Why is this important to me?
Between 40 and 65% of people with MS experience some level of cognitive impairment, which includes changes in memory, planning and attention ability, and other mental tasks. Cognitive changes can begin early in the disease process, and many individuals ask about how to identify these changes if and when they begin. Although there are several tests of cognitive ability, no one has determined which are the most reliable.
What is the objective of this study?
The authors reviewed features of cognitive impairment in MS and the neuropsychological tests that are used to assess cognitive ability. Studies suggest that:
- Cognitive problems can occur very early in MS. About a quarter of people with a type of pre-MS called clinically isolated syndrome (or CIS) are experiencing some cognitive changes.
- Cognitive problems occur in all types of MS (relapsing-remitting, primary progressive, secondary).
- Cognitive decline is unlikely to improve on its own, and instead progresses slowly.
- Anxiety and depression, which are common among some individuals with MS, may make cognitive problems worse. It is important to note, however, that this relationship has not been proven.
- Children with MS may experience cognitive changes more rapidly than do adults with MS.
- Behavioral interventions instead of drug treatment may be more effective for individuals experiencing cognitive changes.
These are the test most frequently used to assess changes in cognition in individuals with MS. These tests may be administered singly or in clusters:
- Mini-mental state examination evaluates your awareness of time and place, attention, calculation, recall, language, and other aspects of cognitive function.
- Montreal Cognitive Assessment tests memory, orientation, and language fluency (ability to list as many words as possible beginning with a particular letter in a set amount of time).
- Symbol digit modalities, digit symbol substitution, faces symbol, and paced auditory serial addition tests assess attention and information processing speed abilities.
- Free recall and recognition test evaluates memory and adjusts for individuals who have varying severity or progression of MS.
- Everyday problems test assesses ability to perform such everyday tasks as in meal preparation, medication use, phone use, shopping, money management, transportation, and household management.
- Patient-reported outcomes measurement information system allows patient self-assessment of physical health, mental health, and social well-being.
It is important to understand that the tests described in this article provide an overall assessment of an individual's cognitive status, and they are not useful in monitoring the effectiveness of behavioral intervention. The tests are taken in a standardized environment such as a clinic and may not assess your everyday cognitive abilities. Future research should work to develop test(s) that will detect cognitive problems early in the disease course and that will not be impacted by anxiety, depression, or other psychiatric conditions. In addition, tests are needed that evaluate cognitive changes over time after drug or behavioral interventions.
How did the authors study this issue?
The authors reviewed cognitive problems in MS, the tests used to assess them, and some advantages and disadvantages.
Cognitive Impairment in Multiple Sclerosis: A Review of Neuropsychological Assessments
Nikolaos Korakas, MD, and Magda Tsolaki, MD, PhD
Cognitive & Behavioral Neurology
Of the more than two million people worldwide with multiple sclerosis, 40% to 65% experience cognitive impairment, many of them early in the course of the disease. Cognitive impairment has been found in patients with all subtypes of multiple sclerosis. Because both pharmacologic and nonpharmacologic interventions may improve patients’ brain function, cognitive assessment should be a routine part of the clinical evaluation. Traditional paper-and-pencil neuropsychological tests and batteries can help detect and monitor patients’ cognitive problems. Computerized cognitive batteries also show promise. Controversy continues over which test is most reliable at assessing cognitive impairment in both everyday clinical practice and research. Each battery has possible disadvantages, such as practice effects, poor sensitivity and specificity, and questionable applicability to multiple sclerosis. Based on our review of the literature, we describe the tests that are currently being used or that might be used in assessing cognitive deficits in patients with multiple sclerosis, and we summarize the strengths and limitations of each.
BICAMS=Brief International Assessment of Cognition for Multiple Sclerosis. BRB-N=Brief Repeatable Battery–Neuropsychology. EPT=Everyday Problems Test. FST=Faces Symbol Test. MMSE=Mini-Mental State Examination. MoCA=Montreal Cognitive Assessment. MS=multiple sclerosis. NPSBMS=Neuropsychological Screening Battery for Multiple Sclerosis. PASAT=Paced Auditory Serial Addition Test. PROMIS=Patient-Reported Outcomes Measurement Information System. RBANS=Repeatable Battery for the Assessment of Neuropsychological Status. SDMT=Symbol Digit Modalities Test.
Multiple sclerosis (MS) is an inflammatory neurodegenerative disease of the central nervous system (Hernández-Pedro et al, 2013) in which inflammatory cells attack and destroy the myelin, the protective insulation surrounding the nerves (Rahn et al, 2012). Typically, MS symptoms first appear between the ages of 20 and 40 years (Ortiz et al, 2013), although 2.7% to 5% of patients are diagnosed as children (MacAllister et al, 2013). Newly developed magnetic resonance imaging sequences and other studies have shown gray matter demyelination and microglia activation in the disease’s earliest stages (Klaver et al, 2013) and even at its very start (Schutzer et al, 2013). As a result, many people with MS experience physical, sensory, and emotion-regulation difficulties (Katz Sand and Lublin, 2013).
Diagnosis of MS follows the 2010 McDonald criteria (Polman et al, 2011), which focus on symptoms and neurologic findings. The diagnosis starts with a patient’s description of symptoms typical of an MS attack, eg, optic neuritis, intranuclear ophthalmoplegia, and partial transverse myelitis. These symptoms should be combined with objective findings from a neurologic examination and testing. Important to the diagnosis are a determination that multiple areas of the central nervous system are involved and evidence that disease activity is continuing (Katz Sand and Lublin, 2013).
Cognitive Impairment in MS
In addition to their physical symptoms, an estimated 40% to 65% of people with MS experience some degree of cognitive impairment (Klaver et al, 2013; Rahn et al, 2012). This impairment can have serious effects on daily life, interfering with such basic and instrumental activities of daily living as housekeeping, social life, and employment (Chiaravalloti and DeLuca, 2008). The most commonly affected cognitive domains are complex attention, information processing, executive function, processing speed, and long-term memory (Helekar et al, 2010). Patients first show deficits in verbal fluency and verbal memory, followed by a decline in visuospatial and recall skills, and, still later, a deterioration in attention and information processing speed (Achiron et al, 2005). Verbal learning can also be affected; this deficit has been associated with apolipoprotein E epsilon 4 (Koutsis et al, 2007). Helekar and colleagues (2010) speculated that people with MS overcome their primary structural deficits through a neuroplastic reorganization of their functional networks, or they use alternative strategies to cope with cognitive tasks.
