Introduction

Hepatitis B virus-related cirrhosis (HBV-RC) is a serious health problem in Asia with high infection, morbidity, and mortality rates (Moriwaki et al. 2010). Hepatic encephalopathy (HE) is the most common neuropsychiatric complication in end-stage HBV-RC. It is characterized by a wide spectrum of mental status, ranging from subtle cognitive dysfunction to severe coma. Clinically, overt hepatic encephalopathy (OHE) is easily identified in patients with cirrhosis by the appearance of neuropsychiatric symptoms and is usually staged according to the West Haven criteria. In contrast, patients with cirrhosis without OHE always present as essentially normal on clinical neuropsychiatric examination. However, increasing data (Dhiman and Chawla 2009; Li et al. 2004; Moriwaki et al. 2010) indicate that neurocognitive dysfunction of varying degrees is common in patients without any signs of OHE.

Neurocognitive dysfunction in cirrhosis without OHE has recently become more relevant because it has been associated with OHE, mortality, poorer quality of life, and deterioration in daily functioning (Cordoba 2011). As a result, several approaches have been used to explore cerebral function in these patients, including neuropsychological tests, electroencephalography, critical flicker frequency, and so on (Cordoba 2011; Weissenborn 2008). These studies reported cognitive deficits mainly in visual perception, visuoconstructive abilities, impaired fine motor skills and attention, whereas general intelligence and speaking are preserved (Dhiman and Chawla 2009; Li et al. 2004; Weissenborn 2008). However, most of these tests have been developed to diagnose minimal HE and do not reveal the underlying pathophysiology mechanisms.

MRI is a non-invasive technique that can be used to assess metabolic changes in the brain, impaired brain function, and brain biochemistry. Magnetic resonance spectroscopy studies revealed that myoinositol and choline signals are reduced and that glutamine-glutamate signals are increased in several brain regions in patients with cirrhosis (Huda et al. 2008; McPhail and Taylor-Robinson 2010). Magnetization transfer imaging and diffusion tensor imaging studies, which reflect the integrity of the white mater tract, also reveal mild and diffuse edema in cirrhotic brains (Huda et al. 2008; McPhail and Taylor-Robinson 2010). A recent study (Chen et al. 2012a) combining voxel-based morphometry with voxel-based diffusion tensor imaging found that patients with HBV-RC without OHE exhibited diffuse white matter abnormalities, including decreased white matter volume and fractional anisotropy values. However, the underlying functional changes in brains of these patients are still not fully understood.

Functional MRI (fMRI) has been extensively used to investigate various brain functions. As a new branch of fMRI, resting-state fMRI can reflect the baseline brain activity. Biswal et al. (Biswal et al. 1995) reported that the low-frequency (0.01–0.08 Hz) fluctuations of the resting-state fMRI signal are physiologically important and suggested that they reflect spontaneous neuronal activity (Biswal et al. 1995). Evidence increasingly indicates that the pathophysiology of many neurocognitive dysfunctions may be associated with the spontaneous changes in brain activity fluctuations as measured with resting-state fMRI (Fox et al. 2005; Mantini et al. 2007). More recently, amplitude of low-frequency fluctuation (ALFF) (Zang et al. 2007), a method developed to analyze these low-frequency fluctuations, has also been used as an effective fMRI algorithm to detect a wide range of brain disorders, such as the Alzheimer’s disease (Dai et al. 2012), depressive disorder (Zhu et al. 2012), and hepatic encephalopathy (HE) (Chen et al. 2012b; Qi et al. 2012). Several recent fMRI studies (Chen et al. 2012b; Qi et al. 2012) with ALFF have found that cirrhotic patients had wide-spreadly abnormal spontaneous brain activity. Nevertheless, the results of these studies were not entirely consistent, perhaps because of differences in the causes of cirrhosis, which may have different effects on cerebral alterations (Burra et al. 2004). Additionally, most of these studies were focused on abnormal baseline brain activity in cirrhotic patients with low-grade HE or OHE.

Abnormalities in ALFF in patients with HBV-RC have not been studied intensively. Moreover, little is known about the relationships between these neurocognitive impairments and any functional abnormalities in those patients. The purpose of this study was to identify modulations in neural activity in patients with HBV-RC without OHE in the resting state with the ALFF method. More importantly, we also sought to determine whether changes of neural activity were related to the psychometric hepatic encephalopathy score (PHES), a validated battery of psychometric tests used to assess the peculiar cognitive impairments in patients with cirrhosis (Duarte-Rojo et al. 2011; Weissenborn 2008).

