Severe Falciparum Malaria in Children: Current Understanding of Pathophysiology and Supportive Treatment

Associate Editor: P. Winstanley Severe Falciparum Malaria in Children: Current Understanding of Pathophysiology and Supportive Treatment

Charles R. J. C. Newton*‡ and Sanjeev Krishna†‡ *NEUROSCIENCE’S UNIT, INSTITUTE OF CHILD HEALTH, WOLFSON CENTRE, MECKLENBURGH SQUARE, LONDON WC1N 2AP, UK †DIVISION OF INFECTIOUS DISEASES, DEPARTMENT OF CELL AND MOLECULAR SCIENCES, ST. GEORGE’S HOSPITAL MEDICAL SCHOOL, CRANMER TERRACE, LONDON SW17 0RE, UK

ABSTRACT. Severe falciparum malaria is one of the most lethal parasitic infections in the world and is responsible for more than one million deaths in African children per year. Changes to management over the last 40 years have not improved survival. A reduction in the mortality and morbidity may only come about by a better understanding of the pathophysiological processes that are responsible for severe disease and that determine the outcome before antimalarials have had time to work. This review discusses potential adjunctive therapies for severe malaria that are under development following such detailed clinical and pathophysiological studies. pharmacol. ther. 79(1):1–53, 1998. © 1998 Elsevier Science Inc. KEY WORDS. Falciparum malaria, pathophysiology, children, adjunct therapy.

CONTENTS 1. INTRODUCTION ...... 2 6. CLINICAL DEFINITIONS AND 2. DEVELOPMENT OF INFECTION ...... 3 PRESENTATIONS ...... 14 2.1. MULTIPLICATIVE CAPACITY ...... 3 6.1. SEVERE MALARIA ...... 14 2.2. SEQUESTRATION OF PARASITISED 6.2. CEREBRAL MALARIA ...... 16 RED BLOOD CELLS ...... 4 6.2.1. DEFINITION OF CEREBRAL 3. PATHOLOGY ...... 4 MALARIA ...... 16 3.1. PATHOLOGY OF INFECTION ...... 4 6.2.2. COMA SCORES . . . . .18 3.2. MACROSCOPIC APPEARANCES .....5 6.2.3. CLINICAL PRESENTATION 3.3. MICROSCOPIC APPEARANCES .....5 OF CEREBRAL MALARIA IN 3.3.1. VASCULAR CONGESTION ....5 AFRICAN CHILDREN . . .18 3.3.2. RING HAEMORRHAGES 6.2.4. NEUROLOGICAL AND GRANULOMA SEQUELAE ...... 19 FORMATION ...... 5 6.3. METABOLIC ABNORMALITIES IN 3.3.3. CELLULAR ELEMENTS ...... 6 SEVERE MALARIA ...... 23 3.3.4. PIGMENT ...... 6 6.4. SEVERE MALARIAL ANAEMIA, 3.4. RELATIONSHIP BETWEEN JAUNDICE, AND HAEMOLYSIS ... 24 SEQUESTRATION AND CEREBRAL 6.5. THE COMMONEST MODES OF DISEASE ...... 6 PRESENTATION OF SEVERE 4. PATHOGENESIS—HOST FACTORS ...... 7 MALARIA IN CHILDREN ...... 24 4.1. MECHANICAL HYPOTHESIS ...... 7 6.6. MODERATE MALARIA ...... 25 4.2. TOXIN HYPOTHESIS ...... 8 7. PATHOPHYSIOLOGY ...... 25 4.3. CYTOKINE HYPOTHESIS ...... 8 7.1. FEVER ...... 25 4.4. OTHER CYTOKINES AND MARKERS OF 7.2. COMA ...... 25 ENDOTHELIAL CELL ACTIVATION ...9 7.3. SEIZURES ...... 26 4.5. NITRIC OXIDE ...... 9 7.4. RAISED INTRACRANIAL 4.6. REACTIVE OXYGEN SPECIES ..... 10 PRESSURE ...... 26 4.7. THE PERMEABILITY HYPOTHESIS ... 11 7.5. CAUSES OF NEUROLOGICAL 4.7.1. EXPERIMENTAL EVIDENCE .. 11 SEQUELAE ...... 27 4.7.2. BLOOD-BRAIN BARRIER 7.5.1. ISCHAEMIA ...... 27 IN HUMANS ...... 11 7.5.2. REACTIVE OXYGEN 4.8. THE IMMUNOLOGICAL SPECIES ...... 27 HYPOTHESIS ...... 12 7.5.3. EXCITOTOXINS . . . . .27 4.8.1. EXPERIMENTAL EVIDENCE .. 12 7.5.4. APOPTOSIS ...... 28 4.8.2. CLINICAL STUDIES ...... 12 7.5.5. HEMIPARESIS ...... 28 5. PATHOGENESIS—PARASITE FACTORS .. 12 7.5.6. VISUAL IMPAIRMENT . .28 5.1. CYTOADHERENCE ...... 12 7.6. HYPOGLYCAEMIA ...... 29 5.2. ROSETTING ...... 13 7.7. LACTIC ACIDOSIS ...... 30 5.3. RHEOLOGY ...... 14 7.8. ANAEMIA AND THROMBOCYTOPAENIA ...... 31 7.9. ELECTROLYTE ABNORMALITIES AND FLUID BALANCE ...... 32 ‡ Corresponding authors. 7.10. RENAL IMPAIRMENT AND 2 C. R. J. C. Newton and S. Krishna

BLACKWATER FEVER ...... 33 9.5. ACIDOSIS ...... 38 7.11. SECONDARY INFECTIONS ..... 33 9.6. ANAEMIA AND EXCHANGE 8. ANTIMALARIAL TREATMENT ...... 33 TRANSFUSION ...... 39 8.1. INTRODUCTION ...... 33 9.7. DESFERRIOXAMINE ...... 39 8.2. CINCHONA ALKALOIDS ...... 34 9.8. ANTI-INFLAMMATORY AGENTS .. 40 8.3. 4-AMINOQUINOLINES ...... 35 9.8.1. CORTICOSTEROIDS . . .40 8.4. ARTEMISININ DERIVATIVES ... 35 9.9. AGENTS THAT IMPROVE 9. ADJUNCTIVE MEASURES ...... 35 MICROCIRCULATORY FLOW .... 40 9.1. ANTIPYRETICS ...... 35 9.9.1. PENTOXIFYLLINE . . . .40 9.2. ANTICONVULSANTS ...... 36 9.9.2. MISCELLANEOUS AGENTS .40 9.3. MEASURES TO REDUCE RAISED 9.10. OTHER AGENTS ...... 40 INTRACRANIAL PRESSURE .... 37 10. DISCUSSION AND FUTURE STUDIES ... 40 9.4. CORRECTION OF ACKNOWLEDGEMENTS ...... 41 HYPOGLYCAEMIA...... 37 REFERENCES ...... 41

ABBREVIATIONS. BBB, blood-brain barrier; BCS, Blantyre coma score; CM, cerebral malaria; CPP, cerebral perfusion pressure; CSF, cerebrospinal fluid; CT, computerised tomography; CVP, central venous pressure; DAT, direct antiglobulin test; DCA, dichloroacetate; E-selectin, endothelial selectin; ICAM-1, intercellular adhesion molecule-1; ICP, intracranial pressure; Ig, immunoglobulin; IH, intracranial hypertension; IL, interleukin; iNOS, inducible nitric oxide synthase; NCM, non-cerebral malaria; NO, nitric oxide; NPRBC, nonparasitised red blood cell; Pfemp-1, Plasmodium falciparum erythrocyte membrane protein-1; PRBC, parasitised red blood cell; ROS, reactive oxygen species; TNF, tumour necrosis factor; TPI, triose phosphate isomerase; VCAM-1, vascular cell adhesion molecule-1; WHO, World Health Organisation.

1. INTRODUCTION cesses in the 1960s, insecticides have failed to curb trans- Malaria is one of the most common and important parasitic mission by anopheline mosquitoes (World Health Organi- diseases worldwide. About 40% of the world’s population sation, 1993). Insecticide-impregnated bednets have reduced lives in malaria-endemic areas (Sturchler, 1990), and malaria mortality and morbidity in malarious areas (Nevill et al., is responsible for up to 500 million episodes of clinical infec- 1996; D’Alessandro et al., 1995c), but their efficacy may tion and 2.7 million deaths every year (World Health Organ- not be sustainable (D’Alessandro et al., 1995b; Greenwood, isation, 1996). Plasmodium falciparum is the principal cause 1997). Results of recent trials with the current generation of of severe disease, since the other species of malaria rarely malaria vaccines have been unimpressive (D’Alessandro et cause death or persistent sequelae. P. falciparum may infect al., 1995a; Alonso et al., 1994), although newer prototypes humans at any time from conception to adulthood. Malarial are under continuous investigation (Stoute et al., 1997; Tar- infection probably results in 3.5 million low birth-weight gett, 1995). Therefore, it probably will require a concerted infants every year (Steketee et al., 1996), since an estimated combination of measures, including antimalarials, vaccines, 24 million pregnant women live in malaria-endemic areas. and vector control, before a reduction in infection and mortal- Children living in sub-Saharan Africa bear the brunt of the ity can be maintained. Furthermore, reducing the transmission disease, as they are exposed to malaria frequently after birth, of malaria in an endemic area runs the concomitant risk of and either die from complications or experience clinical epi- reducing the development of antimalarial immunity in the sodes of infection for many years until the slow and capricious population. This reduction in “herd immunity” eventually development of antimalarial immunity (Edington, 1967). may result in the rapid spread of severe infection if control This balance between acquisition of immunity and develop- measures fail, so that longer-term study of preventive inter- ment of severe disease has continued for thousands of years. ventions will be crucial to assessment of their true and sus- Since the first description of malarial parasites in a suf- tainable worth. The goal of effective prevention is not yet ferer by Laveran in Constantine, Algeria in 1880, there achievable, and consequently, we will still need to manage have been considerable advances in the understanding of cases of severe malaria in the foreseeable future. disease mechanisms in malaria (Laveran, 1880; White and The early effective treatment of falciparum infection is Ho, 1992). These advances have resulted from detailed critical in preventing the progression to severe, life-threat- studies in patients, animal models of infection, biochemi- ening disease. Falciparum malaria has become increasingly cal, cellular, and molecular investigations. These studies refractory to chloroquine, the cheapest and most widely have illuminated our understanding of disease processes, al- available antimalarial (Krishna and White, 1996; Zucker et though none have succeeded so far in reducing mortality al., 1996). In Southeast Asia, multidrug resistance is rapidly from severe infection. Mortality from treated severe malaria spreading (Pukrittayakamee et al., 1994a; White, 1992), in- in children is between 5 and 15% (Waller et al., 1995; creasing the likelihood of severe disease. In Africa, wide- Marsh et al., 1995), and the current management of severe spread chloroquine resistance has increased the incidence malaria has changed surprisingly little, in spite of rapid sci- of some complications such as anaemia (because of inability entific advances in malariology. to cure infections) (Lackritz et al., 1992), and other related Measures to eradicate malaria have been ineffective problems are anticipated to worsen as higher-grade resis- (World Health Organisation, 1993). After the initial suc- tance becomes entrenched. Pathophysiology and Supportive Treatment of Severe Malaria 3

Newer antimalarial treatment regimens testing, for ex- cytes are asymptomatic. The cause of P. falciparum’s ample, the efficacy of artemether against quinine in chil- virulence in comparison with other human parasites is un- dren with severe malaria, have not confirmed any advan- known, but its multiplicative capacity and ability to seques- tage for artemether in terms of survival (van Hensbroek et ter in the deep vascular beds are thought to contribute. al., 1996a). Yet, in vitro, the artemisinin derivatives are some of the most rapidly parasiticidal drugs, with the broad- 2.1. Multiplicative Capacity est stage-specificity of action (Murphy et al., 1995b; ter The median number of falciparum sporozoites initiating in- Kuile et al., 1993), and indeed, in vivo, they clear circulat- fection is 8–15, but this may reach up to 100 sporozoites ing parasites much faster than other antimalarials (Hien (White and Ho, 1992). In comparison with other malaria and White, 1993). This observation, that the use of more parasites, P. falciparum has a shorter pre-erythrocytic stage rapidly parasiticidal drugs may not affect mortality, points (5–7 days), prepatent period (interval between infection and to alternative directions for further research designed to re- the appearance of parasites in the erythrocytes), and incuba- duce mortality in children with malaria. These alternative tion period (interval between infection and onset of symp- approaches are based on detailed studies of the pathophysi- toms). It also produces more merozoites from liver schizonts ology of infection (White and Ho, 1992), and are aimed at and after erythrocytic merogony. reversing or ameliorating those disease processes that may Each infecting exo-erythrocytic merozoite can yield up contribute directly to a fatal outcome, in spite of the rapid to 36 daughter meronts once the erythrocytic stage of infec- administration of effective antimalarials to the patient. tion is initiated from the liver. Untrammelled multiplica- These adjunctive therapies are designed to support severely tion rates in this stage of infection can exceed 10 per cycle, ill children until the underlying disease can be reversed by logarithmically expanding the parasite burden in an indi- the antimalarial treatment. This review focuses on key vidual. Within a few days, a few thousand parasites liber- pathophysiological observations made in children with se- ated from the liver can progress to a total parasite burden of vere malaria that may guide the eventual development of Ͼ1012 parasites in adults (Fig. 1) (White et al., 1992a; such adjunctive therapies. White and Krishna, 1989). The threshold of microscopic detection is reached after 3–4 asexual cycles, and a lethal parasite burden may be reached in another 3–4 cycles. 2. DEVELOPMENT OF INFECTION Young children will begin the asexual phase of parasite de- Clinical symptoms and signs of malaria occur when P. falci- velopment at a higher parasitaemia, as they have a smaller parum-infected erythrocytes multiply asexually (Marchi- blood volume in which to dilute the merozoites liberated by afava and Bignami, 1894). The hepatic stages and gameto- hepatic merogony (White and Krishna, 1989). Thus, in a

FIGURE 1. Estimates of the increase in malaria parasites infecting an adult during the exponen- tial growth phase of a relatively synchronous infection; effects of sporozoite load and multiplication rate on time to reach potential lethal parasitaemia (as- suming unchanged clearance and multiplication rates). Reproduced from White and Krishna (1989), with permission of the copyright holder, Royal Society of Tropi- cal Medicine and Hygiene. 4 C. R. J. C. Newton and S. Krishna young child with a blood volume of 500 mL, a lethal parasi- tion is that circulating parasitaemia is not representative of taemia could develop in 8 days from hepatic merogony the total parasite biomass in the patient. Indeed, the para- (White et al., 1992a). P. falciparum can invade red cells of sites visible on microscopy have not adhered to host tissues any age, and since more than one merozoite can infect a and are unlikely to be contributing directly to severe symp- single erythrocyte, multiply infected erythrocytes are com- toms of disease (White and Krishna, 1989). Furthermore, mon in falciparum malaria. studies carried out on these circulating parasites, especially ex vivo studies, may not represent the pathogenic potential of sequestered parasites. As sequestered parasites are much 2.2. Sequestration of Parasitised Red Blood Cells more metabolically active when mature, their increased re- During the first half (24 hr) of the parasite’s life cycle, the quirements for synthetic precursors, such as glucose and nu- parasitised red blood cell (PRBC) is metabolically quies- cleosides, may compete directly with the needs of adjacent cent. There is a modest increase in parasite size and the ex- host tissue. This potential for metabolic diversion is an im- pression of some parasite antigens on the surface of the in- portant additional consideration in the pathogenesis of se- fected erythrocyte. These PRBCs continue to circulate and vere malaria, particularly cerebral malaria (CM). are visible by microscopic examination of a patient’s blood. By contrast, the remaining 24 hr of development are char- acterised by intense synthetic and metabolic activity (Sher- 3. PATHOLOGY man, 1979). The PRBCs express molecules on the erythro- The finding of relatively mature parasites within the deep cyte surface, which allows the cells to sequester in vascular vascular bed at post mortem was made soon after the dis- beds. These include the expression of a recently identified covery of the parasite (Marchiafava and Bignami, 1894), family of parasite-encoded genes (the var genes), which are and gave rise to the “mechanical” hypothesis for the patho- primarily responsible for increasing the adhesiveness of genesis of severe malaria. This hypothesis has dominated PRBCs to host ligands expressed on capillary and post-cap- the thinking of severe malaria during this century. We will illary venules (Su et al., 1995). The adhesion of cells infected first discuss the pathological basis of this theory, describing with mature stages of parasites to capillary beds removes the pathology of severe malaria, after which we will further most of them from the circulation, thereby preventing their discuss the mechanistic theory and other theories of the destruction by the spleen. It also allows asexual growth of pathogenesis. the parasite and division to occur in a favourable hypoxic environment, and perhaps allows more efficient invasion of erythrocytes after schizogony (Marsh et al., 1988). 3.1. Pathology of Infection Observations on the age distribution of different stages of One of the difficulties of interpreting pathological findings parasite development on admission blood films suggest that has been heterogeneity in the clinical features of malaria a preponderance of more mature forms (Ͼ20% trophozoites (discussed in Section 6), particularly age-dependent and or schizonts, for example) is a prognostic indicator for fatal- geographical disparities. The lack of clinical definition in ity (Silamut and White, 1993). These circulating mature most pathological studies has particularly hampered inter- forms easily can be quantitated on peripheral blood films pretation. This review focuses on pathological findings ob- and probably represent a fraction of the total sequestered served in CM as a clinically identifiable syndrome of severe parasite burden. disease. Clinico-pathological correlates in severe non-CM At this later developmental stage, there is a surge in the (NCM) still require further detailed studies, which should uptake of essential synthetic precursors, such as glucose (up be guided by validated clinical definitions. to 25- to 50-fold increase compared with uninfected eryth- Most pathologists have defined cerebral involvement as rocytes), amino acids, and nucleosides (Elford et al., 1995). the presence of sequestered parasites in the brain (Turner et The uptake of these metabolites and the disposal of lactate, al., 1994; Aikawa et al., 1980; Edington and Gilles, 1976; the principal waste product resulting from the parasite’s Thomas, 1971; Lemercier et al., 1966; Spitz, 1946). How- anaerobic glycolysis, are mediated through poorly under- ever, patients who die of noncerebral complications, e.g., stood transport processes (Elford et al., 1995; Kanaani and renal failure, without clinical evidence of cerebral involve- Ginsburg, 1991). These biochemical changes are associated ment, may also have cerebral sequestration (MacPherson et with a rapid increase in parasite size, the visible deposition al., 1985). Resolving these apparent discrepancies requires of haemozoin in the PRBC, and subsequently syncytial nu- careful quantitative comparisons of parasite density in cere- clear division, which produces daughter merozoites (Leete bral and other tissues obtained from patients with meticu- and Rubin, 1996). The factors that signal cell rupture when lously documented clinical histories. These studies are now multiplication is complete are also not understood, but sub- in progress and will be valuable in further establishing the sequent invasion of red cells takes place within a few min- validity of the mechanical model for malaria (Turner, 1997). utes, and the developmental cycle is re-initiated. Alternatively, a method of quantitating the total parasite Sequestration of PRBCs is thought to give rise to the burden in an individual with malaria would also allow a di- complications of falciparum malaria (White and Ho, 1992), rect test of the mechanical hypothesis, but this has not which define severe disease. A consequence of sequestra- proved possible by simple methods (Davis et al., 1990a). Pathophysiology and Supportive Treatment of Severe Malaria 5

Most post-mortem studies [reviewed by Turner (1997)] crovascular obstruction. These changes can cause ventricu- have been conducted on adults, and in those series that in- lar compression without associated tissue oedema. In view clude children, differences between adults and children of the potential therapeutic implications, the presence of have not been highlighted. The original autopsy studies in cerebral oedema and the delineation of its role in causing children were from Africa (Lemercier et al., 1966; Thomas, death from CM is an important area for future pathological 1971; Edington and Gilles, 1976), and few have been per- studies in African children. Studies are in progress to ad- formed elsewhere. dress some of these issues (T. Taylor, web site: http:// www.niaid.gov./ictdr/Blantyre.htm). 3.2. Macroscopic Appearances 3.3. Microscopic Appearances In CM, brains packed with PRBC have a slate-grey discol- oration, which is evident on the cut surfaces of the brain in The earliest microscopic observations have dictated much both adults (Toro and Roman, 1978; Schmid, 1974; Khan of the current thinking about the pathogenesis of severe fal- and Durham, 1945; Dhayagude and Purandare, 1943; Thom- ciparum malaria. In 1949, Clark observed: son and Annecke, 1926; Aikawa et al., 1980) and children There is intense congestion of the vessels through- (Thomas, 1971; Lemercier et al., 1966; Edington and Gilles, out the brain and spinal cord with special promi- 1976). The colour reflects the presence of malarial pigment nence in the cortical layer of both the cerebrum (haemazoin). Meningeal vessels are often congested, with and cerebellum. These capillaries contain both surrounding haemorrhage and some infiltration of leuco- parasitised and non-parasitised erythrocytes, de- cytes (Spitz, 1946; Khan and Durham, 1945; Dhayagude pending on the terminal degree of parasitaemia and Purandare, 1943; Boonpucknavig and Boonpucknavig, and localisation. Frequently one area of the cortex 1988; Thomson and Annecke, 1926). Haemorrhages, de- will show scattered parasites while other areas will scribed as “punctiform” or “ring haemorrhages,” are a com- reveal almost complete parasitisation of the eryth- mon pathological feature of CM and are distributed rocytes and no satisfactory explanation for this throughout the brain (including the brainstem). They are phenomenon is available (Clark and Tomlinson, more common in children who have had convulsions asso- 1949). ciated with CM than in other children (Thomas, 1971). A recent controversy has been the frequency with which As discussed in Section 5, molecular explanations for this cerebral oedema complicates CM, and how this swelling phenomenon have emerged recently. may contribute to mortality. The macroscopic features of oedema, such as increase in brain weight and compression 3.3.1. Vascular congestion. Cerebral capillaries and venules of cerebrospinal fluid (CSF) spaces (e.g., gyri, ventricles, distended with PRBCs are the microscopic hallmark of se- and pericisternal spaces), are observed in CM. The brains vere falciparum malaria (Fig. 2). In contrast to the periph- of African children (Walker et al., 1992; Edington and eral blood, all mature stages of the parasite are seen within Gilles, 1976) and adults (Janota and Doshi, 1979; Rigdon these vessels, both in adults (Spitz, 1946; MacPherson et and Fletcher, 1945) dying with CM frequently appeared al., 1985; Cropper, 1908; Aikawa et al., 1980), as well as in swollen with flattened gyri. Thomas reported macroscopic African children (Lemercier et al., 1966). The distended oedema and an increase in brain weight by more than 10% venules are more prominent in the gray matter, where they in 10/13 (77%) Ugandan children with CM (all of whom appear evenly distributed. Arteriolar dilatation has been were less than 5 years old), with white matter oedema more noted (Dudgeon and Clarke, 1917), although arteriolar marked in the heavier brains (Thomas, 1971). However, in constriction was also described in one study (Polder et al., a Nigerian study, only one child had a heavy brain (Walker 1991). et al., 1992), although severe oedema was observed in 5/7 (71%) children with CM. Ventricular compression (with- 3.3.2. Ring haemorrhages and granuloma formation. out accompanying oedema) was seen in 79% of children Ring haemorrhages contain a blocked central capillary, with P. falciparum sequestered in their brains (Lucas et al., with an agglutinated mass of PRBCs surrounded by brain 1996), although these children may not have had CM since tissue that is necrotic and contains demyelinated fibres clinical details were lacking. Another study reported the in- (Spitz, 1946; Boonpucknavig and Boonpucknavig, 1988) or cidence of cerebral oedema as similar in adult patients dy- a glial reaction. More recently, it has been suggested that ing of CM and those dying without cerebral disease (Ri- the “haemorrhages” are a specific response to blockage of a ganti et al., 1990), suggesting that it did not contribute to vessel by PRBCs and reperfusion, resulting in the concen- cerebral disease. tric pattern of haemorrhage surrounding a necrosed vessel It has often proved difficult to distinguish oedematous (Turner, 1997), and are, therefore, different from petechial changes seen at post-mortem from those occurring in life, haemorrhages seen after a variety of hypoxic insults, e.g., or those arising in the agonal state. Cerebral swelling may typhus fever (Vietze, 1978). Small malarial granulomata also occur because of increased blood volume in cerebral (Dürck’s nodules) are a distinctive pathological feature of tissues (Newton et al., 1991a) due to sequestration and mi- malaria (Dhayagude and Purandare, 1943; Thomson and 6 C. R. J. C. Newton and S. Krishna

these features are post-mortem artefacts. The ultrastructure of capillaries appears well preserved, with only a few vessels showing patchy degenerative changes (MacPherson et al., 1985). Most pathologists have commented on the lack of inflammatory cells both in adults (MacPherson et al., 1985; Fitz-Hugh, 1944; Dudgeon and Clarke, 1917) and African children (Thomas, 1971). However, an accumulation of macrophages (with active phagocytosis of PRBCs and pigment), neutrophils, and plasma cells has been seen in areas of extravasated PRBCs (Boonpucknavig and Boonpuck- navig, 1988), and has been interpreted as evidence of in- flammation. Some immunofluorescence studies have shown the deposition of falciparum antigens and anti-falciparum antibody on the cerebral vessels (Aikawa, 1988), often associated with haemorrhages in the white matter (Oo et al., 1987; Nagatake et al., 1992; Boonpucknavig and Boon- pucknavig, 1988), although other studies have not found evidence of such deposition (Turner et al., 1994; MacPher- son et al., 1985). An increase in leukocytes in the cerebrum was reported in a single case of CM, but the patient also had acquired immunodeficiency syndrome (Porta et al., 1993), which complicates interpretation of these findings.

3.3.4. Pigment. Malarial pigment or crystalline haemozoin is present in all erythrocytes that contain mature stages of parasites. Pigment accumulates in monocytes and macrophages in those tissues where there is significant parasitaemia, including in the peripheral circulation where its FIGURE 2. Cerebral capillary packed with parasitised red presence is a prognostic indicator (Metzger et al., 1995; Phu blood cells in an African child. Courtesy of Professor S. Lucas. et al., 1995). In a rodent model of infection, the longer the preceding malarial infection, the more pigment is seen in Annecke, 1926) associated with ring haemorrhages. They the reticuloendothelial system (Sullivan and Meshnick, are not found in patients who die shortly after the onset of 1996). The presence of pigment in the brain of patients symptoms (Dhayagude and Purandare, 1943), and probably with CM when PRBCs are no longer present implies that represent a more advanced stage of repair following haem- coma was prolonged, and death ensued even after tissue orrhage in which necrotic tissue has been replaced by neu- parasitaemia had begun to clear. roglial cells and microglial tissue (Spitz, 1946; Edington Pigment is composed of sheets of ␤-haematin, which de- and Gilles, 1976; Dhayagude and Purandare, 1943). rive from haem polymerisation after haemoglobin break- The presence of thrombotic lesions in CM is controver- down by parasites. Most of the metabolised haemoglobin in sial. Some authors have described thrombi in the white and the cell contributes to pigment formation (Morrison and gray matter (Dudgeon and Clarke, 1917; Aikawa et al., Jeskey, 1948). Chloroquine, and perhaps other antimalari- 1980), often associated with ring haemorrhages or granulo- als, were thought to interfere with the process of haem po- mata, although they do not contain PRBC (Spitz, 1946). In lymerisation by inhibiting a purported enzyme (Slater and contrast, other pathologists have commented on the lack of Cerami, 1992). More recently it has been shown that poly- evidence for organising thrombi in adult brains (Janota and merisation can proceed without protein present (Dorn et Doshi, 1979) or in children (Edington and Gilles, 1976). al., 1995), although parasite histidine-rich protein II accel- These haemorrhagic deposits do not contain platelets erates the process of haemozoin formation (Sullivan et al., (MacPherson et al., 1985) or fibrin (Dudgeon and Clarke, 1996). Pigment was believed to be relatively inert and non- 1917). Since there is no evidence of fibrinolysis in most toxic, but it can clearly modulate important biological ac- cases of severe malaria (Warrell, 1987), these lesions are tivities. The quantitation of pigment in human tissues unlikely to be true thrombi. eventually may prove to be as interesting as the quantitation of parasites. 3.3.3. Cellular elements. Capillary endothelial cells are described as swollen (Sein et al., 1993; Pongponratn et al., 3.4. Relationship between 1991; Fitz-Hugh, 1944), necrotic, or desquamated (Aikawa Sequestration and Cerebral Disease et al., 1980). These changes appear only in samples taken The degree and distribution of sequestered parasites varies more than 5 hr after death (Oo et al., 1987), suggesting that according to the species of malaria parasite and host. For Pathophysiology and Supportive Treatment of Severe Malaria 7 example, in the night monkey (Aotus trivirgatus), P. falci- served increase in expression of receptors in severe malaria parum sequestration affects in descending order myocar- may be related to induction by high levels of so-called pro- dium, adipose tissue, and skeletal muscle, with no parasites inflammatory cytokines (see Section 4.3), which are found visible in the brain (Miller, 1969). This difference in sein severe, particularly cerebral disease. These findings begin questration in animal models and humans is one of the key to provide a mechanistic explanation for the development factors that complicates the extrapolation of findings in an- of severe malaria, as well as linking the mechanical hypoth- imal models to understanding of human disease. esis with the cytokine hypothesis. A more recent larger The first semiquantitative assessment provided light and post-mortem study of 50 Vietnamese adults dying of severe electron-microscopic comparisons of samples obtained from and cerebral disease (Turner, 1997), found heterogeneity of Thai adults dying with strictly defined CM (n ϭ 7) and distribution of parasites within cerebral tissue. Patients dy- those with multi-organ involvement without predominant ing later in the course of infection did not show sequestra- cerebral symptoms (n ϭ 6) (MacPherson et al., 1985). An tion, although pigment was detectable. Some observers have advantage of this study was early collection of specimens reported that vessels congested with parasites are more (~90 min after death), although only needle biopsies were prominent in gray rather than white matter, and also that obtainable, which may have caused a sampling bias in the the cerebellum is affected more than other parts of the CNS results. Nevertheless, statistically significant differences (Sein et al., 1993). were observed between clinical groups. Patients with CM had more PRBCs and more vessels affected than NCM patients. The degree of sequestration was greatest in the 4. PATHOGENESIS—HOST FACTORS brain, followed by the heart and then the liver, lung, and 4.1. Mechanical Hypothesis kidney, all of which contained larger numbers of PRBCs The mechanical hypothesis assumes that organs will be af- than the blood. There was little evidence for haemorrhagic fected in proportion to the overall number of PRBCs se- or inflammatory exudation in tissues, immune complex questered in tissues, as well as their relative proportions deposition, or widespread thrombosis. Knobs on PRBCs ap- within different tissue capillary beds. Thus, CM arises be- peared to be the site of attachment to endothelial cells, cause of sequestration of PRBCs in cerebral capillaries and some of which were swollen. postcapillary venules, whereas patients who may have simi- A subsequent study also carried out in Thailand of 39 adults lar numbers of sequestered parasites in other tissues, but not confirmed that the percentage of PRBCs in organs was the brain, would not be expected to develop a full-blown higher in CM patients than in NCM patients (Pongponratn cerebral syndrome. Patients, of course, may have other signs et al., 1991). Again in the CM group, the sequestration of of severe disease, as many other organs, particularly the PRBCs in the brain was significantly higher than in other or- liver and spleen are frequently congested with PRBCs. gans, and there was a strong correlation between peripheral The mechanical hypothesis has been questioned by a parasitaemia and PRBC sequestration in brain (r ϭ 0.7031). number of investigators. The two main objections have This organ difference was not seen in the NCM group, al- been the lack of sequestration in some patients with cere- though some degree of PRBC sequestration was visible in bral symptoms associated with falciparum malaria, and the 50% of the NCM cases. Interestingly in this study, lung tis- fact that most patients with CM recover without evidence sue contained mononuclear cells rather than parasites. of ischaemic damage. In another study (Turner et al., 1994), based upon exper- The lack of massive intracerebral sequestration of para- imental observations on in vitro models of cytoadherence, a sites in some patients with CM can be rationalised in a detailed quantitative analysis of expression of endothelial number of ways (Sections 4.3–4.8) (Berendt et al., 1994). cell markers in tissues from adult patients (n ϭ 9) with fatal Since CM is a diagnosis of exclusion (Section 6.2), other CM were compared with control patients (n ϭ 9). There diagnoses (e.g., encephalitis) (White et al., 1992b) may not was increased expression of intercellular adhesion mole- have been excluded, and the coexistence of factors such as cule-1 (ICAM-1) and endothelial selectin (E-selectin) on hypoglycaemia may also cause impairment of conscious- cerebral vessels from malaria patients compared with con- ness. Furthermore, the mechanical hypothesis may be un- trols. CD36 and thrombospondin staining was sparse in all able to explain why all patients with malaria die. In some cerebral tissue. Parasites co-localised to areas expressing treated cases of malaria and coma, patients have died with- ICAM-1, CD36, and E-selectin, supporting the notion that out any parasites visualised on histological examination, al- these molecules may be important vascular ligands in vivo. though changes in host tissues may be similar to those seen There was additional evidence of endothelial cell activa- in patients dying with parasite-engorged capillaries (Dud- tion in malaria with increased expression of other markers geon and Clarke, 1917). An obvious explanation for these such as human leucocyte antigen Class II antigens, but findings is that parasites have responded successfully to these changes were not accompanied by increased local ac- treatment, but that pathophysiological processes have pro- cumulation of inflammatory cells. The most marked seques- gressed to an irreversible stage, and resulted in a fatal out- tration of parasites was in the brain, confirming findings come. There are, however, additional hypotheses that at- from previous studies. There were also large accumulations tempt to explain severe malaria and more particularly CM, of parasites and pigment in the spleen and liver. The ob- and these are discussed below. 8 C. R. J. C. Newton and S. Krishna

