Mitsuru Kawamura, Division of Neurology, Department of Internal Medicine, School of Medicine, Showa University, 1-5-8 Hatanodai, Shinagawa-ku, Tokyo 142-8666, Japan. Email: firstname.lastname@example.org
Diagnosis of mild cognitive impairment (MCI) in Parkinson's disease (PD; PD-MCI) can be difficult. We examined whether the Japanese version of the Montreal Cognitive Assessment (MoCA-J) and the Neurobehavioral Cognitive Status Examination (COGNISTAT-J) were suitable to screen PD-MCI.
A total of 50 patients with PD and PD with dementia, took the Mini-Mental State Examination (MMSE), MoCA-J and COGNISTAT-J (except 3 patients) tests. Impairment in each cognitive domain on the tests was then compared between groups.
Of 25 patients with a high MMSE score of 27 or above, 13 had a MoCA-J score of below 25, and showed significantly lower scores than 12 patients with MoCA-J scores of 25 or above in the visuospatial, executive and memory domains of MoCA. Of the 25 patients with a high MMSE score, seven had four or more impaired subtests of the 10 subtests of COGNISTAT-J. The seven patients showed significantly lower scores in the subtests of construction, calculation and similarity compared with the 18 patients with less than four impaired subtests. In patients with a high MMSE score and less than four impaired subtests on the COGNISTAT-J, construction and memory were more impaired. Executive, visuospatial and memory abilities are characteristically impaired in PD-MCI, therefore both MoCA-J and COGNISTAT-J detected characteristics of PD-MCI in patients with a high MMSE score.
Both MoCA and COGNISTAT are sensitive tools for screening for PD-MCI.
Cognitive impairment in patients with Parkinson's disease (PD) is well known. PD with mild cognitive impairment (PD-MCI) is also known to be a predementia state, and patients with PD-MCI derive benefit from anti-dementia drugs. Executive, attentional, visuospatial and memory abilities,[1-3] especially executive abilities,[4, 5] are frequently impaired in the early phase of PD-MCI. However, the definition and detection of PD-MCI can be difficult. Diagnostic criteria for PD-MCI were published by the Movement Disorder Society (MDS) in 2012, and require at least 10 domain-specific neuropsychological tests for precise and comprehensive assessment of PD-MCI. However, such assessments are hard to carry out in daily outpatient clinics. In contrast, the criteria require recognition of impairment on a scale of global cognitive abilities validated for use in PD to make an abbreviated assessment. The abbreviated assessment can be carried out in an outpatient clinic, and brief and sensitive tools to screen MCI are much in demand. The Mini-Mental State Examination (MMSE) is one of the most frequently used tools to evaluate cognitive function. However, MMSE is not sensitive enough to detect PD-MCI. Use of the Montreal Cognitive Assessment (MoCA) to screen PD-MCI is recognized outside Japan. It was developed to identify patients with mild cognitive impairment, and evaluates more domains, including executive function, than the MMSE. MoCA has also gained currency outside Japan,[8-12] and is translated into a number of languages including Japanese. In contrast, the Neurobehavioral Cognitive Status Examination (COGNISTAT) is more sensitive than MMSE to detect cognitive impairment in patients with cerebrovascular disease and brain tumor, and geriatric inpatients with various medical diseases, but not as yet PD. COGNISTAT is also translated into Japanese. The object of the present study was to examine whether the Japanese versions of MoCA (MoCA-J) and COGNISTAT (COGNISTAT-J) are suitable for screening mild cognitive impairment in patients with PD.
A total of 50 PD or PD with dementia (PDD) patients were recruited from outpatients and inpatients diagnosed at Showa University Hospital or Showa University East Hospital in Tokyo, Japan. All patients except four were being treated with levodopa or a dopamine agonist. The other four patients were de novo, and no medication had been started before the present study. The experiments were carried out in the on state of medication. No patient had a history of impulse control disorder or had undergone deep brain stimulation. No patient had a disease that affected motor and cognitive functions except PD or PDD. While our present study was underway, three patients were taking antidepressants (mianserin hydrochloride, sertraline hydrochloride and escitalopram oxalate), but none had severe depression likely to affect the cognitive assessment.
The participants took three tests: MMSE, MoCA-J and COGNISTAT-J. They were carried out separately with sufficient rest time or on different days. However, three patients took only the MMSE and MoCA, because they died before the day of the COGNISTAT test.
