JOURNAL OF MAGNETIC RESONANCE IMAGING 20:496 –500 (2004)

Clinical Note

Metabolites in Ventricular Cerebrospinal Fluid

Detected by Proton Magnetic Resonance
Spectroscopic Imaging
Lidia M. Nagae-Poetscher, MD,1,2 Michael McMahon, PhD,2 Nancy Braverman, MD,1
William T. Lawrie, Jr., AB,1 Ali Fatemi, MD,1 Mahaveer Degaonkar, PhD,1,2
Alena Horská, PhD,2 Martin G. Pomper, MD, PhD,2 Vaddapuram P. Chacko, PhD,2 and
Peter B. Barker, DPhil1,2*

tively insensitive techniques (2). Three cases are pre-

Normally, ventricular cerebrospinal fluid (CSF) contains
sented here that showed unusual elevation of signals in
low levels of all metabolite signals on proton magnetic res-
onance spectroscopic imaging (MRSI). We present here the lateral ventricles, with or without a smaller eleva-
three cases (two with seizure disorders, one with a central tion in the brain parenchyma, on in vivo proton MRSI.
nervous system lymphoma) who presented with unusually
elevated CSF signals on MRSI. Based on chemical shifts
and in vitro studies (in one case), the signals were assigned MATERIALS AND METHODS
to propan-1,2-diol (PD), acetone, and lactate, respectively. All scanning was performed at 1.5 Tesla. Routine brain
These compounds were either exclusively, or more readily,
MRI was performed in all cases. For proton MRSI, four
detected in CSF than in brain. Proton MRSI conveniently
screens both brain and CSF for abnormal metabolism si-
oblique-axial sections (thickness, 15 mm; intersec-
multaneously. tional gap, 2.5 mm) were recorded parallel to the ante-
rior commissure-posterior commissure line (TR/TE,
Key Words: brain; spectroscopy; CSF; acetone; propan- 2300/272 msec; field of view (FOV), 24 cm; matrix,
1,2-diol; lactate 32 ⫻ 32 size; 1 signal average; nominal voxel size, 0.8
J. Magn. Reson. Imaging 2004;20:496 –500.
cm3). At least one of the MRSI sections was placed at
© 2004 Wiley-Liss, Inc.
the level of the lateral ventricles. Outer volume satura-
tion pulses were used for the suppression of lipid sig-
nals originating from the skull and scalp, and a chem-
ANALYSIS OF THE CEREBROSPINAL FLUID (CSF) col-
ical shift-selective saturation pulse was used for water
lected by means of lumbar puncture (LP) has long been
suppression. Full data acquisition (3) and processing
one of the major diagnostic tools in neurology. Although
(4) details are given elsewhere. Chemical shifts were
ex vivo magnetic resonance spectroscopy (MRS) has
measured relative to the brain N-acetyl asparate (NAA)
successfully been used for biochemical analysis of CSF
signal set to 2.02 ppm. Patients either were scanned
(1), in vivo MRS and MR spectroscopic imaging (MRSI)
after informed consent was obtained under protocols
generally show either no or very small metabolite sig-
approved by the local Internal Review Board (IRB), or
nals in CSF, since the concentrations of compounds in
underwent an MRSI scan as part of their routine clini-
CSF are usually too low to be detected by these rela-
cal evaluation. Retrospective review of clinical data was
approved by the local IRB following abstraction of clin-
ical and imaging data to a database with all subject
1
Kennedy Krieger Institute, Baltimore, Maryland. identifiers removed.
2
Russell H. Morgan Department of Radiology and Radiological Science, In case 1, a CSF sample obtained by LP 10 months
Johns Hopkins University School of Medicine, Baltimore, Maryland. prior to the MRSI examination was available for analy-
Contract grant sponsor: NIH; Contract grant numbers: P41RR15241,
R01MH61438; RZ1EB00099J.
sis by two-dimensional MRS (two-dimensional correla-
Presented as a poster at ISMRM 2002. tion spectroscopy, 2D-COSY) at 500 MHz and by gas
The current address of William T. Lawrie, Jr., is Department of Epide- chromatography-mass spectrometry (GC-MS).
miology, Johns Hopkins University School of Public Health, 615 N.
