CA S E S TU DY
www.nature.com/clinicalpractice/neuro

BOLD-MRI cerebrovascular reactivity findings
in cocaine-induced cerebral vasculitis
Jay S Han, Daniel M Mandell, Julien Poublanc, Alexandra Mardimae, Marat Slessarev, Cheryl Jaigobin,
Joseph A Fisher and David J Mikulis*
S U M M ARY
Background An 18-year-old woman presented to a regional stroke center

with dysphasia and right hemiparesis 2 days after consuming alcohol and
inhaling cannabis and—for the first time—cocaine.
Investigations Physical examination, blood tests for inflammatory
markers, vasculitis and toxicology screen, echocardiography,
electrocardiography, CT scanning, brain MRI, magnetic resonance
angiography, magnetic resonance vessel wall imaging, catheter angiography,
and correlation of blood oxygen level-dependent (BOLD)-MRI signal
intensity with changes in end-tidal partial pressure of carbon dioxide.
Diagnosis Cocaine-induced cerebral vasculitis.
Management No specific therapy was initiated. The patient’s vital signs
and neurological status were monitored during her admission. Follow-up
medical imaging was performed after the patient’s discharge from hospital.
keywords cerebral vasculitis, cerebrovascular reactivity, cocaine, MRI, stroke

JS Han is an MSc candidate in the Department of Physiology at the University
of Toronto and the Department of Anesthesia at the University Health Network,
and DM Mandell is a Resident in the Department of Medical Imaging at the
University Health Network, Toronto. A Mardimae is a graduate, M Slessarev is
a medical student, C Jaigobin is a Stroke Neurologist and Assistant Professor of
Medicine, and JA Fisher is an Anesthesiologist and Professor, all at the University
of Toronto, Toronto. DJ Mikulis is Full Professor and Co-Director of Medicine
Imaging Research in the Department of Medical Imaging at the University
Health Network and the University of Toronto, Toronto, ON, Canada.
J Poublanc is a graduate of the University of Toulouse, Toulouse, France.
Correspondence
*Department of Medical Imaging, The Toronto Western Hospital, MCL 3-431, 399 Bathurst Street,
Toronto, ON M5T 2S8, Canada
mikulis@uhnres.utoronto.ca
Received 28 January 2008 Accepted 18 August 2008 Published online 7 October 2008
www.nature.com/clinicalpractice
doi:10.1038/ncpneuro0918

628 nature clinical practice NEUROLOGY

THE CASE

An 18-year-old female was brought to the emergency department of a community hospital by
paramedics after her mother found it difficult to
rouse her from sleep. The patient’s vital signs were
as follows: blood pressure 127/69 mmHg, heart
rate 69 beats/min, respiration rate 16 breaths/min,
temperature 36.9 °C, and pulse oxygen saturation 96%. Two days previously, the patient had
ingested alcohol and inhaled cannabis and—for
the first time—cocaine. The only positive physical findings on examination of the patient were
non-fluent speech with impairment of naming,
repetition and comprehension, and weakness in
the right side of her face and in her right arm and
leg. Non-enhanced CT of the brain revealed an
acute infarction in the left middle cerebral artery
(MCA) territory. Toxicology screens were positive only for cocaine and cannabis. The patient
was transferred to a regional stroke center having
spent less than 12 h at the community hospital.
On arrival at the stroke center, the patient was
alert but dysphasic. Motor examination confirmed the presence of right-sided weakness and
demonstrated an extensor plantar response on
the right side. There was no evidence of joint
swelling, skin rash or any other peripheral mani­
festations of vasculitis. The following laboratory investigations were performed: complete
blood count; prothrombin time–international
normalized ratio; partial thromboplastin time;
liver function tests; and measurement of levels of
glucose, bilirubin, amylase, serum sodium, potassium, chloride, bicarbonate, creatinine, calcium
(total and corrected), magnesium, total protein,
albumin, and phosphate. Abnormal test results
were limited to the leukocyte count (14.0 × 109/l;
reference range 4.0–11.0 × 109/l), neutrophils
(9.8 × 109/l; reference range 2.0–7.5 × 109/l),
monocytes (0.9 × 109/l; reference range 0.2–
0.8 × 109/l), total calcium (2.01 mmol/l; reference
range 2.20–2.62 mmol/l), albumin (33 g/l; reference range 38–50 g/l) and bicarbonate (21 mmol/l;
reference range 25–35 mmol/l). A 12-lead
­electrocardiogram showed that the patient had
November 2008 vol 4 no 11

