Prostaglandins, Leukotrienes and Essential Fatty Acids 70 (2004) 357–360

In vivo MR spectroscopy in diagnosis and research of

neuropsychiatric disorders

I.J. Cox*, B.K. Puri

Faculty of Medicine, Imaging Sciences Department, Imperial College London, Division of Clinical Sciences, Robert Steiner Magnetic Resonance Unit,

Hammersmith Campus, Du Cane Road, London W12 0HS, UK

Accepted 18 December 2003

Abstract

Magnetic resonance spectroscopy is one of the most important tools for quantitative analysis of chemical composition and

structure, and this non-invasive technique is now being applied in vivo to study biochemical processes in those neuropsychiatric
disorders that are part of the phospholipid spectrum. Interpretation of a clinical magnetic resonance spectrum can provide
information about membrane phospholipid turnover, cellular energetics, neuronal function, selected neurotransmitter activity and
intracellular pH. Cerebral proton and phosphorus magnetic resonance spectroscopy findings are summarized in relation to
schizophrenia, dyslexia and chronic fatigue syndrome.
r

2004 Elsevier Ltd. All rights reserved.

1. Introduction

Nuclear magnetic resonance (NMR) spectroscopy is

one of the most important tools for quantitative analysis
of chemical composition and structure. Indeed four
Nobel prizes have been awarded in the field: to Rabi in
1944 for his resonance method for recording the
magnetic properties of atomic nuclei; to Purcell and
Bloch in 1952 for their development of new methods for
nuclear magnetic precision measurements and discov-
eries in connection therewith; to Ernst in 1991 for his
contributions to the development of the methodology of
high resolution NMR spectroscopy; and to W

.uthrich in

2002 for his development of NMR spectroscopy for
determining the 3D structure of biological macromole-
cules in solution. In addition, the Nobel Prize in
Physiology or Medicine 2003 was awarded jointly to
Lauterbur and Mansfield ‘for their discoveries concern-
ing magnetic resonance imaging’.

In the last two decades the NMR technique has been

applied in vivo to image the human brain (magnetic
resonance imaging (MRI)) and also to study biochem-
ical processes in the human brain (clinical magnetic
resonance spectroscopy (MRS)). Specifically, certain

atomic nuclei, such as hydrogen-1 (proton,

1

H),

phosphorus-31 (

31

P), carbon-13 (

13

C), fluorine-19 (

19

F)

and nitrogen-15 (

15

N), can be imagined to act like tiny

bar magnets when placed in a magnetic field. Each
nucleus type resonates at a characteristic frequency
when placed in the same magnetic field. For example at
1.5 T the resonant frequencies for

1

H and

31

P nuclei are

63.7 and 25.8 MHz, respectively. During relaxation
following excitation, radiofrequency signals can be
detected which contain information about the magnetic
environment experienced by each nucleus. In a MRI
study this information is related to the spatial position
of the nucleus and in MRS studies this information is
related to the molecules in which the nuclei are
contained. Considering MRS applications the resulting
signal, the free induction decay (FID), can be resolved
into a frequency spectrum by the mathematical function
of Fourier transformation. In the absence of any
magnetic field gradients the local magnetic environment
of a nucleus, and therefore it resonance frequency, is
influenced by its immediate chemical environment. The
relative frequency position can be described by a
parameter known as chemical shift, a dimensionless
unit accounting for the strength of the static magnetic
field and measured in parts per million.

In principle, the intensity of the metabolite signal is

directly related to its concentration, but in practice there
are so many variables that influence the signal intensity

ARTICLE IN PRESS

*Corresponding author. Tel.: [phone]; fax: [phone]-

3038.

E-mail address:

[e-mail] (I.J. Cox).

0952-3278/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.plefa.2003.12.010

that absolute quantitation is difficult to achieve in vivo

[1]

. Factors that need to be considered include the

relaxation parameters for each resonance, reference of
signal levels to a calibrated standard (for example,
internal water or an external phantom) and the
separation of overlapping peaks into individual compo-
nent signals. In general many of these factors can be
adequately considered with long examination times, but
since a limiting factor for a clinical examination is the
length of time the subject can be still, time is generally of
the essence.