Patients and methods

Patients

Thirty patients (25 male; mean age, 45.6 ± 9.4 years; age range, 32 ~ 66 years) with chronic liver cirrhosis caused by HBV infection were included in this prospective study. All patients had HBV-RC diagnosed by biopsy or on the basis of case history, clinical examination, biochemical and imaging findings. Overt hepatic encephalopathy was diagnosed when the West Haven criteria indicated stage I disease or higher. Patients were excluded if they had current symptoms of OHE at the time of recruitment, any history of OHE, other types of viral hepatitis, gastrointestinal hemorrhage or bacterial infection (within 1 month before the study), a transjugular intrahepatic portosystemic or a surgical portocaval shunt, diffuse hepatocellular carcinoma, or were taking drugs that could alter cerebral function. All patients underwent a detailed clinical examination. The severity of liver disease was determined according to the Child-Pugh score.

For comparison, 30 healthy controls (25 male; mean age, 45.2 ± 9.1 years; age range, 32 ~ 63 years) matched for age (within 3 years), sex and years of education (within 3 years), without disease of liver and other systems, were recruited through advertising within the hospital. All controls received detailed clinical and neurological examinations on the same day as the fMRI scans.

Exclusion criteria for all patients and controls included age lower than 18 or greater than 70 years, alcoholism, neurological or psychiatric diseases, a history of substantial head trauma, infection with human immunodeficiency virus, hypertension, diabetes, poor vision, other major medical illness, left-handedness, and any focal abnormality detected on routine brain MRI examinations.

This study was approved by the Medical Research Ethics Committee of Nanfang Hospital, Southern Medical University. Written informed consent was obtained from all the participants before the study. The clinical and demographic data of patients with HBV-RC without OHE and of healthy controls are shown in Table 1.

Table 1 Demographic and clinical characteristics of patients with HBV-Related cirrhosis without overt hepatic encephalopathy and of healthy controls
Full size table

Neuropsychological tests

The PHES is calculated from a standardized battery of psychometric tests developed to assess the degree of cognitive impairment in patients with cirrhosis (Weissenborn 2008). All 30 patients and 30 controls completed the five paper-and-pencil psychometric tests that comprise the PHES—number connection tests A and B, the digit symbol test, the serial dotting test, and the line-tracing test—after an appropriate explanation and demonstration. The physicians participating in the study were trained in conducting and evaluating these psychometric tests. The PHES was generated from the sum of the scores of the five tests results from the five psychometric tests, with a range from +5 to −15 points. The method of calculating the PHES is reported in detail elsewhere (Lv et al. 2013).

Imaging protocol

All MRI data were obtained with a 1.5-T MR imager (Achieva Nova-Dual; Philips, Best, the Netherlands) in the Department of Medical Imaging, Guangdong No. 2 Provincial People’s Hospital. Each participant was placed in a standard head coil fitted with foam padding to reduce head motion. Each was instructed to close both eyes, to keep as still as possible, and not to think of anything in particular or to fall asleep in the scanner. Routine imaging studies consisted of T1-weighted images and T2-FLAIR images acquired to detect clinically silent lesions. FMRI was conducted using an echo-planar imaging sequence with the following settings: repetition time, 3,000 ms; echo time, 50 ms; flip angle, 90°; field of view, 230 × 230 mm; matrix, 64 × 64; and total volumes, 160. Axial slices (n = 33) were collected at a thickness of 4.5 mm and no interslice gap. Resolution was 3.59 × 3.59 mm in-plane. Each fMRI scan lasted 8 min. After the examination, all participants were asked some questions to verify the degree of their adherence to instructions.

Image analysis

Imaging data were mainly preprocessed with Statistical Parametric Mapping (SPM8, http://www.fil.ion.ucl.ac.uk/spm). The first 10 time points for each participant were discarded to avoid transient signal changes before magnetization reached steady-state and to allow the participant to get used to the fMRI scanning noise. Raw data were corrected for slice-timing and realigned for head movement correction (data from participants with movement more than 1.5 mm of translation or more than 1.5° of rotation in any direction were excluded). Afterwards, all of the realigned images were spatially normalized into the Montreal Neurological Institute template, and each voxel was resampled to isotropic 3 × 3 × 3 mm3. The resting-state images were then spatially smoothed with an isotropic Gaussian kernel (full-width at half-maximum, 8 mm).