CA S E S TU DY
www.nature.com/clinicalpractice/neuro

A

C

E

B

D

F

Figure 1 Results from brain imaging studies performed at initial presentation of a patient with cocaine-induced cerebral vasculitis.
(A) An axial T2 FLAIR image and (B) an apparent diffusion coefficient map show intensity changes in the left middle cerebral artery territory,
consistent with acute infarction. (C) Coronal maximum intensity projection from a dynamic contrast-enhanced magnetic resonance
angiogram of the circle of Willis shows concentric beaded stenosis of the distal left internal carotid artery, and of the M1 and proximal M2
segments of the left middle cerebral artery. There is also narrowing of the distal left internal carotid artery. There is a narrow A1 segment of
the left anterior cerebral artery with a more-normal-appearing distal portion at the junction with the anterior communicating artery (arrow),
suggesting that this vessel is also affected by the vasculopathy rather than representing congenital hypoplasia. (D) Post-gadolinium
coronal T1 FLAIR imaging shows diffuse enhancement of the left middle cerebral artery vessel wall (arrows). Catheter angiography with
injection of the (E) right and (F) left internal carotid arteries confirms the findings seen on magnetic resonance angiography but also shows
involvement of the left A1 segment (arrow in Figure 1F). Abbreviation: FLAIR, fluid-attenuated inversion recovery.

a normal sinus rhythm. Normal heart echo­
cardio­gram findings diminished the likelihood
of a cardio­embolic source of stroke. T2 fluidattenuated­ inversion recovery (FLAIR) imaging
of the brain at 3 T showed an acute infarc­tion in
the left MCA territory (Figure 1A). An apparent­
diffusion coefficient map showed restricted diffusion corresponding to the area of T2 hyper­
intensity on the FLAIR image (Figure 1B).
Magnetic resonance angiography (MRA)
showed a beaded pattern of concentric stenosis
involving the distal portion of the left internal
carotid artery and left MCA (Figure 1C). The
A1 segment of the left anterior cerebral artery
was also diminished in size; it was difficult to
determine whether this narrowing was the
result of a congenital hypoplasia or an acquired
vasculo­pathy, but the latter seemed more probable because of the more normal appearance in
caliber of the distal aspect of the A1 segment.
The distal left internal carotid artery was also

narrowed. T1 FLAIR imaging performed after
intravenous administration of gadolinium
showed diffuse concentric enhancement of the
wall of the left MCA (Figure 1D). This beaded
pattern of wall enhancement would not have
been visible on a 1.5 T system without considerable lengthening of the sequence time (from 5 to
20 min) to compensate for the loss of signal at
the higher spatial resolution required. Findings
from catheter angiograms of the patient’s right
and left internal carotid arteries were concordant­
with the MRA results (Figure 1E,F).
The vasculitic pattern on imaging prompted
laboratory investigations for indicators of pri­
mary vasculitis (anti-nuclear antibody, and
complements C3 and C4) and hyper­coagulability
(antithrombin III, protein C, protein S, homocysteine, and antiphospholipid antibody). The
results of these tests were within normal limits.
With the imaging findings suggestive of a vascu­
litis­, clinical history of cocaine use, and lack of

ncpneuro_2008_022f1.eps
November 2008 vol 4 no 11 HAN ET AL. 

nature clinical practice NEUROLOGY 629

CA S E S TU DY
www.nature.com/clinicalpractice/neuro

+0.7

A

0

–0.7
+0.7

B

0

–0.7

Figure 2 Whole-brain BOLD-MRI cerebrovascular reactivity maps obtained after admission of a patient
with cocaine-induced cerebral vasculitis. (A) BOLD-MRI cerebrovascular reactivity maps acquired
2 days after admission show reduced reactivity in the left middle cerebral artery territory. Absence of
color indicates exhausted reactivity, blue represents the steal phenomenon,1 and red, orange and yellow
represent normal reactivity. (B) BOLD-MRI cerebrovascular reactivity maps at 4-week follow-up show
definite interval improvement in BOLD cerebrovascular reactivity on the left side. There is a small focus of
persisting paradoxical reactivity in the region of a left corona radiata infarct. Units are percentage change
in BOLD magnetic resonance signal per mmHg change in end-tidal partial pressure of carbon dioxide.
Abbreviation: BOLD, blood oxygen level-dependent.

