Demonstration of interstitial cerebral edema with diffusion tensor MR imaging in type C hepatic encephalopathy


  • Potential conflict of interest: Nothing to report.


Brain water may increase in hepatic encephalopathy (HE). Diffusion tensor imaging was performed in patients with cirrhosis with or without HE to quantify the changes in brain water diffusivity and to correlate it with neuropsychological (NP) tests. Thirty-nine patients with cirrhosis, with minimal (MHE) or overt HE, were studied and compared to 18 controls. Mean diffusivity (MD) and fractional anisotropy (FA) were calculated in corpus callosum, internal capsule, deep gray matter nuclei, periventricular frontal, and occipital white matter regions in both cerebral hemispheres. The MD and FA values from different regions in different groups were compared using analysis of variance and Spearman's rank correlation test. In 10 patients with MHE, repeat studies were performed after 3 weeks of lactulose therapy to look for any change in MD, FA, and NP scores. Significantly increased MD was found with insignificant changes in FA in various regions of brain in patients with MHE or HE compared with controls, indicating an increase in interstitial water in the brain parenchyma without any microstructural changes. A significant correlation was found between MD values from corpus callosum, internal capsule, and NP test scores. After therapy, MD values decreased significantly and there was a corresponding improvement in NP test scores. Further analysis showed that MD values were different for different grades of minimal or overt HE. In conclusion, the increase in MD with no concomitant changes in FA in cirrhosis with minimal or early HE indicates the presence of reversible interstitial brain edema. (HEPATOLOGY 2006;43:698–706.)

Cerebral edema (CE) is a well-known feature of acute liver disease; however, recently it has also been implicated in chronic liver disease.1, 2 CE in chronic liver disease differs from that in acute liver failure in terms of the temporality of disease. In the former, there is sufficient time for effective compensation and stabilization of osmolyte shift to counteract the osmotic imbalance induced by intra-astrocytic accumulation of glutamine, whereas in the later, evolution of the syndrome is more rapid and does not allow the system to compensate for metabolic changes.3 CE in acute liver failure is cytotoxic,4 whereas in chronic liver disease, low-grade CE is associated with Alzheimer type II change.5

Magnetic resonance (MR) imaging studies including MR spectroscopy, magnetization transfer (MT), and diffusion-weighted imaging (DWI) have improved our understanding of basic neuroanatomical and pathophysiological alterations in patients with hepatic encephalopathy (HE).6–12 High signal intensity in the globus pallidus, putamen, and portions of the internal capsule on T1-weighted imaging along with brain atrophy, particularly in frontal lobes, has been described in 50% to 88% of patients with liver cirrhosis and HE.8, 13, 14 Although various causes have been proposed8, 14 for this hyperintensity, deposition of manganese is regarded as the most likely explanation,12 with no direct correlation of pallidal hyperintensity and grade of encephalopathy.15

MT imaging has shown decreased MT ratio (MTR) in all brain regions in patients with chronic HE.10, 16 The likely explanation proposed is Alzheimer type II change and increased water content in astrocytes.16 MR spectroscopy (MRS) permits detection and quantification of brain metabolites in vivo. In HE, an increase in glutamate/glutamine (Glx), taurine, and decreases in myo-inositol (mI) and choline (Cho) are observed on MRS.12, 17 The changes in the mI in MRS are considered highly suggestive of cellular edema.7, 18

Use of DWI allows assessment of intracellular and extracellular water content in the brain, which would help in differentiating cytotoxic from vasogenic edema.19 Diffusion tensor imaging (DTI) measures different DTI metrics: mean diffusivity (MD), an index of water movement across cell membranes, and fractional anisotropy (FA), an index of microstructural integrity of the brain white matter.20 Using DWI, Lodi et al.21 have studied patients with postnecrotic cirrhosis and HE and reported increased diffusivity of water in brain parenchyma.21

We prospectively performed DTI in 39 patients with cirrhosis with and without HE to look for possible microstructural changes in the white matter as suggested by measuring FA and MD values in the brain. Furthermore, we correlated these indices with grades of encephalopathy as well as the neuropsychological (NP) test scores. In 10 of 24 patients with minimal HE (MHE), the imaging and NP test were repeated after 3 weeks of treatment with lactulose22 to look for any significant reversibility of these parameters.


