Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Sep 1.
Published in final edited form as: Neuropsychology. 2018 Jun 25;32(6):654–663. doi: 10.1037/neu0000453

Cognition in older patients with multiple sclerosis compared to patients with amnestic mild cognitive impairment and healthy older adults

Alexandra K Roth a, Douglas R Denney a, Jeffrey M Burns b, Sharon G Lynch b
PMCID: PMC6126957  NIHMSID: NIHMS974308  PMID: 29939057

Abstract

Objective:

Progress in the treatment of multiple sclerosis (MS) has resulted in larger numbers of patients living to an advanced age, but little is known about the cognitive status of these individuals. The primary purpose of this study was to identify differences in the cognitive performance between elderly individuals with MS and those with amnestic mild cognitive impairment (aMCI).

Method:

Three groups ranging in age from 60-80 were compared: patients with MS (n=64), patients with aMCI (n=58), and healthy adults (n=70). All participants completed a standard neuropsychological test battery that evaluated domains of attention, processing speed, executive function, memory, language, and visual spatial function.

Results:

Compared to age-and sex-matched healthy controls, elderly MS patients exhibited a pattern of cognitive impairment centering on information processing speed and memory that was consistent with the deficits observed in other studies of MS patients regardless of age. Compared to aMCI patients, the MS patients exhibited worse performance on measures of processing speed, but better performance on a measure of memory under cued conditions (Selective Reminding Test), a non-speeded measure of language (Boston Naming Test), and measures of executive function with processing speed statistically controlled (Trail Making Test, Stroop Test).

Conclusions:

Differences on neuropsychological measures can serve to distinguish aMCI from MS-related cognitive impairment in older patients, but it is essential that these measures control for the deficit in processing speed that is such a primary feature of MS.

Keywords: Secondary progressive multiple sclerosis, amnestic mild cognitive impairment, Alzheimer’s disease, neuropsychology

Introduction

Advances in the treatment of multiple sclerosis (MS) have led to improvements in prognosis and life expectancy (Finlayson, 2009) and thus a larger cohort of older individuals with MS. Approximately 9% of MS patients are older than 65 years (Awad & Stüve, 2010). Cognitive impairment remains one of the least investigated areas in these individuals (Vargas & Arnett, 2014). Both cross-sectional (Bodling, Denney, & Lynch, 2009; Smestad, Sandvik, Landrø, & Celius, 2010) and longitudinal studies (Amato, Ponziani, Siracusa, & Scorbi, 2001; Bergendal, Fredrikson, & Almkvist, 2007; Schwid, Goodman, Weinstein, McDermoot, & Johnson, 2007; Strober, Rao, Lee, Fischer, & Rudick, 2014) show that deficits in cognitive performance worsen over the course of MS. A study of Norwegian patients diagnosed with MS between 1940 and 1980 found that 48% met the investigators’ criteria for cognitive impairment after 30 years of disease duration. Longitudinal studies have reported the percentage of patients who shift from an intact to an impaired cognitive profile over a 10-year period ranges from 5% (Schwid et al., 2007) to 30% (Amato et al., 2001). Other studies report shifts of 18% over 18 years (Strober et al., 2014) and 16% over eight years (Bergendal et al., 2007). Comparisons across these studies are complicated by differences in the operational definitions of impairment. However, all confirm that cognitive performance declines in MS patients as they age, and furthermore, that the declines generally occur in the two areas most commonly affected by MS from the outset of the disease, processing speed and memory (Chiaravalloti & DeLuca, 2008; DeLuca, Yates, Beale, & Morrow, 2014).

An older patient with an existing diagnosis of MS who raises concerns about cognitive function poses a challenge in differentiating these concerns from normal aging, symptoms associated with the established medical history, or, alternatively, the onset of a comorbid dementing condition. Little is known about how the cognitive profile of older patients with MS differs from those of healthy older adults or individuals with mild cognitive impairment. Only one previous study (Müller et al., 2013) bears on this topic. It featured a comparison among patients with secondary progressive MS (SPMS), patients with amnestic mild cognitive impairment (aMCI), and healthy older individuals, with all samples equivalent in age (M = 61). Healthy individuals performed better than both patient groups on all tests comprising the neuropsychological battery. No differences were observed between SPMS and aMCI patients on any measure, except delayed recognition of a word list. On this measure, aMCI patients had significantly lower scores than SPMS patients, whose performance was comparable to that of healthy controls.

The present study is a further attempt to characterize the cognitive profile of older adults with MS using an extensive set of neuropsychological measures. The measures were derived from version 2 of the Uniform Data Set (UDS) neuropsychological test battery (Weintraub et al., 2009), a collection of 10 standardized tests used by Alzheimer’s Disease Centers across the country. At the site of the present study, the UDS battery was supplemented with four additional tests that were also included in this study. The battery was administered to a sample of older patients with MS, and performance was compared to archival data for a sample of healthy older adults and a sample of aMCI patients. Thus the present study was comparable to the preceding one by Müller et al. (2013) but with a larger, older, and more diverse sample of MS patients consisting of all three major subtypes of MS. We hypothesized that MS patients would show the greatest impairment on tests of processing speed and memory, relative to healthy older adults. Of primary interest, however, were cognitive variables that could distinguish between the MS and aMCI samples. Based on Müller et al.’s findings, we expected to find that aMCI patients differed from MS patients on measures of memory. Some investigators have argued that the memory problems in MS patients involve deficits in retrieval processes rather than acquisition or storage (Chiaravalloti & DeLuca, 2008; DeLuca, Gaudino, Diamond, Christodoulou, & Engel, 1998; Guimarães & Sá, 2012). On the other hand, aMCI typically involves a broader impairment in memory, encompassing deficits in free and delayed recall (Chen et al., 2000; Tabert et al., 2006). Although other cognitive abilities are thought to remain basically intact in aMCI (Petersen et al., 2001), additional deficits have been observed in verbal fluency, object naming, psychomotor speed, visual spatial processing, and attention (Tabert et al., 2006). Any of these domains might offer measures for distinguishing between aMCI and MS patients.

Method

Participants

Sixty-four individuals between the ages of 60 and 80 (M=66.1; SD=4.5) who met the revised McDonald criteria for clinically definite MS (Polman et al., 2011) participated in the study. The patients were diagnosed with relapsing (n=23), primary progressive (n=19), or secondary progressive MS (n=22). All were under the care of the same neurologist (SGL) at the University of Kansas Medical Center. The length of time since being diagnosed with MS ranged from 2 to 47 years (M=21.2; SD=11.5). Disability ratings based on the Expanded Disability Status Scale (EDSS; Kurtzke, 1983) administered by the neurologist at the time of recruitment ranged from 1.0 to 8.0 (Md=5.0). Eighty-three percent of the patients were retired, with half indicating retirement was due to disability. Exclusionary criteria were the following: relapse within the past three months; neurological disorder other than MS; history of drug or alcohol abuse; severe visual impairment or no color vision; impairment in the use of their dominant hand; severe cognitive or psychiatric impairment of sufficient magnitude to interfere with the ability to comprehend testing instructions or provide informed consent; or less than a high school education.

