Editorials Paul
A. Bottomley,
Proton Hepatic
T
PhD
MR Spectroscopy Encephalopathy?’
HE 1991
meeting of the Society of Magnetic Resonance in Medicine in San Francisco last August will be remembered for the burgeoning of studies of water-suppressed proton (hydrogen-1) magnetic resonance (MR) spectroscopy in groups of patients representing a broad spectrum of brain disease. Most findings were documented as changes in the relative levels of one or more of four commonly detectable chemical moieties: lactate (La), N-acetylaspartate (NAA), the total creatme pool (Cr), and choline (Cho). Most findings in diseases that involve acute destruction
of brain
tissue
were
charac-
terized by a decrease in NAA level, an increase in Cho level, and perhaps an elevation in La (Arnold D, oral communication, 1991). Physiologic interpretations beyond that of neuronal loss (which NAA levels may index) are cornplicated by uncertainties about the precise biochemical role of NAA and now-in chronic stroke-by possible elevations in metabolically active La derived from infiltrating rnacrophages, as distinct from elevated La in potentially viable brain tissue (1). Moreover, what remains to be seen is just how the clinical research findings can be of benefit to patients, especially in the context of existing diagnostic tools. The article by Kreis et al on the detection of chronic hepatic encephalopathy in this issue of Radiology (2) breaks out of the mold of these studies and advances clinical H-i brain spectroscopy in three important ways: First, in measuring relative glutamate and glutamine, rnyoinositol, and cerebral glucose (noted in passing) levels in
Index MR,
terms:
761.794 copy Radiology
r
Brain,
diseases,
10.599
#{149} #{149} #{149}
10.1214
Liver,
Editorials
Magnetic
1992;
resonance
#{149} Brain,
cirrhosis,
(MR),
spectros-
182:6-7
From the GE Research
and Development
Center, NMR Building, Rm 130, River Rd. Schenectady, NY 12309. Received and accepted September 20, 1991. Address reprint requests the C
RSNA,
See also this issue.
6
to
author.
1992 the
article
by Kreis
et al (pp
19-27)
in
for
Diagnosing
a respectably sized group of 58 patients, the authors have doubled the number of biologically important moieties that can be detected and quantified by means of localized H-i spectroscopy with a clinical imager. Their demonstration quashes any doubts about the potential for routine measurement of these compounds in clinical populations (3) and marks the transition from technique development (4-6) to clinical research phases for these particular moieties, as heralded by recent letters to the Lancet (7,8). The detection of glutamate with spectroscopy is important because the metabolic turnover of MR-labeled glucose or other substrates to glutamate can provide a measure of flux through the tncarboxylic acid (TCA) cycle, linking it to energy metabolism (9) and diseases thereof, and because it is a putative neurotransmitter (iO,li). Glutamate denves from transamination of a-ketoglutanate, a TCA cycle intermediary, and from glutamine via glutaminase, which releases ammonia. The ammonia leaves the brain by diffusion or is used in the conversion of glutamate back to glutamine, which requires energy and glutamine synthetase (iO,ii). Glucose is also important as an energy index, and spectroscopy might be of some value as a noninvasive monitor in patients with diabetes (2,8). Second, the authors have identified and meticulously documented a possible first clinical application for millimole-per-liter-level H-i spectroscopy: that of the diagnosis and monitoring of chronic hepatic encephalopathy. This disease is most commonly associated with chronic liver disease and manifests as episodes of neurologic dysfunction. Its cause appears to be the accumulation of toxins in the brain, especially ammonia, resulting from portal venous hypertension that leads to shunting of products of intestinal origin into the systemic circulation, and possibly from depression of liver function (ii). With an elevated ammonia level, brain glutamine levels increase via the glutamine synthetase reaction, which would presumably also tend to deplete glutamate (ii). Elevated glutamine and a-ketoglutarate can be detected in the cerebrospinal fluid (10).
