Cerebral oedema in minimal hepatic encephalopathy due to extrahepatic portal venous obstruction

CL IN ICAL STUDIES

Cerebral oedema inminimal hepatic encephalopathydue toextrahepatic portal venous obstructionAmit Goel1, Santosh Yadav2, Vivek Saraswat1, Arti Srivastava2, M. Albert Thomas3, Chandra M. Pandey4,Ramkishore Rathore5 and Rakesh Gupta2

1 Department of Gastroenterology, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, India

2 Department of Radiodiagnosis, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, India

3 Department of Radiology, Geffen School of Medicine at UCLA, Los Angeles, USA

4 Department of Biostatistics, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, India

5 Department of Mathematics and Statistics, Indian Institute of Technology, Kanpur, India

Keywords

extrahepatic portal vein obstruction –1H-magnetic resonance spectroscopy –

hyperammonaemia – minimal hepatic

encephalopathy – MR imaging

Correspondence

Dr Vivek A Saraswat, MD, DM, Professor,

Department of Gastroenterology, Sanjay

Gandhi Post Graduate Institute of Medical

Sciences, Lucknow, India

Tel: 191 522 249 4407

Fax: 191 522 266 8017

e-mail: [email protected]

Received 18 January 2010

Accepted 4 May 2010

DOI:10.1111/j.1478-3231.2010.02289.x

AbstractBackground: Minimal hepatic encephalopathy (MHE) has recently beenreported in patients with extrahepatic portal venous obstruction (EHPVO).Aims: To evaluate brain changes by magnetic resonance studies in EHPVOpatients. Methods: Blood ammonia level, critical flicker frequency (CFF),brain metabolites on 1H-magnetic resonance (MR) spectroscopy and brainwater content on diffusion tensor imaging and magnetization transfer ratio(MTR) were studied in 31 EHPVO patients with and without MHE, asdetermined by neuropsychological tests. CFF and magnetic resonance imagingstudies were also performed in 23 controls. Results: Fourteen patients (14/31,45%) had MHE. Blood ammonia level was elevated in all, being significantlyhigher in the MHE than no MHE group. CFF was abnormal in 13% (4/31)with EHPVO and in 21% (3/14) with MHE. On 1H-MR spectroscopy,increased Glx/Cr, decreased mIns/Cr, and no change in Cho/Cr were notedin patients with MHE compared with controls. Significantly increased meandiffusivity (MD) and decreased (MTR) were observed in the MHE group,suggesting presence of interstitial cerebral oedema (ICE). MD correlatedpositively with blood ammonia level (r=0.65, P=0.003) and Glx (r=0.60,P=0.003). Discussion: MHE was detected in 45% of patients with EHPVOwhile CFF was abnormal in only 13%. ICE was present in 7/10 brain regionsexamined, particularly in those with MHE. Hyperammonaemia elevatedcerebral Glx levels correlated well with ICE. Conclusions: MHE was commonin EHPVO; CFF could identify it only in a minority. ICE was presentin EHPVO, particularly in those with MHE. It correlated with bloodammonia and Glx/Cr levels. Hyperammonaemia seems to contribute to ICEin EHPVO.

Extrahepatic portal venous obstruction (EHPVO) ischaracterized by occlusion of the extrahepatic portalvein, with or without the involvement of its tributariesor branches and in the absence of cirrhosis or neoplasia,resulting in portal–systemic shunts (PSS). Liver functionis normal and the portal vein is usually replaced by acavernoma.

Minimal hepatic encephalopathy (MHE) is a well-established entity in cirrhosis, which results in a poorquality of life, impaired working and driving capabilitiesand poor prognosis (1, 2). MHE has recently beenreported in 35–50% of adults with EHPVO (3, 4).

Magnetic resonance (MR) imaging studies including1H-MR spectroscopy (1H-MRS), magnetization transfer

ratios (MTR) and diffusion tensor imaging (DTI) havebeen used to understand cerebral alterations in cirrhosisand hepatic encephalopathy (HE) (5–9). Increased T1signal intensity (T1SI) in basal ganglia in cirrhosis isthought to be because of manganese (Mn) depositionsecondary to PSS (10). A characteristic triad of increasedGlx/Cr, decreased mIns/Cr and decreased Cho/Cr seenon 1H-MRS is considered to be the hallmark of HE incirrhosis (7, 9, 11). Low MTR has been used as a markerof increased brain water in cirrhosis (5). Minguez et al.(4) have described 1H-MRS and MT changes in fivepatients with EHPVO. However, a detailed profile ofMR changes in the brain of EHPVO patients with orwithout MHE has not been reported.

Liver International (2010)c� 2010 John Wiley & Sons A/S 1143

Liver International ISSN 1478-3223

mailto:[email protected]

The aims of this study were to assess MHE in EHPVOpatients and compare brain MR changes between MHEand no-MHE groups using T1SI quantification, metabo-lites on 1H-MRS and cerebral oedema using DTI metricsand MTR imaging.

