12
Metabolic Brain Disease, Vol. 17, No. 1, March 2002 ( C 2002) Correlations between Cerebral Glucose Metabolism and Neuropsychological Test Performance in Nonalcoholic Cirrhotics Alan H. Lockwood, 1,5 Karin Weissenborn, 2 Martin Bokemeyer, 2 U. Tietge, 3 and Wolfgang Burchert 4 Received January 29, 2001; accepted May 15, 2001 Many cirrhotics have abnormal neuropsychological test scores. To define the anatomical– physiological basis for encephalopathy in nonalcoholic cirrhotics, we performed resting- state fluorodeoxyglucose positron emission tomographic scans and administered a neu- ropsychological test battery to 18 patients and 10 controls. Statistical parametric mapping correlated changes in regional glucose metabolism with performance on the individual tests and a composite battery score. In patients without overt encephalopathy, poor performance correlated with reductions in metabolism in the anterior cingulate. In all patients, poor per- formance on the battery was positively correlated ( p < 0.001) with glucose metabolism in bifrontal and biparietal regions of the cerebral cortex and negatively correlated with metabolism in hippocampal, lingual, and fusiform gyri and the posterior putamen. Similar patterns of abnormal metabolism were found when comparing the patients to 10 controls. Metabolic abnormalities in the anterior attention system and association cortices mediat- ing executive and integrative function form the pathophysiological basis for mild hepatic encephalopathy. Key words: Cirrhosis; hepatic encephalopathy; tomography; emission computed; neuropsychological tests. INTRODUCTION For over two decades, clinicians have known that a substantial number of patients with cirrhosis of the liver who appeared to have a normal mental status, on the basis of a conventional bedside examination, often scored in the impaired range on certain neu- ropsychological tests (Gilberstadt et al., 1980; Groeneweg et al., 2000; Rikkers et al., 1978; Schomerus and Schreiegg, 1993). However, little is known of the anatomical–physiological basis of hepatic encephalopathy (HE). Earlier PET studies showed that the patterns of cere- bral blood flow (CBF) and the cerebral metabolic rate for glucose (CMRglucose) were abnormal in cirrhotics (Lockwood et al., 1991). The methods employed did not identify 1 Center for Positron Emission Tomography, Veterans Affairs Western New York Healthcare System, and De- partments of Neurology and Nuclear Medicine, State University of New York, University at Buffalo, Buffalo, New York. 2 Department of Neurology, Medizinische Hochschule, Hannover, Germany. 3 Departments of Gastroenterology and Hepatology, Medizinische Hochschule, Hannover, Germany. 4 Department of Nuclear Medicine, Medizinische Hochschule, Hannover, Germany. 5 To whom correspondence should be addressed at Center for Positron Emission Tomography (115P), Veterans Affairs Western New York Healthcare System, 3495 Bailey Avenue, Buffalo, New York 14215. E-mail: alan@ petnet.buffalo.edu 29 0885-7490/02/0300-0029/0 C 2002 Plenum Publishing Corporation

Correlations Between Cerebral Glucose Metabolism and Neuropsychological Test Performance in Nonalcoholic Cirrhotics

Embed Size (px)

Citation preview

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

Metabolic Brain Disease, Vol. 17, No. 1, March 2002 (C© 2002)

Correlations between Cerebral Glucose Metabolismand Neuropsychological Test Performancein Nonalcoholic Cirrhotics

Alan H. Lockwood,1,5 Karin Weissenborn,2 Martin Bokemeyer,2 U. Tietge,3

and Wolfgang Burchert4

Received January 29, 2001; accepted May 15, 2001

Many cirrhotics have abnormal neuropsychological test scores. To define the anatomical–physiological basis for encephalopathy in nonalcoholic cirrhotics, we performed resting-state fluorodeoxyglucose positron emission tomographic scans and administered a neu-ropsychological test battery to 18 patients and 10 controls. Statistical parametric mappingcorrelated changes in regional glucose metabolism with performance on the individual testsand a composite battery score. In patients without overt encephalopathy, poor performancecorrelated with reductions in metabolism in the anterior cingulate. In all patients, poor per-formance on the battery was positively correlated (p < 0.001) with glucose metabolismin bifrontal and biparietal regions of the cerebral cortex and negatively correlated withmetabolism in hippocampal, lingual, and fusiform gyri and the posterior putamen. Similarpatterns of abnormal metabolism were found when comparing the patients to 10 controls.Metabolic abnormalities in the anterior attention system and association cortices mediat-ing executive and integrative function form the pathophysiological basis for mild hepaticencephalopathy.

