300

CLINICAL ARTICLES

Magnetic Resonance Imaging of the Brain in Patients with Cerebral Malaria

s. Looareesuwan, P. Wilairatana, S. Krishna, B. Kendall, From the Faculty of Tropical Medicine,
S. Vannaphan, C. Viravan, and N. J. White Mahidol University, Bangkok, Thailand; and the
Centre for Tropical Medicine, Nuffield Department of Clinical Medicine,
John Radcliffe Hospital, Headington, Oxford; and the
National Hospital for Nervous Diseases,
Queen Square, London, United Kingdom

In a prospective study of cerebral malaria, 24 adults with this disease underwent magnetic

Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021

resonance imaging (MRI) of the brain. Four patients died. Two of these patients (nos. 17 and 24)
had breathing abnormalities requiring ventilatory support followed by clinical signs of brain death.
Four days later MRI of patient 17 showed gross swelling of the brain, and 5 hours later MRI of
patient 24 showed foramen magnum herniation. Twenty-two patients had no evidence of cerebral
edema, but MRI revealed that brain volume during acute cerebral malaria was slightly greater than
that during the convalescent phase of the disease. This difference was attributed to an increase in
the volume of intracerebral blood. The cerebral volume was lower during early convalescence than
several months later. The volume of the brain in patients with cerebral malaria is increased. This
increased volume probably results from sequestration of parasitized erythrocytes and compensatory
vasodilatation rather than from edema. Brain stem herniation may occur, but its temporal relation
to brain death in cases of cerebral malaria remains uncertain.

Unrouseable coma in patients with severe or cerebral malaria anatomy of the brain stem because, unlike CT, signals from
is caused by diffuse symmetrical encephalopathy associated bones do not interfere with imaging. MRI is particularly useful
with the sequestration of erythrocytes containing mature stages for assessing the complications of tentorial and foramen mag-
of Plasmodium falciparum in the cerebral microvasculature [1]. num herniation. Therefore, we conducted a prospective study of
The mortality rate associated with this disease is 15%-20%. 24 adult patients in Thailand who had strictly defined cerebral
In many respects cerebral malaria is similar to a metabolic malaria; MRI was used to examine cerebral anatomy.
encephalopathy of short duration, with the majority of survivors
making full neurological recoveries [2]. However, ,..., 10% of
Methods
children and a smaller proportion of adults with cerebral ma-
laria [3] will have residual neurological sequelae following Patients
recovery of consciousness. Although microvascular obstruction
All patients were admitted to the intensive care unit of the
compromises delivery of nutrients and oxygen to the brain [4],
Hospital for Tropical Diseases, Faculty of Tropical Medicine,
the exact causes of coma and death in patients with cerebral
Mahidol University, Bangkok, Thailand. Adult patients with
malaria are not known. Acute respiratory arrest is a common
cerebral malaria (defined as parasitemia with asexual P. falci-
terminal event, particularly in children, and it has been sug-
parum and unrouseable coma, i.e., a Glascow coma score
gested that acute respiratory arrest results from raised intracran-
[GCS] of ~8 and the absence of other causes of coma) were
ial pressure (ICP) precipitating fatal brain stem herniation [5].
enrolled in the study if their relatives or guardians had given
MRI is currently the preferred technique for visualizing the
informed consent.

Procedures
Received 19 September 1994; revised 14 December 1994.
Informed consent was obtained from the relatives or guardians of the pa-
At the time of admission to the intensive care unit, a full
tients. clinical and neurological assessment was made, and blood was
Financial support: This study was supported by the Wellcome Trust of Great taken for routine hematologic and biochemical analyses, cul-
Britain as part of the Wellcome-Mahido1 University, Oxford Tropical Medicine
Research Programme, and by the National Science and Technology Develop-
ture, and determination of quantitative parasite count, plasma
ment Agency, Bangkok, Thailand. lactate concentration, glucose concentration, and arterial blood
Reprints or correspondence: Dr. N. J. White, Faculty of Tropical Medicine, gas and pH levels. The diagnosis of malaria was confirmed by
Mahidol University, 420/6 Rajvithi Road, Bangkok 10400, Thailand.
examination of a peripheral blood film stained with Field's
Clinical Infectious Diseases 1995;21:300-9
© 1995 by The University of Chicago. All rights reserved.
stain. Other infections of the CNS were excluded by examina-
1058--4838/95/2102-0021 $02.00 tion of CSF. Lumbar punctures were performed with the patient
ern 1995;21 (August) MRI of the Brain in Cerebral Malaria 301

in the left lateral position with use of a 22G/20G needle. Cere- Table 1. Clinical and laboratory findings for 24 patients with cere-
brospinal pressures were measured with a glass manometer bral malaria at the time of admission to the hospital.
(internal volume, 0.5 mL; length, 30 em). CSF samples were
Finding Value*
collected for microscopy, culture, and measurement of protein,
glucose, and lactate concentrations. Male:female ratio 20:4
All patients received antimalarial treatment with intravenous Age (y) 27 ± 9
quinine dihydrochloride (Government Pharmaceutical Organi- Weight (kg) 53 ± 7
Oral temperature (oq 38.4 ± 1.1
sation, Bangkok, Thailand; loading dose of 20 mg/kg followed
Pulse (beats/min) 109 ± 14
by 10 mg/kg every 8 hours), intravenous artesunate (Guilin Systolic blood pressure" (mm Hg) 111 ± 20
No.2 factory, Guilin, People's Republic of China; 120 mg Diastolic blood pressure" (mm Hg) 66 ± 14
initially followed by 60 mg every 12 hours [total dose, 600 Respirations' (breaths/min) 25 ± 4
mg]), or intramuscular artemether (Kunming Pharmaceutical, Hematocrit (%) 33 ± 6
WBe count (XI0 91L) 12 ± 5
Kunming, People's Republic of China; initial dose of 3.2 mg/

Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021

Median platelet count, X 109/L (range) 54 (11-178)
kg followed by 1.6 mg/kg daily) as part of a separate series of Median serum creatinine level, mg/dL
studies on drug treatment for severe malaria. Intramuscular (range) 1.8 (0.7-8.4)
phenobarbitone (May & Baker, Dagenham, Essex, United Serum albumin level; (g/dL) 3.3 ± 0.5
Serum globulin level! (g/dL) 2.8 ± 0.6
Kingdom; 200 mg) was given to 11 patients at the time of
Median serum aspartate aminotransferase
admission. Management in the intensive care unit included level.t UIL (range) 96 (27-450)
careful volume replacement with hemodynamic monitoring and Median serum bilirubin level, g/dL
hemodialysis when necessary. No patients received any drugs (range) 6.4 (1.2-31)
(e.g., steroids or mannitol) used in the treatment of cerebral Serum alkaline phosphatase level, II UIL
(range) 35 (25-72)
edema complicating other conditions. Convulsions were treated
Plasma glucose level (mg/dL) 164 ± 98
with intravenous diazepam (Government Pharmaceutical Or- Geometric mean parasitemia, /f.lL (range) 118,900 (260-1,394,200)
ganisation).
* Values are means ± SD unless stated otherwise.
t Finding for 23 patients.
t Finding for 22 patients.
§ Normal range, 0-40 UIL.
MRI of the Brain II Normal range, 9-35 UIL.

During the acute phase of cerebral malaria, MRI of the brain

was performed only when the patient's clinical condition was
stable. MRI was then repeated at the time of discharge from
the hospital and again 6-18 months later. All patients under-
went MRI on the same machine: a 0.2T Hitachi MRP20 super-
conducting system (Phyathai X-Ray Computer Center, Bang-
kok, Thailand). Image slices of 7.5 mm in the sagittal plane
and 10 mm in the axial and coronal planes were used. T1-
weighted images for the axial views were obtained with 550/
25 (TRITE), and T2-weighted images for the coronal and axial
planes were generated with 2,000/38 and 2,000/110, respec-
tively, with dual echo analysis. In each case the Evans ratio
(the ratio of the width of the frontal hom to the short-axis
diameter of the internal skull) was calculated to assess brain
swelling. For quantitating shifts in the brain stem, the perpen-
dicular distance between Twining's line (a line extending from
the internal occipital protuberance to the anterior superior mar-
gin of the sella turcica) and the pontomesencephalic junction
(T-PMJ) or the top of the sylvian aqueduct along the dorsal
margin of the brain stem (T-A) was evaluated on hard copies
of midsaggital scans that were obtained as described previously
[6]. A correction factor for image reduction was applied for
ensuring that measurements were comparable between scans.
The scans were assessed independently by two neuroradiolo-
gists who did not have knowledge of the clinical details of the
cases before quantitative evaluation.
Statistical Analysis of Data
Data were analyzed with use of SYSTAT (version 5.2, Sys-
tat, Evanston, IL). Normally distributed data were analyzed
with use of the Student's t-test. Nonnormally distributed data
either were transformed logarithmically for normalizing distri-
butions or were analyzed by the Wilcoxon rank-sum test and
the Kruskal-Wallis one-way analysis of variance. The correla-
tion coefficient was used for evaluating the relation between
normally distributed variables, and Spearman's rank correlation
was used for evaluating the remaining variables.
Results
Clinical Features During Admission
Twenty-four adult patients with cerebral malaria were stud-
ied; four (17%) of these patients died. The demographic and
clinical features of the patients are listed in table 1. At the time
of the first MRI, all of the patients except one (patient 10)
were in an unrouseable coma and had GCSs of ~8. Patient 10
had a GCS of 8 during admission, but at the time of the first
MRI, the GCS was 9. No patients had neurological signs indi-
cating a focal deficit in either cerebral hemisphere at the time
of MRI, although several patients had focal seizures at other
times. Most patients usually had evidence of involvement of
302 Looareesuwan et al. cm 1995;21 (August)

Table 2. Clinical and CSF findings for 24 patients with cerebral malaria at the time of admission to the hospital.

CSF
Opening Closing CSF CSF Plasma CSF Plasma total
Outcome, Duration cerebrospinal cerebrospinal protein glucose glucose lactate lactate WBC
patient Retinal of coma pressure pressure level level level level level count
no. Convulsions GCS hemorrhage (h) (mm) (mm) (mg/dL) (mg/dL) (mg/dL) (mmollL) (mmol/L) (zmm')

Survival
1 Yes 5 Yes 24 190 110 138 87 106 8.8 9
2 No 8 No 20 170 120 10 76 111 3.5 2
3 No 8 No 24 193 140 118 118 149 5.0 6
4 No 7 No 24
6 No 6 No 36 165 135 65 103 243 2.4 3.2 8
7 No 8 Yes 12 100 60 60 90 164 4.9 6.1 20

Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021

8 No 8 No 36 180 140 18 74 167 2 2.3 3
9 Yes 8 No 40 195 100 32 92 130 4.9 3.3 6
10 9 No 24 110 70 58 69 166 2.5 2 12
11 8 No 24 210 112 76 90 150 3.6 3.7 6
12 Yes 5 Yes 12 180 140 52 117 237 5
13 No 7 No 24 195 140 97 26 156 4.4 4.3 4
14 Yes 8 No 10 170 95 33 6.3 6.4 2
15 Yes 7 No 10 110 80 82 90 140 2.8 4 2
16 No 8 No 72 55 0 82 92 188 3.7 17.7 10
18 No 8 No 10 110 70 96 18 223 2.9 5.1 2
19 No 6 No 6 80 65 80 56 150 3.4 4.1 5
20 Yes 8 No 10 130 80 30 66 207 4.2 3.8 100
21 No 8 No 62 150 48 68 117 10
22 No 5 No 12 190 36 102 143 72
Death
5 No 7 Yes 60 140 85 117 71 112 76
17 Yes 3 No 120
23 Yes 8 Yes 24 120 70 172 63 122 4
24 No 3 Yes 5 300 120 44 84 346 0

NOTE. GCS = Glascow coma score; ... = data not available.

multiple vital organs. In particular, liver dysfunction and renal
dysfunction were common in addition to cerebral malaria. At
the time of admission or soon after, the majority of patients
were jaundiced (18 [75%] of24) and anemic (21 [88%] of 24).
Nine patients (38%) presented with hepatomegaly, and three
patients (13%) presented with splenomegaly. None of the pa-
tients were hypotensive, in shock (table 1), or septicemic during
admission.
Neurological Findings
The mean (± SD) length of coma before admission was 29
± 26 hours, and the mean (± SD) GCS on admission was 7
± 1.3. The GCS had not changed significantly when it was
reevaluated at the time MRI was performed during the acute
phase of the disease (P > .5). Patients who died had a signifi-
cantly lower mean GCS (6) during admission than did survivors
(7.3) (P < .05). Eight patients (33%) had a history of seizures,
and five patients (21 %) had one or more episodes of generalized
convulsions after admission. These convulsions were con-
trolled adequately with appropriate anticonvulsant therapy. Six
patients (25%) presented with or had retinal hemorrhages dur-
ing admission (table 2). The mean (± SD) opening cerebrospi-
nal pressure was 157 ::!::: 53 mm, which correlated with the
closing cerebrospinal pressure (r = .8) (table 2). All CSF sam-
ples were negative for pathogens. The results of CSF examina-
tion are shown in table 2.
Clinical Course
The median time to recovery of consciousness for survivors
was 4 days (range, 2-8 days), the mean (± SD) duration of
fever for 15 patients in whom complicated secondary bacterial
infections did not develop was 114 ± 76 hours (range, 21-
338 hours), and the mean (± SD) time of parasite clearance
for 23 patients was 70 ± 29 hours (range, 24-152 hours). Six
patients (25%) required hemodialysis because of renal failure,
and azotemia developed during the course of infection in eight
patients (33%). Sixteen patients (67%) required blood transfu-
sions; four of these patients received more than five units of
blood because of disseminated intravascular coagulation or gas-
trointestinal bleeding. Twelve patients (50%) had one or more
episodes of hypoglycemia after admission.
Fatal Cases
Patient 5. A 19-year-old male patient with cerebral malaria
was treated with intravenous artesunate (120 mg followed by
ern 1995;21 (August) MRI of the Brain in Cerebral Malaria
60 mg every 12 hours; total dose, 600 mg). His illness was
complicated by severe hepatic dysfunction (peak total serum
bilirubin level, 34 mg/dL; peak serum aspartate aminotransfer-
ase level, 4,520 U/L) and renal failure (peak serum creatinine
level, 7.7 mg/dL) necessitating hemodialysis. Parasites cleared
from his blood in 117 hours. Upper gastrointestinal bleeding,
which had developed shortly after admission, worsened with
hemodialysis. Therefore, hemodialysis was changed to perito-
neal dialysis. However, the bleeding could not be controlled,
in spite of 12 units of fresh blood being transfused, and he
died of disseminated intravascular coagulation and hemor-
rhagic shock. His neurological condition did not deteriorate
following the initial MRl that was performed 5 days before
his death.
Patient 17. A 22-year-old man with acute falciparum ma-
laria was admitted to the hospital with a 3-day history of fever
followed by sudden loss of consciousness (GCS of 5 on admis-
sion). His pulse was 118 beats/min, blood pressure was 141/
88 mm Hg, respirations were 26 breaths/min, temperature was
37.8°C, and central venous pressure was 5 em. Auscultation
revealed that his chest was clear. He was hypoglycemic (blood
glucose level, 35 mg/dL). He was given 25 g of 50% dextrose
intravenously and was treated immediately with intramuscular
artemether (160 mg). Lumbar puncture was attempted, but the
fluid was bloody. His vital signs remained stable, except for
his core temperature (which rose to 40°C at 6 hours). Eight
hours after admission his respiratory rate increased, and he
started decorticate posturing. At this time arterial blood gas
determination when the patient was receiving 35% inspired
oxygen revealed the following: pH, 7.565; PC02, 33.2 mm Hg;
P02> 157 mm Hg; and base excess, 7.3.
Ten hours after admission more-obvious respiratory diffi-
culties with labored breathing and cyanosis developed, and he
was noted to have unreactive pupils. The blood pressure and
pulse rate had not changed significantly before his condition
changed. He was intubated and ventilated electively. Twenty
hours after admission papilledema developed, and 1 hour later
his blood pressure became unstable and decreased. Subse-
quently, his neurological state worsened, as evidenced by fixed
dilated pupils, no responses to deep pain stimulus, negative
caloric tests, and no electrical activity revealed by electroen-
cephalography. MRl was performed 4 days after admission. It
showed a grossly swollen brain with occlusion of the cerebral
circulation. The sulci, cisterns, and third ventricle were absent,
and there was considerable descent of the cerebellar tonsils
and foramen magnum herniation. He died shortly afterward.
Patient 23. A 57-year-old unconscious (GCS of 7) man
was admitted to the hospital with fever and jaundice (total
serum bilirubin level, 4.1 mg/dL; aspartate aminotransferase
level, 58 U/L). MRl was performed the following day. He
subsequently recovered consciousness sufficiently to be able
to talk to his relatives. However, his jaundice worsened pro-
gressively (peak total serum bilirubin level, 33 mg/dL; peak
aspartate aminotransferase level, 1,290 U/L), and he became
anuric (peak serum creatinine level, 5.2 mg/dL). Hemodialysis
303
was started, but gastrointestinal bleeding developed. Endos-
copy confirmed gastric erosions, but the bleeding could not be
controlled despite transfusions with fresh blood. He died of
intractable hemorrhage 9 days after admission.
Patient 24. A 23-year-old conscious man was admitted to
the hospital with heavy parasitemia (1,390,000 parasites/rd.
of blood) and a temperature of 38.6°C. He was treated with
intramuscular artemether (160 mg). Three hours later he be-
came very restless and agitated and did not respond to verbal
commands. One hour later he had generalized seizures that
were treated with an intravenous injection of diazepam (10
mg) followed by coma (GCS of 6) with tonic posturing and
divergent upward gaze. He was then transferred to the intensive
Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021
care unit. The parasite count had not fallen, and he was treated
with an intravenous injection of artesunate (120 mg).
Six hours after admission he was noted to be hyperventilat-
ing. One hour later he was intubated and ventilated. Focal
clonic seizures of the right leg developed. Pupillary and brain
stem reflexes at that time were normal. There was no papil-
ledema. At the time of lumbar puncture, the opening and clos-
ing cerebrospinal pressures were 300 mm and 120 mm, re-
spectively. Arterial blood gas analysis while the patient was
receiving 40% inspired oxygen (tidal volume, 500 mL; respira-
tory rate, 20) revealed uncompensated metabolic acidosis (pH,
7.268; PC02, 36 mm Hg; Po 2, 229 mm Hg; and base ex-
cess, -10).
Ten hours after admission to the hospital, respiratory distress
with profuse sweating and extensor posturing developed. The
endotracheal tube was repositioned, and he was supported by
temporary manual ventilation. Ventricular ectopia was noted
on the electrocardiographic monitor. The respiratory rate was
15, pulse rate was 95, and blood pressure was 151/95 mm
Hg. Eleven hours after admission he had further generalized
seizures that were treated with intravenous diazepam. One hour
later he had fixed dilated pupils. Direct ophthalmoscopy re-
vealed "cattle trucking" in the retinal vessels. A neurological
examination confirmed brain stem death.
MRl was performed 17 hours after admission (5 hours after
the signs of brain stem death developed). The findings were
similar to those for patient 17: gross brain swelling; absent
sulci, cisterns, and ventricles; and tonsillar descent (figure 1).
He died 2 hours later. An autopsy was performed 5 hours after
death; it confirmed the MRl findings of tonsillar herniation
and resultant pressure on the medulla oblongata. The cerebral
microvasculature was packed with erythrocytes containing ma-
ture forms of P. falciparum.
MRI Findings
In the majority of cases (20 of 24), MRl performed during
the acute phase ofillness revealed normal-appearing brains when
they were viewed in isolation (table 3). However, detailed com-
parisons of scans suggested swelling of the brain during the
acute phase of cerebral malaria as opposed to early convales-
cence (figure 2). The Evans ratio was slightly decreased during
304 Looareesuwan et al. eID 1995;21 (August)

Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021

Figure 1. MRIs of the brain of a patient with cerebral malaria who
died (patient 24). A, transverse T2-weighted image was obtained with
2,000/110 (TRITE) 5 hours after the development of signs of brain
stem death; this scan shows the absence of sulci and small ventricles.
E, T2-weighted image showing no fluid in the basal cisterns, except
for a small quantity around the chiasmata. C, sagittal scan generated
with 550/25 that reveals tonsillar herniation and downward displace-
ment of the fourth ventricle. A line has been drawn between the poste-
rior clinoid process (A) and the internal occipital protuberance (B).

the acute phase of illness, thus reflecting slight compression of
the frontal horns (mean Evans ratio ± SD: .244 ± .022 [acute
phase] vs..250 ± .027 [early convalescence]; n = 16; P =
.037). The Evans ratios for patients who died and patients who
survived were not significantly different. Except for patients 17
and 24 who had gross brain swelling, no patients had features
of cerebral edema. In particular, there was no increase in signal
from cerebral matter on T2-weighted scans of the other 22 pa-
tients. The increase in brain size during acute cerebral malaria
was attributed to an increase in the volume of intracerebral
blood. The brain stems were not swollen, and there was no
evidence of acute hydrocephalus.
MRI performed during early convalescence showed mild
global shrinkage of brain substance. During follow-up of
four patients in late convalescence, MRI revealed that all
areas showing changes had reverted to normal. One patient
 o
a
Table 3. Findings of MRI of the brain for 24 patients with cerebral malaria. :a
1.0
~~
Finding during acute phase of disease Finding during early convalescence of disease Finding during late convalescence of disease
~

Outcome, T-A T-PMJ T-A T-PMJ T-A T-PMJ

patient distance distance Evans Day of distance distance Evans Day of distance distance i
no. (mm) (mm) ratio * Comment MRI (mm) (mm) ratio* Follow-up comment MRI (mm) (mm) Comment ~