Subjects and methods

Patients with EHPVO attending the Gastroenterologydepartment and fulfilling eligibility criteria were prospec-tively enrolled in this study after obtaining writteninformed consent. Those who initially presented withvariceal bleeding were managed as per hospital protocoland received antibiotics during hospital stay. At the time ofdischarge, none of the patients were advised lactulose/probiotics/or antibiotics. After 6 weeks of discharge fromthe hospital, they were enrolled in the study and underwentbiochemical, neuropsychological and MR evaluation. Thisstudy was approved by the ethics committee of SanjayGandhi Post Graduate Institute of Medical Sciences, Luck-now, India. EHPVO was diagnosed if obstruction of theextrahepatic portal vein was detected with cavernomaformation, with or without the involvement of its tribu-taries or branches, on Doppler ultrasound and withnormal liver function tests. Patients having isolated occlu-sion of superior mesenteric or splenic vein, or any of thefollowing conditions like neurological illness, cirrhosis,pancreatitis, malignancy, any positive aetiological markerfor cirrhosis, were excluded. Other exclusion criteria were ahistory of gastrointestinal bleeding, infection, treatmentwith antibiotics or lactulose within the previous 6 weeks,alcoholism, drug abuse, use of psychoactive drugs anduncorrected visual impairment. For comparison age-,gender- and education-matched healthy controls wereselected among relatives of the patients.

Clinical methods

All patients underwent blood tests for haemogram, liverfunction tests, prothrombin time, HBsAg, antibody toHCV, serum ceruloplasmin, antinuclear, antiliver kidneymicrosomal and antismooth muscle antibodies. Othertests included abdominal ultrasonography with Dopplerexamination and oesophago-gastroduodenoscopy.

Neuropsychological assessment

After excluding overt HE using the West Haven criteria(12) and the mini mental state examination (score � 24being abnormal), all patients were subjected to a batteryof nine NP tests, which included four trail-making tests,i.e. number connection tests A and B (NCT A and B) andfigure connection tests A and B (FCT A and B); and fivetests of performance subset of the modified WechslerAdult Intelligence Scale (WAIS-P), i.e. picture comple-tion (PC), digit symbol (DS), picture arrangement (PA),object assembly (OA) and block design (BD) tests. NCT-A and FCT-A assess visual motor coordination, concen-tration and mental speed, while NCT-B and FCT-B test

additionally assess memory alteration and attention. Theprocedure of performing WAIS-P tests and functionsassessed by these tests are described elsewhere (8, 13).The time taken for completing the NP test battery rangedfrom 35 to 45 min. Test scores were considered abnormalif patients’ values lay beyond mean� 2 SD, from normsestablished in age-, gender-, and education-matchedhealthy controls. MHE was diagnosed if Z2 NP testswere abnormal. Based on NP test results, patients weregrouped into MHE and no-MHE.

Critical flicker frequency

Critical flicker frequency (CFF) was determined usingthe HEPAtonorm analyzer (Accelab GmbH, D-72127Kusterdingen, Germany) after dark adaptation, in a quietroom with constant light intensity, after proper instruc-tions and after at least five trial readings had beenobtained. The analyzer emitted defined pulses of lightwith a wavelength of 650 nm, a luminance of 270 cd/m2

and a luminous intensity of 5.3 mcd for intrafoveal lightstimulation. Frequency of stimulus was graduallyreduced from 60 Hz; patients’ CFF were the values atwhich flickering was first observed. The procedure wasrepeated eight times in each patient and the mean CFFfrequency was calculated. The entire procedure tookabout 15 min. Cut-off for the abnormal CFF value wastaken as mean –1 SD to maintain consistency withpublished literature (14).

Blood ammonia level measurement

Blood sample was collected in a heparinized syringe afterovernight fasting. The collected sample was immediatelytaken to the laboratory in an ice bath and serum wasseparated within 30 min of sample collection followed byammonia estimation using the enzymatic-UV method(Randox Laboratories Ltd, County Antrim, UK; normalrange 10–47 mmol/L).

Magnetic resonance imaging

Magnetic resonance imaging and 1H-MRS were per-formed on a 1.5 T MR system (Signa, General ElectricMedical Systems, Milwaukee, WI, USA) equipped witha quadrature birdcage receive and transmit radio-frequency head coil. The routine imaging studies in-cluded the following: fast spin-echo T2-weighted imageswith repetition time (TR)/echo time (TE)/number ofexcitations (NEX) = 6000 ms/85 ms/4 and T1-weighted fastspin-echo images with TR/TE/NEX = 825 ms/9.4 ms/4. Atotal of 36 contiguous 3-mm-thick axial sections wereacquired with a 240� 240 mm field-of-view and an imagematrix of 256� 256. The MT imaging study was per-formed in a spin-echo pulse sequence (TR/TE/flip an-gle = 950/minimum full/651) with the same slice thicknessof T1-weighted imaging, except for the off-resonancepulse (TR/TE/flip angle = 950/minimum full/651). DTIdata were acquired using a single-shot echo-planar dual