Key words: Cirrhosis; hepatic encephalopathy; tomography; emission computed; neuropsychological tests.

INTRODUCTION

For over two decades, clinicians have known that a substantial number of patientswith cirrhosis of the liver who appeared to have a normal mental status, on the basis ofa conventional bedside examination, often scored in the impaired range on certain neu-ropsychological tests (Gilberstadtet al., 1980; Groeneweget al., 2000; Rikkerset al., 1978;Schomerus and Schreiegg, 1993). However, little is known of the anatomical–physiologicalbasis of hepatic encephalopathy (HE). Earlier PET studies showed that the patterns of cere-bral blood flow (CBF) and the cerebral metabolic rate for glucose (CMRglucose) wereabnormal in cirrhotics (Lockwoodet al., 1991). The methods employed did not identify

1Center for Positron Emission Tomography, Veterans Affairs Western New York Healthcare System, and De-partments of Neurology and Nuclear Medicine, State University of New York, University at Buffalo, Buffalo,New York.

2Department of Neurology, Medizinische Hochschule, Hannover, Germany.3Departments of Gastroenterology and Hepatology, Medizinische Hochschule, Hannover, Germany.4Department of Nuclear Medicine, Medizinische Hochschule, Hannover, Germany.5To whom correspondence should be addressed at Center for Positron Emission Tomography (115P), VeteransAffairs Western New York Healthcare System, 3495 Bailey Avenue, Buffalo, New York 14215. E-mail: [email protected]

290885-7490/02/0300-0029/0C© 2002Plenum Publishing Corporation

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

30 Lockwood, Weissenborn, Bokemeyer, Tietge, and Burchert

abnormalities in specific brain regions. Refinements in statistical methodology permitted anextension of these studies, and a second small group of alcoholic cirrhotics was comparedto controls (Lockwoodet al., 1993). These patients, who had mild HE, had a reductionin the CMRglucose in bifrontal cortices and the anterior cingulate gyrus, a critical part ofthe anterior attention system (Posner and Petersen, 1990). Thus, a deficit in attention, theability to focus neural resources on a specific task, was postulated as a critical element inthe deficits that characterize minimal cognitive impairment in cirrhotics.

The present study was undertaken to seek relationships between the degree of impair-ment on neuropsychological tests and the regional CMRglucose in a larger group of patientswith nonalcoholic cirrhosis. The results of this study have been presented, in part, in abstractform (Lockwoodet al., 1998).

METHODS

Subjects and Their Evaluation

Nonalcoholic cirrhotics were recruited and studied in Hannover, Germany. Controlsubjects were studied in Buffalo, NY. Written informed consent was obtained from allsubjects in accord with regulations specific to each imaging center and the Declaration ofHelsinki. Patients were examined by a neurologist (KW) and graded as to the severity oftheir encephalopathy (Holmet al., 1980) with criteria similar to those of Parsons-Smithet al.(1957). T1 and T2 MRI scans of the brain were obtained in 15 patients (1.0 T, Magnetom,Siemens). Each patient was given a battery of German versions of the Trailmaking tests Aand B, line drawing (time and error), serial dotting, a modified digit–symbol test (Schomeruset al., 1999), and a cancelingd test (two results) yielding a total of 8 scores. All but thecancelingd test have been used to devise a portal-systemic encephalopathy score (PSEsyndrome test) (Schomeruset al., 1999). Each result was age-corrected and converted to aZ score to standardize differences from mean normal values (where aZ = 1 represents 1standard deviation from the mean,Z = 2 represents 2 standard deviations from the mean,etc.). In addition, a composite score was computed for each subject by assigning a valueof 0 for Z scores within 1 standard deviation from the mean,±1 for scores between 1 and2 standard deviations from the mean,±2 for scores between 2 and 3 standard deviationsfrom the mean, and±3 for scores 3 or more standard deviations from the mean. Subjectsunable to complete a test were assigned a value of−3.

The Z scores from 16 patients able to complete all tests were subjected to furthercorrelational and factor analysis (Systat 8.0, SPSS Inc, Chicago, IL). The two patientsexcluded were too severely affected to complete individual tests.