evidence for primary vasculitis, a diagnosis of
cocaine-induced cerebral vasculitis­ was made.
The patient was enrolled in an ongoing
functional MRI study designed to assess
cerebro­vascular reactivity (CVR) with blood
oxygen level-dependent (BOLD)-MRI during
controlled changes in end-tidal partial pressure of carbon dioxide (PETCO2).1 Iso-oxic
(~100 mmHg) and normocapnic (~40 mmHg)
to hypercapnic­ (~50 mmHg) step changes
in PETCO2 were actuated with a computercontrolled custom-built gas delivery system
(RespirAct™, Thornhill Research Inc., Toronto,
Canada).2 MRI scans were performed on a GE
Signa® 3.0 T scanner (General Electric Health
Care, Milwaukee, WI) with an eight-channel
phased array head coil. T1-weighted anatomical
images were acquired through the entire brain
with a three-dimensional­ spoiled gradient­-echo
pulse sequence (slice thickness 2.2 mm, matrix
size 128 × 128). Whole-brain BOLD-MRI CVR
data—percentage change in BOLD magnetic
resonance signal per mmHg change in PETCO2
between the normocapnic and hypercapnic
states—were acquired with a BOLD pulse

sequence (T2*-weighted two-dimensional echoplanar gradient-echo). The BOLD-MRI CVR
map was then color coded on a voxel by voxel
basis and overlaid onto the anatomical images.
The BOLD-MRI CVR study performed in this
patient 2 days after admission showed that she
had reduced reactivity in the left MCA territory
(Figure 2A), in which several foci of paradoxical
(negative) reactivity could also be seen.
Immunosuppressive therapy was withheld
as the working diagnosis was cocaine-induced
cerebral vasculitis rather than primary vascu­
litis. Over the next few days, the patient’s speech
and extremity strength improved. One week
after her initial presentation to hospital, she was
discharged. At this time she was coherent, had
normal speech, and had only mild residual right
arm weakness.
After discharge, the patient continued to
improve clinically. She returned to the stroke
cen­ter for follow-up imaging studies 4 weeks later.
At this time she denied having had any exposure
to alcohol, cannabis or cocaine since her last hospital visit. Her clinical deficits­ consisted only of
occasional word-finding ­difficulties. Repeat MRI

ncpneuro_2008_022f2.eps
630 nature clinical practice NEUROLOGY

HAN ET AL. November 2008 vol 4 no 11

CA S E S TU DY
www.nature.com/clinicalpractice/neuro

of the brain showed the expected evolution of a
left MCA territory infarction (Figure 3A). Vessel
wall imaging showed interval resolution of the
previously seen MCA vessel wall enhancement
(Figure 3B). However, despite clini­cal improvement, there was progressive narrowing­ of the left
MCA on MRA (Figure 3C). Given the disparity
between the MRA findings­ and the patient’s
clinical course, a repeat BOLD-MRI CVR examination was performed (Figure 2B). This investigation showed almost complete recovery of CVR
in those regions that were impaired on the initial
study (Figure 2A).

A

DISCUSSION OF DIAGNOSIS

Figure 3 Follow-up MRI findings 1 month after discharge of a patient with
cocaine-induced cerebral vasculitis. (A) An axial T1 FLAIR image shows
hyperintensity in the left insular cortex (arrow), consistent with laminar necrosis.
(B) A coronal post-gadolinium T1 FLAIR image shows interval resolution of
the middle cerebral artery vessel wall enhancement, as well as insular cortex
laminar necrosis as seen on the pre-gadolinium T1 FLAIR image. (C) A coronal
maximum intensity projection from a dynamic contrast-enhanced magnetic
resonance angiogram of the circle of Willis shows interval worsening of distal
left internal carotid artery and middle cerebral artery stenosis, and interval
narrowing of the A1 segment of the left anterior cerebral artery. Abbreviation:
FLAIR, fluid-attenuated inversion recovery.