Liver International (2010)1144 c� 2010 John Wiley & Sons A/S

Cerebral oedema in extrahepatic portal venous obstruction Goel et al.

spin-echo sequence with ramp sampling. The DTI acquisi-tion parameters were as follows: b-factor set to 0 and 1000s/mm2, field-of-view = 240� 240 mm2, TR = 8 s, TE = 100ms, and NEX = 8. A total of 36 axial sections were acquiredwith an image matrix of 256� 256 (following zero-filling)and a slice thickness of 3 mm with no interslice gap. Abalanced rotationally invariant (15) diffusion encodingscheme with 10 uniformly distributed directions over theunit sphere was used for obtaining the diffusion-weighteddata. 1H-MRS was obtained using a water-suppressedlocalized single-voxel spin-echo sequence with TR/TE =3000 ms/35 ms in both patients and controls. A voxel of2� 2� 2 cm3 was placed in the right parietal region of thebrain in all cases (16).

Data quantification

T1 signal intensity

IMAGE J, a JAVA-based software, was used for the quanti-fication of T1SI. Elliptical region-of-interest(s) varyingfrom 4� 4 to 6� 6 pixels was placed on the globuspallidus on the T1-weighted image.

Magnetization transfer ratio

Magnetization transfer ratio was quantified as a percen-tage of signal loss according to the following equation:MTR = 100 (S0� SS)/S0, where S0 is the mean signalintensity for a particular region obtained from the spin-echo sequence without the saturation pulse and SS isthe mean signal intensity for the same region withthe saturation pulse. Pixel-by-pixel MTR maps wereconstructed from the two sets of spin-echo images. Themean value of the MTR within selected areas wasobtained by averaging the pixel values in the regions-of-interest on the MTR map.

1H-magnetic resonance spectroscopy

LC-MODEL software package (Version 6.0; Stephen Pro-vencher, Oakville, ON, Canada) was used for the proces-sing and quantification of the 1H-MRS data. The processof determining peak intensities of the different metabo-lites is described in detail elsewhere (17). The metaboliteratios of various metabolites were calculated with respectto creatine (Cr).

Diffusion tensor imaging

Diffusion tensor imaging data were processed and ana-lysed, as described previously in detail (8). Analysis ofconventional MRI as well as DTI images was undertakenby an experienced neuroradiologist. DTI-derived frac-tional anisotrophy (FA) and mean diffusivity (MD) mapswere displayed and overlaid on images with T2 contrastto facilitate the region(s)-of-interest placement. Ellipticaland rectangular region(s)-of-interest were placed bilater-ally at the level of the third ventricle to obtain FA and MDvalues in the anterior limb of the internal capsule and theposterior limb of the internal capsule, thalamus, puta-men, globus pallidus, cingulate gyrus, frontal whitematter and occipital white matter. Rectangular region(s)-of-interest were also placed on the genu and splenium ofcorpus callosum. In addition, co-ordinates of spectro-scopy were co-registered with MD maps in all subjects,and MD and FA values from this large voxel were alsomeasured to correlate these with spectroscopy-derivedmetabolites and blood ammonia (Fig. 1A). The size ofthe region(s)-of-interest for other regions varied from2� 2 to 6� 6 pixels.

Statistical analysis

Numerical data were expressed as mean� SD andcompared using the Student’s independent t-test and one-way analysis of variance (ANOVA) tests. Bonferroni’s

Fig. 1. (A) Axial gray scale fractional anisotropy map fused with the mean diffusivity map at the level of the third ventricle shows voxelplacement for spectra (square on right hemisphere) and regions-of-interest placement for diffusion tensor imaging measures in the lefthemisphere. (B, C) Localized proton spectra from the 2�2� 2 cm3 voxel placed on the right parietal region of control (B) and patient (C).


Goel et al. Cerebral oedema in extrahepatic portal venous obstruction

multiple comparison post hoc analysis was performedamong controls and patient groups. Pearson’s correlationcoefficients were calculated to check for any correlationamong metabolite ratios, blood ammonia and NP testscores. Correlations were considered as high, moderate andlow according to r-value 40.6, 0.6–0.4 and o0.4 respec-tively. A probability of Po0.05 was chosen to establishsignificance between the groups. All statistical analyseswere performed using the Statistical Package for SocialSciences (SPSS, version 16.0, SPSS Inc., Chicago, IL, USA).

