—all of which contribute to localized tissue hypoxia and ischemia. But it is still not clear what additional factors determine polarization to the clinical phenotypes of cerebral malaria, acidosis, or severe anemia.


One of the first pieces to fit into the SM puzzle was the phenomenon of parasite sequestration—retention of parasitized red blood cells (pRBCs) in small blood vessels—discovered more than 100 years ago (1). Postmortem specimens of individuals dying from malaria revealed crowding of pRBCs in some microvasculature beds, most notably in the brain. This observation has been repeatedly confirmed for CM (1), but there are scarce data on the extent of sequestration in the other SM syndromes, and proving that sequestration causes SM is challenging (4). Extensive sequestration of pRBCs is also a feature that distinguishes P. falciparum from the other five Plasmodium species causing malaria in humans and hence is believed to be the reason why P. falciparum is the most deadly. The extent and nature of parasite sequestration in the brain is one feature that is very different in the mouse model of CM (so-called experimental CM, or ECM) and has been the main detractor from the relevance of this model (3).

Another candidate for the pathogenesis of SM is an excessive host immune response. Production of cytokines such as tumor necrosis factor–α (TNF-α), interleukin-6 (IL-6), and IL-10 is generally higher in CM and respiratory distress than in uncomplicated malaria, but considerable heterogeneity exists between and within SM syndromes (5, 6). In ECM, the causal role of immunopathology has been clearly established (3), but causality has not been proven in human SM.

More recently, a pathological triumvirate has emerged, as vascular endothelial dysfunction has gained prominence as a central mechanism linking inflammation and sequestration (Fig. 1B) (7). Regulation of blood flow, barrier function, and coagulation in the microcirculation is controlled by endothelial cells, and these are profoundly perturbed in SM by reduced nitric oxide bioavailability and inflammatory endothelial activation. Proinflammatory cytokines, components released from pRBCs (such as hemoglobin, glycosylphosphatidylinositol, and histones), and cytoadherent pRBCs may all contribute to endothelial activation, resulting in loss of barrier function, localized derangement of coagulation, and amplification of the inflammatory response (710). Furthermore, endothelial activation may amplify the pathological process by increasing expression of endothelial surface receptors to which pRBCs can adhere (7).


Although one or more elements of the triumvirate may initiate the pathogenesis of SM, their reciprocal interactions make it difficult to determine their relative importance in established disease. We suspect that once SM is initiated, redundancy between mechanisms develops, which will make therapeutic intervention particularly challenging. Yet, there must be some differences between individuals and/or parasites that explain why some develop SM and others do not; or, why some develop CM, others hyperlactatemia, and others SA (Fig. 1B). For a long time, it has been suspected that differences in the ability of pRBCs to adhere to microvascular endothelium would provide the explanation, and a flurry of recent discoveries has almost realized this expectation (8, 1113).

The best-characterized adhesion ligands expressed on the surface of pRBCs are the members of the highly polymorphic P. falciparum erythrocyte membrane protein 1 (PfEMP1) family, encoded by the var gene family. A series of recent studies have shown that particular PfEMP1 subtypes containing sequences known as domain cassette 8 (DC8) and DC13 are strongly associated with binding to human cerebral microvascular endothelium and occur much more frequently in subjects with CM than in subjects with uncomplicated malaria (1113). A subsequent study screened a large library of human plasma membrane–expressed proteins to show that DC8-PfEMP1 binds preferentially to the endothelial protein C receptor (EPCR) and blocks the binding site for activated protein C (APC) (8). Binding of APC to the EPCR is important in maintaining integrity and quiescence of the vascular endothelium (14), and the blocking of this interaction may link pRBC cytoadhesion to endothelial activation (Figs. 1B and 2).

Fig. 2. Linking parasite biomass with pathology.

Parasite biomass is determined by the replication rate, the clearance rate, and the number of replication cycles. Replication rate may increase in response to the metabolic environment or environmental stress, resulting in more daughter parasites per generation and a high circulating biomass relative to the sequestered biomass. Extensive sequestration, mediated by specific PfEMP1 variant expression, reduces clearance of pRBCs by the spleen and allows more parasites to survive each generation, resulting in a high total parasite biomass and relatively low circulating biomass. Although not shown, replication and sequestration can vary simultaneously. Parasite-derived endosome-like vesicles may act as a sensor, and potentially a negative regulator of biomass in vivo, to prevent excessive replication and host death, but they are also proinflammatory.


