284

Am. J. Trop. Med. Hyg., 62(2), 2000, pp. 284–290Copyright� 2000 by The American Society of Tropical Medicine and Hygiene

CEREBROSPINAL FLUID STUDIES IN CHILDREN WITH CEREBRAL MALARIA: ANEXCITOTOXIC MECHANISM?

MICHAEL DOBBIE, JANE CRAWLEY, CATHERINE WARUIRU, KEVIN MARSH,AND ROBERT SURTEESInstitute of Child Health, University College London Medical School, London, United Kingdom; Kenya Medical Research InstituteCentre for Geographical Medicine Research Coast, Kilifi, Kenya; Nuffield Department of Clinical Medicine, John Radcliffe Hospital,

Headington, Oxford, United Kingdom

Abstract. The pathogenesis of cerebral malaria is poorly understood. One hypothesis is that activation of microgliaand astrocytes in the brain might cause the cerebral symptoms by excitotoxic mechanisms. Cerebrospinal fluid wassampled in 97 Kenyan children with cerebral malaria, 85% within 48 hr of admission. When compared with an age-matched reference range, there were large increases in concentrations of the excitotoxin quinolinic acid (geometricmean ratio cerebral malaria/reference population [95% confidence limits]� 14.1 [9.8–20.4],P � 0.001) and totalneopterin (10.9 [9.1–13.0],P � 0.001) and lesser increases in tetra-hydrobiopterin, di-hydrobiopterin, and 5-hydrox-yindoleacetic acid. There was no change in tryptophan concentration. In contrast, nitrate plus nitrite concentrationswere decreased (geometric mean ratio� 0.45 [0.35–0.59],P � 0.001). There was a graded increment in quinolinicacid concentration across outcome groups of increasing severity. The increased concentration of quinolinic acidsuggests that excitotoxic mechanisms may contribute to the pathogenesis of cerebral malaria.

Cerebral malaria is the most serious complication of in-fection withPlasmodium falciparum, which causes the deathof more than 1 million children in sub-Saharan Africa eachyear.1 The pathogenesis of childhood cerebral malaria is notwell understood.2 A central occurrence in the production ofthe clinical syndrome is adherence of parasitized erythro-cytes to endothelial cells of the cerebral vasculature. Thisinitiates a cascade of events that cause intense endothelialactivation on the luminal side of the blood-brain barrier.3

These events include cytoadherence itself,4 and the produc-tion of inflammatory cytokines,5–7 malarial toxins,8 and nitricoxide.9–12 Much less is known about events occurring on theabluminal side of the blood-brain barrier in cerebral malariaand the cause of cerebral symptoms.

The clinical syndrome of childhood cerebral malaria ischaracterized by the rapid onset and recovery from a diffuseencephalopathy.13 The coma is complicated by raised intra-cranial pressure and seizures occur in more than half thepatients.14,15 Between 5% and 15% of survivors of cerebralmalaria have neurologic sequelae suggesting focal braindamage.16,17 These clinical findings could be explained by anexcitotoxic mechanism.

Experimental stimulation of cells of the macrophage/monocyte lineage by various cytokines causes the parallelinduction of indoleamine-pyrrole 2,3-dioxygenase (EC1.13.11.42, indoleamine 2,3-dioxygenase), GTP cyclohydro-lase I (EC 3.5.4.16), and nitric oxide synthase (EC1.14.13.39).18,19 These enzymes catalyze the first step ofpathways that lead to the formation of quinolinic acid, neop-terin and tetrahydrobiopterin, and nitric oxide, respectively.In the brain, microglia constitute the resident macrophage/monocyte cells. Experimental stimulation of microglia withcytokines also causes induction of these enzymes.20,21

Quinolinic acid is an endogenous excitotoxin that is a se-lective agonist ofN-methyl-D-aspartate (NMDA) glutamatereceptors.22 When applied to the central nervous system ofexperimental animals or to cerebral tissue culture, quinolinicacid causes seizures, reversible neuronal swelling caused bysodium influx, and delayed neuronal disintegration causedby calcium influx.23–27 Thus, the toxic action of quinolinicacid could explain the neurologic complications of childhood

cerebral malaria: convulsions and reversible cerebral edema,and permanent neurologic damage.

Nitric oxide synthases catalyze the formation of the gasnitric oxide from arginine (Figure 1).28 The NMDA receptorsare modulated by endogenous nitric oxide;29–31 increasingconcentrations of nitric oxide cause receptor blockade whiledecreasing concentrations potentiate NMDA-induced calci-um influx. Nitric oxide synthases require tetrahydrobiopterinas an essential cofactor.32 GTP-cyclohydrolase catalyzes thefirst step in the tetrahydrobiopterin synthesis pathway (Fig-ure 1), in which dihydroneopterin triphosphate is an inter-mediary. Human macrophage/monocyte lineage cells are rel-atively deficient in the enzyme that catalyzes the formationof 6-pyruvoyltetrahydropterin from dihydroneopterin tri-phosphate,33 and when activated, cause the accumulation ofthe breakdown product neopterin. Neopterin concentrationin biological fluids is widely used as a marker of immuneactivation.34 Tetrahydrobiopterin is also required as a cofac-tor for tryptophan monooxygenase, which is the first andrate-limiting step of the serotonin biosynthetic pathway. 5-Hydroxyindoleacetic acid is an acidic metabolite of seroto-nin whose concentration in cerebrospinal fluid (CSF) reflectsserotonin turnover.