Results

Thirty-one patients were included in this study. MHE waspresent in 14 patients (45%) and was absent in 17 (55%;no-MHE). NP tests results have been summarized inFigure 2. At least two of the four trail-making tests wereabnormal in all 14 patients with MHE. The WAIS-P subsetdid not detect any additional patients with MHE. Forcontrols, the CFF was 46.2� 3.8 Hz (mean� 1 SD), whichplaced the lower cut-off for normal CFF values at 42.4 Hz

Fig. 2. Bar plots show the scores of various neuropsychological tests in controls, no-MHE and MHE groups. �Denotes significant differencebetween MHE and no-MHE groups as compared with controls (Po 0.05); # denotes the significant difference (Po 0.05) between no-MHE andMHE groups, using Bonferroni’s multiple comparison post hoc analysis. BD, block design; DS, digit symbol; FCT, figure connection test; MHE,minimal hepatic encephalopathy; NCT, number connection test; OA, object assembly; PA, picture arrangement; PC, picture completion.

Table 1. Clinical and biochemical parameter of extrahepatic portal vein obstruction patients with no-minimal hepatic encephalopathy andminimal hepatic encephalopathy

Parameters

No-MHE (n = 17) MHE (n = 14)

P-valueMean� SD Range Mean� SD Range

Age (years) 22.3�5.3 16–33 25.9� 9.2 18–48 0.222Sex (M/F) 13/4 – 9/5 – 0.693Bleed/splenomegaly 12/5 – 8/6 – 0.436BMI (kg/m2) 19.2�1.5 15.8–22.0 19.6� 2.1 15.5–22.9 0.596Spleen size(cm) 8.2�5.9 1–20 7.7� 3.7 2–14 0.786Haemoglobin (g/dl) 11.5�2.3 8.0–15.1 11.0� 2.8 5.0–14.5 0.43Platelets (1000/mm3) 82.0�58 27–289 70.0� 47 22–208 0.50Serum bilirubin (mg/dl) 1.7�0.9 0.8–3.4 1.8� 0.9 0.6–3.2 0.87Serum albumin (g/dl) 4.20�0.6 2.4–5.0 4.20� 0.5 3.5–5.0 0.94AST (IU/dl) 38.0�13 19–68 44.0� 22 20–89 0.34ALT (IU/dl) 31.0�13 17–74 32.0� 13 15–58 0.80Alkaline phosphatase (IU/dl) 170�123 73–622 169� 113 79–515 0.99Serum creatinine (mg/dl) 0.8�0.2 0.5–1.1 0.8� 0.2 0.5–1.1 0.82PT prolongation(s) (Control 11.8 s) 2.10�1.6 0.0–5.1 1.70� 1.3 0.0–4.2 0.41Blood ammonia (mmols/L) 121�42.8 52–190 154.7� 34.9 110–214 0.02

ALT, alanine transaminase; AST, aspartate transaminase; BMI, body mass index.



(mean-1 SD). CFF was abnormal in four (13%) patients(three with MHE and one with the no-MHE group).

Clinical and laboratory profile

Twenty (64%) patients initially presented with acutevariceal bleeding, while 11 (36%) presented with asymp-tomatic splenomegaly. On examination, all patientsshowed splenomegaly. Clinical and laboratory para-meters of MHE and no-MHE groups were comparable,except for blood ammonia, which was higher in MHEthan in no-MHE (Table 1).

Globus pallidus T1 signal intensity

T1 signal intensity showed a sequential increase fromcontrols to no-MHE to MHE (Table 2). It significantlyincreased in EHPVO as compared with controls withoutany significant difference between MHE and no-MHEgroups (Table 2).

1H-magnetic resonance spectroscopy

Brain glutamine was increased in all patients, as indicatedby an increased Glx/Cr, which showed a gradual increasefrom controls to no-MHE to MHE groups (Table 2). Asignificant increase in Glx/Cr was observed in EHPVOand MHE group compared with controls (Table 2); asignificant decrease in mIns/Cr occurred only in patientswith MHE compared with controls. Cho/Cr and NAA/Crdid not show any statistically significant difference be-tween the different groups (Table 2).

Magnetization transfer ratio values

Magnetization transfer ratio values showed a significantdecrease in frontal white matter, genu, putamen andposterior limb of internal capsule regions for the MHEgroup and in genu for the no-MHE group as compared

with controls. The MHE group had significantly de-creased MTR values in cingulate gyrus, putamenand caudate nucleus regions as compared with no-MHE(Fig. 3).