MoCA was developed as a more sensitive tool than MMSE to screen mild cognitive impairment, and it contains tests assessing frontal-executive functions and complex visuospatial processing – both insensitive to the MMSE.
MoCA-J scores were compared with MMSE scores for all participants, and patients were classified into three groups according to MMSE and MoCA-J scores. Patient backgrounds and mean scores for each cognitive domain on MoCA-J were compared among the three groups. To categorize the cognitive assessment tests of MoCA, we used the “domains-based subscore” presented in the original MoCA proposal, and the recombinant of the “subscores-based subscore” shown in the current MoCA test paper. The domains-based subscores are based on six domains: visuospatial, executive, attention/concentration/working memory, language, memory and orientation.
Classification of the participants using MoCA-J
We set the MMSE cut-off score at 27 or above, because 26 had been used in previous studies,[8, 16] and we considered patients with a 1-point higher score would have better cognitive function. The cut-off score of MoCA-J was set at 25 for the present study, because it combines sensitivity with specificity for diagnosis of cognitive disorder. The three groups were thus defined as follows. Group A: MMSE score of 27 or above and MoCA-J score of 25 or above. Group B: MMSE score of 27 or above and MoCA-J score of below 25. Group C: MMSE score of below 27 with any MoCA-J score.
COGNISTAT is a cognitive screening test of separate subtests assessing major domains of cognitive functions. These are: orientation, attention, language (comprehension, repetition and naming), construction, memory, calculation, similarity and judgement. In COGNISTAT-J, the raw score of each subtest is converted to a standard score, in which average scores in normal controls are set at 10, and standard deviations (SD) in healthy controls are set at 1. A standard score of 8 thus signifies a record of mean minus 2SD. A standard score of 8 or below is defined as impairment. The results of 10 subtests are presented with their standard scores in the form of a graph, so intact or impaired subtests are visually clear. We calculated the number of impaired subtests in each patient, and compared the MMSE score and number of impaired COGNISTAT-J subtests for each patient. The 47 patients were then divided into three groups based on the results. Patient backgrounds, mean standard scores and impairment prevalence for each COGNISTAT subtest were compared among the three groups.
Classification of patients using COGNISTAT-J
The cut-off value of the number of impaired subtests was set at 4 based on the assumption that patients with a MMSE score of 27 (the MMSE cut off score) lost 1 point per cognitive domain for three domains, and patients with a MoCA score of 25 (the MoCA cut-off score) lost 1 point per cognitive domain for five domains. The three groups were thus defined as follows. Group D: MMSE score of 27 or above with the number of impaired subtests less than four. Group E: MMSE score of 27 or above with the number of impaired subtests four or more. Group F: MMSE below 27 with any results on COGNISTAT.
Statistical analysis of the three groups was carried out by one-way analysis of variance (Tukey's method for multiple comparison) to compare age, duration, Hoehn and Yahr stage, unified Parkinson's disease rating scale (UPDRS), levodopa equivalent dose, and cognitive assessment scores. The χ2-test and residual analysis were used to compare impairment prevalence in each COGNISTAT subtest in the three groups. Correlation between two indices was carried out by calculating Spearman's correlation coefficient.
Comparison of MMSE and MoCA-J scores for all 50 participants is shown in Fig. 1. Significant correlation was seen between MMSE and MoCA-J (r = 0.88, P = 0.000). A total of 25 of the 50 patients had high MMSE scores of 27 or above. Of these 25 patients, the MoCA-J score was below 25 in 13 (group B) and 25 or above in 12 (group A). All 25 patients with a MMSE score of below 27 (group C) had MoCA-J scores under 25. Backgrounds of the three groups (A, B and C) are shown in Table 1. A significant difference between groups A and C was seen on current age, but there were no other significant background differences.
Data were expressed by mean ± SD(standard deviation).
P < 0.05 vs Group A.
70.7 ± 10.5
65.3 ± 10.5
5.3 ± 4.1
2.7 ± 0.8
34.9 ± 16.0
13.0 ± 2.8
489.2 ± 356.6
71.3 ± 5.4
58.7 ± 20.3
12.6 ± 15.8
3.0 ± 0.8
45.4 ± 19.9
12.1 ± 2.0
440.4 ± 243.2
75.7 ± 7.7
66.5 ± 8.7
9.0 ± 6.6
3.2 ± 0.9
46.5 ± 21.6
12.8 ± 2.4
502.5 ± 361.3
MoCA-J mean score for each cognitive domain is shown in Table 2. Group C showed significantly lower scores than group A in all domains. Group B showed significantly lower scores than group A in the visuospatial, executive and memory domains. No significant difference was seen in attention/concentration/working memory, language and orientation domains between groups A and B.
