Prevalence and significance of neurocognitive dysfunction in hepatitis C in the absence of correlated risk factors

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Neurocognitive morbidity has been reported in individuals with chronic hepatitis C virus (HCV) infection, but the magnitude of such dysfunction in the absence of disease-correlated factors known to affect the central nervous system (e.g., substance abuse, cirrhosis, depression, interferon treatment) and the impact of any such change on functioning is unclear. We investigated a cohort of individuals with HCV, all of whom were carefully screened to exclude relevant comorbidities, to elucidate virus-related changes in the brain using neuropsychological tests and magnetic resonance spectroscopy (MRS). A cohort of 37 patients with chronic HCV infection was culled from 300 consecutive patients presenting to a tertiary care liver clinic. A comparison group of healthy controls (n = 46) was also assessed. Of 10 neurocognitive measures evaluated, the HCV group showed marginally poorer learning efficiency compared with controls; only 13% of patients demonstrated a clinical level of impairment on this test (defined as 1.5 SD below the normative standard). Although patients reported greater levels of fatigue and symptoms of depression, these factors did not correlate with the degree of learning inefficiency. With respect to MRS, the HCV group demonstrated increased choline and reduced N-acetyl aspartate relative to controls in the central white matter. Indicators of liver disease severity did not correlate with either memory or MRS abnormalities. In conclusion, while our findings support an association between hepatitis C and indicators of central nervous system involvement in a cohort of patients carefully screened to eliminate other factors influencing neurocognitive integrity, the clinical significance of these effects is limited. (HEPATOLOGY 2005;41:801–808.)

Individuals with chronic hepatitis C virus (HCV) infection often complain of “brain fog,” even in the absence of clinically significant liver disease. Specific complaints include forgetfulness, difficulty concentrating, fatigue, and depression. Impaired central nervous system function has been documented in individuals with chronic HCV. Neuropsychological tests, electrophysiological event-related potentials, and proton magnetic resonance spectroscopy (1H MRS) have been used to investigate whether cerebral dysfunction exists in these patients even in the absence of severe liver disease. An initial study reported by Forton et al. using 1H MRS revealed elevated choline in the basal ganglia and cerebral white matter in a cohort of individuals with chronic HCV infection who had histologically mild liver disease, compared with individuals with hepatitis B and control subjects.1 More recently, this group reported deficits in concentration and working memory speed in viremic patients with hepatitis C, and the degree of impairment was correlated with the level of choline.2 Similar deficits in sustained attention and psychomotor speed have been reported in other studies,3–5 and subclinical impairment in the P300 event-related potential has been described.6

The recent detection of viral sequences in postmortem brain tissue adds credence to the hypothesis that direct central nervous system infection may be the cause of these neurocognitive and metabolic changes.7–8 Before one can directly relate central nervous system changes to HCV infection in vivo, it is essential to exclude other comorbid factors in this population. Of particular importance are factors such as substance abuse, major affective disorder, and minimal hepatic encephalopathy (MHE), as each of these conditions can independently affect performance on cognitive tests and other indicators of brain dysfunction. MHE occurs only in the presence of cirrhosis and is associated with EEG and neuropsychological abnormalities similar to those reported in patients with HCV.9–12 Several of the previous studies have included patients with cirrhosis,4, 6 and most of these have included individuals with a history of substance abuse and/or affective disturbance. Statistical control or inferential testing of these effects was typically done, although often not for all factors simultaneously. Clearly, the ability to eliminate or control for these comorbidities is of crucial importance in determining the cause of the impairments.

Although there is consensus about the kind of mild cognitive dysfunction observed, there is inconsistency regarding the extent to which such effects correlate with degree of liver disease. Hilsabeck et al. found a pattern of cognitive deficits that was identical in patients with HCV alone and in patients with other forms of chronic liver disease.4 Although up to 50% of patients without cirrhosis were impaired (more than 1 SD below population means) on measures of psychomotor speed and processing efficiency, a significant association was found between the magnitude of cognitive deficit and the severity of liver disease. In another study, HCV patients with decompensated cirrhosis exhibited impaired attention, executive function, and psychomotor speed, whereas patients without cirrhosis did not differ from healthy controls.3 In contrast, Kramer et al. did not find an association between abnormal P300 characteristics and the degree of histological severity or biochemical activity of hepatitis.6 Likewise, Forton et al. reported no association between cognitive impairment and biochemical markers of liver disease activity as such (serum alanine aminotransferase).1, 2 Perhaps the most striking evidence of dissociation comes from the studies demonstrating elevated choline-containing compounds in the brain in patients with HCV,1, 2 because this is the opposite of the pattern seen in hepatic encephalopathy.13, 14