Diffusion tensor imaging metrics

As compared with controls, the no-MHE group showedsignificantly increased MD in cingulate gyrus, while the

Table 2. Summary of brain metabolite ratios in all patients with extrahepatic portal vein obstruction, no-minimal hepatic encephalopathy,minimal hepatic encephalopathy and healthy controls

Groups EHPVO (n = 31) No-MHE (n = 17) MHE (n = 14) Control (n = 23)

P-value

� w z ‰

Glx/Cr 2.48�0.59 2.43�0.65 2.54� 0.53 2.03� 0.55 0.007 0.03 1.00 0.11Cho/Cr 0.22�0.06 0.23�0.04 0.23� 0.07 0.21� 0.04 0.68 0.82 0.35 1.00mIns/Cr 0.41�0.16 0.46�0.15 0.34� 0.15 0.48� 0.17 0.10 0.03 1.00 1.00NAA/Cr 1.11�0.41 1.13�0.35 1.09� 0.48 1.30� 0.17 0.10 0.37 0.69 0.69GP T1 SI 554�26 537�28 570� 25 462� 24 0.05 0.001 0.58 0.001

All values are expressed as the Mean� SD. 1H-Magnetic resonance spectroscopy peaks (Glx/Cr, Cho/Cr, mIns/Cr, NAA/Cr) are expressed as metabolite

ratios with respect to Cr.�EHPVO vs control.

wMHE vs control.

zMHE vs no-MHE.

‰No-MHE vs control.

cho, choline; Cr, creatine; GP T1 SI, globus pallidus T1 signal intensity; Glx, glutamine/glutamate; MHE, minimal hepatic encephalopathy; mIns, myo-

inositol; NAA, N-acetylaspartate.

Fig. 3. Bar chart comparing the magnetization transfer ratio (MTR)values of different study groups in different regions. �Significantdifference (Po0.05) between the MHE and no-MHE groups ascompared with controls; #Significant difference (Po0.05) betweenMHE and no-MHE patient groups, using Bonferroni’s multiplecomparison post hoc analysis. CG, cingulate gyrus; CN, caudatenucleus; FWM, frontal white matter; P, putamen; PLIC, posteriorlimb of internal capsule.



MHE group showed significantly increased MD in fron-tal white matter, cingulate gyrus, caudate nucleus, puta-men, globus palidus, posterior limb of internal capsuleand occipital white matter regions. There was a signifi-cant difference between MHE and no-MHE groups(Fig. 4) for caudate nucleus, putamen and posterior limbof internal capsule regions. No significant change wasobserved in FA values in any region in patients with MHEand no-MHE as compared with controls.

Correlation among different parameters (NP tests, bloodammonia, Glx/Cr, magnetization transfer ratio, diffusiontensor imaging metric, globus pallidus T1SI)

To assess the effect of low-grade cerebral oedema onNP test performance, MD and MTR values in variousregion-of-interest(s) were correlated with different NPtests, and their Pearson’s correlation coefficients aresummarized in Tables 3 and 4 respectively. A significant

Fig. 4. Bar chart comparing mean diffusivity (MD) values of different study groups in different regions. �Significant difference (Po0.05)between the MHE and no-MHE groups as compared with controls; #Significant difference (Po0.05) between MHE and no-MHE patientgroups, using Bonferroni’s multiple comparison post hoc analysis. CN, caudate nucleus; FWM, frontal white matter; GP, globus pallidus; OWM,occipital white matter; P, putamen; PLIC, posterior limb of internal capsule.

Table 3. Pearson’s correlation of mean diffusivity (MD) of different region-of-interest with neuropsychological Tests in patients withextrahepatic portal vein obstruction

Regions NCTA NCTB FCTA FCTB PC DS BD PA OA

CN 0.351 0.219 0.370� 0.284 �0.201 �0.176 � 0.008 0.090 0.090P 0.434w 0.377� 0.277 0.308 �0.103 �0.394� � 0.100 0.245 0.245GP 0.472w 0.335 0.265 0.190 �0.300 �0.441w � 0.105 �0.073 �0.073CG 0.269 0.251 0.110 0.193 0.049 �0.251 0.044 0.069 0.069FWM 0.460w 0.379� 0.310 0.337 �0.302 �0.331 � 0.161 �0.134 �0.134OWM 0.656z 0.229 0.331 0.276 �0.099 �0.150 0.188 0.137 0.137G 0.175 0.053 0.268 0.238 �0.179 �0.413w � 0.123 �0.053 �0.053S 0.100 0.154 0.110 0.070 �0.227 0.062 0.322 0.144 0.144PLIC 0.424w 0.251 0.442w 0.315 �0.377� �0.090 � 0.156 �0.025 �0.025

�, w, zHigh (r40.6), moderate (r = 0.6–0.4) and low (ro 0.4) significant (Po 0.05) correlation of neuropsychological test with mean diffusivity in

various brain regions respectively.

BD, block design; CG, cingulate gyrus; CN, caudate nuclei; DS, digit symbol; FCT, figure connection tests A and B; FWM, frontal white matter; G, genu;

GP, globus pallidus; NCT, number connection tests A and B; OA, object assembly; OWM, occipital white matter; P, putamen; PA, picture arrangement;

PC, picture completion; PLIC, posterior limb of internal capsule; S, splenium.



moderate positive correlation was observed among MDvalue from the spectroscopy voxel with Glx (r = 0.60,P = 0.003) and blood ammonia (r = 0.65, P = 0.003).

