Effect of choline alfoscerate in older adult patients with dementia: an observational study from the claims data of national health insurance

Abstract

Background

Choline alfoscerate, a cholinergic precursor with limited evidence of efficacy in dementia management, has been used for various cognitive impairments in Korea. Partly due to its insurance coverage, this agent appears to incur significant expense for the insurance system. Thus, we aimed to describe choline alfoscerate prescription patterns and analyze their long-term effects in an older adult cohort with dementia.

Methods

This observational study used the National Health Insurance Service Senior Cohort Dataset. Choline alfoscerate -naïve patients who were diagnosed with dementia between 2003 and 2014 with at least 12 months of follow-up were selected. Time-dependent Cox regression was employed to estimate the association between drug exposure and the risk of treatment failure events.

Results

There were 11,463 eligible participants, of whom approximately 73% were female, and 19% had been exposed to choline alfoscerate. According to the main regression survival analysis, the association between longitudinal choline alfoscerate use and the risk of progression events related to treatment failure was unclear. However, a significant decrease of nearly 20% in the risk of all-cause mortality was associated with choline alfoscerate exposure, and a slight reduction in progression regarding treatment failure was observed with CA use only during the early stages of diagnosis. Age, sex, insurance premiums, several comorbidities and concurrent medications were significantly associated with the probability of the events according to the multivariate models.

Conclusions

Further analyses are needed to confirm the early-stage and long-term effectiveness of choline alfoscerate in specific populations, which will help in considering its reimbursement.

Graphical abstract

Background

Dementia affects cognitive and emotional functions beyond the expected consequences of biological aging [1]. The most common form of dementia is Alzheimer’s disease followed by vascular dementia [2,3,4]. The number of reported cases of dementia continues to increase worldwide [1, 5], as does the health economic burden for patients and society, especially in fast-aged countries such as the Republic of Korea [2, 4]. More than half of the burden of this disease is due to the indirect cost of productivity decline due to treatment, informal care, and early death; medication claims account for less than 12% [4]. This could be mainly due to the lack of pharmacologic agents with clear evidence for dementia treatment. Acetylcholinesterase inhibitors (AChEIs, such as donepezil, galantamine, and rivastigmine) are currently recommended to manage mild to moderate dementia; and memantine, a N-methyl-D-aspartate receptor antagonist, is considered for treating patients with moderate to severe dementia [6, 7].

Most approved agents only offer temporary relief from dementia symptoms [8]. Therefore, there is a need for disease-modifying therapies that not only slow the progression of dementia but also provide neuroprotection to enhance cognitive function, behaviors, and quality of life [9]. Choline alfoscerate (CA, choline alphoscerate or α-glycerylphosphorylcholine, α-GPC), a cholinergic precursor, is believed to increase the level of endogenous choline by promoting its formation and release [10, 11]. This mechanism of action may be primarily responsive to its beneficial effects in patients with mild-to-moderate dementia over a short period [12,13,14,15,16,17,18]. Several preclinical and clinical studies have shown that CA is effective and safe in preventing [19] and improving cognitive functional disorders like dementia [9], whether used alone or in combination with other treatments. CA has been approved as a medication in countries like Poland and Russia. In the Republic of Korea, it was introduced as Gliatilin^® in 1998 for managing various conditions, including “secondary symptom and metamorphosis or degenerative brain-organic psychiatric syndrome by cerebrovascular deficiency, emotional and behavioral change, and senile pseudodepression” [20].

Although there is limited evidence assessing the cost-benefit balance of CA with public health insurance [21]; its cost is covered by Korean insurance system, making it a significant expense for patients with Alzheimer’s dementia, secondly to donepezil [20]. In 2020, the Korean Ministry of Food and Drug Safety decided that pharmaceutical companies manufacturing CA oral formulations must re-evaluate and submit their clinical development plans for all indications for this agent [22]. They decided to reduce the insurance payment rate for CA to 80% and recollect the remainder from companies [23]. In addition to trials, learning about the real-world use of this agent is necessary. The aim of our pharmaco-epidemiological study was thus to (1) describe choline alfoscerate prescription patterns and (2) analyze its long-term effects in an older adult cohort with dementia.

