Personality Disorder in Multiple Sclerosis Correlates With Cognitive Impairment

Thirty-four patients with laboratory or clinically definite MS²⁰ were referred for evaluation of cognitive dysfunction or responded to an advertisement for MS patients with emotional problems. In addition to neuropsychological evaluation, patients underwent magnetic resonance imaging (MRI) and neurological examination to derive an Expanded Disability Status Scale (EDSS) rating.²¹ Some patients also participated in an experimental trial of a newly developed psychological intervention designed to diminish the frequency of socially aggressive behaviors. The results of the MRI and the psychological study will be presented in separate reports.

Twenty-one patients had a progressive form of MS and 13 had relapsing disease. EDSS scores ranged from 1.0 to 8.5 (mode=3.0). There were 32 Caucasian patients and 2 African Americans. The male/female ratio was 13/21 (62% female).

Fourteen normal volunteers (12 women, 2 men), all Caucasian, matched to the MS group on age and education, were also studied as control subjects.

All participants were seen with an informant who provided collateral information for the history and completed questionnaires pertaining to the primary participant's personality. Among patients, 68% (23) of the informants were spouses, 23% (8) were parents, and 9% (3) were domestic partners or friends. Among control subjects, the informant roles were 83% (12) spouse, 7% (1) parent, 14% (2) domestic partner or friend. The proportion of informants in each category did not differ across groups by chi-square test (χ²=3.3, P=0.50).

Exclusion criteria included history of neurological disease other than MS (or for control subjects, any neurologic disease), drug or alcohol dependence, psychiatric disease other than psychological problems attributable to MS, or clinical MS relapse (or steroid treatment) within 3 weeks prior to participation.

Depression was quantified by use of the Beck Depression Inventory (BDI).²² Cognitive tests, based on previous studies of dementia and MS,^23,24 were reduced to 10 variables for statistical analysis (Table 1). Language was assessed by the University of Victoria version of the Token Test²⁵ and the Boston Naming Test.²⁶ Spatial processing was assessed with the Judgment of Line Orientation Test²⁷ and the copy administration of the Complex Figure Test.²⁸ The California Verbal Learning Test²⁹ and the Brief Visuospatial Memory Test–Revised³⁰ assessed new learning and memory. Attention and/or rate of processing was examined by the Trail Making Test, Part B,³¹ and the Paced Auditory Serial Addition Test.³² The Wisconsin Card Sorting Test³³ and the Booklet Category Test³⁴ were employed to measure higher executive ability.

Interpersonal empathy was assessed with a 15-item, adapted version of the Hogan Empathy Scale (HES).³⁵ Higher scores indicate a greater degree of empathy in social situations. Personality was further evaluated with the revised NEO Personality Inventory (NEO-PI),¹² including the five domain scores as follows: Neuroticism, Extraversion, Openness, Agreeableness, Conscientiousness. In accordance with the five-factor model, Neuroticism refers to the degree of emotional responsiveness and proclivity for negative affect. Extraversion refers to dependence upon external stimulation for arousal; that is, extroverts are outgoing and sensation seeking. Openness reflects one's desire for new knowledge, ideas, and experiences. Agreeableness is the desire for socialization, honesty, and altruism in relationships. Conscientiousness is the proclivity to be well organized and deliberate. Domain subscale scores (six for each domain) were also examined in some analyses, as described below. The raw scores derived from the NEO-PI were converted to T-scores (mean=50, SD=10), with higher scores indicating a stronger endorsement of each personality trait. All informants for MS patients and control subjects completed these tests without assistance. Some MS patients had questions read to them to avoid fatigue.

Discrepancy scores were calculated for the HES and each NEO-PI domain score. In each case, differences between self and informant report were calculated and the absolute value of the difference recorded. Absolute differences in raw scores were employed for the HES, and differences in T-scores were employed for the NEO-PI.

All participants underwent an initial structured telephone screening interview, emphasizing DSM-IV³⁶ criteria for psychiatric disorders. Mood and substance-related disorders were covered in detail to rule out previously untreated disease. After screening, participants underwent neuropsychological and neurological examinations. (Control subjects completed only the personality and neuropsychological tests.) Neuropsychological testing was performed by a trained technician supervised by a board-certified clinical neuropsychologist. All participants provided signed, Institutional Research Board–approved, informed consent.