CE, cerebral edema; MR, magnetic resonance; MT, magnetization transfer; DWI, diffusion-weighted imaging; HE, hepatic encephalopathy; MTR, MT ratio; MRS, MR spectroscopy; Glx, glutamate/glutamine; MI, myo-inositol; Cho, choline; DTI, diffusion tensor imaging; MD, mean diffusivity; FA, fractional anisotropy; NP, neuropsychological; MHE, minimal hepatic encephalopathy; MRI, magnetic resonance imaging; NCT, number connection test; FCT, figure connection test; WAIS-P, Wechsler Adult Intelligence Scale; PC, picture completion; OA, object assembly; BD, block design; TR, repetition time; TE, echo time; NEX, number of excitations; ROI, region of interest; CC, corpus callosum; LIC, left internal capsule; RIC, right internal capsule; CN, caudate nucleus; FWM, frontal white matter; OWM, occipital white matter; ADC, apparent diffusion coefficient.

Subjects and Methods


Patients with liver cirrhosis attending the gastroenterology department were enrolled according to criteria mentioned later. Thirty-nine patients [35 men, median age 45 years; 4 women, median age 44 years (range, 16-65)] with liver cirrhosis of various causes [alcohol (n = 14), hepatitis b (n = 7), hepatitis c (n = 5), combined hepatitis C virus + alcohol (n = 2), hepatitis B virus + alcohol (n = 1), autoimmune (n = 2), and cryptogenic (n = 8)] with and without HE were studied. Those with a history of alcohol consumption within the past 6 months, drug abuse, psychiatric or neurological illness, head injury, advanced pulmonary disorders, or renal or metabolic disorders were excluded from the study. Patients with grades 3 and 4 encephalopathy and patients requiring sedation for MR imaging (MRI) also were excluded. Cirrhosis was diagnosed on the basis of suggestive clinical and imaging features.23 The functional status of patients was assessed by the Child-Pugh score.24 Eight patients were classified as Child A, 12 as Child B, and 19 as Child C class. The study protocol was discussed in detail with all the patients, or guardians in case of patients who were unable to give consent, and healthy controls. Written consent was obtained from patients as well as healthy controls. The study was approved by the institutional ethics committee. All patients underwent detailed clinical examination, including neurological examination. Overt HE was graded as per West Haven criteria.25 NP tests were performed in all patients with normal neurological examination results, and those with two or more abnormal NP test results were classified as MHE.26, 27

Eighteen healthy controls [13 men and 5 women, median age 45 years (range, 24-60 years)] were recruited as healthy controls. None had any history of neurological or psychiatric illness, alcohol or drug abuse, head injury, or liver disease. All underwent detailed clinical assessment, including neurological examination.

Ten MHE patients and 10 age- and sex-matched healthy controls (part of the 18 healthy controls) had repeat imaging and NP tests after 3 weeks of lactulose to look for reversibility of abnormalities in these parameters, using the same protocol as described later. Lactulose was used in 30- to 90-mL/day divided doses to produce two to three loose stools daily.

NP Tests.

NP tests included number connection tests (NCT A and B), figure connection tests (FCT A and B),26, 28 and the performance subset of the modified Wechsler Adult Intelligence Scale (WAIS-P, modified for Indian population).29 Normal values for these tests were derived from 250 healthy volunteers, and the test score was considered as abnormal if values were more than the mean + 2 SD. These tests have subsequently been validated and used in several other studies.26, 30

WAIS-P tests visuospatial capacity and visuomotor speed on relatively novel problems. Tests included in WAIS-P were picture completion (PC), digit-symbol (DS), picture arrangement (PA), object assembly (OA) and block design (BD). The procedure of performing these tests and functions assessed by these tests are described elsewhere.26, 28

In NCT and FCT, lower scores represent better performance, whereas in the other NP tests a higher score represents better performance. Total duration for the entire battery of NP tests ranged from 45 to 60 minutes in patients and from 35 to 45 minutes in healthy controls.