Even though cognitive impairment is not a core symptom of MS and the McDonald criteria do not require it for diagnosis (Katz Sand and Lublin, 2013), patients may have cognitive impairment in the early stages of the disease, even before the first characteristic physical symptoms appear (Lovera and Kovner, 2012). In fact, a 2013 survey by Achiron et al estimated that cognitive impairment may precede other symptoms by 1.2 years.
Clinically isolated syndrome, the mildest form of MS, can be a precursor to a more severe form. Reuter et al (2011) studied patients with clinically isolated syndrome and found that 24% had cognitive dysfunction at the very onset of MS. In general, patients with clinically isolated syndrome have cognitive problems similar to those in patients with the more severe forms of MS, although with relatively intact verbal learning and memory capacity (Potagas et al, 2008).
Cognitive dysfunction affects patients with all MS subtypes. The highest frequency is in those with secondary progressive MS; the rate is lower in primary progressive MS, still lower in relapsing-remitting MS, and lowest in clinically isolated syndrome (Achiron et al, 2013). Amato et al (2001) reported that cognitive decline in MS is unlikely to subside; rather, it progresses slowly. Borghi et al (2013) wrote that the frequency of impaired cognition gradually increases as patients progress through their disease course, and that patients with the progressive forms have more severe cognitive impairment than do patients with the relapsing-remitting form. In a recent survey of 1500 patients with MS, Achiron et al (2013) found that those with the secondary progressive form showed greater cognitive decline than those with the other forms, in all domains except for the visuospatial. In a 6-year follow-up study of patients newly diagnosed with MS, Hankomäki et al (2014) observed quite stable overall cognitive function, but deterioration in attention and processing speed.
Borghi et al (2013) looked at each subtype of MS but were unable to discern subtype-specific patterns of cognitive impairment. This finding suggests a global pattern of cognitive dysfunction independent of disease course.
Cognitive impairment correlates strongly with brain lesions. Brain imaging studies of patients with MS have identified several potential sources of cognitive impairment. Tsolaki et al (1994)examined patients with MS using magnetic resonance imaging and nine neuropsychological scales, and found that cognitive dysfunction correlated with lesions in the hippocampus and with enlargement of the third ventricle. Cognitive impairment is also related to brain atrophy. Implicated in particular have been thalamic atrophy and hippocampal damage (Sicotte, 2011), as well as atrophy of the basal ganglia and cerebral cortex (Batista et al, 2012). Furthermore, elevated volume of the third ventricle is a strong biomarker of cognitive decline (Houtchens et al, 2007). Riccitelli et al (2011) reported that patients with the progressive form of MS lost gray matter in cortical regions, while patients with the relapsing-remitting form lost deep gray matter structures. Specifically for the relapsing-remitting form, however, Papadopoulou et al (2013) found that white matter lesions seemed to be an essential factor for cognitive problems.
No significant relationship has been found between overall cognitive impairment and physical disability in MS (Amato et al, 2008b), duration of the disease (Amato et al, 2010), or education or age (Solari et al, 2002). However, a recent survey reported a connection between cognitive problems and patients’ levels of education and physical disability (Ben Ari Shevil et al, 2014).
About half of patients with MS have depression (Marrie et al, 2009; Sadovnick et al, 1996), with 25.7% of all patients suffering from major depression (Patten et al, 2003). Anxiety has been reported in 23.5% to 41% of patients with MS (Wood et al, 2013). Anxiety has been shown to interfere with cognitive performance in patients with MS (Goretti et al, 2014). Researchers are divided, however, on whether depression does so: The COGIMUS study found no significant correlation between depression and cognitive dysfunction in patients with MS (Patti et al, 2009), but Borghi et al (2013) reported that impaired cognitive performance and depression might be linked in MS, and Sundgren et al (2013)found that even mild depressive symptoms affected cognition in MS.
Even if depression does not affect the objective neuropsychological performance of patients with MS, some evidence suggests that it influences their subjective perception of cognitive impairment (Kinsinger et al, 2010). According to Brassington and Marsh (1998), however, other researchers claim that depression does not affect neuropsychological efficiency and that objective cognitive function and depression vary according to an individual patient’s coping style. Although this issue is not resolved, Kinsinger et al (2010) concluded that specific domains, such as speed of information processing, may be related to depressive symptoms in MS.
Cognitive Impairment in Pediatric-Onset MS
Cognitive dysfunction is also a core symptom of pediatric-onset MS (Till et al, 2013), and may be found in children with clinically isolated syndrome (Julian et al, 2013). Within the first 2 years after onset of clinically isolated syndrome, 30% of children experience significant cognitive dysfunction (Bigi and Banwell, 2012).
Children diagnosed with MS at a very young age have an especially high risk of experiencing cognitive impairment within the first years of the disease (Bigi and Banwell, 2012). Cognition in children with MS worsens relatively fast (Jongen et al, 2012), and the impairment can be more dramatic than in adults with MS (Ozakbas et al, 2012). Because children’s central nervous system myelin is still developing, the pathologic process can interfere with the ongoing maturation of the white matter pathways, destroying the neural networks involved in cognition (Ozakbas et al, 2012). However, children have significant brain plasticity and repair mechanisms, which might help them recover from the neurodegeneration more efficiently than can adults (Ozakbas et al, 2012).
The cognitive domains most likely to be affected in children with MS are verbal and visuospatial memory, complex attention, and executive function (by contrast, most patients with adult-onset MS retain their verbal skills) (Bigi and Banwell, 2012). Processing speed seems to be one of the most sensitive and disturbed domains in both pediatric and adult-onset MS (Penner et al, 2013).
Not only may patients with childhood-onset MS show cognitive deficits during the first years of the disease, but they are also at risk for a lower intelligence quotient. Children’s cognitive impairment and low intelligence quotient scores can jeopardize their current and future academic performance as well as their ability to cope with life difficulties and psychosocial challenges (Amato et al, 2008a).
Clinicians should be aware that a child with MS may have cognitive impairment caused by the MS, but may also have depression and/or anxiety that can complicate the diagnosis of the cognitive impairment; alternatively, a child with MS may have depression and/or anxiety that can manifest as cognitive impairment (Weisbrot et al, 2014). Cognitive fatigue may lead children to perform poorly on cognitively effortful tasks, and they may need counseling to help with their everyday activities (Goretti et al, 2012). Nevertheless, Goretti et al (2012) found no connection between general or cognitive fatigue (evaluated with subjective scales) and overall cognitive impairment. Bigi and Banwell (2012) found no connection between cognitive dysfunction in children with MS and their disability status, number of relapses, or disease duration.