PET Imaging

Both laboratories have Siemens ECAT 951/31 tomographs. Subjects were scanned atrest with their eyes open. At the onset, subjects were given an intravenous injection of 10 mCi(370 MBq) or less of18F-fluorodeoxyglucose (FDG). Periodic timed arterial blood sampleswere drawn and used to calculate CMRglucose (Huanget al., 1980; Patlaket al., 1983).

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

Cerebral Metabolism and Function in Cirrhosis 31

Data Analysis

PET Images

PET images were edited to remove extracerebral activity, normalized to conform to astandard brain size and shape based on the 1988 Talairach Atlas (Talairach and Tournoux,1988), smoothed using a 15 mm kernel, and normalized to a common value, using 1995statistical parametric mapping (SPM) software (Frackowiaket al., 1997). Significant rela-tionships were sought using the 1996 version of SPM using a omnibus threshold ofp =0.001 (Frackowiaket al., 1997). This software generatesZ score images that show thedeviations from a mean control value, where±1 indicates 1 standard deviation from themean,±2 indicates 2 standard deviations from the mean, etc. In the “glass brain” for-mat, SPM{Z} image planes are stacked and viewed from the side, top, and front of thebrain: subthreshold values are not shown, while suprathreshold values are depicted with agreyscale (higher values are darker). The most significantZ value is projected on to eachof the three orthogonally oriented planes. Coordinates (x, y, andz, in mm) and Brodmannareas (BA) of SPMZ maxima are referenced to the Talairach Tournoux Atlas (Talairach andTournoux, 1988).

The correlation analysis in the SPM package identified brain regions with positive ornegative correlations between the CMRglucose and the composite neuropsychological testscore andZ scores for each individual neuropsychological test. A secondary, confirmatoryanalysis was performed by comparing the patients to the controls.

RESULTS

Subjects

The 18 patients were 52± 11 years old. Thirteen patients were completely normal,according to standard bedside examination criteria. Eight patients had PSE scores that werealso in the normal range, i.e., they scored−3 or higher. Five had normal clinical examinationsbut scored in the impaired range on one or more of the subtests. Their PSE scores ranged from−4 to−14 and have minimal encephalopaty. One of these subjects was Russian and wasunfamiliar with the German alphabet. He had a pathological score on the Trailmaking A testand it was assumed that he would score at least−1 on the Trailmaking B test. Three had grade1 encephalopathy. Their PSE scores ranged from−3 to−14. Two patients were untestablebecause of grade 2 encephalopathy and were arbitrarily assigned a−3 score on all subtests ofthe PSE examination. All but one of the patients had elevated venous plasma ammonia levels(mean: 89.8µmol/L; SD: 26.5µmol/L; range 44–136µmol/L). In each case, the diagnosis ofcirrhosis was established by liver biopsy. Four patients had cirrhosis of unknown origin, fivehad hepatitis C, four had hepatitis B and C, one had primary sclerosing cholangitis, three hadprimary biliary cirrhosis, and one had autoimmune hepatitis. The Child classification was:A= 4, B= 7, andC = 7. For the portion of the analysis restricted to 16 subjects, the Childclassifications were:A= 4, B= 6, andC= 7. At the time of the neuropsychological testingand PET imaging (done within 24 h of each other), most patients were receiving lactuloseand were clinically stable, i.e., not recovering from an episode of severe encephalopathy.

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

32 Lockwood, Weissenborn, Bokemeyer, Tietge, and Burchert

Table 1. Z Score Data and Factor Loadings for Neuropsychological Tests

Test Max Min Mean SD Factor 1 Factor 2

Trailmaking A 3.01 −1.7 0.79 1.28 0.963 0.022Trailmaking B 6.18 −1.2 1.35 2.16 0.893 −0.443Line drawing error 1.64 −2.8 −0.53 1.35 0.911 0.364Line drawing time −0.73 4.45 1.49 1.61 0.871 −0.286Dotting 3.4 −0.9 1.16 1.38 0.949 0.232Digit–symbol −7.64 1.31 −1.57 2.39 −0.985 −0.083Composite test score 3 −24 6.1 8 −0.973 0.034Percent variance explained by factor 86.3 6.7

An analysis of variance showed no differences in the whole-brain CMRglucose amongthe four patient groups (normal, minimal, grade 1, grade 2 HE) (group by CMRglucose,degrees of freedom= 3, 14, F = 1.209, p = 0.343). Similarly, there was no correlationbetween whole-brain CMRglucose and performance on the composite battery: Pearsoncorrelation coefficient= 0.3436,p= 0.163.