On the basis of NP tests, patients were grouped as no HE (n = 5) and MHE (n = 24). Patients with overt HE were grouped into HE grade 1 (n = 5) and HE grade 2 (n = 5) on the basis of West Haven criteria.25 All except one patient with overt HE had type C, episodic, and recurrent HE, having experienced two to four episodes of HE since developing liver decompensation. One patient had persistent HE and was receiving treatment with L-ornithine L-aspartate. Among patients with MHE, six had experienced prior episodes of type C recurrent HE, and 18 patients had never experienced overt HE. Patients were followed for a median duration of 12 months (range, 1.5-27 months).

All patients and healthy controls underwent brain MRI either before or after NP testing. NP tests were performed by two observers (R.A.K. and P.R.) and interpreted by an experienced gastroenterologist (V.A.S.). MRI was analyzed by two separate observers (R.T. and A.M.M.) who were blinded to the NP test results. Final verification and correlation of results were performed separately by two observers (R.K.G. and V.A.S.).

MRI Protocol.

Brain MRI was performed on a 1.5 tesla GE scanner (General Electric Medical System, Milwaukee, WI) equipped with an actively shielded gradient set with a maximum strength of 33 mT/m. A standard quadrature birdcage receive and transmit radio frequency head coil was used. The conventional MRI protocol included T2-weighted fast spin echo sequence with repetition time (TR in ms)/echo time (TE in ms)/echo train length/number of excitations (NEX) = 6,000/200/16/2 and T1-weighted spin echo sequence with TR/TE/NEX = 500/14/1. Fast spin echo T2-weighted MRI was done to reduce total MRI time for sick patients (grade 1, grade 2). This was maintained for the remaining groups as well as healthy controls for the sake of consistency in imaging protocol.

DTI was acquired using a single-shot echo planar dual spin echo sequence with ramp sampling.20 A balanced rotationally invariant31 dodecahedral diffusion encoding scheme with 10 uniformly distributed directions over the unit sphere was used for obtaining the diffusion-weighted and encoded data. The b-factor was set to 1,000 s/mm2, TR = 8 seconds, TE = 100 ms. A total of 36 axial sections were acquired with image matrix of 256 × 256 (following zero-filling) slice thickness of 3 mm with no inter-slice gap, and a field-of-view of 240 × 240 mm2. To enhance the signal-to-noise ratio and reduce the phase fluctuations, magnitude-constructed images were repeated (NEX = 8) and temporally averaged. Total duration for the entire MRI protocol was 12 to 15 minutes.

DTI Data Analysis.

The DTI data were processed as described in detail elsewhere.31 Briefly, after image cropping and distortion corrections, the data were interpolated to attain isotropic voxels and decoded to obtain the tensor field for each voxel. The tensor field data were then diagonalized using the analytical diagonalization method31 to obtain the eigenvalues (λ1, λ2, and λ3) and the three-orthonormal eigenvectors. The tensor field data and eigenvalues were used to compute the DTI metrics such as MD and FA for each voxel.

We selected major white matter and deep gray matter regions for region-of-interest (ROI) analysis, because it has been reported in the earlier study that the changes are widespread in both the gray and the white matter in patients with HE.17 The corpus callosum (CC) was divided into seven segments, based on the scheme proposed by Witelson (Figure 2 in Hasan et al.31), to measure the specific regions in a manner that approximately represents the CC connections hypothesized across cortical brain regions.32–34 The CC is associated with connecting the hemispheres, and any changes in this largest commissural fiber bundle may be associated with NP abnormalities.35 A single midsagittal slice where the fornix and massa intermedia were clearly visible was used to measure the seven regional areas within the CC. To facilitate ROI placement for quantitative analysis, the DTI-derived FA and MD maps were displayed and overlaid on images with different contrasts in three orthogonal planes. Rectangular ROIs were placed in the seven segments of CC for FA and MD quantification. The FA and MD values in different regions of CC (rostrum, genu, rostral body, anterior mid-body, posterior mid-body, isthmus, and splenium) internal capsule (left and right posterior limb of internal capsule, left and right anterior limb of internal capsule, left and right genu of internal capsule), right and left caudate nuclei, right and left putamen, right and left frontal and occipital white matter in patients as well as in healthy controls were measured (Fig. 1A). The size of the ROI varied from 2 × 2 and 6 × 6 pixels, with shape varying from elliptical to rectangular.