Management of Cognitive Impairment in MS
Amato and colleagues (2013) published a position paper that examined in great detail the state of treatment for cognitive impairment in MS. They reported that some studies showed benefit from symptomatic drug treatment, but could not be replicated. As for disease-modifying treatments, relevant research is slight, and the few published studies show minimal benefit (Amato et al, 2013).
Despite the limitations of the relevant studies, behavioral techniques have markedly improved learning and memory performance in people with MS-related cognitive impairment. Even though more data are needed, clinical trials to date have shown that, for some patients, behavioral interventions may slow the deterioration of cognition or even improve the already affected brain functions (Amato et al, 2013; Portaccio et al, 2013). Surprisingly, Amato et al (2013) found no studies evaluating behavioral treatment for processing speed deficits in patients with MS, despite the fact that processing speed is one of the major affected domains in MS. In the domain of long-term memory, Leavitt et al (2014) showed that aerobic exercise may be effective treatment.
Rationale for This Review
Even when people have been diagnosed with MS by the McDonald criteria, determining cognitive impairment is difficult. Health care practitioners need universal agreement on what constitutes cognitive impairment and tools to help them assess cognition in early MS (Achiron and Barak, 2006). Because most of the methods currently used to assess mental ability require both time and special training to give, cognitive impairment in MS is underdiagnosed (Patti, 2012). Borghi et al (2013) strongly support neuropsychological screening for patients with MS, especially those at high risk for cognitive impairment based on demographic, clinical, and radiographic evidence. Zarei and colleagues (2003), arguing for a cortical variant of MS, even proposed including MS in the differential diagnosis of dementia with cortical features and the differential diagnosis of depression when it is early onset, unusual, and persistent.
Our aim in this study was to examine the current state of knowledge about methods used to assess neurocognitive function in patients with MS. Based on a review of the literature, we describe the individual tests and batteries that measure cognitive deficits in MS and we describe the advantages and limitations of each. We group noncomputerized and computerized tests separately, and within each category we list established cognitive assessments, proposed cognitive assessments, and functional assessments.
Between June and December 2013, we performed an online search of the PubMed database using the key words cognitive impairment, multiple sclerosis, and neuropsychological assessment. We searched the key words first alone and then in combination. Next, we manually searched all relevant sources cited in the identified articles.
Later we added articles suggested by the anonymous reviewers of our study, and we kept updating the manuscript with new articles published until June 2015. We tried to include every article that tested or validated a neuropsychological test or test battery for patients with MS.
Noncomputerized Cognitive Assessments
Mini-Mental State Examination (MMSE)
The MMSE (Folstein et al, 1975) evaluates orientation to time and place, registration, attention and calculation, recall, language, and visuoconstruction. A resident or attending physician, neuropsychologist, or speech therapist (Benaim et al, 2015) can easily give the 11-question interview with pencil and paper in at most 10 minutes (Folstein et al, 1975). The test is easily scored, from 0 to 30 (Beatty and Goodkin, 1990). Cutoff scores for cognitive dysfunction differ.
While the MMSE is not very sensitive in detecting well-defined dementia in people with MS, the scale is useful in predicting focal cognitive impairment, especially in patients with relapsing-remitting MS who also have minor physical disabilities (Beatty and Goodkin, 1990).
Montreal Cognitive Assessments (MoCA)
The MoCA (Nasreddine et al, 2005) is also a screening tool for identifying cognitive impairment. The short edition is recommended by the US National Institute of Neurological Diseases and Stroke and by the Canadian Stroke Network. The MoCA has been validated for cognitive screening in patients with Alzheimer disease, human immunodeficiency virus, Huntington disease, parkinsonism, tremors, and epilepsy (Kaur et al, 2013). The test has also been validated for cognitive evaluation of patients with MS (Dagenais et al, 2013).
The MoCA is a 5-minute verbal test that consists of an immediate and a delayed five-word memory task, an orientation test, and a language fluency test of words starting with the letter F. Patients are awarded one point for each correct answer, with a maximum score of 30 (Kaur et al, 2013). To determine the value of the MoCA in MS, Dagenais et al (2013) have suggested that it be given to a large sample of patients with MS and the results compared to other sensitive neuropsychological batteries such as the Minimal Assessment of Cognitive Function in MS and the Brief Repeatable Battery–Neuropsychology (BRB-N).
Symbol Digit Modalities Test (SDMT)
The SDMT (Smith, 1982) measures attention and information processing speed, and can be administered in 5 minutes by nursing or technical personnel. The SDMT is part of many batteries, such as the Screening Examination for Cognitive Impairment, BRB-N, and Minimal Assessment of Cognitive Function in MS.
The top of the SDMT form shows nine typographic symbols, each matched with a digit in sequence from 1 through 9. Beneath this display is a row of 15 symbols, all taken from the original set, in a pseudo-randomized sequence. Participants are given 90 seconds to say which digit is paired with which symbol. Scherer et al (2007) noted that use of the SDMT is limited to cultures that use Arabic numerals.
The SDMT has been found to correlate with degree of cerebral atrophy in patients with MS (Christodoulou et al, 2003). Some researchers have suggested that if only one neuropsychological test could be given to screen for cognitive impairment, the SDMT could stand alone (Amato et al, 2006). Sonder et al (2014) claimed that in both everyday clinical practice and clinical trials, the SDMT can be used as a sole neuropsychological tool in assessing the cognition of patients with MS. The SDMT has even been found to be a significant predictor of cognition and automobile driving performance in people with MS (Schultheis et al, 2010). Furthermore, the test is relatively free of practice effects.
Digit Symbol Substitution Test
The Digit Symbol Substitution Test (Wechsler, 1981) is similar to the SDMT both in format and in evaluating information processing speed and attention, as well as concentration. As with the SDMT, the top of the Digit Symbol Substitution Test form shows symbols, each paired with a digit. In this test, however, beneath the display the form shows the digits pseudo-randomized, and participants have to draw (instead of say) the correctly paired symbol underneath each digit. Like the SDMT, the Digit Symbol Substitution Test is limited to cultures that use Arabic numerals (Scherer et al, 2007).