Methods

Data source and setting

The National Health Insurance Service Senior Cohort (NHIS-senior) represents 10% of a total of 5.5 million Korean insurance beneficiaries from the age of 60 at the time of entry, for which longitudinal claims data were collected from 2002 to 2015 [24,25,26]. All patient data were provided anonymously, without identifying information. The Korean Standard Classification of Diseases 7, equivalent to the International Classification of Diseases 10 (ICD-10), was used to identify the primary and secondary diagnoses for each claimed prescription [24, 25] and was listed in Supplementary Materials – Appendix Table A1. The prescribed medications were represented by the common name drug code of each specific dosage formulation according to the Medical Code of Korea (see more coding details in Supplementary Materials – Appendix Table A2). Other information of demographics and mortality is also provided in the database [27].

Study design

This retrospective cohort study investigated CA prescription and its effect on older adult Korean patients with dementia using a time-dependent Cox regression model to alleviate immortal time bias [28]. Landmark time analysis, another method for controlling this bias [29], was applied for the sensitivity analysis (Fig. 1). A one-year washout period (2002) was excluded from the database to include only newly diagnosed users.

Fig. 1

Study design with the main and sensitivity analyses, timepoints, and corresponding variables

Participant eligibility

Patients were selected when the diagnosis of dementia was confirmed from January 01, 2003, to December 31, 2014, at the first date of medical claims (index date) for AChEI prescription (including donepezil, galantamine, and rivastigmine, which lasted for at least 90 days) in the NHIS-Senior database for the primary or second diagnosis of dementia (including corresponding ICD10 of F00, F01, F02, F03, and G30) [26]. We did not select patients with a CA prescription before the index date to exclude those who used CA for other purposes, such as dementia prevention, or who had insufficient follow-up time in the database (under 12 months).

Study time point and exposure definition

In the primary (time-dependent) analysis, the follow-up time for each patient was set from the index date (the first date of dementia diagnosis confirmation) to the end of follow-up. End-of-follow-up was defined as the date of the corresponding endpoint event or the last date of claims in the database (censoring) until December 31, 2015, whichever occurred first. The exposed patients received CA claims for at least 90 days, whereas the others were assigned to the unexposed arm.

In the sensitivity analysis, the landmark period was established from the entry date (day 0, the date of dementia diagnosis confirmation) to multiple landmark time points defined as index dates (Fig. 1). Based on our preliminary results from the NHIS-Senior database, most patients began CA treatment within the first 2 years after diagnosis, with the duration of use typically less than 2 years. To more accurately define CA exposure and avoid immortal-time bias, given the challenge of classifying patients who experience outcomes before CA exposure, we have established landmark time points at 360, 720, and 1080 days (corresponding to the first three years). Beyond that time, the rate of CA use was obviously lower compared to what was observed in previous research [20]. The data of eligible patients for sensitivity analysis had to be available in the database for at least 6 months from the index date; CA-exposed patients had to start CA prescriptions during the landmark period.

Variables and dementia management endpoints

The baseline demographic characteristics of the patients included age, sex, and higher insurance premiums (≥ 6th quantiles) in the same year as the index date. The predefined comorbidities, use of donepezil as the main treatment for at least 90 days, and use of other concurrent medications for at least 60 days were also clarified (see more detail in Fig. 1 and Supplementary Materials - Appendix Table A1). CA use features, including onset time and treatment duration, were described.