Figure 1.

Healthy control and a patient with grade 1 hepatic encephalopathy (HE). (A) Region-of-interests (ROIs) placement on the color-coded fractional anisotropy (FA) map fused with mean diffusivity (MD) map in brain (red, blue, and green represent the right to left, superior-inferior, and anterior-posterior directions of the white matter fibers, respectively). (B) MD map from the healthy control shows normal signal intensity in all the ROIs visualized on the color-coded FA map fused with MD map in (A). (C) MD map from the grade 1 HE patient shows subtle increase in the signal intensity in all the ROIs with significantly increased MD values on quantification compared with MD values of healthy control (B) in all the ROIs (Table 1).

Table 1. MD and FA Values in Healthy Controls and Patients With Cirrhosis With No HE and With Different HE Grades
RegionsHealthy Controls (n = 18)No HE (n = 5)MHE (n = 24)HE Grade 1 (n = 5)HE Grade 2 (n = 5)P-value*
  • Abbreviations: HE, hepatic encephalopathy; MHE, minimal hepatic encephalopathy; CC, corpus callosum; RIC, right internal capsule; LIC, left internal capsule; CN, caudate nuclei; P, putamen; FWM, frontal white matter; OWM, occipital white matter.

  • *

    Using analysis of variance (ANOVA).

Mean (±SD) MD values (× 10−3 mm2/s)
CC1.05 ± 0.011.17 ± 0.011.11 ± 0.011.12 ± 0.011.13 ± 0.00.001
RIC1.04 ± 0.011.09 ± 0.011.11 ± 0.011.12 ± 0.011.16 ± 0.00.009
LIC1.06 ± 0.011.12 ± 0.011.09 ± 0.011.10 ± 0.011.18 ± 0.00.025
CN1.08 ± 0.011.17 ± 0.011.17 ± 0.011.21 ± 0.011.22 ± 0.01.000
P1.03 ± 0.011.06 ± 0.011.10 ± 0.011.13 ± 0.011.18 ± 0.01.000
FWM1.10 ± 0.011.10 ± 0.011.15 ± 0.011.23 ± 0.011.25 ± 0.01.000
OWM1.05 ± 0.011.08 ± 0.011.10 ± 0.011.13 ± 0.011.15 ± 0.01.050
Mean (±SD) FA values (arbitrary unit)
CC0.65 ± 0.050.51 ± 0.290.63 ± 0.050.64 ± 0.060.62 ± 0.08.247
RIC0.51 ± 0.080.42 ± 0.230.53 ± 0.550.59 ± 0.020.54 ± 0.03.143
LIC0.50 ± 0.070.40 ± 0.220.53 ± 0.050.57 ± 0.030.54 ± 0.03.060
CN0.12 ± 0.010.11 ± 0.050.14 ± 0.030.14 ± 0.020.12 ± 0.02.079
P0.10 ± 0.020.07 ± 0.040.11 ± 0.010.12 ± 0.010.11 ± 0.02.016
FWM0.28 ± 0.040.22 ± 0.120.27 ± 0.050.29 ± 0.050.27 ± 0.06.696
OWM0.35 ± 0.060.31 ± 0.170.32 ± 0.510.33 ± 0.070.31 ± 0.08.985

Statistical Analysis.