Faces Symbol Test (FST)
The FST (Scherer et al, 2007) evaluates working memory and sustained attention using the same testing principle as the SDMT, except that participants match symbols to faces rather than to Arabic numerals. The test is brief, easily administered as part of a routine neurologic examination, and well accepted by patients. Before participants take the FST, their hand function and visual acuity are evaluated to ensure that they are adequate for the test. During the actual FST, participants are given 90 seconds to draw each symbol as quickly as possible beneath the corresponding face. The sensitivity is 84% and the specificity is 85% (Scherer et al, 2007).
Because the FST uses faces instead of digits, patients’ education and culture have minimal effect on their performance. However, the use of faces instead of digits means that participants cannot reply orally but must draw the symbols (Scherer et al, 2007). Thus, patients who have hand or vision dysfunction cannot take the test and should be given the SDMT or Paced Auditory Serial Addition Test (PASAT) instead. Researchers with the Berlin Multi-Centre FST Validation Study have recommended using the FST as a screening tool for MS-related cognitive decline, and the patients who show deficits should undergo a more comprehensive neuropsychological assessment (Scherer et al, 2007).
Paced Auditory Serial Addition Test (PASAT)
Like the SDMT and the Digit Symbol Substitution Test, the PASAT (Gronwall, 1977; Gronwall and Sampson, 1974) evaluates information processing speed and attention. Also like the SDMT, the PASAT is included in many batteries, such as the Neuropsychological Screening Battery for Multiple Sclerosis (NPSBMS) and BRB-N.
The PASAT is a complex test involving language functions, visualization of numbers, and calculation (Yaldizli et al, 2014). The test asks participants to add together two randomized single digits: the digit they just heard and the one they had heard immediately before it. The digits are presented by a voice on a compact disk or audiocassette. In the PASAT-3, the voice presents a digit every 3 seconds; in the PASAT-2, the rate is one digit every 2 seconds. The maximum number of correct answers is 60.
Like the other tests that use Arabic numerals, the PASAT is not culture-free (Scherer et al, 2007). It is such a difficult test for patients to take that in Aupperle et al’s (2002) study, 11 (17%) of participants refused to take the PASAT and four (6%) tried but quit before completing it.
Scherer et al (2007) found the SDMT to be more sensitive than the PASAT in differentiating between patients with MS and healthy controls, and reported that the two tests had similar reliability and practice effects. Other studies, however, have shown substantial practice effects for the PASAT (Parmenter et al, 2007). Researchers do not agree on which of the two tests is better for cognitive screening, and whether either of them can be given alone (Amato et al, 2006). Brooks et al (2011)argued that the PASAT is unpleasant to take and its results in patients with MS can be compromised by sleep disorders, fatigue, depression, anxiety, and systemic diseases. However, Williams et al (2006) maintained that the SDMT is not an equal substitute for the PASAT. Sonder et al (2014) studied the long-term validity of the PASAT and SDMT, and found the SDMT to be superior.
Free Recall and Recognition Test
The Free Recall and Recognition Test (Wahlin et al, 1995) has two parts. In the recall section, the tester shows the patient a list of 12 concrete nouns while reading them aloud at a rate of 3 seconds per noun (Claesson et al, 2007). Then the tester withdraws the list, asks the patient to say the nouns in any order, and records the correct answers. In the recognition section, the tester presents the same nouns in the same format as in the recall section, but now the nouns are mixed randomly with 12 distractors. The tester asks the patient to answer “yes” or “no” as to whether each noun appeared during the earlier presentation (Claesson et al, 2007).
The Free Recall and Recognition Test has several advantages over other tests. It takes less than 5 minutes to give, has no floor or ceiling effects, and is accepted by most patients (Claesson et al, 2007). In contrast to the MMSE and PASAT, this test can discriminate among mild, moderate, and severe MS (Claesson et al, 2007). The test is easily administered by master's- and doctoral-level staff with knowledge of neuropsychology and proper training, and the results may indicate whether the patient needs a more comprehensive neuropsychological assessment (Claesson et al, 2007).
Multiple Sclerosis Neuropsychological Screening Questionnaire
This Questionnaire (Benedict et al, 2003) consists of 15 items assessing neuropsychological performance in daily activities. The Questionnaire has two forms, one for patients and the other for informants. It is self-administered and takes only 5 minutes.
To evaluate the validity of the Questionnaire, Benedict and colleagues (2003, 2004) gave it to patients and informants, and also gave the patients a comprehensive neuropsychological test battery. The patients’ Questionnaire responses correlated strongly with measures of depression, but not with neuropsychological function. The informants’ scores, by contrast, correlated with the patients’ cognitive function but not with depression. For this reason, the Questionnaire would be most valid if the patient and an informant completed it at the same time (Parmenter et al, 2007).
Brief International Assessment of Cognition for Multiple Sclerosis (BICAMS)
The BICAMS (Langdon et al, 2012) is a 15-minute battery comprising, in the following sequence, the SDMT, the California Verbal Learning Test 2nd ed (Delis et al, 2000) (the first five recall trials), and the Brief Visuospatial Memory Test Revised (Benedict, 1997) (the first three recall trials). Advantages of the BICAMS are that it can be administered by appropriately trained nonspecialist staff, it is given easily with just pencil and paper, and the only prerequisite is background information on the patient.
Clinicians should evaluate BICAMS results taking into account that performance is influenced by demographic factors, physical impairment (eg, dysarthria, impaired vision, pain), concurrent medical disorders, medications, fatigue, and, to some extent, depression (Langdon et al, 2012).
A limitation of the BICAMS is that it should not be given within 1 month after a patient recovers from a relapse or within 1 month after a patient takes a steroid medication, since these would affect the scores.
Neuropsychological Screening Battery for Multiple Sclerosis (NPSBMS)
The NPSBMS (Rao et al, 1991) consists of, in the following sequence, a verbal learning test, a spatial learning test, the PASAT, and a letter fluency task. An option is to add the SDMT, which slightly increases the battery’s sensitivity, even though it adds time. Advantages of the NPSBMS are a short administration time (mean=31.7 minutes) (Aupperle et al, 2002) and high sensitivity (71%) and specificity (94%) (Rao et al, 1991).
Two major drawbacks: Administration of the NPSBMS requires trained subdoctoral professionals, and, as noted earlier, a component of the battery, the PASAT, is a very difficult test for patients to take.
