Abstract
Background: Brain edema and increased intracranial pressure worsen prognosis in patients with end-stage chronic cirrhosis.
Objective: To use diffusion-weighted imaging (DWI) to quantify water apparent diffusion coefficient (ADC) in different brain regions of patients with chronic liver failure with or without hepatic encephalopathy.
Methods: The authors studied 14 patients with viral liver cirrhosis and 12 sex- and age-matched healthy volunteers. Seven patients had no clinical evidence of hepatic encephalopathy; six had grade I hepatic encephalopathy; and one had grade II hepatic encephalopathy. Brain DWI was obtained using a single-shot echo-planar imaging sequence, and four gradient strengths (b values = 0, 300, 600, and 900 s/mm2) were applied to calculate the average diffusivity maps.
Results: Mean ADC values in the brains of patients with cirrhosis were significantly increased in all selected regions of interest (caudate, putamen, and pallidus nuclei; occipital, parietal, and frontal lobe white matter) except in the thalamus. Venous ammonia was linearly related to ADC values in deep gray and white matter regions of interest.
Conclusions: Brain water apparent diffusion coefficient is increased in patients with chronic liver disease and may be useful in monitoring patients with hepatic encephalopathy.
Brain edema and increased intracranial pressure, a major complication of fulminant hepatic failure,1 also occur in patients with end-stage cirrhosis and adversely affect prognosis.2 There is no effective treatment for brain edema in patients with cirrhosis, and liver transplantation is the only successful intervention.2 More recently, the presence of mild brain edema even in patients with cirrhosis with no evidence of encephalopathy has been hypothesized based on increased T2-weighted signal intensity3 and reduced magnetization transfer ratios4 in the hemispheric white matter (WM).
MR spectroscopy (MRS) studies have identified neurometabolic abnormalities that could well contribute to brain edema development in patients with cirrhosis.5–8⇓⇓⇓ MRS studies of the brains of patients with cirrhosis with clinical and subclinical hepatic encephalopathy (HE) consistently found an increase in glutamine5–7⇓⇓ associated with a reduction in glutamate7 and myo-inositol, a cerebral osmolyte.6 Glutamine increase, the result of ammonia detoxification in astrocytes, has been implicated as the central event determining cell swelling and increased brain water content in patients with HE.9
Diffusion-weighted imaging (DWI) allows the assessment of the water apparent diffusion coefficient (ADC), a measure of tissue water diffusivity. ADC depends on the interactions between water molecules and the chemical environment and the structural barriers at the cellular and subcellular level hindering their motion in vivo.10
Given the limited efficacy of therapy when the stage of overt cerebral edema and intracranial hypertension is reached, detection of early signs of brain edema or changes in water distribution at the microscopic level can be critical for the treatment of patients with chronic liver disease.2 In the present study, we used DWI to assess water ADC in different brain regions in patients with chronic liver failure either with or without clinical evidence of HE.
Methods.
Patients.
Fourteen patients with viral liver cirrhosis and 12 sex- and age-matched healthy volunteers (9 men; aged 58 ± 8 years) were studied (table). Patients with a history of drug abuse and those with neurologic or psychiatric diseases unrelated to liver failure were excluded. Etiology was hepatitis C virus (HCV) in eight patients, hepatitis B virus (HBV) in five patients, and HBV and hepatitis D virus (HDV) in one. Based on Parsons-Smith criteria, HE was grade 0 in seven patients, grade I in six patients, and grade II in one patient. No patients with cirrhosis with severe HE (grades III or IV) were included in the study. The degree of liver failure was moderate in five patients (Child–Pugh B) and severe in nine patients (Child–Pugh C). Laboratory screening, which included albumin, bilirubin, prothrombin time, and serum venous ammonia levels, was performed the same day as the MRI scans.
Table Clinical/laboratory features of the cirrhotic patients at the time of the MRI scan and clinical follow-up
MRI.
MRI was performed using a 1.5-T GE Signa Horizon LX system (Milwaukee, WI) equipped with a birdcage head radiofrequency coil for signal reception and an EchoSpeed gradient system providing a maximum gradient strength of 22 mT/m and maximum slew rate of 120 mT/m/ms.
Structural images and DWIs were obtained from the same 18 to 24 axial slices with 5-mm thickness and 1-mm interslice gap.
T1-weighted structural imaging was performed with a spin-echo sequence with a flip angle alpha of 90°, a repetition time (TR) of 500 ms, an echo time (TE) of 10 ms, an isotropic spatial in-plane resolution of 0.94 mm, and two signal averages.
DWI10 was conducted using a spin-echo single-shot echo planar imaging (EPI) technique11 with a pair of Stejskal–Tanner diffusion-weighting gradient pulses.12 The EPI sequence was performed with alpha = 90°, TR = 10 seconds, TE = 100 ms, an in-plane resolution of 2.5 mm, and phase encoding in anterior-posterior direction. The diffusion-weighting gradients were applied on each of the three physical axes x, y, and z in separate scans. Three different gradient strengths were chosen corresponding to b-factor values of 300, 600, and 900 s/mm2. In addition, images without diffusion weighting were acquired corresponding to b = 0 s/mm2 and exhibiting a T2-weighted contrast.
