Cerebrospinal fluid and serum biomarkers of cerebral malaria mortality in Ghanaian children

Malaria is an important neglected disease and one of the most important global health problems, potentially affecting more than one third of the world's population. Cerebral malaria (CM) is a deadly complication of Plasmodium falciparum infection, associated with a 10–14% mortality rate and approximately 1–2 million annual deaths among young children predominantly in sub-Saharan Africa and Southeast Asia, yet its pathogenesis remains incompletely understood. In Ghana, malaria has a wide spectrum of presentations: from asymptomatic carriers to mild malaria to multifactorial severe disease, including CM and severe malarial anemia (SMA) [1‐5].

CM, a clinically complex syndrome of coma and potentially reversible encephalopathy, is associated with increased levels of proinflammatory cytokines like tumor necrosis factor (TNF)-α, interferon (IFN)-γ and lymphotoxin [6, 7], and increasingly recognized long-term sequelae in survivors [7‐15]. Although the physiopathology of CM has been extensively investigated, the exact cellular and molecular basis of the neuropathology is still unclear. Recent studies have shown that mechanical blockage caused by sequestration of parasitized red blood cells (pRBCs), leukocytes and platelets [3, 4, 8‐12], secretion of cytokines and chemokines [2, 6, 7, 13], angiogenic failure [14, 15], immune status and the genetic background of the host, and parasite factors [7, 16] are involved in the pathogenesis of CM. However, it is generally accepted that two major factors are involved: (i) metabolic insufficiencies due to the sequestration of pRBCs, leukocytes and platelets within brain vessels via upregulated adhesion molecules [3, 4, 8‐12], and (ii) immunological reactions with the local involvement of T cells and monocytes activated by Plasmodium antigens [7, 11]. These two major mechanisms appear to act together under the control of cytokines [7], and chemokines [2] to exacerbate CM.

Postmortem data in humans and murine models of CM show that neuronal damage in brain tissue occurs in CM, although the parasites remain confined to the intravascular space (with no contact with neurons). This strongly suggests that the blood-brain barrier (BBB) is perturbed, and thus the BBB represents a key interface between the intraerythrocytic stages of the parasite and the human host. The functional and morphological evidence supports mild-to-moderate impairment of the BBB, but whether this is sufficient to cause neurological complications such as CM is inconclusive [16, 17]. Mechanical blockage could occur from the ability of pRBC's to adhere to unparasitized erythrocytes and endothelial cells, and sequester in the deep cerebral microvasculature [3, 4, 8‐12]. Parasite sequestration alone, however, cannot account for fatal CM pathogenesis as there is evidence that survivors of CM have the same degree of sequestration during comas as those succumbing to disease. It is evident that host immune factors play an important role in the pathogenesis of CM. [8, 16]. Although parasite sequestration during severe malaria is central to pathogenesis of severe malaria, the role of cytokines, chemokines, apoptotic, and angiogenic factors in exacerbating disease severity remains unclear.

The balance between specific cytokines and chemokines produced in response to infection with Plasmodium falciparum is thought to play an important role in CM and other forms of severe malaria. Severe malaria has been associated with high TNF-α plasma levels in conjunction with increased production of IFN-γ and IL-1β [18, 19] and decreased production of anti-inflammatory cytokines, notably IL-10 and TGF-β [1, 19‐22]. Pro-inflammatory Th1-type cytokines (eg. TNF-α, IFN-γ, interleukin IL-1β, and IL-6) are thought to be critical to the control of exoerythrocytic and erythrocytic Plasmodium falciparum infection [23, 24], but their exaggerated production may also contribute to organ damage, particularly in the brain. It is widely accepted that anti-inflammatory Th2-type cytokines down-regulate Th1-derived cytokines. Th2-type cytokines, such as IL-10, has been shown to regulate Th1-cytokines and prevent CM in some animal models [18]. The regulation of TNF-α levels by IL-10 appears to contribute to the prevention of severe malarial anemia in humans [1, 20, 21]. However, the role that IL-10 plays may depend on its levels, since very high levels of IL-10 have been associated with severe malaria in humans [19] and some animal models [25]. Therefore, cytokines appear to maintain a delicate balance between the control of infection and contribution to disease in falciparum malaria infection. However, the expression of Th1 and Th2 cytokines in CSF, either from the peripheral circulation via the BBB or from neuronal immune cells (glia) has not been adequately addressed.