MD and FA values for different regions of CC and genu, and anterior and posterior limbs of right and left internal capsule were averaged to get values for CC, LIC (left internal capsule), and RIC (right internal capsule). Left and right measurements for all other regions were pooled together and averaged values for caudate nuclei (CN), putamen (P), frontal white matter (FWM) and occipital white matter (OWM) calculated for the final data set for statistical analysis. To see the changes between values of the initial study and values after lactulose therapy in MD and FA, a paired t test was applied to the observations obtained from all ROIs. Independent sample t test was applied to compare these values between healthy controls and MHE. ANOVA and post-hoc tests were performed to detect the overall difference in the FA and MD values for each ROI between various study groups. Spearman's rank correlation coefficient was used to explore the association between NP scores and FA and MD values after ranking the NP scores and FA and MD values. A P-value less than .05 was considered statistically significant. Statistical Package for the Social Sciences 13.0 (SPSS Inc, Chicago, IL) was used for statistical computations.


All patients showed no abnormality on conventional imaging, except one who had increased signal intensity in the globus pallidus on T1-weighted images. After neurological and neuropsychiatric evaluation, 24 of 39 patients studied were found to have MHE whereas 5 each had no HE, grade 1 HE, and grade 2 HE.

Imaging With Grades of HE.

We compared MD values from all groups of patients and healthy controls using ANOVA, and the P-value obtained was indicative of significant differences among the groups (Table 1).

Only caudate nuclei of no HE patients showed a significant increase in the MD values relative to healthy controls (Fig. 1B). MD values of MHE patients were significantly higher as compared with healthy controls (Fig. 1B) in five brain regions (CC, RIC, LIC, CN, and OWM). All seven brain regions (CC, RIC, LIC, CN, P, FWM, and OWM) showed significant increased MD in patients with HE grade 1 (Fig. 1C) and grade 2 compared with healthy controls (Fig. 1B). MD values showed an increase from no HE to HE grade 2 (Table 1).

We did not observe a significant difference in FA values between no HE, different HE grades, and healthy controls except in P (P = .016) using ANOVA (Table 1). In all brain regions, it showed an initial decrease from healthy controls to no HE followed by an increase/decrease (insignificant) in the rest of the grades, suggesting an inconsistent pattern across the grades of HE (Table 1).

NP Tests Relation With DTI Metrics.

In all patients with no HE (n = 5) and MHE (n = 24), the mean scores of NCT A, NCT B, FCT A, and FCT B were 82, 98, 73, and 111 seconds, respectively. Scaled scores for WAIS-P tests PC, DS, BD, PA, and OA were 7.5, 6.4, 10, 8, and 8, respectively. The Spearman's rank correlations between MD and FA values of various ROIs with different NP tests are summarized in Table 2.

Table 2. Spearman's Rank Correlation Coefficients of MD and FA Values With Neuropsychological Tests in Patients With No HE (n = 5) and MHE (n = 24)
  • Abbreviations: HE, hepatic encephalopathy; MHE, minimal hepatic encephalopathy; NCT A, number connection test A; NCT B, number connection test B; FCT A, figure connection test A; FCT B, figure connection test B; PA, picture arrangement; OA, object assembly; PC, picture completion; BD, block design; DS, digit symbol; CC, corpus callosum; RIC, right internal capsule; LIC, left internal capsule; CN, caudate nuclei; P, putamen; FWM, frontal white matter; OWM, occipital white matter.

  • *

    Significant at .05 level.

Correlation With MD
Correlation With FA

NCT A test showed significant strong positive correlation with the mean MD values for CC and RIC, whereas NCT B showed a significant strong positive correlation only with CC. FCT A showed a significant strong correlation with MD values for CC and RIC, and FCT B test showed significant strong correlation with CC. OA, PC, BD, and DS tests showed significant strong correlations with MD values of CC. The PA test did not showed a significant correlation with MD values from any regions (Table 2).

The correlation between FA values of various ROIs with different NP tests is also summarized in Table 2. There was no significant and strong correlation of NP tests with FA values from different regions.

Reversibility of Brain Edema in 10 MHE Patients on Serial Study.