Wolfe St., Baltimore, MD 21205.
*Address reprint requests to: P.B.B., Department of Radiology, MRI CASE REPORTS
143C, Johns Hopkins University School of Medicine, 600 N. Wolfe St.,
Baltimore, MD 21287. E-mail: barker@mri.jhu.edu Case 1
Received May 14, 2003; Accepted May 7, 2004.
DOI 10.1002/jmri.20128 A 19-month-old female with irritability, seizures, and
Published online in Wiley InterScience (www.interscience.wiley.com). severe developmental delay was diagnosed with molyb-
© 2004 Wiley-Liss, Inc. 496
MRS Imaging of Metabolites in CSF
Figure 1. Case 1. A 19-month-old fe-
male with molybdenum cofactor defi-
ciency syndrome. a: FLAIR MRI showing
hyperintensity of the lentiform nuclei bi-
laterally, brain parenchymal volume
loss, and large CSF spaces. b: T1-
weighted MRI showing MRSI voxel loca-
tions in CSF and left posterior frontal
lobe gray matter. c: Metabolic image
from the spectral region centered on
1.14 ppm, showing elevated PD signal in
the lateral ventricles (bright signal at the
edges of the brain originates from sub-
cutaneous lipid). d: Spectrum from CSF
shows a doublet peak of PD at 1.14 ppm.
e: Spectrum from left posterior frontal
lobe shows a relatively low ratio of NAA/
Cho, but no definitive PD signal.
denum cofactor deficiency syndrome, on the basis of
elevated urinary S-sulphocysteine, taurine, sulfites and
purine metabolites, increased blood taurine, decreased
blood cysteine and uric acid, deficiency of fibroblast
sulfite oxidase (5), and detection of both mutant alleles
in the MOCS1A gene (6). She was placed on polyvita-
mins, phenobarbital, and a diet with restriction of sul-
fur-containing amino acids.
MRSI of CSF demonstrated a doublet (spin-spin cou-
pling constant J ⫽ 6.3 Hz) with chemical shift centered
at 1.14 ppm, assigned to propan-1,2-diol (PD) (Fig. 1).
PD was only visible in CSF; no signal in brain paren-
chyma could be identified. No lactate was detectable in
brain or CSF. Brain spectra showed NAA to choline
(Cho) ratios of approximately 1 to 1 (Fig. 1e) in multiple
brain regions, which is somewhat low for a 19-month-
old (7), suggesting widespread neuronal and axonal
damage or dysfunction. MRI showed bilateral hyperin-
tense signal in the putamen on T2 and fluid-attenuated
inversion recovery (FLAIR) MRI, atrophy of the cerebel-
lar vermis, and prominent CSF spaces associated with
significant brain volume loss (Fig. 1a). The patient died
one month after the imaging study.
Two-dimensional-COSY spectra of CSF (obtained 10
months prior to MRSI) were also consistent with an
assignment to PD, with cross peaks at (1.14, 3.92) (f1,
f2) ppm, (3.75, 3.92) ppm, and (3.48, 3.92) ppm. The
specificity of the assignment of PD was further investi-
gated by comparing the CSF spectrum to that of the
pure compounds PD, mercapto-propanol, and di-pro-
pylene glycol. Only PD showed a close resemblance to
the CSF spectrum. Finally, a small quantity of PD was
added (spiked) to the CSF sample to confirm assign-
ment.
Case 2
A 12-year-old female presented with intractable tonic-
clonic seizures, following head trauma and a three-
497
minute period of unconsciousness one and a half years
previously. Neurological examination was normal, and
video-electroencephalogram revealed left parietal and
frontal epileptic foci. Ketogenic diet for seizure control
was initiated, and brain MRSI was performed three
days later. A singlet peak at 2.22 ppm assigned to ace-
tone was seen in both brain and CSF, but with a higher
signal intensity in CSF (Fig. 2). Structural MRI and
brain MRSI (Fig. 2e) were normal.
Case 3
A 40-year-old male, with diagnosis of AIDS for two
years, presented with a two-week history of altered
mental status after discontinuation of antiretroviral
medication because of possible drug-induced hepatitis.