Discussion

Minimal hepatic encephalopathy was detected in 45% ofpatients with EHPVO in the present study. Blood ammo-nia was increased in all patients and was significantlyhigher in MHE than in no-MHE. Globus pallidus T1SIwas significantly increased in both groups of patients.1H-MRS revealed a significant increase in Glx/Cr and adecrease in mIns/Cr, while Cho/Cr and NAA/Cr re-mained unchanged in MHE compared with controls.Significantly increased MD, decreased MTR and un-changed FA values in various region(s)-of-interest inpatients with MHE as compared with controls suggestedthe presence of cerebral oedema without any microstruc-tural changes in these patients.

NP tests battery detected MHE in 45% of the EHPVOcohort. The trail-making tests were most useful fordetecting MHE as NCT-A was abnormal in 9/14, whileNCT-B, FCT-A and FCT-B were abnormal in 13/14 MHEpatients. WAIS-P tests failed to yield any extra case ofMHE. Trail-making tests seem to be the most usefulpaper–pencil tests for MHE detection. In the only otherreport evaluating MHE in EHPVO, Sharma et al. (3)reported a frequency of 26.4% based on abnormal NCT-A and NCT-B for literate patients while using FCT-A andFCT-B only for illiterate patients. The higher yield ofMHE in the present study can be attributed to the use ofFCT-A and B in all EHPVO patients, irrespective of theireducational status. None of our patients were illiterate; ifonly NCT-A and B had been used, MHE would have beenidentified only in 9/31 patients (29%), comparable toSharma et al. (3). On the other hand, using only FCT-Aand B would have identified MHE in 13/31 (42%). Thus,FCT-A and B appeared to be more useful than NCT-Aand B, suggesting that they should be used in all patientsand not reserved only for illiterates. The psychometric

hepatic encephalopathy score (PHES) battery was notused in these patients. Currently, the PHES battery isgaining popularity for detection of MHE in cirrhosis; itsperformance in detecting MHE in EHPVO is awaitedwith interest.

CFF has been positioned as a simple, sensitive andreproducible method for assessing MHE and early HE incirrhosis (14, 18). Its role in evaluating EHPVO-relatedMHE is not known. In the present study, the mean� SDvalue in controls was 46.2� 3.8 Hz. With the cut-off atmean-1 SD, i.e. 42.4 Hz, as proposed by Kircheis et al.(14), only 13% of the patients had abnormal CFF andsensitivity was 21% for detecting MHE compared withpsychometric tests. Moreover, in the no-MHE group, 7%had abnormal CFF. These observations suggested thatCFF had limited usefulness for identifying MHE inpatients with EHPVO. A recent study reported a sensi-tivity of 75% in detecting MHE in EHPVO, using the cut-off value of 38 Hz reported by Kircheis et al. (14) derivedfrom a German cohort of healthy volunteers (3). Thisdifference may be related to the different cut-off valuesfor abnormal CFF used. In the present study, healthycontrols with social, cultural and educational back-grounds similar to the patients were used. We speculatethat the lower detection rate of MHE by CFF in EHPVOas compared with cirrhosis might be because of presentlyunexplored differences in the pathophysiological basis ofMHE in these two conditions.

Increased T1SI in the globus pallidus is well documen-ted in patients with cirrhosis with or without MHE and isthought to be because of an increased deposition ofmanganese (19) that bypasses the hepatic filter due toPSS. We observed a significant increase in globus pallidusT1SI in the entire group of patients, when compared withcontrols. This increase was observed in those with MHEas well as those with no-MHE and was similar to thatreported in cirrhosis (10), suggesting that the degree ofPSS in EHPVO is at least similar to that in cirrhosis.However, there was no significant difference in globuspallidus T1SI between MHE and no-MHE, suggesting

Table 4. Pearson’s correlation of magnetization transfer ratio of different regions of interests with neuropsychological tests in patients withextrahepatic portal vein obstruction

Regions NCTA NCTB FCTA FCTB PC DS BD PA OA

CN 0.091 � 0.088 � 0.268 � 0.336 0.039 0.300 0.169 0.240 0.054P � 0.423� � 0.599� � 0.493� � 0.304 0.314 0.420� 0.341 0.222 0.237GP 0.068 � 0.102 0.070 0.096 �0.189 � 0.030 � 0.035 �0.151 0.053CG � 0.265 � 0.288 � 0.414� � 0.335 0.138 0.294 0.043 0.120 0.006FWM � 0.086 � 0.231 � 0.307 � 0.153 0.046 0.321 0.128 �0.073 0.142OWM 0.049 � 0.070 0.027 0.105 �0.073 0.233 0.109 �0.177 0.180G 0.048 � 0.134 0.107 0.289 �0.079 0.111 0.000 �0.267 0.100S � 0.100 � 0.107 � 0.130 0.090 0.198 0.014 0.066 0.216 0.294PLIC 0.114 � 0.163 � 0.104 0.090 �0.070 0.156 0.106 �0.172 0.285

�Moderate (r = 0.6–0.4) significant (Po 0.05) correlation of neuropsychological test with magnetization transfer ratio of various brain regions.