The MRI scans showed slight pallidal high signal intensities on T1 images in nine pa-tients, and a more pronounced hyperintensity in five patients. Only one patient had a normalsignal intensity in the pallidum. Slight cortical atrophy was present in four of the 15 patients.

Neuropsychological Tests

Values for the mean, maximum, minimum, and standard deviation of theZ scores forthe Trailmaking A and B, line drawing (yielding time and error scores), dotting, and digit–symbol tests and the composite test are shown in Table 1. The scores on the composite batteryranged from+3 to−24. None of the patients performed better than 1 standard deviationabove the age-corrected mean on any subtest. The digit–symbol test yielded the greatestmean deviation from normal (meanZ = −1.57± 2.39) while the line drawing error scoreresults were closest to normal (meanZ=−0.53±1.35). An analysis seeking correlations ofthese tests with each other and with the composite test was performed with results as shownin Table 2. Since the Bartlett chi-square test for the overall significance of the correlations

Table 2.Pearson Correlation and Bonferroni Probability Matrix

CompositeTest TMA TMB LDE LDT DOT DSYM test

Trailmaking A (TMA) ****** 0.001 <0.001 0.001 <0.001 <0.001 <0.001Trailmaking B (TMB) 0.802 ****** 0.076 0.008 0.036 0.006 <0.001Line drawing error (LDE) 0.882 0.648 ****** 0.078 <0.001 <0.001 <0.001Line drawing time (LDT) 0.800 0.745 0.647 ****** 0.002<0.001 0.001Dotting (DOT) 0.890 0.685 0.93 0.789 ****** <0.001 <0.001Digit–symbol (DSYM) −0.935 −0.759 −0.911 −0.873 −0.961 ****** <0.001Composite test −0.940 −0.842 −0.875 −0.817 −0.859 0.946 ******

Note: The bottom left half of the table shows the Pearson correlation coefficients among the tests indicated. TheBartlett Chi-square testing the global hypothesis concerning the significance of all correlations isχ2 = 180.173,df = 21, p < 0.001. The top right half shows the probability that the correlation is significant, corrected formultiple comparisons using the Bonferroni correction procedure.

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

Cerebral Metabolism and Function in Cirrhosis 33

yielded ap < 0.001, multiple pairwise comparisons of the probabilities were performed,using a Bonferroni correction. These results are shown in Table 2. Although all subtestshad a significant correlation with the composite score, the correlations were highest for thedigit–symbol and Trailmaking A tests.

A factor analysis with computation of principal components was also performed (vari-max rotation). Factor loadings for the first two factors associated with each test are shownin Table 1 along with the fraction of the total variance explained by the factors. Factor 1explains 86.3% of the variance in the test score results.

Correlations between Neuropsychological Tests and CMRglucose

Figure 1 shows the loci of positive and negative correlations between the compositescore on the neuropsychological test battery and cerebral glucose metabolism. The imagesof positive correlations define sites in the brain where higher scores (better performance)are associated with a higher, more normal CMRglucose (it also shows correlations betweenpoor performance and low CMRglucose). The images of negative correlations define sitesin the brain where poor performance is associated with a high CMRglucose. Part A ofthe figure shows the correlations in the subset of 13 patients who had no evidence ofovert encephalopathy. Eight were clinically normal, five were clinically normal but hadone or more test scores in the impaired range, and none were classified as grade 1 or 2.The image of positive correlations contains six clusters, predominantly in inferior frontalregions, containing a total of 1216 pixels with aZ maximum of 4.49 at cordinates 22,38,−12 (medial frontal gyrus, BA 11). The image of negative correlations contains threeclusters with a total of 3734 pixels and has aZ maximum of 4.42 in a cluster of 3602 pixelsat coordinates−28,−28,−8 in the hippocampus. Part B of the figure shows the loci ofpositive and negative correlations between metabolism and test scores in the 16 patients withgrade 1 or better HE. The image of positive correlations includes 2365 pixels in six clusters.The Z maximum of 4.75 was at−34, 22,−8 is in the inferior frontal gyrus. The image ofnegative correlations contains 3579 pixels with aZ maximum of 4.95 at−26,−18,−16 inthe hippocampus. Part C of the figure shows the loci of positive and negative correlationsin all 18 patients. The image of positive correlations contains 3090 pixels with one largebiparietal–occipital cluster containing 2354 pixels with a Z maximum of 5.39 in the superioroccipital gyrus at coordinates 34,−84, 24 (BA 39-19). The image of negative correlationscontains 3989 pixels. All but 149 are in a single cluster with a Z maximum of 5.21 in thehippocampal gyrus at coordinates−26,−14,−20. The overall appearance of this correlationimage is very similar to the images comparing the patients to the controls shown in Fig. 3.