Given the few studies to date, the small sample sizes, and the lack of control over some of the potential confounding variables, our first aim in the present investigation was to employ rigorous screening criteria to exclude individuals with known risk factors for cognitive impairment. We used two different measures of central nervous system integrity—neuropsychological measures and 1H MRS—to test the hypothesis that there is a distinct pattern of neurocognitive and metabolic abnormalities associated with HCV that is not directly related to the severity of liver disease or comorbid psychosocial factors.

Abbreviations

HCV, hepatitis C virus; MRS, proton magnetic resonance spectroscopy; MHE, minimal hepatic encephalopathy; CVLT-II, California Verbal Learning Test, 2nd Edition; FAI, Fatigue Assessment Inventory; CHO, choline; MYO, myoinisotol; NAA, N-acetyl aspartate.

Patients and Methods

Patient Groups.

Patients were recruited from a prospective series of individuals presenting at the (tertiary care) Liver Clinic at University Health Network, Toronto Western Hospital. Controls were recruited from the hospital community or were relatives of patients in this study. The study protocol was approved by the Research Ethics Board of the University Health Network, and all individuals provided written informed consent prior to their participation.

A comprehensive history identified potential risk factors for cognitive impairment in all prospective study candidates. Patients were excluded from participation if any of the following conditions were present: a history of injection drug use, substance abuse disorder, or drug dependence within the past 7 years; major depression or other psychiatric disorders within the past 2 years; current use of antidepressant medication or other pharmaceuticals known to affect cognitive function; coinfection (hepatitis B, HIV); evidence of cirrhosis on liver biopsy within the past 18 months; cryoglobulinemia greater than 1%; history of head injury (with loss of consciousness greater than 30 min), stroke, dementia, seizure disorder, hypothyroidism, diabetes, vitamin B12 deficiency, syphilis, cerebrovascular disease; history of learning disability, nonfluency in English, or Full-Scale IQ less than 70; and other unstable medical conditions. Of the 300 treatment-naïve patients screened, over 70% were excluded for reasons such as coinfection or other medical conditions likely to influence the central nervous system (15%); comorbid psychiatric diagnoses (11%); cirrhosis on liver biopsy within the preceding 18 months (16%); lifetime alcohol abuse or use of any recreational drug within the past 7 years (9%); or other factors including lack of fluency in English, refusal, or age greater than 60 years (21%). Criteria were similar for controls, although the former had not undergone liver biopsy.

Thirty-seven patients with HCV met eligibility criteria for the study. Overlapping subsets of the cohort participated in the two parts of the study: 31 underwent neuropsychological testing, and 33 had 1H MRS. A total of 46 healthy controls met eligibility criteria. Forty-two controls underwent neuropsychological testing, and 34 participated in the MRS.

Neuropsychological Assessment.

A neuropsychological assessment was conducted using tests that are sensitive to cerebral dysfunction and are used to screen for cognitive impairment. Descriptions of all measures can be found in Lezak.15 The Wechsler Abbreviated Scale of Intelligence estimates general intellectual functioning. Learning efficiency, reflected in the total number of items across trials, was assessed using the California Verbal Learning Test, 2nd Edition (CVLT-II), and the Brief Visual Memory Test. Measures of attention included the Digit Span subtest from the Wechsler Adult Intelligence Scale, Third Edition, and the Paced Auditory Serial Addition Test (2.0-s presentation rate). The Trail Making Test, parts A & B, and the Digit Symbol subtest of the Wechsler Adult Intelligence Scale measured psychomotor speed. Cognitive flexibility was evaluated using the interference condition of the Stroop Interference Test and the perseverative responses of the Wisconsin Card Sorting Test. The Beck Depression Inventory was administered to assess symptoms of depression, and the Fatigue Assessment Inventory (FAI) was used to evaluate the level of daytime fatigue. The first subscale (FAI-I), which characterizes global severity, was used because it has been shown to be the best discriminator between patients with medical disorders and healthy controls.16

1H MRS.