BD, block design; CG, cingulate gyrus; CN, caudate nuclei; DS, digit symbol; FCT, figure connection tests A and B; FWM, frontal white matter; G, genu;

GP, globus pallidus; NCT, number connection tests A and B; OA, object assembly; OWM, occipital white matter; PA, picture arrangement; P, putamen;

PC, picture completion; PLIC, posterior limb of internal capsule; S, splenium.



that increased globus pallidus T1SI may not have abearing on the development of MHE in EHPVO.

The 1H-magnetic resonance spectroscopy pattern ofincreased Glx/Cr, along with decreased mIns/Cr as well asCho/Cr, is the hallmark of HE in cirrhosis (6, 7, 11).A pattern of significantly increased Glx/Cr, decreasedmIns/Cr with unchanged Cho/Cr and NAA/Cr wasobserved in EHPVO as compared with controls. In-creased Glx/Cr is a reflection of the increased cerebralexposure to ammonia. A normal Cho/Cr ratio on1H-MRS has been reported in EHPVO earlier (4) andsuggests the absence of cerebral choline depletion inEHPVO, a finding that differentiates EHPVO fromcirrhosis. Reduced Cho/Cr has been consistently re-ported in cirrhosis with or without HE (7, 20); however,the exact mechanism for this reduction is not clear.Hypotheses proposed to explain this metabolite changeinclude decreased nutritional intake (21), altered trans-port of choline (22) and impaired brain metabolism,because of systemic metabolic alterations caused by liverdisease (23). None of these factors are operative inEHPVO patients, who have normal synthetic functionof the liver and none of the metabolic alterations ofcirrhosis. Thus, they are able to produce adequate levelsof choline, explaining the lack of choline depletion inEHPVO. In this study, brain metabolites changes be-tween MHE and no-MHE groups was not significantlydifferent, suggesting that 1HMRS may not be able todifferentiate between these two groups.

A significant decrease in MTR values in both grey andwhite matter areas in patients with MHE as comparedwith no-MHE and controls was noted in the present study.Low MTR values have been reported in various regions ofthe brain in cirrhosis as well in EHPVO and are reportedto be because of low-grade cerebral oedema (4, 5).

Significantly increased MD in most of the grey andwhite matter areas studied in EHPVO compared withcontrols suggests the presence of cerebral oedema. Cere-bral oedema was more extensive, involving the corticaland deep grey matter as well as deep white matter, inMHE than in no-MHE. The cingulate gyrus appeared tobe one of the structures that was consistently involved andaffected at the earliest stage. No significant change wasobserved in FA values in patients with MHE and no-MHEas compared with controls, suggesting that there were nomicrostructural changes in the grey or white matter tracts.A significant low to moderate correlation between MDvalues and NP tests suggested that generalized cerebraloedema was responsible for abnormalities in NP tests inthese patients. Increase in MD values on DTI has beenreported earlier in cirrhosis and MHE (8). The presentobservations reinforce these earlier observations andsuggest that the development of cerebral oedema precedesthe appearance of MHE in both cirrhosis and EHPVO.MHE appears as more brain areas are involved withcerebral oedema, and may be related to the developmentof cerebral oedema in some critical regions of the brainsuch as the cingulate gyrus. We observed significant MD

changes in more brain areas than with MTR, in bothMHE and no-MHE groups. This suggests that DTI issuperior to MTR for the assessment of cerebral oedema.Future studies with a large sample size may help inestablishing the role of DTI in the early detection of brainoedema.

Significant elevation of blood ammonia level wasobserved in all patients with EHPVO. It was significantlyhigher in those with MHE than in those with no-MHE,highlighting its role in the pathogenesis of MHE. This isin agreement with a previous report of significantlyelevated ammonia in EHPVO with MHE (3).

In conclusion, the results of this study suggest thatMHE is common in patients with EHPVO, that ICE isusually present in EHPVO with MHE and may evenprecede the appearance of MHE and, finally, that hyper-ammonaemia contributes to the development of cerebraloedema in EHPVO.

Acknowledgements

Guarantor of the article: Dr V. A. Saraswat, MD, DM.Financial support: Santosh K. Yadav received finan-

cial support from Indian Council of Medical Research(ICMR) New Delhi India.

Potential competing interests: None.

References

1. Groeneweg M, Quero JC, De Bruijn I, et al. Subclinicalhepatic encephalopathy impairs daily functioning. Hepa-tology 1998; 28: 45–9.

2. Prasad S, Dhiman RK, Duseja A, et al. Lactulose improvescognitive functions and health-related quality of life inpatients with cirrhosis who have minimal hepatic encepha-lopathy. Hepatology 2007; 45: 549–59.

