Real-World Multimodal Day-Care Intervention for Mild Cognitive Impairment with Lewy Bodies: A Prospective 3-Year Comparative Cognitive Trajectory Study

Abstract

Introduction

Mild cognitive impairment with Lewy bodies (MCI-LB) is generally associated with more rapid cognitive decline than mild cognitive impairment due to Alzheimer’s disease (MCI-AD). However, evidence regarding the potential cognitive trajectories of individuals with MCI-LB participating in structured non-pharmacological multimodal programs remains limited. The aim of the study was to preliminarily examine cognitive changes over time among individuals with MCI-LB and MCI-AD undergoing a multimodal intervention.

Methods

Conducted at the University of Tsukuba Hospital between April 2013 and February 2020, this prospective study enrolled 74 participants (MCI-LB: 14; MCI-AD: 60) in the Cognitive Improvement Day-Care (CIDC) program. The CIDC was a multimodal intervention offering structured sessions including physical exercise, cognitive training, music therapy, and art-based activities. Participants attended the program, mostly once a week, and underwent annual cognitive assessments for up to 3 years using the Japanese version of Mini-Mental State Examination (MMSE-J). Linear mixed-effects models were used to analyze longitudinal changes in MMSE-J scores.

Results

The overall annual cognitive decline was −0.36 points/year (95% CI: −0.63, 0.10). The annual decline was −0.44 points/year (95% CI: −0.95, 0.06) for the MCI-LB group and −0.34 points/year (95% CI: −0.64, −0.03) for the MCI-AD group. No significant group-by-time interaction was observed over the 3-year follow-up (p = 0.97).

Conclusions

These findings suggest that individuals with MCI-LB exhibited longitudinal cognitive trajectories under a structured multimodal intervention that were comparable to those observed in individuals with MCI-AD, at least as assessed by the MMSE-J. Future studies with larger samples and more detailed cognitive assessments are needed to clarify potential subtype-specific responses.

Plain Language Summary

Mild cognitive impairment with Lewy bodies (MCI-LB) is an early stage of Lewy body disease, a brain disorder that can cause problems with thinking, attention, movement, and alertness. People with MCI-LB often experience faster decline in memory and thinking abilities than those with mild cognitive impairment due to Alzheimer’s disease (MCI-AD). Although medicines for dementia exist, there are few nondrug programs specifically tested for people with MCI-LB. This study followed 74 older adults who joined a hospital-based day-care program called the Cognitive Improvement Day-Care (CIDC) at the University of Tsukuba Hospital. Among them, 14 had MCI-LB and 60 had MCI-AD. The CIDC is a structured, nondrug program that combines physical exercise, cognitive (brain) training, music therapy, and art activities in two-hour sessions, usually held once a week. Participants were observed for up to 3 years. Their thinking abilities were measured each year using the Mini-Mental State Examination (MMSE), a common test of global cognitive function. Overall, thinking abilities declined only slightly over time, and there was no meaningful difference between people with MCI-LB and those with MCI-AD. These findings suggest that people with MCI-LB may exhibit a similar pattern of cognitive change to people with MCI-AD when participating in structured, nondrug multimodal programs such as CIDC. Further research with larger groups of participants is needed to confirm these results and to better understand how such programs may support brain health in different types of mild cognitive impairment.

Introduction

Mild cognitive impairment with Lewy bodies (MCI-LB) represents a prodromal stage of dementia with Lewy bodies (DLB) [1, 2], which is the second most common form of neurodegenerative dementia after Alzheimer’s disease (AD) [3, 4]. MCI-LB is characterized by subtle cognitive decline accompanied by core clinical features of Lewy body pathology, including fluctuating attention, rapid eye movement sleep behavior disorder, and recurrent visual hallucinations [1]. As individuals with MCI-LB are at increased risk of accelerated cognitive deterioration and functional decline, early identification and timely intervention during this stage are considered critically important.