The aMCI and healthy control (HC) samples were extracted from archival data collected through the University of Kansas Alzheimer’s Disease Center (ADC). Individuals were self-referred to the ADC to participate in longitudinal research on cognition and aging. Individuals completed neuropsychological testing and a Clinical Dementia Rating interview (CDR; Morris, 1993) annually. Based on these results, participants were classified into disease groups via consensus among a neurologist, neuropsychologist, and nurse clinician. A sample of 58 patients classified as aMCI with a CDR of 0.5 on their first visit to the ADC was used in this study. A sample of 70 healthy adults with a CDR of 0 and no evidence of impairment on testing, who met the relevant inclusion/exclusion criteria for the study, was also selected from the ADC database. Ages ranged from 61 to 80 (M=71.5; SD=4.9) for the aMCI sample and from 60 to 77 (M=66.3; SD=4.5) for the HC sample.

Measures

Cognitive measures were derived from the 10 standardized tests included in version 2 of the UDS (Weintraub et al., 2009) and the four supplemental tests employed by the local ADC. The supplemental tests were Block Design (Wechsler, 1987), Letter Number Sequencing (Psychological Corporation, 1997), the Selective Reminding Test (Buschke, 1984; Grober & Buschke, 1987), and the Kaplan version of the Stroop Test (Comalli, Wapner, & Werner, 1962; Mitrushina et al., 2005). All tests and outcome measures are listed according to their principal cognitive domain in Table 1. Three derived scores noted in the table warrant further explanation. Scores on both trials of the Category Fluency Test were combined into a single score (CF). Also, proportionate “interference” scores were computed for both the Trail Making Test (Trail-I) and the Stroop Test (Stroop-I) to reflect the extent to which performance on the more complicated trial (Trail-B or color-word naming) was impacted by the additional stimulus burden relative to the simpler trial (Trail-A or color naming). In each instance, the difference between the scores on the two trials was divided by the score on the simpler trial. Interference scores were calculated because previous research within the MS literature suggests failing to control for baseline performance, particularly on measures heavily dependent on processing speed, can lead to the appearance of deficits in higher order skills (e.g., executive function; Denney & Lynch, 2009). The derived scores replaced some of the original scores used in their computation, and therefore, as noted in Table 1, the scores for CF-A, CF-V, Trail-B, and Stroop-S were not analyzed in this study.

Table 1.

Cognitive measures organized by domain

Domain Test Order Abbrev. Outcome Score
Screener Mini-Mental State Examination 1 MMSE Total score
Attention Digit Span Forward
Digit Span Backward		 3
4		 DSF
DSB		 Number correct
Number correct
Letter Number Sequencing a 17 LNS Number correct
Processing
speed		 Digit-Symbol 9 D-S Number correct
Trail Making Test, Part A 7 Trail-A Completion time
Stroop Test, color naming a
Stroop Test, word reading a 		13
14		 Stroop-C
Stroop-W		 Number correct
Number correct
Executive
function		 Trail Making Test, Part B b
  Interference Score c 		8 Trail-B
Trail-I		 Completion time
(Trail B - A) / A
Stroop Test, col or-word naming a,b
  Interference Score c 		15 Stroop-S
Stroop-I		 Number correct
(Stroop C - S) / C
Memory Logical Memory, Story A (WMS-R)
 Immediate Recall
 Delayed Recall
2
10
LM-I
LM-II
Number correct
Number correct
Selective Reminding Testa
 All Trials, free recall
 All Trials, free & cued recall		 16
SRT-F
SRT-FC
Number correct
Number correct
Language Boston Naming Test 11 BNT Number correct
Category Fluency, animals b
Category Fluency, vegetables b
  Total Scorec 		5
6
 CF-A
CF-V
CF		 Number correct
Number correct
CF-A + CF-V
Visual
spatial
function		 Block Design a 12 BD Total points
Open in a new tab

a Supplemental test, not part of the Uniform Data Set.

b Scores for this measure were used in computing the derived score that follows and were not analyzed.

c Derived score

In addition to the cognitive tests, all participants completed the short form of the Geriatric Depression Scale (GDS), a 15-item questionnaire designed for use with older adult populations (Sheikh & Yesavage, 1986). The scale consists of yes/no questions pertaining to how the individual has felt during the past week.

Procedure

All participants provided signed informed consent for this study. MS patients meeting inclusion and exclusion criteria were introduced to the study during the course of their regular appointment at the MS Clinic. If interest was expressed, a research assistant met with the patient to obtain written consent and schedule an appointment for later testing. The tests were administered in the same order, beginning with the UDS battery followed by the supplemental tests and the GDS. The order of administration for the cognitive tests is indicated in Table 1. The testing session lasted approximately 60 minutes.

Statistical Analyses

Data analysis was conducted using IBM SPSS Version 23. Overall group differences were assessed with one-way analyses of variance. Significant findings were then followed by two a priori planned comparisons: MS vs. HC and MS vs. aMCI. Comparisons between aMCI patients and healthy controls were not considered relevant to this study.

Binary logistic regression was used to identify variables that best differentiated between MS and aMCI patients. The variables consisted of age, sex, depression score, and certain cognitive measures selected on the basis of the univariate comparisons between the two groups. Multicollinearity among these variables was examined using multiple regression. In order to mitigate suppression effects, variables were entered in the logistic analysis using a backward stepwise method with probabilities of .05 to enter and .10 to remove. The Hosmer & Lemeshow R2 coefficient (RHL2) served to reflect the accuracy of the final model.

Scaled scores were used to investigate patterns of impairment across different cognitive domains for the MS and aMCI samples. The scores from the UDS battery were converted to age-corrected scaled scores (M = 10, SD = 3) using an online calculator that incorporates normative UDS data from a large national sample (Shirk et al., 2011). Four other normative samples were used to convert the raw scores on the supplemental tests into comparable age-corrected scaled scores (Ivnik et al., 1992; Ivnik, Malec, Smith, Tangalos, & Peterson, 1996; Ivnik et al., 1997; Psychological Corporation, 1987). Impairment was defined as a scaled score of 6 (z-score −1.33) or lower – thus, approximately 1.5 standard deviations below the normative mean. The number of tests with impaired performance was summed both within and across cognitive domains in order to classify an individual’s cognitive status as intact or impaired. Cognitive impairment, at an individual level, was defined as impaired performance on at least one test within two or more cognitive domains. Impairment data was examined using 2X2 chi-square analyses with p-values based on Fisher’s Exact Test (FET).