Data processing and evaluation.
In general, diffusion-weighted EPIs suffer from distortions because of eddy currents generated by the large gradients applied for diffusion weighting.13 In this study, distortions were corrected by slice-wise registration of the DWIs onto the T2-weighted EPIs using the image registration software FLIRT (www.fmrib.ox.ac.uk/fsl). Because of the nature of the distortions, the degrees of freedom were restricted to translation, scaling, and shearing along the phase-encoding direction, as Haselgrove et al. also reported.13
Assuming a signal attenuation depending monoexponentially on the b value, the ADC of each direction was determined pixel-wise using a least-squares fit. By calculating the mean of the three directions, the ADC trace map was generated.
To avoid contamination of the ADC values for gray matter and WM by the much higher values of CSF during further evaluation, CSF was removed from the ADC map. This was accomplished using the FAST algorithm (www.fmrib.ox.ac.uk/fsl) for a two-class segmentation based on the corresponding T2-weighted EPIs. Finally, using FLIRT, the diffusion data were registered onto the T1 scan, and a region of interest (ROI) then was selected. Figure 1 illustrates the brain areas defined for ADC calculation: left and right caudate, putamen, pallidus, and thalamus nuclei, and occipital, parietal, and frontal WM. The spatial resolution of standard DWIs does not allow a reliable measurement of ADC values in the cerebral cortex because partial volume effects cannot be avoided. Therefore, cortical ROIs were not selected in the present study.
Figure 1. Evaluation of the apparent diffusion coefficient (ADC) maps. Two slices from the T1-weighted spin-echo scan, the T2-weighted echoplanar imaging scan, and the CSF-masked ADC map from a healthy subject are shown. The regions of interest (ROIs) were initially selected on the T1-weighted image, controlled with the T2-weighted image, which has the same distortions from field inhomogeneities as the diffusion images, and then pasted onto the ADC map to determine the mean values in each ROI. 1 = caudate nucleus; 2 = putamen; 3 = pallidus; 4 = thalamus; 5 = occipital white matter; 6 = frontal white matter; and 7 = parietal white matter.
Significance, determined by the nonparametric Mann–Whitney U test, was taken as p < 0.05. Linear regression analysis was used to calculate correlation coefficients.
Results.
T2-weighted EPIs did not show signal intensity abnormalities in any of the patients with cirrhosis except for Patient 2. This patient had hemispheric WM high-signal intensity (figure 2) consistent with brain edema and died 2 weeks after the scan.
Figure 2. T2-weighted echoplanar imaging scan and the CSF-masked apparent diffusion coefficient map obtained in Patient 2. Areas of increased signal intensity are evident in the hemispheric white matter on the T2-weighted image.
Water ADC values were similar in corresponding right and left hemisphere ROIs in healthy subjects and patients and therefore are reported as mean values.
In patients with cirrhosis, mean ADC values were increased in the caudate nucleus (0.79 ± 0.06 vs 0.74 ± 0.02 in control subjects; p = 0.009), putamen (0.77 ± 0.05 vs 0.72 ± 0.01; p < 0.0001), and in the pallidus (0.80 ± 0.05 vs 0.72 ± 0.02; p < 0.0001), and WM ROIs such as occipital (0.83 ± 0.10 vs 0.77 ± 0.03; p = 0.01), parietal (0.79 ± 0.09 vs 0.73 ± 0.02; p = 0.01), and frontal lobes (0.80 ± 0.07 vs 0.75 ± 0.03; p = 0.04; figure 3). In the patients, ADC increase did not reach significance in the thalamus (0.76 ± 0.05 vs 0.74 ± 0.02; p = 0.1; see figure 3). Patient 2, the only patient with grade II encephalopathy, presented the highest ADC values in all ROIs. Conversely, no significant difference was found in brain ADC values between patients with encephalopathy grades 0 and I (data not shown).
Figure 3. Mean ± SD apparent diffusion coefficient values in the deep gray and white matter regions of interest in patients and control subjects (***p < 0.0001; **p < 0.01; and *p < 0.05).
In the patients, venous ammonia showed a linear relationship with ADC values in the putamen (r = 0.65; p = 0.01), pallidus (r = 0.55; p = 0.04), thalamus (r = 0.54; p = 0.04), occipital WM (r = 0.71; p = 0.004), parietal WM (r = 0.82; p < 0.001), and frontal WM (r = 0.61; p = 0.02), whereas it failed to reach the significance in the caudate nucleus (r = 0.42; p = 0.1). A linear relationship also was found between albumin concentration and ADC values in the putamen (r = −0.56; p = 0.03) and pallidus (r = −0.60; p = 0.02) and between ADC values in the pallidus and the Child–Pugh scores (r = 0.68; p = 0.007). No correlation was found between ADC values in the different brain areas and the other clinical scores and laboratory tests reported in the table (data not shown).
Mean ADC values in all selected brain areas were similar in patients with HBV- and HCV-positive etiology (data not shown).
Discussion.