Chemokines, or chemoattractant cytokines, and their corresponding receptors have also been shown to mediate mobilization and coordination of immune responses to malaria. Chemokines have lympho-chemotactic activity and modulate many infectious and inflammatory diseases, including malaria. The recent demonstration of leukocyte sequestration, in addition to pRBC sequestration [3, 4, 8‐12], within brain vessels in human CM suggests a more important role for leukocytes, including eosinophils, in CM immunopathology than previously thought. Thus, chemokines, including eotaxin, may play an important role in human CM by attracting leukocytes to sequestration sites. Chemokines are less well studied in severe malaria, but recent studies have associated severe malaria infection with increased production of chemokines of the C-C or β subfamily, including regulated upon activation, normal T cell expressed and secreted (RANTES), monocyte chemotactic protein (MCP)-1, macrophage inflammatory protein (MIP)-1α, MIP-1β, and IL-8 [2, 13, 26‐28]. MCP-1, MIP-1α and MIP-1β are potent chemoattractants for monocytes to produce TNF-α and IL-6. IL-8 preferentially recruits neutrophils and plays an important role in inflammatory diseases. Recently, low levels of RANTES have been associated with severe malaria [26, 27, 29], and specifically associated with mortality in children with CM [26]. The low levels of RANTES in severe malaria have been associated with malaria-induced thrombocytopenia [26, 29], given that platelets are a major reservoir of RANTES in the peripheral circulation. In contrast, increased mRNA and protein expression of RANTES and CCR5 was found in localized brain regions of children dying of CM [2]. Interferon inducible protein 10 (IP-10) is a member of the CXC or α subfamily of chemokines, and is induced in response to IFN-γ attracting activated Th1 cells [30]. IP-10 levels have been shown to increase in cultured intervillous blood mononuclear cells isolated from placenta's infected with malaria [31, 32], although no studies have characterized IP-10 levels in human cerebral malaria. Hanum and colleagues recently demonstrated the induction of IP-10 expression in the brain of both CM-susceptible (C57BL/6) and CM-resistant (BALB/c) mice as early as 24 hours post-infection with Plasmodium berghei ANKA, and in KT-5 astrocyte cell line in vitro upon stimulation with a crude antigen of malaria parasites [33]. Currently, the role of chemokines in clinical severity and outcome of malaria, especially the development of CM and SMA in children remains poorly defined.

Angiogenic factors, long implicated as prognostic factors in cerebral ischemia or stroke [34], have been suggested to play a role in the petechial hemorrhages and BBB dysfunction associated with CM pathology [14, 15]. Vascular endothelial growth factor (VEGF) stimulates endothelial cell growth and migration as well as enhancing vascular permeability. VEGF levels (more VEGF+ astrocytes) were higher in CM patients as compared with controls in a post-mortem immunohistology study of CM patients [15]. Platelets, that accumulate with pRBCs in the brain miscrovasculature in CM patients [10, 12], are also implicated in CM pathology through TGF-β induced apoptosis in TNF activated human brain endothelial cells [35, 36]. Platelet derived growth factor (PDGF) is another angiogenic factor that stimulates vascular growth, and has been implicated as a neuroprotective factor inducing regeneration of damaged axons and neuronal growth after ischemia [37]. These angiogenic factors play a dominant role in the recovery from stroke and may be applicable in CM due to their effect on the endothelium, which is central to CM pathology. These angiogenic factors most probably impact the regenerative potential of the parasite-induced BBB damage, rather than impacting neoangiogenesis, since CM is an acute neurological syndrome.

Parasite-induced apoptosis in the host may also mediate the severity of malaria. High levels of Fas-Ligand in sera of human [38, 39], monkey [40], and mice [41] are associated with severity of malaria. Lymphocytes and macrophages express increased levels of Fas and Fas-Ligand during an acute Plasmodium chabaudi infection [40]. TNFR2-deficient mice are resistant to experimental cerebral malaria (ECM), Fas-deficient mice showed 50% reduction in ECM incidence, and TNFR1-deficient mice showed the least reduction in ECM incidence [42‐46]. Lpr & Gld mice, deficient in Fas & Fas-Ligand, are protected from fatal ECM [47]. The presence of these apoptotic factors in CSF and serum, and their relevance in CM and CM-associated mortality has not been fully investigated. The rapid reversibility of the clinical symptoms of CM suggests that tissue necrosis is unlikely to occur [9, 16, 48], making apoptosis a more likely pathogenic mechanism.