3. Sharma P, Sharma BC, Puri V, et al. Minimal hepaticencephalopathy in patients with extrahepatic portal veinobstruction. Am J Gastroenterol 2008; 103: 1406–12.

4. Minguez B, Garcia-Pagan JC, Bosch J, et al. Noncirrhoticportal vein thrombosis exhibits neuropsychological andMR changes consistent with minimal hepatic encephalo-pathy. Hepatology 2006; 43: 707–14.

5. Rovira A, Grive E, Pedraza S, et al. Magnetization transferratio values and proton MR spectroscopy of normal-appearing cerebral white matter in patients with livercirrhosis. Am J Neuroradiol 2001; 22: 1137–42.

6. Miese F, Kircheis G, Wittsack HJ, et al. 1H-MR spectro-scopy, magnetization transfer, and diffusion-weighted ima-ging in alcoholic and nonalcoholic patients with cirrhosiswith hepatic encephalopathy. Am J Neuroradiol 2006; 27:1019–26.

7. Kreis R, Ross BD, Farrow NA, et al. Metabolic disorders ofthe brain in chronic hepatic encephalopathy detected withH-1 MR spectroscopy. Radiology 1992; 182: 19–27.

8. Kale RA, Gupta RK, Saraswat VA, et al. Demonstration ofinterstitial cerebral edema with diffusion tensor MR



imaging in type C hepatic encephalopathy. Hepatology2006; 43: 698–706.

9. Haussinger D, Kircheis G, Fischer R, et al. Hepatic ence-phalopathy in chronic liver disease: a clinical manifestationof astrocyte swelling and low-grade cerebral edema?J Hepatol 2000; 32: 1035–8.

10. Rose C, Butterworth RF, Zayed J, et al. Manganese deposi-tion in basal ganglia structures results from both portal–systemic shunting and liver dysfunction. Gastroenterology1999; 117: 640–4.

11. Laubenberger J, Haussinger D, Bayer S, et al. Protonmagnetic resonance spectroscopy of the brain in sympto-matic and asymptomatic patients with liver cirrhosis.Gastroenterology 1997; 112: 1610–6.

12. Atterbury CE, Maddrey WC, Conn HO. Neomycin-sorbitoland lactulose in the treatment of acute portal-systemicencephalopathy: a controlled, double-blind clinical trial.Am J Dig Dis 1978; 23: 398–406.

13. Das A, Dhiman RK, Saraswat VA, et al. Prevalence andnatural history of subclinical hepatic encephalopathy incirrhosis. J Gastroenterol Hepatol 2001; 16: 531–5.

14. Kircheis G, Wettstein M, Timmermann L, et al. Criticalicker frequency for quantification of low-grade hepaticencephalopathy. Hepatology 2002; 35: 357–66.

15. Hasan KM, Gupta RK, Santos RM, et al. Diffusion tensorfractional anisotropy of normal appearing seven segmentsof corpus callosum in healthy adult and relapsing-remittingmultiple sclerosis patients. J Magn Reson Imaging 2005; 21:735–43.

16. Verma A, Saraswat VA, Radha Krishna Y, et al. In vivo 1Hmagnetic resonance spectroscopy-derived metabolite var-iations between acute-on-chronic liver failure and acuteliver failure. Liver Int 2008; 28: 1095–103.

17. Provencher SW. Automatic quantitation of localized invivo 1H spectra with LC Model. NMR Biomed 2001; 14:260–4.

18. Sharma P, Sharma BC, Puri V, et al. Critical flickerfrequency: diagnostic tool for minimal hepatic encephalo-pathy. J Hepatol 2007; 47: 10–1.

19. Inoue E, Hori S, Narumi Y, et al. Portal-systemic en-cephalopathy: presence of basal ganglia lesions withhigh signal intensity on MR images. Radiology 1991; 179:551–5.

20. Geissler A, Lock G, Frund R, et al. Cerebral abnormalitiesin patients with cirrhosis detected by proton magneticresonance spectroscopy and magnetic resonance imaging.Hepatology 1997; 25: 48–54.

21. Klein J, Gonzalez R, Koppen A, et al. Free choline andcholine metabolites in rat brain and body fluids: sensitivedetermination and implications for choline supply to thebrain. Neurochem Int 1993; 22: 293–300.

22. Galea E, Estrada C. Quabain-sensitive choline transportsystem in capillaries isolated from bovine brain. J Neuro-chem 1992; 59: 936–41.

23. Rao VL, Therrien G, Butterworth RF. Choline acetyltrans-ferase and acetylcholinesterase activities are unchanged inbrain in human and experimental portal-systemic ence-phalopathy. Metab Brain Dis 1994; 9: 401–7.



Documents

Cerebral oedema in minimal hepatic encephalopathy due to extrahepatic portal venous obstruction