4.2. Toxin Hypothesis multiplication and synergising with other factors to pro- Gaskell and Miller (1920) were the first to propose that pa- duce gametocidal effects (de Naotunne et al., 1993). tients with malaria died from an overwhelming toxaemia. Parasite-derived material is responsible for TNF release The kallikrein-kinin system was shown to be important in from host cells (Bate and Kwiatkowski, 1994), and in one murine and simian models of Plasmodium infection (Mae- study, parasites varied in their ability to induce TNF release graith and Fletcher, 1972), but similar evidence in humans (Allan et al., 1993). This variation correlated with the is lacking (Warrell, 1987). Clark suggested alternative tox- source of the parasites; those obtained from patients with ins were responsible for the clinical features of severe ma- cerebral disease induced more TNF release ex vivo than laria, initially suggesting endotoxins (Clark, 1978), then re- those from patients with uncomplicated disease (Allan et active oxygen species (ROS) (Clark et al., 1986), cytokines al., 1995). There was considerable overlap in TNF releas- (Clark et al., 1987), and nitric oxide (NO) (Clark et al., ing activity between disease groups, and ~60% of parasites 1992). Although endotoxins are detectable in patients with isolated from patients grew sufficiently to allow assay (Al- falciparum malaria, they are associated with parasitaemia lan et al., 1995). It is possible that TNF may also be an ef- and leucocytosis, but not specifically with severe disease, fector arm of some parasite virulence factors if it has a cen- including CM (Usawattanakul et al., 1985; Aung-Kyaw- tral role in the pathophysiology of severe malaria. Zaw et al., 1988). ROS are discussed in Section 4.6. More The consensus that has emerged from a vast literature on recent studies are attempting to isolate and characterise experimental studies on the role of cytokines in malaria is malarial material, possibly a glycosylphosphatidylinositol that the timing and amounts of different cytokines that are moiety associated with malarial antigens, which can cause released, particularly TNF, may be important determinants release of cytokines from host cells (Schofield and Hackett, of subsequent pathophysiological events and perhaps mor- 1993) and up-regulate adhesion ligands (Schofield et al., tality. Thus, increased production of TNF early in malarial 1996) and NO (Tachado et al., 1996) (Section 4.5). infection may be protective, whereas prolonged, high TNF levels may be detrimental. This temporal relationship has been difficult to identify in humans. Furthermore, one of 4.3. Cytokine Hypothesis the difficulties in interpreting immunologically measured Clark and co-workers suggested that cytokines caused CM elevations in cytokines, particularly when receptor binding (Clark et al., 1991; reviewed in Clark and Rockett, 1994). can influence the unbound concentrations, is that the bio- They suggested that the host overproduction of “pro-inflam- logically active moieties (either circulating or local) may matory” cytokines, such as tumour necrosis factor (TNF) not be represented in these measurements. Thus, bioassays and interleukin (IL)-1, is capable of inducing cerebral syn- of circulating cytokines may be more relevant, but are more dromes in patients with malaria. Localisation of parasites to laborious to perform. certain tissues by sequestration would result in higher local In African children, circulating cytokines (particularly synthesis and release of these potent mediators, and cause TNF) are much higher in severe malaria than controls more marked local derangements in function and metabo- (Krishna et al., 1994b; Nyakundi et al., 1994; Shaffer et al., lism of tissues than elevations in circulating cytokines. 1991; Kwiatkowski et al., 1990; Kern et al., 1989; Grau et Clearly, in those patients in whom there is no significant al., 1989; Scuderi et al., 1986). In some studies, TNF levels sequestration, but cerebral symptoms are still present, other are correlated with parasitaemia (Nyakundi et al., 1994; mechanisms would have to be invoked to explain coma. Shaffer et al., 1991), but most studies detected higher levels These have not been defined further in the cytokine hy- in children with CM rather than NCM. Based on this cy- pothesis. tokine theory, studies have focused on TNF in CM to sub- P. falciparum-infected erythrocytes produce pyrogenic stantiate if elevations in TNF are deleterious. These eleva- material that triggers the release of TNF (and other cyto- tions in TNF may exacerbate the tendency to sequestration kines) from host mononuclear cells (Kwiatkowski et al., by the up-regulation of host ligand molecules responsible 1989). Merogony (soon after which parasite material is lib- for cytoadherence of parasites, especially ICAM-1, vascular erated in largest quantities into the circulation in vivo) pro- cell adhesion molecule-1 (VCAM-1), and E-selectin (see vokes the largest pulses of TNF release in vitro (Kwiat- Section 5.1), increasing mechanical obstruction of cerebral kowski, 1989). Elevations of TNF result in pyrexia in both or other blood vessels. As localised TNF production may falciparum (Kwiatkowski et al., 1989) and vivax malaria not be detected by plasma assays, but nevertheless may (Karunaweera et al., 1992). A monoclonal antibody against have important pathophysiological consequences, TNF TNF attenuates fever in children with CM, confirming the measurements have been carried out on CSF specimens importance of TNF in causing fever (Kwiatkowski et al., from patients with CM, but the results from large series are 1993). In the relatively benign vivax malaria (without cere- lacking. Increased TNF may also have consequences on the bral symptoms), very high (up to 3000 pg/mL) TNF levels metabolic status of patients (see Section 7.7) or in the have been measured transiently in serum (Karunaweera et pathophysiology of anaemia (Section 7.8). al., 1992), suggesting that such high systemic levels do not Host factors are also important in determining the TNF themselves cause serious disease. Furthermore, elevations in released in response to infection. A TNF-␣ promoter poly- TNF can exert antiparasitic effects by inhibiting parasite morphism has been shown to be associated with disease se- Pathophysiology and Supportive Treatment of Severe Malaria 9 verity in Gambian children (McGuire et al., 1994), with a first overcome the simplest interpretation of these findings homozygous polymorphism at Ϫ308 bp relative to the start that increased TNF production is an appropriate response of TNF transcription associated with a 7.7-fold increase in to malaria infection, and abrogation of this response, there- the relative risk of death or neurological sequelae in CM. fore, may be harmful. This interpretation is consistent also Other TNF promoter polymorphisms are now the subject of with the genetic studies reported on meningococcal disease detailed assessments in case control studies in malaria, as (see above). well as other diseases. For example, in meningococcal disease, relatively low TNF production assessed ex vivo in first-degree 4.4. Other Cytokines and relatives of patients was associated with a 10-fold increased Markers of Endothelial Cell Activation risk of fatal outcome (Westendorp et al., 1997). Interest- TNF is recognised as being one of the most important of ingly, this variation in TNF production was unrelated to the pro-inflammatory cytokines, and elevations in TNF polymorphisms at Ϫ308 bp and Ϫ238 bp in the promoter. concentrations trigger many secondary cytokine cascades in A further risk factor in meningococcal disease was in- severe malaria. There are associated increases in IL-1 and creased production of IL-10, which synergised with the ef- interferon-␥ concentrations, as well as increases in circulat- fects of low TNF production, and this was associated with a ing TNF receptors in acute infection. Circulating IL-6 20-fold increase in the risk of fatality. There is increased IL- (Molyneux et al., 1991) and IL-8 levels (Friedland et al., 10 production in severe malaria associated with elevations 1993) are increased in falciparum infection. Interestingly, in TNF levels (Anstey et al., 1996), which may contribute in relatively uncomplicated infections, IL-8 levels remained to severe disease. elevated for up to 4 weeks after the acute infection had An important test of the central role of TNF, therefore, been cured (Friedland et al., 1993). was undertaken using anti-TNF monoclonal antibodies as Both falciparum and vivax infections are associated with adjunctive therapy to either quinine or artemether (in a increases in circulating markers of endothelial cell activa- factorial design) in children with CM, and assessing mortal- tion (serum ICAM-1, serum VCAM-1, and serum ELAM-1), ity and neurological sequelae as endpoints (van Hensbroek and these markers were significantly higher in Gambian et al., 1996b). In this largest double-blind placebo con- children with severe falciparum malaria compared with un- trolled study of severe malaria in African children, 302 complicated infections (Jakobsen et al., 1994). These mark- children were assigned to receive a mouse monoclonal anti- ers were also elevated in nonimmune subjects, suggesting TNF antibody (B-C7) and 308 received placebo. Mortality that elevations are a consequence of cytokine activation by was similar in the two groups (19.9% vs. 20.8%), adjusted TNF. However, in a recent study in Gambian children, ma- odds ratio 0.9 (95% CI 0.57–1.42). However, residual neu- laria was associated with elevations in circulating ICAM-1 rological sequelae (assessed at 6 months) were detected in levels (which correlated with TNF and IL-1a levels), but el- 15 (6.8%) survivors in the study group, compared with 5 evations were not related to disease severity (McGuire et (2.2%) of those receiving placebo, adjusted odds ratio 3.35 al., 1996). Thrombomodulin, another marker of endothe- (95% CI 1.08–10.4). The biological effectiveness of the lial cell damage, was also significantly higher in nonim- anti-TNF antibody was confirmed in this study group by mune adults with severe malaria compared with those with the more rapid decline in rectal temperature in the treat- uncomplicated infections (Hemmer et al., 1994), and ele- ment compared with the placebo group. Four hours after vations were positively correlated with peripheral parasi- treatment, mean fall in temperature was 0.6ЊC in the BC-7 taemia and TNF-␣ levels. group compared with 0.2ЊC in placebo (P ϭ 0.02). Thus, in this rigorous test designed to establish a causal role in mortality due to increased TNF production, neutralisation of 4.5. Nitric Oxide circulating TNF did not reduce mortality, but did produce Since the cytokine hypothesis was first elaborated, NO has an appropriate antipyretic effect. been identified as a potential mediator for TNF action. NO The simplest explanation for these findings is that in- is a short-lived, highly reactive molecule that has a wide creased circulating TNF associated with CM is an epiphe- spectrum of biological activities. NO is involved in defence nomenon in the complex array of pathophysiological pro- by killing intracellular microorganisms (Vouldoukis et al., cesses and does not itself increase mortality. Indeed, 1995; Wei et al., 1995), in maintaining circulatory status by interfering with elevated TNF levels may be harmful. Al- its action on endothelial cells, and in neurotransmission ternative explanations have been suggested, including the (Vallance and Collier, 1994). It is produced both constitu- possibility that antibody prolonged the action of TNF on tively in certain tissues and in response to inflammatory endothelium by retaining it within the circulation. It may stimuli through the action of cytokines that up-regulate the also be that the antibody was administered too late in the synthesis of inducible NO synthase (iNOS or NOS2). disease process, when secondary cascades of mediators had Clark et al. (1992) proposed that TNF increased NO pro- already begun exerting their biological effects. In any case, duction, which caused the coma of CM. This NO is the increased incidence of neurological sequelae after anti- thought to be produced in cerebral endothelial cells and to TNF antibody administration suggests that future studies diffuse into brain tissue, interfering with neurotransmission designed to investigate this aspect of pathophysiology must (Clark et al., 1992). It may also cause neurological damage 10 C. R. J. C. Newton and S. Krishna and sequelae by forming peroxynitrite (Lipton et al., 1993), In contrast, a study from Papua New Guinea noted eleva- although this was not part of the original hypothesis. An al- tions in reactive nitrogen intermediates in children with ternative suggestion for the role of NO in malaria is that it CM compared with conscious patients and higher levels in may be important in host defence, particularly in intracel- fatal cases compared with survivors (Al Yaman et al., lular killing of parasites. This dual pathogenic or protective 1996). Other smaller studies have also noted an association role for NO is reminiscent of similar roles suggested for cy- between elevated levels of circulating nitrogen oxides and tokines, and neither is mutually exclusive. However, testing severe malaria. However, African studies have failed to de- these hypotheses is challenging because of the evanescent tect such an association (Agbenyega et al., 1997; Cot et al., nature of NO and the inapplicability of results from many 1994). In Ghanaian children, hyperlactataemia, a meta- animal studies to human infection. Instead, indirect mea- bolic marker of disease severity, was not correlated with sures of increased NO synthesis have been used as markers plasma nitrogen oxide levels, although it was clearly a prog- for the activity of iNOS. NO is metabolised to nitrate and nostic marker for a fatal outcome and was correlated with nitrite, and since these products are stable in plasma and depth of coma (Agbenyega et al., 1997). These findings are urine, they may be used as endpoints that reflect increased illustrated in Fig. 3. In order to assess the intracerebral syn- iNOS activity. The effects of diet and renal impairment on thesis of nitrogen oxides, which may arise through local dif- nitrate levels have to be taken into account when these infusion of NO if it is produced by cerebral endothelial cells, direct measures are employed (Al Yaman et al., 1997; Prada CSF measurements were performed in a subset of patients and Kremsner, 1997; Anstey et al., 1996). The evidence for with CM. Again, no elevations in nitrogen oxides were ob- an important role of reactive nitrogen intermediates in se- served, suggesting that coma in malaria may not be due to vere and CM vary with the populations that have been ex- derangements in NO metabolism. amined. Clearly, further studies are warranted to establish the im- In the largest study to date (Anstey et al., 1996), de- portance of changes in NO metabolism in malaria. How- creased NO synthesis was noted in Tanzanian children ever, the balance of evidence suggests that changes associ- with CM compared with patients with uncomplicated or ated with severe disease will be harder to establish than subclinical infections. Changes in plasma TNF profiles were they have been with cytokines, and the role of NO in consistent with previous reports that progressively higher pathophysiology and defence is more subtle than results levels were associated with increasing disease severity with from plasma and CSF measurements can disentangle. highest levels in fatal cases. Levels of the anti-inflammatory cytokine IL-10 were also increased in more severe disease, suggesting a mechanism by which NO synthesis could be 4.6. Reactive Oxygen Species suppressed in the patients with CM. In this study, increased Clark et al. (1983) proposed that ROS may play a role in NO synthesis was interpreted as being protective of the de- the pathogenesis of severe malaria. They based this hypoth- velopment of cerebral and fatal disease, and, therefore, im- esis on experimental studies in which they showed that the portant in defence rather than as a contributor to cerebral ROS were produced during malaria infections of mice pathology. Some support for this suggestion comes from (Stocker et al., 1984). They also showed that antioxidants work with desferrioxamine in children with CM, where NO prevent cerebral involvement (Thumwood et al., 1989), in- synthesis was increased in desferrioxamine recipients com- cluding the breakdown of the blood-brain barrier (BBB) pared with placebo recipients (Thuma et al., 1996). (Thumwood et al., 1988), and attenuated the fall in hae-

FIGURE3. Relationship between (A) admission plasma lactate concentration and (B) plasma concentration of nitrogen oxides and the Blantyre score. ᭺, survivors; ᭹, fatal cases. Reproduced from Agbenyega et al. (1997), with permission of the copyright holder, Royal Society of Tropical Medicine and Hygiene, London. Pathophysiology and Supportive Treatment of Severe Malaria 11 moglobin in mice (Clark and Hunt, 1983). Clark et al. tine, with hardly any accumulation within the brain. Thus, (1986) suggested that extravasated erythrocytes may lead to despite rising parasite counts, infected monkeys retain con- ROS generation, causing coma and damage to local tissues, sciousness until just before death (Anonymous, 1987). including the brain. In humans, there is some evidence that ROS play a role in the pathogenesis. Monocyte generation of ROS was in- 4.7.2. Blood-brain barrier in humans. The BBB is thought creased in adults with falciparum malaria, especially those to be grossly intact, both in African children (Badibanga et with severe disease (Descamps Latscha et al., 1987; Dubey al., 1986) and Thai adults (Warrell et al., 1986). In Zairian et al., 1991). In Indian children with CM, products of lipid children with CM, the concentrations of CSF albumin, im- peroxidation were found in the CSF (Das et al., 1991), but munoglobulin (Ig)G, and IgM were not raised compared the lack of an assessment of the BBB made this interpreta- with normal European children, and were significantly lower tion difficult. In African children, desferrioxamine, an ROS than another group of Zairian children, most of whom had scavenger, appears to reduce the duration of deep coma bacterial meningitis (Badibanga et al., 1986). These find- (Gordeuk et al., 1992), although it has not been shown to ings contrast with those in Thai adults with CM, in whom improve the outcome or decrease the incidence of neuro- 78% had evidence of intrathecal production of IgG (Chapel logical sequelae. Desferrioxamine may act by other mecha- et al., 1987), although the studies are not directly compara- nisms (Section 9.7). Thus, ROS appear to be generated in ble. The use of European controls may not be appropriate, acute malarial infections, but their role in the pathogenesis since Kenyan children with febrile convulsions have lower of complications is unclear. CSF protein concentrations than European children (Mac- Dougall, 1962). Furthermore, in 101 Kenyan children with CM, 25% had increased CSF protein concentrations 4.7. The Permeability Hypothesis (Ͼ0.45 g/L), 60% of those who developed permanent se- In the 1960s, Maegraith and colleagues proposed that CM quelae and all (n ϭ 5) those who died had high concentra- was caused by the stasis of blood secondary to an inflamma- tions (Newton, 1995). This would suggest that CSF find- tory state (Maegraith and Fletcher, 1972). They suggested ings in African children need to take into consideration that kinins increase the permeability of the BBB, causing that the BBB is probably impaired and that elevated CSF an efflux of plasma out of the vessels, thereby concentrating concentrations of substances may reflect plasma levels. the red blood cells within the cerebral vasculature and ulti- In the nonimmune adults, the data are difficult to inter- mately producing stasis of the blood. They based their hy- pret. In adult Thais with CM, the CSF levels of larger pro- pothesis on the cerebral oedema found at post mortem in teins such as ␣2-macroglobin were not raised, but the humans and studies of animals infected with malaria. CSF:serum albumin ratio was significantly higher than in British controls, although there was no significant difference in the ratios during coma compared with recovery nor 4.7.1. Experimental evidence. In the Rhesus monkey in fatal cases compared with those who survived (Warrell et (Macaca mulatta), Maegraith and colleagues showed an ef- al., 1986). Less radiolabelled albumin entered the CSF while flux of albumin from the blood into the CSF (Migasena and the patients were unconscious than when they recovered, Maegraith, 1968b), and since albumin is the protein re- but the BBB appeared to be more permeable than in Euro- sponsible for 75% of the oncotic pressure, they suggested peans. The increase in the CSF:serum ratio of 77bromide re- that the efflux of protein would draw water into the brain ported in this study was thought to be caused by hypoxia in- interstitium, causing cerebral oedema. However, they did terfering with the active transport of the bromide out of the not find a significant increase in CSF albumin, but showed CSF, but could be explained by an impairment of the BBB. that the protein was transported back into the blood at an Warrell and colleagues (1986) suggested that their find- almost equally fast rate (Migasena and Maegraith, 1968a). ings, together with normal lumbar puncture CSF pressures The increase in albumin transport did not appear to be as- in most patients and the lack of cerebral oedema on com- sociated with a raised intracranial pressure (ICP), although puterised tomography (CT) scans of adults who survive CM the pressure was not measured. Furthermore, they showed (Looareesuwan et al., 1983), implies that the BBB is grossly that chloroquine and hydrocorticosterone reduced the flow intact in adults. However, these studies do not exclude the of protein in both directions in the monkeys (Migasena and possibility that an abnormal BBB (permeable to smaller Maegraith, 1968b; Maegraith and Fletcher, 1972), suggest- molecules) and mild cerebral oedema, not detectable by ing that the breakdown of the BBB was caused by inflam- CT, may contribute to the pathophysiology of CM. Indeed, mation. These experiments led to the widespread use of ste- Davis et al. (1992) have shown that there is an increase in roids in human CM (Rees, 1982; Woodruff and Dickinson, permeability of the retinal vessels to fluorescein in adults 1968), an intervention subsequently shown to be inappro- with acute falciparum malaria, although this finding was as- priate (Warrell et al., 1982b). sociated with biochemical indices of severity, but not the There are important differences between monkeys dying presence of coma. Recent studies of Kenyan children with of knowlesi malaria and humans with CM. P. knowlesi se- severe malaria have not demonstrated an increased perme- questers preferentially within the liver and the small intes- ability of retinal vessels (Hero et al., 1997). 12 C. R. J. C. Newton and S. Krishna

4.8. The Immunological Hypothesis circulating immune complexes and the depletion of com- In the 1950s, immunological mechanisms were invoked to plement occurs in severe falciparum malaria, and in some explain the low incidence of CM in malnourished African studies, is more common in CM than NCM (Warrell, children since it was thought that they were immunodefi- 1987). As mentioned in Section 3.3.3, immune complex cient (Edington, 1967). The effectiveness of chloroquine in disposition on the cerebral vessels is not a consistent fea- treating CM was also partly attributed to its anti-inflamma- ture of CM (Boonpucknavig and Boonpucknavig, 1988; tory nature (Maegraith, 1969). Furthermore, immunologi- MacPherson et al., 1985). Anti-inflammatory agents, such cal mechanisms have been invoked to explain the rela- as steroids (Hoffman et al., 1988; Warrell et al., 1982a), Ig tively late peak incidence of CM in children living in (Taylor et al., 1992), or the immunosuppressive agent cy- endemic areas, despite the fact that these children have closporin A (Warrell et al., 1990), have not been shown to been exposed to malaria prior to the development of this be beneficial in humans with CM. complication. 5. PATHOGENESIS—PARASITE FACTORS 4.8.1. Experimental evidence. Experimental work in the The macroscopic consequences of falciparum infection have 1960s demonstrated that golden hamsters that had a neo- been recognised for over a century, but the cellular and mo- natal thymectomy (Wright, 1968) or those that were given lecular mechanisms that give rise to the fundamental patho- antithymocyte serum were resistant to P. berghei infections physiological process of sequestration have only begun to un- (Wright et al., 1971). During the last decade, Grau and col- ravel in the past few years. Infection with P. falciparum leagues (1989) have developed a murine model of CNS in- changes the host red cell in many and profound ways. One volvement using P. berghei ANKA in genetically susceptible of these is by altering the surface property of infected cells to mice. Initially, they showed that CD4ϩ T-cells were essen- make it “sticky.” This increased adhesiveness manifests itself tial for the development of neurovascular lesions (Grau et in ways that can be studied in laboratory models designed to al., 1986) and then went on to demonstrate that T-cell- represent the in vivo microenvironment of the PRBC. How- dependent macrophage activation led to the release of cy- ever, there is considerable functional redundancy in the par- tokines, including TNF, which mediated the development asite and host-encoded factors that mediate the attachment of these lesions (Grau et al., 1987a). Also, they showed that of PRBCs to other cells, making study of such mechanisms cyclosporin prevented the development of neurological com- not straightforward. Parasites can be selected to exhibit in- plications in mice (Grau et al., 1987b). There are funda- creased (or decreased) rosetting and cytoadherence pheno- mental differences between rodent models of CNS involve- types relatively easily in vitro, suggesting that similar pheno- ment and humans with CM. In rodents, monocytes are the typic plasticity may be important during the few cycles of principal cells that adhere to the endothelium, with little multiplication necessary to produce a severe infection in an sequestration of PRBC. Igs are consistently deposited on individual. Furthermore, demonstrating a potential patho- the cerebral endothelium in mice, but not in humans. Coma genic mechanism in vitro is only the first step in assessing its is not a prominent feature of murine malaria; rather, the eventual in vivo importance. The principal mechanisms neurological manifestations are limb paralysis, deviation of that are responsible for parasite virulence, the capacity to the head, ataxia, and convulsions (Grau et al., 1989). harm the host, are discussed in the following sections.

4.8.2. Clinical studies. Although immunological mecha- 5.1. Cytoadherence nisms are favoured mainly by experimentalists, they have Understanding of the mechanisms of cytoadherence of also been proposed by some investigators studying human PRBCs to endothelial cells has grown from observations CM. Malaria infections induce cellular and humoral im- made at post-mortem in malaria patients to the develop- mune responses in humans, and proliferative glomerulone- ment of in vitro models that have helped to identify host phritis, a manifestation of immunological disease, has been ligands, and eventually to the characterisation of responsi- reported in falciparum malaria (Boonpucknavig and Boon- ble parasite genes that are expressed at the erythrocyte sur- pucknavig, 1988). Based upon pathologic specimens of face. These more recent studies have reconciled the phe- adults dying in South America, Toro and Roman (1978) nomenon of antigenic variation with the heterogeneity of suggested that CM is a form of disseminated vasculomyelin- parasite-encoded adhesion molecules expressed in asexual opathy resulting from a CNS hyperergic reaction to a mas- stages of development, and now allow for more detailed sive antigenic challenge during a falciparum infection. The studies on their roles in pathogenesis of disease. perivascular inflammatory response, however, is not a fea- Cytoadherence describes the property of PRBCs to stick ture of CM, and the other features of a disseminated vascu- to nonerythrocytic cells or cell lines, and is the key patho- lomyelinopathy that are commonly seen in CM are not spe- physiological process that distinguishes falciparum from cific and could also arise from terminal hypoxia (Warrell, other human malarias. Cytoadherence depends on a high 1987). The more specific features of disseminated vasculo- molecular mass family of proteins encoded by var genes in myelinopathy were not found in all patients. An increase in P. falciparum. These var-encoded proteins [the P. falciparum Pathophysiology and Supportive Treatment of Severe Malaria 13 erythrocyte membrane protein-1 (Pfemp-1) family] are ex- tion 4.4) (Turner, 1997). Some of these molecules have pressed on the surface of the infected erythrocyte and vary been assayed in heterologous expression systems and confer in molecular mass (between 200,000 and 350,000 Mr) (Smith, upon expressing cells the ability specifically to attach to J. D. et al., 1995; Baruch et al., 1995) in parallel with parasite PRBCs. Many of these potential ligands are inducible or are strain-specific antigenic variation. They are also responsible up-regulated when endothelial cells are exposed to appro- for agglutination of infected cells when exposed to strain- priate cytokine stimuli (Section 4.3). The important specific antibodies. These immunological properties helped ligands in vivo in severe infection have only been studied in to confirm the relationship between the var gene family identi- a small number of cases, primarily in adult patients, so the fied during the hunt for the chloroquine-resistance gene and key mediators of cytoadherence in African children dying the properties of culture-adapted parasite clones varying in of severe malaria have yet to be identified. their cytoadherence and rosetting phenotypes, as well as in their variant surface antigens (Smith, J. D. et al., 1995). Interestingly, sequence analysis of var genes shows a 2-exon 5.2. Rosetting structure (with an approximately 1-kb intron), which, when Rosetting is the phenomenon observed when PRBCs con- translated, contains degenerately homologous domains, with taining relatively mature stages of parasites (those with pig- similarity to previously identified proteins implicated in the ment visible with light microscopy) become surrounded invasion of erythrocytes (Adams et al., 1992). These regions and bound to non-PRBCs (NPRBCs) (Udomsangpetch et of the var genes contain cystein-rich motifs similar to the al., 1989). Rosettes commonly consist of up to 10 NPRBCs, Duffy-binding protein of P. vivax responsible for red cell in- but may be much larger (“giant rosettes”). At first sight, the vasion and, therefore, have been called “Duffy binding like” formation of rosettes is puzzling because it may compete regions (Su et al., 1995). Thus, entry into red cells, antigenic with the process of cytoadherence, which sequesters para- variation, and evasion of host clearance by cytoadherence sites from the circulation (indeed, in laboratory studies, it is are probably all properties mediated by similar primary se- necessary to disrupt rosettes to measure binding to cell quences. Antigenic variation in a cloned parasite line is de- lines). However, rosetting can also be viewed as a mecha- tectable at a rate of ~2.4% per asexual generation (Roberts nism for increasing the chances of parasites multiplying by et al., 1992). It has been estimated that there are up to ~150 surrounding them with NPRBCs for reinvasion in the next copies of the var gene family in the haploid genome of falci- replicative cycle (Wahlgren et al., 1991). This view is not parum, accounting for between 2 and 6% of the whole para- shared by all investigators (Pasvol et al., 1995). site genome underlining the potential functional redundancy Rosetting has many features in common with cytoadher- and significance of these gene products for the parasite. ence, such as the parasite-stage specificity of expression of Pfemp-1 is localised primarily to parasite-induced “knobs” the phenotype, dependency on divalent cations, and sensi- on the surface of red cells. These knobs are small (60–100 tivity to trypsin. There are, however, also important differ- nm in diameter) electron-dense structures consisting of ences, such as disruption of rosettes (but not of cytoadher- submembranous accretions of parasite proteins, such as his- ence) by heparin and antibodies from patients with malaria tidine-rich protein-1 and the mature parasite-infected and the lack of positive correlation between the two adhe- erythrocyte surface antigen protein (Howard et al., 1990). sive properties in parasitised erythrocytes obtained from pa- Knobs form the supramolecular structures that allow appo- tients (Ho et al., 1991). Certain blood groups (A/AB or sition to host ligands on endothelial cells, and fine “pro- B/AB) favour rosetting over others (blood group O), and cesses” are visible on electron microscopy that span these this observation has been used to explain the relative pro- putative attachment sites. However, not all natural infec- tection of patients with blood group O from CM. The dogma tions are associated with knobs, and parasite clones isolated that only sequestering species of malaria (P. falciparum, in vitro can still bind without knobs, suggesting that knobs P. fragile, and P. chabaudi) form rosettes recently has been contribute to, but are not essential for, the process of cyto- challenged by the observations that P. vivax and P. ovale adherence (Petersen et al., 1989). PRBCs can bind to a va- can also form rosettes (Angus et al., 1996; Udomsangpetch riety of host cells, such as monocytes, neutrophils, and et al., 1995). Thus, the rosetting phenotype may be neces- platelets, confirming the phenotypic heterogeneity of cyto- sary, but is not sufficient, to cause severe malaria, and rela- adherence. A number of studies have attempted to identify tionship of this phenotype to cytoadherence is not clear. the host ligands both in in vitro models of cytoadherence In ex vivo assays, the mean number of parasites that formed and in vivo by immunocytochemical techniques. rosettes was significantly higher in isolates from Gambian Initially studies used endothelial cells and subsequently, children with CM (n ϭ 48, 31.7%) compared with those the amelanotic C32 melanoma cell line as an in vitro model from uncomplicated malaria (11.4%, P Ͻ 0.00000001), and for cytoadherence (Sharma, 1991). Studies using mono- the rosettes were much larger in the cerebral group (Treuti- clonal antibodies to a variety of ligands established that ger et al., 1992). Rosetting was not related to parasite den- CD36 (formally platelet glycoprotein 4) was a potential en- sity. Autologous (and some heterologous) testing of sera dothelial cell ligand for PRBC. Subsequently, a variety of suggested that disruption of rosettes by sera from CM pa- other host molecules were also identified as potential tients was less efficient than sera from uncomplicated cases. ligands, including ICAM-1, VCAM, and E-selectin (Sec- In a smaller subset of isolates, no correlation was observed 14 C. R. J. C. Newton and S. Krishna between disease severity and capacity to bind to melanoma 6. CLINICAL cells in vitro. Studies on adults in Thailand (Ho et al., 1991) DEFINITIONS AND PRESENTATIONS and children in Papua New Guinea have not found similar 6.1. Severe Malaria relationships between rosetting and disease severity (Al Ya- The parasite and host features that determine why a child man et al., 1995). The Thai study observed an inverse cor- who has been living in a malaria-endemic area and who is relation between rosetting frequency and cytoadherence to likely to have had extensive previous exposure to infection C32 melanoma cell lines, but may have been too small to progresses to develop severe disease are still largely unde- detect a difference in rosetting between groups of differing fined. The clinical manifestations of severe malaria are de- disease severity. In Papua New Guinea, although there was termined by the degree of immunity in the affected child, as no association with disease severity or inhibition of roset- well as genotypic predispositions. African children growing ting by autologous sera in uncomplicated cases, an associa- up in endemic areas are exposed to malaria from infancy, tion between blood group AB and rosetting was observed and in some areas, may receive up to 100 infected bites per (Al Yaman et al., 1995). Another study from Kenya has year (Mbogo et al., 1993). However, the relationship be- confirmed that increased rosetting is a feature of severe ma- tween malaria transmission and severe disease is complex, laria and is underrepresented in patients with blood group and the manifestations of severe disease vary with region O (Rowe et al., 1995). Haemoglobinopathies, such as sickle (Snow et al., 1997). Initially, the children are protected by cell disease and thalassaemia, may also impair rosetting maternal factors (e.g., transplacental acquisition of mater- ability, the latter because of microcytosis (Carlson et al., nal antibodies) and intrinsic factors (e.g., foetal haemoglo- 1994). A variety of sulphated glycoconjugates disrupt ro- bin). The protective effect of these factors, however, begins settes with varying degrees of effectiveness dependent on to wane during the first 6 months, so that severe disease the parasite isolates tested (Rogerson et al., 1994). Roset- leading to death is maximal between the ages of 1 and 4 ting is resistant to metabolic inhibitors (Carlson, 1993). years and is more likely in younger children. Thereafter, the The identification of parasite-encoded molecules responsi- children acquire immunity, so that adults who have lived ble for rosetting, following the recent successes with the continuously in endemic areas rarely develop severe dis- isolation of the var gene family, undoubtedly will help to ease. In contrast, children and adults who have not been clarify the sometimes conflicting experimental observations exposed to malaria previously can rapidly develop severe on this phenomenon. Both rosetting and cytoadherence al- disease after exposure. ter the rheological characteristics of PRBC in laboratory Severe malaria is a spectrum of clinical syndromes uni- models of flow (Kaul et al., 1991). These alterations are dis- fied by the single causative organism P. falciparum. It can be cussed in the next section. operationally defined as any malaria syndrome that is associated with a high mortality (Ͼ5%), even after appropriate hospital treatment. The clinical spectrum of severe malaria is 5.3. Rheology different in African children growing up in malaria-endemic Early studies of P. knowlesi PRBCs showed infection in- areas compared with nonimmune adults (Table 1). The rea- creased resistance to flow through 5-␮ polycarbonate sons for these differences are not clear, although they appear sieves, and obstructed pores at high parasitaemia. Infection to be related to exposure to malaria rather than age, since was also associated with a decrease in red cell deformability, nonimmune children have similar features to nonimmune and these changes were suggested to contribute to the de- adults. The clinical picture, however, is changing, as in velopment of microcirculatory obstruction (Miller et al., Southeast Asia jaundice is seen less frequently (Lalloo et 1971). Subsequently, more sophisticated studies have con- al., 1996). This review largely concentrates on the clinical firmed that there is a stage-dependent decrease in the de- picture of African children since most of the recent work formability of red cells as P. falciparum matures (Nash et al., has been conducted on this group, who bear the brunt of the 1989), and that mature parasites require correspondingly disease worldwide. Two studies have reported the admission of larger pressures (4- to 6-fold, compared with controls) to al- children to units in North America, but the number of chil- low entry of PRBCs into small (3 ␮) capillaries. These dren with severe disease was too small to make meaningful changes could reduce the circulatory flow in downstream comparisons (McCaslin et al., 1994; Lynk and Gold, 1989). (postcapillary) venules, and contribute to other pathophys- In African children, the commonest presenting syn- iologically important processes such as cytoadherence. The dromes associated with significant mortality are CM; recur- molecular basis for adhesion at physiological flow rates has rent convulsions; metabolic dysfunction, which manifests also been studied in cultured cell receptors (Wick and itself as hypoglycaemia or lactic acidosis; or symptomatic Louis, 1991; Nash et al., 1992), but relating changes ex vivo anaemia. Children with the greatest risk of mortality can be to patient’s clinical status is not straightforward (Cooke et identified as those with impaired consciousness or respira- al., 1993, 1995). Recently, reduction in red cell deformabil- tory distress. Although severe anaemia is a common cause ity assessed using ektacytometry has been shown to be of admission, if blood transfusions are given quickly to strongly associated with a fatal outcome in severe malaria those with cardiorespiratory distress, the mortality is low (Dondorp et al., 1997). (Fig. 4). The age distribution of these syndromes is differ- Pathophysiology and Supportive Treatment of Severe Malaria 15

TABLE 1. WHO Criteria for Severe Falciparum Malaria: Comparison of African Children and Papua New Guinean Adults Admitted to Hospital African children Nonimmune adults (Marsh et al., 1995) (Lalloo et al., 1996) Prevalence Mortality Prevalence Mortality

Defining criteria Coma 10.0 16.8 17.1 41.7 Severe anaemia 17.6 4.7 10 0 Respiratory distress 13.7 13.9 — — Hypoglycaemia 13.2 21.7 5.7 75 Circulatory collapse 0.4 71.4 0 — Renal failure 0.1 0 22.9 37.5 Spontaneous bleeding 0.1 0 0.1 100 Haemoglobinaemia 0.1 50.0 0.1 0 Acidosis 63.6 21.4 NA NA Repeated convulsions 18.3 6.8 0.3 0 Supporting criteria Impaired consciousness 8.2 6.0 37.1 11.5 Jaundice 4.7 11.9 45.7 25 Prostration 12.2 5.2 — — Hyperpyrexia 10.6 1.6 20 7.1 Hyperparasitaemia 8.9 4.3 40 28.6

NA, not applicable.

ent, so that severe anaemia affects younger children than icant overlap (Fig. 4), and CM is frequently complicated by CM, with respiratory distress overlapping both syndromes lactic acidosis and/or hypoglycaemia. In cases where lactate (Fig. 5). These syndromes are discussed in greater detail in levels or acid-base status cannot be assessed, clinical signs Sections 6.2–6.4. of respiratory distress suggest severe disease and are often Children who present with severe malaria often have rel- associated with underlying acidosis. These clinical corre- atively short histories of illness, emphasising the rapidity with which illness may progress. Parents accompanying their children to hospital typically report fever beginning 1–3 days prior to admission, with neurological manifestations often beginning within 12 hr of admission (Musoke, 1966; Schmutzhard and Gerstenbrand, 1984; Waller et al., 1995; Gordeuk et al., 1992; Molyneux et al., 1989b). Chil- dren with severe anaemia may present with longer histories, although acute decompensation, manifesting as respiratory distress, often precipitates admission to hospital. Most deaths from severe malaria occur within 24 hr of starting treatment (Waller et al., 1995; Krishna et al., 1994b; Walker et al., 1992; Newton et al., 1991a; Marsh et al., 1995; Molyneux et al., 1989b) (Fig. 6), and most of those who survive make full recoveries within 48 hr of starting treatment (Krishna et al., 1994b; Waller et al., 1995; Walker et al., 1992; Gordeuk et al., 1992; Molyneux et al., 1989b). The syndrome of CM is clinically discrete and relatively easily identified and, therefore, has been the subject of most of the published studies on severe malaria. However, it is important to appreciate that CM is one of the commonest manifestations of severe disease, but that severe malaria is FIGURE 4. Prevalence, overlap, and mortality for the major not synonymous with CM. It is only within the past few clinical syndromes of severe malaria in 1844 children presenting to Kilifi District Hospital, Kenya. Mortality is given as a years that metabolic derangements have been identified percentage of the total number. Adapted from Marsh et al. and subsequently incorporated into the definition of severe (1995), with permission of the copyright holder, Massachusetts disease. Clinical syndromes of severe malaria exhibit signif- Medical Society, Boston. 16 C. R. J. C. Newton and S. Krishna

FIGURE 5. Ages of 1844 Ken- yan children admitted with CM, respiratory distress, and severe anaemia to Kilifi District Hos- pital, Kenya. Data from Marsh et al. (1995).