Previous studies have demonstrated that individuals with MCI-LB tend to exhibit faster cognitive decline (especially, attention, visuospatial function, and executive processes), greater neuropsychiatric burden, and more severe functional impairments than those with MCI due to Alzheimer’s disease (MCI-AD) [4–6]. Patients with DLB exhibit heightened drug sensitivity and are more prone to adverse effects compared to those with AD [7], underscoring the particular importance of non-pharmacological interventions in this population. In response to the need for early, non-pharmacological approaches, multimodal intervention programs – incorporating physical activity, cognitive training, and psychosocial support – have gained empirical support in delaying cognitive deterioration. Landmark studies such as FINGER, MAPT, and AgeWell.de have shown promising results; however, these trials have largely focused on heterogeneous MCI populations without distinguishing between etiological subtypes [8–10].

Despite growing evidence on the effectiveness of multimodal interventions in MCI, most previous studies have not distinguished underlying etiological subtypes. As a result, it remains unclear whether individuals with MCI-LB exhibit cognitive trajectories under multimodal interventions that are similar to, or different from, those observed in individuals with MCI-AD. Given the clinical and neurobiological differences between these subtypes, a direct comparative examination of intervention-related cognitive outcomes across MCI etiologies is needed.

The present preliminary study aimed to address this knowledge gap by exploring longitudinal cognitive changes in individuals with MCI-LB and MCI-AD who participated in a structured, non-pharmacological multimodal program. By examining both groups under identical intervention conditions, we sought to determine whether similar patterns of cognitive change were observed or whether differential trajectories emerged between the two groups.

Methods

Study Design

This longitudinal study was implemented at the University of Tsukuba Hospital, Tsukuba City, Ibaraki, Japan, between April 2013 and February 2020. Each participant was followed for up to 3 years from their enrollment date. Not all participants completed the full 3-year follow-up; however, participants with at least one follow-up assessment were included in the longitudinal analyses. Conversion to dementia and reversion to normal cognition during follow-up were recorded based on clinical evaluation. Because this study was conducted as a prospective study within a routine clinical day-care program, it was not registered in a clinical trials registry.

Participants

Between April 2013 and February 2020, participants were recruited from the Cognitive Improvement Day-Care (CIDC) program through the Department of Psychiatry at the University of Tsukuba Hospital. Upon referral, all underwent a clinical evaluation by a board-certified psychiatrist and neuropsychological testing by a licensed psychologist. Program details were explained to participants and families before enrollment. Eligibility for study participation required meeting all of the following conditions: (1) outpatient diagnosis of MCI-LB or MCI-AD based on McKeith et al. [1] and the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) core criteria [11]; (2) formal diagnosis by a trained psychiatrist at enrollment; and (3) CIDC enrollment during the study period, allowing up to 3 years of follow-up. Exclusion criteria included (1) terminal illness; (2) significant musculoskeletal disorders (e.g., advanced joint disease, recent fractures, or chronic pain syndromes); (3) serious cardiovascular diseases (e.g., congestive heart failure, poorly controlled hypertension, or recent myocardial infarction); (4) neurological conditions such as stroke, epilepsy, or advanced-stage Parkinson’s disease; and (5) absence of medical clearance to participate.

The CIDC Program

The CIDC program, a multimodal intervention developed by our research group and first described by Boku et al. [12], has been offered at the University of Tsukuba Hospital since 2013. Subsequent studies using the same program have demonstrated beneficial effects on cerebral blood flow, preservation of gray matter volume, and long-term cognitive trajectories [13, 14]. The CIDC aims to improve cognitive function in older adults with MCI through structured, non-pharmacological multimodal interventions. The program takes place in a dedicated hospital gymnasium and operates 3 days per week, offering both morning and afternoon sessions, each lasting 2 h. While participants are encouraged to attend a full day (i.e., both sessions) once a week, some attend more frequently depending on personal availability, interest, and caregiver involvement.

The intervention comprised six core domains: (1) physical exercise (40% of total time; aerobic exercise, yoga, resistance training), (2) cognitive training (20%; game-based memory, attention, problem-solving tasks), (3) music therapy (20%; singing, instrument playing), (4) art-based activities (12%; painting, drawing, art appreciation), (5) aromatherapy (4%; herbal olfactory stimulation), and (6) recreational activities (4%; seasonal events, cooking, outdoor activities, breathing exercises). Each session focuses on one domain to ensure balanced weekly coverage.