SPMZ images depicting correlations between low CMRglucose metabolism and poorperformance for the six tests in the battery that form the PSE syndrome test are shown inFig. 2. The correlations with frontal CMRglucose metabolism varied substantially amongthe tests. To depict these differences systematically, we arranged the SPMZ images to showincreasing correlations with frontal lobe sites.

Table 3 shows more detailed results from these correlation analyses. The highest cor-relation between performance and metabolism was found with the digit–symbol subtestof the WAIS-R. TheZ score of 6.17 is in the right occipital gyrus, BA 19, a visual asso-ciation area. This same pixel was the site of theZ maximum correlating performance

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

34 Lockwood, Weissenborn, Bokemeyer, Tietge, and Burchert

Figure 1. Correlations between the composite test battery score and CMRglucose. Part A shows theSPM {Z} projections in the sagittal, coronal, and transaxial orientations demonstrate brain regionswhere there are positive and negative correlations between CMRglucose and the composite test scorein a subset of 13 patients with no overt evidence of HE. Clinically, all appeared normal. Test scoresranged from+3 to−14. Part B shows the results of this same analysis for the 16 patients with grade1 HE or better. Test scores again ranged from+3 to−14. Part C shows the results of this analysis forall 18 subjects, including those with grade 2 HE. Test scores in the entire group ranged from+3 to−24. Sites and values of the SPMZ maxima are in the text.

of the time for the line drawing test (negative correlation,Z = 5.42). The maximumcorrelation with the performance on the Trailmaking A test was found in BA 40 of theleft hemisphere at the site of the inferior parietal lobule–supramarginal gyrus (Z = 5.41).

Comparison of Cirrhotic Patients to Controls

Figure 3 shows the results obtained from the comparison of the cirrhotic patients to 10control subjects aged 29± 9 years (SD). The region where the CMRglucose is higher in

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

Cerebral Metabolism and Function in Cirrhosis 35

Table 3.Correlations between Age-Corrected Test Performance and CMRglucose

Size, largest Z maximumTest, positive or Clusters cluster (number Z p value coordinates

negative correlation identified of voxels) maximum (corrected)x, y, z (mm)

Line drawing time, negative 1 1888 5.42 <0.001 32,−82, 28Dotting, negative 2 664 5.67 <0.001 36,−82, 24Line drawing errors, negative 3 596 4.41 0.008 40,−78, 20Trailmaking B, negative 4 260 4.12 0.045−40,−44, 32Symbol–digit, positive 2 2860 6.71 <0.001 32,−82, 28Trailmaking A, negative 2 3078 5.41 <0.001 −46,−38, 28

Figure 2. Correlations between performance, as measured by age-correctedZ scores, and metabolismfor various subtests in the composite battery. Only those subjects able to complete the test are includedin the analyses. The SPMZ image projections are as in other figures. They all show significantcorrelations with bilateral parietal associative cortex with increasing correlations with frontal regions.The disparity of these results, particularly with regard to the correlations with frontal lobe metabolism,like the factor loading scores, suggest that the different tests appear to make different demands oncerebral resources. See Table 3 for details.

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

36 Lockwood, Weissenborn, Bokemeyer, Tietge, and Burchert

Figure 3. Comparison of glucose metabolism in 18 patients with cirrhosis to 10 controls. SPMZimage orientations are as in the other figures.

the patients than controls is shown in part A of the figure and consists of a cluster of 2087voxels with aZ maximum of 4.46 at coordinates−30,−44,−8. These clusters occupy theposterior portion of the putamen and extend to include portions of the hippocampal, lingualand fusiform gyri, particularly on the left, and some portions of the vermis of the cerebellum.The CMRglucose is lower in the patients in three clusters confined to the cerebral cortexcontaining a total of 6350 pixels with aZ maximum of 5.41 at 34, 10, 40. These clustersoccupy relatively symmetrical cortical sites in both frontal and parietal–occipital regions, theanterior cingulate, and a medial portion of primary and visual association cortex includingthe precuneus.