lates of severe malaria have been recognised for decades such as hypoglycaemia. Whilst there is no single clinical or (Guignard, 1965), although they have been more formally laboratory marker that adequately identifies all patients assessed recently (Waller et al., 1995; Marsh et al., 1995). with severe disease, most high-risk patients are identified Multi-organ involvement is relatively uncommon in using a combination of clinical and laboratory criteria. children compared with adults with severe disease. In our These criteria are summarised in the following sections. experience, none have required renal replacement therapy for significant impairment, and severe jaundice is uncom- 6.2. Cerebral Malaria mon unless it is associated with haemolysis. Pulmonary in- CM is a syndrome of impairment of consciousness associ- volvement has not been studied in detail in children, and the ated with a falciparum malaria infection. Impairment of distinction between secondary infection and oedema requires consciousness caused by other species of malaria have been more detailed examination. In a study of 180 children, 25 described, but in these case reports, other causes of an en- of whom died of severe malaria, the principal clinical causes cephalopathy have not been excluded and rigorous post- of death were identified and summarised in Fig. 7. mortem studies are lacking. Recognition of severe malaria is, therefore, of paramount importance in determining subsequent management deci- sions such as the use of a loading dose of quinine and the 6.2.1. Definition of cerebral malaria. There are no patho- frequency of post-admission monitoring for complications gnomonic pathological features that differentiate CM from

FIGURE 6. Cumulative percentage survival of 2155 children admitted to three African hospitals with severe falciparum malaria (Molyneux et al., 1989b; Marsh et al., 1995; Waller et al., 1995). Pathophysiology and Supportive Treatment of Severe Malaria 17

NCM (Section 3). Furthermore, consciousness can be de- seizure is common, but children with malaria often have a pressed by a variety of pathophysiological processes, includ- prolonged post-ictal state after long or multiple seizures ing seizures [either nonconvulsive status or post-ictal states (Crawley et al., 1996). More than one seizure in a child with (Crawley et al., 1996; Kirkham, 1991)], metabolic derange- falciparum infection is indicative of significant cerebral in- ments [e.g., severe acidosis (Allen et al., 1996a)], and by volvement, as these children frequently progress to coma. the systemic effects of the infection [e.g., fever or hy- Asexual falciparum parasitaemia is a sine qua non of se- ponatraemia (Arieff and Griggs, 1994)]. To exclude sys- vere infection, since the liver stages and gametocytes do temic causes of impaired consciousness and make studies of not contribute to symptoms. However, this definition does CM comparable, Warrell and colleagues (1982a) originally not include cases of blood slide-negative CM (Walker et proposed a strict definition of CM based upon the response al., 1992; Gopinathan and Subramanian, 1986; Edington to a painful stimulus. This definition, slightly modified by a and Gilles, 1976), i.e., malaria patients who have profound panel of experts at the World Health Organisation (WHO) CNS disturbance, without asexual parasites detectable in (Warrell et al., 1990), is now widely accepted. the peripheral blood, but sequestered parasites found in the CM exists in a patient who: brain at post-mortem. The commonest cause of slide-negative falciparum malaria is pretreatment with antimalarials • is unable to localise a painful stimulus (White et al., 1992b), which remove circulating evidence • has peripheral asexual falciparum parasitaemia of infection, but not the parasites sequestered in tissues. In • has no other causes of an encephalopathy these cases, the diagnosis may be suspected by rapid re- The inability to localise a painful stimulus was chosen for sponse to continuing antimalarial therapy, although this is two reasons. First, because it is associated with sufficiently relatively imprecise, or confirmed by finding sequestered deep coma as to exclude systemic effects of infection, and parasites in the brain at post-mortem. It is likely that as the second, because it is a sign that can be reliably elicited use of artemisinin derivatives becomes more widespread, (Newton et al., 1997a). Lesser degrees of CNS dysfunction slide-negative malaria will be observed more commonly, since (confusion, stupor) are frequently caused by falciparum in- this class of antimalarial clears circulating parasites rapidly. fection and should also be managed as severe malaria. To Since a considerable proportion (up to 90%) of children exclude patients with uncomplicated falciparum infection living in endemic areas are infected with P. falciparum, but and a complicating febrile seizure who are unrousable post- are asymptomatic, children admitted with other encephalo- ictally, the child should be reassessed at least 30 min after a pathies, which require exclusion (e.g., encephalitis or Reye’s seizure. Warrell and colleagues originally stipulated that syndrome), may have a concomitant peripheral parasi- this assessment period should be after 6 hr, based on experi- taemia and thus, be misdiagnosed as having CM. This has ence with adults, but this has now been shortened to 30 led to inclusion of quantitative assessment of parasitaemia min since children often regain consciousness after febrile to establish the probability of malaria as a cause of symp- seizures within this time (Warrell, 1989). Our own observa- toms in a patient. The parasitaemia cutoffs that have been tions also suggest that recovery 30 min following a febrile used range from 2000 to 20,000/␮L, depending on the geographical area and the criteria used for establishing positive predictive values for infection. However, most of these studies are community-based case-control studies and, therefore, are inapplicable to children who present to hospital with signs of moderate or severe malaria (as opposed to children presenting with uncomplicated infection). This may be particularly important in complicated falciparum malaria when a low peripheral parasite count may not reflect the parasites sequestered in the deep vascular beds. Thus, the absolute parasitaemia per se is not particularly useful in discriminating severe from uncomplicated or moderate malaria, and any patient who presents with symptoms of severe disease and the presence of asexual P. falciparum parasitaemia, therefore, must be managed as a case of severe malaria. The stages of parasite development on a peripheral FIGURE 7. Causes of death in 25 Gambian children with severe malaria. The causes of death were classified according to blood film may be a more important parameter to assess in the following criteria: CM, highest BCS р2, no biochemical severe infections (Cropper, 1908). The importance of de- evidence of lactic acidosis, pulmonary complications, or sepsis; tecting these mature stages recently has been emphasised lactic acidosis, plasma lactate у5 mmol/L, BCS Ͼ2 after admis- by studies that show that Ͼ20% mature stages in the pression, and no evidence of pulmonary complications or sepsis; pul- ence of a high parasitaemia (pigmented trophozoites or monary complications, chest examination revealed signs compatible with infection or oedema; sepsis, proven by blood later) is associated with a significantly higher mortality culture. Reproduced from Waller et al. (1995), with permission than patients who do not have such a high proportion of of the copyright holder, University of Chicago Press, Chicago. late stages (Silamut and White, 1993). 18 C. R. J. C. Newton and S. Krishna

The exclusion of other causes of coma is determined by a reflexes (Plum and Posner, 1980) or seizures (Crawley et al., careful history and by the available investigative facilities. 1996). Likewise, withdrawal from a painful stimulus may be In most hospitals in malaria-endemic areas, this entails the a spinal reflex requiring little cortical involvement and exclusion of other CNS infections (by obtaining clear and thus, is not a useful sign in assessing the depth of coma. Fur- culture-negative CSF) and hypoglycaemia (either by per- thermore, there are considerable differences between clini- forming a blood glucose estimation or observing the clinical cians in assessing this scale, particularly the verbal compo- response to glucose administration). The CSF needs to be nent, which will limit its use (Newton et al., 1997a). examined in all children presenting with CM, since this Despite these reservations, however, this scale appears to condition cannot be differentiated clinically from bacterial assess the overall degree of compromise due to falciparum meningitis (Wright et al., 1993), but the timing of the lum- malaria, and can be used to predict mortality, rather than bar puncture is controversial (Greenwood, 1991; White, the neurological complications (Newton et al., 1997a). The 1991; Kirkham et al., 1991; Kwiatkowski et al., 1991; BCS is used by many researchers who define CM in chil- Coulter, 1991). If examination is delayed or not possible, dren as a summated score of 2 or less: a total that is mostly then children should empirically be managed with both an- derived from a motor and verbal score of 1 each and thus, is timalarial and antibacterial therapy (to cover the possibility equivalent to the strict WHO definition. Other researchers of meningitis) and the diagnosis confirmed when permissi- include children with scores of 4 or less to define CM; how- ble. Drug toxicity may confuse the clinical picture of CM , ever, this score would not exclude systemic effects and is but clearly the presence of parasites dictates immediate an- not directly comparable with the strict WHO definition. timalarial treatment. Other causes of encephalopathy, e.g., These definitions do not account for the fact that very encephalitis without a pleocytosis or Reye’s syndrome, can- young children may not have acquired the ability to loca- not be confidently excluded. lise pain and thus, may not satisfy the formal definition for coma, but nonetheless may be comatose. Thus, the defini- 6.2.2. Coma scores. The inability to localise a painful tion of CM should be restricted to children over 8 months, stimulus identifies patients who are deeply unconscious, but until more appropriate criteria are found for younger chil- not those with more subtle neurological impairments. A dren (Newton et al., 1997a). Overall, the WHO criteria of- summated Glasgow coma score less than 8 (Table 2) has fer a pragmatic definition that can be used for research and been proposed as part of the definition of CM, instead of in the management of children older than 8 months. Coma the inability to localise pain (Leaver et al., 1990). How- scores may be useful in categorising the severity of CM, but ever, this criterion is not applicable to children since chil- their use as a criterion for diagnosis needs to be considered dren do not have the same range of summated scores (New- carefully. ton et al., 1990) and thus, it has not been widely accepted. Molyneux and Taylor developed the Blantyre coma score (BCS) for the assessment of young Malawian children with 6.2.3. Clinical presentation of cerebral malaria in African CM (Molyneux et al., 1989b). The BCS is based upon the children. The clinical presentation of CM has features of a Glasgow coma scale (Table 2), but measures different re- diffuse metabolic encephalopathy, as described by Plum and sponses. In particular, the eye responses do not take into Posner (1980). The pathological basis of this picture has consideration that some children may be blind and that di- been discussed in Section 3 and its pathophysiology is dis- rected eye responses can be caused by primitive subcortical cussed in Section 7.

TABLE 2. Blantyre and Glasgow Coma Scales Blantyre coma scale Glasgow coma scale (Molyneux et al., 1989a) (Teasdale and Jennett, 1974)

Verbal 2. Appropriate cry 5. Able to give name and age 1. Inappropriate cry or moan 4. Recognisable and relevant words 0. No cry 3. Incomprehensible, but complex vocalisation 2. Cry or grunt 1. No response Motor 2. Localises pain 6. Obeys commands 1. Withdrawal from pain 5. Localises pain 0. Nonspecific or no response to pain 4. Withdrawal from pain 3. Flexes to pain 2. Extends to pain 1. No response Eye 1. Directed eye movements 4. Spontaneous eye opening 0. Not directed 3. Opens eyes to voice 2. Opens eyes to pain 1. No eye opening Pathophysiology and Supportive Treatment of Severe Malaria 19

African children usually present in coma with an almost sions, and concentrations of lactate Ͼ4.5 mmol/L predict a invariable 2- to 3-day history of fever (Rey et al., 1966; particularly high mortality. In comparison with acute bac- Molyneux et al., 1989b; Maguire, 1983; Ikpatt et al., 1990; terial meningitis, the CSF glucose concentration is higher Commey et al., 1980). Alterations in conscious level range and the protein lower (Table 4). The CSF in CM should be from drowsiness and/or convulsions to deep coma and clear (no more than 100 lymphocytes/mm3) and 25% have brainstem abnormalities (Musoke, 1966; Schmutzhard and elevated protein levels. There are no elevations in CSF cre- Gerstenbrand, 1984; Waller et al., 1995; Newton et al., atinine phosphokinase levels in cerebral disease (Miller et 1991a; Walker et al., 1992; Molyneux et al., 1989b). Fifty al., 1989b). These findings are discussed more fully in Sec- to 80% of children have a history of convulsions (Schmutz- tion 4.5. hard and Gerstenbrand, 1984; Molyneux et al., 1989b; In African research settings, the mortality rate for strictly Maguire, 1983; Ikpatt et al., 1990; Commey et al., 1980), defined CM is 18.6% (95% CI 16.3–21.0) (see Table 3). one of which often precipitates coma (White and Looaree- There are no series large enough to give a true estimate of suwan, 1987; Warrell et al., 1990; Edington and Gilles, mortality in children treated in developed countries. 1976). Seizures are witnessed at presentation in up to 50– 80% of African children with CM (Table 3). If seizures per- 6.2.4. Neurological sequelae. Neurological damage has sist after admission and particularly if they are prolonged or only been recently recognised as a common sequela of CM resistant to anticonvulsants, they are associated with the in African children (Brewster et al., 1990). In nonimmune development of neurological sequelae (Bondi, 1992) or adults, neurological sequelae are uncommon, although a death (Kirkham et al., 1994; Brewster et al., 1990; Moly- variety of postmalarial syndromes have been described neux et al., 1989b). Occasionally, we have also observed (Newton and Warrell, 1998). Although most neurological both children and adults presenting with bizarre behavioural sequelae follow an episode of CM, neurological deficits can disturbances, such as confusion, clouding of consciousness, be found in children with seizures without sustained loss of and stereotyped behaviour, such as tongue protrusion. Of- consciousness (C. R. J. C. Newton, personal observation). ten, these symptoms progress rapidly to coma, soon after The studies of neurological sequelae are often difficult to the initiation of antimalarial therapy. interpret since the definition of the severe falciparum ma- Children with CM frequently develop abnormal postures laria, the thoroughness of the examinations, and the extent (decorticate rigidity, decerebrate rigidity, opisthotonos), of follow-up varies considerably, so that the incidence also pupillary changes, absent corneal reflexes, abnormal respi- varies considerably (Table 5). Recent studies of African ratory rhythms (Kussmaul, Cheyne-Stokes, periodic ap- children with CM (van Hensbroek et al., 1996a; Marsh et noea), and gaze abnormalities (eyes wide open, conjugate al., 1995; Bondi, 1992; Brewster et al., 1990; Molyneux et gaze deviation, nystagmus) (Newton et al., 1998). Signs of al., 1989a), which used similar criteria for the definition of frontal lobe release are often seen. Other brainstem signs, CM, report an incidence of neurological sequelae rate to be including dysconjugate and conjugate eye signs, and decer- 10.9% (95% CI 8.3–13.5% of survivors). ebrate posturing movements (all of which may represent 6.2.4.1. Factors predisposing to sequelae. In African chil- seizure activity), occur in about one-third of children dren, the development of sequelae is associated with pro- (Schmutzhard and Gerstenbrand, 1984; Molyneux et al., tracted seizures (van Hensbroek et al., 1996a; Brewster et 1989b; Gelfand, 1973). Retinal haemorrhages are present al., 1990; Bondi, 1992), deep coma (van Hensbroek et al., in 6–36% of African children during admission (Newton et 1997), prolonged coma (Sanohko et al., 1968; Brewster et al., 1991b; Molyneux et al., 1989b; Kayembe et al., 1980). al., 1990; Bondi, 1992; van Hensbroek et al., 1997), and Haemorrhages are associated with a high mortality in adults anaemia in some studies (Brewster et al., 1990), but not in (Looareesuwan et al., 1983), but in children, they do not ap- others (Bondi, 1992). Other smaller studies have shown an pear to have prognostic significance (Kayembe et al., 1980). association between the development of sequelae and In both groups, they resolve without long-term visual de- pathophysiological processes such as raised ICP (Newton et fects (Kayembe et al., 1980; Bell et al., 1976). Papilloedema al., 1997b). Most of these factors are also associated with is present in between 2 and 12.5% of children (Newton et death, and may simply reflect the severity of the underlying al., 1991a; Molyneux et al., 1989a; Kayembe et al., 1980), insult, rather than a specific neuropathogenic process. The and was associated with increased opening pressures on diverse pattern of neurological damage detected clinically CSF examination (Lewallen et al., 1993). Papilloedema and and by CT suggests that there are likely to be a variety, retinal oedema outside the vascular arcades were both inde- probably interacting, of mechanisms responsible for neuro- pendent indicators of fatal outcome in children. logical damage. CSF examination is useful not only in excluding infections that may mimic CM, but also in providing measure- 6.2.4.2. Patterns of sequelae. In the studies of African ments of CSF lactate concentrations, which are one of the children, there is a difference between the sequelae found best independent prognostic indicators of fatality (Krishna on discharge and those that persist (Table 5). About 50– et al., 1994b; White et al., 1985, 1987a). In children, CSF 84% of children recovered completely, usually within the lactate (and glucose) concentrations are correlated with first 6 months after discharge, and 9–55% had recovered 18 plasma levels, are not associated with a history of convul- months after discharge, although this may be an overesti- 20 C. R. J. C. Newton and S. Krishna 24 hr (%) Mortality Ͻ (%) 28 64 Mortality Neurological sequelae (%) 24 3 20 Ͻ of coma (hr) Mean duration C033 Q3440 ϩ ϩ (%) Treatment Seizures L) ␮ (/ Parasite count (%) Hypoglycaemia (years) Mean age 29 3 Q 36.7 7 21 of cases Number , 1990 The Gambia 24 C 25 1962 Senegal 88 Q or C 14 25 , 1991 Ghana 20 4.8 C 10 ., 1993 The Gambia 41 4 7 40,000 C 24 12 29 58 1989b Malawi 131 3.6 37 147,910–229,087 82 Q 31 9 15 70 1969 Senegal 235 3 Q/C 5 9 , 1986 Angola 254 3 3500 20 Q/C 27 , 1991 Ghana 113 5 2 75 C 34 8 4 , 1980 Ghana 43 4.5 100 C 36 5 5 , 1977 Zaire 121 3.7 87 C , 1991a Kenya 61 3.25 28 84 Q 20 100 ., 1968 Senegal 34 3 Q/C 21 9 et al et al. ., 1992 Zambia 83 2.8 45,000 81 Q 68 2 19 , 1990 The Gambia 36 1.8 11 19,960 Q 0 30 ., 1990 The Gambia 377 18 15 C 9 16 et al. , 1993 Nigeria 25 3 AM 40.1 12 12 , 1993 Liberia 97 2 66 Q 16 et al., ., 1993 Nigeria 31 3.3 22.6 , 1993 Congo 170 Q 9 15 64 ., 1992, 1993 Nigeria Malawi 61 65 5.7 2.8 26 15 158,489 88 93 Q or AM C 14(Q), 8(AM) 11 51 11 16 18 7 , 1988 India 40 60 Q , 1987 Congo 140 ., 1988 Malawi 95 3.3 20 250,000 78 Q 36 11 11 et al. ., 1991 Kenya 49 3.4 6 297,000 94 Q 15 10 16 , 1990 Tanzania 110 3.8 210,000 C 13 25 67 , 1982 PNG 68 5.1 Q 4 6 et al., et al., ., 1990 Nigeria 75 4.8 57 C/Q 0 20 et al. , 1989 Malawi 55 5.3 18 273,400 95 Q 20 4 18 et al. et al et al et al et al. et al. et al et al. ., 1966 Senegal 227 4.3 65 Q 31 et al. et al et al. et al. et al et al et al. et al et al. et al. et al. et al. et al. et al Gerstenbrand, 1984 Tanzania 66 95 C 17 18 Kwiatkowski Akpede Wright Walker Pasvol Trape Taylor Molyneux Thapa Newton Bondi, 1992Taylor Carme Nigeria 78 5.2 23 67 C 68 14 21 Wolf-Gould Gordeuk Walker Rothe, 1956Armengaud Kenya 97Bernadino Q, C 22 Schmutzhard and Ahmad and River, 1986 India 30 7 C Grau Brewster Ikpatt Kawo TABLE 3. Summary of Features CM in Children Author, year Country Carlson Kwiatkowski Neequaye Guignard, 1965Rey Musoke, 1966Sanohko Lemercier Madagascar Uganda 60 20 77 C C 34 0 38 5 30 83 Berner, 1975Omanga Malawi 39 2 C 3 28 Commey Stace Pathophysiology and Supportive Treatment of Severe Malaria 21 70 70 ف ف 16.8 25.221 21.5 20.5 2 2 111 29 0 Ͼ P 6 7.7 0 ϩ 56.8 32.3 10.9 18.6 1 3.8 19.9 39,198 67 8974 NRD 12 26 5.6 26.9 65.4 Q 288 4 26.3 57,106 38.5 AM 26 4677 3319 2079 1709 3133 1401 3199 4439 . (1995). et al , 1994 Thailand 96 6.5 55 Q 22 ϭ , 1994 Mozambique 19 2.9 26 17,400 84 Q 11 100 , 1995 Nigeria 24 11,379 RD 46 et al. , 1994 Burundi 31 28.6 11,920 Q 22 , 1995 Zambia 129 2.9 13 68,700 93 ., 1994 Kenya 14 3.2 28 1200–1,108,800 93 Q or AM ., 1995 Burundi 30 5.1 26.6 73.3 Q 46 8 16.7 , 1996b PNG 75 3.7 10 6760 53 Q 6.1 8.5 83 ., 1995 Kenya 185 et al. et al. et al. et al et al. et al et al 0.046. et al. ϭ

Geometric mean. P Calculations of weighted means and confidence intervals were carried out in STATA without including ranges. C, chloroquine; Q, quinine; AM, artemether; PNG, Papua New Guinea; RD, respiratory distress; NRD, no P, pentoxifylline. Modified from Waller 1 2 ϭ Weighted 95% CI 3.5–4.2 16.1–23.6 18,857–81,486 46.1–67.4 23.4–41.1 8.3–13.5 16.3–21 Newton Solomon Marsh di Perri Brown, 1995Mabeza PNG 17 5 0 88 Q 6 18 Wattanagoon Olumese van Hensbroek, 1996 The Gambia 288 3.8 22.3 55,471 28.1 Q 20 Deloron Allen Weighted mean n 22 C. R. J. C. Newton and S. Krishna

TABLE 4. CSF Findings in CM and Acute Bacterial Meningitis CM Acute bacterial meningitis (White et al., 1987a; Newton, 1995) (Eross et al., 1981; Bonadio, 1992)

Appearance Clear Cloudy White cell count (ϫ109/L) Ͻ1001 Ͼ100 neutrophils Ͼ1 ϫ 109/L NR 99% Opening CSF pressure (mm Hg) 1.8–7.1 15 (4–70) Glucose (mmol/L) 2.0–5.6 Ͻ1–3.8 Ͻ2.2 mmol/L 8% 81% Lactate (mmol/L) 2.4–Ͼ10 3.9–Ͼ10.0 Protein (g/L) 0.39 (0–4.0) 1.3–3.0 Ͼ0.45 g/L 25% 60%

1CM is usually defined as Ͻ10 white blood cells. NR, not reported.

mate, since the follow-up was incomplete in some of these children (Sowunmi, 1993). The incidence of cognitive im- studies and children with severe sequelae often die at home pairments is also not known. A case control study of 36 from an intercurrent illness. Other sequelae, e.g., epilepsy Gambian children who recovered from CM without any or behavioural difficulties, may develop after discharge and neurological deficits showed that cases may have had im- persist. However, some children with major sequelae paired balance, but there were no significant impairments showed dramatic resolution, particularly if they had cortical of memory, perception, reasoning, learning, visiomotor co- blindness. ordination, dexterity, or manual speed (Muntendam et al., Ataxia, common on discharge, rapidly improves over a 1996). This study, however, only compared the results of period of weeks. The cause of the ataxia is not clear, since single test scores that have a large standard deviation and most children do not have the features of cerebellar in- thus, are unlikely to detect significant differences (and, volvement. It may be caused by muscle weakness following therefore, subject to Type II error). Furthermore, there were damage by cytokines, depletion of glycogen, or even para- no tests of long-term memory, and language was not ade- site sequestration in the muscles (Miller et al., 1989b). quately measured. Recent studies from Kenya suggest that Although clearly definable neurological deficits are now children have difficulties in attention, expressive language, well recognised, the importance of more subtle defects, and behaviour following CM. manifesting as speech and behavioural difficulties, is be- The pattern of neurological sequelae in adults is different coming increasingly recognised. Psychoses with visual and from children (Newton and Warrell, 1998). There are no auditory hallucinations have been reported in some African studies of nonimmune children, but they may have similar

TABLE 5. Incidence and Pattern of Neurological Sequelae Following CM Malawi Gambia Kenya Gambia (Molyneux (Brewster Nigeria (Murphy (van Hensbroek, et al., 1989b) et al., 1990) (Bondi, 1992) et al., 1996)1 1997)2 Total

Number studied 131 308 72 160 6243 1295 Survivors 111 265 62 134 490 1062 Sequelae/number followed up (%) 12 (11%) 35 (13%) 11 (18%) Sequelae detected 1 month after discharge 7 (6%) — — 13 (10%) 114 (24%) Persistent sequelae NR 11/23 NR 20/452 Duration of follow-up NR 6 months 12–16 months NR 18 months Ataxia 3 6 [NR] NR 3 13 [3] 25/1000 Monoparesis/hemiparesis 4 23 [6] 4 [2] 3 11 [4] 44/1066 Severe deficit 5 9 [2] 1 [1] 5 NR 20/578 Hearing impairment NR NR 2 [1] 6 5 [3] 13/666 Visual impairment NR 11 [2] 3 [0] 5 3 [2] 22/931 Dyphasia/dysarthria NR 9 [2] 3 [0] 7 7 [4] 20/931 Behavioural difficulties NR 2 [NR] NR 1 8 [4] 11/865 Cognitive deficit NR NR 1 [1] 4 4 [3] 9/931 Epilepsy NR NR [3] 1 [1] NR 6/797 5 [1]

1Examined 7 days after discharge. 2Sequelae assessed 1 month after discharge. 3Two patients were re-admitted (with convulsions and parasitaemia) within 48 hr of initial discharge. NR, not reported; [ ], persistent sequelae. Pathophysiology and Supportive Treatment of Severe Malaria 23 patterns of damage as the adults. In adults, sequelae are not monary auscultatory or chest X-ray findings and thus, is confined to CM, but are often associated with laboratory probably a manifestation of a systemic acidosis. Similar evidence of severe disease. The major deficits are isolated conclusions also apply to Gambian children with hyperp- cranial nerve lesions, mononeuritis multiplex, polyneurop- nea, where acidosis is an underlying identifiable cause in -van Hensbroek, 1996). A study of 24 Nigerian chil) %70ف athy, extrapyramidal tremor, and other cerebellar signs (de Silva et al., 1992; White and Looareesuwan, 1987). Some dren also identified respiratory distress as a poor prognostic adults have transient psychosis or delirium during recovery, indicator, but a high proportion of children (37.5%) had whilst others develop focal epilepsy, sometimes associated coarse crepitations on auscultation (Olumese et al., 1995). with transient tomographic opacities in the brain (Looaree- In nonimmune adults with uncomplicated falciparum suwan et al., 1983). More recently, a postmalarial syn- malaria, pulmonary involvement, which is mild and asymp- drome, consisting of confusion, convulsions, and tremor, tomatic, is frequent and disappears when the malaria is associated strongly with mefloquine treatment in Southeast treated (Gozal, 1992). Community-based studies have Asia has been described (Nguyen et al., 1996). shown that malaria may be associated with interstitial shad- owing on a chest X-ray, independent of lower respiratory tract infections (Byass et al., 1991), although oxygen desat- 6.3. Metabolic Abnormalities in Severe Malaria uration is not a feature of severe malaria in hospitalised Hypoglycaemia and lactic acidosis both frequently co-exist children with respiratory symptoms (Taylor et al., 1988). and are also complications of CM (Krishna et al., 1994b). The clinical differentiation of pneumonia- and malaria- There are few clinical features that distinguish hypoglycae- associated respiratory abnormalities can be difficult (O’Demp- mia from other manifestations of malaria, since conven- sey et al., 1993). Also, the nomenclature of the “respiratory tional signs, such as decreased consciousness level, sweat- syndrome” is debatable, as respiratory distress suggests a ing, pupillary dilatation, tachycardia, and seizures, are primary pulmonary pathology, rather than the underlying common to both CM and hypoglycaemia. Hypoglycaemia acidosis. Hyperpnoea (defined as an increase in depth and (defined as blood or plasma glucose Ͻ2.2 mmol/L), there- rate of breathing) may be more appropriate as it does not fore, should be suspected in all patients who look severely presuppose the underlying cause(s) (Newton et al., 1998). ill, and rapidly excluded using bedside tests such as the BM- There are no rapid simple laboratory methods to confirm stix™ (a colorimetric assay for blood glucose that relies on lactic acidosis when it is suspected on clinical grounds, and glucose oxidase). Unfortunately, BMstix™ are not always diagnosis currently relies on dedicated equipment that is available and are relatively expensive (although their cost- expensive. However, manufacturers are developing simpler effectiveness can be enhanced by longitudinally cutting methods equivalent to the BMstix™ for glucose estima- them into 2 or 3 strips). If diagnosis of hypoglycaemia is un- tions, and these will prove particularly useful in the identi- confirmed or likely to be delayed more than a few minutes, fication of patients at risk of death with severe malaria. Of a then empirical administration of hypertonic glucose (1 mL/ variety of admission clinical and laboratory parameters as- kg, 50% glucose) is indicated. The majority of patients will sessed by logistic regression analysis in a Gambian study of not improve dramatically after correction of hypoglycae- 115 children with severe malaria, deep coma and the lac- mia, indicating that hypoglycaemia is not the main cause of tate/glucose ratio were independently found to be predic- symptoms such as coma. tive of mortality. Figure 8 shows the relationship among Lactic acidosis (defined as blood or plasma lactate у5 hypoglycaemia, hyperlactataemia, and mortality in these mmol/L) may be equally difficult to recognise clinically. It patients. should be suspected in any severely ill patient who has When all clinical and laboratory variables were exam- markedly abnormal respiratory patterns, such as tachypnoea ined together in a regression model, sustained hyperlacta- (Ͼ50 respirations/min in children), deep gasping respira- taemia (у5 mmol/L 4 hr after admission) proved to be the tions, or the use of secondary muscles of respiration (Marsh single most sensitive indicator of a fatal outcome, followed et al., 1995; Krishna et al., 1994b). In Kenyan children, by the BCS at 4 hr and presence of hypoglycaemia on ad- “respiratory distress” was defined as alar flaring, chest reces- mission (Krishna et al., 1994b). In many children, hyper- sion (intercostal or subcostal), the use of accessory muscles of lactataemia resolves soon after admission, resuscitation, respiration, or abnormally deep (acidotic) breathing, whilst and the start of antimalarial treatment. In a proportion of “severe respiratory distress” was used for children with chest children, hyperlactataemia remains even after vigorous at- recession or abnormally deep breathing (Marsh et al., 1995). tempts at resuscitation, correction of electrolyte abnormali- In this study, respiratory distress was observed in 7% of ties, and hypoglycaemia (English et al., 1996c; Krishna et 1844 children admitted with malaria to a district hospital. al., 1994b). The severity of lactic acidosis in this latter group Half were fully conscious, 19% died (13.9% died with any is clearly much greater than in children who recover quickly. symptom of respiratory distress and 24.8% with severe res- This is reflected in blood gas measurements made in Malaw- piratory distress). In the children with respiratory distress in ian children who presented with severe malaria (Taylor et whom plasma bicarbonate was measured, 81% were severely al., 1993). Sustained acidaemia was the most specific indi- acidotic (plasma bicarbonate concentration Ͻ15 mmol/L). cator of a poor prognosis in this group of children, again Respiratory distress was not associated with abnormal pul- suggesting that the buffering capacity of the patient could 24 C. R. J. C. Newton and S. Krishna not cope with an increased acid load. Approximately 42% decrease in haemoglobin than the absolute level of haemo- (25/60) of children with CM were acidaemic (capillary globin, but children with chronic anaemia may decompen- blood pH Ͻ7.3) compared with 5% (4/81) with uncompli- sate rapidly when challenged with a febrile illness. cated disease (P Ͻ 0.0001), and some had inadequate respi- The symptoms and signs of life-threatening severe malarial ratory compensation for the metabolic acidosis. The rela- anaemia are those of respiratory distress and hyperdynamic tive risk of a fatal outcome for acidaemic patients with CM circulation, which may be caused by the underlying acidosis was 8.5 (95% CI 1.1–64). Acidaemia, attributable to a lac- from inadequate oxygenation (English et al., 1997) or im- tic acidosis, was associated with hypoglycaemia, elevated pairment of cardiac function. The contribution of impaired white cell count, and witnessed convulsions. Although lac- cardiac function to respiratory distress is not determined. tate levels were lower in uncompensated metabolic acidosis However, most African children with severe falciparum than in those who had partially compensated, this finding malaria are fluid depleted, and respond well to rapidly ad- probably represents the inability of some patients with CM ministered blood transfusions (English et al., 1996c). to mount an effective ventilatory response to acidosis. Malaria is the most common cause of severe anaemia in children admitted to hospitals in Africa and is responsible 6.4. Severe Malarial Anaemia, Jaundice, and Haemolysis for many blood transfusions (Newton et al., 1997d; Slutsker The incidence of severe malarial anaemia has increased et al., 1994; Greenberg et al., 1989). In children with CM, the with the spread of chloroquine-resistant P. falciparum para- lowest haematocrit is reached within the first 24 hr after sites across the world (Zucker et al., 1996). Patients may treatment (Molyneux et al., 1989b) and 23% of the children improve symptomatically following chloroquine treatment, required blood transfusions during the first 24 hr (Moly- but unless the parasites are cleared completely, the haemo- neux et al., 1989b). In some studies, African children with globin is lowered and the patient becomes increasingly vul- CM who also have severe anaemia (haemoglobin Ͻ50 g/L) nerable to a life-threatening anaemia (Bloland et al., 1993). are less likely to die (Marsh et al., 1995), but in others are Destruction of PRBCs is an inevitable consequence of P. falci- more likely to die or survive with sequelae (Brewster et al., parum infection and usually produces a fall in haemoglobin. 1990). They may take longer to recover consciousness (Moly- The WHO defines severe malarial anaemia as a haemoglo- neux et al., 1989b). Whether severe anaemia contributes bin concentration Ͻ50 g/L (or a haematocrit Ͻ15%) in a directly to impaired metabolism (by reducing oxygen deliv- patient with a P. falciparum parasitaemia in excess of 10,000 ery) or simply is a marker of severe disease is unclear. trophozoites/mm3 (Warrell et al., 1990), with normocytic indices. However, this definition may not be appropriate in 6.5. The Commonest Modes of malarial-endemic areas, since anaemia is common, caused Presentation of Severe Malaria in Children mainly by iron and folate deficiency or haemoglobinopathies (Newton et al., 1997d), and children often adjust Identifying children with malaria who are at high risk of a physiologically to the low haemoglobin concentrations. fatal outcome depends on recognition of clinical signs, as Life-threatening anaemia is determined more by the rate of well as access to laboratory variables. When only simple clinical criteria and bedside tests are available, then children who are deeply unconscious, hyperpnoeic, and hypoglycaemic represent the most severely ill group of patients. If lactate measurements or blood gas data are available, then the additional presence of lactic acidosis (lactate у5 mmol/L) or acidaemia (pH Ͻ7.3, or base excess Ͻ15 mmol/L) identifies the majority of severely ill children of whom will die if they have 3 risk factors) (Taylor %80ف) et al., 1993). However, the largest experience with these prognostic variables is from Africa, and there may be considerable geographic variation in their relative importance. In Papua New Guinea, depth of coma was surprisingly not an important discriminant of a fatal outcome, whereas lactic acidosis was in one study (Allen et al., 1996b), but in another study, deep coma, anaemia, and leucocytosis were associated with death (Genton et al., 1997). In our experience with African children, parasitaemia is not a useful prognostic variable (for reasons discussed in Section 2). FIGURE 8. The relationship between venous metabolite con- The distribution of stages of development on a peripheral centrations and mortality of 115 Gambian children with severe blood film may prove to be an additional simple bedside malaria: glucose (ٗ), lactate (᭺) at admission, and lactate after 4 hr (᭹). Reproduced from Krishna et al. (1994), with permis- marker of severe disease, as in Southeast Asia (Silamut and sion of the copyright holder, Royal Society of Tropical Medicine White, 1993), but further studies are needed to confirm the and Hygiene, London. utility of this test in African children. Likewise, the pres- Pathophysiology and Supportive Treatment of Severe Malaria 25 ence of pigment in leucocytes was found to be a useful prog- the fact that the threshold at which fever occurs appears to nostic index in Vietnamese adults (Nguyen et al., 1995), be influenced by age (Greenwood, 1996). Thus, recent stud- and was associated with severe malaria in Gabonese chil- ies confirmed older findings that the parasitaemic threshold dren (Metzger et al., 1995), but its usefulness remains to be for fever varied by a factor of 5 with age and the threshold was established in other areas. Detailed consideration of the highest in 1- to 2-year-old children (Rogier et al., 1996). pathophysiology of these clinical syndromes is discussed in The rise in temperature during malaria infections is Section 7 and their management in Section 9. thought to be mediated by TNF (Kwiatkowski et al., 1989). TNF is an endogenous pyrogen, acting on the hypothala- 6.6. Moderate Malaria mus to reset the set-point for temperature homeostasis. The Moderate malaria is a disease category that identifies chil- ablation of fever in children administered anti-TNF mono- dren who require parenteral treatment, but are unlikely clonals supports the importance of TNF as the cause of fe- (Ͻ10%) to progress to severe disease once appropriate anti- ver (Kwiatkowski et al., 1993), although the fact that not malarial treatment has begun. Patients with moderate ma- all fevers are associated with elevations in TNF suggests laria have none of the defining features of severe disease that other mechanisms may also be involved. (no hypoglycaemia, lactic acidosis, coma, or recurrent sei- The role of fever in the pathophysiology of severe ma- zures), but have other features that suggest the requirement laria is undetermined. Fever may synchronise the stages of for parenteral treatment: parasites (Kwiatkowski, 1989), promoting the survival of parasites. In Gambian children with uncomplicated malaria, • a history of frequent and/or recent vomiting the increased resting energy expenditure is 37% higher com- • drowsiness pared with when they have recovered, and this is associated • obtundation, i.e., an apathetic response to painful stim- with a doubling of whole body protein turnover assessed by uli such as capillary blood sampling or venepuncture stable isotope techniques (Berclaz et al., 1996). Elevations in • prostration, i.e., difficulty or inability to suckle, sit up, body temperature are important in increasing energy expen- stand up, or walk unaided according to the child’s age diture (by between 8.5 and 13.2%/ЊC), but are not the only Moderate malaria, like severe malaria, does not have a mechanism causing these rises. There is a negative protein predictable relationship with the level of peripheral asexual balance during acute infection, which is reversed 15 days parasitaemia or vital signs in individual cases (Sowunmi et after the infection has been treated. Similar studies are al., 1992). Whilst the majority of children with moderate lacking in children with severe malaria, but even greater disease improve rapidly with treatment and are discharged perturbations of energy expenditure are likely during the from hospital within 1–2 days, deterioration in clinical sta- acute phases of infection. The role of fever on cerebral me- tus can occur within a few hours. Apart from the require- tabolism is not known and is unlikely to be predicted from ment for parenteral therapy, this potential for rapid disease whole body metabolism in unconscious children (Matthews progression necessitates admission and close observation. et al., 1995). The effect of temperature on the brain is However, in-hospital mortality in this category of patients potentially important because increases in temperature pro- is by definition low (Ͻ3%). mote ischaemic damage to neurons and may contribute to the Uncomplicated malaria is infection not associated with development of neurological sequelae in children with CM. any of the features of moderate or severe disease. It most commonly presents with fever, and the child is able to tol- 7.2. Coma erate oral antimalarial medication and can be treated as an outpatient with an appropriate antimalarial. Mortality is The cause of the depressed level of consciousness in falci- not anticipated in uncomplicated malaria, providing appro- parum malaria is unknown. Any theory of causation needs to priate antimalarials have been administered. explain why coma develops relatively late in the illness in adults, compared with children; the fact that most patients recover without any evidence of neurological sequelae; and that coma is caused by more than one mechanism. Part of 7. PATHOPHYSIOLOGY the difficulty has been the definition of impairment of con- 7.1. Fever sciousness, since some of the “coma” scales used may not Fever is a characteristic feature of falciparum malaria, although measure the depth of coma (Newton et al., 1997a). parasitisation is not inevitably associated with fever. In ma- Consciousness may be impaired by depression of the re- laria-endemic areas, many children are afebrile despite being ticular activating system within the brainstem or a diffuse parasitaemic, even with relatively high parasitaemias. There process affecting the cerebrum (Plum and Posner, 1980). is, however, a broad relationship between the level of parasi- Sequestration of PRBC within the cerebral venules, partic- taemia and fever in these areas, so that it is possible to deter- ularly within the brainstem, is the most widely accepted mine a threshold of parasitaemia (usually 2000–20,000 para- mechanism. The patchy involvement of blood vessels in sites/␮L) that has an optimum sensitivity and specificity for the brain may explain why most patients with CM develop the definition of a clinical episode of malaria (Greenwood, a completely reversible coma. The impaired delivery of nu- 1996; Smith et al., 1995). The relationship is complicated by trients to certain areas of the brain may be compensated for 26 C. R. J. C. Newton and S. Krishna by normal or increased delivery to microscopically adjacent age, particularly if they are focal in nature. Distinguishing regions (luxury perfusion), preserving the integrity and cause from effect clearly has important ramifications, since function of the brain as a whole, in spite of coma. Global anticonvulsant prophylaxis may prevent seizures, but not cerebral blood flow is not decreased in adults or children neurological damage. Seizures may cause death by aggravat- with CM (Newton et al., 1996a; Warrell et al., 1988), but ing intracranial hypertension (IH) (Newton et al., 1991a), whether it is appropriate for metabolic demands is un- may cause neuronal loss, and may precipitate aspiration of known. The sequestered parasites may also produce toxins, the gastric contents into the lungs (Waller et al., 1995). either directly or stimulate host cells to produce substances The cause of the seizures in CM is unclear. Intracranial that interfere with neuronal metabolism. sequestration of metabolically active parasites is a potential The common occurrence of seizures may contribute to mechanism, but difficult to investigate. The seizures are un- the impairment of consciousness, although the contribu- likely to be simply febrile convulsions (Akpede et al., 1993), tion is difficult to determine. Subtle seizures and noncon- since 54% of seizures occurred when the rectal tempera- vulsive status have been documented in Kenyan children tures were less than 38.0ЊC, 48% were partial, and over (Crawley et al., 1996; Kirkham et al., 1994). In the later 70% were repetitive (Waruiru et al., 1996). Seizures may study, only half of the children with subtle seizures (i.e., be caused by hypoglycaemia and hyponatraemia, although 7.5% of children with CM) appeared to regain conscious- studies of Malawian and Kenyan children have not identi- ness within 6 hr of the termination of the electrical activity fied these mechanisms as clinically important in malaria with diazepam. Prolonged post-ictal states may also contrib- (Crawley et al., 1996; Waruiru et al., 1996; Molyneux et al., ute to the impairment of consciousness in other children 1989b). Other potential mechanisms include toxins, e.g., (Crawley et al., 1996), but there are no specific EEG fea- NO, excitotoxins, as discussed in Section 7.5.2. tures that differentiate a post-ictal state from a rapidly improving encephalopathy. The inappropriate administration of benzodiazepines (intramuscular administration of high 7.4. Raised Intracranial Pressure doses) may also contribute to impaired consciousness and Raised ICP is an important cause of poor outcome in paedi- apparently prolonged post-ictal states. atric encephalopathies, particularly in CNS infections In a few children with hypoglycaemia, the conscious level (Goitein et al., 1983). It can cause death by transtentorial improves with the administration of glucose, although most herniation or a reduction in cerebral perfusion pressure do not (Taylor et al., 1988). Other metabolic derangements (CPP), where CPP ϭ mean arterial pressure Ϫ ICP. Raised [e.g., severe acidosis (Allen et al., 1996a)] are associated ICP was thought not to be important in adults with CM, with coma, but whether they are directly responsible for coma since over 80% of opening CSF pressures were in the nor- is unclear. Systemic effects of the infection [e.g., fever or mal range, there was no difference in the CSF pressures be- hyponatraemia (Arieff and Griggs, 1994)] may also contrib- tween those who died and survived (Warrell et al., 1986), ute, although they are unlikely to be responsible for deep and cerebral oedema was probably an agonal event (Looa- coma. The role of toxins in the pathogenesis of coma requires reesuwan et al., 1983, 1995). However, opening CSF pres- further investigation. NO may impair neurotransmission, sures do not predict the maximum ICP when studied in but there is little evidence to support this as a mediator in other encephalopathies, and raised ICP may be caused by CM (see Section 4.5). Desferrioxamine, a ROS scavenger, an increase in cerebral blood volume (Newton et al., appears to reduce the duration of deep coma (Gordeuk et al., 1991a) rather than cerebral oedema. Opening CSF pres- 1992), although the mechanism is unclear. More recently, sures are raised in most African children with CM (Lew- excitotoxic mechanisms have been invoked (Dobbie et al., allen et al., 1993; Waller et al., 1991; Newton et al., 1997), particularly to explain the incidence of seizures. 1991a). The CSF pressures are similar to those found in Thus, it is likely that the impairment of consciousness pro- adults with malaria, but are above the normal range for duced by falciparum malaria is caused by a number of inter- children. Most Kenyan children dying with CM have clini- acting mechanisms (Newton et al., 1994) that cannot be eas- cal signs compatible with transtentorial herniation (New- ily differentiated from each other, and the contribution of ton et al., 1991a), although these signs may be caused by each mechanism cannot be easily determined. Further light other mechanisms (Kwiatkowski et al., 1991; Kirkham et on these mechanisms may be shed by specific interventions. al., 1991; Newton et al., 1991a). ICP monitoring confirmed that children deeply unconscious from CM all had raised ICP (with maximum ICP higher than the upper limit 7.3. Seizures of adults) and that opening ICP did not predict maximum Seizures, particularly those witnessed after admission in chil- ICP (Fig. 9) (Newton et al., 1997b). Furthermore, all children with CM, are associated with death (Waller et al., 1995; dren who developed severe IH (maximum ICP Ͼ40 mm Molyneux et al., 1989b), and prolonged and multiple seizures Hg, minimum CPP Ͻ40 mm Hg) either died or survived are associated with neurological sequelae. The relationship with severe neurological sequelae. One child, who devel- may not be causal, since seizures may be a marker of neuro- oped an ICP of 158 mm Hg, died with signs of transtento- logical damage rather than the cause. In other encephalo- rial herniation. In another study, transcranial Doppler stud- pathies, seizures often herald the onset of neurological dam- ies showed that about one-half of the children dying from Pathophysiology and Supportive Treatment of Severe Malaria 27