Interpretation of a clinical MR spectrum provides

information about membrane turnover, cellular ener-
getics, neuronal function, selected neurotransmitter
activity and the fate of anaesthetics and of certain
drugs. The first set of 1.5–2.0 T horizontal magnets in
the early 1980s had a bore of approximately 20 cm, so it
was feasible to study cerebral metabolism only in
newborn infants

[2]

. It is now possible to obtain clinical

MR spectra using whole-body 1.5–3.0 T MR systems

[3]

,

either as an adjunct to a MRIexamination or as a
separate study. The availability of routine clinical MRI
systems with localized spectroscopy capabilities has
considerably widened the applicability of clinical
MRS. There is scope for obtaining spectra from a
number of different nuclei, generally on research MR
systems, which allows different aspects of in vivo
biochemistry to be studied.

The phosphorus nucleus has proved to be particularly

valuable in clinical in vivo MRS since resonances from
phosphocreatine (PCr), nucleoside triphosphate (NTP)
and inorganic phosphate (Pi) are readily observed. The
chemical shift of Pi is dependent on intracellular pH.
These parameters are of central importance in energy
metabolism and have been invaluable in defining the
sequence of events termed as ‘‘secondary energy failure’’
in infants with neonatal encephalopathy due to intra-
partum asphyxia. Whilst the

31

P MR spectrum was

often normal within the first few hours following
resuscitation, after 8–24 h a progressive decline in the
PCr/Pi ratio and an alkaline pH

i

have been observed

despite adequate oxygenation and circulation in the
infant

[4]

. The magnitude of the fall in PCr/Pi correlated

with the subsequent neurodevelopmental abnormality

[4]

and, for example the extent of brain alkalosis in the

first 2 weeks after birth has also been associated with the
severity of brain injury on MR imaging and neurode-
velopmental outcome at 1 year

[5]

. This pattern of

secondary energy failure after hypoxia-ischemia has
contributed to the development of neuroprotective
strategies

[6]

.

Of particular relevance to the study of membrane

synthesis are the composite peaks in the

31

P MR

spectrum labelled phosphomonoester (PME) and phos-
phodiester (PDE). The PME peak includes major
contributions from phosphocholine (PC), phosphoetha-

nolamine (PE), and l-phosphoserine, which are impor-
tant precursors of membrane phospholipids. However,
many other metabolites, including sugar phosphates,
can contribute to this region of the spectrum, and
separation of these different peaks cannot be achieved
with the present in vivo methodology. The PDE peak
includes

contributions

from

glycerphosphocholine

(GPC) as well as glycerophosphoethanolamine (GPE),
which are products of membrane breakdown.

The proton is the commonest nucleus in biological

systems and has the highest absolute sensitivity. In vivo
MRS studies are a technical challenge for a number of
reasons: the water signal is 10 000 times larger than
signals from metabolites of interest; the chemical shift
covers a narrow range, so peak overlap is a problem and
there are stringent demands on magnetic field homo-
geneity; interaction between nearby protons within a
molecule (spin–spin coupling) complicates the spectral
pattern; the scalp lipid signal is many times larger than
the metabolite signals, and produce broad features
which can overlap and partially obscure the sharper
resonances from smaller, more mobile species. Never-
theless resonances can be assigned to N-acetylaspartate
(NAA), an amino acid derivative thought to be located
in neurones, choline-containing compounds (Cho) such
as phosphoryl- and glycerophosphoryl-choline which
participate in membrane synthesis and breakdown, and
creatine and phosphocreatine (Cr). In addition more
complex peaks can be identified from protons in a range
of metabolites, if present, including lactate, inositols,
alanine, glutamine and glutamate.

2. Overview of clinical MRS findings

In vivo cerebral phosphorus-31 magnetic resonance

spectroscopy (

31

P MRS) is probably the best available

technique for investigating membrane phospholipid
metabolism. Here, a summary is given of the cerebral
proton and phosphorus magnetic resonance spectro-
scopy findings in relation to schizophrenia, dyslexia and
chronic fatigue syndrome.

2.1. Schizophrenia

The first published in vivo

31

P MRS study of brain-

membrane phospholipid metabolism was that of Pette-
grew and colleagues

[7]

. In this study of the dorsal

prefrontal cortex decreased levels of PME and Pi, and
increased levels of PDE were found in a group of 11
antipsychotic drug-na

.ıve, first-episode patients, com-

pared with a group of 10 matched normal controls. This
finding is entirely consistent with Horrobin’s membrane
phospholipid model of schizophrenia, as it points to a
reduction in neuronal membrane phospholipid bio-
synthesis and an increase in phospholipid breakdown.