CD4 levels were 5 cells/mm3 and viral load was
242,000 copies/mL. Analysis of CSF obtained by LP
showed 9 white cells/mm3, no red cells, 161 mg/dL
protein, and 44 mg/dL glucose. Cryptococcal antigen
was negative.
Brain MRI showed ring-enhancing lesions within the
left frontal lobe (1.5 cm) and right cerebellar hemi-
sphere (0.8 cm) and a nonenhancing lesion superior to
the right lateral ventricle. The periventricular region
also showed increased T2 signal, which enhanced after
contrast administration. The patient was placed on
treatment for both cytomegalovirus and toxoplasmosis.
A thalium single photon emission computed tomogra-
phy (SPECT) scan showed increased uptake in all three
lesions. The patient’s clinical condition worsened and a
follow-up MRI performed eight days later showed that
the left frontal lesion had increased in size to 3 cm with
increased mass effect and subependymal nodular en-
hancement along both lateral ventricles. A presumed
diagnosis of CNS lymphoma was established. MRSI at
this time showed a markedly elevated doublet (chemical
shift ⫽ 1.33 ppm, J ⫽ 7 Hz) assigned to lactate in CSF
(estimated concentration of 8.1 mM using phantom re-
498
placement methodology (4)), with a smaller lactate sig-
nal (estimated concentration of 4.8 mM) in the left fron-
tal mass (Fig. 3d and e). The left frontal lesion (Fig. 3e)
also exhibited increased Cho and decreased NAA and
creatine (Cr) signals compared to the contralateral
hemisphere (Fig. 3f). No biopsy or radiotherapy was
performed because of the patient’s poor clinical condi-
tion, and the patient died five days later.
DISCUSSION
All three cases had abnormally elevated signals in the
ventricular CSF. In patient 1, it was unclear whether
the elevated signal was due to either exogenous admin-
Nagae-Poetscher et al.
Figure 2. Case 2. A 12-year-old female
on day 3 of the ketogenic diet for seizure
control. a: FLAIR MRI showing no ab-
normalities. b: T1-weighted MRI demon-
strating MRSI voxel localizations in CSF
and left posterior frontal gray matter. c:
Metabolic image from the spectral re-
gion centered on 2.22 ppm, showing el-
evated acetone signal in the lateral ven-
tricles. d: Spectrum from predom-
inantly CSF showing a signal from ace-
tone at 2.22 ppm (Cho, Cr, and NAA
signals arise from partial volume with
surrounding brain tissue, since the pa-
tient had small ventricles). e: Left pos-
terior frontal lobe showing a normal
brain spectrum. The acetone peak has
the largest signal intensity in CSF.
istration or endogenous production of PD. PD is a com-
mon drug delivery vehicle and has previously been de-
tected by single-voxel (SV) MRS in the brain
parenchyma (and in vitro MRS of CSF) of neonates
receiving intravenous phenobarbital for seizure control
(8). PD can readily be distinguished from lactate by its
chemical shift (1.14 vs. 1.33 ppm for lactate). Retro-
spective chart review of the patient presented here
failed to demonstrate recent administration of any med-
ications or other substances known to contain PD, al-
though we cannot rule out recent administration by
another institution. In the absence of any identifiable
exogenous source of PD, and the fact that elevated CSF
PD was observed on separate measurements made 10
Figure 3. Case 3. A 40-year-old male
with CNS lymphoma. a: T2-weighted MRI
showing mass lesion in left frontal lobe
and subependymal nodules. b: Unen-
hanced T1-weighted MRI demonstrating
MRSI voxel locations in CSF, left frontal
mass, and contralateral hemisphere. c:
Metabolic image from the spectral region
centered on 1.33 ppm, showing a large
lactate signal in the lateral ventricles
and a smaller elevation of lactate in the
left frontal lesion. Spectrum from CSF
(d) showing a large lactate doublet at
1.33 ppm. Spectrum from the left frontal
mass (e) showing elevated Cho and lac-
tate signals and reduced levels of NAA
and Cr compared to normal proton spec-
trum from the right frontal lobe (f).