Baseline Characteristics

Demographic and clinical data, including age, sex, and educational attainment, were collected from standardized clinical assessment records at the enrollment. Education was recorded as a total year of formal schooling, based on self-report and confirmed by caregivers when needed. Anti-dementia drug use, such as donepezil, rivastigmine, galantamine, and memantine, was extracted from baseline clinical records and subsequently verified through prescription lists maintained by the treating psychiatrist. Comorbidity data, including hypertension, diabetes mellitus, dyslipidemia, and cardiovascular disease, were obtained from clinical records. The total number of prescribed medications was determined by counting all regularly administered systemic medications at enrollment, excluding topical treatments, dietary supplements, and as-needed medications. Prescription data were cross-checked for accuracy.

Adherence Assessment

Adherence to the intervention was assessed using attendance rate, calculated as the proportion of the number of sessions attended to the number of sessions scheduled for each participant over the study period. Attendance data were extracted from institutional logs maintained by program staff. While participants were generally encouraged to attend once weekly (comprising two sessions per day), a subset chose to attend more frequently based on personal preference, availability, or caregiver support, resulting in attendance rates that could exceed 100%.

Cognitive Function Assessments

Global cognition was assessed annually using the Japanese version of the Mini-Mental State Examination (MMSE-J), administered by licensed clinical psychologists [15, 16]. The MMSE-J has been shown to have good validity and reliability, including equivalence to the original MMSE procedures [16]. The widely used 30-point screening tool evaluates temporal and spatial orientation, immediate memory encoding, attention and calculation, delayed memory recall, language functions, and visuospatial ability. Although less sensitive to subtle or domain-specific deficits, it was selected for its clinical utility, feasibility for longitudinal follow-up, and comparability with previous studies.

Statistical Analysis

We included participants who completed at least two cognitive evaluations (one at baseline and one follow-up) to ensure sufficient longitudinal data. Participants were divided into two diagnostic groups: MCI-LB and MCI-AD.

Descriptive statistics were computed as means and standard deviations for continuous variables, and as frequencies and proportions for categorical variables. Baseline characteristics were compared between the MCI-LB and MCI-AD groups using independent sample t tests for continuous variables with approximately normal distributions, chi-square tests for categorical variables, and Mann-Whitney U tests for variables with skewed distributions, as appropriate.

Longitudinal changes in cognitive function were analyzed using a propensity score (PS)-adjusted linear mixed-effects model as the primary analysis. A single composite PS was calculated based on age, sex, education, baseline MMSE-J score, and anti-dementia medication use, and was included as a covariate to control for potential confounding. The model specified time (categorical), diagnostic group, and the group × time interaction as fixed effects, and incorporated random intercepts to account for intra-individual variability while accommodating missing data under the missing-at-random assumption. In a supplementary analysis to estimate annual rates of change, we re-specified time as a continuous covariate in a PS-adjusted mixed model; the coefficient for time (and the time × group term) was interpreted as the yearly change in MMSE-J (points/year). Participants who converted to dementia or reverted to normal cognition during follow-up were retained in the longitudinal analyses up to the time of their last available assessment.

Model fit was evaluated using the Akaike information criterion and Bayesian information criterion, with lower values indicating better fit (online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000551066). All analyses were conducted using IBM SPSS Statistics (version 29.0.2.0). A two-tailed p value of <0.05 was considered significant. Finally, a post hoc power analysis was performed using GPower (version 3.1.9.7) [17, 18] to assess the power for detecting the group × time interaction.

Results

A total of 198 individuals were enrolled in the CIDC program between April 2013 and February 2020. After excluding those with missing baseline and follow-up data (n = 64), non-eligible diagnoses (n = 56), and missing educational history (n = 4), 74 participants met the inclusion criteria and were included in the final analysis (Fig. 1). Among them, 14 were diagnosed with MCI-LB and 60 with MCI-AD.

Fig. 1.

Flowchart of the study patients. Among the 198 individuals enrolled in the CIDC program, 74 fulfilled the inclusion criteria and had adequate data to be included in the final analysis. CIDC, Cognitive Improvement Day-Care; MCI-LB, mild cognitive impairment with Lewy body disease; MCI-AD, mild cognitive impairment due to Alzheimer’s disease.