DWI was used to measure water ADC in the brains of patients with cirrhosis with or without encephalopathy. ADC values were significantly increased in all WM and deep gray matter cerebral areas examined except for the thalamus, in which ADC increase failed to reach significance. ADC values in all the examined brain areas except the caudate nucleus showed a positive correlation with serum venous [NH4+] in the patients. Our present findings suggest that ammonia and related glutamine brain accumulation contribute to changes in brain water motility and content.
The increased ADC in the brains of our patients with cirrhosis can be caused by several pathologic changes occurring at the ultrastructural level, which may all have contributed to the resulting altered ADCs.
Interpretation of water diffusion in vivo is commonly derived from a two-compartment model comprising the extracellular space (ECS) and the intracellular space (ICS) with exchange between the two. It is in general assumed that water diffusion in the ECS is fast because of a low concentration of macromolecules and absence of membranous organelles. Conversely, water diffusion in the ICS is believed to be slow, being strongly restricted by physical (i.e., macromolecules and organelles) and chemical (specific binding and protein transitions and movements) factors.14 However, a number of studies, performed using b values higher than that used in the present study, showed that the relative fraction of fast and slow water apparent diffusion does significantly deviate from that of extracellular and intracellular compartment volumes, suggesting that other factors also may contribute to ADC changes.15–18⇓⇓⇓ Besides changes in extracellular vs intracellular volume ratio, alterations in cell membrane permeability to water and in intracellular and extracellular intrinsic ADC could contribute to the increased ADC found in the brains of the patients with cirrhosis.
Thus far, no studies have investigated changes in brain ECS in the presence of liver failure. Conversely, there is a large body of evidence indicating that cell volume homeostasis can be impaired in patients with chronic liver disease and that, in particular, astrocyte swelling occurs (see Haussinger et al.9 for a review). It has been suggested that ammonia-induced increase in brain water is mainly mediated by the intracellular accumulation of the organic osmolyte glutamine.19,20⇓ Astrocytes are the only brain cells containing glutamine synthetase.21 Increased synthesis of glutamine in patients with cirrhosis is the result of brain exposure to increased ammonia concentration and represents the major route of brain ammonia detoxification.5–7⇓⇓ Experimental studies on rats have shown that inhibition of glutamine synthesis with methionine-sulfoximine prevents the development of ammonia-induced brain edema19 and decreases astrocyte swelling.20 In patients with cirrhosis, the reduction of cerebral myo-inositol, an important organic osmolyte contributing to cell volume regulation in astrocytes22 and other osmolytes, counterbalances the effect of glutamine accumulation in the attempt to prevent astrocytes swelling.5,6,22⇓⇓
The positive correlation we found between serum [NH4+] and brain ADC values indicates that ADC changes in patients with cirrhosis are, at least in part, ammonia- and glutamine-mediated and that they can be related to brain cell swelling.23 A magnetization transfer study consistently found a correlation between globus pallidus contrast and blood ammonia levels in patients with cirrhosis.24 An increase in cell volume reduces the influence of restriction effects on intracellular diffusion pathways, and this is expected to increase ADC values. For patients with chronic liver disease, histopathologic abnormalities are found in astrocytes, which exhibit signs of Alzheimer type II degeneration (cellular swollen shape, nuclear enlargement, prominent nucleolus, and chromatin margination). In addition, astrocytes of patients with chronic liver failure display a reduced expression of the specific glial fibrillary acid protein (GFAP),25 a cytoplasmic filamentous protein that constitutes the major component of intermediate filaments in differentiated astrocytes.26 GFAP is a key component of the cellular cytoskeleton that plays a central role in astrocyte volume regulation and shape.27 Morphologic changes affecting astrocytes could contribute to the brain ADC changes found in patients with cirrhosis because diffusion in the ECS can be heavily influenced, besides by volume fraction, by cell shape.28
The dependence of the increased brain ADC values found in our patients on changes in cell membrane permeability is difficult to estimate because simulations have shown either a large18 or small29 influence of water membrane permeability on ADC values.
Consistent with the aforementioned biochemical abnormalities, recent MRI studies have suggested that the water content in the brains of patients with chronic liver failure may be increased. Magnetization transfer (MT) measurements showed reduced MT ratios in normal-appearing WM of patients with cirrhosis,4 and T2-weighted signal hyperintensities have been detected in the WM of patients with cirrhosis using the fast fluid-attenuated inversion recovery sequence.3 In our study using a T2-weighted EPI sequence, we detected T2-weighted signal hyperintensities in the hemispheric WM only in Patient 2. This patient had the highest ADC values in all examined brain areas and died shortly after the scan. This indicates that increased brain ADC in patients with cirrhosis is related to increased brain water content and that the measurement of brain ADC is critical for the management of chronic liver failure patients with or without MRI evidence of brain edema.
Acknowledgments
Supported by EU contract QLK4–2002–01763 (V Framework), Fondazione Cassa di Risparmio in Bologna, and Progetto Pluriennale di Ricerca E.F. 2000.
- Received July 30, 2003.
- Accepted October 6, 2003.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within the last 8 weeks.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.