No significant difference was found in the MD values of healthy controls on initial and follow-up study after lactulose therapy. However, a significant decrease occurred in MD values on follow-up study compared with initial study in 10 patients of MHE after lactulose therapy (Table 3) (Fig. 2A). MD values of MHE patients were significantly higher compared with healthy controls on initial study, whereas no significant difference was seen in MD values between MHE patients and healthy controls after lactulose therapy (final) (Table 3).

Figure 2.

Reversibility of MD values and neuropsychological tests in patient with minimal HE (MHE) following lactulose therapy. (A) Significant decrease in MD values in all seven ROIs (CC, corpus callosum; RIC, right internal capsule; LIC, left internal capsule; CN, caudate nuclei; P, putamen; FWM, frontal white matter; OWM, occipital white matter) is shown in 10 patients with MHE after 3 weeks of lactulose therapy (*P = .01; **P = .005). (B) Figure connection tests (FCT A and B) show significant reduction in score in patients with MHE after 3 week of lactulose therapy, suggesting neuropsychological improvement (**P < .005).

Table 3. MD and FA Values in Healthy Controls and Patients With Minimal Hepatic Encephalopathy (MHE) Before (Initial) and After (Final) 3 Weeks of Lactulose Therapy
RegionsHealthy Controls (n =10)Minimal Hepatic Encephalopathy (n = 10)Statistical Significance (P)
(A) Initial(B) Final(C) Initial(D) FinalA vs. B*C vs. D*A vs. CB vs. D
  • Abbreviations: CC, corpus callosum; RIC, right internal capsule; LIC, left internal capsule; CN, caudate nuclei; P, putamen; FWM, frontal white matter; OWM, occipital white matter.

  • *

    Paired sample t test.

  • Independent sample t test.

Mean (±SD) MD values (×10−3 mm2/s)
CC1.23 ± 0.071.16 ± 0.081.35 ± 0.101.04 ±
RIC1.03 ± 0.061.00 ± 0.041.29 ± 0.151.02 ± 0.04.591.000.000.348
LIC1.02 ± 0.059.98 ± 0.041.37 ± 0.191.05 ± 0.07.591.000.000.070
CN1.11 ± 0.051.07 ± 0.041.19 ± 0.061.11 ±
P1.05 ± 0.041.02 ± 0.041.05 ± 0.040.96 ± 0.06.343.010.001.330
FWM1.11 ± 0.041.09 ± 0.041.11 ± 0.031.02 ±
OWM1.09 ± 0.041.05 ± 0.061.18 ± 0.091.09 ±
Mean (± SD) FA values (arbitrary unit)
CC0.51 ± 0.040.49 ± 0.030.52 ± 0.040.51 ± 0.04.300.308.481.321
RIC0.48 ± 0.060.44 ± 0.020.49 ± 0.060.48 ±
LIC0.48 ± 0.040.46 ± 0.010.48 ± 0.030.48 ±
CN0.12 ± 0.010.12 ± 0.010.11 ± 0.020.12 ± 0.02.854.191.643.933
P0.10 ± 0.020.11 ± 0.010.10 ± 0.020.10 ±
FWM0.29 ± 0.040.29 ± 0.040.28 ± 0.040.29 ± 0.03.815.380.423.797
OWM0.37 ± 0.080.36 ± 0.060.33 ± 0.030.37 ± 0.08.348.571.122.793

There were no significant differences in FA values between MHE (initial vs. final), healthy controls (initial vs. final), initial (healthy controls vs. MHE), and final (healthy controls vs. MHE) (Table 3) studied.

On initial study, in 10 patients with MHE, mean scores of NCT A, NCT B, FCT A, FCT B were 112, 140, 227, and 194 seconds, respectively, and scaled scores for the WAIS-P tests PC, DS, BD, PA, and OA were 7, 6, 6.5, 8, and 4, respectively. After lactulose therapy in 10 patients with MHE, mean scores of NCT A, NCT B, FCT A, and FCT B were 77, 85, 127, and 106 seconds, respectively, and scaled scores for the WAIS-P tests PC, DS, BD, PA, and OA were 13, 6.5, 12, and 10, respectively. Significant improvement (paired t test) was seen after 3 weeks of lactulose therapy in mean score of FCT A (P = .001), FCT B (P = .004) (Fig. 2B), PC (P = .0001), BD (P = .001), PA (P = .008), and OA (P = .0001) except for NCT A (P = .06), NCT B (P = .08), and DS (P = .5) tests, which showed insignificant improvement.