The primary endpoint of dementia management is a composite endpoint that comprises disease progression events regarding treatment failure, including need for AChEI treatment optimization due to reduced treatment response, such as starting multiple AChEI combinations, changing to memantine, or starting the maximum recommended AChEI daily dose (donepezil ≥ 10 mg, galantamine ≥ 24 mg, transdermal rivastigmine ≥ 27 mg (13.3 mg/24 h), and oral rivastigmine ≥ 12 mg) [30] and all-cause death [31, 32]. Other clinical outcomes related to disease progression or complications (such as delirium, ICD-10 code F05) are at high risk of underdiagnosis in this dataset. Therefore, we excluded these conditions from our endpoints. The secondary endpoint was all-cause death events solely [33], to better provide insights into the mechanism of action and the role of CA in patients with dementia. The survival time (including progression-free interval and overall survival) was defined as the duration from the index date to the date of censoring or any event occurrence, whichever came first.

Statistical method

Patient characteristics are presented as the means and standard deviations (SDs) or medians and interquartile ranges (IQRs) for continuous variables, and as frequencies and percentages for categorical variables. A Cox proportional hazards regression model was constructed to quantify the hazard ratio (HR) of events associated with CA prescription and its 95% confidence interval (95% CI). The significance threshold of the difference and correlation was set at a p value of 0.05.

Results

Baseline characteristics of the participants

Of the 37,007 patients in the database who confirmed their dementia diagnosis during the study period, 11,463 patients were eligible for our primary survival analysis (Fig. 2). There were 2173 (19%) patients exposed to CA, and the remaining patients were assigned to the nonexposed group.

Fig. 2

Participant selection flowchart. AChEIs: acetylcholinesterase inhibitors; CA: choline alfoscerate; N: number

The baseline characteristics of all the participants and each CA-exposed group are shown in Table 1. Patients with CA were slightly younger (mean value, 77.5 years; SD, 6 years) than those without CA (78.5 years, 6.4 years). Approximately 73% of the participants were female. The percentage of patients with high insurance premiums was significantly lower in the CA use group (58.3% vs. 60.9%, p = 0.027). The three most frequently recorded comorbidities in patients with dementia were hypertension (77.3%), dyslipidemia/ atherosclerosis (45.5%), and diabetes mellitus (38.8%). The proportion of patients with dyslipidemia/ atherosclerosis was greater in the exposed arm than in the unexposed arm (50.5% vs. 44.3%, respectively). Concurrent medications related to these chronic diseases were also prescribed at higher rates for users than for nonusers, except for antidiabetic drugs. Approximately 80% of the patients used donepezil for dementia treatment.

Table 1 Patient baseline characteristics overall and of the choline alfoscerate-exposed group versus the nonexposed group

Fewer apparent differences were observed between the study groups in the sensitivity analysis (Supplementary Materials - Appendix Tables A3). According to baseline characteristics of multiple landmark timepoints, the trend of antithrombotic use and insurance premiums was confirmed in the study groups.

Choline alfoscerate prescription characteristics

Regarding CA prescription patterns, all study patients started their choline alfoscerate prescription orally, which included soft capsules and tablets. Eligible patients initiated this prescription with a median onset time of 116 days (IQR = [0-516] days) from baseline and a median treatment duration of 496 days (IQR = [302–832] days).

Effect of choline alfoscerate on the study endpoints of patients with dementia

In general, lower rates of primary (54.9% vs. 64%) and secondary endpoint events (24.3% vs. 35.7%) were observed in the CA-exposed group than in the nonexposed group. Cox regression analysis revealed that a nearly 20% decrease in the risk of all-cause mortality was associated with CA exposure (HR = 0.82, 95% CI = [0.77–0.87]), but no significant difference was detected (p > 0.05) in the incidence of disease progression regarding failure to treatment response between the two groups (Table 2).

Table 2 Study endpoints in each group of choline alfoscerate-exposed/ nonexposed in the time-dependent analysis

The results of the sensitivity analysis for the survival of each endpoint at each landmark time point are shown in Fig. 3. Almost all the analyses comparing progression-free interval and overall survival between CA-exposed and nonexposed groups did not show any significant difference. Only in the analysis of the 1080-day landmark time (Fig. 3Ic), a significant increase in the probability of a dementia progression-free interval was observed in the CA-exposed group, most clearly observed during the first 5 years of follow-up. The median survival time was moderately greater in patients with disease progression events related to treatment failure (3.97 vs. 3.2 years). It was similar to all-cause mortality (5.87 vs. 5.84 years) among CA users compared with nonusers.