Recently, new strategies including magnetic resonance imaging and ophthalmological evaluation of children with CM have been proposed as clinically useful predictors of CM severity, but their reliability is being evaluated. The study hypothesis was that parasite-induced dysregulation in the levels of inflammatory, apoptotic and angiogenic factors at the time of CM death would predict mortality risk of CM. The goal of this study was to identify factors that are tightly associated with CM mortality in Ghanaian children for further development as biomarkers of CM disease. The present study employed a high throughput multiplexed immunoassay to evaluate the predictive value of serum and CSF levels of key immunomodulators (inflammatory, apoptotic and angiogenic proteins) in determining mortality risk in severe malaria in Ghanaian children. We investigated the serum and CSF profiles of 36 different biomarkers (IL-1β, IL-1ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12 (p70), IL-13, IL-15, IL-17, Eotaxin, FGF basic protein, CRP, G-CSF, GM-CSF, IFN-γ, TNF-α, IP-10, MCP-1 (MCAF), MIP-1α, MIP-1β, RANTES, SDF-1α, CXCL11 (I-TAC), Fas-ligand [Fas-L], soluble Fas [sFas], sTNF-R1 (p55), sTNF-R2 (p75), MMP-9, TGF-β1, PDGF bb and VEGF) in order to identify the immune factors which influence progression to fatal outcomes associated with CM.

The present study examined a broad range of disease associated inflammatory mediators, including cytokines, chemokines, and markers of apoptosis and angiogenesis, in postmortem serum and CSF samples of children with CM, SMA, and NM. The study was conducted in an area of moderate Plasmodium falciparum transmission where all the life-threatening complications of malaria occur, namely, coma, severe anemia, and respiratory distress [1‐5]. Although post-mortem studies have provided a wealth of detailed information they reflect, at best, pathology at a single time point after death in the most severely ill patients and may be potentially biased by post-mortem artifacts (agonal changes that may simulate disease-induced pathology). The concurrent studying of these inflammatory, apoptotic and angiogenic biomarkers in appropriate time-matched post-mortem controls from other disease causes helps place the results in context. Due to these limitations, further studies would be required to confirm the functional roles of these host factors in CM.

Cerebral malaria (CM) is a major life-threatening complication of Plasmodium falciparum infection in humans. The mechanisms underlying the fatal cerebral complications are still not fully understood. However, two predominant hypotheses are generally proposed to explain the neuropathology of CM, namely the sequestration and immunological hypotheses. The sequestration hypothetical model suggest that the adhesion of pRBCs to the cerebral vasculature leads to obstruction of the microcirculation, metabolic depletion, BBB breakdown, and alteration in brain function resulting in coma [16]. The immunological hypothetical model suggest that hyperimmune responses (originally evolved for the destruction of the parasite and protection of the host) and Th1/Th2 cytokine or chemokine dysregulation results in localized recruitment of immune effectors cells (T cells, monocytes, etc) and BBB impairment resulting in the development of cerebral complications [7]. However, recent studies indicate that parasite induced apoptosis and tissue degeneration, as well as angiogenic factors may be involved in the pathogenesis of CM [14, 15, 36, 38]. Understanding the cytokine/chemokine cascade, parasite induced apoptotic pathways, and dysregulation of angiogenic factors in CM patients will elucidate the underlying pathogenesis and identify potential predictive prognostic biomarkers for CM mortality.

In the present study, evidence is provided indicating that the serum levels of various cytokines and chemokines are altered in children with CM compared to SMA and NM. The elevated serum level of IP-10 is particularly remarkable since it was the only independent predictor of CM mortality. Eight (8) CSF inflammatory biomarkers (IL-1ra, IL-8, IP-10, PDGFbb, MIP-1β, Fas-L, sTNF-R1, and sTNF-R2) were independently predictive of CM mortality, when compared to SMA and NM deaths. The significant increase in CSF levels of IL-1ra, IL-8, IP-10, MIP-1β, sTNF-R1, and sTNF-R2 in CM compared to SMA and NM suggests a critical role for the brain parenchymal expression of these biomarkers in CM pathogenesis and mortality. In the present study, both the serum and CSF levels of RANTES were not predictive of CM mortality, although low serum levels of RANTES have recently been associated with mortality in Ugandan children with CM [26].

TNF-R1 and TNF-R2 are key mediators of the classical extrinsic apoptotic pathway, as well as in inflammation. The increased expression of sTNF-R1 and sTNF-R2 in CSF of CM non-survivors when compared to SMA and NM suggests that parasite-induced apoptosis in host CNS is critical to CM pathogenesis and mortality. Recent studies in murine experimental CM have shown that the TNF receptor super family also plays a role in CM pathogenesis [42‐47]. Mice deficient in TNF-R2 (TNF-R2-/-) and Fas (Fas-/-) survived significantly longer than wild type in experimental CM, and TNFR2-/- mice survived the longest in the absence of anti-malarial treatment [42, 47]. Additionally, the serum levels of sTNF-R1 and sTNF-R2, which act as binding proteins for TNF, were elevated in patients with acute Plasmodium falciparum malaria compared to the levels in convalescent children and in healthy controls [38, 39].