First, group means were compared by Student's t (cognitive and informant-reported personality) and paired-sample t-tests (patient vs. informant reports), which were considered reliable only where P<0.01 to control for Type I error. Dependent measures included the cognitive measures; BDI; HES; the primary NEO-PI scale scores for Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness; and self/informant discrepancies for the HES and NEO-PI.

Second, linear regression analyses (forward selection, entrance criterion of 0.05, exit criterion of 0.10) were performed to identify cognitive variables that predict abnormal HES and NEO-PI ratings. EDSS and BDI were entered into each regression analysis first (block 1), to control for the influence of physical disability and depression in estimating the association between cognition (block 2) and each personality domain (dependent variable). These analyses included NEO-PI subscales within each primary scale, but the Openness domain was excluded because this scale did not discriminate patients and control subjects in the first analysis.

Third, a similar regression analytic strategy was used to examine the association between cognitive impairment and self/informant discrepancy scores. Only those personality measures in which the self and informant report values differed significantly were examined (HES, NEO-PI Conscientiousness). Because the specific content within personality domains was considered unimportant in evaluating the relationship between discrepancy scores and cognition, these analyses did not include NEO-PI subscale scores.

Although the groups were well matched on age, education, and estimated premorbid IQ (Table 2),³⁷ MS patients performed more poorly than control subjects on nearly every cognitive test. MS patients were more depressed on the BDI (P<0.01).

MS patients were rated lower on the HES, as well as on NEO-PI Extraversion (P<0.01), Agreeableness (P<0.01), and Conscientiousness (P<0.01). Patients were rated significantly higher on Neuroticism (P<0.01).

There were no significant differences between self and informant reports among control subjects, but MS patients generated higher self-ratings in comparison to those of informants on the HES (P<0.01). Patients also rated themselves significantly higher in Conscientiousness (P<0.001).

Linear regression analyses were performed to identify neuropsychological variables that predict abnormal personality ratings by informant report (Table 3). EDSS was significantly correlated with Conscientiousness (r=−0.40), but there were no other reliable correlations between EDSS or BDI and the informant-report personality indices. When EDSS and BDI were controlled in subsequent regression analyses, significant relationships emerged between personality and tests of executive function (Wisconsin Card Sorting Test perseverative responses, Booklet Category Test errors) and visual learning (Brief Visuospatial Memory Test recall). Executive tasks were significant predictors in regression models for low ratings on HES and Altruism, and elevated ratings on Positive Emotions. Impaired learning significantly predicted increased Vulnerability, and low ratings within the Conscientiousness domain (Competence, Dutifulness). EDSS and BDI were retained in only one final regression model (Conscientiousness), and in this case, visual learning was the strongest predictor.

The strength of association between neuropsychological and personality measures can be shown by way of partial correlations (Table 3), which control for the influence of EDSS and BDI. These partial r's were statistically reliable for HES, Vulnerability, Positive Emotions, Altruism, Conscientiousness, Competence, and Dutifulness.

Both Wisconsin Card Sorting Test perseverations and Complex Figure Test copy scores were retained in the final model predicting mean discrepancy score on HES (R²=0.29, P=0.011). Perseverations, the highest-loading predictor, was associated with greater patient/informant disagreement. The partial correlation between perseverations and HES discrepancy, controlling for EDSS, was marginally significant (r=0.39, P<0.05), and the partial r controlling for BDI was not reliable.

In the model predicting NEO-PI Conscientiousness disagreements, only the Wisconsin Card Sorting Test was retained (R²=0.31, P=0.001). Although EDSS and BDI were not retained in the model, lower BDI scores (r=−0.28) and higher EDSS scores (r=0.31) were modestly associated with greater patient/informant disagreement (R²=0.24, P=0.003). Partial correlations between perseverations and Conscientiousness discrepancy scores, controlling for EDSS (r=−0.57, P<0.01) and BDI (r=0.54, R<0.015), were both reliable.