CM had sonographic features of progressive IH, whilst the remainder had sonographic features similar to children who died of NCM (Newton et al., 1996a). Transtentorial herniation occurs when there is an ICP gradient established between the contents of the skull and the spinal cord (Plum and Posner, 1980). This may not apply in CM, where there may be a uniform increase in pressure in the neuraxis due to swelling and, therefore, displacement of cerebral structures is less likely to occur until ICPs become grossly elevated. The pathological features of tentorial herniation, despite being described in the 1930s, have been reported infrequently in CM until recently (Section 3.2). The most likely cause of raised ICP in CM is an increase in cerebral blood volume (Newton et al., 1991a) aggravated by cytotoxic oedema in those who develop severe neurological sequelae (Newton et al., 1994). In a CT of Kenyan children recovering from CM, the scans were normal in most children, but transient brain swelling occurred in children with intermediate IH, and tomographic features of cytotoxic oedema were seen in children who developed severe IH (Newton et al., 1994). There was no evidence of vasogenic oedema or acute hydrocephalus in any of these children. FIGURE 9. Opening ICP and maximum ICP in 23 children These findings are consistent with observations made in a who have undergone ICP monitoring. Opening ICP does not magnetic resonance imaging study in adults with CM (Looa- predict maximum ICP. Modified from Newton (1995). reesuwan et al., 1995). Cerebral blood volume could be increased by the sequestration of PRBC in the vascular compartment, either acting as a diffuse space occupying lesion or as ROS, NO, or excitoxins. Reduction of cerebral blood obstructing venous outflow. Other features of CM, such as flow has not been associated with diffuse neurological dam- seizures, anaemia, and hyperthermia, would also increase ce- age in either children (Newton et al., 1996a) or adults rebral blood flow and thus, increase cerebral blood volume. (Warrell et al., 1988), although these intermittent mea- Thus, raised ICP appears to be a feature of CM in Afri- surements may have missed critical periods of reduced per- can children, and severe IH, when it occurs, is associated fusion. Reduced CPP, however, is strongly associated with with a poor outcome. Whether IH is a primary pathophysi- severe neurological sequelae (Newton et al., 1997b), and if ological process remains to be established. Mannitol was ef- cerebral autoregulation is impaired, cerebral blood flow will fective in lowering the ICP in Kenyan children (Newton et be determined by CPP. Some children with neurological se- al., 1997b) and may have prevented children with mild de- quelae have evidence of borderzone infarction (Fig. 10), grees of IH from dying or developing neurological sequelae, providing further evidence of reduced cerebral perfusion but it did not prevent the development of intractable IH in (Newton et al., 1994), although this is not a consistent fea- those children with a poor outcome (Newton et al., 1997b). ture. Seizures, either prolonged or repetitive, may promote Further post-mortem studies examining the importance of ischaemic damage, although as yet the evidence is lacking herniation, and clinical studies to test the effectiveness of (Newton et al., 1997c). substances to reduce ICP in children with CM, are needed before treatment recommendations can be made. 7.5.2. Reactive oxygen species. ROS are generated during and after hypoxia-ischaemia. These species damage tis- 7.5. Causes of Neurological Sequelae sues by interacting with the polyunsaturated fatty acids of cellular membranes, damaging proteins and DNA (Siesjo et The pathogenesis of neurological damage is likely to be differ- al., 1989). ROS initiate a vicious cycle of lipid peroxida- ent from the mechanisms that cause death, since the risk fac- tion, which is promoted by the presence of iron and lactic tors are different (Newton, 1995). Death of neuronal tissues acidosis. This mechanism appears to be important during may be caused by necrosis or apoptosis, with similar or differ- reperfusion, particularly after a period of reduced microcir- ing pathophysiological processes that lead to these mecha- culatory flow (Hurn et al., 1996) rather than complete is- nisms (Table 6). In Kenyan children with neurological se- chaemia, as may occur during the clearance of parasites. quelae, CT scans demonstrate a variety of patterns of damage, These species also promote the formation of NO and may suggesting that a number of mechanisms may be responsible. enhance the damage caused by excitotoxins.

7.5.1. Ischaemia. Necrosis is caused by a critical reduc- 7.5.3. Excitotoxins. Excitotoxins are endogenously pro- tion in blood supply (ischaemia) or damage by toxins, such duced amino acids, which, under normal conditions, act as ex- 28 C. R. J. C. Newton and S. Krishna citatory neurotransmitters, but when overproduced, become In 2 out of 7 children with hemiparesis, no flow could be toxic to the neuronal cells. The release of excitatory amino detected in the contralateral middle cerebral artery. acids is stimulated by hypoxia-ischaemia, hypoglycaemia, In a CT scan study of 12 children who had a residual and seizures; the latter two are associated with the develop- hemiparesis 3–40 months following CM, 4 had unilateral ment of neurological sequelae in CM. There is no direct ev- atrophy, 5 asymmetrical atrophy, and 3 diffuse atrophy idence that the above toxins cause neurological damage in (Newton et al., 1996b). All the children with unilateral at- CM, although these substances are difficult to measure, par- rophy had partial motor seizures and tomographic features ticularly in the brain. The studies that have been conducted similar to those seen in hemiconvulsion-hemiplegia syn- demonstrate the raised concentrations of quinolinic acid, drome after prolonged febrile convulsions (Kataoka et al., an excitotoxin in the CSF in children with sequelae (Dobbie 1988). A child with multiple seizure types had severe asym- et al., 1997). These results need to be interpreted within the metrical atrophy. The atrophy was more severe in the pari- light of a possible breakdown in the BBB (Section 4.7.2). eto-occipital regions in 8 children. In none of these scans was the damage confined to a single basal cranial artery ter- ritory. Furthermore, in an electroencephalographic study of 7.5.4. Apoptosis. Apoptosis is the process of programmed CM, 66% of seizures were partial, often localised over pari- cell death that occurs in vertebrates. The cell activates an eto-temporal regions (Crawley et al., 1996), regions that intrinsic suicide mechanism that leads to characteristic are not typically borderzone areas. These findings suggest morphological changes (Fraser and Evan, 1996). It is in- that hemiparesis in CM may be caused by convulsions, pos- duced by many factors, including TNF, ROS (Baier-Bitter- sibly as part of the hemiconvulsive-hemiparesis syndrome. lich et al., 1995), and proteases. The role of apoptosis in The occlusion of middle cerebral artery found in some chil- neuronal death remains controversial, and as yet, there is dren still requires an explanation since large vessel throm- no evidence to suggest apoptosis is responsible for neurolog- bosis is not a recognised feature of CM. Underlying ana- ical damage in CM, although this mechanism is difficult to tomical abnormalities, disturbances of clotting, or severe investigate noninvasively. vasospasm may contribute.

7.5.5. Hemiparesis. Hemiparesis is one of the most com- 7.5.6. Visual impairment. Cortical visual impairment, demon and uniform neurological sequelae, but the cause is un- fined as visual impairment with intact pupillary responses known. In studies of 10 West African children with hemi- and normal fundi, is common following CM (Brewster et paresis who did not have clearly defined CM (i.e., other al., 1990). Although it can occur in isolation following causes of an encephalopathy were not excluded), angiogra- CM, it is usually associated with evidence of more diffuse phy demonstrated stenosis abnormalities in 5 children damage (Lafaix et al., 1970). Blindness tends to improve, (Omanga et al., 1983; Collomb et al., 1967). Cerebral artery although most children do not recover completely. The occlusion was found in 4 children and segmental narrowing cause of blindness is not clear, although it is often associ- of the internal carotid in a further child; 5 children had com- ated with seizures (Olurin, 1970). The occipital area is vul- pletely normal angiograms. In Kenyan children with strictly nerable to hypoxic-ischaemic insults in other encephalopa- defined CM, over one-half had a marked asymmetry of blood thies [e.g., meningitis (Thun-Hohenstein et al., 1992)]. This flow velocity in the middle cerebral artery, as detected by area is particularly prone to damage following a reduction transcranial Doppler sonography (Newton et al., 1996a). in cerebral perfusion, possibly because of a difference in the

TABLE 6. Possible Mechanisms for Neurological Damage

Anaemia Impaired delivery of oxygen Cytokines Direct toxic effects Stimulate release of NO Diversion of metabolic substances Competition between host and parasites for metabolic substrates Hypoglycaemia Impaired delivery of substrate Hypoxia Microvascular obstruction Anaemia Reduced cerebral blood flow IH Impaired cerebral perfusion Herniation causing compression of posterior cerebral artery Lactic acidosis Aggravating acidosis NO Formation of peroxynitrite ROS Lipid peroxidation NO Seizures Excitotoxic damage Mismatch between cerebral metabolism and oxygen supply Raised ICP Pathophysiology and Supportive Treatment of Severe Malaria 29

FIGURE 10. CT scan of child with CM. (A) acute scan showing diffuse brain swelling with extensive hypodensity of the su- perficial watershed areas (arrows) and the basal ganglia. (B) con- valescent scan showing contrast enhancement of the border zone (arrows) between the left middle and posterior cerebral artery areas. Reproduced from Newton et al. (1994), with permission of the copyright holder, BMJ Pub- lishing Group, London.

autoregulation of the blood flow in the posterior circula- When quinine was infused in normal volunteers, there was tion, making it more vulnerable to changes in perfusion a significant increase in plasma insulin concentrations and pressure. Damage to the thalamic radiations and other ex- a corresponding fall in fasting plasma glucose concentra- tra-striate structures may also occur. tions. These observations confirmed that quinine was responsible for most episodes of hypoglycaemia in these patients, through its capacity to induce hyperinsulinaemia. 7.6. Hypoglycaemia Pregnancy was identified as a major risk factor for the de- Early studies noted that hypoglycaemia was a common velopment of hypoglycaemia. complication of malaria infection in humans (Fitz-Hugh, These studies in adults subsequently led to an examina- 1944) and animals (Mercado and von Brand, 1954). More tion of the incidence and mechanisms of hypoglycemia in recently, there have been studies of the mechanisms of hy- African children with severe malaria (White et al., 1987b), poglycaemia in animal models of malaria, but findings may after 2 index cases presented with hypoglycaemia during not be directly applicable to human disease processes. the course of pharmacokinetic studies on chloroquine. In Observations were occasionally made in patients with se- Gambian children who had P. falciparum parasitaemia and vere disease (Migasena, 1983; Fisher, 1983), but the first required parenteral chloroquine treatment, the incidence of prospective series that identified hypoglycaemia and lactic hypoglycaemia was 15/47 (32%), and its presence was pre- acidosis in malaria and studied its pathophysiological dictive of subsequent mortality (White et al., 1987b,c). Hy- mechanisms was published from Thailand in 1983 (White poglycaemia was significantly associated with CM (52% et al., 1983). Twelve of 151 (8%) adult patients with CM had hypoglycaemia in this subgroup), a higher peripheral experienced hypoglycaemia, although 17 patients were parasite count, and abnormal liver function tests. In con- studied overall, including 3 women with otherwise uncom- trast to Thai adults, insulin levels were appropriately low in plicated infection who were pregnant or who had delivered hypoglycaemic children, and patients were appropriately recently. The mortality in the hypoglycaemic group was sig- ketotic. A later Sudanese study also found that pregnant nificantly higher than in patients without hypoglycaemia women were prone to hypoglycaemia without associated [8/14 (57%) vs. 27/139 (19.4%), respectively; P ϭ 0.004], hyperinsulinaemia (Saeed et al., 1990). These observations and hypoglycaemic patients had higher parasitaemias and are consistent with the rapidity with which starved chil- more severe multi-organ involvement than the remainder. dren (or pregnant women, or rarely other adults) are prone A striking observation was that plasma insulin and C pep- to deplete glycogen reserves and the inability of chloro- tide levels were inappropriately high during episodes of hy- quine to provoke hyperinsulinaemia. They suggest impor- poglycaemia. In all but one patient, plasma quinine was de- tant differences in the pathogenesis of hypoglycaemia be- tectable during the episode of hypoglycaemia, and quinine tween Southeast Asian adults and patients in Africa. and insulin concentrations were correlated at the first hy- Since the first reports on hypoglycaemia, many prospec- poglycaemic episode. There was an unpredictable clinical tive studies have confirmed that children are particularly response to glucagon or glucose treatment in these patients, prone to develop this complication (Tables 1 and 3), but although glucagon increased plasma glucose in all patients. few have examined its pathophysiology in detail. In 30 C. R. J. C. Newton and S. Krishna

