Our objective was to compare the neuropsychological performance of 100 patients with MS, 50 with and 50 without the APOE ε4 allele, in a single, prospective sample. We found that attention, processing speed, working memory, verbal memory, visual memory, visual-spatial processing, verbal fluency, and executive functions were no worse in ε4+ patients with MS than in those without the allele. These negative results endured irrespective of whether group comparisons were made between scores on individual cognitive indices, between subjects failing a given cognitive index, or between subjects with overall cognitive impairment. In addition, no age or gender interactions with the ε4 allele were detected on cognitive outcome measures. The inclusion of relevant covariates did not alter the results, nor did changing the criteria for test failure or cognitive impairment to capture either milder or more severe dysfunction that may have been masked by cutoff effects.
The absence of an ε4 effect on cognitive outcomes cannot be attributed to confounding external variables between ε4+ and ε4− MS groups in this study. The 2 groups were well-matched on demographic (age, gender, ethnicity, education) and disease characteristics (disease duration, disease course, EDSS, 9HPT, TWT). In addition, ε4+ and ε4− groups did not differ with respect to estimates of premorbid IQ, depressive symptoms, or the proportion of patients taking disease-modifying medication. The ε2 allele, suggested to be a protective factor for AD,25
was also similar in frequency in the 2 groups, and additional analyses excluding the 15 ε2 carriers did not reveal cognitive differences between ε4+ and ε4− patients with MS. This study was not, however, powered to detect a possible protective effect of ε2. In the present study, distributions of gender, ethnicity, education, and disease course were representative of the local, community-dwelling MS population. Consistent with previous studies of APOE
, the prevalence of the ε4 allele was 21%.26
The prevalence of cognitive impairment in our sample, 41%, also fits with the published literature.1
The 90-minute cognitive battery employed in this study, the MACFIMS, is more comprehensive than the Brief Repeatable Battery of Neuropsychological Tests (BRNB), which was used in 3 previous studies of ε4 in MS. The MACFIMS addresses 2 shortcomings of the BRNB. First, it includes tests of executive function and visuospatial orientation, cognitive functions that may each be compromised in up to one quarter of patients with MS21,27
and, in some elderly and AD populations, have been associated with ε4.28-31
Second, the MACFIMS replaced the verbal and visual memory tests used in the BRNB with tests possessing better psychometric properties and larger normative samples.20
When compared in patients with MS, the MACFIMS and the BRNB had similar sensitivity with respect to verbal learning tests but the visual memory test in the former possessed greater discriminative validity.32
Criterion validity of the MACFIMS has been amply demonstrated; each constituent test differentiated patients with MS from normal controls with medium to very large effect sizes.21
Finally, it is important to note that the recommended cutoff for a patient to be designated cognitively impaired using the MACFIMS has ecological validity; cognitive impairment defined by the MACFIMS sensitively predicted the employment status of patients with MS.21
Six studies to date have examined the relationship between APOE
and cognition and the results are equivocal. In part, this may reflect methodologic limitations. The first report8
found no association between ε4 and cognitive impairment but with only 12 ε4+ subjects, type II error was a distinct possibility. A subsequent study9
tested 503 patients with MS—of whom 74 were ε4+—using the Mental Deterioration Battery.33
The sensitivity of this battery to cognitive dysfunction in MS is not clear and the ecological validity of cutoffs for impairment are likewise untested.34
No association was found between ε4 and overall cognitive impairment. A series of post hoc subgroup analyses revealed an association between severe cognitive impairment and ε4 in males only, but age, education, and disability were potential confounders in this result. Cognitive impairment in females did not correlate with any variable investigated including age and education. The authors conceded that they could not explain the gender effect, a finding that that raises the possibility of sampling problems. A third study used a variation of the BRNB for cognitive evaluation in ε4+ and ε4− patients with MS matched on demographic and disease variables. A single cognitive composite score did not differ between the groups. In further analyses a more stringent threshold for cognitive impairment found a correlation with ε4. In a fourth study, the BRNB was used with validated cutoffs for failure on individual cognitive tests (i.e., below the fifth percentile of scores obtained by healthy controls).1,2,34
ε4 carriage emerged as a predictor of verbal learning impairment. This finding was limited in part by a modest sample of 23 ε4+ patients and by the absence of data on the potentially confounding variable of premorbid IQ. The latter omission may lead to overestimating memory deficits in individuals with low-average IQ.35
used the Neuropsychological Screening Battery for MS (BRNB minus the SDMT) to study ε4+ subjects who were slightly older than ε4− subjects. Estimates of premorbid IQ and frequency of the ε2 allele were not reported. The odds ratio of failing the verbal memory test in ε4+ vs ε4− patients was 2.1 (p
= 0.035). There were no other cognitive differences. In a secondary analysis, the young cohort of ε4+ patients (age 31–40) was found to be especially susceptible to verbal learning deficits. These secondary results are difficult to interpret, however, since the statistical tests employed and the subgroup sample sizes were not provided. Finally, a recent study13
found no difference between ε4+ and ε4− patients with MS on a single cognitive index, the SDMT.
The limitations of previous studies preclude any firm conclusions. Our study, in contrast, controlled for many of these problems with an adequate sample size; detailed, validated neuropsychological inquiry; close group matching; and analyses that took into account any modifying effect of demographic variables. Our negative result therefore presents the most robust evidence to date that presence of the ε4 allele has no effect on cognition in MS. Notwithstanding these conclusions, this study is not without its drawbacks. Like previous studies, our investigation was cross-sectional in design. An effect of ε4 on the rate of cognitive decline therefore cannot be excluded. In addition, our sample of 100 patients with MS had statistical power to detect a medium but not a small effect size. The clinical relevance of a small effect size may, however, be questionable and would challenge routine screening of patients with MS for possession of the ε4 allele. In the present study, patients with MS older than 65 years were excluded due to the possible confounding influence of age-related cognitive decline. The presence of an ε4 effect on cognition in older patients with MS therefore remains to be determined. It is also possible that ε4 is associated with cognitive deficits in subgroups of patients with MS that were not highly represented in our community-based sample; for example, in individuals with severe physical disability. An additional limitation of this study was that we were unable to determine a possible dose effect of the ε4 allele.
The search for heritable underpinnings of MS is not new.36
Using cognitive dysfunction as a phenotype in genetic studies of MS is, however, a more recent development.37
Our negative results should not limit this line of inquiry. Longitudinal data relating to putative ε4 effects are needed. Furthermore, research exploring the potential relationship between other genes and cognitive dysfunction may prove more fruitful.