Biexponential Analysis of Diffusion-Tensor Imaging of the Brain in Patients with Cirrhosis before and after Liver Transplantation

Discussion

In this study, MR imaging signs consistent with brain edema were present in the white matter of patients with cirrhosis. DTI and biexponential fitting found an increase in the fast MD values in the parietal white matter and corticospinal tract. These results are consistent with those of previous studies, in which DWI¹⁵ or DTI¹⁴ with monoexponential fitting was used; thus, they support the concept that the increased brain water in cirrhosis is mainly extracellular. These observations contrast with a recent hypothesis suggesting that the brain edema in acute liver failure has a cytotoxic origin.¹³

The exact physiologic meaning of the fast and slow components of diffusion is still under discussion.¹⁰ The idea that the slow diffusion component measures the intracellular space and the fast diffusion component, the extracellular space lacks confirmatory data. The volume fractions obtained from biexponential analysis of water signal-intensity decay (F slow ∼ 0.33; F fast ∼ 0.67) do not correspond to the physiologic volume fractions of the extra- and intracellular compartments (F intra ∼ 0.80; F extra ∼ 0.20).^9,26 A more recent theory has suggested that the 2 components could correspond to 2 differently structured water pools rather than to specific extracellular and intracellular water compartments.¹¹ Despite this controversy, we decided to perform a DTI study over a large range of b-values to investigate whether biexponential analysis could potentially differentiate the cytotoxic or vasogenic nature of the edema. This idea was based on previous studies showing that cytotoxic edema, a feature of the early phase of ischemic stroke, causes an increase in the slow MD,²⁷ whereas vasogenic edema, typically present surrounding brain tumors, causes an increase in the fast MD.²⁸

The observation of a fast MD increase that returned to normal after liver transplantation supports the concept that the increased water content was mainly located in the interstitial compartment. The decreased FA in white matter, which occurred parallel to the MD increase, likely represents less restricted movement of water molecules, a typical feature of extracellular edema.²⁸ Liver transplantation restores liver function and results in resolution of brain edema. In accordance, FA in parietal white matter normalized following transplantation.

Brain edema is a well-recognized complication of acute liver failure related to ammonia-induced astrocyte swelling that can cause intracranial hypertension and death.^29,30 In recent years, MR imaging techniques have documented low-grade brain edema in patients with chronic liver failure.³¹ This observation led to the proposal that encephalopathy in cirrhosis is the clinical manifestation of neuronal dysfunction secondary to astrocyte swelling.³² In support of this hypothesis, quantitative cerebral water-content mapping found an increase in total brain water in patients with HE.¹² However, the characteristics of brain edema in cirrhosis appear to differ from those of edema in acute liver failure. Our observations and those of previous studies^14,15 in cirrhosis support a water increase in the interstitial compartment, which could be explained by greater blood-brain barrier permeability, secondary to chronic inflammation and cytokine activation.³³ The increased permeability could result in low-grade interstitial brain edema, which would not affect neuronal function because it is located extracellularly. This could be the reason why we found no differences in the fast MD between patients with and without minimal HE. The absence of differences in MR imaging parameters in these patients has been observed in previous studies by using magnetization transfer imaging,¹⁷ DWI,¹⁵ and DTI.¹⁴

Episodes of HE are usually related to a precipitating factor that causes acute hyperammonemia. In acute liver failure, persistent elevation of ammonia concentration above 200 μmol/L leads to brain edema and intracranial hypertension.³⁴ It is plausible that the precipitating factor of episodic encephalopathy would induce superimposed swelling of astrocytes and cytotoxic edema. This is supported by the observation of an MD decrease in patients with acute-on-chronic liver failure.³⁵ Thus, a better explanation of what occurs during episodes of HE is the presence of mixed edema,¹⁶ consisting of chronic vasogenic edema secondary to cirrhosis and superimposed acute cytotoxic edema secondary to hyperammonemic decompensation.

An additional finding in our study is the differing pattern observed in the parietal white matter and corticospinal tract. Regional differences on MR imaging have been reported for T1-³⁶ and T2-weighted imaging,¹⁸ and for DWI.¹⁴ In a previous study by using fluid attenuated inversion recovery imaging, we found an increased T2 signal intensity in the corticospinal tract in approximately 25% of patients with cirrhosis. This abnormality, which is associated with functional impairment detected by neurophysiologic methods, probably corresponds to the same pathogenic syndrome as hepatic myelopathy,³⁷ but to a lesser extent. In comparison with the parietal white matter, the corticospinal tract showed an increase in fast and slow MD, which suggests coexistence of extracellular and intracellular edema. The higher vulnerability of this tract to the putative toxins that cause HE has been explained by higher energy requirements that may lead to long-term effects.³⁸ FA of the corticospinal tract did not change after liver transplantation, despite MD normalization. The persistent isotropy (lower FA) in the corticospinal tract probably relates to microscopic structural changes, such as those caused by demyelination.³⁹ Myelin loss can explain the persistent functional derangement after liver transplantation in patients with well-defined hepatic myelopathy³⁷ and in those with T2 hyperintensity along the corticospinal tract.³⁸

One important result of our study is that introduction of a biexponential analysis–enabled detection of MD and FA changes that monoexponential analysis did not show, as can be seen in Figs 1–4. Thus, in addition to providing information that could be pathophysiologically relevant, biexponential analysis enables detection of changes in the water pools that escape monoexponential study. Application of this method to other pathologic conditions, such as ischemia or demyelinating disorders, could be useful for assessing the extent of tissue damage.

The current study has several limitations that should be taken into consideration in the interpretation of the results. The MR imaging method (biexponential analysis of DTI) has the previously discussed differences between physiologic and measured volume fractions. Although statistically significant, the intensity of changes in DTI parameters before and after liver transplantation was low (approximately 2%). Thus, it may be difficult to show differences in studies with limited numbers of patients or multiple comorbidities. Cognitive function is influenced by several factors that are present before and after liver transplantation, and some of these may not have been adequately recognized in the study. The diagnosis of minimal HE is based on demonstration of cognitive disturbance in the absence of other causes, but several factors could have participated in the deterioration seen on neuropsychological tests before liver transplantation. Immunosupressors, which are necessary after liver transplantation, may cause white matter lesions.⁴⁰ We did not observe signs of neurotoxicity attributable to immunosupressors, but a possible role in the outcome after liver transplantation cannot be completely ruled out because it is not possible to control for their effect.