Malawi, blood glucose was negatively correlated with the plasma glucose production (Dekker et al., 1996), and did plasma concentrations of the gluconeogenic substrates lac- not increase with the infusion of alanine (Dekker et al., tate and alanine, but interestingly, not with the length of 1997). Thus, in this group of children, it appears that en- time that patients had symptoms or had been fasting (Tay- dogenous glucose production is an important determinant lor et al., 1988). In this study, quinine was infused with of plasma glucose, in contrast to the findings in adults. dextrose solution, and patients who were normoglycaemic This increased demand for glucose in falciparum malaria on admission did not develop hypoglycaemia. Pretreatment results from many stresses, including the effects of fever and hypoglycaemia was associated with higher mortality and re- increased anaerobic glycolysis (which utilises glucose much current episodes of postadmission hypoglycaemia in some less efficiently to produce ATP) by both host tissues and children, and a proportion of this group had inappropriate parasitised erythrocytes. Anaerobic glycolysis is discussed in elevations in plasma insulin levels. In survivors, hypogly- the following section. caemia was associated with a higher incidence of neurological sequelae. There is likely to be depletion of glycogen reserves in the livers of children with severe malaria, but the 7.7. Lactic Acidosis biochemical basis of gluconeogenic defects has not been Systemic lactic acidosis, like hypoglycaemia, is a metabolic identified. Growth hormone concentrations are elevated in endpoint of many potential causes in severe malaria. The hypoglycaemic children, and there is no deficiency in relative importance of these causes may vary from patient plasma cortisol levels, suggesting appropriate counter-regu- to patient, and broadly reflect both increases in lactate pro- latory homeostatic responses. duction and inadequate lactate clearance by normal ho- In adults with severe malaria, the conversion of galactose meostatic mechanisms (Mordes et al., 1996). Lactic acido- (Pukrittayakamee et al., 1991b) or glycerol (Pukrittayakamee sis is one of the best markers of disease severity and et al., 1994b,c) to glucose are not significantly impaired mortality in malaria (Krishna et al., 1994b). In normal indi- when compared with similar rates measured in convales- viduals, gross elevations in plasma lactate occur transiently cence, arguing against gluconeogenic impairment affecting in response to anaerobic strenuous exercise. These eleva- the enzymes responsible for these relatively proximal syn- tions are a short-term solution to the demand for ATP gen- thetic pathways. Similar studies have been completed with erated in skeletal muscle, and lactate is rapidly metabolised the more distal metabolite alanine, and this study will pro- from peaks of up to 20 mM to basal levels of ~1 mM when vide a more comprehensive assessment of gluconeogenesis oxygen becomes available to tissues. The production of lac- in severe malaria (S. Pukrittayakamee, S. Krishna and N. J. tate under anaerobic conditions allows glycolysis to con- White, unpublished observations). Glucose kinetics mea- tinue by maintaining a supply of NADϩ, without which sured in pregnancy (Davis et al., 1994) and in adults with glycolysis will cease. Utilisation of lactate takes place in the uncomplicated (Davis et al., 1990b) and severe infection liver, kidney, skeletal muscle, and CNS by metabolism via (Davis et al., 1993) demonstrate increased consumption of pyruvate. Most pyruvate enters Krebs’ cycle, where it is oxi- glucose in acute infection. In the latter study, a primed in- dised to CO2 and water, utilising both the accumulated lac- fusion of [3-3H]glucose was given to 18 normoglycaemic tate anion and the accompanying acidic proton. Without adults with severe infection, and monitoring of glucose ki- this resolution of lactic acidosis, the buffering capacity of the netics carried out before and after quinine treatment. Glu- blood and tissues is exceeded, and lactic acidaemia supervenes cose turnover was higher than normal, correlated with oral on lactic acidosis (Stacpoole, 1997). The levels of lactic acid temperatures, inversely correlated with plasma glucose levels, where acidaemia is likely to supervene are у5 mM, although and decreased significantly after quinine infusion. Plasma in- there may be significant metabolic dysfunction in patients sulin values rose significantly after quinine. Postquinine with lactate levels below this value. Sustained elevations in and convalescence glucose turnover rates were similar, but plasma lactate, therefore, can be viewed as a marker for in- in view of the elevated insulin levels after quinine, hepatic adequate mitochondrial metabolism of pyruvate, as this is insulin resistance may be important in the acute phase of the principal route whereby lactic acid is cleared in health infection. In adults, peripheral insulin resistance offsets the (Stacpoole, 1997). In severe malaria, it is unlikely that al- hypoglycaemic effects of hyperinuslinaemia in acute infec- ternative routes of lactate utilisation via pyruvate through tion until recovery is taking place when delayed hypogly- synthetic routes (by conversion to oxaloacetate or acetyl caemia becomes a complication. coenzyme A) will be quantitatively important. African children recover much more rapidly than adults, Increased lactate production in severe infection can re- and although delayed hypoglycaemic episodes are more sult from many causes. Fever and anaemia themselves in- common following quinine therapy, compared with other crease metabolic rates (Stettler et al., 1992). Generalised antimalarials, these may be less problematic than in adults seizures cause brief, large bursts of anaerobic glycolysis in (Krishna et al., 1994b). Admission hypoglycaemia is the muscle tissue, and can produce severe, if transient, lactic more important complication of severe malaria in African acidosis in otherwise normal adults (Orringer et al., 1977). children. In Kenyan children with uncomplicated malaria, Their effects are likely to be more severe in malaria where basal glucose production did not correlate with alanine, lac- there are other derangements in glycolytic pathways, and tate, counter-regulatory hormones, or cytokines, but only some host tissues (such as the brain and muscle) are already Pathophysiology and Supportive Treatment of Severe Malaria 31 net exporters of lactic acid to the blood. While muscle tis- infection (both before and after quinine treatment) and in sue damage is well documented in severe malaria in chil- convalescence, although metabolic clearance of lactate was dren (Miller et al., 1989b) and a higher proportion of chil- significantly lower in acute infection. Lactate turnover in dren with abnormal creatine kinase values [12/16 (75%)] acute infection was correlated with simultaneous arterial have a history of seizures when compared with those whose plasma lactate concentrations, but no such correlation was creatine kinase values are in the normal range [21/21 observed with lactate clearance rates. Lactate release from (50%); P ϭ 0.076, Fisher’s exact test], there are no studies skeletal muscle was significantly greater following quinine, that have prospectively evaluated the contribution of sei- and this was much higher than that observed in convales- zures to lactic acidosis in malaria in children. cence. The observed reduction in lactate clearance in these Infected erythrocytes generate 20- to 50-fold lactate (and patients accords with observations made in severe malaria, consume correspondingly large quantities of glucose) com- where there is diminished liver blood flow. This reduction pared with NPRBCs (Warrell, 1989). This increased glycoly- in liver blood flow may limit lactate delivery to the liver sis may not be sufficient to perturb homeostasis in a normal (Molyneux et al., 1989a), although this is less likely in pa- individual, but it may be more important in severe malaria tients with moderately severe disease (Pukrittayakamee et where the PRBC are localised to capillaries, and local met- al., 1994b). Additionally, the liver may have an impaired abolic diversion may be more severe. Severe anaemia de- intrinsic capability to clear lactate in acute infection. creases the oxygen-carrying capacity of blood, although Many of these earlier observations made on lactic acido- compensatory mechanisms can restore oxygen delivery to sis and malaria have been repeated in a recent Kenyan tissues. If anaemia develops over a period of days rather study (English et al., 1997). In this study, a high anion gap than hours, then there may also be changes in the oxygen- (Ͼ11) remained in most patients with severe malaria, even dissociation curve (Jones and McGregor, 1954) induced by after accounting for hyperlactataemia, and alternative causes increased intraerythrocytic synthesis of 2,3-diphosphogly- for this were suggested. Renal impairment was not severe (as cerate (Bohr shift), resulting in increased unloading of oxy- estimated by plasma creatinine values) and was associated gen, although these mechanisms can be blunted in the pres- with acidosis, but could not account for the increase in an- ence of acidosis (Krishna et al., 1983). Anaemia also ion gap, as organic acids retained by the kidney result in a decreases viscosity, which can increase blood flow to tis- relatively slowly developing lactic acidosis, usually when re- sues, although the increase in cerebral blood flow following nal impairment is severe. It is likely that other anions, such a drop in haemoglobin is caused by a decrease in oxygen- as ␤-hydroxybutyrate (not picked up by dipstick analysis of carrying capacity (Brown and Marshall, 1985). These phys- urine) or phosphate or urate, could have contributed to the iological adaptations have not yet been examined in severe anion gap, although these were not investigated in this malaria in humans, although a study in an animal model of study. In summary, lactic acidosis is an important complica- severe malaria, 31P-NMR studies demonstrated adequate tion in severe malaria that is difficult to manage, and may cerebral stores of phosphocreatine, ATP, and pH, in spite contribute directly to a fatal outcome. of severe anaemia and lactic acidosis (Krishna et al., 1983). In adults with severe malaria, adrenaline use, in contrast to dopamine, was associated with a worsening of lactic aci- 7.8. Anaemia and Thrombocytopaenia dosis complicating severe malaria (Day et al., 1996). In Destruction of PRBC with a fall in haemoglobin is almost children, inotropic support is infrequently required in man- an inevitable consequence of falciparum malaria. Severe agement, but dopamine would be the preferred inotrope. anaemia can be one of the life-threatening complications of Chronic salicylate toxicity has been identified in Kenya as P. falciparum infections, although in African children, it is a potential cause of lactic acidosis and hypoglycaemia in associated with a relatively low mortality on its own (Fig. some children presenting with clinical features of severe 4), but much higher when associated with CM or acidosis malaria (English et al., 1996a). Geographical variations in (Marsh et al., 1995). accessibility to aspirin and local prescribing habits may alert The anaemia could be caused by many interacting mech- clinicians to this possible cause, although in Ghana, salicylates anisms. Removal of erythrocytes by the spleen is an impor- are not used and children presenting with lactic acidosis do tant mechanism in adults with splenomegaly, in whom the not have detectable plasma levels (T. Agbenyega, T. Planche spleen is the major site of erythrocyte destruction after the and S. Krishna, personal observations). institution of therapy (Looareesuwan et al., 1987). In non- In Thai adults (n ϭ 8) with malaria and moderate meta- immune adults without CM, using a simple mathematical bolic derangement (mean arterial lactate concentrations of model, the fall in haematocrit could be described by a 3.44 mmol/L and pH values Ͼ7.3), lactate kinetics and three-term equation (Davis et al., 1990a). The initial fall in skeletal muscle lactate metabolism were examined using a haematocrit (during the first few hours) is caused by plasma lactate infusion labelled with a stable isotope (Davis et al., volume expansion; a zero-order fall with a mean half-life of 1996). Quinine treatment did not affect the plasma lactate, 25 hr occurred with the fall in parasitaemia and another fall pyruvate, or alanine concentrations or arteriovenous differ- with a half-life of 43 days is associated with a loss of ences in these metabolites measured across forearm muscle. NPRBC. However, this model relies on some tenuous as- Lactate turnover did not change significantly during acute sumptions, e.g., that the number of PRBC that sequester af- 32 C. R. J. C. Newton and S. Krishna ter treatment is small, although it is unlikely that quinine Thrombocytopaenia commonly accompanies both falci- used in this study affected sequestration during the initial parum and vivax malaria, and is due to increased consump- phases (Watkins et al., 1993). This model has not been ver- tion of platelets during the acute infection with adequate ified in adults or children with CM. numbers of megakaryocytes in the bone marrow (Srichaikul Impaired red cell production, manifesting as dyserythro- et al., 1988; Looareesuwan et al., 1992). Thrombocytopae- poiesis in chronic malarial infections (Abdalla et al., 1980), nia was associated with elevations in macrophage colony- is thought to play a role in the production of severe stimulating factor, which may mediate the increased de- anaemia in malaria. Dyserythropoiesis is a feature of chronic struction (Lee et al., 1997). In some adults, it is associated anaemia caused by P. falciparum in Gambian children (Fig. with a specific IgG attached to platelet-bound malaria anti- 11), but is not a significant factor in the fall of haemoglobin gen (Kelton et al., 1983). It was more profound in severe in acute infections (Abdalla et al., 1980). It is not confined disease and was associated with increases in plasma P-selec- to children with malaria since it is also found in children tin levels (a marker of endothelial cell activation) (Lee et with iron deficiency and sickle cell disease (Newton et al., al., 1997). Although thrombocytopaenia is common and 1997d). In Gambian children and adults with uncompli- may be associated with biochemical markers of disseminated cated malaria, the degree of dyserythropoiesis correlates intravascular coagulation, a haemorrhagic diathesis in severe with the severity of iron deficiency (Abdalla, 1990; Phillips malaria in children is rare. The degree of thrombocytopaenia et al., 1986), but in children with severe anaemia, no such is frequently mild to moderate (Ͼ50 ϫ 109/L), and in Afri- relationship was found (Newton et al., 1997d). The cause can children, is probably unrelated to disease severity. of dyserythropoiesis in these children is not clear. TNF may Changes in platelet aggregability also accompany thromb- lead to dyserythropoiesis (Clark and Chaudhri, 1988), al- ocytopaenia (Srichaikul et al., 1988). though there is no clinical evidence to support this asser- tion in Kenyan children (Newton et al., 1997d). In African 7.9. Electrolyte Abnormalities and Fluid Balance children with severe malarial anaemia, erythropoietin levels Hyponatraemia (Naϩ Ͻ135 mmol/L) occurs in over half of were appropriately raised for the haemoglobin concentration the patients with severe malaria, particularly CM (English (Burchard et al., 1995; Newton et al., 1997d), unlike Thai et al., 1996b). The cause of the hyponatraemia is contro- adults, in whom erythropoietin levels were inappropriately versial. English and colleagues argued that hyponatraemia low (Burgmann et al., 1996). results from an appropriate hormonal response to the loss of As documented by a positive direct antiglobulin test electrolyte and fluid in sweat and stool. Yet, the children (DAT), autoimmune haemolysis has been suggested as an with most severe hyponatraemia were less dehydrated (lower important component of severe malarial anaemia (Facer et urea and less weight gain after treatment). A more recent al., 1979). Several studies have not found a consistent associ- study has shown that some Kenyan children appear to have ation between a positive DAT and parasitaemia (Newton et inappropriate urinary sodium loss for the plasma sodium al., 1997d; Merry et al., 1986; Abdalla et al., 1983; Abdalla level (Sowunmi, 1996a), although osmolality was not mea- and Weatherall, 1982; Greenwood et al., 1978), although in sured. Furthermore, at least one child had features of cere- one study, a positive DAT was associated with clinical ma- bral salt wasting syndrome, and in both of these studies, uri- laria (Abdalla et al., 1983) and in others, it was associated nary sodiums are often inappropriately high. Thus, it is with anaemia (Jeje et al., 1983; Facer et al., 1979). One possible mechanism is the production of autoantibodies to triose phosphate isomerase (TPI), a glycolytic enzyme. These antibodies are associated with haemolytic anaemia in Epstein- Barr virus infection (Ritter et al., 1990), and the hereditary deficiency of this erythrocyte enzyme is associated with a neurological syndrome and increased susceptibility to infections. In P. vivax infection, no anti-TPI antibodies were detected in 5 patients. However, in 4 of 10 nonimmune patients with P. falciparum infection, who had biochemical manifestations of haemolysis (lactate dehydrogenase Ͼ200 U/L and anaemia) continuing for weeks after parasitaemia had cleared, there was a correlation between anti-TPI antibody titres and severity of anaemia (Ritter et al., 1993). Reactiva- tion of Epstein-Barr virus infection was only detectable in 1 of these 4 patients with anti-TPI antibodies, suggesting that reactivation is not necessary for this mechanism of autoimmune haemolysis to operate in malaria. The exact mecha- FIGURE 11. Bone marrow aspirate of a Kenyan child with nism of red cell damage is still unclear, and the significance malarial anaemia, showing malarial pigment (MP) and features of anti-TPI antibodies in the anaemia of malaria in African of dyserythropoiesis, such as karyorrhexis (K) and incomplete children is unknown, but worth exploring. mitosis (IM). (© C. R. J. C. Newton) Pathophysiology and Supportive Treatment of Severe Malaria 33 likely that hyponatraemia results from a combination of death in adults (Warrell, 1987; Tran et al., 1996a; Mate mechanisms that include lack of sodium intake, loss of so- Kole et al., 1996; Trang et al., 1992), but are rarely seen in dium in body fluids, including sweat and gastrointestinal African children (Waller et al., 1995; Marsh et al., 1995; losses, and inappropriate arginine vasopressin secretion. Molyneux et al., 1989b). Elevation in plasma creatinine, Whatever the mechanisms, hyponatraemia does not appear however, is common in childhood CM (Sowunmi, 1996b), to be associated with seizures, neurological damage, or and although it is associated with death (Waller et al., death, and consequently, is of little direct clinical signifi- 1995), none of the children die from renal failure. A dispro- cance. Hypernatraemic dehydration occasionally occurs in portionate increase in urea suggests that the elevated creat- African children and is often associated with a poor out- inine is probably caused by hypovolaemia (English et al., come (C. R. J. C. Newton, personal observations), al- 1996b) and aggressive intravascular volume repletion may though not studied in detail. be required in some children (English et al., 1996c). There Hypokalaemia (Kϩ Ͻ3.5 mmol/L) was present in 14% on have been no clinical trials to determine if rate of rehydra- the day of admission in survivors, but this increased to 39% tion and type of intravenous fluid therapy influence out- the following day. There was speculation that hypokalaemia come in severe disease. There is little evidence that such may contribute to a variety of complications, such as cardiac aggressive fluid therapy is detrimental to the brain in chil- dysrhythmias and late respiratory arrest. However, dysrhyth- dren with CM. Whole blood transfusion, for example, im- mias are not important complications of malaria in children proves the perfusion pressure, with a mild increase in ICP (Bethell et al., 1996), although supraventricular ectopic beats (Newton et al., 1994). Renal impairment can also be caused can be observed frequently in some children with severe dis- by a glomerulonephritis from the circulating immune com- ease, and a prolonged Q-T interval is seen in acute infections plexes, intravascular haemolysis manifesting as blackwater (von Seidlein et al., 1997). Respiratory arrest most commonly fever. The glomerulonephritis is seen in nonimmune patients follows a neurological pattern (with progressive brainstem (Anonymous, 1976; Boonpucknavig and Boonpucknavig, signs) rather than a metabolic one. Hyperkalaemia is uncom- 1988), but rarely documented in African children. mon in severe childhood malaria, and if it occurs, it is often Blackwater fever was often associated with malaria infec- associated with severe haemolysis or renal impairment. tions in the past, but the incidence appears to have fallen Other electrolyte abnormalities have been detected, e.g., (Mate Kole et al., 1996). It is more frequently seen in non- hypophosphataemia (i.e., Ͻ0.8 mmol/L phosphate in 43% immune patients (Warrell, 1987), particularly in Southeast adult patients and Ͻ0.3 mmol/L in 11/14 severely ill pa- Asia (Tran et al., 1996a), where the outcome is better than tients) (Davis et al., 1991) and less commonly hyperphos- earlier reports from Africa. In Africa, it is occasionally seen in phataemia, but their significance in the pathophysiology of children with severe malaria, but is often caused by other infection is unknown. Mildly abnormal plasma calcium infections or haemoglobinopathies (Delacollette et al., 1995). (corrected) levels and magnesium levels were commonly observed, but are unlikely to contribute significantly to pathophysiology (Davis et al., 1991). By contrast, in children 7.11. Secondary Infections with severe malaria, serum calcium, potassium, urate, and Superimposed bacterial infections (pneumonia, septicaemia, phosphate concentrations were significantly elevated in fatal and urinary tract infections) can complicate malaria in cases compared with survivors (Waller et al., 1995). Whilst nonimmune adults (Phillips et al., 1986) and may be de- mean serum calcium and potassium concentrations were tected in African children and contribute to mortality. still within the normal range in fatal cases, mean phosphate Non-typhi Salmonella septicaemia was found in Gambian and urate concentrations were increased above normal. children with malarial parasitaemia (Mabey et al., 1987), Electrolyte abnormalities can assist in choice of intrave- but septicaemia was rarely found in African children with nous fluid in the initial management of children with se- CM in one study (Molyneux et al., 1989b). vere malaria, but accurate assessment of the circulating intravascular volume is necessary to estimate the amounts of fluid needed. Intravascular circulating volume is notori- ously difficult to assess clinically in children and in severe 8. ANTIMALARIAL TREATMENT malaria, especially in hypotensive children. Central venous 8.1. Introduction pressure (CVP) measurements are probably the most accu- This article focuses on supportive care, but would be incom- rate way to estimate fluid requirements. Where possible, plete without a brief description of the main groups of anti- these measurements should be interpreted in conjunction malarial drugs. Often antimalarial treatment is unavailable with plasma electrolytes to determine the nature and rate of at the primary health care level when parents present with fluid replacement in severely ill children. their sick child, frequently after delays induced by arduous travel, the need to care for the remaining children in the family, consultation with traditional healers, and the lack of 7.10. Renal Impairment and Blackwater Fever money. In this context, a simple (nonparenteral) method of Acute renal failure and pulmonary oedema are common administering antimalarials could make the difference be- complications of falciparum malaria and a major cause of tween progression of infection to severe disease and death, or 34 C. R. J. C. Newton and S. Krishna survival. Intrarectal formulations of antimalarials currently cations, recently have been reviewed (Krishna et al., 1996) are being developed to examine this interventional strategy and their usage is briefly covered in the following sections. in the near future (Barennes et al., 1996; Hien, 1994). In essence, modern management of severe malaria con- sists of the rapid administration of an appropriate antima- 8.2. Cinchona Alkaloids larial and the treatment of complications, such as hypogly- Quinine is the preferred antimalarial for the treatment of caemia, anaemia, intravascular volume depletion, and severe malaria in both children and adults. In some coun- convulsions. Administration of a rapidly acting parasiticidal tries such as the United States, quinine has been unavail- antimalarial is, therefore, one of the immediate aims of able since 1991, and quinidine, the diastereomer of qui- management. Which antimalarial and by what dose and route nine, is the alternative (Miller et al., 1989a; Phillips et al., are questions that have to be answered individually for par- 1985). Unfortunately, with the introduction of newer anti- ticular geographic areas. All antimalarials require time in arrhythmic agents, even quinidine availability is becoming which to exert their maximum antimalarial effects. These increasingly and worryingly limited in the United States effects, and their speed of inhibition, vary between antima- (Rosenthal et al., 1996). larials and have been studied in detail in vitro (Watkins et Recent debate on quinine use has focused on the ques- al., 1993; ter Kuile et al., 1993). Thus, parasite killing, in- tion of applicability of a loading dose in severe malaria, and hibition of parasite metabolism (glycolysis, nucleic acid the exact recommended dose. In patients who have not re- synthesis, and protein synthesis), and of virulence factors ceived previous quinine treatment, we favour the use of a (such as rosetting ability and cytoadherence characteristics) loading dose to ensure a rapid increase in plasma concen- are all endpoints that have potential relevance to patho- trations in children with severe disease. In centres that lack physiology of severe disease, as well as variability in their intensive care facilities, quinine can be given by the intra- responsiveness to different classes of antimalarials (Udom- muscular rather than the intravenous route, and at a 12- sangpetch et al., 1996; Kwiatkowski and Bate, 1995). The hourly rather than an 8-hourly maintenance regimen in chil- artemisinin derivatives as a class are undisputedly the most dren for a number of reasons. First, it is simpler, with fewer rapidly acting of antimalarials (Hien and White, 1993). doses required overall. Second, in many settings, intrave- Additionally, they have the broadest stage specificity of an- nous infusions may not be monitored as closely as necessary timalarial effect. These in vitro (or ex vivo) observations are to prevent excessively rapid quinine administration and at- clearly translated into significantly faster parasite clearance tendant cardiotoxicity. Third, it may be cheaper as there is estimates in vivo compared with other antimalarials, in pa- less requirement for sterile disposables. Fourth, intravenous tients with both severe and uncomplicated malaria (Hien access is not a prerequisite, as antimalarial treatment can and White, 1993). However, in studies designed to answer begin without delay and intravenous access can be estab- the crucial question of improved survival when arteminisin lished later. Finally, there are no important pharmacoki- derivatives are compared with quinine, there is no clear netic differences in profiles and behaviour of intramuscular benefit for artemether over quinine in African children compared with intravenous quinine, even in patients who (van Hensbroek et al., 1996a) or Vietnamese adults (Tran are severely ill (Pasvol et al., 1991). A potential disadvan- et al., 1996b). Indeed, clinical measures of recovery from tage of intramuscular quinine is local toxicity, but this can disease (rather than parasitological ones) favoured the use be minimised if the quinine is diluted 1:1 (v:v) with either of quinine over artemether (shorter duration of coma and water for injection or sterile normal saline. The loading hospitalisation time, and significantly fewer seizures) (van dose is then administered divided equally into both thighs, Hensbroek et al., 1996a). These findings highlight a num- and maintenance doses are carefully alternated between ber of important general points about antimalarials. thighs. Tetanus is a rare complication, which should be Relying on parasite clearance estimates as an endpoint in avoidable by scrupulous adherence to sterile technique severe malaria is misleading. Apparent benefits such as (Yen et al., 1994). Quinine can cause hypoglycaemia by speed of parasiticidal activity in vitro do not necessarily stimulating insulin release (Section 7.6). This complication mean benefits in vivo, and there may be unsuspected disad- is much more common in adults (particularly in preg- vantages of artemether use (such as seizures) in children. nancy), but hypoglycaemia should be also excluded in chil- There is little doubt that if and when quinine becomes inef- dren wherever possible by regular monitoring. Children re- fective as an antimalarial for severe infections, then the ceiving glucose supplementation rarely develop refractory next line of treatment will rely on one of the arteminisin hypoglycaemia (Taylor et al., 1988). derivatives for which there is a growing experience in The doses and frequency of administration of different parenteral formulations. Once the toxicity of these com- regimens of quinine are summarised in Table 7. Quinine pounds has been defined better, their availability and usage should be given for 7 days when used as a single antimalar- are, therefore, likely to increase. However, for the moment, ial agent. Shorter regimens (e.g., 5 days) can be completed the drug of choice for severe malaria in most of sub-Saharan with pyrimethamine-sulphadoxine. A recent report of in- Africa, where there is established chloroquine resistance in tramuscular quinine in children with global malnutrition parasites, is quinine. The pharmacokinetics of the cinchona (not kwashiorkor) and malaria suggests that an 8-hourly alkaloids and the aminoquinolines, and their clinical appli- maintenance regimen may be preferred because of in- Pathophysiology and Supportive Treatment of Severe Malaria 35 creased quinine clearance in the malnourished group com- cent large prospective studies in African children and South- pared with controls who were not undernourished (Treluyer east Asian adults have compared intramuscular artemether et al., 1996). The rectal route for quinine administration with quinine. Artemether is similar in efficacy to quinine for was once considered too toxic, but recently, preparations of the treatment of severe malaria and is conveniently adminis- cinchona alkaloids (Ͼ95% quinine in Quinimax™) have tered. Concerns about toxicity have led to caution in recom- been successfully used after dilution in water for uncompli- mendations for use. Currently, toxicological aspects of arte- cated malaria (Barennes et al., 1996). This route has not misinin derivatives are being investigated intensively. been studied in severe malaria. Quinidine is used in lower The availability and usage of artemisinin derivatives doses than quinine because of its greater potential for car- eventually will be determined by drug regulatory bodies and diotoxicity. Optimal dosing regimens have been debated re- the WHO, but currently they are not licensed for use in cently (Krishna et al., 1996). Western Europe or the United States. They are, however, widely available in sub-Saharan Africa, but generally are 8.3. 4-Aminoquinolines more expensive than chloroquine or quinine. Pharmacoki- Chloroquine (and its structural relative amodiaquine) netic studies on these compounds have been hampered by should only be used to treat severe malaria when there is technical difficulties with assays that only recently have confidence that parasites still retain responsiveness to the been overcome, so that dosing regimens have been based drug. In areas of emerging resistance, the acute infection primarily on studies of drug efficacy. In children, arte- may be managed with chloroquine, but cure may no longer be mether can be given in a loading dose (3.2 mg/kg, intra- achievable. The choice of when to change from chloroquine muscularly) followed by a maintenance dose of 1.6 mg/kg to quinine is particularly difficult under these circumstances. daily for a minimum of 5 days. Artesunate can be given in- Quinine is the preferred antimalarial, but the financial im- travenously or intramuscularly in severe malaria (loading plications of a change in policy can be devastating in some dose of 2.4 mg/kg, followed after 12 hr by 1.2 mg/kg, then at sub-Saharan countries. 24 hr by 1.2 mg/kg, and 1.2 mg/kg every 24 hr for 7 days). Chloroquine’s pharmacokinetics dictate that small fre- Artesunate tablets can be substituted when tolerated. quent dosing is preferable to larger, less frequent dosing, as These regimens may be modified as pharmacological and chloroquine has a small central compartment relative to its toxicological experience accumulates. volume of distribution. Higher doses of chloroquine, therefore, can give dangerously high peak concentrations. Chlo- 9. ADJUNCTIVE MEASURES roquine can be administered by a variety of routes in severe Good nursing care is crucial to the management of children malaria (subcutaneously, intramuscularly, intravenously, or with severe malaria (Murphy et al., 1995a; Krishna and even via nasogastric tube) and is absorbed rapidly from White, 1989). Children should be transferred to a dedi- parenteral sites. Our preferred regimen is via the intramus- cated (intensive care type) unit wherever possible and cular or subcutaneous route (3.5 mg base/kg), given every 6 hr cared for by experienced nurses. Those children who de- to a total dose of 25 mg/kg. Amodiaquine is a prodrug for velop signs of deepening coma or abnormal brainstem signs desethylamodiaquine, which has similar (though not iden- should be considered for assisted ventilation, if facilities ex- tical) pharmacokinetic properties to chloroquine. ist for this supportive measure. These children should not be made normocapnic, since this may precipitate transten- 8.4. Artemisinin Derivatives torial herniation (Looareesuwan et al., 1995). Children The artemisinin derivatives, like quinine, are plant extracts should be monitored regularly for vital signs and the com- that have had a long-standing role in the treatment of ma- plications of hypoglycaemia and seizures. A nasogastric laria in China, but are now much more widely available. tube should be inserted in those with impaired conscious- There are a variety of formulations of the arteminisin deriva- ness or those at risk of aspiration pneumonitis, and eye care tives, which are principally distinguished by their solubility and oral hygiene measures should be instituted. in oil or water. These formulations (e.g., artemether or artemisinin, which are oil soluble, and artesunate, which is water 9.1. Antipyretics soluble) have been administered via oral, rectal, intravenous, Reasons to treat fever in malaria include making the pa- and intramuscular routes. In severe malaria, a number of re- tient more comfortable, minimising metabolic stresses of

TABLE 7. Dose Regimen of Quinine Dihydrochloride in Severe Falciparum Malaria Reference Route Loading dose Maintenance dose

Winstanley et al., 1993 Intravenous 7.5 mg/kg in 30 min, then 7.5 mg/kg over 1.5 hr 10 mg/kg over 2 hr every 12 hr Winstanley et al., 1994 Intravenous 15 mg/kg over 2 hr 10 mg/kg over 2 hr every 12 hr Waller et al., 1990 Intramuscular 20 mg/kg (2-site injection) 10 mg/kg every 8 hr Pasvol et al., 1991 Intramuscular 20 mg/kg (2-site injection) 10 mg/kg every 12 hr 36 C. R. J. C. Newton and S. Krishna infection, and perhaps reducing the risk of convulsions and vention of malaria-associated seizures, phenobarbitone is neurological sequelae in children. There are no studies in the only drug that has been studied in any detail. In a ran- malaria that have shown an improvement in these parame- domised double-blind study of 48 nonimmune Thai adults ters in parallel with a reduction in fever, although nonim- with CM, phenobarbitone (3.5 mg/kg, intramuscular) re- mune adults with uncomplicated malaria reported an im- duced the incidence of seizures after admission from 54 to provement in headaches following the use of ibuprofen 12.5% (White et al., 1988). This effect was surprising since (Krishna et al., 1995b). It is not clear if a reduction in core this dose is not usually associated with anticonvulsant ef- temperature benefits cerebral consequences of severe malaria. fects (Shorvon, 1994), although plasma phenobarbitone Whenever they have been studied, antimalarials, such as concentrations were not measured in this study. A slightly chloroquine and quinine, have not been found to have use- higher dose (5 mg/kg, intramuscular) administered to Gam- ful antipyretic properties (Krishna et al., 1995b; Vani- bian children with CM did not cause a deterioration in janonta et al., 1996) contrary to dogma. Specific anti- level of consciousness, and subsequently, studies using 7 mg/ pyretic treatment is often used as an adjunct to antimalarial kg of phenobarbitone showed it was clinically well toler- treatment. In most developing countries, although aspirin ated, even when other anticonvulsants had to be added is frequently used, paracetamol is the antipyretic of choice. (S. Krishna, personal observation). In suppository formulation, paracetamol is convenient to A small study of Southeast Asian children with severe administer. Recently, however, paracetamol has been shown falciparum malaria demonstrated that phenobarbitone (also to prolong parasite clearance in Gabonese children with used at 7 mg/kg, intramuscular) was well absorbed, reaching non-severe malaria, possibly from a decreased production of a peak (7 mg/L) within 4 hr, but failing to reach concentra- TNF-␣ and ROS (Brandts et al., 1997). In this study, it of- tions above 15 mg/L (ter Kuile et al., 1992), the level thought fered no benefit over mechanical antipyresis. However, oral to prevent febrile seizures (Faero et al., 1972). This dose was paracetamol reduces fever in Thai adults with uncompli- associated with depression of consciousness level in some cated malaria (Krishna et al., 1995b), and ibuprofen (10 patients. In 12 Kenyan children with severe malaria, a mg/kg), a nonsteroidal anti-inflammatory drug, is signifi- higher dose (10 mg/kg, intramuscular) only achieved maxi- cantly more effective (Krishna et al., 1995b). Indomethacin mal concentrations above 15 mg/L in 4 patients, and did not is also superior to paracetamol in adults with uncompli- appear to reduce the incidence of seizures or affect the dura- cated infection. Fever begins to recur 3–4 hr after paraceta- tion and depth of coma (Winstanley et al., 1992). Currently, mol (15 mg/kg), so the dosing frequency may have to be up a larger prophylactic dose is undergoing clinical trials. to 4 hourly until the underlying pyretic tendency subsides. Seizures in severe malaria may abort spontaneously, but Ibuprofen has a more sustained antipyretic action and, data from other encephalopathies suggests that seizures lasting therefore, can be given less frequently (6–8 hourly). longer than 5 min should be actively terminated. Diazepam Although salicylates are not recommended in children and paraldehyde are effective drugs for the termination of because of the danger of provoking Reye’s syndrome, in seizures, although there have been no studies conducted on many countries aspirin is still used as an antipyretic. Addi- their relative efficacy in children with malaria. In many tionally, aspirin may aggravate or even cause metabolic aci- hospitals, these drugs, particularly diazepam, may be given dosis (English et al., 1996a) or, rarely, precipitate a bleed- in inappropriate doses by intramuscular injection or too ing tendency (Warrell et al., 1990). Parenteral antipyretics, rapidly by the intravenous route, with the attendant risk of e.g., dipyrone, are not widely used because of a significant respiratory depression. Diazepam-induced respiratory de- risk of fatal agranulocytosis (Warrell et al., 1990). Mono- pression is rapidly reversed by flumazenil, but this drug is clonal antibodies against TNF reduce the temperature in expensive and not widely available. If diazepam overdose is Gambian children, but are unlikely to gain currency as an- suspected, children, therefore, should be ventilated by bag tipyretics. The role of other measures, e.g., fanning and tepid and mask or endotracheal tube until spontaneous respira- sponging, are controversial since the rate of cooling is not tions return. Paraldehyde is well absorbed from intramuscu- significantly different from acetaminophen alone, although lar injections and rectal administration (Shorvon, 1994). It if these measures are used together, then they appear to re- is generally safe (C. R. J. C. Newton, personal observa- duce the temperature more quickly than either alone tions), although it causes sedation and may prolong coma (Steele et al., 1970). Fanning and sponging may not reduce following the post-ictal phase. Other side effects are rare core temperatures [Royal College of Physicians (London), and include hypotension, metabolic acidosis, and pulmo- 1991], and may even aggravate hyperthermia (especially if nary oedema (Shorvon, 1994). The rectal administration of ice water is used) by inducing vasoconstriction. an outdated drug has caused severe proctitis, excoriating rash, and large-bowel perforation. The chemical interac- tion of paraldehyde with plastic syringes has limited the 9.2. Anticonvulsants drug’s use, but if paraldehyde is quickly administered after It is prudent to prevent or terminate seizures since they are being drawn up, there appear to be no adverse effects. associated with a poor outcome in CM and cause neuronal The treatment of repeated or continuous seizures not damage in other encephalopathies, as well as in experimen- controlled with diazepam or paraldehyde should follow the tal models (Wasterlain and Shirasaka, 1994). In the pre- standard treatment of status epilepticus. This may require Pathophysiology and Supportive Treatment of Severe Malaria 37 access to facilities for assisted ventilation. Phenobarbitone in experimental models have been shown to be associated and phenytoin have both been used successfully, although with raised ICP. Also, PRBC are particularly susceptible to not directly compared in malaria. Phenobarbitone can be lysis with osmotic agents (K. Silamut, N. J. White and S. given in a loading dose of 15–20 mg/kg (preferably intrave- Krishna, unpublished observations), although the concen- nously, but may also be given intramuscularly), followed by trations required for lysis are significantly higher than can be 5 mg/kg 12 hr later. Toxic effects of phenobarbitone in- achieved with the safe administration of these agents in hu- clude decreasing the level of consciousness, respiratory de- mans. Osmotic diuretics have been used in African chil- pression, and hypotension (reputably after diazepam); al- dren with CM, but their efficacy is difficult to assess since though these effects are rarely seen in status epilepticus randomised control trials have not been conducted. King- (Shorvon, 1994). Phenytoin (20 mg/kg) should be infused ston (1971) reported an improved outcome in Liberian over 20 min into a line that does not contain dextrose (to children who were given 30% urea in 10% invert sugar af- prevent precipitation), preferably with the patient’s ECG ter the children had deteriorated or were in prolonged monitored for excessive QTc prolongation, which predis- coma. Commey et al. (1980) thought that mannitol (1 g/kg poses to ventricular arrhythmias. Phenytoin has relatively every 8 hr) improved the level of consciousness and per- little respiratory and cerebral depression, although it may haps outcome in Ghanaian children (Commey, 1984). In potentiate the effects of benzodiazepines and barbiturates. If Kenyan children, mannitol (0.5–1.0 g/kg) reduced ICP and seizure control is not achieved with the use of one of these improved CPP in all cases, although in children with severe drugs, the second should be added, although not in a solu- IH, this caused only transient improvements and did not tion containing other drugs, as precipitation may occur. prevent the development of intractable IH. A major ques- Other measures (e.g., thiopentone, hypothermia) may oc- tion to arise from this study was whether osmotherapy con- casionally be necessary. tributed to the outcome in the group with intermediate IH. In 74% of children, the ICP was greater than 20 mm Hg for more than 15 min, a level that would be actively treated in 9.3. Measures to Reduce Raised Intracranial Pressure most intensive care units. All of these children except one Of the many modalities for reducing ICP, osmotic diuretics were given mannitol, and 70% had a good outcome. It was are the only agents that have been tested in malaria. Man- not possible to determine if severe IH was a consequence of nitol and glycerol are thought to reduce ICP by creating an untreated intermediate IH or whether severe IH reflects the osmotic gradient across the BBB, thereby causing extracel- development of widespread cerebral damage secondary to lular fluid to be drawn into the intravascular space (Go, other concurrent pathogenic processes. At present, there 1991). The BBB must be relatively impermeable to the are not enough data to recommend an empirical regimen agent for the gradient to develop. The decrease in ICP is for control of ICP. correlated with the increase in plasma osmolality that these The effectiveness of other modalities that can be used to agents achieve, although the changes in ICP do not parallel reduce ICP has not been tested in CM. Two controlled tri- the changes in osmolality (Heinemeyer, 1987). Thus, these als of steroids have been conducted in Asian adults (Hoff- agents may act by other mechanisms. For example, Muize- man, S. L. et al., 1988; Warrell et al., 1982a), but the effect laar et al. (1984) have suggested that mannitol causes vaso- on ICP was not monitored, and steroids did not reduce mor- constriction by reducing blood viscosity, if autoregulation is tality. Hyperventilation has not been studied in CM since intact. The vasoconstriction reduces cerebral blood volume most patients are not ventilated. Indeed, spontaneous hy- and thus, ICP. In general, the reduction of ICP following perventilation, secondary to metabolic acidosis, may reduce the administration of osmotic agents depends upon the ICP the ICP and, therefore, increasing the PaCO2 in ventilated level at which the treatment is instituted, the rate of patients may increase the blood volume and precipitate tran- change in the osmolality, and the degree of hyperosmolality stentorial herniation in adults (Looareesuwan et al., 1995). induced (McGraw et al., 1978). Osmotic diuretics are used in children with head injury (Muizelaar et al., 1984; Marshall et al., 1978) and Reye’s 9.4. Correction of Hypoglycaemia syndrome. In Reye’s syndrome, mannitol, in doses as low as Theoretically, correcting hypoglycaemia in the presence of 0.25 g/kg, reduces ICP (Brown et al., 1978), although re- tissue hypoxia can worsen tissue acidosis, as original studies peated doses are often required (Marshall et al., 1978), and by Seisjo and colleagues in animal models clearly demon- some clinicians have used continuous infusions of mannitol strated (Siesjo, 1978). However, similar studies monitoring to control raised ICP (Berman et al., 1975). Mannitol can tissue pH responses to correction of hypoglycaemia have make the BBB more permeable and may cause a rebound not been carried out in patients with CM, and would be phenomenon in which the ICP rapidly rises after an initial technically and ethically challenging. Consequently, indi- fall. This is thought to occur because the mannitol passes rect measurements monitoring, for example, plasma lactate into the interstitial space of the brain, drawing fluid from concentrations after administration of hypertonic glucose the intravascular space (Go, 1991). in adult patients with severe malaria suggest this procedure is Osmotic agents may also have other effects that contrib- not associated with major adverse systemic effects (Pukritta- ute to a reduction in ICP. Mannitol scavenges ROS, which yakamee et al., 1991a). 38 C. R. J. C. Newton and S. Krishna

The correction of hypoglycaemia (blood sugar Ͻ2.2 lies on the effective treatment of the underlying disease, as mmol/L) or incipient hypoglycaemia (blood sugar Ͼ2.2–3.3 well as measures designed to optimise oxygen delivery to mmol/L) requires hypertonic glucose (1 mL/kg of 50% dex- tissues and to improve tissue perfusion (Stacpoole et al., trose or 2 mL/kg of 25% dextrose). The maintenance require- 1994). In severe malaria, this involves the rapid administra- ments of glucose in children with severe malaria are much tion of antimalarials, correction of intravascular volume de- higher than for uninfected children. In uncomplicated infec- pletion and severe anaemia, anticonvulsant therapy if indi- tion in Kenyan children, the rate of glucose utilisation was 5.0 cated, and general measures such as reduction in high (4.1–8.4) mg/kg/min (Dekker et al., 1996). Guidelines from temperature. However, in a significant proportion of pa- studies carried out in adults suggest glucose should be replaced tients, severe lactic acidosis still persists. In this group, an mg/kg/min, and are presently a reasonable basis for de- important unresolved question is whether lactic acidosis per 3ف at termining glucose infusion rates (Davis et al., 1993). se can contribute directly to mortality from severe disease. Other options have been suggested for correction of hy- This question can only be adequately resolved by studying poglycaemia in severe malaria, but none are appropriate for lactate metabolism in severe malaria, and then by interven- African children presenting with this complication. Thus, ing with a specific therapy to assess its impact on mortality. in Thai patients with hyperinsulinaemic hypoglycaemia due Dichloroacetate (DCA) acts specifically to stimulate the to quinine, Sandostatin (SMS 201-995, a long-acting so- rate-limiting enzyme complex pyruvate dehydrogenase, matostatin analogue) given with glucagon boluses effec- which regulates the entry of pyruvate into Krebs’ cycle tively restored plasma glucose levels and reduced circulat- (Stacpoole, 1989). Its mechanism of action is shown in Fig. ing insulin concentrations (Phillips et al., 1993). However, 12. DCA maintains pyruvate dehydrogenase kinase in its the treatments are expensive and have not been studied in active dephosphorylated state, allowing greater utilisation African children, where the natural history of postadmis- of pyruvate by aerobic metabolism, concomitantly reducing sion hypoglycaemic episodes is somewhat different. Simi- acid (i.e., free protons), as well as the lactate anion. The larly, galactose infusions in adults are readily converted to pharmacokinetics and pharmacodynamics of DCA have glucose, but similar studies are not available in African been studied in great detail in animal models and a variety children (Pukrittayakamee et al., 1991b). of disease states in humans, and its safety in severe disease is established (Krishna et al., 1995a; Stacpoole, 1989). DCA was first examined in malaria in a P. berghei young 9.5. Acidosis rat model of severe metabolic dysfunction, where it attenu- No human studies have demonstrated an improvement in ated rapidly rising lactate levels when given with quinine mortality when lactic acidosis is treated with adjunctive (Holloway et al., 1991). Subsequent studies established that therapies (Stacpoole et al., 1992). The initial management DCA could ameliorate malaria-associated lactic acidosis sig- of patients with lactic acidosis from any cause, therefore, re- nificantly more rapidly in both adults and children (Krishna