We hypothesized that the intense cerebrovascular endo-thelial activation in cerebral malaria causes activation of mi-croglia and/or astrocytes in the brain. The activated cellsthen produce the excitotoxin quinolinic acid, which may beone cause of the cerebral symptoms. To address this hypoth-esis, we have measured CSF concentrations of quinolinicacid, its precursor tryptophan, 5-hydroxyindoleacetic acid,individual pterin species, and nitrate plus nitrite (stablebreakdown products of nitric oxide) in CSF sampled fromKenyan children recovering from cerebral malaria.

PATIENTS AND METHODS

Ninety-seven children with cerebral malaria were studied.Their median (95th percentile confidence interval) age was2.2 (0.7–6.9) years, and nearly 70% were less than threeyears old. All were admitted to the hospital in Kilifi, a coast-al town in Kenya where malaria is endemic. Before devel-

285CSF BIOCHEMISTRY IN CEREBRAL MALARIA

FIGURE 1. Inter-relationships between tryptophan, pterin, and nitric oxide metabolism and the N-methyl-D-aspartate glutamate receptor. 5-HIAA � 5-hydroxyindoleacetic acid; BH2 � dihydrobiopterin; BH4 � tetrahydrobiopterin; NEO � dihydroneopterin triphosphate; GTP �guanosine triphosphate; NO � nitric oxide; NMDAR � the excitatory N-methyl-D-aspartate glutamate receptor. Enzyme 1 � indoleamine 2,3-dioxygenase; 2 � tryptophan monooxygenase; 3 � GTP cyclohydrolase I; 4 � nitric oxide synthase.

oping cerebral malaria, each child was well and had no fea-tures suggestive of encephalitis due to human immunodefi-ciency virus.

Two different syndromes were defined depending uponthe degree of loss of consciousness.35 Strict cerebral malariawas defined as a febrile encephalopathy with P. falciparumparasitemia in which children �8 months old were unableto localize pain and children �8 months old extended theirlimbs with pain. Children with P. falciparum parasitemia anda febrile encephalopathy causing a lesser disturbance in con-scious level (Blantyre coma scale �4) were classified as hav-ing malaria with impaired consciousness. All children weretreated in a standardized manner with either intravenous qui-nine (a 15 mg/kg loading dose, then 10 mg/kg every 12 hr)or intramuscular artemether (a 3.2 mg/kg loading dose, then1.6 mg/kg every 24 hr) and treatment was completed witha single dose of oral or intramuscular sulfadoxine (500 mg)/pyrimethamine (25 mg). The outcome for each child withcerebral malaria was determined on discharge from hospital:death, survival with neurologic sequelae, and survival withno neurologic sequelae.

Cerebrospinal fluid was obtained from each child once itwas clinically judged to be safe to do so: usually once theconscious level had started to increase. It was also takenpost-mortem within 15 min of death from 11 children. TheCSF was taken in aliquots: the first 1 ml was used for di-agnostic purposes, and the second 1 ml and third 1 ml wasused for this study. The third 1 ml was collected into a bottlecontaining 1 mg each of dithioerythritol (DTE) and diethy-lenetetraaminopentaacetic acid (DETAPAC). The second 1ml and third 1 ml were frozen at the bedside on dry ice andstored at �70�C until analysis. All CSF analyzed had a nor-mal cell count.

These studies were approved by the Ethical Committee ofthe Kenya Medical Research Institute. Lumbar puncture was

performed as a routine clinical practice. Verbal consent wasobtained from the patients for the storage and subsequentanalysis of redundant CSF for research purposes.

The concentrations of the following compounds weremeasured in CSF: quinolinic acid, tryptophan, 5-hydroxyin-doleacetic acid, total neopterin, tetrahydrobiopterin, dihydro-biopterin, and nitrate plus nitrite. Quinolinic acid was mea-sured by gas chromatography electron impact mass spec-trometry after tert-butyldimethylsilyl derivatization.36 Tryp-tophan was measured by high-performance liquidchromatography (HPLC) with fluorometric detection afterprecolumn derivatization with o-phthalaldialdehyde-2-mer-captoethanol.37 5-Hydroxyindoleacetic acid was measured byHPLC with electrochemical detection.38 Total neopterin(neopterin plus dihydroneopterin, breakdown products ofdihydroneopterin triphosphate), tetrahydrobiopterin, and di-hydrobiopterin were measured by HPLC with dual electro-chemical and fluorometric detection.38 Nitrate plus nitritewere measured by spectrophotometry.39 Reference ranges forthe metabolite concentrations were constructed from chil-dren and young adults living in the United Kingdom with avariety of neurologic or metabolic diseases in whom no dis-turbance of the biochemical pathways was expected; CSFwas sampled after at least a 4-hr fast.36,38,39 With the excep-tion of total neopterin, each metabolite shows an age-relateddecrement in CSF concentration and the reference ranges (n� 24–78 depending upon the metabolite assayed) were ex-actly age-matched to the children with cerebral malaria. Wewere also able to collect an age-matched local referencegroup of 9 children who had seizures or staging for Burkitt’slymphoma for comparison of nitrate plus nitrite concentra-tions; they were not suitable as a reference group for quin-olinic acid or the pterin species because the majority wereinvestigated in the context of a fever.