Screening Examination for Cognitive Impairment
The Screening Examination for Cognitive Impairment (Beatty et al, 1995) consists of a vocabulary test, a verbal reasoning test, a verbal memory test, and the oral version of the SDMT. The tests are given as three learning trials and a delayed recall trial of a list of 10 concrete nouns, along with the Shipley Institute of Living Scale Vocabulary Test (Zachary, 1986), the Shipley Institute of Living Scale Verbal Abstraction Test (Zachary, 1986), and the SDMT. The Screening Examination lasts about 30 minutes, with a mean administration time of 22.6 minutes (Aupperle et al, 2002), and can be performed in a single session that participants tolerate well (Solari et al, 2002). The high sensitivity and specificity are similar to those of the NPSBMS.
Drawbacks: The exam must be administered by practitioners with subdoctoral education and brief specific training. Because the practice effects are not known, the exam is recommended for only a one-time screening of patients (Aupperle et al, 2002).
Repeatable Battery for the Assessment of Neuropsychological Status (RBANS)
The RBANS (Randolph et al, 1998) is a short battery (average administration time=23.8 minutes) (Aupperle et al, 2002) that examines immediate and delayed memory, language, attention, and visuospatial function. Professionals use its 12 short tests to identify cognitive impairment in patients with Alzheimer disease, Huntington disease, subcortical vascular dementia, and Parkinson disease with dementia.
Patients with MS or Parkinson disease who have normal scores on the MMSE share similar patterns of cognitive impairment. When patients with Parkinson disease and normal MMSE scores took the RBANS, they did poorly, indicating that the RBANS was more sensitive than the MMSE in revealing the patients’ cognitive deficits (Beatty et al, 2003). For this reason, Aupperle et al (2002) gave the RBANS to a group with MS who had scored normally on the MMSE, but they found that the RBANS was no more sensitive than the MMSE for the patients with MS.
Another important limitation of the RBANS is that it does not estimate executive function. To address executive function, Aupperle et al (2002) added the Wisconsin Card Sorting Test (Berg, 1948). This combination of tests might be completed in an hour (Aupperle et al, 2002).
Brief Repeatable Battery–Neuropsychology (BRB-N)
The BRB-N (Rao and the Cognitive Function Study Group of the National Multiple Sclerosis Society, 1990) is a battery consisting of five tests, carefully selected to assess most cognitive functions with minimal overlap. The Selective Reminding Test (Buschke, 1973) evaluates verbal memory. The 10/36 Spatial Recall Test (Rao et al, 1990) assesses visual memory. The SDMT measures attention, executive function, and information processing speed. The PASAT-2 and PASAT-3, with numbers given every 2 or 3 seconds, assess sustained attention. The Word List Generation Test (Rao et al, 1990) evaluates verbal fluency.
A major strength of the BRB-N is that it is a brief (about 20 minutes) (Amato et al, 2006) and sensitive measure of cognitive impairment in the earliest stages of all MS subtypes, even clinically isolated syndrome (Potagas et al, 2008). The BRB-N’s sensitivity is 71% and its specificity is 94% (Rao et al, 1991).
The battery can be given by appropriately trained nonspecialist staff. Although the other elements of the BRB-N are well accepted by patients (Solari et al, 2002), as noted they find that the PASAT is quite difficult to take (Aupperle et al, 2002). To minimize practice effects, the BRB-N is available in two versions. Since frontal lobe executive function is underrepresented in the BRB-N, the Stroop Color and Word Test (Stroop, 1935) can be added.
The greatest weakness of this battery is that performance is strongly affected by the patient’s age and education. Furthermore, because depression might also affect performance, patients should not be tested when they show mood disorders (Sepulcre et al, 2006).
Minimal Assessment of Cognitive Function in MS
The Minimal Assessment of Cognitive Function in MS (Benedict et al, 2006) is one of the batteries most widely accepted by clinicians. It has seven tests (including the SDMT) that help assess five cognitive domains: language, spatial processing, learning new information and memory, processing speed and working memory, and executive function.
Despite the advantage of high reliability and validity in identifying cognitive impairment, the Assessment has major drawbacks (Becker et al, 2012). It takes about 2 hours to administer and requires highly trained professional testers. Furthermore, its strictly standardized testing environment may fail to reflect cognitive function in everyday life, in which tasks are complex and require more than one ability (Willis et al, 1992). The test results can be affected by psychoeducational interventions that patients have received to improve their everyday lives (Becker et al, 2012).
Proposed Noncomputerized Batteries Not Yet Validated
Battery Proposed by Aupperle et al (2002)
After evaluating the effectiveness of the NPSBMS, RBANS, and Screening Examination for Cognitive Impairment, Aupperle and colleagues (2002) proposed a new battery that would comprise the Consistent Long-Term Retrieval test (part of the Selective Reminding Test [Buschke, 1973]), the PASAT-3, and the SDMT. This pilot battery has the advantage of brevity (about 30 minutes, similar to the Screening Examination for Cognitive Impairment), but its drawbacks include the need to control for practice effects on the PASAT and the lack of studies to measure sensitivity and specificity (Aupperle et al, 2002).
Battery Proposed by the Pediatric Multiple Sclerosis Centers of Excellence (Julian et al, 2013)
The Pediatric Multiple Sclerosis Centers of Excellence (Julian et al, 2013) used the following comprehensive network battery for neuropsychological assessment of children with MS:
- Wechsler Abbreviated Scale of Intelligence (Wechsler, 1999) two subtests for Vocabulary and Matrix Reasoning: to assess general ability level
- Wechsler Individual Achievement Test, 2nd ed (Wechsler, 2001) Pseudoword Decoding, Expressive One-Word Picture Vocabulary Test, and Wechsler Abbreviated Scale of Intelligence (Wechsler, 1999) Vocabulary subtest: to evaluate reading and language
- Digit Span test from the Wechsler Adult Intelligence Scale, 4th ed (Wechsler, 2008) (for ages 16 years and older) or the Wechsler Intelligence Scale for Children, 4th ed (Wechsler, 2003) (for ages younger than 16), and the Digit Symbol-Coding test from the Wechsler Adult Intelligence Scale or Wechsler Intelligence Scale for Children: to measure attention, working memory, and processing speed
- Contingency Naming Test (Taylor et al, 1987) and Delis-Kaplan Executive Function System Trail Making Test (Delis et al, 2001): to evaluate executive function
- California Verbal Learning Test, 2nd ed (Delis et al, 2000) or Children’s Version (Delis et al, 1994): to assess verbal episodic learning and recall
- Beery-Buktenica Developmental Test of Visual-Motor Integration, 6th ed (Beery and Buktenica, 1997) and the Wechsler Abbreviated Scale of Intelligence (Wechsler, 1999) Matrix Reasoning test: to evaluate visuospatial function
- Grooved Pegboard Test (Kløve, 1963) and Delis-Kaplan Executive Function System (Delis et al, 2001) Trail Making Test Motor Speed Condition: to assess fine motor speed and coordination
The battery was given by a clinical neuropsychologist or a trained and supervised psychometrician. The total administration time was 2.5 hours, with breaks as needed (Julian et al, 2013).