Results

Demographic information for the three groups of participants is summarized in Table 2. The groups did not differ in years of education, and the MS and HC groups did not differ in sex or age. However, the MS group consisted of a larger percentage of females (77%) than the aMCI group (40%; χ2(1) = 17.1, p < .001) and was also younger (M = 66.1) than the aMCI group (M = 71.5; t(120) = 6.32; p <.001; d = 1.15). The groups also differed significantly in their scores on the GDS; MS patients endorsed more depressive symptoms than either the HC (t(132) = 5.22; p < .001; d = 0.93) or aMCI group (t(120) = 2.08; p = .040; d = 0.38). To mitigate these differences between the samples, univariate comparisons between MS and aMCI patients on the cognitive measures were performed with age, sex, and depression scores entered as covariates, and similarly univariate comparisons between MS and HC groups were performed with depression scores entered as a covariate. Before proceeding with each analysis of covariance, the covariates were examined with respect to the assumption of homogeneity of slopes. Six instances of failure to meet this assumption were found. In comparing MS and aMCI groups, heterogeneity of slopes occurred for age on the Stroop-I and for depression scores on the Boston Naming Test and Block Design. In comparing MS and HC groups, heterogeneity of slopes occurred for depression scores on Digit Span Forward and Backward and on Category Fluency. In each of these instances, the heterogeneous covariate was excluded from the analysis of the respective cognitive measure.

Table 2.

Between-group comparisons on demographic variables and depression.

Group MS
vs
aMCIa 		MS
vs
HCa
MS aMCI HC
p p
Sex (M/F) 15/49 35/23 13/57 <.001 .529
M SD M SD M SD
Age 66.1 4.5 71.5 4.9 66.3 4.5 <.001 .791
Education
(Yrs)		 15.3 2.3 16.1 3.0 15.6 2.5 .110 .514
Depression b 3.3 3.1 2.1 2.8 1.0 1.6 .040 <.001
Open in a new tab

a Paired comparisons for sex are based on chi-square, with p-values reported for Fisher’s Exact Test. Paired comparisons for age, education, and depression are based on independent t tests.

b Based on the Geriatric Depression Scale

Analysis of Cognitive Measures

One member of both the MS and aMCI groups had a few missing data points on cognitive measures (affecting 0.28% of MS and 0.61% of aMCI data). Consequently, some analyses were performed with a slightly smaller sample size than the overall group, as indicated in the body of the results section and in table footnotes. All omnibus analyses examining group differences on cognitive variables were statistically significant. Table 3 lists the results of a priori contrasts between the MS group and the HC or aMCI groups, with group differences in age, sex, and depressive symptoms controlled for through covariance as previously indicated. Significant differences between the MS and HC groups occurred on most of the cognitive measures except for Block Design, Boston Naming Test, Digit Span Backward, Letter Number Sequencing, total recall on the Selective Reminding Task, and the interference scores on the Stroop and the Trail Making Test. In every instance, MS patients performed worse than the HC group. The most robust differences, based on effect sizes, occurred on the Digit-Symbol test (d = 0.74) followed by the memory tests (LM-I: d = 0.73; LM-II: d = 0.67; SRT-F: d = 0.70). Compared to the aMCI group, MS patients differed significantly on the Boston Naming Test (d = 0.51), MMSE (d = 0.49), the interference scores on both the Trail Making Test (d = 0.44) and the Stroop Test (d = 0.49), and both scores on the Selective Reminding Test (SRT-F: d = 0.38; SRT-FC: d = 0.49). MS patients performed better than aMCI patients on each of these measures.

Table 3.

Between-group comparisons on cognitive scores

Group MS
vs
aMCI a 		MS
vs
HC a
Cognitive
variable		 MS (n=64) aMCI (n=58) HC (n=70)
M SD M SD M SD p d p d
MMSE b 28.5 1.9 27.6 2.1 29.5 0.9 .008 0.49 .008 0.47
DSF 7.9 2.0 8.1 1.8 8.7 1.9 .408 0.15 .012 0.44
DSB 6.0 2.1 5.7 1.9 6.6 2.1 .682 0.07 .104 0.17
LNS c 9.3 2.8 8.5 2.4 10.3 2.4 .603 0.09 .132 0.26
D-S b 39.3 13.6 40.0 11.7 50.9 10.2 .091 0.31 <.001 0.74
Trail-A b 40.9 23.8 37.2 17.4 28.3 9.2 .067 0.33 .005 0.49
Trail-I b 164.2 92.6 211.8 104.9 172.7 80.9 .017 0.44 .577 0.10
Stroop-C b,c,d 65.3 17.9 66.5 12.7 74.0 11.6 .085 0.32 .006 0.49
Stroop-W c,d 86.4 17.1 87.0 15.6 95.8 13.9 .154 0.26 <.001 0.64
Stroop-I b,c,d 49.9 12.0 55.4 12.7 47.5 11.8 .008 0.49 .113 0.28
LM-I 11.7 4.1 9.8 4.3 14.6 3.5 .543 0.11 <.001 0.73
LM-II 9.8 4.5 7.6 4.6 13.0 3.8 .194 0.24 <.001 0.67
SRT-F b,c 26.7 5.6 22.0 8.1 31.3 4.3 .039 0.38 <.001 0.70
SRT-FC b,c 47.9 0.4 45.9 6.1 47.9 0.4 .008 0.49 .846 0.03
BNT b 28.4 1.7 26.9 3.8 28.4 1.5 .006 0.51 .768 0.05
CF b 33.3 8.0 27.9 7.9 38.8 8.0 .078 0.32 <.001 0.68
BD 32.5 11.7 28.4 8.7 35.0 11.7 .142 0.27 .683 0.07
Open in a new tab

a P-values and effect sizes are reported for planned contrasts between the MS and aMCI groups and between the MS and HC groups. In the former comparisons, age, sex, and depression score were included as covariates to correct for differences between the MS and aMCI groups in these demographic variables. In the latter comparisons, only the depression score was used as a covariate because the MS and HC groups did not differ in age or sex.

b Cognitive variable used (along with age, sex, and depression score) in binary logistic regression analyses comparing MS vs. aMCI patients.

c The aMCI sample size for this measure was 57 due to a missing data point.

d The MS sample size for this measure was 63 due to a missing data point.