Fig. 3

Survival curves in landmark analysis with the x-axis of time (year) and the y-axis of survival probability (%), comparing (I) progression-free interval and (II) overall survival between choline alfoscerate-exposed (blue (dark gray) line) and nonexposed (orange (light gray) line) groups at different landmark timepoints of (a) 360 days, (b) 720 days, and (c) 1080 days

Discussion

This study provides an overview of choline alfoscerate prescriptions and their effects on older adult patients with dementia. To our knowledge, this is the first study to quantify the outcomes of long-term CA prescription in patients with dementia using a claims database. The risk of all-cause mortality decreased by approximately 20% with oral CA exposure; however, the association between longitudinal use and the risk of dementia progression events due to failure to response remains unclear. CA exposure is associated with a reduction in dementia progression regarding treatment failure during the early years following dementia diagnosis. Therefore, these results should be interpreted cautiously.

Choline alfoscerate prescription patterns and its effect on dementia outcomes

Our analysis was based on 14-year data from a cohort of Korean older adult beneficiaries. Less than 20% of the study subjects initiated oral CA, which was slightly lower than that reported by Hwang & Park (2019). This may be mainly because we focused on the group that started CA after confirming the diagnosis of dementia, which helped limit the bias in including previously CA-exposed patients.

In our study, the median survival times for mortality were in the same range as those in other studies [33, 34]. Need for AChEI treatment optimization due to a reduced treatment response or symptom exacerbation, and death reflects the long-lasting effectiveness of this disease [32]. We were able to define these as the primary endpoint by utilizing the claims data. In the time-dependent analysis, no significant difference in the probability of disease progression was observed between the two groups, and there was a slight improvement in all-cause mortality in the exposed arm. In sensitivity tests with a smaller number of patients, the difference in mortality disappeared. This indicates that the effect of CA is minimal and may vary among different groups in the population with dementia.

Evidence from animal and human studies suggests that CA may exert further synergic effects with AChEIs via its impact on increasing choline levels. Some research indicates that CA can protect the brain from injury related to vascular or seizure issues [21, 35, 36], with a broader effect observed when this compound is used in combination with galantamine or donepezil [37, 38]. Additionally, patients with dementia often struggle with managing other chronic conditions, such as diabetes-related comorbidities [39]. Consequently, the short-term use of CA may show only minimal improvement in the progression regarding treatment failure of AChEIs, which can be reversed upon discontinuation of CA. If CA does have a neuroprotective effect and aids in managing comorbidities, this could partly explain its long-term impact on mortality. This insight may open avenues for future research into CA’s mechanisms of action and its effects on managing various comorbidities in patients with dementia.

Earlier, the interim results of the ASCOMALVA trial revealed several favorable outcomes of the combination of donepezil and CA in Alzheimer’s disease [12,13,14, 16, 18], including its early effect on nervous system morphology in the first years from the initial years, which subsequently appeared to decrease [15]. Well-designed longitudinal trials are required to confirm the clinical relevance of these findings [21]. This agent is an additional treatment for mild-to-moderate dementia [13, 16]; the efficacy of extended use, especially in more severe conditions in the presence or absence of complications, has not been justified. Therefore, once further evidence confirms the efficacy of CA for this complex chronic disease with multiple comorbidities, physicians should routinely evaluate the patients’ progression. Discontinuing the medication may be considered with this evidence-limited agent to reduce polypharmacy and alleviate the economic burden on patients and the healthcare system.