A small number of reports describing cocaineinduced cerebral vasculitis have been published.3,4 In most instances, the diagnosis of
this disorder is based on angiographic findings,4
an absence of laboratory markers of primary
vascu­litis, and a clinical history of cocaine
use. Cocaine-induced vasculitis is a condition
belonging to the group of reversible cerebral
vaso­constriction syndromes. This group of
syndromes can be distin­guished from primary
angiitis of the CNS by clinical features such as
an acute-onset headache associa­ted with focal
­neuro­logical deficits, and the ­ reversibility of
vascular abnormalities within days to weeks
of presenta­tion. This is in contradistinction to
primary angiitis of the CNS, which typically
presents in a subacute or chronic fashion with
vascular changes that are frequently irreversible.
The exact mechanism that underlies cocaineinduced vasculitis is unknown. Studies have
demonstrated that cocaine has an apoptotic
effect on cerebrovascular smooth muscle cells5
and a capacity to enhance leukocyte migration
across cerebral blood vessel walls,6 both of which
result in vessel wall inflammation. Moreover, it
has also been reported that both cocaine7 and
its metabolites8 directly constrict cerebral blood
vessels. Following uptake into the blood, cocaine
is rapidly converted into two pre­dominant meta­
bolites: benzoylecgonine and ecgonine methyl
ester.9 Ecgonine methyl ester has a mild vaso­
dilatory effect10 but benzoylecgonine has a predominant vasoconstrictive effect8 on cere­bral
vessels. Most of the absorbed cocaine and its
metabolites are excreted by the kidney within
the first 24 h, but residual benzoylecgonine and
ecgonine methyl ester have been detected in urine
6–14 days after the initial exposure.9 The persistence of benzoylecgo­nine in the blood is likely to

B

C

explain the prolonged vasoconstriction seen in
patients after exposure to cocaine.11 Interestingly,
it has been hypothe­sized that prolonged vaso­
constriction could progress to secondary angi­
itis­.12 The late angio­graphic picture seen in such
patients might, therefore, be attributable to a
combination of both the proinflam­matory and
the vasoconstrictive­ effects of cocaine.
To the best of our knowledge, this is the first
report of changes in regional BOLD-MRI CVR
in cerebral vasculitis. While both the BOLD-MRI
CVR maps and the cerebral angiogram findings
were concordant with this patient’s symptoms
in the acute phase, only BOLD-MRI CVR correlated with symptoms at 4-week follow-up. In this
case, we believe the CVR map may have added
objective quantifiable information to what is
otherwise a subjective clinical assessment of the
success in recruitment of collateral blood flow
to a ­ hemodynamically compromised vascular
­territory. BOLD-MRI CVR may also be of value
in the assessment of patients in whom there is a
strong clinical suspicion of cerebral vascu­litis but
no supportive findings on angiography because
of involvement predominantly of vessels too
small to be resolved by current angiographic techniques.13 In these cases, measurement of CVR by
BOLD-MRI would reflect the hemo­dynamics at

ncpneuro_2008_022f3.eps
November 2008 vol 4 no 11 HAN ET AL. 

nature clinical practice NEUROLOGY 631

CA S E S TU DY
www.nature.com/clinicalpractice/neuro

Competing interests
JA Fisher, M Slessarev and
DJ Mikulis have declared
an association with the
following company:
Thornhill Research. See the
article online for full details
of the relationship. The
other authors declared no
competing interests.

the capil­lary level and would be a more sensitive
test of small vessel involvement. Indeed, BOLDMRI CVR should be capable of assessing­ the
impact of vasculopathy at any level of the precapillary vascular system. In the case patient, the
improvement in CVR paralleled the improvement in her clinical condition despite worsening­
of the proximal­ large vessel vascular stenosis.
We attribute the change in CVR map as well
as the improved clinical signs to the successful
recruitment of cortical collateral blood flow. This
finding supple­ments the suggestion of Yuh et al.
that functional imaging modalities should be
tested in the setting of cerebral vasculitis.14
TREATMENT AND MANAGEMENT