Platelet derived growth factor is a key factor that mediates vascular smooth muscle cell proliferation and serves a neuroprotective role by inducing regeneration of damaged axons and neuronal growth after ischemia [37]. In this study, a significant decline in PDGFbb production was independently predictive of CM mortality. Therefore, it seems that the down regulation of this angiogenic growth factor and upregulation of apoptotic factors in CM patients may be a result of parasite-induced damage or depletion of cells producing PDGF, a highly angiostatic microenvironment with high levels of proinflammatory cytokines/chemokines (notably IP-10), or even an unidentified parasite-derived factor that initiates/exacerbates the inflammatory and apoptotic cascades.

IFN-inducible protein of 10 kDa (IP-10) is a chemokine induced by IFN-γ and TNF-α. Although IP-10 was initially shown to have chemotactic activity for activated Th1 lymphocytes, there is growing evidence implicating this chemokine in both infectious and non-infectious causes of neuronal injury, dementia and inhibition of angiogenesis [50‐54]. This is the first report demonstrating that significant elevation of serum and CSF levels of IP-10 is associated with CM mortality. Our finding suggests that IP-10 plays a major role in the CM immunopathology, and begs for further study in other endemic settings. Interestingly, Plasmodium berghei ANKA infection induced IP-10 and monocyte chemotactic protein (MCP)-1 gene expression in the brain of both CM-susceptible (C57BL/6) and CM-resistant (BALB/c) mice as early as 24 hours post-infection [33]. Additionally, the expression of IP-10 and MCP-1 genes in KT-5, an astrocyte cell line, was induced in vitro upon stimulation with a crude antigen of malaria parasites, suggesting astrocytes as the potential cellular source of cytokine and chemokine expression in brain parenchyma in response to plasmodial infection [33]. Therefore, in response to plasmodial infection, the cells that produce these inflammatory mediators may be different in the brain (microglia and astrocytes) and peripheral circulation (platelets, monocytes, and lymphocytes), and their effects may also differ in the two areas.

This study has revealed new associations, underlying pathogenic events, between different biomarkers and CM mortality in Ghanaian children that may be applicable to other malaria endemic populations. The most important finding demonstrates the association between the elevation of serum and CSF factors involved in the classical extrinsic apoptotic pathways (such as IP-10, TNF-α-sTNF-R1-sTNF-R2 and Fas-L) and the reduction of growth factors that confer endothelial and neuronal cell protection (such as PDGF) with CM mortality. We propose the following hypothesis to explain our observations. It appears that both inflammatory and apoptotic mechanisms may be triggered locally in the human brain during CM that result in the damage of the constituent cells of the BBB (glial cells, astrocytes, and endothelial cells) and possibly neurons. Additionally, this localized BBB damage may be further exacerbated by the significantly decreased levels of neuroprotective angiogenic growth factors (such PDGF), induced by the angiostatic effects of the elevated local CSF levels of IP-10, ultimately resulting in death.

Furthermore, we propose that TNF-α and other proinflammatory factors which are activated following the release of malaria antigens after schizont rupture may induce the local production of IP-10 by the constituent cells of the BBB (glial cells, astrocytes, and endothelial cells) [33]. Subsequently, IP-10 in concert with TNF-α may induce apoptosis of endothelial cells leading to BBB breakdown. Additionally, activated circulating immunomodulator cells (T cells, monocytes, etc) attracted to the BBB by IP-10, may also play a pathogenic role in this process. The significantly decreased production of PDGF may further inhibit angiogenesis and negatively impact the regeneration of damaged endothelial cells and blood capillaries at the BBB. Recently, elevated CSF level of IP-10 has been demonstrated in viral meningitis [54]. Elevated CSF level of IP-10 has been shown to be significantly correlated with the neuropsychiatric impairment in HIV-associated dementia [55]. Furthermore, mouse studies have demonstrated that the HIV-1 virus encoded protein gp120 directly activates astrocytes to produce IP-10 using a novel mechanism independent of IFN-γ and STAT-1 pathway of IP-10 induction [52]. Therefore, elevated serum and CSF levels of IP-10 may be an important pathogenic factor in CM neuropathology, as observed in other infectious disease models. Most CM deaths occur within 24 hours of admission before antimalarials have had time to kill the parasites [9, 16, 48], hence new interventions that address pathophysiological processes causing these early deaths is a public health priority, in addition to addressing the public-health problems resulting in delayed presentation to hospital and ensuring children receive prompt and appropriate resuscitation. Thus, this study provides new insights into the processes leading to cerebral malaria and mortality associated with it.

This study has demonstrated an association between CM associated mortality with elevated serum and CSF levels of apoptotic factors (IP-10, IL-1ra, sTNFR1, sTNFR2, sFas) and reduced serum and CSF levels of neuroprotective angiogenic growth factors (PDGFbb). The observations support recent reports that implicate parasite-induced apoptosis and angiogenic factors in CM neuropathology. Further studies in other endemic areas to confirm these findings are necessary.