| |
Cognitive function in hepatitis C patients with advanced fibrosis enrolled in the HALT-C trial
|
| |
| |
Journal of Hepatology May 2005
.....33% of HALT-C patients have evidence of mild cognitive impairment which was highly correlated with subject IQ, education level, and occupation ........these data suggest that there is evidence of mild cognitive inefficiency amongst HALT-C patients which is most apparent on tasks which require difficult verbal encoding and recall, complex working memory, and reaction to visual stimuli......Although the impact of cognitive test scores on daily functioning were not determined, the overall level of impairment observed is mild ....... This pattern of scores may be consistent with a sub-cortical cognitive inefficiency pattern previously reported in patients with HIV infection
.....findings suggest that impairment in cognitive function in HALT-C patients is not a manifestation of alcohol or drug induced brain disease.....50 and 39% meeting criteria for an alcohol or drug use disorder, respectively.....Contrary to our expectations, lifetime psychiatric and substance abuse history as defined by CIDI-LT did not correlate with most of the group mean standard scores....
Robert J. Fontanaa, Linas A. Bieliauskasb, Carla Back-Madrugac, Karen L. Lindsayd, Ziad Kronfolb, Anna S. Loka, Latha Padmanabhane, the HALT-C Trial Group
a Division of Gastroenterology, Department of Internal Medicine, University of Michigan Medical Center, 3912 Taubman Center, Ann Arbor, MI 48109-0362, USA
b Department of Psychiatry, University of Michigan Medical Center, Ann Arbor, MI, USA
c Department of Psychiatry, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA
d Division of Gastrointestinal and Liver Diseases, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA
e New England Research Institutes, Watertown, MA, USA
Received 29 November 2004; received in revised form 16 April 2005; accepted 17 April 2005 published online 31 May 2005.
Uncorrected Proof- Article in Press
ABSTRACT
Background/Aims
Prior studies have demonstrated neuropsychological abnormalities in chronic hepatitis C (CHC) patients even with mild fibrosis. The aim of this study was to determine the frequency, type, and severity of cognitive impairment in a large group of CHC patients with advanced fibrosis.
Methods
Ten validated neuropsychological tests were administered to 201 CHC patients. Standard scores for individual tests were calculated using normative population data that controlled for age, gender, and/or education. Lifetime psychiatric history, alcohol consumption, and mood status were also determined.
Results
33% of patients met criteria for cognitive impairment (i.e. standard score <40 on at least 4 tests). Mild impairment in verbal recall and working memory were noted with other domains remaining intact. Liver disease severity and lifetime psychiatric/substance abuse history did not correlate with group mean cognitive test results or the presence of cognitive impairment. In contrast, IQ and depression scores were significant and independent predictors of cognitive impairment (ROC=0.84).
Conclusions
33% of Patients entering the HALT-C trial have evidence of a mild, non-focal subcortical processing deficit which was highly correlated with IQ, education, and occupation. Future studies of cognitive function in CHC patients should control for general cognitive ability.
INTRODUCTION
In western countries, countries the most commonly reported parenteral risk factor for acquiring hepatitis C virus (HCV) is a history of injection drug use [1,2]. Amongst chronic hepatitis C (CHC) patients with advanced liver disease, a history of heavy alcohol consumption is also frequently reported [3]. With a high prevalence of substance abuse disorders, it is not surprising that there is an increased prevalence of emotional distress and mood disorders amongst CHC patients compared to uninfected controls and other liver disease patients [4-7]. Objective neurophysiologic and radiologic abnormalities have been reported in CHC patients that are not detected in healthy controls nor in patients who have cleared HCV infection [7-9]. In addition, studies of CHC patients with mild liver disease demonstrate impairment in several domains of cognitive function compared to uninfected controls [8-12]. These data suggest that HCV may influence cognitive function via direct infection of brain microglial cells, indirectly via induction of pro-inflammatory cytokines or through other mechanisms [13,14].
The Hepatitis C Antiviral Long-term Treatment against Cirrhosis (HALT-C) trial is a prospective, randomized, controlled study of long-term pegylated interferon for CHC patients with advanced fibrosis who are non-responders to prior interferon [15,16]. In the cognitive ancillary study, the mood status and cognitive function of CHC patients are being assessed prior to and during antiviral therapy. The primary aim of this analysis was to determine the frequency, type, and severity of cognitive impairment in CHC patients with advanced fibrosis not receiving antiviral therapy compared to normative population data that control for age, gender, and/or education level. Our secondary aim was to identify correlates of cognitive function in HALT-C patients.