Noncomputerized Cognitive Functional Assessment
Everyday Problems Test (EPT)
The EPT (Willis, 1993) and its revised edition (Becker et al, 2012) are useful tools for measuring cognitive performance in everyday situations. The original test consists of 42 written problem-related questions that assess performance in seven areas: meal preparation and nutrition, medications, phone use, shopping, financial management, transportation, and household management. The administration time is about 30 minutes. The revised edition, shortened to 30 questions, takes about 21 minutes, thus reducing patient fatigue and burden (Becker et al, 2012).
Both versions of the EPT have yielded reliability coefficients above 80%, and the revised version has correlated moderately with standard neuropsychological tests. Another advantage of the EPT is that testers require little formal training. However, the revised edition still needs to be investigated for its sensitivity to changes in patients’ cognitive function after they have received interventions. Although the EPT can measure everyday cognitive performance, patients may still need standard neuropsychological tests to determine if they have cognitive impairment (Becker et al, 2012).
Computerized Cognitive Assessments
Computerized Version of the SDMT
In 2011, Akbar et al reported their newly developed computerized version of the SDMT designed for patients with MS (Akbar et al, 2011). Their goal was to overcome methodologic weaknesses of validation in paper versions. In Akbar et al’s study of their computerized SDMT, they presented the task to patients with MS and healthy controls on a 19-inch monitor at a distance of 16 inches. As with the paper version, the participants looking at the computer screen were first shown the nine different typographic symbols, each paired with a digit from 1 to 9. In the next screen, the original display remained, and below it appeared the nine symbols in a new sequence. For each symbol, the participants had to say the number that matched the identical symbol above. The participants took a total of eight trials, each with a different sequence of symbols. Participants also completed the 90-second paper version of the SDMT, among other tests.
The computerized SDMT proved promising (Akbar et al, 2011). It had a slightly higher sensitivity and slightly lower specificity than the paper version. The computerized version seemed to be less influenced than the paper version by participants’ head motion and the speed of their eye movements. However, results in both versions could be affected by impaired visual acuity and speech capability (Akbar et al, 2011).
Computerized Test of Information Processing
The Computerized Test of Information Processing (Smith et al, 2012; Tombaugh and Rees, 2008) consists of three reaction-time subtests, each with progressively greater cognitive requirements. Tombaugh and Rees (2008) developed the three subtests. The first is the Simple Reaction Time task, in which participants are asked to react when the letter X appears on the screen. The second subtest is the Choice Reaction Time task, in which participants are asked to press the right key on a keypad when the word DUCK appears, and the left key when they see the word KITE. This subtest requires participants not only to react but also to make a decision. In the third and most difficult subtest, Semantic Search Reaction Time, participants are asked to decide whether or not a stimulus is a member of a presented semantic category. Here participants must not only react to the stimulus but also simultaneously decide and categorize the stimulus according to its meaning.
The test takes an estimated 10 to 15 minutes (Mazerolle et al, 2013; Smith et al, 2012). Tombaugh et al (2010)reported that the Computerized Test of Information Processing can detect defective information processing ability in people with MS as effectively as the PASAT does, and can even replace it.
Automated Neuropsychological Assessment Metrics
The Automated Neuropsychological Assessment Metrics (Reeves et al, 1992, 2007) battery has a particular strength in assessing processing speed, a crucial element in evaluating cognitive impairment in MS. This 30-minute computerized battery includes the Selective Reminding Test and the Procedural Reaction Time test to evaluate basic and complex reaction time, respectively. The battery assesses memory with the Running Memory Continuous Performance Test, Code Substitution Delayed Memory, Delayed Matching to Sample, Mathematical Processing, and Sternberg Memory Search tests. Information processing speed is measured with Code Substitution Learning and the Running Memory Continuous Performance Test. Finally, the Logical Relations test evaluates verbal reasoning.
Scores on the Automated Neuropsychological Assessment Metrics battery are highly concordant (95.8%) with paper-and-pencil measures of cognitive deficits in MS (Wilken et al, 2003).
Pellicano et al (2013) used the battery in comparing cognitive function in patients with MS and healthy controls. The authors chose to test all participants in the morning to try to avoid potential bias from variability in fatigue. Still, they found significantly poorer outcomes in the MS group.
Cognitive Drug Research Assessment System
The Cognitive Drug Research Assessment System (Simpson et al, 1991) is a computerized cognitive battery that assesses power of attention, continuity of attention, quality of working memory, quality of episodic memory, and speed of memory. The battery comprises the Selective Reminding Test, choice reaction time, digit vigilance, numeric working memory, spatial working memory, word recognition, and picture recognition tasks. Participants record “yes” and “no” answers with a simple response box. To minimize the motor demands on patients, the tester records their responses for word recall. Administration time is 15 to 20 minutes.
To study the usefulness and validity of the Cognitive Drug Research battery for patients with MS, Edgar et al (2011) compared the battery with the PASAT and the Digit Symbol Substitution Test in patients with relapsing-remitting MS, and found similar efficiency. The authors reported high test-retest reliability and low practice effects since the computer generates alternative forms of the tests; however, these alternative forms have not been specifically assessed for their equivalence with the original. The authors also cautioned that the Cognitive Drug Research System might not be sensitive in detecting memory deficits, and that the battery has been validated for use only in the relapsing-remitting form of MS.
The Neurotrax Mindstreams® computerized cognitive battery (http://www.neurotrax.com) (Dwolatzky et al, 2003) was designed to detect mild cognitive impairment. This tool assesses verbal and nonverbal memory as well as verbal fluency, executive function, visuospatial function, attention, information processing speed, and motor skills. The test battery is installed on a patient’s local computer system, and the results are uploaded to and stored in a secure web-based patient demographic database. The test battery takes 45 minutes. Patients can take part in mock sessions before the actual testing.