A binary logistic regression analysis was performed to determine how well the cognitive variables collectively distinguished between the MS and aMCI cohorts and which particular variables were most effective in doing so. The MS sample served as the reference group. The combined sample size exceeded the minimum number recommended by Green for multivariate analyses of this kind (Field, 2009). The predictor variables included age, sex, depression scores, and the subset of cognitive variables that fulfilled one of two criteria: (a) the variable significantly distinguished the MS and aMCI groups; or (b) the variable significantly distinguished the MS and HC groups and had a near-significant value (p < .10) for the comparison between the MS and aMCI groups. As noted in Table 3, this subset consisted of the MMSE, Digit-Symbol, Trail-A and interference score from the Trail Making Test, Stroop-C and the interference score for the Stroop Test, both recall scores from the Selective Reminding Test, the Boston Naming Test, and Category Fluency. No issues involving multicollinearity were found when this set of predictors was examined using multiple regression. As stated earlier, variables were entered into the logistic regression analysis in a backward stepwise fashion, with probabilities of .05 to enter and .10 to remove.

The results of the analysis are summarized in Table 4. The resulting model significantly differentiated between MS and aMCI patients (χ2 (7) = 80.7.4,p < .001, RHL2 = .514). Fifty-four of 63 MS patients (86%) and 43 of 57 aMCI patients (75%) were correctly classified. When considering the specific predictors retained in the model, three points are important to keep in mind. First, because the MS sample served as the reference group, the B value for each predictor in Table 4 reflects the probability of membership in the aMCI group. Thus, for example, the positive B value for age indicates that the likelihood of being in the aMCI group increased with age. Second, sex was coded with females as the reference value (i.e., F=0, M=1); therefore, the positive sign on the B value for sex indicates that the likelihood of membership in the aMCI group was greater for males than for females. Finally, interpretation of the B value for each cognitive variable must take into consideration the direction of the score for that variable. The positive B value for D-S indicates the likelihood of membership in the aMCI group increased with better performance on the Digit-Symbol Test; but the positive B value for Trail-I or Stroop-I indicates this likelihood also increased with greater interference on these tests, and the negative B values for SRT-FC and BNT indicates this likelihood decreased with better performance on these measures. To summarize, the likelihood of membership in the aMCI group instead of the MS group was associated with greater age, male gender, better Digit-Symbol score, greater interference on the Trail Making and Stroop Test, worse Selective Reminding Test recall, and worse BNT score.

Table 4.

A logistic regression model to distinguish aMCI patients from MS patients

RHL2 Accuracy Predictors Retained
MS
# (%)		 aMCI
# (%)		 All
# (%)		 Variable B Wald p
.514 54/63 (86%) 43/57 (75%) 97/120 (81%) Age 0.250 14.90 <.001
Sex (M) 2.431 13.00 <.001
D-S 0.096 10.76 .001
BNT −0.345 4.932 .026
SRT-FC −0.950 3.99 .046
Trail-I 0.007 5.83 .016
Stroop-I 0.052 3.77 .052
Constant 38.73
Open in a new tab

Note: RHL2 = Hosmer & Lemeshow R2 coefficient; D-S = Digit-Symbol Test; BNT = Boston Naming Test; SRT-FC = free and cued recall score on the Selective Reminding Test; Trail-I = interference score on the Trail Making Test; Stroop-I = interference score on the Stroop Test.

In addition to analyzing raw scores, performance on each cognitive measure was also converted to an age-corrected scaled score in order to determine impairment in MS and aMCI patients, as described in the Method section. Impairment in two or more domains was the criterion for designating a patient as impaired. The percentage of impaired individuals in the MS and aMCI samples was similar; 36 of 64 (56%) MS patients and 35 of 58 (61%) aMCI patients met this criterion (FET: p = .584).

To determine whether the patterns of impairment differed between the MS and aMCI sample, we examined impairment within each of the six cognitive domains. As described previously, impairment within a particular cognitive domain was defined as approximately 1.5 standard deviations (i.e., a scaled score of 6 or z-score of −1.33) below the normative mean on at least one measure within the domain. The percentages of impairment in each of the six cognitive domains are presented in Figure 1. A significantly greater percentage of MS patients than aMCI patients was impaired in the processing speed domain (57% vs 37%; FET: p = .043). The percentage impairment was higher for MS patients on each of the four measures comprising this domain; however, the difference was only significant in the case of the Stroop-C measure (22% vs 5%; FET: p = .009) and the Stroop-W measure (24% vs 9%; FET: p = .030). Conversely, significantly greater percentages of aMCI patients than MS patients were impaired in the memory domain (61% vs 39%; FET: p = .018) and the language domain (53% vs 31%; FET: p = .017). The percentage impairment was higher for aMCI patients on each of the measures comprising these domains, but the differences were significant on only the Selective Reminding Test (18 % vs 3%; FET: p = .012) and the Category Fluency Test (52% vs 31%; FET: p = .027).

Figure 1.

Figure 1

Open in a new tab

Percentage impairment within each cognitive domain for MS and aMCI groups

Discussion

Cognitive measures successfully distinguished older MS patients from healthy individuals of a comparable age and sex distribution. The most consistent and robust differences between these groups (d ranging from 0.49 to 0.74; Table 3) occurred on measures of processing speed (Digit-Symbol, Trail Making Test-A, Stroop Test – color naming and word reading) and memory (Logical Memory Test–I & II, Selective Reminding Test – free recall). Although a substantial difference also occurred on the Category Fluency Test (d = 0.68), this result does not override the preceding conclusion. The Category Fluency Test was classified within the language domain in this study, but scores on this test are also heavily influenced by processing speed. Indeed, when the Category Fluency scores are adjusted for differences in processing speed (i.e., by dividing by each participant’s score on the Stroop word reading trial), the difference between MS patients (M = .395, S.D. = .091) and healthy controls (M = .411, S.D. = .094) is no longer significant (t(131) = 0.99, p = .323), indicating that the original difference on this test may be attributable to processing speed rather than to language skill per se. It is also relevant to note here that the non-speeded measure within the language domain, the Boston Naming Test, showed no significant difference between MS patients and controls.

This pattern of cognitive deficits in processing speed and memory in the present sample of older patients with MS is consistent with that seen in the MS population at large (Benedict et al., 2012; Bobholz & Rao, 2003; Chiaravalloti & DeLuca, 2008; DeLuca et al., 2014; Guimaräes & Sá, 2012). The consistencies are remarkable in their details. For example, with respect to memory, older MS patients differed from healthy controls in their free recall scores on the Logical Memory Test and on the Selective Reminding Test. However, when cues were provided in the latter test, patients’ scores improved such that their overall accuracy no longer differed from that of the controls. Similar patterns have been reported in other studies of MS patients’ performance on various memory tasks (Rao et al., 1989; DeLuca, Barbieri-Berger, & Johnson, 1994; DeLuca et al., 1998).