FIGURE 12. Site of action of DCA. Pathophysiology and Supportive Treatment of Severe Malaria 39 et al., 1994a, 1995a). These studies also examined the pharma- bolic abnormalities (Krishna et al., 1994a), whatever the cokinetics of DCA in severe disease, as well as its efficacy. postulated mechanisms of action of these interventions In order to justify further studies in patients, a large prospec- (which include the correction of anaemia as one possibil- tive single-blind mortality study was carried out in the ro- ity). In symptomatic children, particularly those with signs dent model of severe malaria (Holloway et al., 1995). Rats of severe disease, it is reasonable to transfuse blood (or with lactic acidosis due to P. berghei infections (n ϭ 183 packed red cells) as soon as practicable. CVP guidance can with lactate Ͼ5 mM) were randomised to receive either a assist in assessing additional requirements for crystalloid loading dose of quinine (and saline placebo, n ϭ 84) or qui- (e.g., if CVP readings are Ͻ5 cm). The risk of transfusion- nine with DCA (n ϭ 99). Mortality was monitored for 2 associated pulmonary oedema appears to be low. parasite life cycles (50 hr). DCA treatment significantly re- The role of exchange transfusion in malaria is more con- duced mortality (by 33%, odds ratio Ͼ2.2, P Ͻ0.021) com- troversial. Exchange transfusions are usually partial rather than pared with placebo. These findings confirmed that DCA as an complete when carried out in the context of malaria treat- adjunct therapy with quinine in severe malaria has the po- ment. Automated equipment to assist exchanges, the require- tential to reduce mortality in patients. Furthermore, the pre- ment for larger quantities of safe blood, difficulties with venous vious studies in patients have already established that in spite access (or the higher cost of central lines), and more inten- of lactic acidosis, aerobic metabolism of lactic acid can be sive monitoring rule out exchange transfusion as an option in augmented by the addition of DCA, implying that oxygen the management of malaria in most children. Even if avail- delivery to tissues may not be the limiting factor in allowing able, should exchange transfusion be used in spite of these aerobic metabolism in some patients. In order to test this disadvantages? In the absence of guidelines from any ade- hypothesis further and to assess the relevance of DCA in se- quately powered prospective studies (Looareesuwan et al., vere malaria, a large placebo-controlled, double-blind mor- 1990), the requirement for exchange is a matter for individual tality study of DCA is now underway in Ghanaian children. clinical judgement. The potential benefits are rapid removal Sodium bicarbonate has been considered useful in the of high circulating parasitaemia at a satisfyingly predictable past in the management of lactic acidosis complicating se- rate (Wilkinson et al., 1994; Van den Ende et al., 1994) (and vere disease states. However, its use may be harmful be- theoretically, perhaps undefined plasma associated “tox- cause of the dangers of volume overload and paradoxical ins”); the rapid correction of severe anaemia, whilst mini- tissue acidification (Stacpoole, 1986). We do not recom- mising the risk of circulatory overload; and in Africa, the mend its use in malaria-associated lactic acidosis or, indeed, infusion of antibodies, which may assist the infant’s response in lactic acidosis complicating any severe disease. Other to malaria infection. However, the circulating parasitaemia agents such as carbicarb have not been studied in malaria. can effectively be managed with antimalarials (e.g., it responds particularly quickly to artemisinin derivatives) and is a poor surrogate marker of prognosis or disease severity, as it does 9.6. Anaemia and Exchange Transfusion not represent the sequestered parasite biomass. Severe The role of blood transfusions in the treatment of severe anaemia is safely corrected by conventional blood transfu- malaria remains controversial. Blood transfusions are ex- sion, and in semi-immune children, high parasitaemias can pensive and potentially dangerous in many malaria-endemic be surprisingly well tolerated. If available, then the next de- areas, since a high proportion of blood donors are hepatitis cision is at what level of parasitaemia should exchange B surface antigen and human immunodeficiency virus- transfusion be contemplated? If the patient is not only hyper- seropositive, and screening for these pathogens is not al- parasitaemic, but clinically very ill (deep coma, hypoglycae- ways possible. Other viruses (such as hepatitis C) are also mia, acidosis, or severe anaemia), then we suggest exchanges common, but not routinely tested for in blood banks. For should be considered at parasitaemias of 20% and should be these reasons, most clinicians prefer to restrict blood trans- more likely at parasitaemias Ͼ50%. However, the level of fusions to individuals with clear signs of cardio-respiratory parasitaemia by itself should never be used as the sole criterion compromise, hyperparasitaemia, or CM and associated on which to base a decision to proceed to exchange transfu- anaemia (English et al., 1996c; Newton, 1992; Lackritz et sion. A controlled clinical trial of partial exchange transfu- al., 1992). sion in severely ill children with high parasitaemias is long In a study of 24 Kenyan children, metabolic acidosis was overdue, but unlikely to be implemented in the near future. attributed to anaemia in 80% of the patients (English et al., 1996b). In many children, the acidosis responded to aggres- of the 9.7. Desferrioxamine %20ف sive intravascular volume repletion, although children continued to have sustained hyperlactataemia 4 hr Desferrioxamine has been studied as an adjuvant agent in after CVP guided treatment was begun, and unfortunately, malaria. It is an iron chelating agent, used in conditions the study lacked a control comparison group. Blood (20 such as haemochromatosis and thalassaemia requiring regu- mL/kg) was transfused if haemoglobin was 5 g/dL or below. lar transfusions, and successfully induces negative iron bal- These findings independently confirm that in the majority ance when used chronically. Desferrioxamine reduces the of survivors with metabolic acidosis due to malaria, simple formation of ROS by reducing the amount of free iron. In resuscitative measures are sufficient to rapidly reverse meta- vitro and animal studies showed that desferrioxamine has 40 C. R. J. C. Newton and S. Krishna antimalarial properties in its own right (Whitehead and ity, decreasing systemic vascular resistance, and impairs Peto, 1990), probably by withholding iron from the parasite platelet aggregation, thus improving microcirculatory flow. It (Peto and Thompson, 1986; Hershko and Peto, 1988). Re- has been shown to be useful in animal models of ischaemia, cently, a third mechanism has been proposed, i.e., desferri- particularly during reperfusion. It is useful in the manage- oxamine enhances interferon-␥ activity and macrophage ment of peripheral vascular disease, but studies in human activation, and those effects promote the destruction of cerebrovascular disease have been unconvincing. It prevents parasites through NO (Weiss et al., 1997). This may lead to the neurological signs in mice infected with P. berghei, possi- enhancement of T1-cell-mediated immunity and possibly bly by inhibiting the release of TNF (Kremsner et al., 1991). to impaired T2-cell responses (Thuma et al., 1996). Furthermore, it inhibits rosette formation and promotes red These observations encouraged the trial of desferriox- blood cell detachment from rosettes in blood taken from Ga- amine in uncomplicated infections, and more recently as an bonese children with nonsevere malaria (Lehman et al., adjunctive therapy in Zambian children with CM (Gor- 1997). It has been studied prospectively in a trial of African deuk et al., 1992). In adults with mild or moderate malaria, children with CM, where it appeared to reduce the duration desferrioxamine (100 mg/kg/day) enhanced the clearance of coma (Di Perri et al., 1995), but this study was too small to of asexual parasites in the absence of other antimalarials study effects on mortality with confidence. A study in non- (Mabeza et al., 1996), although recrudescences were com- immune adults was abandoned because patients had more mon. In a controlled trial as an adjuvant treatment to qui- gastrointestinal complaints and there was no effect on in- nine in Zambian children with CM, it appeared to reduce flammatory markers (Hemmer et al., 1997). the duration of deep coma, and perhaps hasten the clearance of parasitaemia, although the study did not show any 9.9.2. Miscellaneous agents. Heparin and dextran have difference in outcome or neurological sequelae (Gordeuk et been suggested as agents that may improve microcirculatory al., 1992). Gordeuk and colleagues also showed that the du- flow by reducing the viscosity of blood. No large controlled ration of coma was inversely proportional to the transferrin studies have been conducted. In one study, heparin (300 levels in these children (Gordeuk et al., 1995) and that desfer- IU/kg intramuscularly/day) given to 21 Indonesian children rioxamine was associated with persistently raised neopterin with CM, as adjunct therapy to quinine did not signifi- levels (Weiss et al., 1997). The significance of these clinical cantly improve the outcome compared with controls or findings is not entirely clear, particularly because children those given dexamethasone (Rampengan, 1991). It is un- did not receive a loading dose of quinine and control patients clear if these children were adequately heparinised, but it is had unusually prolonged coma recovery times. Desferrioxam- unlikely that a significant difference would be found in this ine has been combined safely with artesunate in the treat- study (Type II error). Similarly, there is no evidence that ment of patients with uncomplicated malaria (Looareesu- dextran improves outcome in CM, although again, proper wan et al., 1996), and further trials in CM are in progress. controlled studies have not been conducted. Indeed, a recent report on 352 children with CM concluded that desferrioxamine given as an adjunct therapy to a loading dose of quinine did not reduce mortality when 9.10. Other Agents compared with placebo (Thurma et al., 1998). Prostacyclin, an inhibitor of platelet aggregation, has been described in a single case report (Weston et al., 1982), but it is impossible to assess the effect on the outcome of the patient. 9.8. Anti-Inflammatory Agents 9.8.1. Corticosteroids. Steroids have long been advocated for the treatment of CM, but two controlled trials in Southeast 10. DISCUSSION AND FUTURE STUDIES Asian adults have shown no benefit (Hoffman, S. L. et al., Many potentially important pathophysiological events have 1988; Warrell et al., 1982a). The earlier trial concluded that been identified in severe malarial infection, but the testing dexamethasone (1–2 mg/kg over 48 hr) was deleterious be- of appropriate interventions to reverse them still remains cause it prolonged coma in survivors and there was an in- one of the most challenging tasks for investigators. In order creased incidence of side effects, such as pneumonia and gas- to test adjunctive therapies rigorously, very large clinical trointestinal bleeding (Warrell et al., 1982a). The other trials are needed, and the infrastructure and expertise to im- study used larger doses (11.4 mg/kg dexamethasone over 48 plement such studies, even if funding were to be made hr), but was smaller and included patients who were not available, is frequently lacking. Thus, to demonstrate a re- strictly defined as having CM. No studies have been reduction in mortality with an adjunctive intervention effec- ported in African children. tive at reducing mortality by 20% in severe disease where Ͼ1000 patients will ,%25ف the underlying mortality is 9.9. Agents that Improve Microcirculatory Flow need to be enrolled. Recent studies have shown that many 9.9.1. Pentoxifylline. Pentoxifylline (or oxpentifylline) is of the surrogate markers that have been considered useful in a methylxanthine derivative that has many effects of poten- monitoring response to therapy in severe malaria may not tial benefit in the management of severe malaria. It is a va- correlate with the most important outcome measure, mortal- sodilator that reduces red cell deformability and blood viscosity. Thus, duration of coma, estimates of parasite clearance Pathophysiology and Supportive Treatment of Severe Malaria 41 time, and even perhaps seizures, although significantly differ- References ent between treatment groups in one large study, were not as- Abdalla, S. H. (1990) Iron and folate status in Gambian children sociated with any difference in mortality (Tran et al., with malaria. Ann. Trop. Paediatr. 10: 265–272. 1996b). It is, therefore, inappropriate to generalise from im- Abdalla, S. and Weatherall, D. J. (1982) The direct antiglobulin provements in surrogate markers of infection (including pe- test in P. falciparum malaria. Br. J. Haematol. 51: 415–425. Abdalla, S., Weatherall, D. J., Wickramasinghe, S. N. and ripheral parasitaemia) and to anticipate corresponding im- Hughes, M. (1980) The anaemia of P. falciparum malaria. Br. J. provements in survival. Therefore, interventional studies on Haematol. 46: 171–183. severe malaria designed to assess usefulness in management Abdalla, S. H., Kasili, F. G. and Weatherall, D. J. (1983) The (as opposed to answering pathophysiological questions or pi- Coombs direct antiglobulin test in Kenyans. Trans. R. Soc. lot Phase II studies), perforce, will have to be sufficiently Trop. Med. Hyg. 77: 99–102. large as to assess mortality, or they should not be conducted Adams, J. H., Sim, B. K., Dolan, S. A., Fang, X., Kaslow, D. C. and at all. Miller, L. H. (1992) A family of erythrocyte binding proteins of Much of what has been learned about disease mecha- malaria parasites. Proc. Natl. Acad. Sci. USA 89: 7085–7089. nisms and management of severe malaria has been derived Agbenyega, T., Angus, B., Bedu-Addo, G., Baffoe-Bonnie, B., from studies carried out in adults in Southeast Asia. As em- Griffin, G., Vallance, P. and Krishna, S. (1997) Plasma nitro- phasised in this review, there are important differences be- gen oxides and blood lactate concentrations in Ghanaian children with malaria. Trans. R. Soc. Trop. Med. Hyg. 91: 298– tween patterns of disease in African children compared 302. with adults. Such differences may reflect age, genotypic Ahmad, S. H., Moonis, R., Kidwai, T., Khan, T. A., Khan, H. M. variations, and patterns of transmission, as well as immuno- and Shahab, T. (1986) Cerebral malaria in children. 53: logical responses. In order to optimise the treatment of se- 409–413. vere malaria in children, more studies are necessary in this Aikawa, M. (1988) Human cerebral malaria. Am. J. Trop. Med. population. These studies should be designed to examine Hyg. 39: 3–10. the pathophysiological basis of severe disease, as well as Aikawa, M., Suzuki, M. and Gutierrez, Y. (1980) Pathology of newer treatment regimens, particularly adjunctive thera- malaria. In: Malaria. Pathology, Vector Studies, and Culture, pies, and are urgently needed. Many prognostic indicators, Vol. 2, pp. 47–102, Kreier, J. P. (ed.) Academic Press, New York. both laboratory and clinical, are now available to identify Akpede, G. O., Sykes, R. M. and Abiodun, P. O. (1993) Convul- children at risk of dying from severe infection. In assessing sions with malaria: febrile or indicative of cerebral involve- potentially new indicators that may then point to future in- ment? J. Trop. Pediatr. 39: 350–355. Allan, R. J., Rowe, A. and Kwiatkowski, D. (1993) Plasmodium fal- terventional studies, it is important to relate these indica- ciparum varies in its ability to induce tumor necrosis factor. tors to those that already have been identified. This ap- Infect. Immun. 61: 4772–4776. proach of assimilating new correlates of severe disease into Allan, R. J., Beattie, P., Bate, C., van Hensbroek, M. B., Morris- what is already known about infection relies on more diffi- Jones, S., Greenwood, B. M. and Kwiatkowski, D. (1995) Strain cult statistical techniques. It may be more statistically de- variation in tumor necrosis factor induction by parasites from manding, but it is also potentially more valuable in assess- children with acute falciparum malaria. Infect. Immun. 63: ing a new prognostic indicator’s true value. 1173–1175. The recent advances in understanding the molecular ba- Allen, S. J., O’Donnell, A., Alexander, N. D. E. and Clegg, J. B. sis of many of the central pathophysiological events in se- (1996a) Causes of coma in children with malaria in Papua New vere malaria suggest interventions that are directed at re- Guinea. Lancet 348: 1168–1169. versing these processes. We know that antimalarial drugs Allen, S. J., O’Donnell, A., Alexander, N. D. E. and Clegg, J. B. (1996b) Severe malaria in children in Papua New Guinea. Q. J. kill parasites, but require hours before they can exert their Med. 89: 779–788. maximum effects. During this time, it may be advantageous Alonso, P. L., Smith, T., Armstrong Schellenberg, J. R., Masanja, to employ interventions that can immediately reverse, for H., Mwankuyse, S., Urassa, H., de Azevedo, I. B., Chongela, J., example, the cytoadherence of PRBCs from capillary beds Kobero, S. and Menendez, C. (1994) Randomised trial of effi- in the brains of patients with CM. Such approaches have cacy of SPf66 vaccine against Plasmodium falciparum malaria in been attempted in the past (the use of hyperimmunoglobu- children in southern Tanzania. Lancet 344: 1175–1181. lin is one example), and a hope for the future is that more Al Yaman, F. M., Genton, B., Mokela, D., Raiko, A., Kati, S., specific therapies guided by current molecular understand- Rogerson, S., Reeder, J. and Alpers, M. (1995) Human cerebral ing may fare better. Before such approaches can be imple- malaria: lack of significant association between erythrocyte mented at the bedside, however, the importance of many rosetting and disease severity. Trans. R. Soc. Trop. Med. Hyg. new and exciting laboratory observations needs to be exam- 89: 55–58. Al Yaman, F. M., Mokela, D., Genton, B., Rockett, K. A., Alpers, ined in geographically different patient populations. M. P. and Clark, I. A. (1996) Association between serum levels of reactive nitrogen intermediates and coma in children with Acknowledgements–We thank all of our colleagues with whom we have car- cerebral malaria in Papua New Guinea. Trans. R. Soc. Trop. ried out investigations into malaria, in Kenya, Thailand, The Gambia, and Med. Hyg. 90: 270–273. Ghana, and Professor M. Bland, St. George’s Medical School, for help in Al Yaman, F. M., Rockett, K. A. and Clark, I. A. (1997) Reactive constructing Table 3. CRJCN holds a Wellcome Trust Career Post in Tropical Medicine (No. 050533) and SK is a Wellcome Trust Senior Re- nitrogen intermediates and cerebral malaria: a reply. Trans. R. search Fellow in Clinical Science. Soc. Trop. Med. Hyg. 91: 92–93. 42 C. R. J. C. Newton and S. Krishna

Angus, B. J., Thanikkul, K., Silamut, K., White, N. J. and Udom- Berner, H. (1975) Malaria bei Kindern unter besonderer berück- sangpetch, R. (1996) Short report: rosette formation in Plasmo- sichtigung der Cerebralen Malaria. Klin. Paediatr. 187: 401–406. dium ovale infection. Am. J. Trop. Med. Hyg. 55: 560–561. Bethell, D. B., Phuong, P. T., Phuong, C. X., Nosten, F., Waller, Anonymous (1976) Renal lesions in human malaria. Br. Med. J. D., Davis, T. M., Day, N. P., Crawley, J., Brewster, D., Pukrit- 272: 132. tayakamee, S. and White, N. J. (1996) Electrocardiographic Anonymous (1987) Man over monkey [editorial]. Lancet i: 1016. monitoring in severe falciparum malaria. Trans. R. Soc. Trop. Anstey, N. M., Weinberg, J. B., Hassanali, M. Y., Mwaikambo, E. D., Med. Hyg. 90: 266–269. Manyenga, D., Misukonis, M. A., Arnelle, D. R., Hollis, D., Bloland, P. B., Lackritz, E. M., Kazembe, P. N., Were, J. B., Steke- McDonald, M. I. and Granger, D. L. (1996) Nitric oxide in tee, R. and Campbell, C. C. (1993) Beyond chloroquine: impli- Tanzanian children with malaria: inverse relationship between cations of drug resistance for evaluating malaria therapy efficacy disease severity and nitric oxide production/nitric oxide syn- and treatment policy in Africa. J. Infect. Dis. 167: 932–937. thetase type 2 expression. J. Exp. Med. 184: 557–567. Bonadio, W. A. (1992) The cerebrospinal fluid: physiologic Arieff, A. I. and Griggs, R. C. (1994) General considerations in aspects and alterations associated with bacterial meningitis. metabolic encephalopathies and systemic disorders affecting Pediatr. Infect. Dis. J. 11: 423–432. the nervous system. In: Metabolic Brain Dysfunction in Sys- Bondi, F. S. (1992) The incidence and outcome of neurological temic Disorders, pp. 1–20, Arieff, A. I. and Griggs, R. C. (eds.) abnormalities in childhood cerebral malaria: a long term follow- Little, Brown & Company, Boston. up of 62 surviviors. Trans. R. Soc. Trop. Med. Hyg. 86: 17–19. Armengaud, M., Louvain, M. and Diop-Mar, L. (1962) Étude Boonpucknavig, V. and Boonpucknavig, S. (1988) The histopa- pourtant sur 448 cas de paludisme chez l’Africain de la région thology of malaria. In: Malaria. Principles and Practice of Dakaroïse. Bull. Soc. Med. Afr. Noire Lang. Fr. 7: 167–169. Malariology, pp. 674–708, Wernsdorfer, W. H. and McGregor, Aung-Kyaw-Zaw, Khin-Maung-U, and Myo-Thwe (1988) Endo- I. A. (eds.) Churchill Livingstone, Edinburgh. toxaemia in complicated falciparum malaria. Trans. R. Soc. Brandts, C. H., Ndjave, M., Graninger, W. and Kremsner, P. G. Trop. Med. Hyg. 82: 513–514. (1997) Effect of paracetamol on parasite clearance time in Plas- Badibanga, B., Dayal, R., Depierreux, M., Pidival, G. and Lambert, modium falciparum malaria. Lancet 350: 704–709. P. H. (1986) Étude des principaux facteurs immunologiques et Brewster, D. R., Kwiatkowski, D. and White, N. J. (1990) Neuro- de la barrière hemato-meniné, au cause de la malaria cérébrale logical sequelae of cerebral malaria in children. Lancet 336: chez l’enfant en pays d’endemie (Zaire). Ann. Soc. Belg. Med. 1039–1043. Trop. 66: 23–37. Brown, D. A., Brown, F., Ganz, E. and Huttenlocher, P. R. (1978) Baier-Bitterlich, G., Fuchs, D., Murr, C., Reibnegger, G., Werner- Treatment of elevated intracranial pressure in Reye syndrome. Felmayer, G., Sgonc, R., Bock, G., Dierich, M. and Wachter, Ann. Neurol. 4: 275–278. H. (1995) Effect of neopterin and 7,8,dihydroneopterin on Brown, M. M. and Marshall, J. (1985) Regulation of cerebral tumor necrosis factor-␣ induced programmed cell death. FEBS blood flow in response to changes in blood viscosity. Lancet i: Lett. 364: 234–238. 604–609. Barennes, H., Pussard, E., Mahaman Sani, A., Clavier, F., Kahia- Brown, N. (1995) Severe malaria in children at Port Moresby gen- tani, F., Granic, G., Henzel, D., Ravinet, L. and Verdier, F. eral hospital, Papua New Guinea. Trop. Geog. Med. 47: 107–110. (1996) Efficacy and pharmacokinetics of a new intrarectal qui- Burchard, G. D., Radloff, P., Phillips, J., Nyemi, M., Knobiock, J. nine formulation in children with Plasmodium falciparum and Kremsner, P. G. (1995) Increased erythropoietin produc- malaria. Br. J. Clin. Pharmacol. 41: 389–395. tion in children with severe malarial anemia. Am. J. Trop. Med. Baruch, D. I., Pasloske, B. L., Singh, H. B., Bi, X., Ma, X. C., Feld- Hyg. 53: 547–551. man, M., Taraschi, T. F. and Howard, R. J. (1995) Cloning the Burgmann, H., Looareesuwan, S., Kapiotis, S., Viravan, C., Vani- P. falciparum gene encoding PfEMP1, a malarial variant anti- janonta, S., Hollenstein, U., Wiesinger, E., Presterl, E., Win- gen and adherence receptor on the surface of parasitized human kler, S. and Graninger, W. (1996) Serum levels of erythropoietin erythrocytes, Cell 82: 77–87. in acute Plasmodium falciparum malaria. Am. J. Trop. Med. Hyg. Bate, C. A. and Kwiatkowski, D. (1994) Inhibitory immunoglobu- 54: 280–283. lin M antibodies to tumor necrosis factor-inducing toxins in Byass, P., Campbell, H., O’Dempsey, T. J. and Greenwood, B. M. patients with malaria. Infect. Immun. 62: 3086–3091. (1991) Coincidence of malaria parasitaemia and abnormal Bell, D. R., Reid, H. A. and Stephens, A. J. (1976) Parasites which chest X-ray findings in young Gambian children. J. Trop. Med. migrate to the brain [letter]. Lancet i: 1292. Hyg. 94: 22–23. Berclaz, P. Y., Benedek, C., Jequier, E. and Schutz, Y. (1996) Carlson, J. (1993) Erythrocyte rosetting in Plasmodium falciparum Changes in protein turnover and resting energy expenditure malaria—with special reference to the pathogenesis of cerebral after treatment of malaria in Gambian children. Pediatr. Res. malaria. Scand. J. Infect. Dis. Suppl. 86: 1–79. 39: 401–409. Carlson, J., Helmby, H., Hill, A. V. S., Brewster, D., Green- Berendt, A. R., Turner, G. D. H. and Newbold, C. I. (1994) Cere- wood, B. M. and Wahlgren, M. (1990) Human cerebral malaria: bral malaria: the sequestration hypothesis. Parasitol. Today 10: association with erythrocyte rosetting and lack of anti-rosetting 412–413. antibodies. Lancet 336: 1457–1460. Berman, W., Pizzi, F., Schut, L., Raphaely, R. and Holtzapple, P. Carlson, J., Nash, G. B., Gabutti, V., al Yaman, F. and Wahlgren, (1975) The effects of exchange transfusion on intracranial pres- M. (1994) Natural protection against severe Plasmodium falci- sure in patients with Reye syndrome. J. Pediatr. 87: 887–891. parum malaria due to impaired rosette formation. Blood 84: Bernadino, L. and River, R. P. (1986) Análise de 254 internamen- 3909–3914. tos por malária cerebral em crianças dos 0–9 anos no Carme, B., Bouquety, J. C. and Plassart, H. (1993) Mortality and primeiro semestre de 1986 no serviço de pediatra do Hospi- sequelae due to cerebral malaria in African children in Brazza- tal Josina Machel de Luanda. Acta. Med. Ang. 5: 61–69. ville. Am. J. Trop. Med. Hyg. 48: 216–221. Pathophysiology and Supportive Treatment of Severe Malaria 43

Chapel, H. M., Warrell, D. A., Looareesuwan, S., White, N. J., morbidity from malaria in Gambian children after introduction Phillips, R. E., Warrell, M. J., Supawanta, V. and Tharavanij, S. of an impregnated bednet programme. Lancet 345: 479–483. (1987) Intrathecal immunoglobulin synthesis in cerebral malaria. D’Alessandro, U., Olaleye, B. O., McGuire, W., Thomson, M. C., Clin. Exp. Immunol. 67: 524–530. Langerock, P., Bennett, S. and Greenwood, B. M. (1995c) A Clark, H. C. and Tomlinson, W. J. (1949) The pathological anat- comparison of the efficacy of insecticide-treated and untreated omy of malaria. In: Clinical Malariology, pp. 874–883, Boyd, M. bed nets in preventing malaria in Gambian children. Trans. R. F. (ed.) W.B. Saunders, Philadelphia. Soc. Trop. Med. Hyg. 89: 596–598. Clark, I. A. (1978) Does endotoxin cause both the disease and Das, B. S., Mohanty, S., Mishra, S. K., Patnaik, J. K., Satpathy, S. K., parasite death in acute malaria and babesiosis? Lancet ii: 75–77. Mohanty, D. and Bose, T. K. (1991) Increased cerebrospinal Clark, I. and Chaudhri, G. (1988) Tumour necrosis factor may fluid protein and lipid peroxidation products in patients with contribute to the anaemia of malaria by causing dyserythropoie- cerebral malaria. Trans. R. Soc. Trop. Med. Hyg. 85: 733–734. sis and erythrophagocytosis. Br. J. Haematol. 70: 99–103. Davis, T. M. E., Krishna, S., Looareesuwan, S., Supanaranond, W., Clark, I. A. and Hunt, N. H. (1983) Evidence for reactive oxygen Pukrittayakamee, S., Attatamsoonthorn, K. and White, N. J. intermediates causing hemolysis and parasite death in malaria. (1990a) Erythrocyte sequestration and anemia in severe falci- Infect. Immun. 39: 1–6. parum malaria. J. Clin. Invest. 86: 793–800. Clark, I. A. and Rockett, K. A. (1994) The cytokine theory of Davis, T. M. E., Pukrittayakamee, S., Supanaranond, W., Looaree- human cerebral malaria. Parasitol. Today 10: 410–412. suwan, S., Krishna, S., Nagachinta, B., Turner, R. C. and Clark, I. A., Cowden, W. B., Butcher, G. A. and Hunt, N. H. White, N. J. (1990b) Glucose metabolism in quinine-treated (1983) Free oxygen radicals in malaria [letter]. Lancet i: 359–360. patients with uncomplicated falciparum malaria. Clin. Endo- Clark, I. A., Hunt, N. M. and Cowden, W. B. (1986) Oxygen- crinol. (Oxf.) 33: 739–749. derived free radicals in the pathogenesis of parasitic disease. Davis, T. M. E., Pukrittayakamee, S., Woodhead, J. S., Holloway, Adv. Parasitol. 25: 1–44. P., Chaivisuth, B. and White, N. J. (1991) Calcium and phos- Clark, I. A., Cowden, W. B., Butcher, G. A. and Hunt, N. H. phate metabolism in acute falciparum malaria. Clin. Sci. 81: (1987) Possible roles of tumor necrosis factor in the pathology 297–304. of malaria. Am. J. Pathol. 127: 192–198. Davis, T. M. E., Suputtamongkol, Y., Spencer, J. L., Ford, S., Clark, I. A., Rockett, K. A. and Cowden, W. B. (1991) Role of Chienkul, N., Schulenburg, W. E. and White, N. J. (1992) TNF in cerebral malaria. Lancet 337: 302–303. Measures of capillary permeability in acute falciparum malaria: Clark, I. A., Rockett, K. A. and Cowden, W. B. (1992) Possible relation to severity of infection and treatment. Clin. Infect. Dis. central role of nitric oxide in conditions clinically similar to 15: 256–266. cerebral malaria. Lancet 340: 894–896. Davis, T. M. E., Looareesuwan, S., Pukrittayakamee, S., Levy, J. C., Collomb, H., Rey, M., Dumas, M., Nouhouayi, A. and Petit, M. Nagachinta, B. and White, N. J. (1993) Glucose turnover in (1967) Les hémiplegies au cours du paludisme aigu. Bull. Soc. severe falciparum malaria. Metabolism 42: 334–340. Med. Afr. Noire. Lang. Fr. 7: 791–795. Davis, T. M. E., Suputtamongkol, Y., Spencer, J. L., Wilson, S. G., Commey, J. O. (1984) Is it cerebral malaria? [letter] Lancet ii: Mekhton, S., Croft, K. D. and White, N. J. (1994) Glucose 1037. turnover in pregnant women with acute malaria. Clin. Sci. 86: Commey, J. O., Mills-Tetteh, D. and Phillips, B. J. (1980) Cere- 83–90. bral malaria in Accra, Ghana. Ghana Med. J. 19: 68–72. Davis, T. M. E., Benn, J. J., Supputamangkol, Y., Weinberg, J., Cooke, B. M., Morris-Jones, S., Greenwood, B. M. and Nash, Umpleby, A. M., Chierakul, N. and White, N. J. (1996) Lac- G. B. (1993) Adhesion of parasitized red blood cells to cultured tate turnover and forearm lactate metabolism in severe falci- endothelial cells: a flow-based study of isolates from Gambian parum malaria. Endocrinol. Metab. 3: 105–115. children with falciparum malaria. Parasitology 107: 359–368. Day, N. P., Phu, N. H., Bethell, D. P., Mai, N. T., Chau, T. T., Cooke, B. M., Morris-Jones, S., Greenwood, B. M. and Nash, G. Hien, T. T. and White, N. J. (1996) The effects of dopamine B. (1995) Mechanisms of cytoadhesion of flowing, parasitized and adrenaline infusions on acid-base balance and systemic red blood cells from Gambian children with falciparum malaria. haemodynamics in severe infection. Lancet 348: 219–223. Am. J. Trop. Med. Hyg. 53: 29–35. Dekker, E., Romijn, J. A., Waruiru, C., Ackermans, M. T., Wever- Cot, S., Ringwald, P., Mulder, B., Miailhes, P., Yap-Yap, J., Nus- ling, G. J., Sauerwein, R. W., Endert, E., Peshu, N., Marsh, K. sler, A. K. and Eling, W. M. (1994) Nitric oxide in cerebral and Sauerwein, H. P. (1996) The relationship between glucose malaria [letter]. J. Infect. Dis. 169: 1417–1418. production and plasma glucose concentration in children with Coulter, J. B. S. (1991) Lumbar puncture for cerebral malaria. Afr. falciparum malaria. Trans. R. Soc. Trop. Med. Hyg. 90: 654–657. Health November: 27. Dekker, E., Romijn, J. A., Moeniralam, H. S., Waruiru, C. M., Crawley, J., Smith, S., Kirkham, F. J., Muthinji, P., Waruiru, C. Ackermans, M. T., Timmer, J. G., Endert, E., Peshu, N., Marsh, and Marsh, K. (1996) Seizures and status epileticus in child- K. and Sauerwein, H. P. (1997) The influence of alanine infu- hood cerebral malaria. Q. J. Med. 89: 591–597. sion on glucose production in “malnourished” African children Cropper, J. (1908) Phenomenal abundance of parasites in a fatal with falciparum malaria. Q. J. Med. 90: 455–460. case of pernicious malaria. Lancet ii: 16–17. Delacollette, C., Taelman, H. and Wery, M. (1995) An etiologic D’Alessandro, U., Leach, A., Drakeley, C. J., Bennett, S., Olaleye, study of hemoglobinuria and blackwater fever in the Kivu B. O., Fegan, G. W., Jawara, M., Langerock, P., George, M. O., Mountains, Zaire. Ann. Soc. Belg. Med. Trop. 75: 51–63. Targett, G. A. and Greenwood, B. M. (1995a) Efficacy trial of Deloron, P., Dumont, N., Nyongabo, T., Aubry, P., Astagneau, P., malaria vaccine SPf66 in Gambian infants. Lancet 346: 462–467. Ndarugirire, F., Menetrier-Caux, C., Burdin, N., Brelivet, J. C. D’Alessandro, U., Olaleye, B. O., McGuire, W., Langerock, P., and Peyron, F. (1994) Immunologic and biochemical alter- Bennett, S., Aikins, M. K., Thomson, M. C., Cham, M. K., ations in severe falciparum malaria: relations to neurological Cham, B. A. and Greenwood, B. M. (1995b) Mortality and symptoms and outcome. Clin. Infect. Dis. 19: 480–485. 44 C. R. J. C. Newton and S. Krishna de Naotunne, T., Karunaweera, N. D., Mendis, K. N. and Carter, R. Faero, O., Kastrup, K. W., Lykkegaard-Nielsen, E., Melchior, J. C. (1993) Cytokine-mediated inactivation of malarial gametocytes and Thorn, I. (1972) Successful prophylaxis of febrile convul- is dependent on the presence of white blood cells and involves sions with phenobarbital. Epilepsia 13: 279–285. reactive nitrogen intermediates. Immunology 78: 555–562. Fisher, C. W. S. (1983) Clinical curio: acidosis and hypoglycaemia Descamps Latscha, B., Lunel Fabiani, F., Kara Binis, A. and in malaria. Br. Med. J. 286: 1261. Druilhe, P. (1987) Generation of reactive oxygen species in Fitz-Hugh, T. (1944) The cerebral form of malaria. Bull. US Army whole blood from patients with acute falciparum malaria. Para- Med. Dept. 83: 39–48. site Immunol. 9: 275–279. Fraser, A. and Evan, G. (1996) A license to kill. Cell 85: 781–784. de Silva, H. J., de Silva, N. R. and Snook, J. A. (1992) Delayed Friedland, J. S., Ho, M., Remick, D. G., Bunnag, D., White, cerebellar ataxia following falciparum malaria: lack of evidence N. J. and Griffin, G. E. (1993) Interleukin-8 and Plasmodium for antibody mediation. Trans. R. Soc. Trop. Med. Hyg. 86: 608. falciparum malaria in Thailand. Trans. R. Soc. Trop. Med. Dhayagude, R. G. and Purandare, M. B. (1943) Autopsy study of Hyg. 87: 54–55. cerebral malaria with special reference to malarial granuloma. Gaskell, J. F. and Miller, W. L. (1920) Studies on malignant Arch. Pathol. 36: 550–558. malaria in Macedonia. Q. J. Med. 13: 381–426. Di Perri, G., Di Perri, I. G., Monteiro, G. B., Bonora, S., Hennig, Gelfand, M. (1973) Neurological complications of parasitic dis- C., Cassatella, M., Micciolo, R., Vento, S., Dusi, S., Bassetti, D. ease. In: Tropical Neurology, pp. 247–251, Spillane, J. D. (ed.) and Concia, E. (1995) Pentoxifylline as a supportive agent in Oxford University Press, London. the treatment of cerebral malaria in children. J. Infect. Dis. 171: Genton, B., Al-Yaman, F., Alpers, M. P. and Mokein, D. (1997) 1317–1322. Indicators of fatal outcome in paediatric cerebral malaria: a Dobbie, M., Crawley, J., Waruiru, C., Marsh, K. and Surtees, R. study of 134 comatose Papua New Guinean children. Int. J. (1997) Cerebrospinal fluid concentrations of quinolinic acid, Epidemiol. 26: 670–676. total neopterin and nitrate plus nitrite in Kenyan children Go, K. G. (1991) Cerebral Pathophysiology: An Integral Approach recovering from cerebral malaria. J. Neurochem. 69: 5181. with Some Emphasis on Clinical Applications. Elsevier Science Dondorp, A. M., Angus, B. J., Hardeman, M. R., Chotivanich, Publishers B.V., Amsterdam. K. T., Silamut, K., Ruangveerayuth, R., Kager, P. A., White, Goitein, K. J., Fainmesser, P. and Sohmer, H. (1983) Cerebral per- N. J. and Vreeken, J. (1997) Prognostic significance of reduced fusion pressure and auditory brain-stem responses in childhood red cell deformability in severe falciparum malaria. Am. J. Trop. CNS diseases. Am. J. Dis. Child. 137: 777–781. Med. Hyg. 57: 507–511. Gopinathan, V. P. and Subramanian, A. R. (1986) Vivax and fal- Dorn, A., Stoffel, R., Matile, H., Bebndorf, A. and Ridley, R. ciparum malaria seen at an Indian service hospital. J. Trop. (1995) Malaria haemozoin/␤-haematin supports haem polymer- Med. Hyg. 89: 51–55. ization in the absence of protein. Nature 374: 269–271. Gordeuk, V., Thuma, P., Brittenham, G., McLaren, C., Parry, D., Dubey, M. L., Rai, S. K., Ganguly, N. K., Kalra, A., Varma, S. C. Backenstose, A., Biemba, G., Msiska, R., Holmes, L., McKin- and Mahajan, R. C. (1991) Generation of reactive oxygen spe- ley, E., Vargas, L., Gilkeson, R. and Poltera, A. A. (1992) Effect cies by blood monocytes in human Plasmodium falciparum and of iron chelation therapy on recovery from deep coma in chil- P. vivax infections. APMIS 99: 210–212. dren with cerebral malaria. N. Engl. J. Med. 327: 1473–1477. Dudgeon, L. S. and Clarke, C. (1917) A contribution to the Gordeuk, V. R., Thuma, P. E., McLaren, C. E., Biemba, G., Zulu, microscopical histology of malaria. Lancet ii: 153–156. S., Poltera, A. A., Askin, J. E. and Brittenham, G. M. (1995) Edington, G. M. (1967) Pathology of malaria in West Africa. Br. Transferrin saturation and recovery from coma in cerebral Med. J. 1: 715–718. malaria. Blood 85: 3297–3301. Edington, G. M. and Gilles, H. M. (1976) Malaria. In: Pathology Gozal, D. (1992) The incidence of pulmonary manifestations dur- in the Tropics, pp. 20–23, Edward Arnold, London. ing Plasmodium falciparum malaria in non immune subjects. Elford, B. C., Cowan, G. M. and Ferguson, D. J. (1995) Parasite- Trop. Med. Parasitol. 43: 6–8. regulated membrane transport processes and metabolic control Grau, G. E., Piguet, P. F., Engers, H. D., Louis, J. A., Vassalli, P. in malaria infected erythrocytes. Biochem. J. 308: 361–374. and Lambert, P. H. (1986) L3T4ϩ T lymphocytes play a major English, M., Marsh, V., Amukoye, E., Lowe, B., Murphy, S. and role in the pathogenesis of murine cerebral malaria. J. Immunol. Marsh, K. (1996a) Chronic salicylate poisoning and severe 137: 2348–2354. malaria. Lancet 347: 1736–1737. Grau, G. E., Fajardo, L. F., Piguet, P. F., Allet, B., Lambert, P. H. and English, M. C., Waruiru, C., Lightowler, C., Murphy, S. A., Vassalli P. (1987a) Tumor necrosis factor (cachectin) as an essen- Kirigha, G. and Marsh, K. (1996b) Hyponatraemia and dehy- tial mediator in murine cerebral malaria. Science 237: 1210–1212. dration in severe malaria. Arch. Dis. Child. 74: 201–205. Grau, G. E., Gretener, D. and Lambert, P. H. (1987b) Prevention English, M., Waruiru, C. and Marsh, K. (1996c) Transfusion for of murine cerebral malaria by low-dose cyclosporin A. Immu- respiratory distress in life-threatening childhood malaria. Am. J. nology 61: 521–525. Trop. Med. Hyg. 55: 525–530. Grau, G. E., Piguet, P. F., Vassalli, P. and Lambert, P. H. (1989) English, M., Sauerwein, R. W., Waruiru, C. M., Mosobo, M., Obi- Involvement of tumour necrosis factor and other cytokines in ero, J., Lowe, B. and Marsh, K. (1997) Acidosis in severe child- immune-mediated vascular pathology. Int. Arch. Allergy Appl. hood malaria. Q. J. Med. 90: 263–270. Immunol. 88: 34–39. Eross, J., Sihnu, M. and Dorman, D. (1981) Cerebrospinal fluid lactic Greenberg, A. E., Ntumbanzondo, M., Ntula, N., Mawa, L., How- acidosis in bacterial meningitis. Arch. Dis. Child. 56: 692–698. ell, J. and Davachi, F. (1989) Hospital-based surveillance of Facer, C. A., Bray, R. S. and Brown, J. (1979) Direct Coombs malaria-related paediatric morbidity and mortality in Kinshasa, antiglobulin reactions in Gambian children with Plasmodium Zaire. Bull. WHO 67: 189–196. falciparum malaria. I. Incidence and class specificity. Clin. Exp. Greenwood, B. M. (1991) Cerebral malaria [letter]. Lancet 337: 1282. Immunol. 35: 119–127. Greenwood, B. (1996) Fever and malaria. Lancet 248: 280–281. Pathophysiology and Supportive Treatment of Severe Malaria 45