DISCUSSION

We tested 18 patients with nonalcoholic cirrhosis using a test battery specifically de-signed to detect and quantify minimal cerebral dysfunction in patients with cirrhosis of theliver (Schomeruset al., 1999) and found significant correlations between cerebral glucosemetabolism and poor performance on the test battery as a whole as well as the individ-ual tests that form the battery. We suggest that this regional hypometabolism forms thepathophysiological basis for the minimal cerebral dysfunction that is often revealed byneuropsychological testing of cirrhotics.

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

Cerebral Metabolism and Function in Cirrhosis 37

Neuropsychological Tests

Neuropsychological testing of patients with cirrhosis invariably demonstrates impair-ment in a substantial proportion of patients with cirrhosis who are thought to be normal,based on conventional bedside evaluations (e.g., see Gilberstadtet al., 1980; Gitlinet al.,1986; Groeneweget al., 2000; Rikkerset al., 1978; Schomerus and Schreiegg, 1993; Tarteret al., 1984). The abnormalities we found in our subjects, including those without evidenceof overt hepatic encephalopathy, are in keeping with these earlier reports.

Since the subtests that were included in the battery were chosen specifically to detectminimal impairment, it is not surprising that there were significant correlations among theresults of the various subtests and with the entire battery. The results of the factor analysis,shown in Table 1 support this conclusion. All of the loading factors had an absolute valueof 0.8 or greater. This suggests that they have something in common. Since the nature of thetasks differs (some stress motor activity while others visouspatial integration) we suggestcalling this the HE factor, account for 86.3% of the variance in the results of the test battery.Thus the results of the correlation and factor analyses are in agreement.

Correlations between Metabolism and Function

The inclusion of all patients with cirrhosis into a single group, for the purposes ofperforming the correlation analyses, was justified by the negative results of statistical testsdesigned to reveal between-group heterogeneity (ANOVA,p= 0.343; Pearson correlationcoefficient=0.3436,p=0.163). The absence of between-group differences is a prerequisiteto the placement of all patients into a single group and the performance of the statisticaltests we report.

We found significant correlations between a low CMRglucose in inferior frontal anddorsolateral frontal regions and poor performance on the composite battery, even in patientsjudged to be normal (i.e., functioning at a level better than grade 1 encephalopathy) as shownin Fig. 1(A). As patients with more severe encephalopathy were added to the comparison,the extent of the brain exhibiting positive correlations between metabolism and functionincreased, as seen in Fig. 1(B) where SPMZ images are shown for all patients with grade1 encephalopathy or better, and Fig. 1(C) where correlations among all patients, includingthe two with grade 2 HE are shown.

The Trailmaking A and B tests and the digit–symbol tests of the WAIS-R have beenused widely to detect minimal encephalopathy in patients with cirrhosis. As seen in Fig. 2,there are many sites where there are significant correlations between poor performanceon these tests and abnormal cerebral metabolism. These results are in agreement with theconclusions drawn by Tarter and his associates whose neuropsychological tests indicatedthat these brain regions were affected in patients with cirrhosis (Tarteret al., 1984). Theprimary sensory–motor areas in the postcentral gyrus were unaffected. This is in keepingwith the nature of the deficits observed in these patients that show impairment of integrativeand executive functions.

It is unclear, from our data, why metabolism should be increased in the brain regionsshown in the figures. There is little evidence, based on clinical or neuropsychologicaltests, to suggest that these brain regions are excessively active. Similarly, neuropathological

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

38 Lockwood, Weissenborn, Bokemeyer, Tietge, and Burchert

examinations (Lockwood, 1992) and neuroimaging techniques, including T1-weighted MRIimages that show pallidal hyperintensity provide no help in understanding this finding(Lockwoodet al., 1997). Increases in glucose metabolism were observed in similar brainregions in an earlier study of patients with alcohol-induced cirrhosis (Lockwoodet al.,1993).