AUTHOR DISCUSSION
The HALT-C patients showed a cognitive profile which was generally within normal limits compared to population controls except for decreased performance on several tests that are particularly sensitive to generalized cognitive inefficiency including the selective reminding test [44-46]. The serial digit learning test has been shown to be a more sensitive measure of cognitive dysfunction than the digit span test [47]. Reaction times have also been described as particularly sensitive to generalized cognitive impairment in patients with HIV infection [33,48]. Taken together, these data suggest that there is evidence of mild cognitive inefficiency amongst HALT-C patients which is most apparent on tasks which require difficult verbal encoding and recall, complex working memory, and reaction to visual stimuli. Although the impact of cognitive test scores on daily functioning were not determined, the overall level of impairment observed is mild [7,8,12].
Using conservative criteria for defining cognitive impairment, 33% of our study population demonstrated evidence of cognitive impairment. In comparison, only 2-6% of the general population would be expected to have cognitive impairment based on these criteria. Amongst CHC patients with cognitive impairment, we identified evidence of decreased attention/concentration, speed of working memory, and verbal memory with additional findings of decreased non-verbal memory, reasoning, and cognitive flexibility. This pattern of scores may be consistent with a sub-cortical cognitive inefficiency pattern previously reported in patients with HIV infection [49] as well as with decreased concentration and speed of working memory reported in smaller studies of CHC patients [7,10]. However, the gold standard for establishing impairment has been reported to be either clinician ratings or empirically determined deficit scores and we are further analyzing our data using these approaches [40].
Contrary to our hypotheses, CHC patients with more severe liver disease, a history of substance abuse and psychiatric disease, and baseline mood disturbance did not have lower group mean test scores. The lack of correlation of test scores with disease severity may have been due to the inclusion of patients with only advanced fibrosis in this study. Interestingly, the Shipley IQ scores as well as educational level and occupation correlated significantly with most of the individual cognitive test scores. On multivariate analysis, only the Shipley IQ scores and BDI-II scores were independent predictors of cognitive impairment with the Shipley IQ scores accounting for the majority of the variance. These observations raise the possibility that cognitive reserve or general, long-standing cognitive abilities may be a more significant contributor to the observed cognitive test scores than liver disease status as has been reported in HIV patients and we plan to explore this further [50,51]. We acknowledge that the battery of tests used in this study are not normed for IQ but rather by age, gender, and/or education. Therefore, it is possible that the significant relationship noted in our patient population between estimated IQ and cognitive test scores may be due to a negative bias in the IQ scores of our patients compared to those comprising the normative population of each individual test. However, we feel that this is unlikely since the distribution of Shipley IQ scores in our patient population was normal and similar to that reported for the Shipley population data in the United States. This indicates that the general intellectual abilities of our study population are normally distributed and do not represent a population with reduced cognitive reserve as a whole.
The absence of a correlation of cognitive test scores with lifetime substance abuse and psychiatric diagnoses was unexpected [52-54]. The lifetime incidence of major psychopathology in our cohort was substantial with 50 and 39% meeting criteria for an alcohol or drug use disorder, respectively. The high rate of alcohol use disorders may in part be explained by the selection of CHC patients with advanced fibrosis and non-response to prior interferon [21,22]. Interestingly, a lifetime diagnosis of a depression or anxiety disorder also did not correlate with cognitive test scores. Furthermore, baseline emotional status as measured by the BSI did not significantly correlate with cognitive test results while baseline BDI-II scores showed significant correlations with only two test scores. The absence of a consistent correlation corroborates findings from other studies demonstrating a lack of association between subjective psychiatric symptoms and objective neuropsychogical test results in CHC patients [55]. However, it should be noted that only 13% of our subjects had BDI-II scores in the moderate to severe range. Furthermore, the lifetime incidence of depression (15%) and anxiety (11%) in our cohort were not significantly greater than that reported in the general US population [21,22].
The pattern of cognitive deficits in our study population bears some resemblance to previously described patterns of subclinical or minimal hepatic encephalopathy [41]. This entity is characterized by decreased concentration and speed of working memory, reduced psychomotor speed, and mental flexibility in patients with liver disease and portosystemic shunting [12,41,56]. However, the lack of uniform diagnostic criteria for minimal hepatic encephalopathy can lead to imprecision in diagnosis [41,57]. The finding of decreased recall on the selective reminding test and on the continuous visual memory test in the impaired group, suggests that there is a non-focal deficit with recall in the context of preserved memory encoding ability which goes beyond that reported for minimal hepatic encephalopathy. Furthermore, although most patients with minimal hepatic encephalopathy have abnormalities in visuomotor scanning, the group mean test results for the trailmaking test Part's A and B and finger tapping test were normal. In addition, there was no significant correlation of group mean test scores with objective markers of liver disease severity. Therefore, we feel the deficit pattern in our patients is more consistent with a subcortical information processing deficit rather than subclinical hepatic encephalopathy [58-62].