Baseline characteristics for participants included in the analysis (n = 74) were mean age 72.5 ± 6.0 years, 55.4% male, and mean baseline MMSE-J score 27.1 ± 1.9. Among the excluded participants (n = 124), the mean age was 71.0 ± 7.5 years and 46.8% were male. Baseline MMSE-J scores were available for 60 excluded participants, with a mean of 23.7 ± 4.0.

Given this sample size (n = 74), the study had an estimated 99.9% statistical power to detect moderate group × time interaction effects (f = 0.25) but only 50.4% power for small effects (f = 0.10), suggesting limited ability to identify subtle between-group differences. Baseline characteristics of the two groups are presented in Table 1. The mean age was 71.4 ± 5.4 years in the MCI-LB group and 72.8 ± 6.1 years in the MCI-AD group. Years of education averaged 14.4 ± 2.2 in the MCI-LB group and 13.5 ± 2.7 in the MCI-AD group; the proportion of participants with a college degree or higher was 57.1% and 36.7%, respectively. Baseline MMSE-J scores were 27.7 ± 1.6 in the MCI-LB group and 27.0 ± 2.0 in the MCI-AD group. The median attendance rate was 80.9% (range: 34.9–103.0) in the MCI-LB group and 82.2% (range: 7.7–140.5) in the MCI-AD group. There were no significant differences in all items.

Table 1.

Baseline characteristics of the patients

​	MCI-LB (N = 14)	MCI-AD (N = 60)	p value
Demographics and anthropometrics
Sex (men, women)a	9, 5	32, 28	0.46
Age, years	71.4 (5.4)	72.8 (6.1)	0.57
Education, year	14.4 (2.2)	13.5 (2.7)	0.42
Participation period, year (over 3 years)	3.3 (0.8)	3.1 (1.0)	0.13
MMSE-J score	27.7 (1.6)	27.0 (2.0)	0.19
Attendance rate, % (over 3 years)b	80.9 (34.9–103.0)	82.2 (7.7–140.5)	0.89
Medical history
Diagnosed comorbidities, n	1.8 (1.9)	1.6 (1.5)	0.10
Medications (including anti-dementia drugs), n	2.6 (2.2)	2.4 (2.3)	0.77
Anti-dementia medications use, %a	71.4	53.3	0.22

Data are shown as mean (standard deviation) for continuous variables, except for attendance rate, which is shown as median [range].

MCI-LB, mild cognitive impairment with Lewy body disease; MCI-AD, mild cognitive impairment due to Alzheimer’s disease.

^aChi-square test.

^bMann-Whitney U test.

Figure 2 shows the trajectories of MMSE-J scores over the 3-year intervention period for each diagnostic group. No significant group × time interaction was observed in either the crude model (p = 0.97) or the PS-adjusted model (p = 0.97).

Fig. 2.

Three-year MMSE-J score changes in MCI-LB and MCI-AD during a multimodal intervention. Participants were categorized into two groups at each diagnosis: MCI-LB (blue line) and MCI-AD (gray line). Error bars represent standard errors. MCI-LB, mild cognitive impairment with Lewy body disease; MCI-AD, mild cognitive impairment due to Alzheimer’s disease; MMSE-J, the Japanese version of the Mini-Mental State Examination; PS, propensity score (adjusted for age, sex, education, baseline MMSE-J score, and anti-dementia medication use).

In overall samples, the estimated annual rate of cognitive decline was −0.36 points/year (95% confidence interval [CI]: −0.63 to 0.10). When stratified by diagnosis, the annual decline was −0.44 points/year (95% CI: −0.95 to 0.06) in the MCI-LB group and −0.34 points/year (95% CI: −0.64 to −0.03) in the MCI-AD group (online suppl. Table S1).

Table 2 presents the estimated effects from the mixed-effects models comparing MMSE-J score trajectories between MCI-AD and MCI-LB (reference group). In both the crude and PS-adjusted models, there were no statistically significant main effects of diagnostic group, time, or group × time interactions.

Table 2.