Survival
1 .241 Normal 40 .246 Shrinkage
2 17.2 3.9 .262 Normal 9 .277 Shrinkage 344 6.05 1.58 Less shrinkage
3 19.2 2.9 .242 Normal 11 .258 Shrinkage
4 .217 Normal 22 .228 Shrinkage
6 17.0 3.2 .228 Normal 14 18.2 5.7 .237 Shrinkage
7 21.9 7.8 .280 Normal, slight swelling 25 20 9.2 .297 Shrinkage
8 19.5 7.6 .283 Normal 15 20.3 8.5 .3 Shrinkage
9 20.3 6.8 .208 Normal, small ventricles, tonsillar 29 17.9 5.4 .237 Shrinkage
descent
10 20 4.3 .208 Normal, slight swelling 10 22.9 7.6 .209 Slight shrinkage
11 19.2 4.8 .254 Swollen, small ventricles with tonsillar 13 19.5 8.5 .25 Normal 102 4.71 0.59 Normal
herniation
12 21.3 7.4 .25 Normal 13 21.7 7.5 .254 Shrinkage 53 6.39 2.22 Less shrinkage 3:::
13 18.6 5.1 .25 Normal 12 20 5 .242 Slight shrinkage and ~
enlargement of Virchow- 0
....,
Robin spaces So
('p
14 20 7 .246 Normal 62 20.3 7.5 .241 Shrinkage 135 25 ILl Normal
tl:J
15 19.3 6.1 .216 Normal brain: thin anterior convexity 33 20.3 4.7 .206 Normal brain: subdural
subdural hematoma hematoma cleared ~.
::s
16 18.5 8.3 .271 Normal but old infarct in right genu of 26 20.3 10.2 .271 Infarct in right genu of internal 54 19.5 10.2 Unchanged S'
internal capsule capsule, otherwise normal infarct in o
right genu @
of internal cr'
capsule e.
18 20.3 6.8 .238 Normal 28 21.7 6.6 .238 Normal 3:::
~
19 17 3.4 .273 Normal
S"
20 18.7 4.5 .233 Normal: cavitas of septum pellucidum ::I.
~
21 18.6 5.1 .242 Normal, small ventricles with tonsillar
herniation
22 20.9 5.2 .246 Normal
Death
5 22.7 7.8 .237 Normal
17 .213 Sulci, cisterns, and third ventricle
obliterated, other ventricles small,
tonsils descended, cerebral
circulation not visible; T2-weighted
scans showed symmetrical high
signals in parietal white matter
23 23.6 7.6 .25 Normal, large cavitas of septum
pellucidum
24 CSF spaces obliterated except around
chiasmata; ventricles small, brain
swollen (except basal ganglia);
tonsils and fourth ventricle
descended

NOTE. T-A = Twining's line (a line extending from the internal occipital protuberance to the anterior superior margin of the sella turcica) and the top of the sylvian aqueduct along the dorsal margin of the brain stem; T-PMJ =
Twining's line and the pontomesencephalic junction.
* The ratio of the width of the frontal hom to the short-axis diameter of the internal skull.
w
oVI

Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021

306 Looareesuwan et al. em 1995;21 (August)

Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021

Figure 2. MRIs of the brain of a patient with cerebral malaria who
survived (patient 14). The axial images were obtained with 550/25
(TRITE). A, during the acute phase of illness, the ventricles are rela-
tively small, but the brain is otherwise normal. B, in early convales-
cence (62 days after admission), the brain has shrunk. C, in late conva-
lescence (135 days after admission), the brain is normal.