A single-voxel technique was used to assay metabolites in three brain regions: the basal ganglia (putamen and globus pallidus combined), central white matter (in the centrum semiovale), and midline frontal grey matter. Each voxel was 2 × 2 × 1 cm. The imaging parameters were: PRESS, TR = 2,000, TE = 30, 256 averages. Spectra were analyzed using the LC model17 and were corrected for T1 and T2 relaxation using known values of these parameters for the regions of interest to derive millimole concentrations. We did not segment voxels according to tissue type, because the two critical regions (the basal ganglia and central white matter) were homogenous within the voxel. The frontal grey matter (which does contain some cerebrospinal fluid and white matter) was considered a control region because it has not been shown to be affected in HCV or MHE. Metabolites of interest were choline (CHO), creatine, glutamine-glutamate, myoinisotol (MYO) and N-acetyl aspartate (NAA).

Data Analyses.

Data were analyzed with SPSS 10.1 for Windows (SPSS Inc., Chicago, IL). Group effects were explored with t tests, unless ANCOVA was required for covariates. Pearson correlations were used to explore relationships between demographic variables and cognitive and MRS outcome measures. Stepwise linear regression was used to explore the relationship between indicators of liver disease severity and the primary outcome variables that differed between patient and control groups.

Results

Table 1 lists demographic, psychosocial, and liver disease severity characteristics of the two groups. The distribution of fibrosis stage scores was as follows: 1 patient was without fibrosis (Metavir stage 0), 8 patients had stage 1 fibrosis, 17 had stage 2 fibrosis, and 11 had stage 3 fibrosis. For purposes of statistical analysis, patients in the stage 0 and 1 groups were combined.

Table 1. Sample Characteristics
 ControlHCV
  1. Abbreviations: BDI, Beck Depression Inventory; FAI-I, Fatigue Assessment Inventory, subscale I; ALT, alanine aminotransferase; NA, not applicable.

Age (yr)35.8 + 9.346.1 + 7.4
Education (yr)15.9 + 2.213.2 + 3.7
BDI score4.3 + 7.110.8 + 9.7
FAI-I score2.64 + 1.254.42 + 1.69
Fibrosis scoreNA2.0 + 0.8
InflammationNA1.8 + 0.5
ALTNA97.0 + 73.6

Several of these variables differed among groups. Patients in the control group were younger [t(81) = 5.50; P < .001] and had a higher level of educational attainment [t(81) = 4.03; P < .001]. While the mean level of depressive symptoms fell within “normal” limits (score 0–15) on the Beck Depression Inventory for both groups, it was significantly higher in the HCV group [t(81) = 3.26; P < .005]. Among the HCV cohort, 29% of patients endorsed symptoms of depression that can be categorized as clinically significant, whereas only 2% of controls did so. Ratings of fatigue severity (Fatigue Assessment Inventory, subscale I) were also significantly higher in the patient group [t(81) = 4.50; P < .001]. The potential effect of these variables on the neurocognitive measures was explored, and the only significant associations (controlling for multiple tests) were the CVLT-II score and education (r = 0.52; P < .01) and education and Wechsler Abbreviated Scale of Intelligence Full-scale IQ (r = .55; P < .01). Given our use of age-adjusted standardized scores, the age difference between groups was not a significant covariate. In the MRS data, age was correlated with central white matter creatine (r = 0.45; P < .01) and MYO (r = 0.46; P < .01) and with basal ganglia MYO (r = 0.46; P < .01). Thus ANCOVA was used to examine group differences for measures in which there was a significant correlation with demographic factors, whereas t tests were used to interrogate other cognitive and MRS measures.

Neuropsychological Assessment.

Performance on neuropsychological testing is presented in Table 2. Standard scores are used throughout so that the clinical relevance of the data can be more easily appreciated. The only measure for which a significant group effect was seen was the total learning score for the CVLT-II [F(1,72) = 5.37; P < .05]. The HCV group showed poorer learning than controls, even when the effect of education (F = 13.66; P < .001) was removed. As Fig. 1 demonstrates, this was not merely attributable to the effect of extreme scores in one group. No other differences approached significance. Of note, the apparent difference between controls and patients with respect to Full-Scale IQ was eliminated becasue of shared overlap with the education covariate.

Table 2. Neuropsychological Test Scores
Cognitive MeasureControlHCV
  • Abbreviations: WASI, Wechsler Abbreviated Scale of Intelligence; PASAT, Paced Auditory Serial Addition Test; TMT, Trail Making Test; WCST, Wisconsin Card Sorting Test; CVLT, California Verbal Learning Test; BVMT, Brief Visual Memory Test.