Mixed-effects model results for MMSE-J score trajectories comparing MCI-AD to MCI-LB (reference group)

Time point/interaction	Crude β (95% CI)	PS-adjusted β (95% CI)
Year 1	−0.32 (−1.87, 1.24)	−0.31 (−1.86, 1.25)
Year 2	−0.96 (−2.57, 0.64)	−0.98 (−2.58, 0.62)
Year 3	−1.25 (−3.15, 0.65)	−1.24 (−3.31, 0.66)
Group (>threshold)	−0.71 (−2.35, 0.93)	−0.14 (−1.76, 1.48)
Group × year 1	−0.14 (−1.87, 1.58)	−0.16 (−1.88, 1.57)
Group × year 2	0.30 (−1.51, 2.11)	0.32 (−1.49, 2.12)
Group × year 3	0.23 (−1.92, 2.39)	0.19 (−1.96, 2.34)

MCI-LB, mild cognitive impairment due to Lewy body disease; MCI-AD, mild cognitive impairment due to Alzheimer’s disease; MMSE-J, the Japanese version of the Mini-Mental State Examination; 95% CI, 95% confidence interval; PS, propensity score (adjusted for age, sex, education, baseline MMSE-J score, and anti-dementia medication use).

Model fit indices are summarized in online supplementary Table S2. The PS-adjusted model showed better overall fit than the crude model, with lower Akaike information criterion (1,008.63 vs. 1,021.58) and Bayesian information criterion (1,015.34 vs. 1,028.30).

During the follow-up period, 26.7% of participants with MCI-AD and 42.9% of those with MCI-LB completed the full 3-year follow-up, while 68.3% and 92.9%, respectively, completed at least 2 years of follow-up. Conversion to dementia occurred in 23.3% of the MCI-AD group and 14.3% of the MCI-LB group, whereas reversion to normal cognition was observed only in the MCI-LB group (14.3%).

Discussion

This study examined longitudinal data to directly compare cognitive outcomes in MCI-LB and MCI-AD under identical multimodal intervention conditions. In this study, no significant differences were observed in the rate of MMSE-J score decline over the 3-year intervention period between MCI-LB and MCI-AD. Although the preliminary nature and small sample size limit generalizability, these findings suggest that individuals with MCI-LB exhibit longitudinal cognitive trajectories under a structured non-pharmacological intervention that are comparable to those observed in individuals with MCI-AD.

During follow-up, some participants in both groups progressed to dementia, and a small number in the MCI-LB group reverted to normal cognition. These clinical transitions were incorporated into the longitudinal analyses rather than treated as competing outcomes. Because linear mixed-effects models accommodate unequal numbers of observations and missing data, these transitions are unlikely to have substantially biased the estimated cognitive trajectories, but should be considered when interpreting the findings.

Most prior research on MCI-LB has used observational designs without active intervention [19, 20], consistently noting marked attentional and visuospatial deficits, greater functional decline, and cognitive fluctuations compared with MCI-AD [21]. This heterogeneity may partly reflect variability in the progression of α-synuclein pathology years before dementia onset, which can shape early clinical features [1]. Evidence on global cognitive decline, however, is mixed: Hamilton et al. [22] reported no difference between MCI-LB and MCI-AD, whereas van de Beek et al. [23] found greater MMSE decline in MCI-AD. In contrast, another longitudinal study reported a rapid mean decline of 2.9 points/year on the Addenbrooke’s Cognitive Examination Revised (ACE-R) in individuals with MCI-LB [24].

In contrast, the present study assessed participants who were engaged in a structured, multimodal intervention. The rapid cognitive decline reported in some observational studies of MCI-LB, including those using broader cognitive batteries such as the ACE-R [24], was not observed in the present study. For MCI-AD, previous estimates range from −0.30 to −1.04 in MCI, compared with −0.34 in this study [25, 26]. While direct comparisons are limited by methodological differences, these results raise the possibility that multimodal interventions may help slow global cognitive decline in both subtypes. Here, this interpretation is based primarily on comparisons with previously reported natural history studies of MCI rather than on direct comparisons with interventional trials. Nevertheless, randomized trials in broader MCI populations have provided evidence that multimodal lifestyle interventions can improve or maintain cognitive function, lending biological plausibility to the potential benefit of structured programs [8, 27]. The absence of group differences further suggests that MCI-LB may derive comparable influence, at least for global cognition. In the absence of a nonintervention control group, it is not possible to determine whether the observed trajectories reflect a beneficial effect of the intervention, no effect, or even a potential negative effect. However, the magnitude of annual MMSE-J decline observed in the present study falls within, and toward the lower end of, the range reported in our prior studies, suggesting that the intervention did not accelerate cognitive decline [14].