MD values of CC of MHE patients (n = 10) showed strong significant correlation with NCT A, NCT B, FCT A, FCT B, OA, PC, BD, and DS tests. NCT A and FCT A correlate significantly with MD values from RIC. MD values of CC of MHE patients (after lactulose therapy) showed a significantly strong correlation only with NCT A, FCT A, FCT B, PA, and OA tests. MD values from rest of the regions did not show significant correlation with any NP tests. There was no significant strong correlation of FA values and NP tests on initial study and after 3 weeks of lactulose therapy.


Our study showed significantly increased MD values in all seven brain regions in patients with no HE and different grades of HE with no significant decrease in FA values compared with healthy controls. This suggests an increase in the interstitial brain water in patients with HE. Comparison of MD values between healthy controls and cirrhosis with no HE and with HE of different grades (results not shown) showed a number of regions with a significant increase in MD values. These values increased progressively, from 1 in patients with cirrhosis with no HE to 7 in grade 2 HE. This suggests that, with increasing grade of encephalopathy, brain water content increases significantly and progressively affects more regions of the brain. Lodi et al.21 noted that the highest apparent diffusion coefficient (ADC) values were observed in all ROIs of a single patient with grade 2 HE and were higher than values in grade 0 and 1 HE. They observed significant elevation in ADC values in all the regions except thalamus in combined data of 14 patients with cirrhosis with different grades of HE.

Hyperammonemia, secondary to hepatic dysfunction, results in profound changes in astrocyte morphology (Alzheimer type II changes in chronic HE) and function.1 Ammonia detoxification takes place in astrocytes through the action of glutamine synthetase.2 In chronic HE, there is enough time for activation of effective compensatory mechanisms of cellular adaptation to the osmotic changes.3 Increase in intracellular glutamine, due to conversion of glutamate to glutamine by the hyperammonemia, triggered glutamine synthetase pathway and resulted in intracellular depletion of mI and choline in an osmoregulatory attempt by the astrocytes.6 We do not know the exact mechanism of this increased extracellular fluid in HE, and this will be the subject of our future studies. However, we speculate that the extracellular migration of the macromolecules during this cellular osmoregulatory attempt may result in increased accumulation of extracellular fluid, accounting for the increase in MD documented in the current study. Lodi et al21 also noted increased ADC but have tried to explain it on the basis of increased astrocyte cell volume. This explanation is in contradiction to the basic understanding of diffusivity, where an increased cellular volume is associated with increased extracellular space tortuosity and a decrease in ADC.36 In fact, increased cell volume is associated with reduced ADC. The validity of this explanation is corroborated by our observation of reduced ADC in fulminant hepatic failure–induced HE that returned to normal after recovery.4

The depletion of mI and Cho along with Glx peak in patients with cirrhosis with or without HE has been well described in the literature using in vivo MRS, an observation best explained on the basis of osmoregulatory shift of osmolytes to compensate for Gln accumulation.12, 17 Moreover, the exposure of cultured astrocytes to ammonia in vitro results in astrocyte swelling with Alzheimer's type II change.37 We suggest that the culture astrocyte model cannot simulate the situation seen in vivo. In addition, astrocytic Alzheimer's type II changes have hitherto not been shown to be reversible, contrary to HE, which is known to change from higher to lower grade, or vice versa, and also recovers completely after liver transplantation.17 We believe that functional changes that are shown by NP tests are probably related to the increased interstitial edema with a possibility of reversibility, as was observed in 10 patients with MHE in the current series.

A significant increase in MD value in one of the seven regions measured in no HE patients compared with healthy controls suggests that even before the appearance of changes on clinical and NP tests, there is detectable brain edema using DTI. This alteration is analogous to changes in mI, Cho, and Glx on in vivo MRS7 that have been described in patients with liver cirrhosis without HE. The observed changes in MD before changes in NP scores are in agreement with the spectroscopy data and suggest that these MR techniques may be the most sensitive tests for detecting MHE.