Table 2. Mean scores of cognitive domains on Montreal Cognitive Assessment for the three groups (A, B, C)
Attention Concentration Working memory
Memory (Delayed recall)
Sum score of Executive and Attention Concentration Working memory
Copy cube Draw clock
Trail making Verbal fluency Abstraction
Digit span Vigilance Serial 7s
Naming of 3 animals Sentence repetition
Delayed recall of 5 words
Data expressed by mean ± standard deviation.
*P < 0.05 versus group A, **P < 0.01 versus group A.
COGNISTAT-J was carried out by 47 patients. Comparison of MMSE scores and number of impaired subtests on COGNISTAT-J for the 47 participants is shown in Fig. 1. There was a significant negative correlation between MMSE score and number of impaired subtests on COGNISTAT-J (r = −0.67 P = 0.000). The significant correlation was retained after excluding patients with a MMSE score of 12 or below, who can be considered as outliers (r = −0.57 P = 0.000). A total of 25 of the 47 patients had a high MMSE score of 27 or above. Of these, all except one had at least one impaired subtest. Although the MMSE scores were high, seven patients had more than four impaired subtests on COGNISTAT-J (group E), and 18 had less than four impaired subtests (group D). The same high MMSE score included varying degrees of impairment detected by COGNISTAT-J. A total of 22 patients had a MMSE score of below 27 (group F). All except three in group F had four or more impaired subtests. COGNISTAT-J showed a ceiling effect in patients with a MMSE score below 12. The COGNISTAT-J was effective in grading cognitive function in patients with impaired cognitive function that was not severe. No significant differences were seen in any backgrounds shown in Table 1 among the three groups (D, E and F). Table 3 shows the mean standard score of each subtest in the three groups. There were significant differences in the construction, calculation and similarity subtests between groups D and E. Group F showed a significantly lower standard score than group D in all subtests except judgement.
Table 3. Mean standard scores for each subtest on Neurobehavioral Cognitive Status Examination for the three groups (D, E, F)
Data were expressed by mean ± standard deviation.
*P < 0.05 versus group D, **P < 0.01 versus group D.
Figure 2 shows the prevalence of impairment in each subtest in COGNISTAT-J of the three groups. In group D, subtests of construction and memory were impaired in over 40% of the patients, and the prevalence was significantly higher than other subtests. Construction and memory domains were also impaired in the majority of patients in groups E and F. Subtests of comprehension, repetition and judgement were not impaired in any patients of group D.
There was significant negative correlation between MoCA-J score and the number of impaired subtests on COGNISTAT-J (r = −0.80, P = 0.000). All patients in group A were assigned to group D on the basis of a comparison between MMSE and the COGNISTAT-J. Of 13 patients in group B, six were assigned to group D and seven were assigned to group E. No patient in group A or B was assigned to group F. All patients in group C (apart from 3 who did not complete the COGNISTAT-J) were assigned to group F.
The MMSE is one of the most common tools to assess cognitive function. Patients with a high MMSE score can be judged normal with regard to cognition, and daily life seems unaffected. However, in the present study, more than half of the patients with a high MMSE score of 27 or above had a MoCA-J score of below 25 (group B). A MoCA score of 27 or 26[10, 12] is the recommended cut-off for screening for PD-MCI. However, all of the patients in group B were unaware of their cognitive impairment. Therefore, MoCA-J has the capacity to detect mild cognitive impairment in the early phase. Although patients in both groups A and B had high MMSE scores, patients in group B had significantly lower scores than group A on visuospatial, executive and memory tests when applying the MoCA-J cut-off score of 25. Executive, visuospatial and memory abilities are the most frequently impaired domains in PD-MCI; however, their reported rankings differ.[1-5] The orientation domain characteristically impaired in Alzheimer's disease, in contrast to PD, was hardly impaired in groups A and B. In patients with a high MMSE score, the derived points of MoCA-J at a cut-off value of 25 might account for the impaired domains in PD-MCI. However, no significant difference was seen in the attention/concentration/working memory domains between groups A and B in the present study. Other studies show impairment in the attention domain together with the executive domain as an attention/executive domain.[2, 3, 5] Cognitive tests on the attention/executive domain in other studies include trail making, verbal fluency, abstraction, digit span, serial 7s and so on, as with MoCA-J. In the present study, the MoCA-J sum of scores for attention/concentration/working memory and executive domains showed significant differences between groups A and B (P < 0.05; Table 2). The ability to carry out these cognitive tests does not require only one cognitive ability, but cooperation between domains, such as the executive, attention and visuospatial ability, and so on. These abilities can be evaluated as summation values. Participants in the present study were also considered cognitively impaired in the attention and executive domains. Scores of attention/concentration/working memory domain in the present study might have been insufficiently high to show a significant difference.
According to the MDS consensus criteria for comprehensive diagnosis of PD-MCI, at least 10 domain-specific cognitive assessment testings are required. Therefore, precise diagnosis with detection of impaired domains is impossible using only the MoCA-J. Combination with another cognitive assessment tool, such as COGNISTAT-J, which includes 10 separate subtests assessing major cognitive domains, is recommended for more precise clinical assessment. However, the MoCA-J that includes presentative tests in each cognitive domain clearly detects the frequently impaired domains in PD-MCI. Inclusion of more domains in MoCA relating to frontal lobe performance, such as executive function, might make it more sensitive. As discussed here, MoCA-J is more sensitive than MMSE, and sharply detects characteristics of PD-MCI. Therefore, MoCA is as useful for screening for PD-MCI in Japan as it is elsewhere.[8-12]
In the present study, comparing the MMSE and the COGNISTAT-J, group E patients had significantly lower scores in the construction subtest compared with group D patients. Furthermore, even in group D patients whose cognitive impairment was considered almost normal or very mildly impaired, the construction subtest showed a significantly higher prevalence of impairment among the 10 subtests. To solve the COGNISTAT-J construction subtest, the patient is asked to draw two designs from memory after a 10-s study period. If this fails, the patient is given more tests, which require construction of three mosaic patterns similar to those in the Kohs blocks test and the block design subtests of the Wechsler Adult Intelligence Scale-Revised (WAIS-R). This subtest requires visuospatial, constructive, executive and problem-solving abilities. These abilities are frequently impaired in PD-MCI,[1-5] Therefore previous results in the construction subtest do not contradict previous knowledge. The mean score in the subtest of similarities also shows significant differences between groups D and E. To assess the subtest of similarity, the patient is required to identify abstraction of four word pairs. This ability is a frontal lobe function. In group D, the memory subtest also showed frequent impairment. Frontal lobe function and memory are frequently impaired in PD-MCI.[1-5] The COGNISTAT-J reflects the early impairment of these cognitive domains, and was more sensitive than MMSE. COGNISTAT-J includes subtests that require the aforementioned abilities and includes subtests those are more difficult than those in MMSE. According to the previous study using COGNISTAT-J, patients in the MCI stage of Alzheimer's disease (AD-MCI) performed significantly worse than the normal control group on subtests of orientation, naming, construction, memory and similarities. In contrast, patients in groups D and E in the present study showed little impairment on the subtests of orientation and naming. The subtests of orientation and naming can distinguish AD-MCI and PD-MCI. Based on these findings, COGNISTAT-J is a useful means of screening for PD-MCI.
The MoCA-J score and number of impaired subtests in COGNISTAT-J were significantly correlated, and classifications of patients with both cognitive assessment tools were similar when using the cut-off value of the present study. Both MoCA-J and COGNISTAT-J are thought to have almost equivalent sensitivity, although the MoCA-J might be more sensitive. Cognitive deterioration in patients belonging to both groups B and D, who had fewer impaired subtests on the COGNISTAT-J, but whose MoCA-J score were below 25, might have been overlooked by the COGNISTAT.
We did not examine cognitive function in healthy people and patients with other diseases, so our previous results are available only for patients who were already diagnosed with PD or PDD. We emphasize that both the MoCA and the COGNISTAT are suitable for screening MCI in Japanese patients with Parkinson's disease.
This study was supported by the Tamagawa University Center of Excellence under the Ministry of Education, Culture, Sports, Science, and Technology (MEXT); a Grant-in-aid for Scientific Research on Innovative Areas, ”Face Perception and Recognition”(MEXT, 23119720), and a Grant-in-Aid for Scientific Research (MEXT, 23591283). This study was also supported in part by the Research Funding for Longevity Sciences(25-13) from National Centre for Geriatrics and Gerontology (NCGG), Japan.