Comparison of FDG PET Data from Patients to Controls

The within-group correlation analyses are the major focus of this study. However, theresults of these correlation analyses are supported by a secondary patient versus controlanalysis. There are many similarities between the images of the positive and negativecorrelations between metabolism and performance on the composite test battery and theimages of between-group (patient versus control) differences. Although the control subjectsare younger than the patients, we do not believe that this is a significant confounding factor.Martin et al. found no significant correlation between global blood flow and age in a groupof 30 subjects ranging from age 30 to 85 and age-dependent changes did not include theareas of difference we observed (Martinet al., 1991). Similarly, Schultzet al. who studiedsubjects aged 19–50, also failed to find age-dependent reductions in the hemispheric surfacesof the parietal and frontal cortices (Schultzet al., 1999). The similarities between the resultsof the present study and the results of the earlier comparison between alcoholic cirrhoticsand normal controls (Lockwoodet al., 1993) lends further support to the validity of ourdata. In that earlier study, which included five patients, reductions in glucose metabolismwere observed in the anterior cingulate gyrus and bifrontally with increases in metabolismin deep central grey matter regions.

The Pathophysiological Basis for Minimal Hepatic Encephalopathy

Our data are consistent with the hypothesis that a defect in attention is a fundamentalaspect of minimal HE. This is supported by the results of correlation analyses that link poortest battery performance with hypometabolism in the anterior cingulate gyrus. This was afeature of the analysis restricted to the patients who were overtly normal (Fig. 1(A)), thepatients who were grade 1 HE or better (Fig. 1(B)), and the comparison of the patients tothe controls (Fig. (3)). Functional neuroimaging techniques have shown that the anteriorcingulate is an essential element in the anterior attentional system, as conceptualized byPosner and his associates (Posner, 1995; Posner and Driver, 1992; Posner and Petersen,1990). According to this model, the anterior portion of the cingulate is particularly involvedwith target monitoring and response formation. Recent data suggest that this region mayalso mediate reward-based motor activity (Shima and Tanji, 1998). The cingulate metabolicdata from this study confirm and extend the earlier observation that glucose metabolismwas reduced in this brain region in patients with alcohol-induced cirrhosis and minimalHE (Lockwoodet al., 1993). Thus, the anterior cingulate is affected in patients with HE,regardless of the etiology of their liver disease, and appears to be a central manifestation ofthe disorder, particularly in the milder forms. While defects in attention are an importantfeature of HE, the links between poor performance and hypometabolism in associative cortex

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

Cerebral Metabolism and Function in Cirrhosis 39

in biparietal and bifrontal locations implicates additional neural systems. This suggests thatthe complete manifestation of HE is due to a defect in attention coupled with impairmentof executive and integrative functions.

ACKNOWLEDGMENTS

We thank our many colleagues whose efforts were essential to the conduct of thisinvestigation. This work was supported in part by grants from the U.S. Department ofVeterans Affairs and the James H. Cummings Foundation, Buffalo, NY.

REFERENCES

Frackowiak, R.S.J., Friston, K.J., Frith, C.D., Dolan, R.J., and Mazziotta, J.C. (1997).Human Brain Function,Academic Press, New York.

Gilberstadt, S.J., Gilberstadt, H., Zieve, L., Beugel, B., Collier, R.O., Jr., and McClain, C.J. (1980). Psy-chomotor performance defects in cirrhotic patients without overt encephalopathy.Arch. Int. Med. 140:519–521.

Gitlin, N., Lewis, D.C., and Hinkley, L. (1986). The diagnosis and prevalence of subclinical hepatic encephalopathyin apparently healthy, ambulant, non-shunted patients with cirrhosis.J. Hepatol. 3:75–82.

Groeneweg, M., Moerland, W., Quero, J.C., Hop, W.C., Krabbe, P.F., and Schalm, S.W. (2000). Screening ofsubclinical hepatic encephalopathy.J. Hepatol. 32:748–753.

Holm, E., Thiele, H., Wolpert, E.M., Heim, M.E., and Reimann-Werle, B. (1980). Neurologische und PsychiatrischeSymptome bei akuten und chronischen Leberkrankheiten.Therapiewoche30:4790–4806.