Strengths of our study include the use of a standardized, comprehensive battery of tests that were prospectively administered to a large group of CHC patients. Since only mild abnormalities were detected, ongoing assessment of enrolled patients in the longitudinal phase of the HALT-C trial will help determine which tests are most useful for detecting impairment due to antiviral therapys. The generalizability of our findings may, however, be limited by the HALT-C study entry criteria. Our study also did not concomitantly test a contemporary group of CHC patients with lesser degrees of hepatic fibrosis, previously untreated CHC patients, or a control group of uninfected patients. However, we selected neuropsychiatric tests that could be converted to standard scores using normative population data.
In summary, 33% of HALT-C patients have evidence of mild cognitive impairment which was highly correlated with subject IQ, education level, and occupation. If these factors are verified to be determinants of cognitive performance, the development of remedial and supportive educational strategies would be worthwhile. The mechanism by which CHC patients may develop cognitive impairment is not clear although reasonable hypotheses regarding prior substance abuse history, psychiatric illness, and liver disease mechanisms were not confirmed in our study. This may in part be due to the high lifetime incidence of substance abuse disorders observed in our patient population and exclusion of patients with mild fibrosis [3]. Therefore, additional studies of previously untreated CHC patients with greater variability in hepatic fibrosis are needed. Longitudinal studies that assess CHC patients prior to and after successful antiviral therapy are also needed to determine the potential role of HCV in mediating alterations in cognition and mood.
RESULTS
Patient population
There was no significant difference in the baseline demographic features of participants at the two sites (99 from USC, 102 from UMICH) and therefore, their data are grouped together (Table 1, see below). The majority of CHC patients were male (71%) and Caucasian (71%). The most frequently identified parenteral risk factor was injection drug use (46%) and 38% had compensated cirrhosis.
Fifty-two percent of subjects met DSM-IV criteria for a lifetime diagnosis of an alcohol use disorder (Table 2, see below). However, only 3 (1.5%) fulfilled criteria for an active alcohol use disorder within 12 months of study entry. Thirty-nine percent of subjects met DSM-IV criteria for a lifetime diagnosis of a drug use disorder which included abuse or dependence of amphetamines (10%), cannabis (27%), cocaine (27%), hallucinogens (20%), opioids (20%), sedatives (16%), and others (13%). None of the subjects fulfilled criteria for a drug use disorder within 12 months of study entry. Overall, 24% of subjects were receiving a prescription antidepressant and 8% were receiving an anxiolytic. There were significant correlations between baseline BDI-II scores and use of antidepressants (r=0.35, P<0.0001) as well as anxiolytics (r=0.24, P=0.0005).
Group mean cognitive test scores
The HALT-C patients demonstrated significant impairment in verbal recall (selective reminding test) and working memory (serial digit learning) compared to population controls. Conversely, the digit symbol test, Trail's A and B, finger tapping test, measures of executive function and verbal processing remained intact. The normal d scores on the selective reminding test and continuous visual memory test indicate that while encoding of verbal information may be affected, storage and recognition of both verbal and non-verbal information was grossly within normal limits. The proportion of patients with abnormal simple and choice reaction times below the cutoff for clear impairment is similar to the distribution of test scores for age-matched adults [37]. Nevertheless, available data on simple reaction time stratified by age on a similar measuring device suggests that the mean score of 351ms (ms) is slower than the mean score of 285ms reported in a large study of healthy adults [42]. The Shipley IQ scores were normally distributed and similar to that of the general US population.
Sixty-six (33%) subjects had evidence of cognitive impairment. For unimpaired patients, the only test with a standard score <40 was the selective reminding test. For impaired patients, the tests that had a group mean standard score <40 were the selective reminding test, continuous visual memory test, serial digit learning, digit span, and trails B. In addition, both simple and choice reaction time mean scores were below the normative levels indicated above in impaired patients (373.8 and 533.44ms, respectively). The number of perseverative responses on the Wisconsin card sort test was also above reported normative levels (24.8 vs 11.0) [39].