Kreis et al (2) have now shown that these glutamine elevations-or at least elevations in the integrals of signals over two chemical shift ranges (denoted Al and A2) that correspond to glutamine and glutamate-are directly detectable in the brains of patients with hepatic encephalopathy. They used a fairly conventional water-suppressed, stimulated-echo, localized H-i spectroscopy
research
sequence
and
a standard
imaging head coil. The increase, measured relative to the Cr resonance, was, on average, a statistically significant 65% for Al and 35% for A2 compared with levels in control subjects. Increases were observed in i4 of iS patients with hepatic encephalopathy. Kreis et al discovered other abnormalities in hepatic encephalopathy as well. Brain myoinositol concentration was significantly reduced to 46% of that in control subjects and Cho to 77%. To test specificity, the authors studied six additional control groups: patients with mild to severe liver disease but without clinical evidence of hepatic encephalopathy, young healthy subjects, aged healthy subjects with atrophy, patients with diabetes, patients with uremia, and patients receiving furosemide, all of whom might have yielded similar abnormalities in their H-i brain spectra. However, these other groups showed no significant elevations in both Al and A2, except for two of four patients with low myoinositol levels from the group of patients with liver disease but no overt hepatic encephalopathy; the authors designated these four exceptions as cases of “ preclinical” hepatic encephalopathy on the basis of their similar spectral patterns of myoinositol, glutamine, and glutamate signals. Moreover, measurements of Al and myoinositol levels, when combined, were sufficient to provide complete discnimination of patients with hepatic encephalopathy from control subjects and from patients with liver disease but without hepatic encephalopathy, as long as the ,, preclinical” patients were reclassified as in fact having hepatic encephalopathy. While it will be important to see whether the patients classified as haying preclinical hepatic encephalopathy and liver disease do in fact progress to,
or are subsequently diagnosed with, hepatic encephalopathy and do not represent false-positive test results, the findings presented by Kreis et al certainly hold promise for the use of H-i spectroscopy in the diagnosis of this disease. The next questions to be answered will be the reproducibility of the results at different sites, how well diagnostic specificity and prediction hold up with larger patient numbers, and whether spectroscopy is an efficacious tool for making the diagnosis compared with existing methods, such as those used by Kreis et al in identifying their patient population. The disorder is treatable with therapies designed to remove excess ammonia, and indeed all patients with hepatic encephalopathy studied by the authors were taking medications for it. Kreis et al compared the effects of lactulose versus neomycin treatment on brain spectra from patients with hepatic encephalopathy, observing some difference in brain Al levels but not in A2, myoinositol, or choline levels. It is hoped that an opportunity will arise to answer the question of possible effects of medication by comparing medicated and unmedicated patients with clinical symptoms of hepatic encephalopathy, and to determine whether successful therapeutic intervention results in abolition of the spectral abnormalities. Third, the study represents an exemplary piece of quantification in a field where quantification will be key to clinical success (12). Seldom is more than a single control group included in a clinical spectroscopy report, let alone seven. Control groups ranged from three to iS subjects, and the significant H-i spectral abnormalities noted and discussed in just the control groups of patients with liver disease but without hepatic en-
Volume
182
#{149} Number
1
cephalopathy, patients receiving furosemide, aged subjects with atrophy, and patients with diabetes would probably fare well on their own as food for four articles, perhaps. To their credit, the authors have resisted the temptation to subdivide and multiply, being intent on making their point, a decision that has resulted in an article with impact. Their method of quantitative analysis is disclosed in detail, and random errors in the quantification process and possible artifacts from relaxation effects, localization errors, and foreign compounds or other metabolites are discussed, documented, or accounted for. They have adopted an interesting, “grass roots” approach to spectral quantification: They mixed standard solutions, acquired spectra from them, fitted the spectra to those of the brain by adjusting the amplitudes of standard spectra (thereby substantially accounting for the entirety of the spectral range of interest), and then subtracted the control spectra from patients to measure any differences. The reliability with which glutamine contributions to the indexes Al and A2 can be distinguished from those of glutamate by this means does, however, remain a little unclear, and this point bears some relevance to the question of whether decreases in glutamate level caused by its conversion to glutamine via glutamine synthetase are reflected in the Al and A2 measurements or whether glutamate concentrations are somewhat buffered as glutamine levels rise. It is to be hoped that the promise of this substantial work by Kreis et al will fulfill their expectations and that a role for H-i spectroscopy in the diagnosis of hepatic encephalopathy will be found, perhaps heralding the arrival of a routine application of in vivo spectroscopy in clinical medicine. U
References Petroff OAC, Graham GD, Blamire AM, et al. ‘H spectroscopic imaging of stroke in man: histopathology correlates of spectral changes (abstr). In: Book of abstracts: Society of Magnetic Resonance in Medicine 1991. Berkeley, Calif: Society of Magnetic Resonance in Medicine, 1991; 227. 2.
Kreis
R, Ross
BD,
Farrow
Metabolic disorders hepatic encephalopathy MR spectroscopy. 3.
27. Bottomley
tool
Radiology
PA.
troscopy
NA,
Human
detected with H-i 1992; 182:19NMR
medicine:
probe?