Another example of the consistency with the broader MS population is found in results for the Stroop Test. Several studies have shown that when patients’ performance on a set of Stroop stimuli is effectively adjusted for their deficit in processing speed as indicated by their performance on the preliminary color-naming trial, no difference in interference is observed between MS patients and healthy controls (Bodling, Denney, & Lynch, 2008; Denney, Gallagher, & Lynch, 2011; Denney & Lynch, 2009; Pujol et al., 2001; van Dijk, Jennekens-Schindel, Caekebeke, & Zwinderman, 1992). In the present study, the proportionate interference scores on the Stroop Test confirmed this finding and also showed the same result to be true for the Trail Making Test. Since these two proportionate interference scores were the only measures within the executive function domain of this study, no differences in executive function were evident between MS patients and controls. Previous studies by our group have demonstrated that deficits in executive function are not characteristic of MS. When the test places no burden on processing speed or when processing speed is properly controlled within the measure of executive function, we have shown MS patients perform on a par with healthy individuals (Denney, Hughes, Owens, & Lynch, 2012; Owens, Denney, & Lynch, 2013; Roth, Denney, & Lynch, 2015). A similar outcome was noted earlier: an apparent difference between MS patients and controls in the language domain was also nullified when scores on the Category Fluency Test were likewise adjusted for processing speed.

The main focus of this study, however, was the comparison between MS and aMCI patients. We used a variety of analyses to examine these differences. On the domain level, individual test scores were converted to age-corrected scaled scores and then combined to determine whether patients were impaired on each of the six domains. This analysis revealed three differences between the MS and aMCI groups. As illustrated in Figure 1, MS patients were classified as impaired more frequently in the processing speed domain, whereas aMCI patients were classified as impaired more frequently in the memory and language domains.

Univariate comparisons between MS and aMCI patients based on individual test scores were computed with age, sex, and depression scores serving as covariates. MS patients exhibited somewhat poorer performance than aMCI patients on measures of processing speed such as Digit-Symbol, Trail Making Test-A, and the Stroop – color naming trial. Although none of these differences reached significance (p ranging from .067 to .091), they were associated with medium effect sizes (d ranging from 0.31 to 0.33), and, collectively, appear to account for the difference in processing speed noted at the domain level. Significant differences were found on the MMSE, the Boston Naming Test, the interference score on both the Trail Making Test and the Stroop Test, and both scores on the Selective Reminding Test. In each instance, aMCI patients had worse scores than the MS patients.

Given that the aMCI patients were individuals experiencing difficulties with memory, it is perhaps not surprising to find their performance on memory measures to be worse than that of the MS patients. However, as noted earlier, MS patients also had problems with memory when compared to healthy controls. One distinction needs to be highlighted. Although neither group performed on a par with healthy controls on the free recall measure of the Selective Reminding Test, when assistance was provided in the form of cues, MS patients’ performance no longer differed from that of controls. Assistance of this kind appears to mitigate memory problems to a greater extent for MS patients than for aMCI patients. Therefore, the more discerning measures of memory for distinguishing aMCI from MS may be those obtained when cues are provided to aid retrieval. This is only a tentative conclusion at present, although it is consistent with the findings reported by Müller and colleagues (2013). In the present study, the difference between the two groups on the free and cued recall score from the Selective Reminding Test was only moderate (d = 0.49) and was likely compromised by ceiling effects. A total of 78% of the patients (91% MS; 65% aMCI) obtained perfect scores on this measure.

Although Müller et al. (2013) found a difference between aMCI and MS patients on only a single measure of memory, we found differences in measures originating from two additional domains. The interference score on the Trail Making Test reflects the amount of difficulty encountered in alternating between letters and numbers when completing Trail B and is independent of performance on the simpler numerical sequencing task on Trail A. This form of switching is commonly viewed as a type of executive function (Chan, Shum, Toulopoulou, & Chen, 2008; Friedman et al., 2008). Similarly, the interference score on the Stroop Test reflects the difficulty in deploying attention to the color of the print while avoiding the more automatic response to the printed word itself. Selective attention and inhibition is another commonly agreed upon type of executive function. Whereas MS patients exhibited no difficulty in this domain relative to controls, aMCI patients had significantly greater interference scores on both tests than MS patients.

The aMCI patients also had lower scores than MS patients on the Boston Naming Test, and this was the most robust difference between the two groups of patients (d = 0.51). As a non-speeded test, the Boston Naming Test is a measure of language skill unaffected by processing speed. When the Category Fluency Test is corrected for processing speed as described earlier, the aMCI group also had significantly lower scores on this measure of language (p = .001, d = 60). The present study demonstrates that language skill is an important domain to evaluate when trying to distinguish aMCI from the cognitive impact of MS, but in doing so, it is essential to avoid measures confounded by processing speed.

Separate univariate comparisons fail to take into account interdependence among the various measures. One advantage of multivariate analysis such as the binary logistic regression performed in the present study is that this interdependence is accounted for. Logistic regression also yields an overall model composed of a combination of weighted variables maximized to distinguish between the MS and aMCI groups. The resulting model in the present study accurately distinguished 81% of the patients in the MS and aMCI groups. Part of this accuracy derives from the fact that the aMCI patients tended to be older and more frequently male. Whether these demographic differences constitute a limitation of the study will be considered later. However, even with differences in age and sex accounted for, several cognitive variables contributed significantly to the distinction between the two groups. All of these variables were also identified through the univariate analyses: Digit-Symbol, Boston Naming Test, the interference scores on the Trail Making Test and Stroop Test, and the free and cued recalled score on the Selective Reminding Test. They provide the basis for summarizing the findings of the present study as they relate to the practical issue raised in the introduction, namely that of a clinician confronting an older MS patient with concerns about his or her current cognitive function.

Although unlikely that the resolution regarding these concerns should rest solely on the age and sex of the patient, it is worth noting that the number of MS patients who were male and beyond age 72 was so small that it was impossible to recruit a sample of MS patients that matched the aMCI sample on these demographics. Further, the 3.3:1 female to male gender ratio in the present study is consistent with those reported in the literature (Harbo, Gold, & Tintoré, 2013). Thus, to the extent that these are population differences, the age and sex of the patient raising concerns over his or her cognitive function are worthy of consideration. Beyond this, the findings of the present study can be encapsulated under four general guidelines pertaining to neuropsychological assessment.