Correlates of group mean cognitive test scores
There was no consistent correlation between individual standard scores and liver histology, serum albumin, and total bilirubin levels. Similarly, the platelet count was not significantly correlated with most standard scores except for the simple and choice reaction time (Data not shown). Subject age correlated with the trail's A and there were also isolated correlations between subject gender and the selective reminding test and the finger tapping test [43].
Contrary to our expectations, lifetime psychiatric and substance abuse history as defined by CIDI-LT did not correlate with most of the group mean standard scores. In particular, a prior history of a mood disorder did not correlate with any of the standard scores. Statistically significant inverse correlations were noted, however, between a prior drug use disorder and the finger tapping test and selective reminding test. Lifetime alcohol consumption also did not correlate with any of the standard scores. These findings suggest that impairment in cognitive function in HALT-C patients is not a manifestation of alcohol or drug induced brain disease. The lack of a consistent correlation of test results with BDI-II and BSI scores may in part be due to the exclusion of patients with severe depression from the trial and the use of anti-depressants in many of the subjects with elevated BDI-II scores.
Overall, the Shipley IQ score was the strongest and most consistent correlate of all of the standard scores. In addition, level of education and occupation correlated strongly with many of the individual group mean test scores. As expected, there was a significant correlation between education level and IQ score (r=0.4, P<0.0001) and occupation and IQ (r=0.38, P<0.0001).
3.4. Correlates of cognitive impairment
On univariate analysis, predictors of cognitive impairment were level of education (P<0.0001), Shipley IQ (P<0.0001), BDI-II score (P=0.004), occupation code (P<0.0001), GSI score (P=0.08), and diabetes mellitus (P=0.07). A multivariate model using logistic regression demonstrated that Shipley IQ scores and to a lesser extent, BDI-II scores, were independent predictors of cognitive impairment with an area under the receiver operating curve (ROC) of 0.84. The Shipley IQ accounted for 40.2% of the variance in cognitive impairment, the BDI-II accounted for 1.4% of the variance, and the variables omitted through backward selection accounted for 5.9% of the variance.
TABLE 1
MEAN
Age: 50
Female; 29%
Ethnicity: 71% white, 14% African-American. 11% Hispanic
69% Married
EDUCATION: 90% at least high school, 24% at least college;
Work status: 82% working
8% not working, illness/disability; 3% unemployed; 6% retired; 1% full time student/not working
OCCUPATION
Semi-skllied: 27%
Not in work force >10 yrs: 11%
Skilled labor/craftsman/foreman: 13%
Professional/technical: 22%
CIRRHOSIS ON BIOPSY: 38%
SERUM ALT: 133
Albumin: 3.9
Total bilirubin: 0.67
Platelets: 167.5 x 100,000/mm3
INR: 1.04
HCV RNA, log 10IU/ml: 6.51
PARENTERAL RISK FACTORS:
Ever received a transfusion: 40%
Ever experienced a needlestick: 16%
Ever used needles for recreational drugs: 46%
ESTIMATED DURATION OF INFECTION: 29 yrs
Mean time since last IFN therapy: 809 days
Diabetes: 25%
Hypertension: 28%
MATERIALS & METHODS
Patient population
All HALT-C patients had detectable serum HCV RNA, a liver biopsy within 12 months of enrollment demonstrating an Ishak fibrosis score of 3-6, and non-response to prior interferon therapy [15,16]. Patients with any other co-existent liver disorder, a Child-Turcotte-Pugh score >6, or a history of decompensation were excluded. Additional exclusion criteria included interferon intolerance, reactivity to anti-HIV, active use of illicit drugs, ongoing excessive alcohol consumption, a suicide attempt or hospitalization for depression within the past 5 years, and a history of a severe or uncontrolled psychiatric condition within the past 6 months. Of the 232 eligible patients enrolled in the lead-in phase at the two sites, 19 were ineligible for the cognitive study due to language or reading difficulties, 6 refused to participate, and 6 withdrew consent leaving 201 patients in this analysis.
Baseline assessment
All subjects underwent a complete medical history, physical examination and laboratory testing [15,16]. Years of education were coded by the highest level achieved and occupation was coded into six hierarchical categories [17,18].
Lifetime psychiatric history
The Composite International Diagnostic Interview Lifetime 2.1 (CIDI-LT) is a computerized diagnostic interview for mental health disorders [19]. After excluding organic illness, a diagnostic hierarchy using DSM-IV criteria is applied to subject responses for the Anxiety, Depression, Alcohol, and Drug modules [20]. The CIDI-LT diagnoses were grouped together as being either an Alcohol (e.g. alcohol abuse or dependence), Drug (e.g. drug abuse or dependence to various substance), Depressive (e.g. major depression), or Anxiety disorder (e.g. generalized anxiety, phobias) and compared to the prevalence of these disorders in the general US population [21,22].