Other observations suggest the story is not quite so straightforward. First, the same DC8-PfEMP1 variant was also found to be strongly associated with severe malarial anemia, a manifestation in which sequestration is not usually thought to be important (12). Neither is binding specific to the brain; it also occurs avidly in microvascular endothelial cells from other organs (13). But most notably, an almost simultaneous report of postmortem findings from Malawian children with CM demonstrated a dramatic association of sequestration in areas of the brain with loss of EPCR (9). This depletion of EPCRs was associated with endothelial dysfunction and local coagulopathy, but it is hard to understand how dense sequestration of pRBCs can occur when the specific receptor is not present. More work is needed to determine whether pRBC binding actually causes shedding of the EPCR, which could derepress the expression of other endothelial surface receptors [such as intercellular adhesion molecule 1 (ICAM1)] and allow them to mediate additional pRBC sequestration. Unfortunately, once CM is established, it may be too late to reverse the process by targeting the EPCR with recombinant APC, but downstream interventions targeting the vascular endothelium may be more successful (7, 9).


Another determinant of malaria severity has recently come into focus: parasite biomass. An obvious question is whether the risk of developing SM is related to the number of blood stage parasites in the body. Conventionally, parasite load is determined by means of microscopic examination of a blood smear and calculation of the percentage of pRBCs (parasitemia). But sequestration of parasites creates a problem, because sequestered pRBCs are not sampled in the circulating blood used for the blood smear. Thus, the circulating parasite biomass is an underestimate of the total biomass, and this might explain why circulating parasitemia is not a consistent predictor of outcome.

Dondorp and colleagues discovered that the plasma concentration of a soluble parasite molecule, histidine rich protein 2 (PfHRP2), is approximately proportional to the total biomass of parasites in the body (15). Since then, this method has been used to show that parasite biomass is an important predictor of severity and outcome in children (1620). Although this is not particularly surprising, it has some important implications. The most immediately relevant is improvement of the definition of SM; a major problem in resource-poor malaria-endemic settings is distinguishing SM from other infections, such as bacterial sepsis or meningitis, with coincidental parasitemia. Defining SM by using threshold concentrations of PfHRP2, appropriate for the intensity of malaria transmission, might help to differentiate patients who truly have SM from those who need treatment with antibiotics (17, 18). Furthermore, such a threshold would improve the specificity of SM diagnosis in pathogenesis studies and clinical trials.

Another important implication of the dependency of SM on high parasite biomass is the potential to prevent SM by restricting parasite replication with a blood-stage vaccine. Blood-stage malaria vaccines have had little success to date because either they have demonstrated limited reductions in parasitemia and clinical malaria, or benefits have been restricted only to the vaccine sequence of polymorphic antigens (21). However, their ability to reduce parasitemia and febrile malaria is probably not the yardstick against which they should be measured; instead, their ability to reduce the total parasite biomass to levels below those associated with severe malaria should be used. The recent discovery that almost all of the 60 parasite var genes can be simultaneously expressed by a single parasite after knockdown of the P. falciparum variant–silencing SET gene (PfSETvs) (Fig. 2) (22) raises the exciting prospect of developing an attenuated blood-stage vaccine that might simultaneously limit cytoadhesion, restrict total parasite biomass, and overcome the problem of antigenic variation.


Returning to the puzzle of SM (Fig. 1B), the association of high parasite biomass with all forms of SM raises additional questions. Why do some individuals develop higher parasite biomass than others? And how do different manifestations of SM arise from the same central feature of high parasite biomass? The rate of expansion of parasite biomass is determined by the opposing processes of parasite replication and parasite clearance. Conceptually, a high parasite biomass can be achieved by increasing the rate of replication [producing more asexual progeny (merozoites) per parasite], by avoiding clearance by the host response, or by increasing the number of cycles of parasite replication (Fig. 2).

Two recent observations have revealed plausible mechanisms controlling inter-individual variation in parasite replication. First, the number of merozoites produced per schizont is controlled by the nicotinamide adenine dinucleotide (NAD+)–dependent histone deacetylase PfSir2a (Fig. 2) (23); here, NAD+ dependence could link variation in the host metabolic environment to control of parasite replication. Second, simple stochastic variation in gene expression, under epigenetic control, causes variation in replication rates between parasite clones under selective pressure, such as heat stress (24). In contrast, sequestration is the archetypal mechanism for the parasite to avoid destruction in the host spleen, by reducing the circulation of mature parasites through this organ (Fig. 2). Particularly rapid rises in parasite biomass might be expected when parasites achieve both rapid replication and sequestration. But an inexorable increase in asexual parasite biomass is unlikely to be in the interests of the parasites, because they will inevitably kill the host and consequently themselves. So, might Plasmodium have evolved to sense and regulate their own biomass within a host and switch to the generation of sexual stages (gametocytes) that are nonpathogenic but essential for transmission to mosquitoes and propagation of infection to other hosts?