ARTICLE IN PRESS

I.J. Cox, B.K. Puri / Prostaglandins, Leukotrienes and Essential Fatty Acids 70 (2004) 357–360

358

Furthermore, an offshoot of the same study also pointed
to the ability of in vivo

31

P MRS to detect presympto-

matic cerebral metabolism changes

[8]

.

Numerous 31-phosphorus and proton studies have

been carried out to investigate spectroscopy changes in
different brain regions in schizophrenia. The findings
from these studies have not always been consistent. For
example, some, but by no means not all, subsequent
in vivo

31

P MRS studies have replicated the above

findings of alterations in membrane phospholipid
metabolites in the prefrontal cortex

[9,10]

and temporal

lobe

[11]

. However, such studies of medicated patients

with chronic schizophrenia have not been consistent,
although the more consistent findings in the prefrontal
region again include decreased PME and increased PDE

[12]

. The range of different findings with

31

P MRS have

recently been reviewed by Pettegrew and colleagues

[12]

.

Interestingly, in the case report by our group

[13]

of

an antipsychotic-na

.ıve patient with schizophrenia who

responded clinically to ethyl eicosapentaenoic acid, it
was found that following such treatment his PME and
PDE levels fell. This suggests that in vivo

31

P MRS may

be useful in indexing the way in which eicosapentaenoic
acid alters neuronal membrane phospholipid metabo-
lism.

Proton MRS studies in schizophrenia have been

reviewed by Puri

[14]

. In general, one of the most

consistent findings of in vivo proton MRS studies of
schizophrenia is a reduction in NAA in the left
(dominant) temporal lobe

[14]

. As mentioned above,

NAA is believed to be located in neurones, so that this
finding is consistent with the temporal lobe atrophy that
has been described in post mortem studies of schizo-
phrenia.

2.2. Dyslexia

In a study of dyslexia, two different possibilities for

MRS findings were considered. First, that the MR
spectrum from dyslexia and schizophrenia would be
similar i.e. PDE would be elevated and PME reduced,
and second that the membrane abnormality in dyslexia
is different to that in schizophrenia i.e. there would be
reduced incorporation of phospholipids into membranes
in dyslexia and therefore an increase in PME would be
expected. In a

31

P MRS study of 12 subjects who had

previously been identified by educational psychologists
as dyslexic were compared with 10 age- and sex-matched
normal controls

[15]

. The relative level of phosphomo-

noesters (PMEs), PME/bNTP and PME/PDE ratios was
increased across the brain in the dyslexic adults
compared with the matched controls in. The relative
level of PDE and PDE/ATP did not differ significantly
between dyslexic and control groups. No other spectral
parameters differed significantly between the groups.
Given that there was already indirect evidence for a

phospholipid abnormality in dyslexia, these findings
provided additional support for this hypothesis and
could be interpreted as difficulties in either the biosynth-
esis of PME and/or their incorporation into membranes
this could lead to an excess of PME precursors.

These studies were complemented by a comparison of

PME and PDE levels with erythrocyte membrane fatty
acid concentrations

[16]

. Levels of PDE were signifi-

cantly correlated with reduced concentrations of the
highly unsaturated fatty acids docosahexaenoic acid
(DHA) (r ¼ 0:68; P

o0:05) and eicosapentaenoic acid

(EPA) (r ¼ 0:78; P

o0:02). No significant correlations

were found between peripheral concentrations of any
highly unsaturated fatty acids and PME levels, nor
between their essential fatty acid precursors and either
PDE or PME levels. Other 31-phosphorus metabolites
also showed no significant correlations with the blood
fatty acid measures. The correlations between central
measures of PDE and peripheral measures of DHA and
EPA provide validation of cerebral

31

P MRS as a non-

invasive technique for the study of membrane phospho-
lipid metabolism in vivo.

The first published proton MRS study consisted of 14

dyslexic men and 15 age-matched control men

[17]

. The

main positive finding was a decrease in Cho/NAA in the
left temporo-parietal lobe, which was considered to
represent a decrease in Cho in this cerebral region. The
authors interpreted this to be indicative of a decrease in
total cell membranes in this brain region in develop-
mental dyslexia, without a concomitant decrease in the
total neuronal volume. However, an alternative expla-
nation of these findings has been put forward, which
relates to the role of essential fatty acids in dyslexia

[18]

.

Since a major component of Cho consists of metabolites
of phosphatidylcholine, which is one of the major
phospholipids of mammalian cell membranes, then
these findings are consistent with reduced catabolism
of phospholipids in the left temporo-parietal lobe in
dyslexia. In turn, this is compatible with the findings of
our group of increased PME on

31

P MRS.