The strengths of this battery include quick data registration and the ability to alter the level of task difficulty according to each patient’s performance, thus minimizing ceiling effects. Dwolatzky et al (2003) gave the Mindstreams® Mild Impairment Battery to older people with cognitive impairment or mild Alzheimer disease and found that the battery successfully discriminated those with mild cognitive impairment from healthy elderly controls. Later, Achiron et al (2007)validated the Mindstreams® Computerized Cognitive Battery for patients with MS, and concluded that it can be used to evaluate cognitive dysfunction in MS.
Computer-Generated Battery Not Yet Evaluated
Cognitive Screening Battery Proposed by Lapshin et al (2013)
In 2013, Lapshin and colleagues reported their formulation and validation of a new computer-generated battery to screen for cognitive impairment in patients with MS. The battery contains five tests to evaluate information processing speed and working memory: the Stroop Color-Word Test (Stroop, 1935), the computerized version of the SDMT (Akbar et al, 2011), the Paced Visual Serial Addition Test (Nagels et al, 2005; Sampson, 1956) (a 2- and 4-second visual analogue of the PASAT), the Simple Reaction Time test (Tombaugh and Rees, 2008), and the Choice Reaction Time test (Tombaugh and Rees, 2008). The battery can be given with standard Windows software and does not require any special hardware. Testing time is estimated at 20 minutes.
A tester was needed to give the Paced Visual Serial Addition Test and to supervise the rest of the battery, even though the computer collected all the participants’ non-Paced Visual Serial Addition Test responses. In the Lapshin et al (2013) study, the tester was a master’s degree student, not a neuropsychologist.
The authors found that the computerized version of the SDMT and Paced Visual Serial Addition Test had good sensitivity (up to 85.7%) and specificity (up to 100%). The battery was limited by the need to exclude patients with significant vision impairment, and possibly by practice effects.
Computerized Functional Assessments
Patient-Reported Outcomes Measurement Information System (PROMIS)
The PROMIS, a program funded by the US National Institutes of Health (http://www.nihpromis.org), is a system of coordinated patient self-assessments of physical and mental health and social well-being (Cella et al, 2007). Patients and informants complete PROMIS questionnaires on paper or a computer screen. Their data can be sent to the Assessment Center website (http://www.assessmentcenter.net), where they are scored and kept in a database for analysis by both clinicians and researchers. Because the self-reports are standardized and validated, researchers can compare across domains and conditions.
Among the PROMIS measures are subscales for Cognitive Concerns and for Cognitive Abilities (Becker et al, 2014). Each consists of eight Likert-type items asking for a self-rating from 1=“not at all” to 5=“very much.” The sum of the scores for individual items becomes the total score.
Becker et al (2014) confirmed the validity and reliability of the two PROMIS subscales as a neuropsychological tool for patients with MS. The authors recommended future studies to evaluate whether these scales are sensitive in measuring meaningful cognitive change after interventions.
Virtual Shopping Test
The Virtual Shopping Test was developed and tested by Okahashi et al (2013) to assess cognitive dysfunction caused by brain injury, as from stroke or trauma. The creators designed the test to require only a personal computer and a touchscreen.
After getting instructions and a demonstration, participants were given 10 seconds to view and memorize a shopping list of four items. Then they had to organize the most effective route through a series of 20 different kinds of shops in order to buy the four items. Ten shops were placed on each side of a virtual street, and each shop contained six items. Each of the items on the shopping list came from a different shop. This meant that four shops had one correct item and five incorrect items, and 16 shops had no correct items. Okahashi et al selected the incorrect items to be similar to the correct item in set, use, phoneme, color, or shape. If participants forgot the items that they needed to buy, they could get a second try. They could also use hints, like referring back to the shopping list.
The Virtual Shopping Test was easy for participants to use. Including the demonstration, it took 30 minutes to complete.
The authors reported that the test could detect attention and memory problems in patients with brain damage, even in one session. A potential advantage of the test was that it could be used to evaluate cognitive function while patients were being rehabilitated. The researchers also emphasized the test’s possible ecological validity, ie, its ability to generalize to real-life situations. They did not discuss possible application of the test to patients with MS.
Computerized Versus Conventional Tests
Neuropsychological evaluation of patients with MS is the focus of substantial research. Along with the available radiologic and clinical data, numerous well-constructed and validated tests give clinicians a means to detect and diagnose cognitive deficits early in the course of the disease. New computerized and online evaluation tools are joining the mix, like Okahashi et al’s (2013) Virtual Shopping Test. The advantages and disadvantages of computerized neuropsychological testing versus standard paper-and-pencil tests are part of an ongoing debate.
The advantages of computers in neuropsychological evaluation include shorter test administration times, faster registering of data (in milliseconds), and automation reducing human error (Woo, 2008). Because computerized batteries are always administered the same way, they standardize assessments, benefiting both clinicians and researchers (Woo, 2008). Some computerized batteries prevent floor and ceiling effects by creating a range of possible scores that keep participants from scoring too high or too low (Woo, 2008). Finally, computerized batteries can create multiple and randomized versions of the same stimuli. In general, Schatz and Browndyke (2002) write that computerized tests are characterized by ease of use.
Computerized tests also have substantial disadvantages. For example, clinicians should be careful when they use Internet-based neuropsychological tools because normative data established for conventional versions of the tests do not necessarily apply, and norms for online versions have not yet been established (Buchanan, 2003). Test developers should give clinicians all specifics on hardware and software requirements (Schatz and Browndyke, 2002). Another potential problem is that older patients who are not familiar with computer technology may experience so much anxiety while undergoing computerized tests that their performance suffers (Leposavić et al, 2010). Computerized batteries generally cannot be used with patients who have low vision or poor motor skills. Importantly, because computerized tests require less interaction between the patient and the examiner, these tests almost exclude the clinical interview that is so important to analyzing the data correctly and planning proper rehabilitation (Leposavić et al, 2010).
Even though computerized batteries do not always require a trained neuropsychologist to interpret the results, Leposavić and colleagues (2010) maintained that a certified neuropsychologist is essential, not only for gathering the data, but also for evaluating the history of illness, the patient’s behavior, and the strategy that the patient uses in taking the test. These three factors are fundamental to diagnosing cognitive impairment (Leposavić et al, 2010). Hence, Leposavić et al concluded that computerized batteries cannot replace traditional paper-and-pencil tests, but they can serve as a primary evaluation tool for clinicians in other fields, who should refer affected patients to a neuropsychologist.