  • *

    Scaled score (mean = 10, SD = 3).

  • t score (mean = 50, SD = 10).

WASI Full-Scale IQ111.3 + 12.8100.5 + 16.0
Digit span*11.9 + 2.911.4 + 3.5
PASAT57.1 + 11.252.5 + 10.9
Digit symbol*10.9 + 2.99.7 + 3.4
TMT-A48.7 + 10.848.3 + 8.1
TMT-B47.6 + 11.946.0 + 11.6
Stroop interference50.9 + 7.550.6 + 9.3
WCST perseverations48.3 + 8.247.8 + 10.3
CVLT total learned57.5 + 9.147.8 + 11.0
BVMT total learned40.0 + 11.838.9 + 10.1
Figure 1.

Distribution of learning efficiency scores (t scores) for control and HCV groups. CVLT-II, California Verbal Learning Test, 2nd Edition; CON, control; HCV, hepatitis C virus.

Associations between variables characterizing liver disease severity and the CVLT-II measure were examined. None of these accounted for sufficient variance to enter the model. The partial correlations were 0.06 for the inflammatory score, −0.19 for the fibrosis score, and 0.21 for alanine animotransferase. Interestingly, these disease severity markers also showed no significant relationship with variables capturing depressive symptomatology (Beck Depression Inventory) or fatigue (Fatigue Assessment Inventory, subscale I); the absolute values for each partial correlation were less than 0.25.

1H MRS.

Cerebral metabolites in the basal ganglia, frontal grey matter, and central white matter for each of the groups are presented in Table 3. Two group differences reached significance in the central white matter. The HCV group showed increased CHO [t(65) = 2.17; P < .05] and reduced NAA [t(65) = 3.27; P < .01]. Patients also showed a significant elevation of creatine in the basal ganglia [t(65) = 3.09; P < .01]. The distribution of concentrations for these variables are shown in Fig. 2. It is worth noting that although MYO means appear different, this is attributable to the age confound, and there was no significant group effect (P > .10 for each) when age was introduced as a covariate. Stepwise regression revealed no significant predictive relationship between disease severity markers and the 1H MRS values that were abnormal in the patient sample (the absolute values for all partial correlations were less than 0.34). In particular, patients with fibrosis scores of 3 were no different than those with scores of 0 through 2. Similarly, there is no significant correlation between the abnormal 1H MRS variables and the CVLT-II score, although there is a trend seen in the relationship between central white NAA and CVLT-II scores (r = 0.24; P = .08 [uncorrected]).

Table 3. MRS Data (Means + SD of Metabolites in mmol Concentration)
Region/MetaboliteControlHCV
  1. Abbreviations: CRE, creatine; GLX, glutamate + glutamine.

Basal ganglia/CHO1.631 + 0.2311.664 + 0.162
Basal ganglia/CRE7.567 + 1.0278.309 + 0.919
Basal ganglia/GLX11.406 + 3.89412.320 + 3.745
Basal ganglia/MYO3.663 + 0.9124.357 + 1.150
Basal ganglia/NAA11.413 + 1.00711.064 + 1.072
Central white/CHO1.937 + 0.2382.054 + 0.254
Central white/CRE6.627 + 0.8706.769 + 0.789
Central white/GLX8.531 + 2.5668.677 + 2.639
Central white/MYO4.346 + 0.8725.045 + 0.819
Central white/NAA10.942 + 1.09810.204 + 0.698
Frontal grey/CHO1.672 + 0.2781.710 + 0.361
Frontal grey/CRE8.523 + 1.4798.841 + 1.122
Frontal grey/GLX15.012 + 2.96415.281 + 3.532
Frontal grey/MYO5.831 + 1.6785.638 + 1.171
Frontal grey/NAA10.041 + 0.85310.215 + 0.894
Figure 2.

Distribution of metabolites (in millimoles) showing significant group effects between control and HCV samples. CHO, choline; NAA, N-acetyl aspartate; CRE, creatine; CON, control; HCV, hepatitis C virus.