Recent evidence suggests that exosome-like vesicles (EVs) containing proteins and genetic material are actively released from pRBCs, taken up by other pRBCs, and instruct changes in behavior (including gametocytogenesis) in a dose-dependent fashion (25, 26). These same EVs are also proinflammatory, triggering IL-1, IL-6, TNF-α, and IL-10 secretion from macrophages and, in turn, neutrophil chemotaxis (26, 27). Whether EVs contribute to control of total parasite biomass in vivo remains to be assessed, but it is tempting to speculate that such a mechanism might be a key determinant of disease severity (Fig. 2). Furthermore, the discovery that release of EVs is an active process dependent on the PfEMP1 trafficking protein 2 identifies a possible therapeutic target for prevention of gametocytogenesis and therefore blockade of malaria transmission (25).

Because high parasite biomass can be achieved in different ways, it is possible that these also contribute to the different clinical manifestations of SM. On one hand, if sequestration is the dominant mechanism permitting high biomass (Fig. 2), pathology caused by microvascular obstruction and endothelial dysfunction might be expected to be prominent and CM more likely (Fig. 1B). On the other hand, if enhanced replication is the dominant mechanism, circulating biomass would be expected to be relatively high compared to sequestered biomass (Fig. 2), and the greater elaboration of parasite-derived molecules might perhaps skew the balance toward systemic inflammation and acidosis (Fig. 1B). Consistent with this, recent evidence suggests that parasites from CM patients preferentially bind vascular endothelium, whereas those from patients with respiratory distress show a greater propensity to bind other RBCs, forming “rosettes” (28). Furthermore, we have recently shown that children with hyperlactatemia have the expected high total parasite biomass composed of a very high circulating parasitemia but have relatively little sequestration (16).

In contrast, several studies have reported very high concentrations of PfHRP2 in children with SA (1618), who often have relatively low circulating parasitemia, indicating a high sequestered parasite biomass. This is of considerable interest because SA actually carries the lowest mortality rate of the SM syndromes, indicating that somehow there is a tolerance to this high sequestered biomass. Such tolerance may simply relate to the location: Sequestration in some tissues (such as adipose tissue) may be less damaging than sequestration in others (such as brain tissue). Alternatively, it may arise from a slower rise in parasite biomass, with extensive sequestration but a low replication rate (Fig. 2) and a more gradually evolving inflammatory response characterized by lower levels of proinflammatory cytokines, much lower levels of the anti-inflammatory cytokine IL-10 (6), and dyserythropoiesis (Fig. 1B).


We now return to the problem that no effective adjunctive therapies have yet been developed to treat SM, despite many failed attempts (1). New treatments may need to target the poorly defined events downstream of the pathogenic triggers (Fig. 1B), perhaps individually tailored to each SM syndrome. Current trials of nitric oxide and l-arginine therapy, aimed at restoring endothelial quiescence (7), go some way to addressing this. Alternatively, combinations of treatments, targeting all components of the pathological triumvirate, may be required to overcome the expected redundancy between these mechanisms. Making such treatments affordable and practical in a resource-poor setting will be a major challenge.

Perhaps the first step should be to ensure that our basic assumptions about pathogenesis (Fig. 1B) are correct. Reevaluation of the nature of host inflammatory and endothelial response in relation to total and sequestered parasite biomass seems to be particularly important. For example, when cytokine concentrations in SM are considered relative to parasite biomass, we suspect that in some SM syndromes there will actually be a relative deficiency of inflammatory cytokines for the number of parasites. It is conceivable that selective enhancement of some components of the inflammatory response, perhaps coupled with strategies to limit ensuing immunopathology, might actually improve parasite clearance and recovery. Like looking at a jigsaw puzzle of a crime scene, we may not be able to deduce the true sequence of events, or the culprits we need to target, until we fit the last few pieces of the SM puzzle into place.


Stay Connected to Science Translational Medicine

Navigate This Article