Mild cognitive impairment (MCI) represents an intermediate state between normal aging and dementia.1 Several longitudinal studies have shown that most persons with MCI are at increased risk for the development of dementia.23 Especially, when memory loss is the predominant symptom, it is termed "amnestic MCI" and is frequently seen as a prodromal stage of Alzheimer's disease (AD).4 Therefore, MCI has been receiving considerable attention in clinical practice and research settings.

A question commonly raised by patients with MCI and their family members concerns the likelihood and time course of progression to dementia. Although the general rate of progression among those with a diagnosis of MCI is estimated at 10–15% per year,1 certain factors predict a more rapid progression. The degree of memory impairment at presentation is a clinical predictor of progression,5 probably because these patients are closer to the threshold for the diagnosis of dementia. Longitudinal data have shown that the progression to dementia is more rapid among carriers of the apolipoprotein (APOE) ε4 allele than among noncarriers.6 Other considerations such as old age and low education were also reported.7

As MCI may represent a prodromal state to clinical AD, treatments proposed for AD, such as cholinesterase inhibitors (ChEIs), may be considered for MCI, too. However, until now, there is no proven treatment for MCI.8 Several potential treatments are still under investigation.9 Although the risk-benefit ratio is questionable, some interventions in patients with MCI are being used in clinical practice without evidence.

We planned this study to investigate the effects of medications without evidence, frequently prescribed in clinical practice, such as ChEIs and neural pills including nootropics and Ginkgo Biloba, on progression to dementia, in addition to determine the potential clinical predictors of progression from MCI in a Korean cohort.

Of the 778 participants with MCI who completed at least one follow-up visit, 215 (27.63%) progressed to dementia (200 AD, 8 VaD, 4 DLB, and 3 FTD) and 74 (9.51%) reversed to normal cognition. The demographic and clinical characteristics of the participants who progressed into dementia, remained in MCI, and revised into normal cognition, are summarized in Table 1. Mean follow-up duration was 1.42±0.72 years, and not different between groups. Participants who progressed to dementia were older and had lower K-MMSE scores at the baseline evaluation, compared with those who did not progress. Even after adjusting for age and K-MMSE scores, CDR-SOB score, BMI value, and frequency of presence of APOE ε4 allele were different. GDS scores were different at follow-up (p=0.0006), although these were not different at baseline. The change of GDS scores during follow-up was 1.79±5.11 in participants who progressed to dementia, which was higher compared to -0.28±4.17 in those remained, and -1.24±3.16 in those reversed (p=0.0001). Distribution of other characteristics including family history of dementia and history of concomitant medical diseases was not different.

Table 2 shows the differences of total medical expenses including medication costs and number of used medical institutions, and the prescription frequency of ChEIs and neural pills during follow-up. During the follow-up period, neural pills were prescribed to 539 (69.28%) MCI participants and ChEIs were used by 395 (50.77%). ChEIs were prescribed to 172 (80.00%) participants with progression to dementia, 208 (42.54%) remained, and 15 (20.27%) reversed (p<0.001). Other differences were not observed.

Initial Cox proportional hazard models, adjusted for age and K-MMSE score, revealed that some factors were associated with a higher risk of MCI progression: lower education years, lower BMI values, amnesia, higher CDR-SOB and S-IADL scores at baseline, increased GDS scores during follow-up, and presence of APOE ε4 allele, in addition to older age and lower K-MMSE score at baseline (Table 3). To further analyze the best predictors of dementia progression, we performed Cox proportional hazard analyses entering all the above factors showing significance at the initial analysis. Older age [hazard ratio (HR), 1.036; 95% confidence interval (CI), 1.006–1.067; p=0.018], APOE ε4 allele (HR, 2.247; 95% CI, 1.512–3.337; p<0.001), higher CDR-SOB score (HR, 1.367; 95% CI, 1.143–1.636; p=0.001), higher S-IADL score (HR, 1.035; 95% CI, 1.003–1.067; p=0.029), and lower K-MMSE score (HR, 0.892; 95% CI, 0.839–0.949; p<0.001) predicted greater hazard (Table 3).