Other potential related factors

Our study population in general had similar epidemiological characteristics to those reported in previous studies conducted in Korea and other countries [2, 4, 5, 40, 41]. The age range of patients with dementia in our study is consistent with findings from studies conducted in China, Germany, and the US [42,43,44]. However, the proportion of female patients in our study is higher compared to other studies [42,43,44]. Donepezil was also the most frequently prescribed AChEI for dementia patients overall and specifically for those with Alzheimer’s disease [43]. Similar to the study by Hwang & Park (2019), the nonuse group seemed to have more female patients than the CA-use group, whereas 2/3 of the patients with dementia were females who had longer expectancies and were thus in a greater age range [45]. A lower proportion of females was reported in the CA-exposed group; consequently, a lower average age was observed. Additionally, our results revealed that patients exposed to CA were less likely to have high insurance premiums. These findings are in line with previous results, which suggest that CA prescriptions are more common in provinces or cities outside the capital, and local health centers, as opposed to medical institutions, tend to primarily prescribe CA [20]. Furthermore, patients receiving these prescriptions are more likely to have medical aid insurance, which is predominant among individuals with lower incomes [20]. Because economic and setting factors can influence access to welfare or healthcare services, these differences may partly contribute to the disparities of healthcare support and even the timing of dementia diagnosis between the exposed and unexposed group. This indicates that CA users and nonusers may not share the same severity at baseline or during long-term care, which is crucial for study design and evaluation [46].

In terms of comorbidities, hypertension, dyslipidemia, and diabetes mellitus are the most frequently reported risk factors for dementia development [47], in line with previous studies [14, 33, 41]. There was a trend toward more comorbidities (but not significantly different) and concomitant medications in the CA group. The appropriate control of comorbidities such as cardiovascular diseases can be associated with slower dementia progression [48,49,50]. Given that our study was not a trial, we can only observe differences in “claimed” comorbidities and comedication use between groups. Managing comorbid conditions can influence the treatment response of dementia. However, accurately assessing this impact is difficult due to the limited clinical data available. Other factors, such as healthcare literacy, may contribute to this chronic disease outcome and need to be considered when interpreting the overall results. In fact, there is little information in the literature about the determinants of dementia progression or treatment failure (mostly the predictors of dementia development). Therefore, high-scale studies should be conducted to identify the key factors that influence treatment outcomes. This helps limit biases, particularly in observational research, and generates study results for the target treatment population by fully reporting the participant characteristics [51].

Study limitations

Our study has several limitations. First, a meta-analysis concluded that non-Alzheimer’s dementias were associated with higher mortality rates than was Alzheimer’s disease [3]. The effect of an anti-dementia agent may vary between different disease forms and indications; however, our study did not separately analyze the effects of CA in each subgroup. In addition, the impact of the progression of dementia on endpoints needs to be considered. Further controlled studies should consider collecting relevant clinical information to stratify or match patients to balance the characteristics of each group at baseline. Given the nature of an administrative claims database, our findings may face accuracy challenges, including potential coding errors, misclassification, underdiagnosis, and limitations in capturing informal care or non-prescription treatments. Moreover, potential confounding factors beyond extracted information from the database (concurrent use of dementia medications for instance), may influence the interpretation of the results. Conclusions about causal relationships cannot be drawn directly from this retrospective data. Our study specifically examines the use of adjuvant CA in dementia patients. Otherwise, NHIS data may not capture all medication treatments, particularly for those with private insurance [52]. Consequently, our findings may not apply to all populations, and further research is needed to explore other clinical scenarios, such as preventing cognitive decline. Finally, a population pharmacokinetic study of oral CA in healthy Koreans showed that the regimen of daily use may affect the choline levels of patients [10]. Since treatment compliance can be a critical issue in patients with dementia, future studies should consider this to more accurately assess the effects of drugs in dementia management.

Conclusions

The use of choline alfoscerate was associated with a slight reduction in dementia progression regarding treatment failure, but this association was only notable during the initial years following diagnosis. Further studies are needed to clarify the effectiveness-risk and effectiveness-cost balance, such as longitudinal studies with clinical outcome measurements in specific populations (e.g., patients with Alzheimer’s dementia stratified by various stages of severity, dementia patients with specific comorbidities, etc.). This will aid in deciding on clinical interventions for medication use in dementia patients and will also contribute to determining the exact reimbursement rate for choline alfoscerate.