Interpretation of MMSE-based comparisons warrants caution because AD and DLB are characterized by distinct cognitive profiles. DLB-related disorders typically show relatively greater impairments in attention/executive function and visuospatial abilities, whereas AD is marked by prominent episodic memory deficits, particularly in the early symptomatic stages [1, 28]. Such domain-level differences can be reflected even within brief cognitive screening instruments; for example, visuoconstructional items such as pentagon copying tend to be disproportionately impaired in DLB, and item-level approaches using the MMSE have been proposed to aid differentiation from AD [29, 30]. Accordingly, MMSE total score should be interpreted in the present study as an index of global cognitive status rather than as a domain-equated measure across etiologies. In this real-world clinical setting, the MMSE-J was used because it was routinely and consistently administered annually within the CIDC program, enabling longitudinal modeling of cognitive trajectories in both diagnostic groups. Importantly, our primary inference concerns similarities or differences in rates of cognitive change over time under the same intervention program (i.e., group-by-time effects), rather than an assumption that baseline MMSE totals represent equivalent cognitive status between MCI-AD and MCI-LB.

Overall, this is among the few studies examining non-pharmacological interventions in MCI while accounting for diagnostic subtype. Given its observational design and small sample size, causal inference is not possible. The small number of participants who reverted to normal cognition, observed only in the MCI-LB group, likely reflects clinical fluctuation or diagnostic instability rather than a sustained disease-modifying effect. Future research should incorporate domain-specific measures and explore tailored interventions to clarify subtype-specific responses.

Clinical Implications

These findings offer preliminary support for the inclusion of individuals with MCI-LB in multimodal non-pharmacological intervention programs. Given its poorer prognosis and limited pharmacological options, non-pharmacological strategies may help slow cognitive and functional decline [5, 31].

While MCI-LB and MCI-AD differ in neuropathology and clinical manifestations [1, 2, 23], the rationale for applying a similar multimodal intervention framework to both groups lies in the fact that both conditions involve multiple interacting pathological and functional domains. These include cognitive impairment, neuropsychiatric symptoms, sleep-wake disturbances, and motor dysfunction, all of which contribute to functional decline even at the prodromal stage [1, 32]. This multidimensional disease structure provides a theoretical basis for multimodal intervention approaches that target several domains simultaneously rather than focusing on cognition alone.

From a mechanistic perspective, the multimodal structure of the CIDC program may be well suited to the complementary cognitive and neuropsychiatric profiles of MCI-AD and MCI-LB. MCI-AD is primarily characterized by episodic memory impairment with gradual involvement of executive functions, for which cognitively oriented components such as memory training and problem-solving activities may be particularly relevant [33, 34]. In contrast, MCI-LB is characterized by fluctuating attention, visuospatial dysfunction, and prominent neuropsychiatric symptoms. Core components of the CIDC program, including physical exercise, rhythm-based activities, and socially engaging programs, have been associated with improvements in attentional control, arousal regulation, and neuropsychiatric symptoms in Lewy body-related disorders [1, 35]. Thus, while cognitively focused components may primarily target amnestic-executive deficits typical of MCI-AD, the inclusion of physical and socially interactive activities may be particularly relevant for MCI-LB. This disease-complementary, multimodal design provides a plausible explanation for the broadly similar global cognitive trajectories observed across the two diagnostic groups in this real-world intervention setting.

Even if the effect size is modest, structured programs that incorporate cognitive, physical, and social activities may help stabilize symptoms and enhance quality of life. These results also underscore the importance of including individuals with MCI-LB in dementia prevention trials and services, rather than excluding them based on assumptions of limited responsiveness.

Strengths and Limitations

This study’s strengths include being among the few preliminary investigations to examine 3-year cognitive trajectories in MCI-LB under multimodal intervention, directly compared with MCI-AD. However, several limitations should be acknowledged. First, the sample size – particularly for the MCI-LB group – was small, which limits statistical power and generalizability. A post hoc power analysis using GPower indicated that the current sample size (N = 74) achieved 99.9% power to detect a moderate interaction effect (f = 0.25), but only 50.4% power to detect a small effect (f = 0.10) at α = 0.05. Thus, the nonsignificant group-by-time interaction may reflect insufficient power to detect subtle group differences, rather than a true absence of effect.