The results were analyzed after logarithmic transformation

286 DOBBIE AND OTHERS

TABLE 1Cerebrospinal fluid metabolite concentrations in cerebral malaria

Geometric mean(95th centile confidence limits)

Reference range Cerebral malaria

Quinolinic acid (nmol/L) 16.2(12.3–21.5)

229*(188–278)

Nitrate plus nitrite (�mol/L) 12.6(10.5–15.1)

5.7†(5.1–6.5)

Total neopterin (nmol/L) 17.0(15.2–19.1)

186*(162–213)

Tetrahydrobiopterin (nmol/L) 33.9(30.2–38.0)

119*(106–134)

Dihydrobiopterin (nmol/L) 3.8(2.9–4.9)

12.7*(11.2–14.4)

5-Hydroxyindoleacetic acid (nmol/L) 194(172–218)

233*(210–259)

Tryptophan (�mol/L) 2.7(2.4–3.0)

2.5(2.1–3.0)

* Significantly increased (Student’s t-test).† Significantly reduced (Student’s t-test).

to equalize the variances. Mean metabolite concentrationswere compared across the diagnostic and treatment groupsusing Student’s t-test. In children with cerebral malaria,mean metabolite concentrations were compared across theoutcome groups using analysis of variance with post hocsignificance testing using Duncan’s new multiple range test.P values �0.05 were considered statistically significant.

RESULTS

Eighty-two children (85%) had CSF taken within 48 hr ofadmission. There were no differences in any metabolite con-centration in CSF fluid taken within 48 hr of admission (n� 82) compared with that taken after 48 hr (n � 15). Therewere also no significant differences in metabolite concentra-tions between children treated with quinine (n � 79) andthose treated with artemether (n � 18). The ratios of thegeometric means (quinine:artemether) (95th percentile con-fidence limits for the ratio) for each metabolite were quin-olinic acid � 0.9 (0.6–1.5), neopterin � 0.8 (0.6–1.2), te-trahydrobiopterin � 1.0 (0.7–1.3), dihydrobiopterin � 0.8(0.6–1.1), 5-hydroxyindoleacetic acid � 1.0 (0.7–1.3), tryp-tophan � 0.7 (0.5–1.1), and nitrate plus nitrite � 1.0 (0.7–1.3). There were no significant differences in metabolite con-centration between the diagnostic groups. Geometric meanratios (strict cerebral malaria/malaria with impaired con-sciousness) (95th percentile confidence limits) for each me-tabolite were quinolinic acid � 1.1 (0.7–1.9), neopterin �0.9 (0.6–1.2), tetrahydrobiopterin � 0.8 (0.6–1.1), dihydro-biopterin � 1.1 (0.8–1.5), 5-hydroxyindoleacetic acid � 1.0(0.8–1.4), tryptophan � 1.2 (0.8–1.8), and nitrate plus nitrite� 1.2 (0.9–1.5). Therefore, for the rest of the analysis, chil-dren with strict cerebral malaria (n � 79) and malaria withimpaired consciousness (n � 18) were grouped together andreferred to as cerebral malaria for simplicity (Table 1).

In comparison with the reference range, there were highlysignificant and very large increases in the CSF concentra-tions of quinolinic acid (ratio of the geometric means) (ce-rebral malaria/reference range [95th percentile confidenceinterval for the ratio] � 14.1 [9.8–20.4], P � 0.001) andneopterin (geometric mean ratio � 10.9 [9.1–13.0], P �

0.001) in children with cerebral malaria. There was also ahighly significant but moderate increase in concentrations oftetrahydrobiopterin (geometric mean ratio � 3.5 [3.0–4.2],P � 0.001) and dihydrobiopterin (geometric mean ratio �3.3 [2.6–4.5], P � 0.001). There was a significant, but smallincrease in 5-hydroxyindoleacetic acid (geometric mean ra-tio � 1.2 [1.0–1.4], P � 0.02). There was no change intryptophan concentration (geometric mean ratio � 0.93[0.77–1.1], P � 0.48). In contrast, CSF nitrate plus nitriteconcentrations were highly significantly and moderately re-duced in cerebral malaria compared with both the UnitedKingdom reference range (geometric mean ratio � 0.45[0.35–0.59], P � 0.001) and the local reference range (geo-metric mean ratio � 0.55 [0.45–0.68], P � 0.001). Therewas no difference between nitrate plus nitrite concentrationsin the local range compared with the United Kingdom ref-erence range (geometric mean ratio � 0.82 [0.60–1.11], P� 0.19).

In children with cerebral malaria, concentrations of CSFquinolinic acid, dihydrobiopterin, and 5-hydroxyindoleaceticacid were significantly different between the 3 outcomegroups (Table 2). The CSF quinolinic acid was significantlyincreased in survivors with abnormal neurologic signs (n �11) and those who died (n � 12) compared with survivorswith no neurologic sequelae (n � 74). The CSF dihydro-biopterin and 5-hydroxyindoleacetic acid concentrationswere significantly increased in the children who died com-pared with survivors with no neurologic sequelae and thosewith a neurologic deficit. There was no significant differencein age across the outcome groups (F2,95 � 0.55, P � 0.58).