One limitation of our study is that we found no data on the internal and external validity of the neuropsychological tests before and after cognitive interventions. Very few of the tests have been validated for their ability to detect minor cognitive changes after intervention.
Another limitation is that most current batteries measure cognition in a highly standardized environment. They do not allow the evaluation of cognitive abilities in everyday life. We could not find data on the ecological validity of the tests we presented, and so could not discuss the tests’ real-world effectiveness.
A further limitation is that we could not find studies that directly addressed the costs of the neuropsychological tests.
An intrinsic limitation of using self-report assessments is that they are somewhat reliable but not sufficient for longitudinal studies (Gold et al, 2003). Gold and colleagues concluded, “The common pattern of poor correlation between self-rated and objective cognitive function thus appears to be a result of the patients’ (adaptive or maladaptive) coping mechanisms rather than being due to inaccurate measurement.”
Van der Hiele et al (2012) found that 29% of their participants with MS underestimated or overestimated their cognitive abilities. It is noteworthy that in the self-report tests, the participants evaluated their cognitive function within the context of a variety of everyday situations, while the neuropsychological tests evaluated cognition in a strictly standardized environment. Furthermore, patients may have cognitive deficits that they and their informants do not report.
Recommendations and Conclusions
Cognitive impairment is a major part of the spectrum of clinical manifestations of MS and can have a serious impact on patients’ quality of life. Thus, cognitive impairment should be identified as soon as possible, even before other functional symptoms appear, so that patients can benefit from early intervention and, possibly, better rehabilitation.
Helping clinicians confirm their suspicions of cognitive decline in their patients with MS are demographic and clinical information (Amato et al, 2008a; Ben Ari Shevil et al, 2014; Bigi and Banwell, 2012; Borghi et al, 2013; Hankomäki et al, 2014; Jongen et al, 2012; Julian et al, 2013; Kinsinger et al, 2010; Koutsis et al, 2007; Langdon et al, 2012; Ozakbas et al, 2012; Patten et al, 2003; Penner et al, 2013; Potagas et al, 2008; Reuter et al, 2011; Solari et al, 2002; Till et al, 2013) and imaging findings (Batista et al, 2012; Houtchens et al, 2007; Papadopoulou et al, 2013; Riccitelli et al, 2011; Sicotte, 2011; Tsolaki et al, 1994). Patients at high risk for cognitive dysfunction in the course of MS can be identified using cognitive percentile curves (Achiron et al, 2005). In view of the high prevalence of brain dysfunction in patients with MS, all diagnosed patients should undergo neuropsychological testing (Klaver et al, 2013; Rahn et al, 2012). For these reasons, we encourage clinicians to consider the advantages and disadvantages of the individual tests and batteries that we have described and to choose the most suitable battery for each patient and each clinical situation.
We recommend that clinicians suspect and evaluate their patients for anxiety and depression, both of which can affect cognitive test performance (Goretti et al, 2014; Patti et al, 2009, Sundgren et al, 2013). Even if clinicians do not detect depression, they should be aware that patients with MS are more likely than the general population to self-report feeling “useless” on the Chicago Multiscale Depression Inventory (Chang et al, 2003). For this reason, we endorse the use of a functional performance test that has ecological validity.
We also think that the cognitive percentile curves built by Achiron et al (2005) could be used in the clinical evaluation of patients’ cognitive performance to predict their risk of cognitive decline and to guide decisions on the possible value of neurorehabilitation.
The neuropsychological tests that we have reviewed can be used to evaluate children with MS. The usefulness of neuropsychological testing to guide pediatric treatment decisions is illustrated by a case report by Penner et al (2013). They affirmed the high sensitivity and clinical relevance of the testing that they used in their care of a 16-year-old boy with relapsing-remitting MS. An interesting finding, however, was that the patient’s SDMT score appeared normal. This finding led the authors to conclude that a better choice than the SDMT would have been a test that measures pure information processing speed without elements of working memory (Penner et al, 2013).
To overcome the current limitations of neuropsychological testing, future research should focus on constructing a battery that will identify cognitive impairment as early as possible and will not be confounded by depression, anxiety, or other psychiatric disorders. The need is great for a test that can differentiate the somatic symptoms of MS from anxiety and depressive vegetative disturbances, and help us identify patients in whom anxiety and depression may affect cognitive evaluation.
Research is needed to examine the validity of the new batteries, such as those described by Aupperle et al (2002) and Lapshin et al (2013). We encourage research, using the tests discussed in this review or even constructing new ones, to meet the great need for tests that can evaluate cognitive changes over time after pharmaceutical and cognitive-behavioral therapy. The ecological validity of new and existing tests is another area that needs further research; any new tests should be designed to evaluate patients’ cognitive performance in their everyday activities. Finally, a review of the costs of cognitive evaluation tools, in both time and money, would be of great benefit to clinicians who seek guidance in choosing the best test for their patient.
In summary, because 40% to 65% of people with MS experience cognitive dysfunction, which may affect their quality of life and may be treatable, cognitive evaluation should constitute a major part of the clinical examination in MS, especially when impairment seems likely.
The authors thank the anonymous reviewers for their insightful comments.
Achiron A, Barak Y. 2006. Cognitive changes in early MS: a call for a common framework. J Neurol Sci. 245:47–51.
Achiron A, Chapman J, Magalashvili D, et al.. 2013. Modeling of cognitive impairment by disease duration in multiple sclerosis: a cross-sectional study. PLoS One. 8:e71058.
Achiron A, Doniger GM, Harel Y, et al.. 2007. Prolonged response times characterize cognitive performance in multiple sclerosis. Eur J Neurol. 14:1102–1108.
Achiron A, Polliack M, Rao SM, et al.. 2005. Cognitive patterns and progression in multiple sclerosis: construction and validation of percentile curves. J Neurol Neurosurg Psychiatry. 76:744–749.
Akbar N, Honarmand K, Kou N, et al.. 2011. Validity of a computerized version of the symbol digit modalities test in multiple sclerosis. J Neurol. 258:373–379.
Amato MP, Goretti B, Ghezzi A, et al.. 2008a. Cognitive and psychosocial features of childhood and juvenile MS. Neurology. 70:1891–1897.
Amato MP, Langdon D, Montalban X, et al.. 2013. Treatment of cognitive impairment in multiple sclerosis: position paper. J Neurol. 260:1452–1468.
Amato MP, Ponziani G, Siracusa G, et al.. 2001. Cognitive dysfunction in early onset multiple sclerosis: a reappraisal after 10 years. Arch Neurol.