A key feature of the treatment of drug-induced
cerebral vasculitis is the identification and withdrawal of the offending agent.15 The clinical
improvement of the present patient during her
hospital admission and following discharge supports such an action. Had the patient’s clinical
condition continued to deteriorate, she would
have been prescribed calcium-channel blockers
such as verapamil or nimodipine, possibly with
added glucocorticoids.16 Prolonged immunosuppressive therapy or treatment with cytotoxic agents was not seriously con­sidered.16 The
manage­ment decisions regarding this patient
were made predominantly on the basis of clinical assessment rather than as the result of any
particular radiological image. The BOLD-MRI
CVR maps are consistent with our understanding­
that clinical assessment reflects the adequacy of
perfusion of the affected part of the brain from
collateral vessels as well as from feeding arteries.
However, clinical examination cannot discriminate between the adequacy of feeding vessels
and the adequacy of the collateral circulation, a
critical point if revascularization (angioplasty or
vascular bypass) is under consideration as a treatment option. Only in a larger series of cocaineinduced vasculitides can the value of exhausted
non-revascularized auto­regulation be determined
in the management­ of these patients.
CONCLUSIONS

The described case of cocaine-induced cerebral
vasculitis resolved following discontinuation of
the offending agent, as would have been expected.

632 nature clinical practice NEUROLOGY

Of particular note, however, is the fact that noninvasive BOLD-MRI CVR maps were consistent
with the clinical assessment and seemed to provide
information above that given by routine clinical
structural imaging of the brain (MRI and MRA)
and by catheter angiography. Adequate assessment
of the value of BOLD-MRI CVR as a clinical tool
for diagnosis and prognostication of cerebral vasculitis requires additional study, but this method
could prove a useful supplement to currently used
imaging methods in this condition.
References
1 Mandell DM et al. (2008) Mapping cerebrovascular
reactivity using blood oxygen level-dependent MRI
in patients with arterial steno-occlusive disease:
comparison with arterial spin labeling MRI. Stroke
39: 2021–2028
2 Slessarev M et al. (2007) Prospective targeting and
control of end-tidal CO2 and O2 concentrations.
J Physiol 581: 1207–1219
3 Kaye BR and Fainstat M (1987) Cerebral vasculitis
associated with cocaine abuse. JAMA 258:
2104–2106
4 Krendel DA et al. (1990) Biopsy-proven cerebral
vasculitis associated with cocaine abuse. Neurology
40: 1092–1094
5 Su J et al. (2003) Cocaine induces apoptosis in cerebral
vascular muscle cells: potential roles in strokes and
brain damage. Eur J Pharmacol 482: 61–66
6 Gan X et al. (1999) Cocaine enhances brain endothelial
adhesion molecules and leukocyte migration. Clin
Immunol 91: 68–76
7 Kaufman MJ et al. (1998) Cocaine-induced cerebral
vasoconstriction detected in humans with magnetic
resonance angiography. JAMA 279: 376–380
8 Schreiber MD et al. (1994) Effects of cocaine,
benzoylecgonine, and cocaine metabolites in
cannulated pressurized fetal sheep cerebral arteries.
J Appl Physiol 77: 834–839
9 Das G and Laddu A (1993) Cocaine: friend or foe?
(Part 1). Int J Clin Pharmacol Ther Toxicol 31: 449–455
10 Madden JA and Powers RH (1990) Effect of cocaine
and cocaine metabolites on cerebral arteries in vitro.
Life Sci 47: 1109–1114
11 Mena I et al. (1994) Cerebral blood flow changes with
acute cocaine intoxication: clinical correlations
with SPECT, CT, and MRI. NIDA Res Monogr 138:
161–173
12 Calabrese LH and Duna GF (1995) Evaluation and
treatment of central nervous system vasculitis. Curr
Opin Rheumatol 7: 37–44
13 Benseler SM et al. (2005) Angiography-negative
primary central nervous system vasculitis in children:
a newly recognized inflammatory central nervous
system disease. Arthritis Rheum 52: 2159–2167
14 Yuh WT et al. (1999) Diagnosis of microvasculopathy
in CNS vasculitis: value of perfusion and diffusion
imaging. J Magn Reson Imaging 10: 310–313
15 Cuellar ML (2002) Drug-induced vasculitis. Curr
Rheumatol Rep 4: 55–59
16 Calabrese LH and Duna GF (1996) Drug-induced
vasculitis. Curr Opin Rheumatol 8: 34–40

HAN ET AL. November 2008 vol 4 no 11