DISCUSSION

Our results show that Kenyan children with cerebral ma-laria have a characteristic neurochemical profile with in-creased CSF quinolinic acid and neopterin concentrationsand a reduced nitrate plus nitrite concentration. Central ner-vous system quinolinic acid and neopterin have previouslybeen shown to be increased in a variety of diseases andexperimental models where there is immune activation,34,40–

48 including cerebral malaria.49,50 Experimentally, this hasbeen shown to be caused by induction of indoleamine 2,3-dioxygenase and GTP cyclohydrolase I, the first enzymes inthe pathways leading to quinolinic acid and neopterin syn-thesis, respectively.34,51,52

There is some experimental support for our finding of in-creased quinolinic acid concentrations in children recoveringfrom cerebral malaria. Mice infected with P. berghei ANKA,who develop fatal malaria with cerebral involvement,showed a 3-fold increase in brain quinolinic acid concentra-tions caused by induction of indoleamine 2,3-dioxygenase.50

They also had normal brain tryptophan concentrations.50

Whether quinolinic acid could be chronically elevated inKenyan children who have a greater exposure to infectiousdisease than our reference population might also be ques-tioned. We think this is unlikely because lumbar CSF con-centrations of quinolinic acid in American children with sep-ticemia but without central nervous system infection are in-creased only 1.3–3.5 times those of non-septic controls.53 Incomparison, much higher values (10–20 times) were foundhere in patients with cerebral malaria. Importantly, we also

287CSF BIOCHEMISTRY IN CEREBRAL MALARIA

TABLE 2Cerebrospinal fluid metabolite concentrations in the different outcome groups

Geometric mean(95th centile confidence limits)

Survivors—no neurologic signs

Survivors—abnormal neurology Died

Analysis of variancestatistics

Quinolinic acid (nmol/L) 194(161–234)

359*(170–759)

429*(188–978)

F2,95 � 5.35P � 0.006

Nitrate plus nitrite (�mol/L) 5.4(4.7–6.1)

6.3(4.3–9.2)

8.1(5.1–12.9)

F2,93 � 2.64P � 0.076

Total neopterin (nmol/L) 180(154–251)

169(114–251)

243(141–419)

F2,95 � 1.12P � 0.33

Tetrahydrobiopterin (nmol/L) 130(116–144)

101(74.8–134)

114(62.1–210)

F2,93 � 1.67P � 0.19

Dihydrobiopterin (nmol/L) 11.0(9.8–12.3)

14.5(9.9–21.2)

28.2*(16.9–47.1)

F2,95 � 15.7P � 0.001

5-Hydroxyindoleacetic acid (nmol/L) 216(194–241)

242(173–338)

335*(220–511)

F2,95 � 3.94P � 0.023

Tryptophan (�mol/L) 2.4(2.0–2.8)

2.4(1.6–3.7)

3.5(1.6–7.8)

F2,93 � 1.18P � 0.31

* Significantly increased (analysis of variance with Duncan’s new multiple range test).

found a graded increment of quinolinic acid with increasingseverity of outcome that would make a physiological in-crease very unlikely.

There has been a single previous report of total neopterinand total biopterin concentrations in CSF from children withcerebral malaria.49 Here, the investigators demonstrated amodest increases in total neopterin and total biopterin (com-pared with our reference ranges, no reference range beinggiven in the paper). Total biopterin refers to the sum of te-trahydrobiopterin, dihydrobiopterin, and biopterin; we foundbiopterin to be variably detectable in our samples and of lowconcentration, and the equivalent of total biopterin from ourstudy would be the sum of tetrahydrobiopterin and dihydro-biopterin. It is not clear why our measurements of the pterinspecies should be almost an order of magnitude greater thanthose of Weiss and others. The clinical and epidemiologiccharacteristics of childhood cerebral malaria are similar inKenya and Zambia. However, there were some methodologicdifferences. We strictly collected the second and third mil-liliter of CSF; the second was collected into tubes containingDTE and DETAPAC to prevent auto-oxidation of tetrahy-drobiopterin,54 and all specimens were frozen immediatelyat the bedside. More importantly, the timing of the CSF sam-ples was different. Our sampling was delayed until it wasjudged clinically safe (usually an improvement in consciouslevel) whereas Weiss and others sampled on admission. It ispossible that the delay in sampling allowed more florid bio-chemical changes to develop.

Conditions that cause induction of indoleamine 2,3-diox-ygenase and GTP cyclohydrolase I have also been shownexperimentally to cause induction of nitric oxide synthase,32

leading to increased synthesis of nitric oxide. However, wefound decreased concentrations of nitrate plus nitrite in ce-rebral malaria when compared with both a United Kingdomand a local reference population. This is in contrast to amodest elevation of CSF nitrate and nitrite levels previouslyfound in cerebral malaria.49 Support for our finding of re-duced concentrations of nitrate plus nitrite comes from care-ful studies of plasma nitrate plus nitrite concentration andurinary nitrate plus nitrite excretion in Tanzanian childrenwith P. falciparum infection.55 In this study, reduced nitrate

plus nitrite levels in both plasma and urine were found inchildren with cerebral malaria compared with controls andchildren with subclinical or uncomplicated infection.

Nitrate and nitrite are stable breakdown products of nitricoxide and the peroxynitrite anion.56 The breakdown of nitricoxide and the peroxynitrite anion is the major source of ni-trate and nitrite in biological fluids of fasted humans.57,58 Inrodents, brain nitrate plus nitrite concentrations were foundto correlate accurately with brain nitric oxide synthase ac-tivity.59 Cisternal CSF concentrations were found to parallelthose of the brain and there was no effect of plasma con-centrations upon brain concentrations.59 In humans, evidencefor a rostro-caudal gradient of nitrate plus nitrite concentra-tions suggests that nitrate and nitrite measured in lumbarCSF are generated higher up the neuraxis.36 The evidencetherefore suggests that concentrations of nitrate plus nitritein lumbar CSF reflect nitric oxide synthase activity in thebrain that is independent of systemic nitric oxide metabo-lism.