This study shows the risk of progression from MCI to dementia and its underlying predictors, in a clinical Korean cohort for dementia research. Older age, worse baseline cognitive and global function, and presence of the APOE ε4 allele are the best predictors. Additionally, lower education years, lower BMI values, presence of amnesia at baseline, and increased GDS scores during follow-up had some influence on the progression to dementia. The present findings are in agreement with those from previous studies.5672324 Furthermore, with linkage to the national health insurance claims database, we show that total medical expenses are not different between MCI participants who progressed to dementia and those who did not. Although it was based on a clinical cohort, we believe that the present evidence will add to the existing knowledge on the risk of dementia, especially AD, in Korea, which has been accumulated mostly from epidemiological studies in this country.24

At present, there has been no medication approved for the treatment of MCI. In several placebo-controlled clinical trials, there was no significant reduction in the rates of progression to dementia among MCI patients who were treated with ChEIs.6252627 In one trial, donepezil significantly reduced the risk of progression to AD for the first 12 months of the study (and for up to 24 months in subgroups who were carriers of APOE ε4) but had no significant effect on the risk of AD at 36 months, which was the primary study outcome.6 The present study showed that, at the initial Cox proportional hazard models, HR for ChEIs during follow-up was 3.546 (95% CI, 2.48–5.08, p<0.001). However, the result was not significant after multivariate analysis. We do not suggest that ChEIs might make patients with MCI progress rapidly. During the follow-up of mean 1.42 years, 396 (50.77%) participants with MCI were prescribed with ChEIs. Eighty percent of MCI participants who progressed into dementia were prescribed with ChEIs, which was higher than in those who did not progress. During follow-up, changes in K-MMSE, CDR-SOB, and S-IADL scores were more severe in participants who progressed to dementia. We do not have information on whether the participants were already taking ChEIs before being enrolled into the claims database. However, mean age at the initial visit was older, and initial and follow-up scores of K-MMSE were lower in participants prescribed with ChEIs during follow-up. Our findings might indicate that substantial MCI patients in Korea, who were considered as severe as dementia by clinicians, could be prescribed with ChEIs, but this medication could not prevent the progression to dementia. In our cohort data, the annualized rate of progression to AD in our study (17.49%) was slightly higher than the original estimate (10 to 15% per year) by Petersen et al.1 Another study performed in Korea showed the annual conversion rate of 15.48%.28

This study has some limitations. The first is regarding the rates of progression to dementia or reversion to normal cognition observed in our study. Within the mean 1.5 years of follow-up, 27.63% of participants progressed to dementia, which represents an annual conversion rate slightly higher than those found by most studies, generally ranging from 10 to 15% (1–3). This might be due to a selection bias. We investigated only those participants with follow-up evaluations. Several individuals who had MCI at baseline were excluded from our analyses for missing a follow-up evaluation. These individuals differed significantly from those remaining in the study. Patients without cognitive decline would not visit the clinic for a follow-up evaluation. Moreover, 9.51% participants showed reversion to normal cognition, which could be by seeking help for their cognitive difficulties through higher cognitive, physical, and social activities. Consequently, we may be reporting an overestimated prevalence of reversion. An association between age and reversion was not found in this study, but there are reports that individuals younger than 70 years may show different predictors of reversion.29 Moreover, the relatively short duration of follow-up in this study provides only a narrow window of investigation. A longer follow-up period is needed to determine the extent to which unstable MCI represents a very early MCI stage of serious cognitive decline. A final limitation is our lack of consideration of transitory cognitive impairment associated with factors like stress, acute illness, or poor motivation, with MCI diagnosed under such conditions at greater than normal chances of reversion to normal cognition.2930 Additionally, we could not exclude the possibility that MCI patients who progressed to dementia already took ChEIs or neural pills at other clinics (not in the database) before the baseline enrollment in the database. K-MMSE score at initial visit was lower in participants prescribed with ChEIs during the follow-up. Although adjustment for K-MMSE score was performed, matching the baseline cognitive status between those who were prescribed with and without ChEIs could further evaluate the effect of ChEIs on the rate of cognitive decline.

However, this Korean multi-hospital based cohort database showed that 215 (27.63%) of 778 participants with MCI progressed to dementia including 200 AD, during mean 1.42±0.72 years of follow-up duration, and the best predictors for progression were older age, lower K-MMSE scores, higher CDR-SOB, and S-IADL scores, and APOE ε4 allele. Lower education years, lower BMI values, and increased GDS scores during follow-up can have some effect on the prognosis.