Z.
in chronic
in vivo
in diagnostic
or research
Ackerman
of the brain
spec-
clinical
Radiology
1989;
170:1-15. 4.
Behar
KL,
den
HollanderJA,
et al. High resolution netic resonance study in vivo.
Proc
Natl
Stromski
‘H nuclear of cerebral
Acad
Sci
ME,
maghypoxia
U S A 1983;
80:
4945-4948. 5.
Rothman
DL,
Shulman
RG.
resonance
Behar
KL,
Hetherington
Homonuclear
difference
HP,
‘H double
spectroscopy
of the
6.
brain in vivo. Proc Nati Acad Sci U S A 1984; 81:6330-6334. Frahm J, Merboldt KD, Hanicke W. Localized proton spectroscopy using stimulated echoes. J Magn Reson 1987; 72:502-508.
7.
Kreis
8.
9.
R, Farrow
N, Ross
BD.
Diagnosis
of
hepatic encephalopathy by proton magnetic resonance spectroscopy (letter). Lancet 1990; 336:635-636. Bruhn H, Michaelis T, Merboldt KD, Hanicke W, Gyngell ML, Frahm J. Monitoring cerebral glucose in diabetics by proton MRS (letter). Lancet 1991; 337:745-746. Chance EM, Seeholzer SH, Kobayashi K, Williamson JR. Mathematical analysis of isotope
labeling
in
the
citric
acid
cycle
with
to ‘3C NMR studies in perrat hearts. J Biol Chem 1983; 258:
applications
fused
13785-13794.
10.
Duffy TE, Plum F. Seizures, coma, and major metabolic encephalopathies. In: Siegel GJ, Albers RW, Agranoff BW, Katzman R, eds.
11.
12.
Basic
neurochemistry.
3rd
ed.
Bos-
ton: Little, Brown, 1981; 707-710. Pulsinelli WA, Cooper AJL. Metabolic encephalopathies and coma. In: Siegel GJ, Agranoff BW, Albers RW, Molinoff PB, eds. Basic neurochemistry. 4th ed. New York: Raven, 1989; 324, 767-772. Bottomley PA. The trouble with spectroscopy
papers.
Radiology
1991;
181:344-350.
Radiology
#{149} 7
A. Bottomley,
Proton Hepatic
T
PhD
MR Spectroscopy Encephalopathy?’
HE 1991
meeting of the Society of Magnetic Resonance in Medicine in San Francisco last August will be remembered for the burgeoning of studies of water-suppressed proton (hydrogen-1) magnetic resonance (MR) spectroscopy in groups of patients representing a broad spectrum of brain disease. Most findings were documented as changes in the relative levels of one or more of four commonly detectable chemical moieties: lactate (La), N-acetylaspartate (NAA), the total creatme pool (Cr), and choline (Cho). Most findings in diseases that involve acute destruction
of brain
tissue
were
charac-
terized by a decrease in NAA level, an increase in Cho level, and perhaps an elevation in La (Arnold D, oral communication, 1991). Physiologic interpretations beyond that of neuronal loss (which NAA levels may index) are cornplicated by uncertainties about the precise biochemical role of NAA and now-in chronic stroke-by possible elevations in metabolically active La derived from infiltrating rnacrophages, as distinct from elevated La in potentially viable brain tissue (1). Moreover, what remains to be seen is just how the clinical research findings can be of benefit to patients, especially in the context of existing diagnostic tools. The article by Kreis et al on the detection of chronic hepatic encephalopathy in this issue of Radiology (2) breaks out of the mold of these studies and advances clinical H-i brain spectroscopy in three important ways: First, in measuring relative glutamate and glutamine, rnyoinositol, and cerebral glucose (noted in passing) levels in
Index MR,
terms:
761.794 copy Radiology
r
Brain,
diseases,
10.599
#{149} #{149} #{149}
10.1214
Liver,
Editorials
Magnetic
1992;
resonance
#{149} Brain,
cirrhosis,
(MR),
spectros-
182:6-7
From the GE Research
and Development
Center, NMR Building, Rm 130, River Rd. Schenectady, NY 12309. Received and accepted September 20, 1991. Address reprint requests the C
RSNA,
See also this issue.
6
to
author.