There is also evidence of a species difference in the abilityof cells to produce parallel induction of indoleamine 2,3-dioxygenase, GTP cyclohydrolase I, and nitric oxide syn-thase. While parallel induction of nitric oxide synthase, GTPcyclohydrolase I, and indoleamine 2,3-dioxygenase can beinduced in rodent microglia by cytokines,19,60 this is not thecase for human microglia. In human cells, cytokine stimu-lation does induce indoleamine 2,3-dioxygenase,21 but notnitric oxide synthase to any great degree.61 However, thereis good evidence that nitric oxide synthase can be inducedin human astrocytes by cytokine stimulation,62 althoughthese cells cannot produce quinolinic acid.21 While speciesdifferences might explain why there is no parallel inductionof nitric oxide synthase with the induction of GTP cyclohy-drolase I and indoleamine 2,3-dioxygenase, it does not ex-plain the reduction in nitrate plus nitrite concentrations. Thereduction can only occur by reduced activity of constitutivenitric oxide synthase or increased scavenging of nitric oxide.We can only speculate about the mechanisms whereby ac-tivity of constitutive nitric oxide synthase is reduced. Theactivity of both constitutive isoforms of nitric oxide synthasein the brain is both calcium-dependent and requires tetrahy-

288 DOBBIE AND OTHERS

drobiopterin as a cofactor.28 We have shown here that tetra-hydrobiopterin concentrations are increased in cerebral ma-laria and cofactor deficiency does not seem likely as a cause;although the increased concentrations of dihydrobiopterinmight inhibit nitric oxide synthase.32 It is possible that freeintracellular calcium is reduced in the brain in cerebral ma-laria and this question needs to be addressed. Concentrationsof the substrate arginine may also be limiting to nitric oxidesynthase activity in the brain and this too needs to be ad-dressed.63 Alternatively, other inhibitors of nitric oxide syn-thase or scavengers of nitric oxide might be produced.

The combination of increased quinolinic acid and reducednitric oxide concentrations in the brain may cause an exci-totoxic mechanism for the production of symptoms in ce-rebral malaria. When quinolinic acid is synthesized in excessit readily enters the extracellular compartment.64 There isneither extracellular metabolism nor active transport of quin-olinic acid in cerebral tissue; therefore, toxic quantities ofquinolinic acid cannot be removed from the synaptic cleft.65

Nitric oxide has been shown to protect against NMDA-me-diated neurotoxicity by reducing conductance of calciumthrough the receptor pore.29–31,66,67 A reduction in nitric oxideproduction could therefore enhance quinolinic acid neuro-toxicity.

The mechanisms by which NMDA-toxicity is mediatedcan be separated into 2 components distinguishable on thebasis of differences in time course and ionic dependence.First, an influx of sodium causes reversible neuronal swell-ing while delayed neuronal disintegration is caused by aninflux of calcium.27 Excitotoxic mechanisms can thus explainthe neurologic symptoms of childhood cerebral malaria:coma, cerebral edema, and seizures that may be reversible,and permanent neuronal damage. That such a mechanismoccurs in cerebral malaria is supported by the graded in-crease in quinolinic acid found across the 3 outcome groups.It is possible that the increased concentration of quinolinicacid found in the children who died is caused by post-mor-tem artefact, but we think that this is unlikely because intra-cellular accumulation of quinolinic acid does not occur inthe central nervous system,64 and the cerebrospinal fluid wassampled within 15 min of death.

In conclusion, we have found biochemical evidence foran excitotoxic mechanism causing the cerebral symptoms incerebral malaria. This is of potential importance because se-lective NMDA-receptor blockers are becoming more widelyused in medicine and would provide a novel approach to thetreatment of cerebral malaria.68

Acknowledgment: This paper is published with the permission ofthe director of the Kenya Medical Research Institute (KEMRI).

Financial support: This study was supported by grants from theWellcome Trust, KEMRI, and Search.

Authors’ addresses: Michael Dobbie, Department of Neurochemis-try, Institute of Neurology, University College London MedicalSchool, Queen Square, London, WC1N 3BG, United Kingdom. JaneCrawley, Nuffield Department of Clinical Medicine, John RadcliffeHospital, Headley Way, Headington, Oxford OX3 9DU, UnitedKingdom. Catherine Waruiru, Paediatric Department, Leicester Roy-al Infirmary, Leicester LE1 5WW, United Kingdom. Kevin Marsh,KEMRI Centre for Geographical Medicine Research Coast, Box230, Kilifi, Kenya. Robert Surtees, Neurosciences Unit, Institute ofChild Health, University College London Medical School, The

Wolfson Centre, Mecklenburgh Square, London WC1N 2AP, UnitedKingdom.

REFERENCES

1. Greenwood BM, Bradley AK, Greenwood AM, Byass P, Jam-meh K, Marsh K, Tulloch S, Oldfield FS, Hayes R, 1987.Mortality and morbidity from malaria among children in arural area of The Gambia, West Africa. Trans R Soc TropMed Hyg 81: 478–486.

2. Newton CR, Krishna S, 1998. Severe falciparum malaria in chil-dren: current understanding of pathophysiology and support-ive treatment. Pharmacol Ther 79: 1–53.

3. Turner GD, Morrison H, Jones M, Davis TM, Looareesuwan S,Buley ID, Gatter KC, Newbold CI, Pukritayakamee S, Na-gachinta B, White NJ, Berendt AR, 1994. An immunohisto-chemical study of the pathology of fatal malaria. Evidence forwidespread endothelial activation and a potential role for in-tercellular adhesion molecule-1 in cerebral sequestration. AmJ Pathol 145: 1057–1069.