with a history of convulsions (no. 15) had a small anterior
convexity subdural hematoma that had resolved at the time
of follow-up. Another patient (no. 16) had evidence of a
small infarct in the right genu of internal capsule, but the
appearance of this infarct did not change at follow-up, which
suggests that it was old.
For quantitative comparison of scans obtained during the
acute phase of illness and follow-up, the degree of downward
displacement of brain stem structures relative to fixed internal
bony landmarks was measured [6]. Such displacement is asso-
ciated with a reduction in the T-PMJ and T-A distances. The
T-A and T-PMJ distances measured on scans obtained during
em 1995;21 (August) MRI of the Brain in Cerebral Malaria
the acute phase of illness were correlated with mean (:::!: SD)
values of 19.7 :::!: 1.8 mm and 5.8 :::!: 1.7 mm, respectively (r
= .67; n = 20; P = .001). In scans obtained during early
convalescence, mean (:::!:SD) T-A and T-PMJ distances had
increased slightly but not significantly to 20.3 :::!: 1.4 mm (P
= .19) and 7.1 :::!: 1.7 mm (P = .08), respectively. Thus, during
convalescence the mean (:::!: SD) increases in the T-A and T-
PMJ distances (0.59 :::!: 1.48 mm [n = 12] and 0.92 :::!: 1.68
mm [n = 12], respectively) were small and not statistically
significant. There were no significant differences between these
distances measured on the fewer scans obtained during delayed
convalescence (table 3). Changes in these measurements did
not correlate with changes in the Evans ratio. The parasite
count in peripheral blood at the time of admission correlated
negatively with the T-PMJ distance (r = .48; P = .032; n =
20) but not with the T-A distance. There was no relation be-
tween the opening cerebrospinal pressure or other variables
presented in table 2 and T-A and T-PMJ distances.
Discussion
The cause of coma and cerebral death in patients with falci-
parum malaria is not known. Soon after Laveran' s discovery
of the malarial parasite over one century ago, the peculiar
discrepancy between the number of parasites seen circulating
inside erythrocytes on peripheral blood smears and the number
of parasites found in the tissues after death was noted by Mar-
chiafava and Bignami [7]. These researchers observed that par-
asites sequestered in the microcirculation were the mature
stages not seen in the peripheral blood. They further surmised
that microcirculatory obstruction resulted from the sequestra-
tion of these erythrocytes containing mature forms of the para-
site P. falciparum and that this obstruction was the cause of
coma and death in patients with cerebral malaria.
Since these early observations several hypotheses explaining
the pathophysiology of cerebral malaria have been proposed.
The "mechanical obstruction" theory prevailed initially; it was
supported by detailed pathological studies conducted during
World War I [8-10]. However, microvascular obstruction was
considered an insufficient explanation of the autopsy finding
of a swollen brain in many patients who died of cerebral ma-
laria. In the 1940s it was suggested that brain swelling may
contribute directly to death [11, 12]. This theory was later
promoted by Migasena and Maegraith [13] and Maegraith and
Fletcher [14], who showed that the capillary permeability in
the cerebrums of simians with Plasmodium knowlesi malaria
who died was increased and suggested that similar processes
might occur in humans. These observations led to the use of
therapeutic agents directed against reducing cerebral edema.
Dexamethasone was the most widely used agent until it was
shown unequivocally to confer no benefit and to cause adverse
effects [3, 15].
In the 1980s several studies on adults with cerebral malaria
from Thailand provided evidence against "increased cerebral
307
capillary permeability." In addition to the failure of dexameth-
asone [3, 15], results of CT of the brain were usually normal
[16], opening cerebrospinal pressures at the time of lumbar
puncture were normal in 80% of cases [17], papilledema was
noted rarely, and a series of studies examining the blood-brain
barrier concluded that there was no significant increase in per-
meability [17]. The common finding of cerebral edema during
autopsy [8-12] was considered to represent an agonal phenom-
enon. Although these findings argued against cerebral edema as
a primary pathological process, the "mechanical obstruction"
theory alone (i.e., cerebral ischemia resulting from microvascu-
lar obstruction) was also insufficient as an explanation for the
complete neurological recovery in the majority of survivors,
Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021
despite the fact that comas lasted for several days in many
cases [2, 3]. To account for these various findings, researchers
have suggested that cerebral malaria is metabolic encephalopa-
thy and results from alterations in blood flow and in local
disturbances in function (possibly mediated by local release of
cytokines or other transmitters) caused by parasitized erythro-
cytes sequestered in the cerebral microvasculature [2]. How-
ever, this theory does not explain why patients die.
The possibility that raised ICP might playa lethal role in the
pathophysiology of cerebral malaria was reanalyzed because of
observations in African children with cerebral malaria. In con-
trast to opening cerebrospinal pressures in adults that are usu-
ally (80%) in the normal range [17], the opening cerebrospinal
pressures in African children with cerebral malaria are usually
elevated [5, 18-20]. In fact, the mean cerebrospinal pressures
(""' 160 mm) are remarkably similar in adults and children with
cerebral malaria; however, the normal range of cerebrospinal
pressures in children is much lower [21], and, consequently,
80% of children have opening pressures above the normal range
[20]. Furthermore, in children and, less commonly, in adults,
respiratory arrest usually secondary to hyperventilation is an
important cause of death for which there is no clear explanation.
In one series from Kenya [5], the neurological signs com-
monly present in patients with cerebral malaria were considered
to reflect brain stem herniation, and it was suggested that the
raised ICP could be fatal. Although there was no significant
difference between opening cerebrospinal pressures measured
during lumbar puncture in children who died of cerebral ma-
laria and those who survived [18-20], the raised ICP hypothe-
sis was not necessarily disproved because these cerebrospinal
pressures were shown not to reflect peak ICPs measured di-
rectly [22]. Some children had normal or slightly elevated
opening cerebrospinal pressures measured during lumbar punc-
ture, but sudden marked rises in pressure were recorded during
continuous direct ICP monitoring. The increases in pressure