MRS Imaging of Metabolites in CSF
months apart, it appears possible that PD could be due
to endogenous production. PD has been shown to be
endogenously produced by glycerol metabolism in ani-
mal models (9), and an elevated level of glycerol was
also detected in vitro by GC-MS in the CSF sample
taken 10 months prior to MRSI. Glycerol has been ob-
served to be elevated in ischemia, trauma, and post-
mortem brain (10). Therefore, one possibility is that the
PD may arise from endogenous glycerol metabolism
possibly related to the underlying metabolic disease,
although further investigation will be required to con-
firm or deny this hypothesis. A number of other sub-
stances were considered as possible assignments for
the 1.14 ppm doublet, including ␣-oxoisovalerate (11),
2,3-butanediol, mercapto-propanol, di-propylene gly-
col, and methylmalonic acid; however, none of these
were consistent with the in vitro 2D-COSY analysis of
CSF.
A prior SV-MRS study of the parietal region in one
patient with molybdenum cofactor deficiency (12) dem-
onstrated an elevated doublet peak in the aliphatic re-
gion of the spectrum, which the authors assigned to
lactate, but MRS of CSF was not reported.
In patient 2, an abnormally elevated acetone signal
(singlet, 2.22 ppm) was observed in both brain and
CSF, with a larger intensity in CSF. Elevated brain
acetone in association with the ketogenic diet (13) and
diabetic ketoacidosis (14) has been previously reported
in brain using SV-MRS, but CSF was not examined in
these studies. The increased signal in CSF compared to
brain may be due to either increased acetone concen-
tration in CSF or increased sensitivity of detection in
CSF by long TE MRSI due to decreased line widths and
longer T2 relaxation times. It is expected that metabolite
T2 and T2* relaxation times will be longer in CSF than
the brain, because of the greater fluidity (lower viscos-
ity) of CSF and the absence of microscopic susceptibil-
ity effects found in the brain due to deoxyhemoglobin in
the microvasculature (15). While metabolite T1 values
may also be increased in CSF compared to the brain,
the long TR and TE MRSI sequence used here is rela-
tively more sensitive to changes in T2 than in T1. There-
fore, even when compounds are equally distributed be-
tween the brain and CSF, they may nevertheless be
more readily detectable in CSF, particularly when using
long echo times for MRS or MRSI.
In patient 3, larger lactate signals were seen in CSF
than in the lymphoma, and this may be the result of
differences in relaxation times as discussed above.
However, it is also possible that CSF lactate concentra-
tions may be higher than those in the brain or lesions,
depending on the rates of production, transport, and
clearance of lactate from the different regions (16). We
speculate that the glycolytic metabolism of the lym-
phoma (particularly that of the periventricular and sub-
ependymal lesions) produced large amounts of lactate
that were transported into CSF, where it accumulated
in high levels. Lactate clearance from CSF is known to
be slow (17). Elevated lactate in CSF (either with or
without concomitant detection in brain parenchyma)
has been observed in several other pathologies, includ-
ing stroke, AIDS dementia (without brain lesions (2)),
mitochondrial diseases (18), choroid plexus carcinoma
499
(19), and methylmalonic academia (20). The case pre-
sented here is notable for the very large increase of
lactate in CSF (estimated concentration of 8.1 mM, at
least an order of magnitude larger than normal), com-
pared to that found both in normal CSF and in these
other pathologies.
In summary, proton MRSI of CSF may occasionally
demonstrate elevated metabolite signals that are either
not visible or are less prominent in brain parenchyma.
Since multislice or three-dimensional MRSI almost al-
ways includes voxel locations in the lateral ventricles,
no additional scan time (unlike SV-MRS) is required to
investigate CSF using this methodology.
ACKNOWLEDGMENTS
We thank Dr. Jeffrey Duyn and Dr. Jan Willem van der
Veen (National Institutes of Health, Bethesda, MD) for
the MRSI pulse sequence. We thank Barbara Amann
(Johns Hopkins University) for performing the in vitro
CSF spectroscopic analysis and 2D-COSY experiments,
Dr. Richard Kelley (Kennedy Krieger Institute) for per-
forming GC-MS and for his helpful comments on the
manuscript, Greg Lukaszczyk, RPh (Johns Hopkins
University), for his help with medicament composition,
and Dr. Dermot O’Hare (Oxford University, UK) for
helpful discussions.