2.3. Chronic fatigue syndrome

The first systematic study of chronic fatigue syndrome

using magnetic resonance spectroscopy was published by
our group and consisted of a proton MRS study in eight
patients with chronic fatigue syndrome and eight age-
and sex-matched healthy control subjects

[19]

. The mean

ratio of Cho/Cr in the occipital cortex in the chronic
fatigue syndrome group was significantly higher than in
the controls. No other metabolite ratios were significantly
different between the two groups in either the frontal or
occipital cortex (the two regions of the brain investi-
gated). In addition, there was a loss of the normal spatial
variation of Cho in chronic fatigue syndrome. Since
increased choline levels are associated with abnormal

ARTICLE IN PRESS

I.J. Cox, B.K. Puri / Prostaglandins, Leukotrienes and Essential Fatty Acids 70 (2004) 357–360

359

membrane phospholipid metabolism, specifically relating
to phospholipid head groups

[20]

, our results suggest that

there may be an abnormality of phospholipid metabolism
in the brain in chronic fatigue syndrome.

In the year following the publication of the above

in vivo proton MRS study of chronic fatigue syndrome,
another such study, also involving eight patients and
eight matched control subjects and specifically investi-
gating the left basal ganglia, has been published by
another, independent, group

[21]

. The key finding from

this study was also an increase in Cho in the chronic
fatigue syndrome patients, so that these results are
consistent with the findings of our group.

3. Discussion

In this paper we have seen that the non-invasive

technique of in vivo MRS provides a powerful
investigative tool that allows the living chemistry of
the brain to be studied. Its application to schizophrenia,
dyslexia and chronic fatigue syndrome has shed light on
changes in phospholipid metabolism that occur in these
disorders of otherwise unknown aetiology. It also
provides a method of following the development of
neonatal encephalopathy following hypoxia-ischaemia.
The applications of MRS as both a research tool and as
a diagnostic investigation in phospholipid spectrum
disorders in psychiatry and neurology are set to grow in
the years ahead.

References

[1] E.B. Cady, Quantitative combined phosphorus and proton

PRESS of the brains of newborn human infants, Magn. Reson.
Med. 33 (1995) 557–563.

[2] E.B. Cady, A.M. Costello, M.J. Dawson, D.T. Delpy, P.L. Hope,

E.O. Reynolds, P.S. Tofts, D.R. Wilkie, Non-invasive investiga-
tion of cerebral metabolism in newborn infants by phosphorus
nuclear magnetic resonance spectroscopy, Lancet 1(8333) (1983)
1059–1062.

[3] I.J. Cox, Development and applications of in vivo clinical

magnetic resonance spectroscopy, Prog. Biophys. Mol. Biol. 65
(1996) 45–81.

[4] S.C. Roth, A.D. Edwards, E.B. Cady, D.T. Delpy, J.S. Wyatt, D.

Azzopardi, J. Baudin, J. Townsend, A.L. Stewart, E.O.R.
Reynolds, Relation between cerebral oxidative metabolism
following birth asphyxia and neurodevelopmental outcome and
brain growth at one year, Dev. Med. Child Neurol. 34 (1992)
285–295.

[5] N.J. Robertson, F.M. Cowan, I.J. Cox, A.D. Edwards, Brain

alkaline intracellular pH after neonatal encephalopathy, Ann.
Neurol. 52 (2002) 732–742.

[6] D.L. Taylor, A.D. Edwards, H. Mehmet, Oxidative metabolism,

apoptosis and perinatal brain injury, Brain Pathol. 9 (1999)
93–117.

[7] J.W. Pettegrew, M.S. Keshavan, K. Panchalingam, S. Strychor,

D.B. Kaplan, M.G. Tretta, M. Allen, Alterations in brain high-
energy phosphate and membrane phospholipid metabolism in

first-episode, drug-naive schizophrenics. A pilot study of the
dorsal prefrontal cortex by in vivo phosphorus 31 nuclear
magnetic resonance spectroscopy, Arch. Gen Psychiatry 48
(1991) 563–568.

[8] M.S. Keshavan, J.W. Pettegrew, K.S. Panchalingam, D. Kaplan,

E. Bozik, Phosphorus 31 magnetic resonance spectroscopy detects
altered brain metabolism before onset of schizophrenia, Arch.
Gen Psychiatry 48 (1991) 1112–1113.

[9] P. Williamson, D. Drost, J. Stanley, T. Carr, S. Morrison, H.

Merskey, Localized phosphorus 31 magnetic resonance spectro-
scopy in chronic schizophrenic patients and normal controls,
Arch. Gen Psychiatry 48 (1991) 578.

[10] J.A. Stanley, P.C. Williamson, D.J. Drost, T.J. Carr, R.J. Rylett,

A. Malla, R.T. Thompson, An in vivo study of the prefrontal
cortex of schizophrenic patients at different stages of illness via
phosphorus magnetic resonance spectroscopy, Arch. Gen Psy-
chiatry 52 (1995) 399–406.

[11] H. Fukuzako, T. Fukuzako, T. Hashiguchi, S. Kodama, M.

Takigawa, T. Fujimoto, Changes in levels of phosphorus
metabolites in temporal lobes of drug-naive schizophrenic
patients, Am. J. Psychiatry 156 (1999) 1205–1208.

[12] J.W. Pettegrew, M.S. Keshavan, J.A. Stanley, R.J. McClure, C.R.

Johnson, K. Panchalingam, Magnetic resonance spectroscopy in
the assessment of phospholipid metabolism in schizophrenia and
other psychiatric disorders, in: M. Peet, I. Glen, D.F. Horrobin
(Eds.), Phospholipid Spectrum Disorders in Psychiatry and
Neurology, 2nd Edition, Marius Press, Carnforth, Lancashire,
2003, pp. 239–255.

[13] B.K. Puri, A.J. Richardson, D.F. Horrobin, T. Easton, N. Saeed,

A. Oatridge, J.V. Hajnal, G.M. Bydder, Eicosapentaenoic acid
treatment in schizophrenia associated with symptom remission,
normalisation of blood fatty acids, reduced neuronal membrane
phospholipid turnover and structural brain changes, Int. J. Clin.
Pract. 54 (2000) 57–63.

[14] B.K. Puri, MRIand MRS in neuropsychiatry, in: I.R. Young,

D.M. Grant, R.K. Harris (Eds.), Methods in Biomedical
Magnetic Resonance Imaging and Spectroscopy, Wiley, New
York, 2000, pp. 1135–1143.

[15] A.J. Richardson, I.J. Cox, J. Sargentoni, B.K. Puri, Abnormal

cerebral phospholipid metabolism in dyslexia indicated by
phosphorus-31 magnetic resonance spectroscopy, NMR Biomed.
10 (1997) 309–314.

[16] A.J. Richardson, S.J. Allen, J.V. Hajnal, I.J. Cox, T. Easton, B.K.

Puri, Associations between central and peripheral measures of
phospholipid breakdown revealed by cerebral 31-phosphorus
magnetic resonance spectroscopy and fatty acid composition of
erythrocyte membranes, Prog. Neuropsychopharmacol. Biol.
Psychiatry 25 (2001) 1513–1521.

[17] C. Rae, M.A. Lee, R.M. Dixon, et al., Metabolic abnormalities in

developmental dyslexia detected by

1

H magnetic resonance

spectroscopy, Lancet 351 (1998) 1849–1852.

[18] B.K. Puri, A.J. Richardson, Brain phospholipid metabolism in

dyslexia assessed by magnetic resonance spectroscopy, in: M.
Peet, I. Glen, D.F. Horrobin (Eds.), Phospholipid Spectrum
Disorders in Psychiatry and Neurology, 2nd Edition, Marius
Press, Carnforth Lancashire, 2003, pp. 501–508.

[19] B.K. Puri, S.J. Counsell, R. Zaman, J. Main, A.G. Collins, J.V.

Hajnal, N.J. Davey, Relative increase in choline in the occipital
cortex in chronic fatigue syndrome, Acta Psychiatr. Scand. 106
(2002) 224–226.

[20] J. Ruiz-Cabello, J.S. Cohen, Phospholipid metabolites as

indicators of cancer cell function, NMR Biomed. 5 (1992)
226–233.

[21] A. Chaudhuri, B.R. Condon, J.W. Gow, D. Brennan, D.M.

Hadley, Proton magnetic resonance spectroscopy of basal ganglia
in chronic fatigue syndrome, Neuroreport 14 (2003) 225–228.

ARTICLE IN PRESS

I.J. Cox, B.K. Puri / Prostaglandins, Leukotrienes and Essential Fatty Acids 70 (2004) 357–360

360