Lifetime alcohol consumption
A semi-quantitative estimate of lifetime alcohol consumption was obtained using an adaptation of the Skinner survey [23]. Lifetime alcohol consumption in drinks was converted to grams of alcohol assuming that one standard drink such as a 12 oz beer, 5oz of wine, and a shot of liquor (i.e. 1.5oz) is equivalent to 12g of alcohol [24,25].
Mood status
The Beck Depression Inventory-II (BDI-II) is a self-administered questionnaire used to screen for depressive symptoms during the past 2 weeks [26]. The BDI-II scores were coded as no depression <10, minimal depression 11-14, mild depression 15-19, moderate depression 20-28, and severe depression 29. The Brief Symptom Inventory (BSI) was used to assess emotional distress [27]. Subjects were classified as having clinically significant emotional distress if the Global Severity Index (GSI) summary T-score was 63 (i.e. 90th percentile) [4].
Neuropsychological tests
Ten neuropsychological tests with alternate forms were administered in a predetermined order. Tests were selected based upon brevity, availability of normative data, and sensitivity for subtle neuropsychological abnormalities [7-12]. The battery was administered in a quiet environment by a team of trained technicians that were blinded to patient clinical information. To assess global fund of information, the Shipley Institute of Living scale was administered and a full scale Intelligence Quotient (IQ) estimate from the Wechsler Adult Intelligence Scale-Revised was calculated using population controls with a mean of 100 and standard deviation of 15 [28,29]. Standard scores (SS) were calculated for each test using normative data where x is the patient's test score [30]:
Verbal memory
The selective reminding test is sensitive to cognitive changes in general [31]. Standard scores were calculated for total words recalled over 12 trials using normative data. A d-prime (d) score was also calculated for true positives versus false positives in a recognition trial at the end of the test [32].
Non-verbal memory
The continuous visual memory test is sensitive to global brain injury and provides scores measuring encoding, storage, and retrieval of geometric shapes using age matched norms [33]. A d score is calculated for the recognition portion of the test.
Speed and efficiency of information processing
The digit span test assesses both basic attention and available working memory compared to age matched controls [34]. The digit symbol test measures the ability to rapidly match geometric symbols and numbers compared to age matched controls [35]. Lastly, the serial digit learning test consists of repeated trials of learning an extended series of 9 digits that can be converted to standard scores using age and education matched normative data [36].
Visuomotor tracking
Simple reaction time captures how quickly an individual presses a telegraph key when a light stimulus appears. It is considered abnormal when the response time exceeds 425ms [36]. The choice reaction time is considered abnormal when the response time exceeds 550ms [37]. The trailmaking test, Parts A and B, are complex assessments of visual scanning with a motor component wherein the respondent is either asked to connect a series of numbers in sequence (A) or letters and numbers (B) [38]. The Trail's A and B are scored using age matched normative data. The finger tapping test for both the dominant and non-dominant hand was converted to a standard score using age and gender matched normative data [38].
Executive function
The Wisconsin card sorting test is a computerized test of one's ability to shift conceptual set without directed instructions and is reported as an age-adjusted standard score [33,39].
Verbal processing
In the controlled oral word association test, subjects are asked to generate as many words as possible that begin with a particular letter in one minute. Standard scores are calculated using age, gender, and education matched normative data [32].
Definition of cognitive impairment
Twelve standard scores were available from the 10 tests administered. As in other studies, a standard score of <40 (i.e. 1 standard deviation below the mean) on an individual test was considered abnormal [11,17]. The standard score for the dominant and non-dominant hands of the finger tapping test were grouped together as well as the standard score for the digit span forward and backward tests. Individual patients were considered cognitively compromised if they had a standard score <40 on at least 4 of the 10 tests [17,40,41].
Data analyses
Descriptive statistics of baseline demographic, clinical, and liver disease parameters are reported. The mean standard scores for the entire group and the percent of patients with a standard score <40 on at least 4 of the 10 scaled tests was determined. Univariate analysis of demographic, liver disease, lifetime psychiatric, alcohol, and mood related correlates of the group mean standard scores was undertaken using correlation coefficients. In addition, univariate and logistic regression analysis with backward removal of variables was performed to identify independent correlates of cognitive impairment. Because of the large number of comparisons and test scores, a significance level of P<0.01 was chosen to minimize Type I error using SASŪ version 8.02.
|
|
| |
|
|
|