REFERENCES
1. Petroff OA, Yu RK, Ogino T. High-resolution proton magnetic reso-
nance analysis of human cerebrospinal fluid. J Neurochem 1986;
47:1270 –1276.
2. Barker PB, Lee RR, McArthur JC. AIDS dementia complex: evalu-
ation with proton MR spectroscopic imaging. Radiology 1995;195:
58 – 64.
3. Duyn JH, Gillen J, Sobering G, van Zijl PCM, Moonen CTW. Multi-
section proton MR spectroscopic imaging of the brain. Radiology
1993;188:277–282.
4. Soher BJ, van Zijl PCM, Duyn JH, Barker PB. Quantitative proton
spectroscopic imaging of the human brain. Magn Reson Med 1996;
35:356 –363.
5. Johnson JL, Rajagopalan KV, Wadman SK. Human molybdenum
cofactor deficiency. Adv Exp Med Biol 1993;338:373–378.
6. Reiss J. Genetics of molybdenum cofactor deficiency. Hum Genet
2000;106:157–163.
7. Kreis R, Ernst T, Ross BD. Development of the human brain: in vivo
quantification of metabolite and water content with proton mag-
netic resonance spectroscopy. Magn Reson Med 1993;30:424 – 437.
8. Cady EB, Lorek A, Penrice J, et al. Detection of propan-1,2-diol in
neonatal brain by in vivo proton magnetic resonance spectroscopy.
Magn Reson Med 1994;32:764 –767.
9. Michal G. Biochemical pathways: an atlas of biochemistry and
molecular biology. New York: Wiley; 1998.
10. Michaelis T, Helms G, Frahm J. Metabolic alterations in brain
autopsies: proton NMR identification of free glycerol. NMR Biomed
1996;9:121–124.
11. Lutz NW, Maillet S, Nicoli F, Viout P, Cozzone PJ. Further assign-
ment of resonances in 1H NMR spectra of cerebrospinal fluid (CSF).
FEBS Lett 1998;425:345–351.
12. Salvan AM, Chabrol B, Lamoureux S, Confort-Gouny S, Cozzone
PJ, Vion-Dury J. In vivo brain proton MR spectroscopy in a case of
molybdenum cofactor deficiency. Pediatr Radiol 1999;29:846 – 848.
13. Seymour KJ, Bluml S, Sutherling J, Sutherling W, Ross BD. Iden-
tification of cerebral acetone by 1H-MRS in patients with epilepsy
controlled by ketogenic diet. MAGMA 1999;8:33– 42.
14. Kreis R, Ross BD. Cerebral metabolic disturbances in patients with
subacute and chronic diabetes mellitus: detection with proton MR
spectroscopy. Radiology 1992;184:123–130.
500
15. Michaeli S, Garwood M, Zhu XH, et al. Proton T2 relaxation study
of water, N-acetylaspartate, and creatine in human brain using
Hahn and Carr-Purcell spin echoes at 4T and 7T. Magn Reson Med
2002;47:629 – 633.
16. Inao S, Marmarou A, Clarke GD, Andersen BJ, Fatouros PP, Young
HF. Production and clearance of lactate from brain tissue, cerebro-
spinal fluid, and serum following experimental brain injury. J Neu-
rosurg 1988;69:736 –744.
17. Valenca LM, Shannon DC, Kazemi H. Clearance of lactate from the
cerebrospinal fluid. Neurology 1971;21:615– 620.
Nagae-Poetscher et al.
18. Lin DD, Crawford TO, Barker PB. Proton MR spectroscopy in the
diagnostic evaluation of suspected mitochondrial disease. AJNR
Am J Neuroradiol 2003;24:33– 41.
19. Horská A, Ulug AM, Melhem ER, et al. Proton magnetic resonance
spectroscopy of choroid plexus tumors in children. J Magn Reson
Imaging 2001;14:78 – 82.
20. Trinh B-C, Melhem ER, Barker PB. Multi-slice proton MR spectros-
copy and diffusion-weighted imaging in methylmalonic acidemia:
report of two cases and review of the literature. AJNR Am J Neu-
roradiol 2001;22:831– 833.