Greenwood, B. M. (1997) Malaria transmission and vector con- Hurn, P. D., Kirsch, J. R.. Helfaer, M. A. and Traystman. R. J. trol. Parasitol. Today 13: 90–92. (1996) Brain damage and free radicals. In: Tissue Perfusion and Greenwood, B. M., Stratton, D., Williamson, W. A. and Moham- Organ Function: Ischemia/Reperfusion Injury, pp. 71–84, med, I. (1978) A study of the role of immunological factors in Kamada, T., Shiga, T. and McCuskey, R. S. (eds.) Elsevier, the pathogenesis of the anaemia of acute malaria. Trans. R. Amsterdam. Soc. Trop. Med. Hyg. 72: 378–385. Ikpatt, N. W., Asindi, A. A., Ekanem, I. A. and Khalil, M. I. Guignard, J. (1965) Le paludisme pernicieux du nourrisson et de (1990) Preliminary observations on cerebral malaria in Nige- l’enfant. Ann. Pediatr. (Paris) 41: 2474–2484. rian children. East Afr. Med. J. 67: 341–347. Heinemeyer, G. (1987) Clinical pharmacokinetic considerations Jakobsen, P. H., McKay, V., Morris-Jones, S. D., McGuire, W., in the treatment of increased intracranial pressure. Clin. Phar- van Hensbroek, M. B., Meisner, S., Bendtzen, K., Schousboe, I., macokinet. 13: 1–25. Bygbjerg, I. C. and Greenwood, B. M. (1994) Increased con- Hemmer, C. J., Bierhaus, A., Riedesel, J., Gabat, S., Lilensiek, B., centrations of interleukin-6 and interleukin-1 receptor antago- Pitronik, P., Lin, J., Grauer, A., Amiral, J., Zielger, R., Schief- nist and decreased concentrations of beta-2-glycoprotein I in fer, S., Kern, P., Egbring, R., Dietricj, M. and Nawroth, P. P. Gambian children with cerebral malaria. Infect. Immun. 62: (1994) Elevated thrombomodulin plasma levels as a result of 4374–4379. endothelial involvement in Plasmodium falciparum malaria. Janota, I. and Doshi, B. (1979) Cerebral malaria in the United Thromb. Haemost. 72: 457–464. Kingdom. J. Clin. Pathol. 32: 769–772. Hemmer, C. J., Hort, G., Chiwakata, C. B., Seitz, R., Egbring, R., Jeje, O. M., Kelton, J. G. and Blajchman, M. A. (1983) Quantita- Gaus, W., Hogel, J., Hassemer, M., Nawroth, P. P., Kern, P. and tion of red cell membrane associated immunoglobulin in chil- Dietrich, M. (1997) Supportive pentoxifylline in falciparum dren with Plasmodium falciparum parasitaemia. Br. J. Haematol. malaria: no effect on tumour necrosis factor alpha levels or clin- 54: 567–572. ical outcome: a prospective, randomized, placebo-controlled Jones, E. S. and McGregor, I. A. (1954) Pathological processes in study. Am. J. Trop. Med. Hyg. 56: 397–403. disease. V. Blood physiology of Gambian children infected with Hero, M., Harding, S. P., Riva, C. E., Winstanley, P. A., Peshu, N. Plasmodium falciparum. Ann. Trop. Med. Parasitol. 48: 95–101. and Marsh, K. (1997) Photographic and angiographic charac- Kanaani, J. and Ginsburg, H. (1991) Transport of lactate in Plas- terization of the retina of Kenya children with severe malaria. modium falciparum-infected human erythrocytes. J. Cell Biol. Arch. Ophthalmol. 115: 997–1103. 149: 469–476. Hershko, C. and Peto, T. E. (1988) Desferroxamine inhibition of Karunaweera, N. D., Grau, G. E., Gamage, P., Carter, R. and Men- malaria is independent of host iron status. J. Exp. Med. 168: dis, K. N. (1992) Dynamics of fever and serum TNF levels are 375–387. closely associated during clinical paroxysms in Plasmodium vivax Hien, T. T. (1994) An overview of the clinical use of artemisinin malaria. Proc. Natl. Acad. Sci. USA 89: 3200–3203. and its derivatives in the treatment of falciparum malaria in Viet Kataoka, K., Okuno, T., Mikawa, H. and Hojo, H. (1988) Cranial Nam. Trans. R. Soc. Trop. Med. Hyg. 88 (Suppl. 1): S7–S8. computed tomographic and electroencephalographic abnormal- Hien, T. T. and White, N. J. (1993) Qinghaosu. Lancet 341: 603– ities in children with post-hemiconvulsive hemiplegia. Eur. 608. Neurol. 28: 279–284. Ho, M., Davis, T. M. E., Silamut, K., Bunnag, D. and White, N. J. Kaul, D. K., Roth, E. F., Ngel, R. L., Howard, R. J. and Handun- (1991) Rosette formation of Plasmodium falciparum-infected netti, S. M. (1991) Rosetting of Plasmodium falciparum-infected erythrocytes from patients with acute malaria. Infect. Immun. red blood cells with uninfected red blood cells enhances microvas- 59: 2135–2139. cular obstruction under flow conditions. Blood 78: 812–819. Hoffman, S. L., Rustama, D., Punjabi, N. H., Surampaet, B., San- Kawo, N. G., Msengi, A. E., Swai, A. B. M., Chuwa, L. M., jaya, B., Dimpudus, A. J., McKee, K. T., Jr., Paleologo, F. P., Alberti, K. G. M. M. and McLarty, D. G. (1990) Specificity of Campbell, J. R., Marwoto, H. and Laughlin, L. (1988) High- hypoglycaemia for cerebral malaria in children. Lancet 336: dose dexamethasone in quinine-treated patients with cerebral 454–457. malaria: a double-blind, placebo-controlled trial. J. Infect. Dis. Kayembe, D., Maertens, K. and De Laey, J. J. (1980) Complica- 158: 325–331. tions oculaires de la malaria cérébrale. Bull. Soc. Belge. Ophtal- Hoffman, W. H., Steinhart, C. M., Gammal, T. E., Steele, S., mol. 190: 53–60. Cuadrado, A. R. and Morse, P. K. (1988) Cranial CT in chil- Kelton, J. G., Keystone, J., Moore, J., Denomme, G., Tozman, E., dren and adolescents with diabetic ketoacidosis. Am. J. Neuro- Glynn, M., Neame, P. B., Gauldie, J. and Jensen, J. (1983) radiol. 9: 733–739. Immune-mediated thrombocytopenia in malaria. J. Clin. Invest. Holloway, P. A. H., Krishna, S. and White, N. J. (1991) Plasmo- 71: 832–836. dium berghei: lactic acidosis and hypoglycaemia in a rodent Kern, P., Hemmer, C. J., Van Dame, J., Gruss, H. J. and Dietrich, model of severe malaria; effects of glucose, quinine, and M. (1989) Elevated tumor necrosis factor alpha and interleu- dichloroacetate. Exp. Parasitol. 72: 1–11. kin-6 serum levels as markers for complicated Plasmodium falci- Holloway, P. A. H., Knox, K., Bajaj, N., Chapman, D., White, N. J., parum malaria. Am. J. Med. 87: 139–143. O’Brien, R., Stacpoole, P. W. and Krishna, S. (1995) Plasmodium Khan, N. U. and Durham, M. D. (1945) Cerebral malaria. J. R. berghei infection: dichloroacetate improves survival in rats with Army Med. Corps 84: 263–267. lactic acidosis. Exp. Parasitol. 80: 624–632. Kingston, M. E. (1971) Experience with urea in invert sugar for the Howard, R. J., Handunnetti, S. M., Hasler, T., Gilladoga, A., de treatment of cerebral malaria. J. Trop. Med. Hyg. 74: 249–252. Aguiar, J. C., Pasloske, B. L., Morehead, K., Albrecht, G. R. Kirkham, F. J. (1991) Intracranial pressure and cerebral blood flow and van Schravendijk, M. R. (1990) Surface molecules on Plas- in non-traumatic coma in childhood. In: Problems of Intracra- modium falciparum-infected erythrocytes involved in adherence. nial Pressure in Childhood, pp. 283–348, Minns, R. A. (ed.) Am. J. Trop. Med. Hyg. 43: 15–29. Mac Keith Press, London. 46 C. R. J. C. Newton and S. Krishna

Kirkham, F. J., Newton, C. R. J. C., Winstanley, P. A., Pasvol, G., Kwiatkowski, D., Molyneux, M. E., Stephens, S., Curtis, N., Klein, Peshu, N., Warrell, D. A. and Marsh, K. (1991) Cerebral N., Pointaire, P., Smit, M., Allan, R., Brewster, D. R., Grau, G. E. malaria. Lancet 337: 1282–1283. and Greenwood, B. M. (1993) Anti-TNF therapy inhibits fever Kirkham, F. J., Newton, C. R. J. C., Winstanley, P. A., Peshu, N. in cerebral malaria. Q. J. Med. 86: 91–98. and Marsh, K. (1994) Crises d’epilepsie dans le paludisme cere- Lackritz, E. M., Campbell, C. C., Ruebush, T. K., 2d, Hightower, bral. In: Neurologie Tropicale, pp. 381–382, Dumas, M., Gior- A. W., Wakube, W., Steketee, R. W. and Were, J. B. (1992) dano, C., Gentilini, M. and Chieze, F. (eds.) John Libbey & Effect of blood transfusion on survival among children in a Company, Paris. Kenyan hospital. Lancet 340: 524–528. Kremsner, P. G., Grundmann, H., Neifer, S., Sliwa, K., Sahlmuller, Lafaix, C., Dumas, M., Nouhouayi, A. and Rey, M. (1970) Cécité G., Hegenscheid, B. and Bienzle, U. (1991) Pentoxifylline pre- consecutive au paludisime cérébrale (d’après sept cas). Bull. Soc. vents murine cerebral malaria. J. Infect. Dis. 164: 605–608. Med. Afr. Noire Lang. Fr. 15: 423–433. Krishna, S. and White, N. J. (1989) Management of severe malar- Lalloo, D. G., Trevett, A. J., Paul, M., Korinhona, A., Laurenson, ial infection. Indian J. Pediatr. 56: 155–163. I. F., Mapao, J., Nwokolo, N., Danga Christian, B., Black, J., Krishna, S. and White, N. J. (1996) Pharmacokinetics of quinine, Saweri, A., Naraqi, S. and Warrell, D. A. (1996) Severe and chloroquine and amodiaquine. Clinical implications. Clin. complicated falciparum malaria in Melanesian adults in Papua Pharmacokinet. 30: 263–299. New Guinea. Am. J. Trop. Med. Hyg. 55: 119–124. Krishna, S., Shoubridge, E. A., White, N. J., Weatherall, D. J. and Laveran, C. A. (1880) Un nouveau parasite trouvé dans le sang Radda, G. K. (1983) Plasmodium yoelii: blood oxygen and brain des malades atteints de fièvre palustre. Origine parasitaire des function in the infected mouse. Exp. Parasitol. 56: 391–396. accidents de l’impaludisme. Bull. Mem. Soc. Méd. Hôp. Paris Krishna, S., Supanaranond, W., Pukrittayakamee, S., Karter, D., 17: 158–164. Supputamongkol, Y., Davis, T. M. E., Holloway, P. A., White, Leaver, R. J., de Baetselier, H., Bagshawe, A. and Watters, D. A. N. J. and Davis, T. M. (1994a) Dichloroacetate for lactic acido- (1990) Cerebral malaria: what is unarousable coma? Lancet sis in severe malaria: a pharmacokinetic and pharmacodynamic 335: 44–45. assessment. Metabolism 43: 974–981. Lee, S. H., Looareesuwan, S., Chan, J., Wilairatana, P., Vani- Krishna, S., Waller, D. W., ter Kuile, F., Kwiatkowski, D., Craw- janonta, S., Chong, S. M. and Chong, B. H. (1997) Plasma ley, J., Craddock, C. F., Nosten, F., Chapman, D., Brewster, D., macrophage colony-stimulating factor and P-selectin levels in Holloway, P. A. and White, N. J. (1994b) Lactic acidosis and malaria-associated thrombocytopaenia. Thromb. Haemost. 72: hypoglycaemia in children with severe malaria: pathophysiolog- 2898-293. ical and prognostic significance. Trans. R. Soc. Trop. Med. Leete, T. H. and Rubin, H. (1996) Malaria and the cell cycle. Par- Hyg. 88: 67–73. asitol. Today 12: 442–444. Krishna, S., Agbenyenga, T., Angus, B. J., Bedu-Addo, G., Ofori- Lehman, L. G., Vu-Quoc, B., Carlson, J. and Kremsner, P. G. Amanfo, G., Henderson, G., Szwandt, I. S. I., O’Brien, R. and (1997) Plasmodium falciparum: inhibition of erythrocyte rosette Stacpoole, P. W. (1995a) Pharmacokinetics and pharmacody- formation and detachment of rosettes by pentoxifylline. Trans. namics of dichloracetate in children with lactic acidosis due to R. Soc. Trop. Med. Hyg. 91: 74–75. severe malaria. Q. J. Med. 88: 341–349. Lemercier, G., Rey, M. and Collomb, H. (1966) Lesions cérébrales Krishna, S., Pukrittayakamee, S., Supanaranond, W., ter Kuile, F., dans le paludisme de l’enfant. Bull. Soc. Pathol. Exot. 59: 533– Ruprah, M., Sura, T. and White, N. J. (1995b) Fever in uncom- 548. plicated Plasmodium falciparum malaria: randomized double- Lemercier, G., Bert, J., Nouhoayi, A., Rey, M. and Collomb, H. ”blind” comparison of ibuprofen and paracetamol treatment. (1969) Le neuropaludisme: aspects électroencéphalographiques, Trans. R. Soc. Trop. Med. Hyg. 89: 507–509. neuropathologiques, problèmes physiopathologiques. Path. Biol. Krishna, S., Supanaranond, W., Pukrittayakamee, S., Kuile, F. T., (Paris) 17: 459–472. Ruprah, M. and White, N. J. (1996) The disposition and effects Lewallen, S., Taylor, T. E., Molyneux, M. E., Wills, B. A. and of two doses of dichloroacetate in adults with severe falciparum Courtright, P. (1993) Ocular fundus findings in Malawian chil- malaria. Br. J. Clin. Pharmacol. 41: 29–34. dren with cerebral malaria. Opthalmology 100: 857–861. Kwiatkowski, D. (1989) Febrile temperatures can synchronize the Lipton, S. A., Choi, Y., Pan, Z., Lei, S. Z., Chen, H. S., Sucher, N. J., growth of Plasmodium falciparum in vitro. J. Exp. Med. 169: Loscalzo, J., Singel, D. J. and Stamler, J. S. (1993) A redox- 357–361. based mechanism for the neuroprotective and neurodestructive Kwiatkowski, D. and Bate, C. (1995) Inhibition of tumour necro- effects of nitric oxide and related nitroso-compounds. Nature sis factor (TNF) production by antimalarial drugs used in cere- 364: 626–632. bral malaria. Trans. R. Soc. Trop. Med. Hyg. 89: 215–216. Looareesuwan, S., Warrell, D. A., White, N. J., Chanthavanich, Kwiatkowski, D., Cannon, J. G., Manogue, K. R., Cerami, A., P., Warrell, M. J., Chantaratherakitti, S., Changswek, S., Dinarello, C. A. and Greenwood, B. M. (1989) Tumour necro- Chongmankongcheep, L. and Kanchanaranya, C. (1983) Reti- sis factor production in falciparum malaria and its association nal hemorrhage, a common sign of prognostic significance in with schizont rupture. Clin. Exp. Immunol. 77: 361–366. cerebral malaria. Am. J. Trop. Med. Hyg. 32: 911–915. Kwiatkowski, D., Hill, A. V., Sambou, I., Twumasi, P., Castra- Looareesuwan, S., Ho, M., Wattanagoon, Y., White, N. J., War- cane, J., Manogue, K. R., Cerami, A., Brewster, D. R. and rell, D. A., Bunnag, D., Harinasuta, T. and Wyler, D. J. (1987) Greenwood, B. M. (1990) TNF concentration in fatal cerebral, Dynamic alteration in splenic function during acute falciparum non-fatal cerebral, and uncomplicated Plasmodium falciparum malaria. N. Engl. J. Med. 317: 675–679. malaria. Lancet 336: 1201–1204. Looareesuwan, S., Phillips, R. E., Karbwang, J., White, N. J., Kwiatkowski, D., Molyneux, M., Taylor, T., Klein, N., Curtis, N. Flegg, P. J. and Warrell, D. A. (1990) Plasmodium falciparum and Smit, M. (1991) Cerebral malaria [letter]. Lancet 337: hyperparasitaemia: use of exchange transfusion in seven patients 1281–1282. and a review of the literature. Q. J. Med. 227: 471–481. Pathophysiology and Supportive Treatment of Severe Malaria 47

Looareesuwan, S., Davis, J. G., Allen, D. L., Lee, S. H., Bunnag, ciparum transmission and the incidence of severe malaria infec- D. and White, N. J. (1992) Thrombocytopenia in malaria. tions on the Kenyan coast. Am. J. Trop. Med. Hyg. 49: 245– Southeast Asian J. Trop. Med. Public Health 23: 44–50. 253. Looareesuwan, S., Wilairatana, P., Krishna, S., Kendall, B., Van- McCaslin, R. I., Pikis, A. and Rodriquez, W. J. (1994) Pediatric naphan, S., Viravan, C. and White, N. J. (1995) Magnetic reso- Plasmodium falciparum malaria: a ten-year experience from Wash- nance imaging of the brain in patients with cerebral malaria. ington DC. Pediatr. Infect. Dis. J. 13: 709–715. Clin. Infect. Dis. 21: 300–309. McGraw, C. P., Alexander, E. and Howard, G. (1978) Effect of Looareesuwan, S., Wilairatana, P., Vannaphan, S., Gordeuk, V. R., dose and dose schedule on the response of intracranial pressure Taylor, T. E., Meshnick, S. R. and Brittenham, G. M. (1996) to mannitol. Surg. Neurol. 10: 127–130. Co-administration of desferrioxamine B with artesunate in McGuire, W., Hill, A. V. S., Allsopp, C. E. M., Greenwood, B. M. malaria: an assessment of safety and tolerance. Ann. Trop. Med. and Kwiatkowski, D. (1994) Variation in the TNF-␣ promoter Parasitol. 90: 551–554. region associated with the susceptibility to cerebral malaria. Lucas, S. K., Hounnou, A., Bell, J., Peacock, C., Andoh, J., Nature 371: 508–510. Honde, M., De Cock, K. M. and Warrell, D. A. (1996) Severe McGuire, W., Hill, A. V., Greenwood, B. M. and Kwiatkowski, D. cerebral swelling is not observed in children dying with malaria. (1996) Circulating ICAM-1 levels in falciparum malaria are Q. J. Med. 89: 351–353. high but unrelated to disease severity Trans. R. Soc. Trop. Med. Lynk, A. and Gold, R. (1989) Review of 40 children with Hyg. 90: 274–276. imported malaria. Pediatr. Infect. Dis. J. 8: 745–750. Mercado, T. I. and von Brand, T. (1954) Glycogen studies on Mabey, D. C., Brown, A. and Greenwood, B. M. (1987) Plasmo- white rats infected wth Plasmodium berghei. Exp. Parasitol. 3: dium falciparum malaria and Salmonella infections in Gambian 259–266. children. J. Infect. Dis. 155: 1319–1321. Merry, A. H., Looareesuwan, S., Phillips, R. E., Chanthavanich, Mabeza, G. F., Moyo, V. M., Thuma, P. E., Biemba, G., Parry, D., P., Supanaranond, W., Warrell, D. A. and Weatherall, D. J. Khumalo, H., Nyarugwe, P., Zulu, S. and Gordeuk, V. R. (1986) Evidence against immune haemolysis in falciparum (1995) Predictors of severity of illness on presentation in chil- malaria in Thailand. Br. J. Haematol. 64: 187–194. dren with cerebral malaria. Ann. Trop. Med. Parasitol. 89: 221– Metzger, W. G., Mordmuller, B. G. and Kremsner, P. G. (1995) 228. Malaria pigment in leucocytes. Trans. R. Soc. Trop. Med. Hyg. Mabeza, G. F., Biemba, G. and Gordeuk, V. R. (1996) Clinical 89: 637–638. studies of iron chelators in malaria. Acta Haematol. 95: 78–86. Migasena, P. and Maegraith, B. G. (1968a) The movement of flu- MacDougall, L. G. (1962) Low cerebrospinal fluid protein in African orescent isothiocyanate (F.I.T.C.) labelled human albumin children with febrile convulsions. Arch. Dis. Child. 37: 309–313. from blood into brain tissue examined by fluorescent technique MacPherson, G. G., Warrell, M. J., White, N. J., Looareesuwan, S. in normal and Plasmodium knowlesi infected Macaca mulatta. and Warrell, D. A. (1985) Human cerebral malaria. A quanti- Med. J. Malaysia 22: 250. tative ultrastructural analysis of parasitized erythrocyte seques- Migasena, P. and Maegraith, B. G. (1968b) Factor affecting on the tration. Am. J. Pathol. 119: 385–401. movement of protein across the blood:brain:C.S.F. barriers in Maegraith, B. (1969) Comments on pathophysiology. Milit. Med. Plasmodium knowlesi infected Macaca mulatta. Med. J. Malaysia 134: 1129–1131. 22: 251. Maegraith, B. and Fletcher, A. (1972) The pathogenesis of mam- Migasena, S. (1983) Hypoglycaemia in falciparum malaria. Ann. malian malaria. Adv. Parasitol. 10: 49–75. Trop. Med. Parasitol. 77: 323–324. Maguire, M. J. (1983) The management of cerebral malaria in Miller, K. D., Greenberg, A. E. and Campbell, C. C. (1989a) African children. East Afr. Med. J. 60: 260–265. Treatment of severe malaria in the United States with a contin- Marchiafava, E. and Bignami, A. (1894) On Summer-Autumn uous infusion of quinidine gluconate and exchange transfusion. Malaria Fevers. The New Sydenham Society, London. N. Engl. J. Med. 321: 65–70. Marsh, K., Marsh, V. M., Brown, J., Whittle, H. C. and Green- Miller, K. D., White, N. J., Lott, J. A., Roberts, J. M. and Green- wood, B. M. (1988) Plasmodium falciparum: the behavior of clin- wood, B. M. (1989b) Biochemical evidence of muscle injury in ical isolates in an in vitro model of infected red blood cell African children with severe malaria. J. Infect. Dis. 159: 139– sequestration. Exp. Parasitol. 65: 202–208. 142. Marsh, K., Forster, D., Waruiru, C., Mwangi, I., Winstanley, M., Miller, L. H. (1969) Distribution of mature trophozoites and schiz- Marsh, V., Newton, C., Winstanley, P., Warn, P., Peshu, N., onts of Plasmodium falciparum in the organs of Aotus trivirga- Pasvol, G. and Snow, R. W. (1995) Indicators of life-threatening tus, the night monkey. Am. J. Trop. Med. Hyg. 18: 860–865. malaria in African children. N. Engl. J. Med. 332: 1399–1404. Miller, L. H., Fremount, H. N. and Luse, S. A. (1971) Deep vascu- Marshall, L. F., Smith, R. W., Rauscher, L. A. and Shapiro, H. M. lar schizogony of Plasmodium knowlesi in Macaca mulatta. Distri- (1978) Mannitol dose requirements in brain-injured patients. J. bution in organs and ultrastructure of parasitized red cells. Am. Neurosurg. 48: 169–172. J. Trop. Med. Hyg. 20: 816–824. Mate Kole, M. O., Yeboah, E. D., Affram, R. K. and Adu, D. Molyneux, M. E., Looareesuwan, S., Menzies, I. S., Grainger, S. L., (1996) Blackwater fever and acute renal failure in expatriates in Phillips, R. E., Wattanagoon, Y., Thompson, R. P. H. and War- Africa. Ren. Fail. 18: 525–531. rell, D. A. (1989a) Reduced hepatic blood flow and intestinal Matthews, D. S., Bullock, R. E., Matthews, J. N., Aynsley Green, malabsorption in severe falciparum malaria. Am. J. Trop. Med. A. and Eyre, J. A. (1995) Temperature response to severe head Hyg. 40: 470–476. injury and the effect on body energy expenditure and cerebral Molyneux, M. E., Taylor, T. E., Wirima, J. J. and Borgstein, A. oxygen consumption. Arch. Dis. Child. 72: 507–515. (1989b) Clinical features and prognostic indicators in paediat- Mbogo, C. N., Snow, R. W., Kabiru, E. W., Ouma, J. H., Githure, ric cerebral malaria: a study of 131 comatose Malawian chil- J. I., Marsh, K. and Beier, J. C. (1993) Low-level Plasmodium fal- dren. Q. J. Med. 71: 441–459. 48 C. R. J. C. Newton and S. Krishna

Molyneux, M. E., Taylor, T. E., Wirima, J. J. and Grau, G. E. (1990) Cerebral malaria: what is unarousable coma? [letter]. (1991) Tumour necrosis factor, interleukin-6, and malaria. Lan- Lancet 335: 472. cet 337: 1098. Newton, C. R., Kirkham, F. J., Winstanley, P. A., Pasvol, G., Mordes, J. P., Tranquada, R. E. and Rossini, A. A. (1996) Lactic Peshu, N., Warrell, D. A. and Marsh, K. (1991a) Intracranial acidosis. In: Intensive Care Medicine, pp. 1330–1337, Rippe, J. M., pressure in African children with cerebral malaria. Lancet 337: Irwin, R. S., Fink, M. P. and Cerra, F. B. (eds.) Little, Brown 573–576. and Company, New York. Newton, C. R., Winstanley, P. A. and Marsh, K. (1991b) Retinal Morrison, D. B. and Jeskey, H. A. (1948) Alterations in some con- haemorrhages in falciparum malaria. Arch. Dis. Child. 66: 753. stituents of the monkey erythrocytes infected with Plasmodium Newton, C. R., Peshu, N., Kendall, B., Kirkham, F. J., Sowunmi, knowlesi as related to pigment formation. J. Nat. Malaria Soc. 7: A., Waruiru, C., Mwangi, I., Murphy, S. A. and Marsh, K. 259–264. (1994) Brain swelling and ischaemia in Kenyans with cerebral Muizelaar, J. P., Lutz, H. A. and Becker, D. P. (1984) Effect of malaria. Arch. Dis. Child. 70: 281–287. mannitol on ICP and CBF and correlation with pressure auto- Newton, C. R. J. C., Marsh, K., Peshu, N. and Kirkham, F. J. regulation in severely head-injured patients. J. Neurosurg. 61: (1996a) Perturbations of cerebral hemodynamics in Kenyan 700–706. children with cerebral malaria. Pediatr. Neurol. 15: 41–49. Muntendam, A. H., Jaffar, S., Bleichrodt, N. and van Hensbroek, Newton, C. R. J. C., Peshu, N., Chong, K., Crawley, J. and Marsh, M. B. (1996) Absence of neuropsychological sequelae following K. (1996b) Computerized tomographic appearances of Kenyan cerebral malaria in Gambian children. Trans. R. Soc. Trop. children with neurological sequelae following cerebral malaria. Med. Hyg. 90: 391–394. Am. J. Trop. Med. Hyg. 55: 239. Murphy, S., English, M., Omar, A., Crawley, J., Waruiru, C., Newton, C. R. J. C., Chokwe, T., Armstrong-Schellenberg, J. R. M., Mwangi, I., Amukoye, E., Peshu, N., Newton, C. R., Winstan- Winstanley, P. A., Forster, D., Peshu, N., Kirkham, F. J. and ley, P. A. and Marsh, K. (1995a) The management of severe Marsh, K. (1997a) Coma scales in children with severe falci- malaria in children: a review. East Afr. Med. J. 72: 536–539. parum malaria. Trans. R. Soc. Trop. Med. Hyg. 91: 161–165. Murphy, S., Watkins, W. M., Bray, P. G., Lowe, B., Winstanley, Newton, C. R. J. C., Crawley, J., Sowunmi, A., Waruiru, C., P. A., Peshu, N. and Marsh, K. (1995b) Parasite viability during Mwangi, I., Murphy, S. A., English, M., Marsh, K. and treatment of severe falciparum malaria: differential effects of Kirkham, F. J. (1997b) Intracranial hypertension in Kenyan artemether and quinine. Am. J. Trop. Med. Hyg. 53: 303–305. children with cerebral malaria. Arch. Dis. Child. 76: 219–226. Murphy, S., English, M., Waruiru, C., Mwangi, I., Amukoye, E., Newton, C. R. J. C., Kirkham, F. J., Cope, M., Sowunmi, A., Crawley, J., Newton, C., Winstanley, P., Peshu, N. and Marsh, Peshu, N. and Marsh, K. (1997c) Changes in cerebral oxyhe- K. (1996) An open randomized trial of artemether versus qui- moglobin and deoxyhemoglobin during seizures in children nine in the treatment of cerebral malaria in African children. with cerebral malaria. J. Cereb. Blood Flow Metab. 17: S56. Trans. R. Soc. Trop. Med. Hyg. 90: 298–301. Newton, C. R. J. C., Warn, P. A., Winstanley, P. A., Peshu, N., Musoke, L. K. (1966) Neurological manifestations of malaria in Snow, R. W., Pasvol, G. and Marsh, K. (1997d) Contribution children. East Afr. Med. J. 43: 561–564. of falciparum malaria to severe anemia in Kenyan children. Nagatake, T., Hoang, V. T., Tegoshi, T., Rabbege, J., Ann, T. K. Trop. Med. Int. Health 2: 165–178. and Aikawa, M. (1992) Pathology of falciparum malaria in Newton, C. R. J. C., Whitten, R. O. and Taylor, T. E. (1998) Vietnam. Am. J. Trop. Med. Hyg. 47: 259–264. Pathophysiology of fatal falciparum malaria in African children. Nash, G. B., O’Brien, E., Gordon-Smith, E. C. and Dormandy, J. A. Am. J. Trop. Med. Hyg., in press. (1989) Abnormalities in the mechanical properties of red blood Nguyen, P. H., Day, N., Pram, T. D., Ferguson, D. J. and White, cells caused by Plasmodium falciparum. Blood 74: 855–861. N. J. (1995) Intraleucocytic malaria pigment and prognosis in Nash, G. B., Cooke, B. M., Marsh, K., Berendt, A., Newbold, C. severe malaria. Trans. R. Soc. Trop. Med. Hyg. 89: 200–204. and Stuart, J. (1992) Rheological analysis of the adhesive inter- Nguyen, P. H., Day, N. P., Ly, V. C., Waller, D., Nguyen, H. P., actions of red blood cells parasitized by Plasmodium falciparum. Bethell, D. B., Tran, T. H. and White, N. J. (1996) Post- Blood 79: 798–807. malaria neurological syndrome. Lancet 348: 917–921. Neequaye, J., Ofori-Adjei, E., Ofori-Adjei, D. and Renner, L. Nyakundi, J. N., Warn, P., Newton, C., Mumo, J. and Jephthah (1991) Comparative trial of oral versus intramuscular chlo- Ochola, J. (1994) Serum tumour necrosis factor in children suf- roquine in children with cerebral malaria. Trans. R. Soc. Trop. fering from Plasmodium falciparum infection in Kilifi District, Med. Hyg. 85: 718–722. Kenya. Trans. R. Soc. Trop. Med. Hyg. 88: 667–670. Nevill, C. G., Some, E. S., Mung’ala, V. O., Mutemi, W., New, L., O’Dempsey, T. J., McArdle, T. F., Laurence, B. E., Lamont, A. C., Marsh, K., Lengeler, C. and Snow, R. W. (1996) Insecticide- Todd, J. E. and Greenwood, B. M. (1993) Overlap in the clini- treated bednets reduce mortality and severe morbidity from cal features of pneumonia and malaria in African children. malaria among children on the Kenyan coast. Trop. Med. Int. Trans. R. Soc. Trop. Med. Hyg. 87: 662–665. Health 1: 139–146. Olumese, P. E., Sodeinde, O., Gbadegesin, R. A., Nafiu, O., Newton, C. R. J. C. (1992) The contribution of malaria to severe Oguche, S. and Walker, O. (1995) Respiratory distress adversely anaemia in children on the Kenyan coast. Br. J. Haematol. 82: affects the outcome of childhood cerebral malaria. Trans. R. Soc. 268–269. Trop. Med. Hyg. 89: 634. Newton, C. R. J. C. (1995) Intracranial hypertension in Kenyan Olurin, O. (1970) Cortical blindness following convulsions and children with cerebral malaria. M.D. Thesis, University of Cape fever in Nigerian children. Pediatrics 46: 102–107. Town. Omanga, U., Ngandu, K., Disengomoka, I. and Badibanga, B. Newton, C. R. J. C. and Warrell, D. A. (1998) Neurological manifes- (1977) Accès pernieux palustre chez l’enfant. Afrique Med. 16: tations of Plasmodium falciparum malaria. Ann. Neurol., in press. 507–516. Newton, C. R., Pasvol, G., Winstanley, P. A. and Warrell, D. A. Omanga, U., Ntihinyurwa, M., Shako, D. and Mashako, M. Pathophysiology and Supportive Treatment of Severe Malaria 49