Second, while MMSE-J is widely used as a global cognitive measure in clinical and research settings, it may lack sensitivity to domain-specific impairments such as executive dysfunction, attentional fluctuation, and visuospatial deficits, which are characteristic of MCI-LB. Thus, relying solely on MMSE-J may have led to an underestimation of cognitive changes specific to the MCI-LB phenotype.

Third, unmeasured confounding variables – such as baseline health status, cognitive reserve, individual motivation, and caregiver involvement – may have influenced the observed trajectories.

Fourth, the fact that examiners were not blinded to participants’ diagnostic classification (MCI-LB or MCI-AD) may have introduced assessment bias during cognitive testing. Future studies should incorporate blinded assessment procedures to reduce the potential influence of diagnostic knowledge on test administration or scoring.

Fifth, given the exploratory nature of the analysis, the findings should be interpreted with caution and confirmed in larger, adequately powered, and controlled studies incorporating domain-specific cognitive assessments.

Finally, although missing data were handled under the missing-at-random assumption, informative missingness cannot be excluded. Participants with greater decline may have reduced attendance, raising the possibility of reverse causation. While PS adjustment mitigated confounding, future studies should consider joint modeling or sensitivity analyses to address nonrandom attrition.

Conclusions

This preliminary study examined cognitive trajectories in individuals with MCI-LB and MCI-AD during a 3-year multimodal intervention. No significant differences in cognitive decline were observed between the two groups. These findings offer preliminary support for the potential utility of multimodal interventions in individuals with MCI-LB. Further large-scale, controlled studies incorporating subtype-specific cognitive assessments are warranted to validate these results.

Acknowledgments

We thank the CIDC participants and families, the clinical and administrative staff, the clinical psychologists for assessments, and the Department of Psychiatry for referral support. The authors used a generative AI tool (ChatGPT, GPT-5; OpenAI, 2025) for minor language editing. All intellectual and analytical content was created and verified by the authors.

Statement of Ethics

The study protocol adhered to the ethical standards outlined in the Declaration of Helsinki and received approval from the Medical Ethics Committee of the University of Tsukuba, Japan (Reference No. R03-148). In accordance with national ethical guidelines for medical and health research involving human participants in Japan, the requirement for obtaining written informed consent was waived. Instead, an opt-out method was utilized: relevant study information was made publicly available on the institution’s website, allowing individuals the opportunity to decline participation. Opt-out informed consent protocol was used for use or collection of participant data for research purposes. This consent procedure was reviewed and approved by the Medical Ethics Committee of the University of Tsukuba, Japan, Approval No. R03-148, date of decision June 18, 2023.

Conflict of Interest Statement

T.A. has received lecture honoraria from Eisai, Eli Lilly, and Otsuka Pharmaceutical. K.N. has received lecture honoraria from Otsuka Pharmaceutical. All other authors declare no conflicts of interest relevant to this study.

Funding Sources

This study was funded by JSPS KAKENHI (Grant Nos. JP 25K14161, JP 25K02576).

Author Contributions

M.N., H.S., and M.O. were responsible for the study’s conceptualization, data analysis, and initial drafting of the manuscript. K.N., H.S., A.K., and H.M. participated in data collection and contributed to the interpretation of findings. A.S. provided critical input on the interpretation of the results and substantially revised the manuscript for intellectual content. T.A. oversaw the overall study design and interpretation process and contributed to the critical revision of the manuscript. All authors were actively involved in the research, reviewed and approved the final manuscript, and agreed to take full responsibility for the integrity and accuracy of the work.

Funding Statement

This study was funded by JSPS KAKENHI (Grant Nos. JP 25K14161, JP 25K02576).

Data Availability Statement

The datasets generated and analyzed during the current study are not publicly available due to ethical restrictions and the sensitive nature of clinical patient information. However, de-identified data may be made available by the corresponding author upon reasonable request and with appropriate institutional approval.