were associated with reduced cerebral perfusion.
These observations have obvious and potentially important
therapeutic implications; if raised ICP is not distributed uni-
formly throughout the neuraxis, then pressure differentials
could lead to fatal brain herniation via the tentorium or foramen
magnum. If raised ICPs kill children with cerebral malaria,
then active measures to reduce raised ICPs should be instituted.
308 Looareesuwan et al.
However, proof of causality is lacking, and, from a clinical
standpoint, there are no clinical indicators that have identified
children with markedly elevated ICPs.
The role of raised ICPs in causing death in patients with
cerebral malaria has not been resolved. Although there was
good evidence that children with cerebral malaria may have
sudden marked increases in ICPs, there is no proof that raised
ICPs commonly cause brain stem herniation or compression.
The neurological signs attributed to brain stem compression
have been observed frequently both in patients who died of
cerebral malaria and in survivors of cerebral malaria, which
suggests that the underlying pathological process responsible
for these signs is common; however, results of CT for adults
[16] and the autopsy findings reported so far (again mainly for
adults) have not suggested that herniation is common. A recent
CT study of a selected group of Kenyan children with cerebral
malaria showed some loss of CSF spaces in six of the 14
patients [22]. Three ofthese children had markedly raised ICPs,
but none had abnormalities of the posterior fossa that are sug-
gestive of coning.
MRI should resolve some of these issues because it provides
excellent views of brain stem anatomy and its spatial relations.
Three recent case reports [23, 25, 26] purported to show MRI
findings for patients with cerebral malaria; however, all three
cases were atypical, and only one case fulfilled the diagnostic
criteria for cerebral malaria at the time of MRI. Examination
of a film of peripheral blood from one patient did not reveal
parasites at the time of MRI [23], and this patient may not
have had cerebral malaria at all [24]. The second patient, a 71-
year-old man, had focal neurological signs (aphasia) without
coma and evidence of a left-parietal-lobe hemorrhage and a
large infarct in the parieto-occipital region [25]. The third pa-
tient had MRI evidence of central pontine myelinolysis 14
days after cerebral malaria with multisystem involvement was
diagnosed; however, CT performed during the acute phase of
illness and results of MRI performed on the 10th day of illness
were both normal [26].
Our specific objective of this MRI study of patients with
acute cerebral malaria was to identify brain stem abnormalities
or displacement in vivo. In Thailand most cases of cerebral
malaria (as in this series) occur in adults [3]. Unlike our earlier
less-sensitive CT studies, MRI did show some brain swelling;
however, with the notable exception of two fatal cases, the
changes were subtle and were evident only when scans obtained
during the acute phase of illness were compared with those
obtained during convalescence. Brain swelling was not caused
by cerebral edema, contrary to some recent predictions [27].
The most plausible explanation for these findings is that the
intracerebral blood volume in patients with cerebral malaria is
increased consistently by sequestration of parasitized erythro-
cytes and the compensatory vasodilatation necessary for main-
taining cerebral blood flow. Indeed, it would be surprising if the
brain volume was not increased in patients with acute cerebral
malaria because the majority ofRBCs in venules and capillaries
in most fatal cases are parasitized and sequestered (i.e., stuck)
em 1995;21 (August)
[2, 7-10]. Yet, we have documented cerebral blood flows of
24 to 70 mL/(100 g. min) (mean, 52 mL/[100 g. min]) in adults
with cerebral malaria, thus indicating considerable compensa-
tory vasodilatation (possibly as a result of local lactic acidosis)
[4]. A doubling of cerebral blood volume would increase the
total volume of the brain by ,..., 10%.
In this series brain swelling was associated in some cases
with a very slight downward displacement of the brain stem
(,..., 1 mm on average), but there was no evidence of tentorial
or foramen magnum compression in survivors of cerebral ma-
laria. In comparison with MRI performed several months after
discharge, MRI performed shortly after recovery of conscious-
ness in patients with cerebral malaria indicated that brain vol-
Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021
ume was decreased slightly. Thus, the sequence of events in
patients with cerebral malaria was acute swelling of the brain
(probably as a result of intracerebral vascular sequestration and
vasodilatation), shrinkage of the brain after parasite clearance
and recovery from coma, and a return to normal. The cause
of brain shrinkage during early convalescence is not known,
although changes in osmolality (particularly in the patients with
renal failure) would provide a plausible explanation.
Two of the four patients who died in this series had gross
brain swelling and coning that were demonstrated by MRI. In
both cases MRI during the acute phase of illness was performed
after respiratory difficulties, hyperventilation, and/or erratic la-
bored breathing followed by clinical signs of brain death. MRI
of patient 17 was first performed 4 days after acute respiratory
arrest and clinical signs ofbrain death, thereby making interpre-
tation of events difficult. However, patient 24 underwent MRI
only 5 hours after the diagnosis of brain death. In both cases
emergency-assisted ventilatory support was given because of
the breathing abnormalities. It cannot be ascertained if coning
after acute brain swelling was the fatal event or ifgross swelling
occurred after respiratory difficulties, possible transient hyp-
oxia, and brain death (i.e., as a secondary event).
Two explanations for these fatal cases are possible. First,
some patients with cerebral malaria and an unstable ICP (which
is controlled partly by acidosis-driven hypoventilation and hy-
pocapnia) may suddenly decompensate because of a lethal,
unevenly distributed rise in the ICP that causes fatal coning,
particularly if patients are receiving ventilatory support and the
Paco, is not reduced. This condition might be more likely in
children whose brains fit more tightly inside the rigid cranium
(where there is less room for expansion) than in adults. Lumbar
puncture might be dangerous for these patients because it could
lead to a differential in craniospinal pressure and cause coning.
However, in other larger series, there has been no evident
relation between the time of lumbar puncture and the time of
death, and most authorities recommend that a lumbar puncture