Increases in MD values have also been described secondary to gliosis38 as well as neuronal loss.20 Preservation of axonal density and lack of demyelinating process has been supported by the normal N-acetylaspartate levels observed in patients with HE.17 The observation of an increase in MD with normal FA values in various regions in this study also supports the absence of gliosis in these patients, because gliosis is associated with an increase in MD and decrease in FA.38

Astrocytes in patients with chronic liver failure display a reduced expression of specific glial fibrillary acid protein (GFAP).39 This is a cytoplasmic filamentous protein that constitutes a major component of the cellular cytoskeleton in mature astrocytes.40 GFAP plays a central role in astrocyte volume and shape regulation.41 Morphological changes affecting astrocytes also could contribute to the ADC changes found in patients with cirrhosis because diffusion in the extracellular space can be heavily influenced by the cell shape.42 Substantiating the contribution of altered membrane permeability to MD values observed in patient groups is difficult.

In the current study, the significant changes in the FA values were not observed in patients with HE relative to healthy controls. FA is considered to be an indicator of white matter microstructural integrity and can be used as a predictor of microstructural changes in biological systems.20 Our observations of unaltered FA values along with high MD suggest an increased brain edema without a concomitant alteration in white matter structural integrity. This has been further confirmed by the reversibility of the MD values in 10 patients with MHE with insignificant change in FA value. The change in MD value with insignificant change in FA may be used as an indicator of reversibility in this group of patients.

Low MTRs have been reported in various regions of the brain in patients with HE.16 A number of pathological conditions such as gliosis and demyelination are known to result in decreased MTR.16, 43 Restoration of initial low MTR values after liver transplantation support the hypothesis that the changes in the brain parenchyma are attributable to mild reversible cerebral edema rather than any structural lesions.17

We have found a significantly strong correlation between increased MD values and NP test scores mainly in the CC of patients with HE. The speed of information processing, executive functions, and visual spatial functions were affected in these patients as assessed by NP tests and showed significant correlation with increased MD in CC on initial study. Because different lobes of the brain contribute to the different regions of CC, the dysfunction of the lobe may be reflected in the corresponding region of CC.35, 44 Our results suggest that the abnormality in CC reflects the dysfunction in various lobes of the brain parenchyma in patients with chronic HE, which are involved in the execution of these NP tests. Persistence of significant strong correlation of NP tests with CC in these 10 patients with MHE showing reversibility of function after lactulose therapy further confirms that these NP tests analyze the functionality of the CC.

Our results indicate that corticospinal tract shows increased MD values in different grades of HE (Table 1). Selective involvement of the white matter within or close to the corticospinal tract has been observed in patients with cirrhosis using transcranial magnetic stimulation, with subsequent improvement after liver transplantation consistent with reversibility of mild edema in chronic liver disease.45 Pyramidal signs are frequently observed in HE, and the selective involvement of the corticospinal tract in a preclinical state of HE is an expected result.46 The reasons for this greater vulnerability of the corticospinal tract are unknown, but they may include higher energy demands and higher susceptibility to excitotoxicity.47

Overt encephalopathy can be diagnosed using clinical criteria, but MHE requires NP tests for its diagnosis. Results of NP tests can be influenced by variables such as age, educational status, and mental condition of the patient during the test, and learning effects. In addition, no consensus has been reached on the type and number of tests required for diagnosis. DTI may be well suited for evaluating the efficacy of any therapeutic intervention, because it is more objective compared with NP tests. However, more studies are required to address these issues.

In conclusion, an increase in MD with no significant changes in FA values in different grades of HE suggests reversible interstitial cerebral edema. These changes in MD values, especially in CC, show significant correlation with various NP tests.


Richa Trivedi, Asht Mangal Mishra, and Piyush Ranjan acknowledge the financial assistance from Council of Scientific and Industrial Research, New Delhi, India.