1992 the
article
by Kreis
et al (pp
19-27)
in
for
Diagnosing
a respectably sized group of 58 patients, the authors have doubled the number of biologically important moieties that can be detected and quantified by means of localized H-i spectroscopy with a clinical imager. Their demonstration quashes any doubts about the potential for routine measurement of these compounds in clinical populations (3) and marks the transition from technique development (4-6) to clinical research phases for these particular moieties, as heralded by recent letters to the Lancet (7,8). The detection of glutamate with spectroscopy is important because the metabolic turnover of MR-labeled glucose or other substrates to glutamate can provide a measure of flux through the tncarboxylic acid (TCA) cycle, linking it to energy metabolism (9) and diseases thereof, and because it is a putative neurotransmitter (iO,li). Glutamate denves from transamination of a-ketoglutanate, a TCA cycle intermediary, and from glutamine via glutaminase, which releases ammonia. The ammonia leaves the brain by diffusion or is used in the conversion of glutamate back to glutamine, which requires energy and glutamine synthetase (iO,ii). Glucose is also important as an energy index, and spectroscopy might be of some value as a noninvasive monitor in patients with diabetes (2,8). Second, the authors have identified and meticulously documented a possible first clinical application for millimole-per-liter-level H-i spectroscopy: that of the diagnosis and monitoring of chronic hepatic encephalopathy. This disease is most commonly associated with chronic liver disease and manifests as episodes of neurologic dysfunction. Its cause appears to be the accumulation of toxins in the brain, especially ammonia, resulting from portal venous hypertension that leads to shunting of products of intestinal origin into the systemic circulation, and possibly from depression of liver function (ii). With an elevated ammonia level, brain glutamine levels increase via the glutamine synthetase reaction, which would presumably also tend to deplete glutamate (ii). Elevated glutamine and a-ketoglutarate can be detected in the cerebrospinal fluid (10).
Kreis et al (2) have now shown that these glutamine elevations-or at least elevations in the integrals of signals over two chemical shift ranges (denoted Al and A2) that correspond to glutamine and glutamate-are directly detectable in the brains of patients with hepatic encephalopathy. They used a fairly conventional water-suppressed, stimulated-echo, localized H-i spectroscopy
research
sequence
and
a standard
imaging head coil. The increase, measured relative to the Cr resonance, was, on average, a statistically significant 65% for Al and 35% for A2 compared with levels in control subjects. Increases were observed in i4 of iS patients with hepatic encephalopathy. Kreis et al discovered other abnormalities in hepatic encephalopathy as well. Brain myoinositol concentration was significantly reduced to 46% of that in control subjects and Cho to 77%. To test specificity, the authors studied six additional control groups: patients with mild to severe liver disease but without clinical evidence of hepatic encephalopathy, young healthy subjects, aged healthy subjects with atrophy, patients with diabetes, patients with uremia, and patients receiving furosemide, all of whom might have yielded similar abnormalities in their H-i brain spectra. However, these other groups showed no significant elevations in both Al and A2, except for two of four patients with low myoinositol levels from the group of patients with liver disease but no overt hepatic encephalopathy; the authors designated these four exceptions as cases of “ preclinical” hepatic encephalopathy on the basis of their similar spectral patterns of myoinositol, glutamine, and glutamate signals. Moreover, measurements of Al and myoinositol levels, when combined, were sufficient to provide complete discnimination of patients with hepatic encephalopathy from control subjects and from patients with liver disease but without hepatic encephalopathy, as long as the ,, preclinical” patients were reclassified as in fact having hepatic encephalopathy. While it will be important to see whether the patients classified as haying preclinical hepatic encephalopathy and liver disease do in fact progress to,