  1. Processing speed is seriously compromised in patients with MS, to an even greater extent than in aMCI patients. Therefore, neuropsychological measures used to indicate co-morbid aMCI in patients with MS must have minimal loadings on processing speed. If speed is an inherent feature of the test, then some method of adjusting the score for the patient’s deficit in processing speed must be employed. We have seen illustrations of this guideline in the contrast between speeded and non-speeded measures of language and the proportionate interference scores on the Trail Making and Stroop tests. However, the issue is relevant in virtually all neuropsychological assessments of patients with MS. For example, studies have shown that deficits in processing speed can confound the assessment of memory when the items of the test (e.g., words, story elements) are presented too quickly to allow patients an opportunity to process the information with a sufficiency comparable to their peers (Arnett, 2003).

  2. Although both MS and aMCI patients may be expected to exhibit problems with memory, MS patients appear to derive greater benefit from cues designed to aid retrieval of information from storage. Therefore, memory tests used to indicate co-morbid aMCI in patients with MS should assess recall under aided conditions of this kind. For the same reason, recognition measures may constitute an especially useful alternative, as Müller et al.’s (2013) study indicated.

  3. Whereas aMCI patients perform poorly on measures of executive function, MS patients are comparable to controls on these measures. Therefore, interference measures on tasks such as the Trail Making Test and the Stroop Test are useful for indicating co-morbid aMCI in patients with MS, provided that the patient’s deficit in processing speed is effectively controlled within the measure.

  4. Patients with aMCI also perform poorly on measures of language. Here again MS patients are comparable to controls on these measures. Therefore, a non-speeded naming task such as the Boston Naming Test is useful for indicating co-morbid aMCI in patients with MS.

Like Müller et al.’s study (2013), the present study was aligned with a large, multi-center investigation of Alzheimer’s disease using a mandated neuropsychological battery explicitly designed to evaluate this form of dementia. Several limitations occurred as a result of this alignment. Because the two studies were aligned with different consortium studies, the batteries used in the studies were not the same. The most obvious deficiency in the present battery was that it had no measure of delayed recognition, the one area in which MS and aMCI patients differed in Müller et al.’s study. Furthermore, in neither of these studies was the battery optimized for individuals with MS. For example, the Symbol Digit Modalities Test is much preferred over Digit-Symbol Test used in the present study because participants are able to respond verbally, and therefore difficulties with manual dexterity frequently found in patients with MS can be avoided. The Trail Making Test is also not typically used with MS patients for the same reason. Fortunately, the Stroop Test, which requires a verbal response, was included as a supplemental test by the local ADC, and scores on the preliminary word reading and color naming trials of this test could be used as suitable measures of processing speed. Finally, it would have been useful to have included a non-speeded measure of executive function in the battery since, as noted earlier, performance in this domain is another likely candidate for differentiating aMCI from MS patients.

Another potential limitation of the present study is the disparity between the MS and aMCI groups in age, sex, and depression scores. We would argue that, to the extent they reflect true differences between the two patient populations, the disparities are a valid feature of the study design. As previously mentioned, female to male sex ratios ranging between 2:1 and 3:1 are common within MS samples (Harbo, Gold, & Tintoré, 2013). Similarly, elevated rates of depression, relative to both the general population as well as other chronically ill patient populations, is a common finding in MS (Feinstein, 2001; Patten et al., 2003; Viner et al., 2014). Nevertheless, these demographic disparities pose problems when it comes to determining differences between MS and aMCI groups in cognitive performance. Efforts were made to control for these differences in the univariate analysis of the present data, but the application of covariance in such instances is not an ideal solution. Likewise, age, sex, and depression scores were included in the logistic regression analysis, and both age and sex accounted for much of the accuracy in predicting group membership. However, even with statistical controls of this kind imposed on differences between samples of MS and aMCI patients, several cognitive variables contributed significantly to the distinction between the two groups in both the univariate and multivariate analyses. Another less obvious problem stemming from the disparity between the two groups in sex distribution needs to be noted. Normative data stratified by sex was unavailable for some of the measures in the cognitive battery, and therefore only age was considered when determining whether patients were impaired on these measures. As a result, the analysis of impairment within each of the domains does not completely parallel that of the test scores in which both age and sex were controlled.

The size of the two samples of patients is also a limitation in this study. Although a sample size exceeding 50 for each group is clearly sufficient for univariate comparisons, larger samples would allow much greater confidence in the reliability of the model emerging from the logistic regression analysis. In the present study, the number of participants barely exceeded the recommended minimums for logistic regression analysis. Also, a larger MS sample would have permitted comparisons between separate subtypes of MS (e.g., relapsing or progressive MS) versus aMCI.

Probably the most important limitation in the design adopted by the present study as well as in the earlier study by Müller et al. is that it is only indirectly aligned with the ultimate objective of the research. The MS patients in this study were not individuals raising concerns over recent changes in their cognitive status, and therefore, their comparability to the clinical cohort for which the results are intended to generalize is questionable. Nevertheless, the studies yield some indications of the form that cognitive impairment assumes in older individuals with existing diagnoses of MS and how this may differ from that characterizing individuals with aMCI. The differences can then serve to guide clinicians in their efforts to determine the presence of comorbid aMCI in conjunction with MS.

Public Significance Statement.

When older adult multiple sclerosis (MS) patients report new or worsening cognitive symptoms, multiple etiologies must be considered. This study documents typical cognitive performance in older MS patients and identifies neuropsychological tasks that differentiate the cognitive profiles of patients with MS and amnestic mild cognitive impairment. Differences between patient groups are most readily apparent when processing speed demands are reduced or controlled for on measures of memory, language, and executive function.

Acknowledgements

This study was supported in part by the NIH (P30AG035982). We wish to thank Drs. Amber Watts and David Johnson for the assistance they provided to this project.