(1983) Les hémiplegies au cours de l’acces pernicieux a plasmo- Pukrittayakamee, S., White, N. J., Davis, T. M. E., Looareesuwan, dium falciparum de l’enfant. Ann. Pediat. 30: 294–296. S., Supanaranond, W., Desakorn, V., Chaivisuth, B. and Will- Oo, M. M., Aikawa, M., Than, T., Aye, T. M., Myint, P. T., Iga- iamson, D. H. (1991b) Hepatic blood flow and metabolism in rashi, I. and Schoene, W. C. (1987) Human cerebral malaria: severe falciparum malaria: clearance of intravenously adminis- a pathological study. J. Neuropathol. Exp. Neurol. 46: 223–231. tered galactose. Clin. Sci. 82: 63–70. Orringer, C. E., Eustace, J. C., Wunsch, C. D. and Gardner, L. B. Pukrittayakamee, S., Supanaranond, W., Looareesuwan, S., Vani- (1977) Natural history of lactic acidosis after grand mal seizures. janonta, S. and White, N. J. (1994a) Quinine in severe falci- N. Engl. J. Med. 297: 796–797. parum malaria: evidence of declining efficacy in Thailand. Pasvol, G., Newton, C. R., Winstanley, P. A., Watkins, W. M., Trans. R. Soc. Trop. Med. Hyg. 88: 324–327. Peshu, N. M., Were, J. B., Marsh, K. and Warrell, D. A. (1991) Pukrittayakamee, S., Vanijanonta, S., Chantra, A., Clemens, R. Quinine treatment of severe falciparum malaria in African chil- and White, N. J. (1994b) Blood stage antimalarial efficacy of dren: a randomized comparison of three regimens. Am. J. Trop. primaquine in Plasmodium vivax malaria. J. Infect. Dis. 169: Med. Hyg. 45: 702–713. 932–935. Pasvol, G., Clough, B., Carlsson, J. and Snounou, G. (1995) The Pukrittayakamee, S., White, N. J., Davis, T. M., Supanaranond, pathogenesis of severe falciparum malaria. In: Malaria, pp. 249– W., Crawley, J., Nagachinta, B. and Williamson, D. H. (1994c) 270, Pasvol, G. (ed.) Baillière Tindall, London. Glycerol metabolism in severe falciparum malaria. Metabolism Petersen, C., Nelson, R., Magowan, C., Wollish, W., Jensen, J. 43: 887–892. and Leech, J. (1989) The mature erythrocyte surface antigen of Rampengan, T. H. (1991) Cerebral malaria in children. Compara- Plasmodium falciparum is not required for knobs or cytoadher- tive study between heparin, dexamethasone and placebo. Pae- ence. Mol. Biochem. Parasitol. 36: 61–66. diatr. Indones. 31: 59–66. Peto, T. E. and Thompson, J. L. (1986) A reappraisal of the effects of Rees, P. (1982) Dexamethasone deleterious in cerebral malaria iron and desferrioxamine on the growth of Plasmodium falciparum [letter]. Br. Med. J. 285: 1357. ‘in vitro’: the unimportance of serum iron. Br. J. Haematol. 63: Rey, M., Nouhouayi, A. and Diop Mar, L. (1966) Les expressions 273–280. cliniques du paludisme a plasmodium falciparum chez l’enfant Phillips, R. E., Warrell, D. A., White, N. J., Looareesuwan, S. and noir African. Bull. Soc. Pathol. Exot. 58: 683–704. Karbwang, J. (1985) Intravenous quinidine for the treatment of Riganti, M., Pongponratn, E., Tegoshi, T., Looareesuwan, S., Pun- severe falciparum malaria. Clinical and pharmacokinetic stud- poowong, B. and Aikawa, M. (1990) Human cerebral malaria ies. N. Engl. J. Med. 312: 1273–1278. in Thailand: a clinico-pathological correlation. Immunol. Lett. Phillips, R. E., Looareesuwan, S., Warrell, D. A., Lee, S. H., Kar- 25: 199–205. bwang, J., Warrell, M. J., White, N. J., Swasdichai, C. and Rigdon, R. H. and Fletcher, D. E. (1945) Lesions in the brain asso- Weatherall, D. J. (1986) The importance of anaemia in cere- ciated with malaria. Pathologic study on man and on experi- bral and uncomplicated falciparum malaria: role of complica- mental animals. Arch. Neurol. Psychiatry 53: 191–198. tions, dyserythropoiesis and iron sequestration. Q. J. Med. 58: Ritter, K., Brestrich, H., Nellen, B., Kratzin, H., Eiffert, H. and 305–323. Thomssen, R. (1990) Autoantibodies against triosephosphate Phillips, R. E., Looareesuwan, S., Molyneux, M. E., Hatz, C. and isomerase. A possible clue to pathogenesis of hemolytic anemia Warrell, D. A. (1993) Hypoglycaemia and counterregulatory in infectious mononucleosis. J. Exp. Med. 171: 565–570. hormone responses in severe falciparum malaria: treatment Ritter, K., Kuhlencord, A., Thomssen, R. and Bommer, W. (1993) with Sandostatin. Q. J. Med. 86: 233–240. Prolonged haemolytic anaemia in malaria and autoantibodies Phu, N. H., Day, N., Diep, P. T., Ferguson, D. J. P. and White, against triosephosphate isomerase. Lancet 342: 1333–1334. N. J. (1995) Intraleucocyte malaria pigment and prognosis in Roberts, D. J., Craig, A. G., Berendt, A. R., Pinches, R., Nash, G., severe malaria. Trans. R. Soc. Trop. Med. Hyg. 89: 200–204. Marsh, K. and Newbold, C. I. (1992) Rapid switching to multi- Plum, F. and Posner, J. B. (1980) The Diagnosis of Stupor and ple antigenic and adhesive phenotypes in malaria. Nature 357: Coma, 4th ed. F. A. Davis, Philadelphia. 689–692. Polder, T. W., Jerusalem, C. R. and Eling, W. M. (1991) Morpho- Rogerson, S. J., Reeder, J. C., al Yaman, F. and Brown, G. V. (1994) logical characteristics of intracerebral arterioles in clinical Sulfated glycoconjugates as disrupters of Plasmodium falciparum (Plasmodium falciparum) and experimental (Plasmodium berghei) erythrocyte rosettes. Am. J. Trop. Med. Hyg. 51: 198–203. cerebral malaria. J. Neurol. Sci. 101: 35–46. Rogier, C., Commenges, D. and Trape, J. F. (1996) Evidence of an Pongponratn, E., Riganti, M., Punpoowong, B. and Aikawa, M. age-dependent pyrogenic threshold of Plasmodium falciparum (1991) Microvascular sequestration of parasitized erythrocytes parasitaemia in highly endemic population. Am. J. Trop. Med. in human falciparum malaria: a pathological study. Am. J. Trop. Hyg. 54: 613–619. Med. Hyg. 44: 168–175. Rosenthal, P. J., Petersen, C., Geertsma, F. R. and Kohl, S. (1996) Porta, J., Carota, A., Pizzolato, G. P., Wildi, E., Widmer, M. C., Mar- Availability of intravenous quinidine for falciparum malaria gairaz, C. and Grau, G. E. (1993) Immunopathological changes [letter]. N. Engl. J. Med. 335: 138. in human cerebral malaria. Clin. Neuropathol. 12: 142–146. Rothe, H. (1956) One hundred cases of cerebral malaria. East Afr. Prada, J. and Kremsner, P. G. (1997) Reactive oxygen intermedi- Med. J. 33: 405–407. ates and cerebral malaria. Trans. R. Soc. Trop. Med. Hyg. 91: Rowe, A., Obeiro, J., Newbold, C. I. and Marsh, K. (1995) Plasmo- 92–93. dium falciparum rosetting is associated with malaria severity in Pukrittayakamee, S., Davis, T. M., Levy, J., Looareesuwan, S., Kenya. Infect. Immun. 63: 2323–2326. Roche, R. J., Jeerawathanapun, P. and White, N. J. (1991a) Royal College of Physicians (London) (1991) Guidelines for the The metabolic response to rapid intravenous glucose injection management of convulsions with fever. Br. Med. J. 303: 634– in acute falciparum malaria. Trans. R. Soc. Trop. Med. Hyg. 85: 636. 189–193. Saeed, B. O., Atabani, G. S., Nawwaf, A., Nasr, A. M., Abdul- 50 C. R. J. C. Newton and S. Krishna

hadi, N. H., Abu Zeid, Y. A., Alrasoul, M. A. and Bayoumi, R. A. Smith, T., Hurt, N., Teuscher, T. and Tanner, M. (1995) Is fever a (1990) Hypoglycaemia in pregnant women with malaria. Trans. good sign for clinical malaria in surveys of endemic communi- R. Soc. Trop. Med. Hyg. 84: 349–350. ties? Am. J. Trop. Med. Hyg. 52: 306–310. Sanohko, A., Dareys, J. P. and Charreau, M. (1968) État encépha- Snow, R. W., Omumbo, J. A., Lowe, B., Molyneux, C. S., Obiero, litique prolongé et acces pernicieux palustres. Bull. Soc. Med. J., Palmer, A., Weber, M. W., Pinder, M., Nahlen, B., Obonyo, C., Afr. Noire. Lang. Fr. 3: 662–669. Newbold, C., Gupta, S. and Marsh, K. (1997) Relation between Schmid, A. H. (1974) Cerebral malaria. On the nature and signif- severe malaria morbidity in children and level of Plasmodium icance of vascular changes. Eur. Neurol. 12: 197–208. falciparum transmission in Africa. Lancet 349: 1650–1654. Schmutzhard, E. and Gerstenbrand, F. (1984) Cerebral malaria in Solomon, T., Felix, J. M., Samuel, M., Dengo, G. A., Saldanha, Tanzania. Its epidemiology, clinical symptoms and neurological R. A., Schapira, A. and Phillips, R. E. (1994) Hypoglycaemia long term sequelae in the light of 66 cases. Trans. R. Soc. Trop. in paediatric admissions in Mozambique. Lancet 343: 149–150. Med. Hyg. 78: 351–353. Sowunmi, A. (1993) Psychosis after cerebral malaria in children. Schofield, L. and Hackett, F. (1993) Signal transduction in host J. Natl. Med. Assoc. 85: 695–696. cells by a glycosylphosphatidylinositol toxin of malaria para- Sowunmi, A. (1996a) Hyponatraemia in severe falciparum sites. J. Exp. Med. 177: 145–153. malaria: a clinical study of 19 comatose African children. Afr. J. Schofield, L., Novakovic, S., Gerold, P., Schwarz, R. T., McCon- Med. Med. Sci. 25: 47–52. ville, M. J. and Tachado, S. D. (1996) Glycosylphosphatidyli- Sowunmi, A. (1996b) Renal impairment in children with severe nositol toxin of Plasmodium up-regulates intercellular adhesion falciparum malaria. Arch. Dis. Child. 74: 293–298. molecule-1, vascular cell adhesion molecule-1, and E-selectin Sowunmi, A., Walker, O. and Salako, L. A. (1992) Hyperparasi- expression in vascular endothelial cells and increases leukocyte taemia: not a reliable indicator of severity or poor prognosis in and parasite cytoadherence via tyrosine kinase-dependent sig- falciparum malaria in children in endemic African countries. nal transduction. J. Immunol. 156: 1886–1896. Ann. Trop. Paediatr. 12: 155–158. Scuderi, P., Sterling, K. E., Lam, K. S., Finley, P. R., Ryan, K. J., Spitz, S. (1946) The pathology of acute falciparum malaria. Milit. Ray, C. G., Petersen, E., Slymen, D. J. and Salmon, S. E. (1986) Surg. 99: 555–572. Raised serum levels of tumour necrosis factor in parasitic infec- Srichaikul, T., Pulket, C., Sirisatepisarn, T. and Prayoonwiwat, tions. Lancet ii: 1364–1365. W. (1988) Platelet dysfunction in malaria. Southeast Asian J. Sein, K. K., Brown, A. E., Maeno, Y., Smith, C. D., Corcoran, K. D., Trop. Med. Public Health 19: 225–233. Hansukjariya, P., Webster, H. K. and Aikawa, M. (1993) Stace, J., Bilton, P., Coates, K. and Stace, N. (1980) Cerebral Sequestration pattern of parasitized erythrocytes in cerebrum, malaria in children: retrospective study of admissions to mid-brain, and cerebellum of Plasmodium coatneyi-infected rhesus Madang hospital. PNG Med. J. 25: 230–234. monkeys (Macaca mulatta). Am. J. Trop. Med. Hyg. 49: 513–519. Stacpoole, P. W. (1986) Lactic acidosis: the case against bicarbon- Shaffer, N., Grau, G. E., Hedberg, K., Davachi, F., Lyamba, B., High- ate therapy. Ann. Intern. Med. 105: 276–279. tower, A. W., Breman, J. G. and Phuc, N. D. (1991) Tumor Stacpoole, P. W. (1989) The pharmacology of dichloroacetate. necrosis factor and severe malaria. J. Infect. Dis. 163: 96–101. Metabolism 38: 1124–1144. Sharma, Y. D. (1991) Knobs, knob proteins and cytoadherence in Stacpoole, P. W. (1997) Lactic acidosis and other mitochondrial falciparum malaria. Int. J. Biochem. 23: 775–789. disorders. Metabolism 46: 306–321. Sherman, I. W. (1979) Biochemistry of Plasmodium (malarial par- Stacpoole, P. W., Wright, E. C., Baumgartner, T. G., Bersin, R. M., asites). Microbiol. Rev. 43: 453–495. Buchalter, S., Curry, S. H., Duncan, C. A., Harman, E. M., Hend- Shorvon, S. (1994) Status Epilepticus. Its Clinical Features and erson, G. N., Jenkinson, S., Lachin, J. M., Lorenz, A., Schneider, Treatment in Children and Adults. Cambridge University S. H., Siegel, S., Summer, S., Thompson, D., Wollfe, C. L., Press, Cambridge. Zorovich, B. and the Dichloroacetate-Lactic Acidosis Study Siesjo, B. K. (1978) Brain Energy Metabolism. John Wiley & Group (1992) A controlled clinical trial of dichloroacetate for Sons, Chichester. treatment of lactic acidosis in adults. The Dichloroacetate-Lac- Siesjo, B. K., Agard, C. and Bengtsson, F. (1989) Free radicals and tic Acidosis Study Group. N. Engl. J. Med. 327: 1564–1569. brain damage. Cerebrovasc. Brain Metab. Rev. 1: 165–211. Stacpoole, P. W., Wright, E. C., Baumgartner, T. G., Bersin, R. M., Silamut, K. and White, N. J. (1993) Relation of the stage of para- Buchalter, S., Curry, S. H., Duncan, C., Harman, E. M., Hender- site development in the peripheral blood to prognosis in severe son, G. N., Jenkinson, S., Lachin, J. M., Lorenz, A., Schneider, falciparum malaria. Trans. R. Soc. Trop. Med. Hyg. 87: 436– S. H., Siegel, S., Summer, S., Thompson, D., Wollfe, C. L., 443. Zorovich, B. and the DCA-Lactic Acidosis Study Group (1994) Slater, A. F. G. and Cerami, A. (1992) Inhibition by chloroquine Natural history and course of acquired lactic acidosis in adults. of a novel haem polymerase activity in malaria trophozoites. DCA-Lactic Acidosis Study Group. Am. J. Med. 97: 47–54. Nature 355: 167–169. Steele, R. W., Tanaka, P. T., Lara, R. P. and Bass, J. W. (1970) Slutsker, L., Taylor, T. E., Wirima, J. J. and Steketee, R. W. Evaluation of sponging and of oral antipyretic therapy to reduce (1994) In-hospital morbidity and mortality due to malaria-asso- fever. J. Pediatr. 77: 824–829. ciated severe anaemia in two areas of Malawi with different pat- Steketee, R. W., Wirima, J. J., Slutsker, L., Heymann, D. L. and terns of malaria infection. Trans. R. Soc. Trop. Med. Hyg. 88: Breman, J. G. (1996) The problem of malaria and malaria con- 548–551. trol in pregnancy in sub-Saharan Africa. Am. J. Trop. Med. Hyg. Smith, J. D., Chitnis, C. E., Craig, A. G., Roberts, D. J., Hudson- 55: 2–7. Taylor, D. E., Peterson, D. S., Pinches, R., Newbold, C. and Stettler, N., Schutz, Y., Whitehead, R. and Jequier, E. (1992) Miller, L. H. (1995) Switches in expression of Plasmodium falci- Effect of malaria and fever on energy metabolism in Gambian parum var genes correlate with changes in antigenic and cytoad- children. Pediatr. Res. 31: 102–106. herent phenotypes of infected erythrocytes. Cell 82: 101–110. Stocker, R., Hunt, N. H., Clark, I. A. and Weidemann, M. J. Pathophysiology and Supportive Treatment of Severe Malaria 51

(1984) Production of luminol-reactive oxygen radicals during Thuma, P. E., Mabeza, G. F., Biemba, G., Bhat, G. J., McLaren, Plasmodium vinckei infection. Infect. Immun. 45: 708–712. C. E., Moyo, V. M., Zulu, S., Khumalo, H., Mabeza, P., Stoute, J. A., Slaoui, M., Heppner, G., Momin, P., Kester, K. E., M’Hango, A. M., Parry, D., Poltera, A. A., Brittenham, G. M. Desmons, P., Wellde, B. T., Garcon, N., Krzych, U., Ballou, R. and Gordeuk, V. R. (1998) Effect of iron chelation therapy on and Cohen, J. D. for the RTSS Malaria Vaccine Evaluation mortality in Zambian children with cerebral malaria. Trans. R. Group (1997) A preliminary evalution of a recombinant cir- Soc. Trop. Med. Hyg. 92: 214–218. cumsporozoite protein vaccine against Plasmodium falciparum Thumwood, C. M., Hunt, N. H., Clark, I. A. and Cowden, W. B. malaria. N. Engl. J. Med. 336: 86–91. (1988) Breakdown of the blood-brain barrier in murine cerebral Sturchler, D. (1990) How much malaria is there worldwide? Para- malaria. Parasitology 96 (Part 3): 579–589. sitol. Today 5: 12. Thumwood, C. M., Hunt, N. H., Cowden, W. B. and Clark, I. A. Su, X. Z., Heatwole, V. M., Wertheimer, S. P., Guinet, F., Herr- (1989) Antioxidants can prevent cerebral malaria in Plasmo- feldt, J. A., Peterson, D. S., Ravetch, J. A. and Wellems, T. E. dium berghei-infected mice. Br. J. Exp. Pathol. 70: 293–303. (1995) The large diverse gene family var encodes proteins Thun-Hohenstein, L., Schmitt, B., Steinlin, H., Martin, E. and involved in cytoadherence and antigenic variation of Plasmo- Boltshauser, E. (1992) Cortical visual impairment following bac- dium falciparum-infected erythrocytes. Cell 82: 89–100. terial meningitis: magnetic resonance imaging and visual evoked Sullivan, A. D. and Meshnick, S. R. (1996) Haemozoin: identifi- potentials findings in two cases. Eur. J. Pediatr. 151: 779–782. cation and quantification. Parasitol. Today 12: 161–163. Toro, G. and Roman, G. (1978) Cerebral malaria. A disseminated Sullivan, D. J., Gluzman, I. Y. and Goldberg, D. E. (1996) Plasmo- vasculomyelinopathy. Arch. Neurol. 35: 271–275. dium hemozoin formation mediated by histidine rich proteins. Tran, T. H., Day, N. P., Ly, V. C., Nguyen, T. H., Pham, P. L., Science 271: 219–222. Nguyen, H. P., Bethell, D. B., Dihn, X. S. and White, N. J. Tachado, S. D., Gerold, P., McConville, M. J., Baldwin, T., Qui- (1996a) Blackwater fever in southern Vietnam: a prospective lici, D., Schwarz, R. T. and Schofield, L. (1996) Glycosylphos- descriptive study of 50 cases. Clin. Infect. Dis. 23: 1274–1281. phatidylinositol toxin of Plasmodium induces nitric oxide Tran, T. H., Day, N. P., Nguyen, H. P., Nguyen, T. H., Pham, P. L., synthase expression in macrophages and vascular endothelial Dinh, X. S., Ly, V. C., Ha, V., Waller, D., Peto, T. E. and cells by a protein tyrosine kinase-dependent and protein kinase White, N. J. (1996b) A controlled trial of artemether or qui- C-dependent signaling pathway. J. Immunol. 156: 1897–1907. nine in Vietnamese adults with severe falciparum malaria [see Targett, G. A. T. (1995) Malarial vaccines. Curr. Opin. Infect. comments]. N. Engl. J. Med. 335: 76–83. Dis. 8: 322–327. Trang, T. T., Phu, N. H., Vinh, H., Hien, T. T., Cuong, B. M., Taylor, T. E., Molyneux, M. E., Wirima, J. J., Fletcher, A. and Chau, T. T., Mai, N. T., Waller, D. J. and White, N. J. (1992) Morris, K. (1988) Blood glucose levels in Malawian children Acute renal failure in patients with severe falciparum malaria. before and during the adminstration of intravenous quinine for Clin. Infect. Dis. 15: 874–880. severe falciparum malaria. N. Engl. J. Med. 319: 1040–1047. Trape, J. F., Quinet, Q. M. C., Nzigoula, S., Senga, P., Tchichelle, Taylor, T. E., Molyneux, M. E., Wirima, J. J., Borgstein, A., Gold- F., Carme, B., Candito, D., Mayanda, H. and Zoulani, A. (1987) ring, J. D. and Hommel, M. (1992) Intravenous immunoglobu- Malaria and urbanization in central Africa: the example of Braz- lin in the treatment of paediatric cerebral malaria. Clin. Exp. zaville. Part V: pernicious attacks and mortality. Trans. R. Immunol. 90: 357–362. Soc. Trop. Med. Hyg. 81 (Suppl. 2): 34–42. Taylor, T. E., Borgstein, A. and Molyneux, M. E. (1993) Acid- Treluyer, J. M., Roux, A., Mugnier, C., Flouvat, B. and Lagardere, base status in paediatric Plasmodium falciparum malaria. Q. J. B. (1996) Metabolism of quinine in children with global mal- Med. 86: 99–109. nutrition. Pediatr. Res. 40: 558–563. Teasdale, G. and Jennett, B. (1974) Assessment of coma and Treutiger, C. J., Hedlund, I., Helmby, H., Carlson, J., Jepson, A., impaired consciousness. A practical scale. Lancet ii: 81–84. Twumasi, P., Kwiatkowski, D., Greenwood, B. M. and Wahlgren, ter Kuile, F., Nosten, F., Chongsuphajaisiddhi, T., Holloway, P., M. (1992) Rosette formation in Plasmodium falciparum isolates Maelankirri, L. and White, N. J. (1992) Absorption of intra- and anti-rosette activity of sera from Gambians with cerebral or muscular phenobarbitone in children with severe falciparum uncomplicated malaria. Am. J. Trop. Med. Hyg. 46: 503–510. malaria. Eur. J. Clin. Pharmacol. 42: 107–110. Turner, G. (1997) Cerebral malaria. Brain Pathol. 7: 569–582. ter Kuile, F., White, N. J., Holloway, P., Pasvol, G. and Krishna, S. Turner, G. D. H., Morrison, H., Jones, M., Davis, T. M. E., Looa- (1993) Plasmodium falciparum: in vitro studies of the pharmaco- reesuwan, S., Buley, I. D., Gatter, K. C., Newbold, C. I. and dynamic properties of drugs used for the treatment of severe Pukrittayakamee, S. (1994) An immunohistochemical study of malaria. Exp. Parasitol. 76: 85–95. the pathology of fatal malaria. Am. J. Pathol. 145: 1057–1069. Thapa, B. R., Marwaha, R. K., Kumar, L. and Mehta, S. (1988) Udomsangpetch, R., Wåhlin, B., Carlson, J., Berzins, K., Torii, M., Cerebral malaria in children: therapeutic considerations. Aikawa, M., Perlmann, P. and Wahlgren, M. (1989) Plasmo- Ind. Paediatr. 25: 61–65. dium falciparum-infected erythrocytes form spontaneous eryth- Thomas, J. D. (1971) Clinical and histopathological correlation of rocyte rosettes. J. Exp. Med. 169: 1835–1840. cerebral malaria. Trop. Geogr. Med. 23: 232–238. Udomsangpetch, R., Thanikkul, K., Pukrittayakamee, S. and Thomson, J. G. and Annecke, S. (1926) Observations of the White, N. J. (1995) Rosette formation by Plasmodium vivax. pathology of the central nervous system in malignant tertian Trans. R. Soc. Trop. Med. Hyg. 89: 635–637. malaria, with remarks on certain clinical phenomena. J. Trop. Udomsangpetch, R., Pipitaporn, B., Krishna, S., Angus, B., Pukrit- Med. Hyg. 29: 313–346. tayakamee, S., Bates, I., Suputtamongkol, Y., Kyle, D. E. and Thuma, P. E., Weiss, G., Herold, M. and Gordeuk, V. R. (1996) White, N. J. (1996) Antimalarial drugs reduce cytoadherence and Serum neopterin, interleukin-4, and interleukin-6 concentra- rosetting of Plasmodium falciparum. J. Infect. Dis. 173: 691–698. tions in cerebral malaria patients and the effect of iron chela- Usawattanakul, W., Tharavanij, S., Warrell, D. A., Looareesuwan, tion therapy. Am. J. Trop. Med. Hyg. 54: 164–168. S., White, N. J., Supavej, S. and Soikratoke, S. (1985) Factors 52 C. R. J. C. Newton and S. Krishna

contributing to the development of cerebral malaria. II. Endo- Waller, D., Krishna, S., Crawley, J., Miller, K., Nosten, F., Chap- toxin. Clin. Exp. Immunol. 61: 562–568. man, D., ter Kuile, F. O., Craddock, C., Berry, C., Holloway, P. A., Vallance, P. and Collier, J. (1994) Biology and clinical relevance Brewster, D., Greenwood, B. M. and White, N. J. (1995) Clini- of nitric oxide. Br. Med. J. 309: 453–456. cal features and outcome of severe malaria in Gambian chil- Van den Ende, J., Moorkens, G., Van Gómpel, A., Demey, H., dren. Clin. Infect. Dis. 21: 577–587. Lins, R., Maldague, P., Pelfrene, E., Van den Enden, E., Tael- Warrell, D. A. (1987) Pathophysiology of severe falciparum man, H., Van der Stuyft, P. and Colebunders, R. (1994) Twelve malaria in man. Parasitology 94 (Suppl.): S53–S76. patients with severe malaria treated with partial exchange trans- Warrell, D. A. (1989) Cerebral malaria [editorial]. Q. J. Med. 71: fusion. Comparison between mathematically predicted and 369–371. observed effect on parasitaemia. Trop. Geogr. Med 46: 340–345. Warrell, D. A., Looareesuwan, S., Warrell, M. J., Kasemsarn, P., van Hensbroek, M. B. (1996) Cerebral malaria in children: patho- Intaraprasert, R., Bunnag, D. and Harinasuta, T. (1982a) Dexam- genesis, treatment and outcome. Ph.D. Thesis, University of ethasone proves deleterious in cerebral malaria. A double-blind Amsterdam. trial in 100 comatose patients. N. Engl. J. Med. 306: 313–319. van Hensbroek, M. B., Onyiorah, E., Jaffar, S., Schneider, G., Warrell, D. A., White, N. J. and Warrell, M. J. (1982b) Dexa- Palmer, A., Frenkel, J., Enwere, G., Forck, S., Nusmeijer, A., methasone deleterious in cerebral malaria [letter]. Br. Med. J. Bennett, S., Greenwood, B. and Kwiatkowski, D. (1996a) A [Clin. Res.] 285: 1652. trial of artemether or quinine in children with cerebral malaria. Warrell, D. A., Looareesuwan, S., Phillips, R. E., White, N. J., N. Engl. J. Med. 335: 69–75. Warrell, M. J., Chapel, H. M., Areekul, S. and Tharavanij, S. van Hensbroek, M. B., Palmer, A., Onyiorah, E., Schneider, G., (1986) Function of the blood-cerebrospinal fluid barrier in Jaffar, S., Dolan, G., Memming, H., Frenkel, J., Enwere, G., Ben- human cerebral malaria: rejection of the permeability hypothe- nett, S., Kwiatkowski, D. and Greenwood, B. (1996b) The effect sis. Am. J. Trop. Med. Hyg. 35: 882–889. of a monoclonal antibody to tumor necrosis factor on survival Warrell, D. A., White, N. J., Veall, N., Looareesuwan, S., Chantha- from childhood cerebral malaria. J. Infect. Dis. 174: 1091–1097. vanich, P., Phillips, R. E., Karbwang, J., Pongpaew, P. and Krishna, van Hensbroek, M., Palmer, A., Jaffar, S., Schneider, G. and Kwi- S. (1988) Cerebral anaerobic glycolysis and reduced cerebral atkowski, D. (1997) Residual neurologic sequelae after child- oxygen transport in human cerebral malaria. Lancet ii: 534–538. hood malaria. J. Pediatr. 131: 125–129. Warrell, D. A., Molyneux, M. E. and Beales, P. F. (1990) Severe Vanijanonta, S., Chantra, A., Phophak, N., Chindanond, D., Cle- and complicated malaria. Trans. R. Soc. Trop. Med. Hyg. 84 mens, R. and Pukrittayakamee, S. (1996) Therapeutic effects of (Suppl. 2): 1–65. chloroquine in combination with quinine in uncomplicated fal- Waruiru, C. M., Newton, C. R. Forster, D., New, L., Winstanley, ciparum malaria. Ann. Trop. Med. Parasitol. 90: 269–275. P., Mwangi, I., Marsh, V., Winstanley, M., Snow, R. W. and Vietze, G. (1978) Malaria and other protozoal diseases. In: Infec- Marsh, K. (1996) Epileptic seizures and malaria in Kenyan chil- tions of the Nervous System, Part III, Handbook of Clinical dren. Trans. R. Soc. Trop. Med. Hyg. 90: 152–155. Neurology, Vol. 35, pp. 143–160, Vinken, P. J. and Bruyn, G. W. Wasterlain, C. G. and Shirasaka, Y. (1994) Seizures, brain damage (eds.) Elsevier/North Holland Biomedical Press, Amsterdam. and brain development. Brain Dev. 16: 279–295. von Seidlein, L., Jaffar, S. and Greenwood, B. M. (1997) Prolon- Watkins, W. M., Woodrow, C. and Marsh, K. (1993) Falciparum gation of the QTc interval in African children treated for falci- malaria: differential effects of antimalarial drugs on ex vivo para- parum malaria. Am. J. Trop. Med. Hyg. 56: 494–497. site viability during the critical early phase of therapy. Am. J. Vouldoukis, I., Rivernos-Moreno, V., Dugas, B., Ouaaz, F., Trop. Med. Hyg. 49: 106–112. Becherael, P., Debre, P., Moncada, S. and Mossalayi, M. D. Wattanagoon, Y., Srivilairit, S., Looareesuwan, S. and White, N. J. (1995) The killing of Leishmania major by human macrophages (1994) Convulsions in childhood malaria. Trans. R. Soc. Trop. by nitric oxide induced after the ligation of FCeRII/CD23 sur- Med. Hyg. 88: 426–428. face antigen. Proc. Natl. Acad. Sci. USA 92: 7804–7808. Wei, X., Charles, I. G., Smith, A., Ure, J., Feng, G., Huang, F., Wahlgren, M., Carlson, J., Udomsangpetch, R. and Perlmann, P. Xu, D., Muller, W., Moncada, S. and Liew, F. Y. (1995) Altered (1991) Why do Plasmodium falciparum-infected erythrocytes form immune response in mice lacking inducible nitric oxide syn- spontaneous erythrocyte rosettes. Parasitol. Today 5: 183–185. thase. Nature 375: 408–411. Walker, O., Salako, L. A., Sowunmi, A., Thomas, J. O., Sodeine, Weiss, G., Thuma, P. E., Mabeza, G., Werner, E. R., Herold, M. O. and Bondi, F. S. (1992) Prognostic risk factors and post mor- and Gordeuk, V. R. (1997) Modulatory potential of iron chela- tem findings in cerebral malaria in children. Trans. R. Soc. tion therapy on nitric oxide formation in cerebral malaria. J. Trop. Med. Hyg. 86: 491–493. Infect. Dis. 175: 226–230. Walker, O., Salako, L. A., Omokhodion, S. I. and Sowunmi, A. Westendorp, R. G. J., Langermans, J. A. M., Huizinga, T. W. J., (1993) An open randomized comparative study of intramuscu- Elouali, A. H., Verweij, C. L., Boomsma, D. I. and Vanden- lar artemether and intravenous quinine in cerebral malaria in borouke, J. P. (1997) Genetic influence in cytokine production children. Trans. R. Soc. Trop. Med. Hyg. 87: 564–566. and fatal meningococcal disease. Lancet 349: 170–173. Waller, D., Krishna, S., Craddock, C., Brewster, D., Jammeh, A., Weston, M. J., Jackman, N., Rudge, C., Bowles, J., Brady, C., Kwiatkowski, D., Karbwang, J., Molunto, P. and White, N. J. Bielawska, C. and O’Grady, J. (1982) Prostacyclin in falci- (1990) The pharmacokinetic properties of intramuscular qui- parum malaria. Lancet ii: 609. nine in Gambian children with severe falciparum malaria. White, N. J. (1991) Lumbar puncture in cerebral malaria. Lancet Trans. R. Soc. Trop. Med. Hyg. 84: 488–491. 338: 640–641. Waller, D., Crawley, J., Nosten, F., Chapman, S., Krishna, S., White, N. J. (1992) Antimalarial drug resistance: the pace quick- Craddock, C., Brewster, D. and White, N. J. (1991) Intracra- ens. J. Antimicrob. Chemother. 30: 571–585. nial pressure in childhood cerebral malaria. Trans. R. Soc. White, N. J. and Ho, M. (1992) The pathophysiology of malaria. Trop. Med. Hyg. 85: 362–364. Adv. Parasitol. 31: 83–173. Pathophysiology and Supportive Treatment of Severe Malaria 53

White, N. J. and Krishna, S. (1989) Treatment of malaria: some Winstanley, P. A., Newton, C. R. J. C., Pasvol, G., Kirkham, F. J., considerations and limitations of the current methods of assess- Mberu, E., Peshu, N., Ward, A. A., Were J. B. O., Warrell, D. A. ment. Trans. R. Soc. Trop. Med. Hyg. 83: 767–777. and Marsh, K. (1992) Prophylactic phenobarbitone in young White, N. J. and Looareesuwan, S. (1987) Cerebral malaria. In: children with severe falciparum malaria: pharmacokinetics and Infections of the Nervous System, pp. 118–143, Kennedy, P. G. E. clinical effects. Br. J. Clin. Pharmacol. 33: 149–154. and Johnson, R. T. (eds.) Butterworths, London. Winstanley, P., Newton, C., Watkins, W., Mberu, E., Ward, S., White, N. J., Warrell, D. A., Chanthavanich, P., Looareesuwan, Warn, P., Mwangi, I., Waruiru, C., Pasvol, G., Warrell, D. and S., Warrell, M. J., Krishna, S., Williamson, D. H. and Turner, Marsh, K. (1993) Towards optimal regimens of parenteral qui- R. C. (1983) Severe hypoglycemia and hyperinsulinemia in fal- nine for young African children with cerebral malaria: the ciparum malaria. N. Engl. J. Med. 309: 61–66. importance of unbound quinine concentration. Trans. R. Soc. White, N. J., Warrell, D. A., Looareesuwan, S., Chanthavanich, Trop. Med. Hyg. 87: 201–206 P., Phillips, R. E. and Pongpaew, P. (1985) Pathophysiological Winstanley, P. A., Mberu, E. K., Watkins, W. M., Murphy, S. A., and prognostic significance of cerebrospinal-fluid lactate in Lowe, B. and Marsh, K. (1994) Towards optimal regimens of cerebral malaria. Lancet i: 776–778. parenteral quinine for young African children with cerebral White, N. J., Miller, K. D., Brown, J., Marsh, K. and Greenwood, malaria: unbound quinine concentrations following a simple load- B. M. (1987a) Prognostic value of CSF lactate in cerebral ing dose regimen. Trans. R. Soc. Trop. Med. Hyg. 88: 577–580. malaria [letter]. Lancet i: 1261. Wolf-Gould, C., Osei, L., Commey, J. O. O. and Bia, F. J. (1991) White, N. J., Miller, K. D., Marsh, K., Berry, C. D., Turner, R. C., Pediatric cerebral malaria in Accra, Ghana. J. Trop. Pediatr. 38: Williamson, D. H. and Brown, J. (1987b) Hypoglycaemia in 290–294. African children with severe malaria. Lancet i: 708–711. Woodruff, A. W. and Dickinson, C. J. (1968) Use of dexametha- White, N. J., Watt, G., Bergqvist, Y. and Njelesani, E. K. (1987c) sone in cerebral malaria. Br. Med. J. 3: 31–32. Parenteral chloroquine for treating falciparum malaria. J. Infect. World Health Organisation (1993) A global strategy to control Dis. 155: 192–201. malaria. Report No. 839, Geneva. White, N. J., Looareesuwan, S., Phillips, R. E., Chanthavanich, P. World Health Organisation (1996) Malaria. Report No. 94 Fact and Warrell, D. A. (1988) Single dose phenobarbitone prevents sheet, Geneva. convulsions in cerebral malaria. Lancet ii: 64–66. Wright, B. D. (1968) The effect of neonatal thymectomy on the White, N. J., Chapman, D. and Watt, G. (1992a) The effects of survival of golden hamsters infected with Plasmodium berghei. multiplication and synchronicity on the vascular distribution of Br. J. Exp. Pathol. 49: 379–384. parasites in falciparum malaria. Trans. R. Soc. Trop. Med. Hyg. Wright, D. H., Masembe, R. M. and Bazira, E. R. (1971) The 86: 590–597. effect of anti-thymocyte serum on golden hamsters infected White, N. J., Krishna, S. and Looareesuwan, S. (1992b) Encepha- with Plasmodium berghei. Br. J. Exp. Pathol. 52: 465–477. litis, not cerebral malaria, is likely cause of coma with negative Wright, P. W., Avery, W. G., Ardill, W. D. and McLarty, J. W. blood smears [letter]. J. Infect. Dis. 166: 1195–1196. (1993) Initial clinical assessment of the comatose patient: cere- Whitehead, S. and Peto, T. E. (1990) Stage-dependent effect of bral malaria vs. meningitis. Pediatr. Infect. Dis. J. 12: 37–41. deferoxamine on growth of Plasmodium falciparum in vitro. Yen, L. M., Dao, L. M., Day, N. P., Waller, D. J., Bethell, D. B., Blood 76: 1250–1255. Son, L. H., Hien, T. T. and White, N. J. (1994) Role of quinine Wick, T. M. and Louis, V. (1991) Cytoadherence of Plasmodium in the high mortality of intramuscular injection tetanus. Lancet falciparum-infected erythrocytes to human umbilical vein and 344: 786–787. human dermal microvascular endothelial cells under shear con- Zucker, J. R., Lackritz, E. M., Ruebush, T. K., Hightower, A. W., ditions. Am. J. Trop. Med. Hyg. 45: 578–586. Adungosi, J. E., Were, J. B., Metchock, B., Patrick, E. and Wilkinson, R. J., Brown, J. L., Pasvol, G., Chiodini, P. L. and Campbell, C. C. (1996) Childhood mortality during and after Davidson, R. N. (1994) Severe falciparum malaria: predicting hospitalization in western Kenya: effect of malaria treatment the effect of exchange transfusion. Q. J. Med. 87: 553–557. regimens. Am. J. Trop. Med. Hyg. 55: 655–660.