or are subsequently diagnosed with, hepatic encephalopathy and do not represent false-positive test results, the findings presented by Kreis et al certainly hold promise for the use of H-i spectroscopy in the diagnosis of this disease. The next questions to be answered will be the reproducibility of the results at different sites, how well diagnostic specificity and prediction hold up with larger patient numbers, and whether spectroscopy is an efficacious tool for making the diagnosis compared with existing methods, such as those used by Kreis et al in identifying their patient population. The disorder is treatable with therapies designed to remove excess ammonia, and indeed all patients with hepatic encephalopathy studied by the authors were taking medications for it. Kreis et al compared the effects of lactulose versus neomycin treatment on brain spectra from patients with hepatic encephalopathy, observing some difference in brain Al levels but not in A2, myoinositol, or choline levels. It is hoped that an opportunity will arise to answer the question of possible effects of medication by comparing medicated and unmedicated patients with clinical symptoms of hepatic encephalopathy, and to determine whether successful therapeutic intervention results in abolition of the spectral abnormalities. Third, the study represents an exemplary piece of quantification in a field where quantification will be key to clinical success (12). Seldom is more than a single control group included in a clinical spectroscopy report, let alone seven. Control groups ranged from three to iS subjects, and the significant H-i spectral abnormalities noted and discussed in just the control groups of patients with liver disease but without hepatic en-
Volume
182
#{149} Number
1
cephalopathy, patients receiving furosemide, aged subjects with atrophy, and patients with diabetes would probably fare well on their own as food for four articles, perhaps. To their credit, the authors have resisted the temptation to subdivide and multiply, being intent on making their point, a decision that has resulted in an article with impact. Their method of quantitative analysis is disclosed in detail, and random errors in the quantification process and possible artifacts from relaxation effects, localization errors, and foreign compounds or other metabolites are discussed, documented, or accounted for. They have adopted an interesting, “grass roots” approach to spectral quantification: They mixed standard solutions, acquired spectra from them, fitted the spectra to those of the brain by adjusting the amplitudes of standard spectra (thereby substantially accounting for the entirety of the spectral range of interest), and then subtracted the control spectra from patients to measure any differences. The reliability with which glutamine contributions to the indexes Al and A2 can be distinguished from those of glutamate by this means does, however, remain a little unclear, and this point bears some relevance to the question of whether decreases in glutamate level caused by its conversion to glutamine via glutamine synthetase are reflected in the Al and A2 measurements or whether glutamate concentrations are somewhat buffered as glutamine levels rise. It is to be hoped that the promise of this substantial work by Kreis et al will fulfill their expectations and that a role for H-i spectroscopy in the diagnosis of hepatic encephalopathy will be found, perhaps heralding the arrival of a routine application of in vivo spectroscopy in clinical medicine. U
References Petroff OAC, Graham GD, Blamire AM, et al. ‘H spectroscopic imaging of stroke in man: histopathology correlates of spectral changes (abstr). In: Book of abstracts: Society of Magnetic Resonance in Medicine 1991. Berkeley, Calif: Society of Magnetic Resonance in Medicine, 1991; 227. 2.
Kreis
R, Ross
BD,
Farrow
Metabolic disorders hepatic encephalopathy MR spectroscopy. 3.
27. Bottomley
tool
Radiology
PA.
troscopy
NA,
Human
detected with H-i 1992; 182:19NMR
medicine:
probe?
Z.
in chronic
in vivo
in diagnostic
or research
Ackerman
of the brain
spec-
clinical
Radiology
1989;
170:1-15. 4.
Behar
KL,
den
HollanderJA,
et al. High resolution netic resonance study in vivo.
Proc
Natl
Stromski
‘H nuclear of cerebral
Acad
Sci
ME,
maghypoxia
U S A 1983;
80:
4945-4948. 5.
Rothman
DL,
Shulman
RG.
resonance
Behar
KL,
Hetherington
Homonuclear
difference
HP,
‘H double
spectroscopy
of the
6.
brain in vivo. Proc Nati Acad Sci U S A 1984; 81:6330-6334. Frahm J, Merboldt KD, Hanicke W. Localized proton spectroscopy using stimulated echoes. J Magn Reson 1987; 72:502-508.
7.
Kreis
8.
9.
R, Farrow
N, Ross
BD.
Diagnosis
of
hepatic encephalopathy by proton magnetic resonance spectroscopy (letter). Lancet 1990; 336:635-636. Bruhn H, Michaelis T, Merboldt KD, Hanicke W, Gyngell ML, Frahm J. Monitoring cerebral glucose in diabetics by proton MRS (letter). Lancet 1991; 337:745-746. Chance EM, Seeholzer SH, Kobayashi K, Williamson JR. Mathematical analysis of isotope
labeling
in
the
citric
acid
cycle
with
to ‘3C NMR studies in perrat hearts. J Biol Chem 1983; 258:
applications
fused
13785-13794.
10.
Duffy TE, Plum F. Seizures, coma, and major metabolic encephalopathies. In: Siegel GJ, Albers RW, Agranoff BW, Katzman R, eds.
11.
12.
Basic
neurochemistry.
3rd
ed.
Bos-
ton: Little, Brown, 1981; 707-710. Pulsinelli WA, Cooper AJL. Metabolic encephalopathies and coma. In: Siegel GJ, Agranoff BW, Albers RW, Molinoff PB, eds. Basic neurochemistry. 4th ed. New York: Raven, 1989; 324, 767-772. Bottomley PA. The trouble with spectroscopy
papers.
Radiology
1991;
181:344-350.
Radiology
#{149} 7