References

  1. Amato MP, Ponziani G, Siracusa G, & Scorbi S (2001). Cognitive dysfunction in early-onset multiple sclerosis: A reappraisal after 10 years. Archives of Neurology, 58, 1602–1606. [DOI] [PubMed] [Google Scholar]
  2. Arnett PA (2003) Speed of presentation influences story recall in college students and persons with multiple sclerosis. Archives of Clinical Neuropsychology, 19, 507–523. doi: 10.1016/j.acn.2003.07.006 [DOI] [PubMed] [Google Scholar]
  3. Awad A & Stüve O (2010). Multiple sclerosis in the elderly patient. Drugs & Aging, 27, 283–294. doi: 1170-229X/10/0004-0283 [DOI] [PubMed] [Google Scholar]
  4. Benedict RH, Amato MP, Boringa J, Brochet B, Foley F, Fredrikson S, ... Langdon D (2012). Brief international cognitive assessment for MS (BICAMS): International standards for validation. BMC Neurology, 12, 1–8. doi: 10.1186/1471-2377-12-55 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bergendal G, Fredrikson S, & Almkvist O (2007). Selective decline in information processing in subgroups of multiple sclerosis: An 8-year longitudinal study. European Neurology, 57, 193–202. doi: 10.1159/000099158 [DOI] [PubMed] [Google Scholar]
  6. Bobholz JA, & Rao SM (2003). Cognitive dysfunction in multiple sclerosis: A review of recent developments. Current Opinion in Neurology, 16, 283–288. doi: 10.1097/01.wc.0000073928.19076.84 [DOI] [PubMed] [Google Scholar]
  7. Bodling AM, Denney DR, & Lynch SG (2008). Rapid serial processing in patients with multiple sclerosis: The role of peripheral deficits. Journal of the International Neuropsychological Society, 14, 646–650. doi: 10.1017/S1355617708080739 [DOI] [PubMed] [Google Scholar]
  8. Bodling AM, Denney DR, & Lynch SG (2009). Cognitive aging in patients with multiple sclerosis: A cross-sectional analysis of speeded processing. Archives of Clinical Neuropsychology, 24, 761–767. doi: 10.1093/arclin/acp076 [DOI] [PubMed] [Google Scholar]
  9. Buschke Η (1984). Cued recall in amnesia. Journal of Clinical Neuropsychology, 6, 433–440. [DOI] [PubMed] [Google Scholar]
  10. Chan RC, Shum D, Toulopoulou T, & Chen EY (2008). Assessment of executive function: Review of instruments and identification of critical issues. Archives of Clinical Neuropsychology, 23, 201–216. doi: 10.1016/j.acn.2007.08.010 [DOI] [PubMed] [Google Scholar]
  11. Chen P, Ratcliff G, Belle SH, Cauley JA, DeKosky ST, & Ganguli M (2000). Cognitive tests that best discriminate between presymptomatic AD and those who remain nondemented.Neurology, 55, 1847–1853. [DOI] [PubMed] [Google Scholar]
  12. Chiaravalloti ND & DeLuca J (2008). Cognitive impairment in multiple sclerosis. Lancet Neurology, 7, 1139–51. doi: 10.1016/S1474-4422(08)70259-X. [DOI] [PubMed] [Google Scholar]
  13. Comalli PE Jr., Wapner S, & Werner H (1962). Interference effects of Stroop color-word test in childhood, adulthood, and aging. The Journal of Genetic Psychology, 100, 47–53. [DOI] [PubMed] [Google Scholar]
  14. DeLuca GC, Yates RL, Beale EL, & Morrow SA (2014). Cognitive impairment in multiple sclerosis: Clinical, radiologic, and pathologic insights. Brain Pathology, 25, 79–98. doi: 10.1111/bpa.12220 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. DeLuca J, Barbieri-Berger S, & Johnson SK (1994). The nature of memory impairments in multiple sclerosis: acquisition versus retrieval. Journal of Clinical and Experimental Neuropsychology, 16, 183–189. [DOI] [PubMed] [Google Scholar]
  16. DeLuca J, Gaudino EA, Diamond BJ, Christodoulou C, Engel RA (1998). Acquisition and storage deficits in multiple sclerosis. Journal of Clinical and Experimental Neuropsychology, 20, 376–390. [DOI] [PubMed] [Google Scholar]
  17. Denney DR, Gallagher KS, & Lynch SG (2011). Deficits in processing speed in patients with multiple sclerosis: Evidence from explicitly-timed and covertly-timed measures. Archives of Clinical Neuropsychology, 26, 110–119. doi: 10.1093/arclin/acql04. [DOI] [PubMed] [Google Scholar]
  18. Denney DR, Hughes AJ, Owens EM, & Lynch SG (2012). Deficits in planning time but not performance in patients with multiple sclerosis. Archives of Clinical Neuropsychology, 27, 148–158. doi: 10.1093/arclin/acrl05. [DOI] [PubMed] [Google Scholar]
  19. Denney DR, & Lynch SG (2009). The impact of multiple sclerosis on patients’ performance on the Stroop Test: Processing speed vs. interference. Journal of the International Neuropsychological Society, 75,451–458. doi: 10.1017/S1355617709090730 [DOI] [PubMed] [Google Scholar]
  20. Feinstein A (2011). Multiple sclerosis and depression. Multiple Sclerosis Journal, 17, 1276–1281. doi: 10.1177/1352458511417835 [DOI] [PubMed] [Google Scholar]
  21. Field A (2009). Discoverins Statistics Usins SPSS (3rd Edition). London: Sage. [Google Scholar]
  22. Finlayson M (2009). Multiple sclerosis and aging: Complexities, concerns and considerations for care. Aging Health, 5, 89–102. doi: 10.2217/1745509X.5.1.89 [DOI] [Google Scholar]
  23. Friedman NP, Miyake A, Young SE, Defries JC, Corley RP, & Hewitt JK (2008) Individual differences in executive functions are almost entirely genetic in origin. Journal of Experimental Psychology: General, 137, 201–225. doi: 10.1037/0096-3445.137.2.201 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Grober E, & Buschke H (1987). Genuine memory deficits in dementia. Developmental Neuropsychology, 3, 13–36. [Google Scholar]
  25. Guimarães J & Sá MJ (2012). Cognitive dysfunction in multiple sclerosis. Frontiers in Neurology, 3, 1–8. doi: 10.3389/fneur.2012.00074 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Harbo HF, Gold R, & Tintoré M (2013). Sex and gender issues in multiple sclerosis. Therapeutic Advances in Neurological Disorders, 6, 237–248. doi: 10.1177/1756285613488434 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Ivnik RJ, Malec JF, Smith GE, Tangalos EG, Petersen RC, & Kokmen E (1992). Mayo’s older Americans normative studies: WAIS-R norms for ages 56 to 97. The Clinical Neuropsychologist, 6:S1, 1–30. doi: 10.1080/13854049208401877 [DOI] [Google Scholar]