Further data preprocessing and ALFF analysis were performed with the REST software (http://resting-fmri.sourceforge.net). Preprocessing also included removing linear trends and temporally filtered (band pass, 0.01–0.08 Hz) to remove the effects of very-low-frequency drift and high-frequency noise. The procedure for calculating ALFF is described elsewhere (Chen et al. 2012b; Qi et al. 2012; Zang et al. 2007). The filtered time series was transformed to a frequency domain, and the power spectrum was obtained. The square root was thus calculated at each frequency of the power spectrum and the averaged square root was obtained across 0.01–0.08 Hz at each voxel. This averaged square root was taken as the ALFF. For standardization purposes, the ALFF of each voxel was divided by the global mean ALFF value. The standardized ALFF (mALFF) of each voxel should have a value of about 1. This standardization procedure is analogous to that used in positron emission tomography studies (Raichle et al. 2001).

Statistical methods

Wilcoxon signed-rank tests were used to analyze the differences in the PHES between two groups. Analyses were conducted using software (SPSS, version 13.0; Chicago, III), and a P value less than 0.05 was deemed significant.

To investigate differences in ALFF between the patients and controls, a two-sample t-test was executed on the individual mALFF maps in a voxel-by-voxel manner using age and years of education as covariates. Multiple comparisons were corrected using the AlphaSim program in the AFNI software determined by Monte Carlo simulations. Statistical maps of the two-sample t-test were created using a combined threshold of P < 0.001 and a minimum cluster size of 22 voxels, yielding a corrected threshold of P < 0.05.

We also performed a whole-brain voxel-based correlation analysis between mALFF values and the PHES using age and years of education as covariates. Given the exploratory nature of the study, we adopted a relatively liberal statistical threshold (A corrected significance level of P < 0.05 was obtained by clusters with a minimum cluster size of 74 voxels at an uncorrected individual voxel height threshold of P < 0.01). Then, mean mALFF values of all significantly different clusters revealed by voxel-based correlation analysis were extracted separately using the extract time series in REST and were input into SPSS. Finally, using age and years of education as covariates, the relationship between mean mALFF values in significant different areas and the PHES in patients was assessed with Spearman’s correlation coefficient. P < 0.05 (two-tailed) was used to determine significant correlations.

Results

Differences in PHES

Compared with healthy controls, cirrhotic patients performed significantly worse on the PHES test (−3 [−9 to 1] vs. [−2 to 3] respectively; P < 0.001).

ALFF changes in patients with HBV-RC without OHE

Differences in mALFF values between the patients and controls were widespread (Fig. 1 and Table 2). Compared with controls, patients had significantly lower mALFF values in the bilateral lingual gyrus, cuneus/precuneus/middle occipital gyrus, precentral gyrus, inferior parietal lobule and paracentral lobule, left inferior temporal gyrus and right postcentral gyrus. The values of mALFF were higher in the bilateral orbital gyrus/rectal gyrus.

Fig. 1
figure 1

Regions showing different mALFF values between patients with HBV-related cirrhosis and healthy controls. Regions showing higher mALFF values in the bilateral orbital gyrus in patients with HBV-related cirrhosis and lower mALFF values in the bilateral lingual gyrus, cuneus/precuneus/middle occipital gyrus, precentral gyrus, inferior parietal lobule and paracentral lobule, left inferior temporal gyrus cuneus and right postcentral gyrus. Threshold was set P < 0.05 (AlphaSim corrected). mALFF standardized amplitude of low-frequency fluctuation

Full size image
Table 2 Regions of the brain in which the amplitudes of low-frequency fluctuation differ significantly between patients with HBV-related cirrhosis without overt hepatic encephalopathy and healthy controls
Full size table

Correlations between ALFF and PHES

The regions in which mALFF values were significantly and positively correlated with the PHES were the right middle occipital gyrus (r = 0.66, P < 0.001) and bilateral precentral gyrus (left: r = 0.73, P < 0.001; right: r = 0.64, P < 0.001; Fig. 2 and Table 3).

Fig. 2
figure 2

The results of correlation analysis in patients with HBV-related cirrhosis (P < 0.05, AlphaSim corrected): regions in which mALFF values were significantly correlated with the PHES (a). Red means a positive correlation. The PHES positively correlated with ALFF values in the right MOG (b), bilateral PCG (c, d). mALFF standardized amplitude of low-frequency fluctuation, PHES psychometric hepatic encephalopathy score, MOG middle occipital gyrus, PCG precentral gyrus

Full size image
Table 3 Correlations between amplitude of low-frequency fluctuation and psychometric hepatic encephalopathy scores in patients with HBV-related cirrhosis
Full size table

Discussion

Compared with controls, the lower mALFF values in HBV-RC patients without OHE were widely distributed, mainly in the visual association areas (the bilateral lingual gyrus and the middle occipital gyrus and the left inferior temporal gyrus), motor-related areas (the bilateral precentral gyrus and the paracentral lobule, and the right postcentral gyrus), and the default-mode network (the bilateral cuneus/precuneus and inferior temporal gyrus). Higher mALFF values were also found mainly in the prefrontal cortex (the bilateral orbital gyrus/rectal gyrus). More importantly, mALFF values in right middle occipital gyrus and bilateral precentral gyrus correlated with the PHES.

The blood oxygenation level-dependent (BOLD) signal indirectly reflects neural activity and low-frequency fluctuations in the resting state. Although the exact biologic mechanisms of ALFF remain unclear, many studies have reported that the ALFF changes in the BOLD signal are associated with local neuronal spontaneous activity (Biswal et al. 1995; Zang et al. 2007). Different fMRI algorithms may reflect different aspects of comprehensive brain functions. Compared with other methods, such as functional connectivity analysis, ALFF has the advantage of directly reflecting the amplitude or intensity of spontaneous activity (Zang et al. 2007). Thus, to explore the alterations of brain activity by measuring the resting-state ALFF across the whole cerebral gray matter may provide more information on the dysfunction of cirrhotic brains.

The lingual gyrus, as the key part of visual cortex, is the visual processing center in the brain. This area is important in color perception, visual discrimination, and visual attention (Lee et al. 2000). The middle occipital gyrus is considered to be a part of the visual dorsal stream in sighted subjects (Wandell et al. 2007). The inferior temporal cortex is part of the ventral pathway, also called the “what” pathway, for visual object processing (Hitomi et al. 2012). The precentral gyrus cortex (M1, BA4) is intimately related to motor control. The paracentral lobule has been reported to be involved with lower extremity movements and attention to somatosensory stimulation (Forss et al. 1996; Lim et al. 1994). The postcentral gyrus is the location of the primary somatosensory cortex (SI, BA3, 1, 2), which is initially involved in processing information related to tactile awareness.

Our finding of markedly lower mALFF values in these visual-association and motor-related areas in cirrhotic brains is consistent with the findings from a recent study (Lv et al. 2013) in which a regional homogeneity method in another group of patients with HBV-RC without OHE detected decreased coherence of spontaneous neuronal activity in visual association (in the left lingual gyrus, middle temporal gyrus, and right middle occipital gyrus) and motor-related areas (the bilateral precentral gyrus and paracentral lobule). Positron emission tomography also reveals reduced bilateral cerebral glucose utilization in a medial portion of the primary and visual association cortices (Lockwood et al. 2002), as well as motor cortices (Weissenborn et al. 2007) in patients with cirrhosis.

Recently, resting-state studies (Chen et al. 2012b; Qi et al. 2012) with the ALFF analysis algorithm also revealed decreased mALFF values in the visual (such as the lingual gyrus, middle occipital gyrus, and middle temporal gyrus) and motor associated areas (such as the supplemental motor area and right postcentral gyrus) in patients with low-grade or minimal HE. Given the findings of these neuroimaging studies, the reduced mALFF values in the visual association cortex and motor-related areas in patients with HBV-RC in the present study suggest dysfunction in those regions.

An important finding in this study was the positive correlation between the PHES and mean mALFF values in the visual-association area (the right middle occipital gyrus) and motor-related areas (the bilateral precentral gyrus). The PHES reliably reflects most minimal HE-related neuropsychological impairments because it assesses visual perception, construction, visual/spatial orientation, motor speed and accuracy, concentration, and attention in patients with cirrhosis (Weissenborn 2008; Weissenborn et al. 2001). The lower the cognitive abilities, the lower the mALFF values in those brain regions in patients with HBV-RC without OHE. Thus, we speculate that the decreased neuronal activity at baseline in the visual association and motor-related areas may contribute to the deficits in processing visual information and motor control, leading to some typical cognitive impairments in these patients, such as deficits in visual perception, visuoconstructive abilities, and impaired fine motor performance (Amodio et al. 2008; Weissenborn 2008). In turn, the significant correlation between the mALFF values and the PHES in spontaneous activity in the right middle occipital gyrus and the bilateral precentral gyrus indicate that differences in these regions could provide a new and objective biomarker of the progression of the cognitive impairment in these patients.

In the present study, mALFF values were lower in the bilateral cuneus/precuneus and inferior temporal gyrus, which are part of the default mode network (Raichle et al. 2001). This network helps maintain baseline brain activities related to self-awareness, episodic memory, and interactive modulation between internal mind activities and external tasks (Raichle et al. 2001). Neuroimaging studies (Chen et al. 2012b, c; Lv et al. 2013; Qi et al. 2012) consistently find that patients with cirrhosis have abnormal resting-state brain activity in regions within the default mode network, such as the precuneus and inferior parietal lobule. Thus, our study provides more evidence that the default mode network is involved in the cognitive deficits in these patients.

Interestingly, compared with healthy controls, patients with HBV-RC without OHE showed higher mALFF values in the bilateral orbital gyrus/rectal gyrus and those regions belong to a part of prefrontal cortex, which is the most complex and highly evolved neurocortex region. Abnormities of the prefrontal cortex have been reported in many neuroimaging studies (Huda et al. 2008; Weissenborn et al. 2007). However, as noted before (Chen et al. 2012c; Lv et al. 2013; Ni et al. 2012), performance on neuropsychological tests was not significantly correlated with indices of abnormal brain activity in this critical area in cirrhotic patients without OHE. Decreased functional activity in the resting state could be related to functional impairment, whereas an increase could be interpreted as compensatory reallocation or recruitment of cognitive resources. Recently, a compensatory neural mechanism during the visual judgment (Zafiris et al. 2004) and in the resting state (Qi et al. 2012; Chen et al. 2012b, c) has also been reported in cirrhotic patients without OHE. Thus, the higher mALFF values of the prefrontal cortex (the bilateral orbital gyrus/rectal gyrus) may compensate for the deficits in visual processing and impaired motor function in these patients. However, this concept needs to be confirmed.

Some limitations in our study are worth mentioning. First, as a preliminary study, our results are limited to a small patient cohort. A large-cohort study is needed. Also, a proper cutoff PHES for diagnosing minimal HE has not been established in our country. Thus, further group analysis is needed. Third, the fMRI of the BOLD signal relies on susceptibility-sensitive gradient-echo sequences, such as the echo-planar imaging sequence (Weiskopf et al. 2007). Thus, certain areas of the brain, especially the inferior brain regions, such as the orbital gyrus/rectal gyrus, are subject to signal loss and distortion from susceptibility artifacts. This loss is difficult to avoid, and an optimal echo-planar imaging settings are needed (Weiskopf et al. 2007). Finally, the slow sampling rate, which is frequently used in resting-state fMRI studies because it permits scanning the entire brain, could not be avoided. At a repetition time of slow sampling rates (as in this study TR = 3,000 ms) for multisection acquisitions, the cardiac and respiratory fluctuation effects might be aliased into the low-frequency BOLD MR signal fluctuations (Lowe et al. 1998). In future studies, simultaneous cardiac recording may provide a more direct correction.

Conclusions

In summary, we found that patients with HBV-RC but without OHE have diffuse abnormalities in intrinsic brain activity and lower mALFF values mainly distributed in the visual-association areas, motor-related areas, and the default mode network, as well as higher mALFF values in the prefrontal cortex. In addition, the right middle occipital gyrus and bilateral precentral gyrus intrinsic activity were related to the PHES. Resting-state fMRI with ALFF analysis provided more evidence that the pathogenesis of neurocognitive dysfunction in patients with HBV-RC without OHE may be attributed to abnormal neural activity in multiple brain regions.