should be performed in all suspected cases of cerebral malaria.
In patient 24, who had the highest opening cerebrospinal pres-
sure obtained during lumbar puncture in this series, respiratory
difficulties developed 3 hours after lumbar puncture, and it is
not possible to say whether these two events are related. The
second explanation is that there is a fatal progression in an
ern 1995;21 (August) MRI of the Brain in Cerebral Malaria
intrinsic pathophysiological process in the brain stem (such as
microvascular sequestration) that initially causes respiratory
abnormalities and finally causes respiratory arrest. This process
leads to CO2 retention, hypoxic cerebral damage, and secondary
brain swelling and herniation.
In support of the second explanation, respiratory abnormalit-
ies and other signs that could be interpreted as indicating coning
do occur in the absence of a raised ICP or brain stem compres-
sion. However, MRI did not show brain stem swelling in this
series, and neuropathological observations do not indicate pref-
erential brain stem sequestration [28]. Although in two of our
four fatal cases brain swelling was gross and sufficient to cause
considerable foramen magnum herniation (confirmed in one
case during autopsy), in a separate autopsy series of62 prospec-
tively studied Vietnamese adults dying of severe falciparum
malaria who were not receiving ventilatory support (N.T.H.
Mai, unpublished data), we observed evidence of herniation in
only two patients. All of these patients with cerebral malaria
underwent lumbar punctures. This fact suggests that swelling
may worsen after brain death in patients who receive ventila-
tory support. These studies confirm that in the majority of cases
of cerebral malaria the brain is not markedly swollen. If raised
ICP causes coning, then it must develop very rapidly in fatal
cases. Whether it happens before or after a lethal brain stem
event in such cases remains unresolved.
From a practical therapeutic standpoint, further studies on
cerebral malaria will be needed to determine the indications
for treatment with osmotic agents, such as mannitol. However,
if patients are receiving artificial ventilatory support, it would
be prudent to maintain hypocarbia to reduce ICP. Because the
clinical pattern of cerebral malaria in adults and children is
different, further information on children with this disease will
be of great interest and importance.
Acknowledgments
The authors thank Professor T. Chongsuphajaisiddhi, Associate
Professor P. Charoenlarp, and the staff of the intensive care unit
(Hospital for Tropical Diseases, Bangkok) for their support; Dr.
C. Sriphojanart (Phyathai X-Ray Computer Center, Bangkok) for
her expertise and assessment; Dr. C. R. 1. C. Newton for advice;
and Nucharee Cholvilai for preparing the manuscript.
References
1. Brewster DR, Kwiatkowski D, White NJ. Neurological sequelae of cere-
bral malaria in children. Lancet 1990; 336: 1039-43.
2. White NJ, Ho M. The pathophysiology of malaria. In: Advances in parasi-
tology. New York: Academic Press, 1992:84- 175.
3. Warrell DA, Looareesuwan S, Warrell MJ, et aI. Dexamethasone proves
deleterious in cerebral malaria: a double-blind trial in 100 comatose
patients. N Engl J Med 1982;306:313-9.
309
4. Warrell DA, White NJ, Veall N, et al. Cerebral anaerobic glycolysis and
reduced cerebral oxygen transport in human cerebral malaria. Lancet
1988;2:534-8.
5. Newton CRJC, Kirkham FJ, Winstanley PA, et aI. Intracranial pressure
in African children with cerebral malaria. Lancet 1991;337:573-6.
6. Feldmann E, Gandy SE, Becker R, et aI. MRI demonstrates descending
transtentorial herniation. Neurology 1988;38:697-701.
7. Marchiafava E, Bignami A. On summer-autumnal fever. London: The
New Sydenham Society, 1894.
8. Dudgeon LS, Clarke C. A contribution to the microscopical histology of
malaria. Lancet 1917; 193: 153-6.
9. Dudgeon LS, Clarke C. An investigation on fatal cases of pernicious
malaria caused by Plasmodium falciparum in Macedonia. Q J Med
1918; 12:372-90.
10. Gaskell JF, Millar WL. Studies on malignant malaria in Macedonia. Q J
Downloaded from https://academic.oup.com/cid/article/21/2/300/292219 by guest on 07 April 2021
Med 1920; 13:381-426.
1I. Rigdon RH. The pathological lesions in the brain in malaria. South Med
J 1944; 37:687 -94.
12. Rigdon RH, Fletcher DE. Lesions in the brain associated with malaria.
Pathological studies on man and on experimental animals. Arch Neurol
Psychiatry 1945;53:191-8.
13. Migasena P, Maegraith BG. Pharmacological action of antimalarial drugs:
action of chloroquine and hydrocortisone on blood-brain barrier in Plas-
modium knowlesi malaria [abstract]. Trans R Soc Trop Med Hyg
1967;61:6.
14. Maegraith BG, Fletcher A. The pathogenesis of mammalian malaria. Adv
Parasitol 1972; 10:49- 76.
15. Hoffman SL, Rustama D, Punjabi NH, et aI. High-dose dexamethasone in
quinine-treated patients with cerebral malaria: a double-blind, placebo-
controlled trial. J Infect Dis 1988; 158:325-31.
16. Looareesuwan S, Warrell DA, White NJ, et aI. Do patients with cerebral
malaria have cerebral oedema? A computed tomography study. Lancet
1983; 1:434-7.
17. Warrell DA, Looareesuwan S, Phillips RE, et aI. Function of the blood-
cerebrospinal fluid barrier in human cerebral malaria: rejection of the
permeability hypothesis. Am J Trop Med Hyg 1986;35:882-9.
18. Waller D, Crawley J, Nosten F, et aI. Intracranial pressure in childhood
cerebral malaria. Trans R Soc Trop Med Hyg 1991; 85:362-4.
19. Kwiatkowski D, Molyneux M, Taylor T, et aI. Cerebral malaria. Lancet
1991; 337: 1281-2.
20. White NJ. Lumbar puncture in cerebral malaria [letter]. Lancet 1991;
338:640-1.
21. Minns RA, Engleman HM, Stirling H. Cerebrospinal fluid pressure in
pyogenic meningitis. Arch Dis Child 1989;64:814-20.
22. Newton CRJC, Peshu N, Kendall B, et aI. Brain swelling and ischaemia
in Kenyans with cerebral malaria. Arch Dis Child 1994;70:281-7.
23. Chia JKS, Nakata MM, Co S. Smear-negative cerebral malaria due to
mefloquine-resistant Plasmodium Jalciparum acquired in the Amazon
[letter]. J Infect Dis 1992; 165:599-600.
24. White NJ, Krishna S, Looareesuwan S. Encephalitis, not cerebral malaria,
is likely cause of coma with negative blood smears [letter]. J Infect Dis
1992; 166:1195-6.
25. Millan 1M, San Millan JM, Munoz M, Navas E, Lopez-Velez R. CNS
complications in acute malaria: MR findings. Am J Neuroradiol
1993; 14:493-4.
26. Kampfi AW, Birbamer GG, pfausler BE, Haring HP, Schmutzhard E.
Isolated pontine lesion in algid cerebral malaria: clinical features, man-
agement, and magnetic resonance imaging findings. Am J Trop Med
Hyg 1993;48:818-22.
27. Poser CN, Roman G. Cerebral malaria [letter]. Lancet 1991;337:1282.
28. Turner G, Mai NTH, Phu NH, et aI. The neuropathology of fatal malaria
in Viet Nam [abstract]. Trans R Soc Trop Med Hyg 1995; 89: 138.