  28. Ivnik RJ, Malec JF, Smith GE, Tangalos EG, & Peterson RC (1996). Neuropsychological tests’ norms above age 55: COW AT, BNT, MAE Token, WRAT-R Reading, AMNART, STROOP, TMT, and JLO. The Clinical Neuropsychologist, 10, 262–278. doi: 10.1007/s00702-014-1288-x [DOI] [Google Scholar]
  29. Ivnik RJ, Smith GE, Lucas JA, Tangalos EG, Kokmen E, & Petersen RC (1997). Free and cued selective reminding test: MOANS norms. Journal of Clinical and Experimental Neuropsychology, 19, 676–691. [DOI] [PubMed] [Google Scholar]
  30. Kurtzke JF (1983). Rating neurological impairment in multiple sclerosis: an Expanded Disability Status Scale (EDSS). Neurology, 33, 1444–1452. [DOI] [PubMed] [Google Scholar]
  31. Mitrushina M, Boone KB, Razani J, & D’Elia LF (2005). Handbook of Normative Data for Neuropsychological Assessment. Oxford: Oxford University Press. [Google Scholar]
  32. Morris JC (1993). The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology, 43, 2412–2414. [DOI] [PubMed] [Google Scholar]
  33. Müller S, Saur R, Greve B, Melms A, Hautzinger M, Fallgatter AJ, & Leyhe T (2013). Recognition performance differentiates between elderly patients in the long term course of secondary progressive multiple sclerosis and amnestic mild cognitive impairment. Multiple Sclerosis, 19, 799–805. doi: 10.1177/1352458512461392 [DOI] [PubMed] [Google Scholar]
  34. Owens EM, Denney DR, & Lynch SG (2013). Difficulties in planning among patients with multiple sclerosis: A relative consequence of deficits in information processing speed. Journal of the International Neuropsychological Society, 19, 613–620. doi: 10.1017/S1355617713000155. [DOI] [PubMed] [Google Scholar]
  35. Patten SB, Beck CA, Williams JVA, Barbui C, & Metz LM (2003). Major depression in multiple sclerosis: A population-based perspective. Neurology, 61, 1524–1527. [DOI] [PubMed] [Google Scholar]
  36. Petersen RC, Doody R, Kurz A, Mohs RC, Morris JC, Rabins PV, ... Winblad B (2001). Current concepts in mild cognitive impairment. Archives of Neurology, 58, 1985–1992. [DOI] [PubMed] [Google Scholar]
  37. Polman CH, Reingold SC, Banwell B, Clanet M, Cohen J, Filippi M, ... Wolinsky JS (2011). Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald Criteria. Annuals of Neurology, 69, 292–302. doi: 10.1002/ana.22366 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Corporation Psychological. (1997). Wechsler Adult Intelligence Scale (3rd ed.). San Antonio, Texas: Harcourt Assessment. [Google Scholar]
  39. Pujol J, Vendrell P, Deus J, Junque C, Bello J, Marti-Vilalta JL, & Capdevila A. (2001), The effect of medial frontal and posterior parietal demyelinating lesions on Stroop interference. Neuroimage, 13, 68–75 [DOI] [PubMed] [Google Scholar]
  40. Rao SM, Leo GJ, Aubin-Faubert P St. (1989). On the nature of memory disturbance in multiple sclerosis. Journal of Clinical and Experimental Neuropsychology, 11, 699–712. doi: 10.1080/01688638908400926 [DOI] [PubMed] [Google Scholar]
  41. Roth AK, Denney DR, & Lynch SG (2015). Information processing speed and attention in multiple sclerosis: Reconsidering the Attention Networks Test. Journal of Clinical and Experimental Neuropsychology, 37, 518–529. doi: 10.1080/13803395.2015.1037252. [DOI] [PubMed] [Google Scholar]
  42. Schwid SR, Goodman AD, Weinstein A, McDermott MP, & Johnson KP (2007). Cognitive function in relapsing multiple sclerosis: Minimal changes in a 10-year clinical trial. Journal of the Neurological Sciences, 255, 57–63. [DOI] [PubMed] [Google Scholar]
  43. Sheikh JI & Yesavage JA (1986). Geriatric Depression Scale (GDS). Recent evidence and development of a shorter version In Brink TL (Ed.), Clinical Gerontology: A Guide to Assessment and Intervention (pp. 165–173). NY: The Haworth Press, Inc. [Google Scholar]
  44. Shirk SD, Mitchell MB, Shaughnessy LW, Sherman JC, Locascio JJ, Weintraub S, & Atri A (2011). A web-based normative calculator for the uniform data set (UDS) neuropsychological test battery. Alzheimer’s Research & Therapy, 3, 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Smestad C, Sandvik L, Landrø NI, & Celius EG (2010). Cognitive impairment after three decades of multiple sclerosis. European Journal of Neurology, 17, 499–505. doi: 10.1111/j.1468-1331,2009.02889.x [DOI] [PubMed] [Google Scholar]
  46. Strober LB, Rao SM, Lee J-C, Fischer E, & Rudick R (2014). Cognitive impairment in multiple sclerosis: An 18 year follow-up study. Multiple Sclerosis and Related Disorders, 3, 473–481. doi: http://dx.doi.Org/10.1016/j.msard.2014.03.004 [DOI] [PubMed] [Google Scholar]
  47. Tabert MH, Manly JJ, Liu X, Pelton G, Rosenblum S, Jacobs M, ... Devanand DP (2006). Neurological prediction of conversion to Alzheimer’s disease in patients with mild cognitive impairment. Archives of General Psychiatry, 63, 916–924. [DOI] [PubMed] [Google Scholar]
  48. van Dijk JG, Jennekens-Schindel A, Caekebeke JFV, & Zwinderman AH (1992). Are event-related potentials in multiple sclerosis indicative of cognitive impairment? Evoked and event-related potentials, psychometric testing and response speed: a controlled study. Journal of the Neurological Sciences, 109, 18–24 [DOI] [PubMed] [Google Scholar]
  49. Vargas G & Arnett P (2014). Aging with multiple sclerosis: Cognitive, emotional and neuropathological considerations. Neur ode generative Disease Management, 4, 187–194. doi: 10.2217/NMT.14.12 [DOI] [PubMed] [Google Scholar]
  50. Viner R, Fiest KM, Bulloch AGM, Williams JVA, Lavorato DH, Berzins S, ... Patten SB (2014). Point prevalence and correlates of depression in a national community sample with multiple sclerosis. General Hospital Psychiatry, 36, 352–354. doi: http://dx.doi.Org/10.1016/j.genhosppsych.2013.12.011 [DOI] [PubMed] [Google Scholar]
  51. Wechsler A (1987). Wechsler Adult Intelligence Scale-Revised. San Antonio, Texas: Psychological Corporation, 1987. [Google Scholar]
  52. Weintraub S, Salmon D, Mercaldo N, Ferris S, Graff-Radford NR, Chui FL, ... Morris JC (2009). The Alzheimer’s Disease Centers’ Uniform Data Set (UDS): The neuropsychologic test battery. Alzheimer Disease & Associated Disorders, 23, 91–101. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES