Feasibility and usefulness of cognitive monitoring using a new home-based cognitive test in mild cognitive impairment: a prospective single arm study

This study was prospectively performed at a university-affiliated dementia clinic from July 2019 to December 2021. Elderly subjects who visited the hospital due to persistent cognitive worsening and who were diagnosed with MCI were consecutively recruited during the study period. Clinical diagnosis of MCI was based on evaluations including detailed neuropsychological test battery named Seoul Neuropsychological Screening Battery (SNSB) [4], brain MRI, and routine blood chemistry including syphilis, thyroid function, vitamin B12/folate/thiamine levels, and apolipoprotein epsilon (APOE) genotyping. SNSB includes tests evaluating attention, language, praxis, visuospatial function, verbal and visual memory, and executive function as well as Korean version of instrumental activities of daily living (K-IADL) [4]. Plasma oligomeric amyloid beta level [5, 6] was examined at baseline. At baseline, MRI results within 1 year and blood test results within 3 months were accepted as the baseline findings. If there were no results within the period, retesting was performed. The study inclusion criteria were the following: (1) any cognitive complaint reported by participant or family members; (2) neuropsychological test shows impairment in at least 1 domain of cognitive function below − 1.0 standard deviation (SD) compared with age-, gender-, and education-adjusted norms; [4] and (3) ADL is normal based on MCI criteria according to Petersen and colleagues [7]. The physicians determined whether ADLs are normal based on K-IADL and Clinical Dementia Rating (CDR) scores [8]. K-IADL score below the cut-off score [9] of 0.43 and CDR score of 0.5 (sum of boxes scores below 3.5) were considered to be preserved ADLs. Exclusion criteria were (1) dementia status; (2) acute poor medical condition that may be associated with cognitive impairments such as acute kidney injury, hepatic encephalopathy, or systemic infection; and (3) participant or a family member refused to participate in the study.

For cross-sectional comparisons of HCT scores between cognitively normal, MCI, and mild Alzheimer’s disease (AD) dementia subjects [10], individuals with mild dementia CDR score 0.5 or 1.0⁷ who fulfilled the diagnostic criteria of National Institute on Aging-Alzheimer’s Association (NIA-AA) workgroups on diagnostic guidelines for AD [10] (n = 17) and participants with normal cognition based on SNSB [4] results ( > − 1.0 SD compared with age-, gender-, and education-adjusted norms in all domains; n = 30) were additionally enrolled. For cross-sectional assessment of test-retest reliability, 10 of the normal cognition subjects underwent HCT repeatedly with a 1-week interval. For cross-sectional comparison of inter-rater reliability, 2 independent raters examined the HCT in the same participant with normal cognition (n = 10) with a 1-week interval.

This study was approved by the institutional review board of Uijeongbu St. Mary’s Hospital (approval number: UC19OESI0069), and written informed consent was obtained from all participants. The study was conducted in accordance with the Declaration of Helsinki and principles of Good Clinical Practice. The personal information of the research participants was managed in an anonymized manner and stored in a location with restricted access except for the researcher in charge.

HCT was developed based on the Korean version of the Mini-Mental State Examination (K-MMSE) [11] to assess general cognitive functions over phone calls. The two tests are similar because they can be performed in approximately 5–10 min and scores range from 0 to 30, with a higher score indicating better cognitive function. However, several differences exist between HCT and K-MMSE; we minimized tests for orientation, increased tests for registration and recall, and added tests for frontal functions because the HCTs should be performed in non-demented participants using phone calls. The HCT was composed of the following: orientation to time (3 questions) and place (3 questions); verbal memory registration (5 words) and free recall (5 words); attentional tests that included serial 7 subtraction and backward digit span (4 numbers, e.g., 3-2-7-3); 2 frontal function tests using Controlled Oral Word Association Test (COWAT) phonemic (2 points) and 1 question for abstract thinking (2 points, e.g., what should you do if your room catches fire? or why do you wash your clothes?); and language function tests including naming (2 questions), comprehension (2 questions), repetition (1 question), and fluency (1 question).To prevent a potential learning effect, 7 sets were prepared for verbal memory registration and recall, language function tests, and tests for abstract thinking, and 1 among them was chosen for each test. Test-retest consistency and inter-rater reliability in the 10 cognitively normal controls were assessed.

Brain MRI was performed using a 3.0-Tesla scanner (GE Medical Systems, Milwaukee, WI, USA), including fluid attenuated inversion recovery (FLAIR), gradient-echo, and three-dimensional (3D) T1-weighted images. The white matter hyperintensities (WMHs) were rated using a visual rating scale of axial FLAIR images. In brief, periventricular WMHs and deep WMHs were evaluated separately and rated as minimal (grade 1), moderate (grade 2), or severe (grade 3) [12]. Lacunes were defined as small lesions (3–15 mm in diameter) hyperintense on T2- and hypointense on T1-weighted images with a perilesional halo on FLAIR [13]. Cerebral cortical microbleeds were defined as round and homogeneously low-signal lesions < 10 mm in diameter in lobar areas on gradient echo images [13]. Hippocampal atrophy was rated on coronal T1-weighted images using Scheltens’ visual rating scale [14]. The mean of the left and right hippocampal atrophy scores was used. The number of lacunes, number of microbleeds, degree of WMH, and degree of hippocampal atrophy were measured by a neurologist blinded to the data.

Processing and analysis of MRIs were performed using an artificial intelligence (AI)-based program named AQUA 2.0 (Neurophet, South Korea). The details of the MRI analysis including segmentation and normative data analysis were described in previous studies [15]. A normative percentile of the dataset was obtained using the East-Asian database described in a previous report [16] and calculated as the adjusted volume corrected to total intracranial volume as suggested in a previous study [17]. Brain MRI analyses showed regional standardized volumes (percentile) adjusted based on age, gender, and total intracranial volume [18].

All participants were diagnosed with MCI using the formal neuropsychological test battery SNSB [4], including the K-MMSE, CDR, K-IADL, tests for attention (digit span test), language (Boston Naming Test, tests for comprehension/repetition/fluency), visuospatial function (Rey Complex Figure Test, RCFT), verbal and visual memory function (Seoul Verbal Learning Test, SVLT and RCFT recall test), and frontal executive function (contrasting program, go-no-go, COWAT test, and Stroop test) [4]. Age-, sex-, and education-specific norms were used to interpret the SNSB results. The percentile scores, standardized scores adjusted by age, sex, and education, are based on a large nationwide Korean sample (1100 people), making it possible to perform comparisons with the population averages. Scores ≥ 16th percentile, which were compared to − 1 standard deviation (SD) of the norm, were defined as normal. Scores ≥ 16th percentile, which were comparable to − 1 SD of the norm, were defined as normal. Follow-up neuropsychological evaluations of clinical progression/cognitive decline included SNSB, K-MMSE, CDR, K-IADL, neurological and physical examinations, and history taking to assess clinical progression to dementia. A trained neuropsychologist administered the cognitive tests. Progression to dementia was evaluated based on follow-up neuropsychological tests and history taking in outpatient clinics at endpoint. Participants with CDR sum of boxes score ≥ 3.5 or K-IADL scores ≥ 0.43 were considered to have progressed to dementia.

Plasma oligomeric amyloid beta values were measured using the Multimer Detection System-oligomeric Aß (MDS-OAß) method [5] at baseline. In brief, the inBloodTM™ OAß test (People Bio Inc., Gyeonggi-do, Republic of Korea) was used to quantify MDS-OAß values in heparin vacutainer tubes; higher values indicate stronger amyloid oligomeric tendencies with vigorous amyloidosis [5, 6]. The plasma amyloid values using MDS-OAß method were higher in AD patients than in normal controls or non-AD dementia patients showing good discriminations [5, 6]. Plasma amyloid beta values correlated well with conventional AD biomarkers including [11]C-Pittsburgh compound B (PIB) positron emission tomography (PET), cerebrospinal fluid Aß42 and phosphorylated tau levels.

Test-retest consistency and inter-rater reliability were assessed using Cronbach’s alpha values. Associations between HCT scores and neuropsychological tests scores were investigated. In addition, whether the HCT scores correlated with AD biomarkers was assessed. To assess relevance of the HCT scores and neuropsychological tests scores/AD biomarkers, partial correlation coefficient adjusted for age, sex, and education was calculated. ANOVA was used to compare continuous variables (demographics, HCT test scores, and biomarker findings) among normal cognition, MCI, and mild dementia patients because basic demographics such as age, gender, and education levels were similar among the groups. Independent t-test or nonparametric Mann–Whitney U test (based on normal distribution patterns) was used for comparison of continuous variables between dementia progressors and non-progressors. Chi-square test was used to compare categorical variables. All statistical analyses were performed using SPSS (version 18.0; SPSS Inc., Chicago, IL, USA). P-values < 0.05 were considered to indicate statistically significant differences.