Huang, S.-C., Phelps, M.E., Hoffman, E.J., Sideris, K., Selin, C.J., and Kuhl, D.E. (1980). Noninvasive determi-nation of local cerebral metabolic rate of glucose in man.Am. J. Physiol.238:E69–E82.

Lockwood, A.H. (1992).Hepatic Encephalopathy, Butterworth Heinemann, Stoneham, Massachusetts.Lockwood, A.H., Murphy, B.W., Donnelly, K.Z., Mahl, T.C., and Perini, S. (1993). Positron-emission tomo-

graphic localization of abnormalities of brain metabolism in patients with minimal hepatic encephalopathy.Hepatology18:1061–1068.

Lockwood, A.H., Weissenborn, K., Burchert, W., Bokemeyer, M., and Wack, D.S. (1998). Neuropsychological testdeficits correlate with altered cerebral glucose metabolism in patients with non-alcoholic cirrhosis.Neurology50:A253.

Lockwood, A.H., Weissenborn, K., and Butterworth, R.F. (1997). An image of the brain in patients with liverdisease.Curr. Opinion Neurobiol.10:525–533.

Lockwood, A.H., Yap, E.W.H., Rhoades, H.M., and Wong, W.-H. (1991). Altered cerebral blood flow and glucosemetabolism in patients with liver disease and minimal encephalopathy.J. Cereb. Blood Flow Metab. 11:331–336.

Martin, A.J., Friston, K.J., Colebatch, J.G., and Frackowiak, R.S.J. (1991). Decreases in regional cerebral bloodflow with normal aging.J. Cereb. Blood Flow Metab.11:684–689.

Parsons-Smith, B.G., Summerskill, W.H.J., Dawson, A.M., and Sherlock, S. (1957). The electroencephalogramin liver disease.Lancetii: 867–871.

Patlak, C.S., Blasberg, R.G., and Fenstermacher, J.D. (1983). Graphical evaluation of blood-to-brain transferconstants from multiple-time uptake data.J. Cereb. Blood Flow Metab. 3:1–7.

Posner, M.I. (1995). Attention in cognitive neuroscience: An overview. In (M.S. Gazzaniga, ed.),The CognitiveNeurosciences, MIT Press, Cambridge Massachusetts, pp. 615–624.

Posner, M.I. and Driver, J. (1992). The neurobiology of selective attention.Curr. Opionion Neurobiol.2:165–169.

Posner, M.I. and Petersen, S.E. (1990). The attention system of the human brain.Annu. Rev. Neurosci. 13:25–42.Rikkers, L., Jenko, P., Rudman, D., and Freides, D. (1978). Subclinical hepatic encephalopathy: Detection, preva-

lence, and relationship to nitrogen metabolism.Gastroenterology75:462–469.Schomerus, H. and Schreiegg, J. (1993). Prevalence of latent portasystemic encephalopathy in an unselected

population of patients with liver cirrhosis in general practice.Zeitschrift fur Gastroenterologie31:231–234.Schomerus, H., Weissenborn, K., Hecker, H., Hamster, W., and R¨uckert, N. (1999).PSE Syndrom Test: Psychodi-

agnostisches Verfahren zur quantitativen Erfassung der (minimalen) portosystemischen Encephalopathie(PSE), Swets Test Services, Frankfurt.

P1: MRM/RKP P2: MRM

Metabolic Brain Disease [mebr] PP355-365681 January 29, 2002 10:32 Style file version Nov. 19th, 1999

40 Lockwood, Weissenborn, Bokemeyer, Tietge, and Burchert

Schultz, S.K., O’Leary, D.S., Boles Ponto, L.L., Watkins, G.L., Hichwa, R.D., and Andreasen, N.C. (1999). Age-related changes in regional cerebral blood flow among young to mid-life adults.NeuroReport10:2493–2496.

Shima, K. and Tanji, J. (1998). Role for cingulate motor area cells in voluntary movement selection based onreward.Science282:1335–1338.

Talairach, J. and Tournoux, P. (1988).Co-Planar Stereotaxic Atlas of the Human Brain, Georg Thieme Verlag,Stuttgart and New York.

Tarter, R.E., Hegedus, A.M., Van Thiel, D.H., Schade, R.R., Gavaler, J.S., and Starzl, T.E. (1984). Nonalco-holic cirrhosis associated with neuropsychological dysfunction in the absence of overt evidence of hepaticencephalopathy.Gastroenterology86:1421–1427.