4. Esslinger CW, Picot S, Ambroise Thomas P, 1994. Intra-eryth-rocytic Plasmodium falciparum induces up-regulation of in-ter-cellular adhesion molecule-1 on human endothelial cellsin vitro. Scand J Immunol 39: 229–232.

5. Kwiatkowski D, Cannon JG, Manogue KR, Cerami A, DinarelloCA, Greenwood BM, 1989. Tumour necrosis factor produc-tion in falciparum malaria and its association with schizontrupture. Clin Exp Immunol 77: 361–366.

6. Kwiatkowski D, Hill AV, Sambou I, Twumasi P, Castracane J,Manogue KR, Cerami A, Brewster DR, Greenwood BM,1990. TNF concentration in fatal cerebral, non-fatal cerebral,and uncomplicated Plasmodium falciparum malaria. Lancet336: 1201–1204.

7. Mantovani A, Sica A, Colotta F, Dejana E, 1990. The role ofcytokines as communication signals between leukocytes andendothelial cells. Prog Clin Biol Res 349: 343–353.

8. Schofield L, Novakovic S, Gerold P, Schwarz RT, McConvilleMJ, Tachado SD, 1996. Glycosylphosphatidylinositol toxin ofPlasmodium up-regulates intercellular adhesion molecule-1,vascular cell adhesion molecule-1, and E-selectin expressionin vascular endothelial cells and increases leukocyte and par-asite cytoadherence via tyrosine kinase-dependent signaltransduction. J Immunol 156: 1886–1896.

9. Ghigo D, Todde R, Ginsburg H, Costamagna C, Gautret P, Bus-solino F, Ulliers D, Giribaldi G, Deharo E, Gabrielli G, Pes-carmona G, Bosia A, 1995. Erythrocyte stages of Plasmodiumfalciparum exhibit a high nitric oxide synthase (NOS) activityand release an NOS-inducing soluble factor. J Exp Med 182:677–688.

10. Prada J, Kremsner PG, 1995. Enhanced production of reactivenitrogen intermediates in human and murine malaria. Paras-itol Today 11: 409–410.

11. Kremsner PG, Winkler S, Wildling E, Prada J, Bienzle U, Gran-inger W, Nussler AK, 1996. High plasma levels of nitrogenoxides are associated with severe disease and correlate withrapid parasitological and clinical cure in Plasmodium falci-parum malaria. Trans R Soc Trop Med Hyg 90: 44–47.

12. Tachado SD, Gerold P, McConville MJ, Baldwin T, Quilici D,Schwarz RT, Schofield L, 1996. Glycosylphosphatidylinositoltoxin of Plasmodium induces nitric oxide synthase expressionin macrophages and vascular endothelial cells by a proteintyrosine kinase-dependent and protein kinase C-dependentsignaling pathway. J Immunol 156: 1897–1907.

13. Molyneux ME, Taylor TE, Wirima JJ, Borgstein A, 1989. Clin-ical features and prognostic indicators in paediatric cerebralmalaria: a study of 131 comatose Malawian children. Q J Med71: 441–459.

14. Crawley J, Smith S, Kirkham F, Muthinji P, Waruiru C, MarshK, 1996. Seizures and status epilepticus in childhood cerebralmalaria. Q J Med 89: 591–597.

15. Newton CR, Crawley J, Sowumni A, Waruiru C, Mwangi I,English M, Murphy S, Winstanley PA, Marsh K, Kirkham FJ,

289CSF BIOCHEMISTRY IN CEREBRAL MALARIA

1997. Intracranial hypertension in Africans with cerebral ma-laria. Arch Dis Child 76: 219–226.

16. Brewster DR, Kwiatkowski D, White NJ, 1990. Neurologicalsequelae of cerebral malaria in children. Lancet 336: 1039–1043.

17. Bondi FS, 1992. The incidence and outcome of neurologicalabnormalities in childhood cerebral malaria: a long-term fol-low-up of 62 survivors. Trans R Soc Trop Med Hyg 86: 17–19.

18. Werner ER, Werner Felmayer G, Fuchs D, Hausen A, Reibneg-ger G, Wachter H, 1989. Parallel induction of tetrahydrobiop-terin biosynthesis and indoleamine 2,3-dioxygenase activityin human cells and cell lines by interferon-gamma. BiochemJ 262: 861–866.

19. Sakai N, Kaufman S, Milstien S, 1995. Parallel induction ofnitric oxide and tetrahydrobiopterin synthesis by cytokines inrat glial cells. J Neurochem 65: 895–902.

20. Sakai N, Kaufman S, Milstein S, 1993. Tetrahydrobiopterin isrequired for cytokine-induced nitric oxide production in a mu-rine macrophage cell line (RAW 264). Mol Pharmacol 43: 6–10.

21. Heyes MP, Achim CL, Wiley CA, Major EO, Saito K, MarkeySP, 1996. Human microglia convert l-tryptophan into the neu-rotoxin quinolinic acid. Biochem J 320: 595–597.

22. Stone TW, 1993. Neuropharmacology of quinolinic and kynu-renic acids. Pharmacol Rev 45: 309–379.

23. Lapin IP, 1982. Convulsant action of intracerebroventricularlyadministered l-kynurenine sulphate, quinolinic acid and otherderivatives of succinic acid, and effects of amino acids: struc-ture-activity relationships. Neuropharmacology 21: 1227–1233.