Discussion

The question of whether HCV is associated with central nervous system dysfunction receives a qualified positive response in this study. On the one hand, our findings can be characterized as being similar to those reported by other groups in that we showed measurable effects on parameters of cerebral integrity. Specifically, we found a significant difference between patients and healthy controls on a measure of learning efficiency. In contrast, other measures assessing attention, nonverbal learning, problem-solving, and speed of processing were not impaired. Most previous investigations did not assess verbal learning, instead observing deficits in tests of attention and psychomotor speed.2, 4–5 Tests tapping these domains of power and speed of processing have a high sensitivity but poor specificity for detecting central nervous system dysfunction because they frequently accompany emotional disorders and situational changes in arousal.15 Our failure to find impairments in these domains may be attributed to our use of standard scores (T scores) rather than raw scores to characterize the data, because normalization moderates the effects of outliers, particularly for speed-dependent measures. Furthermore, screening and exclusion of comorbid factors that are known to influence these particular parameters (e.g., concurrent substance abuse, depression, or cirrhosis) and our assessment of controls tested in the same manner, rather than comparison with normative data bases, may also account for the lack of a significant amount of variance attributable to HCV status.

One other study examined a broad spectrum of cognitive variables in patients with HCV.3 They compared neuropsychological profiles in healthy controls, patients without cirrhosis but with HCV, HCV patients with compensated cirrhosis, and a group of HCV patients with decompensated cirrhosis. They only observed impairments in the latter group, in whom attention, executive functioning, and motor speed were compromised. Learning efficiency was marginally but not significantly poorer in this group. Of note, the group in their study that is most similar to the one we tested—patients without cirrhosis but with HCV—did not differ from controls in any of the cognitive domains.

Cognitive impairment may be expressed in a wide variety of medical and psychiatric conditions (e.g., fatigue, depression, substance abuse) and in association with some psychotropic medications. In addition, short-term neuropsychiatric complications are not uncommon in individuals with chronic HCV infection during treatment with interferon.18–20 If participants are not screened thoroughly for these factors, the estimated prevalence of cognitive impairment associated with the virus per se may be inflated, particularly if such risk factors are more prevalent in the index population. In our study, the majority of consecutive cases presenting to the clinic (73% after eliminating refusals and individuals who did not meet age and linguistic competency criteria) had to be excluded for such reasons. It is probable that many excluded cases would demonstrate greater cognitive impairment attributable to some interaction between viral status and other factors. Because our goal was to characterize morbidity associated with HCV status in the absence of such correlated factors, our findings likely underestimate the presence of cognitive dysfunction in the natural cohort.

Whereas we employed rigorous screening criteria to rule out other possible causes of cerebral dysfunction, a majority of the sample (67%) described by Hilsabeck et al. reported a history of illicit drug abuse with no information regarding recency, and at the time of testing 28% met the criteria for a diagnosis of depression, 27% were taking psychiatric medications, and 23% were undergoing interferon therapy.4 In their later study, 33% of the sample had cirrhosis, 19% had alcoholic hepatitis, 71% reported a history of intravenous drug use, and 82% were under treatment for depression.5 A possible lack of statistical power in assessing the influence of these effects was recognized by the authors. Forton et al. reported a lack of correlation between measures of depressive symptoms and the cognitive measures they used, suggesting that this was not a primary factor differentiating patients with current and past HCV infection (determined by their HCV RNA status).2 Nonetheless, the significant difference in affective status between these groups could have been confounded in the determination of their primary dependent measure—the number of tests impaired (relative to normative data). Thus the cognitive deficits reported by these authors may have been related to several causes.

It is important to clarify how neuropsychological impairment is identified to minimize overinterpretation of positive findings. When using population norms as a reference, a key issue is the extent to which the testing context affects the viability of the comparison. For example, the study sample may undergo multiple tests in which influences of fatigue or cross-contamination could be applicable and the normative data may have been collected on one measure in isolation. Determining cut-off points for a designation of impairment is also crucial. In clinical practice, the typical value for cutting scores is at least 1.5 SD units from the control mean, and the validity of these scores requires that the reference sample be sufficiently large and matched demographically to the population of interest.15 In our study, only 13% of the HCV group (4 cases) obtained a score below the age-matched normative cut-off for the learning efficiency measure. Although this proportion exceeds the 2% (1 case) of the control group, it does not indicate widespread cognitive morbidity. Furthermore, the predictive value of any clinical “sign” (such as an impaired score on the CVLT-II) is determined by both the sign's association with the diagnosis of interest (its valid-positive rate in identifying true cognitive impairment) and the base rate of the event of interest (memory impairment in HCV). If the base rate is low in the target group, as our data suggest, even a strong sign may decrease diagnostic accuracy.20

Although performance on neurocognitive tests is likely to be influenced by a host of psychosocial and demographic variables, these (except for age) are likely to have little impact on metabolites measured with 1H MRS. Therefore, such data may be considered a less “contaminated” indicator of cerebral integrity. Like Forton et al., we observed an increase in CHO in central white matter for the HCV group. We also observed a reduction in white matter NAA in the HCV group, which has not been reported to date, perhaps because of the high sensitivity of our analysis technique or because our sample size is almost twice as large as the previously published series.2 There was also a trend toward association between NAA levels and memory efficiency, similar to the results for CHO and the working memory scores reported by Forton. We provide converging evidence of a direct impact of HCV on brain function in a cohort that can be considered relatively free of comorbid influences, but the clinical correlates or pathophysiological mechanisms for these identified differences are unclear. NAA appears to be a marker of the functional integrity of neurons and processes, rather than an indication of viable neurons, in that reversible decreases in NAA have been documented in brain injury.21–22 The significance of elevated CHO in white matter is also unclear. The possible mechanism of glial activation due to oxidative stress has been suggested in disorders such as chronic fatigue syndrome and human immunodeficiency virus infection.23–24 However, the amount of CHO-containing compounds in the brain detected by MRS can also be enhanced by choline ingestion,25 so that the sources of intergroup variability and the clinical relevance of any differences cannot be immediately ascribed to pathological states.

The differences we observed in learning efficiency and brain metabolites cannot simply be ascribed to the severity of liver disease, and none of our patients had cirrhosis on recent liver biopsy (most had mild fibrosis). Therefore, it is unlikely that MHE was responsible for the cognitive impairment observed. Although the absence of histological evidence of cirrhosis in a single liver biopsy does not demonstrate unequivocally that an individual is free of cirrhosis,26–27 several findings are contrary to the typical pattern in MHE. First, MHE is diagnosed using neuropsychological tests of psychomotor speed and/or EEG slowing, while the cognitive deficit in our sample was characterized by a decrement in learning efficiency.9, 28 Second, our MRS findings are quite different from those typical of hepatic encephalopathy, which include elevated glutamine-glutamate and a reduction in CHO and MYO.13–14, 29 Finally, we found no association between the key differentiating variables and fibrosis stage, which might have been expected if the effects were due to low-grade encephalopathy secondary to degree of liver dysfunction, as has been reported elsewhere.4 Nonetheless, it remains possible that there is some aspect of liver–brain interaction, not captured by our measures, that contributes to neurocognitive dysfunction in our patients.

How the limited impairment we observed may be reflected in quality of life and functional disability is an important issue for patients and society. The majority of our patients were either employed or were students; two were unemployed, two were on long-term disability for reasons related to HCV, and three were on long-term disability for unrelated reasons. Only one of these individuals was impaired by clinical criteria on the learning efficiency measure. There does not appear to be a strong relationship between cognitive morbidity and disability status in our sample, but this likely underestimates the relationship in the larger population, given the moderating effects of comorbid conditions. Furthermore, there may be more subtle indicators of functional disability present, even in our specific sample of relatively unimpaired patients. Although health-related quality of life has been reported to be reduced in HCV and to improve with treatment,30–32 much of this literature is based on reports in patients with significant chronic liver disease. Several recent studies suggest that, among individuals with mild clinical morbidity, labeling or awareness of HCV serological status is itself a significant cause of poor health-related quality of life.33, 34 Such work underscores the interaction of disease burden and context in determining health-related quality of life in the larger population.

In summary, the limited prevalence and functional significance of neurocognitive abnormalities we found in a carefully screened sample suggests that “brain fog” described in association with HCV may be more a function of comorbid factors than a direct impact of the virus on the central nervous system.

Note added in proofs

Subsequent to the acceptance of this paper, we became aware of a recent publication in which decreased NAA in cortical grey matter and reduced cognitive functioning was reported in a cohort of patients with HCV. (Weissenborn K, Krause J, Bokemeyer M, Hecker J, Schuler A, Ennen JC, et al. Hepatitis C virus infection affects the brain—evidence from psychometric studies and magnetic resonance spectroscopy. J Hepatol 2004;41:845–851.)

Ancillary