24. Schwarcz R, Whetsell Jr WO, Mangano RM, 1983. Quinolinicacid: an endogenous metabolite that produces axon-sparinglesions in rat brain. Science 219: 316–318.

25. Kim JP, Choi DW, 1987. Quinolinate neurotoxicity in corticalcell culture. Neuroscience 23: 423–432.

26. Giulian D, Vaca K, Corpuz M, 1993. Brain glia release factorswith opposing actions upon neuronal survival. J Neurosci 13:29–37.

27. Choi DW, 1992. Excitotoxic cell death. J Neurobiol 23: 1261–1276.

28. Nathan C, Xie QW, 1994. Regulation of biosynthesis of nitricoxide. J Biol Chem 269: 13725–13728.

29. Manzoni O, Prezeau L, Marin P, Deshager S, Bockaert J, FagniL, 1992. Nitric oxide-induced blockade of NMDA receptors.Neuron 8: 653–662.

30. Lei SZ, Pan ZH, Aggarwal SK, Chen HS, Hartman J, SucherNJ, Lipton SA, 1992. Effect of nitric oxide production on theredox modulatory site of the NMDA receptor-channel com-plex. Neuron 8: 1087–1099.

31. Manzoni O, Bockaert J, 1993. Nitric oxide synthase activityendogenously modulates NMDA receptors. J Neurochem 61:368–370.

32. Mayer B, Werner ER, 1995. In search of a function for tetra-hydrobiopterin in the biosynthesis of nitric oxide. NaunynSchmiedebergs Arch Pharmacol 351: 453–463.

33. Werner ER, Werner Felmayer G, Fuchs D, Hausen A, Reibneg-ger R, Yim JJ, Wachter H, 1991. Biochemistry and functionof pteridine synthesis in human and murine macrophages.Pathobiology 59: 276–279.

34. Wachter H, Fuchs D, Hausen A, Reibnegger G, Werner ER,1989. Neopterin as marker for activation of cellular immunity:immunologic basis and clinical application. Adv Clin Chem27: 81–141.

35. Marsh K, Forster D, Waruiru C, Mwangi I, Winstanley M,Marsh V, Newton C, Winstanley P, Warn P, Peshu N, PasvolG, Snow R, 1995. Indicators of life-threatening malaria inAfrican children. N Engl J Med 332: 1399–1404.

36. Dobbie MS, Surtees RAH, 1997. Concentrations of quinolinicacid in cerebrospinal fluid measured by gas chromatographyand electron-impact ionisation mass spectrometry. Age-relat-ed changes in a paediatric reference population. J ChromatogrB 696: 53–58.

37. Surtees R, Hyland K, 1990. Cerebrospinal fluid concentrations

of S-adenosylmethionine, methionine, and 5-methyltetrahy-drofolate in a reference population: cerebrospinal fluid S-adenosylmethionine declines with age in humans. BiochemMed Metab Biol 44: 192–199.

38. Hyland K, Surtees RAH, Heales SJ, Bowron A, Howells DW,Smith I, 1993. Cerebrospinal fluid concentrations of pterinsand metabolites of serotonin and dopamine in a pediatric ref-erence population. Pediatr Res 34: 10–14.

39. Clelland JD, Brand MP, Bolanos JP, Surtees RAH, Land JM,Heales SJ, 1996. Age dependent changes in the cerebrospinalfluid concentration of nitrite and nitrate. Ann Clin Biochem33: 71–72.

40. Heyes MP, 1993. Metabolism and neuropathologic significanceof quinolinic acid and kynurenic acid. Biochem Soc Trans 21:83–89.

41. Heyes MP, Saito K, Milstien S, Schiff SJ, 1995. Quinolinic acidin tumors, hemorrhage and bacterial infections of the centralnervous system in children. J Neurol Sci 133: 112–118.

42. Achim CL, Heyes MP, Wiley CA, 1993. Quantitation of humanimmunodeficiency virus, immune activation factors, and quin-olinic acid in AIDS brains. J Clin Invest 91: 2769–2775.

43. Brouwers P, Heyes MP, Moss HA, Wolters PL, Poplack DG,Markey SP, Pizzo PA, 1993. Quinolinic acid in the cerebro-spinal fluid of children with symptomatic human immuno-deficiency virus type 1 disease: relationships to clinical statusand therapeutic response. J Infect Dis 168: 1380–1386.

44. Gisslen M, Chiodi F, Fuchs D, Norkrans G, Svennerholm B,Wachter H, Hagberg L, 1994. Markers of immune stimulationin the cerebrospinal fluid during HIV infection: a longitudinalstudy. Scand J Infect Dis 26: 523–533.

45. Chatelet P, Hautecoeur P, Dhondt JL, Forzy G, Gallois P, HayteJM, Warot P, Dereux JF, 1993. Pteridines non cojuguees etinfections neuro-meningees. Rev Neurol (Paris) 149: 331–335.

46. Ott M, Demisch L, Engelhardt W, Fischer PA, 1993. Interleukin-2, soluble interleukin-2-receptor, neopterin, L-tryptophan andbeta 2-microglobulin levels in CSF and serum of patients withrelapsing-remitting or chronic-progressive multiple sclerosis.J Neurol 241: 108–114.

47. Hagberg L, Dotevall L, Norkrans G, Larsson M, Wachter H,Fuchs D, 1993. Cerebrospinal fluid neopterin concentrationsin central nervous system infection. J Infect Dis 168: 1285–1288.

48. Millot F, Dhondt JL, Mazingue F, Mechinaud F, Auget JL, Bau-ters F, Harousseau JL, 1993. Cerebrospinal fluid neopterin lev-els in children with central nervous system leukemia. LeukLymphoma 9: 351–356.

49. Weiss G, Thuma PE, Biemba G, Mabeza G, Werner ER, Gor-deuk VR, 1998. Cerebrospinal fluid levels of biopterin, nitricoxide metabolites, and immune activation markers and theclinical course of human cerebral malaria. J Infect Dis 177:1064–1068.

50. Sanni LA, Thomas SR, Tattam BN, Moore DE, Chaudhri G,Stocker R, Hunt NH, 1998. Dramatic changes in oxidativetryptophan metabolism along the kynurenine pathway in ex-perimental cerebral and noncerebral malaria. Am J Pathol152: 611–619.

51. Naritsin DB, Saito K, Markey SP, Chen CY, Heyes MP, 1995.Metabolism of L-tryptophan to kynurenate and quinolinate inthe central nervous system: effects of 6-chlorotryptophan and4-chloro-3-hydroxyanthranilate. J Neurochem 65: 2217–2226.

52. Heyes MP, Saito K, Major EO, Milstien S, Markey SP, VickersJH, 1993. A mechanism of quinolinic acid formation by brainin inflammatory neurological disease. Attenuation of synthe-sis from L-tryptophan by 6-chlorotryptophan and 4-chloro-3-hydroxyanthranilate. Brain 116: 1425–1450.

53. Batshaw ML, Robinson MB, Hyland K, Djali S, Heyes MP,1993. Quinolinic acid in children with congenital hyperam-monaemia. Ann Neurol 34: 676–680.

54. Howells DW, Hyland K, 1987. Direct analysis of tetrahydro-biopterin in cerebrospinal fluid by high-performance liquidchromatography with redox electrochemistry: prevention ofautoxidation during storage and analysis. Clin Chim Acta 167:23–30.

290 DOBBIE AND OTHERS

55. Anstey NM, Weinberg JB, Hassanali MY, Mwaikambo ED,Manyenga D, Misukonis MA, Arnelle DR, Hollis D, Mc-Donald MI, Granger DL, 1996. Nitric oxide in Tanzanian chil-dren with malaria: inverse relationship between malaria se-verity and nitric oxide production/nitric oxide synthase type2 expression. J Exp Med 184:557–567.

56. Ignarro LJ, Fukuto JM, Griscavage JM, Rogers NE, Byrns RE,1993. Oxidation of nitric oxide in aqueous solution to nitritebut not nitrate: comparison with enzymatically formed nitricoxide from L-arginine. Proc Natl Acad Sci USA 90: 8103–8107.

57. Rhodes P, Leone AM, Francis PL, Struthers AD, Moncada S,Rhodes PM, 1995. The L-arginine:nitric oxide pathway is themajor source of plasma nitrite in fasted humans. Biochem Bio-phys Res Commun 209: 590–596.

58. Surtees R, Heales S, 1997. A rostrocaudal gradient of nitrateplus nitrite concentrations in CSF. J Neurol Neurosurg Psy-chiatry 62: 100–101.

59. Salter M, Duffy C, Garthwaite J, Strijbos PJ, 1996. Ex vivomeasurement of brain tissue nitrite and nitrate accurately re-flects nitric oxide synthase activity in vivo. J Neurochem 66:1683–1690.

60. Alberati Giani D, Ricciardi Castagnoli P, Kohler C, Cesura AM,1996. Regulation of the kynurenine metabolic pathway byinterferon-gamma in murine cloned macrophages and microg-lial cells. J Neurochem 66: 996–1004.

61. Peterson PK, Hu S, Anderson WR, Chao CC, 1994. Nitric oxide

production and neurotoxicity mediated by activated microgliafrom human versus mouse brain. J Infect Dis 170: 457–460.

62. Brosnan CF, Battistini L, Raine CS, Dickson DW, CasadevallA, Lee SC, 1994. Reactive nitrogen intermediates in humanneuropathology: an overview. Dev Neurosci 16: 152–161.

63. Westergaard N, Beart PM, Schousboe A, 1993. Transport of L-[3H]arginine in cultured neurons: characteristics and inhibi-tion by nitric oxide synthase inhibitors. J Neurochem 61:364–367.

64. Speciale C, Schwarcz R, 1993. On the production and disposi-tion of quinolinic acid in rat brain and liver slices. J Neuro-chem 60: 212–218.

65. Foster AC, Collins JF, Schwarcz R, 1983. On the excitotoxicproperties of quinolinic acid, 2,3-piperidine dicarboxylic acidsand structurally related compounds. Neuropharmacology 22:1331–1342.

66. Jones NM, Loiacono RE, Moller M, Beart PM, 1994. Diverseroles for nitric oxide in synaptic signalling after activation ofNMDA release-regulating receptors. Neuropharmacology 33:1351–1356.

67. Connop BP, Boegman RJ, Jhamandas K, Beninger RJ, 1995.Excitotoxic action of NMDA agonists on nigrostriatal dopa-minergic neurons: modulation by inhibition of nitric oxidesynthesis. Brain Res 676: 124–132.

68. Lipton SA, 1993. Prospects for clinically tolerated NMDA an-tagonists: open-channel blockers and alternative redox statesof nitric oxide. Trends Neurosci 16: 527–532.