Simulation of Alzheimer's diagnostic flows with blood biomarker test options in Japan

Abstract

INTRODUCTION

This study projected the diagnostic testing landscape for lecanemab treatment in Japan under different workflows.

METHODS

A dynamic simulation estimated wait times and treatment‐eligible patient numbers under four scenarios: current diagnostic workflow, blood biomarker (BBM) tests as triage tools, BBM tests for confirmatory diagnostics, and both combined. Willingness‐to‐pay (WTP) and intangible costs were assessed via an online survey to estimate testing demand.

RESULTS

The maximum mean wait time under the current workflow was projected at 6.4 months, decreasing with BBM integration. The number of treatment‐eligible patients increased considerably with BBM‐based confirmatory diagnostics. BBM triage testing reduced wait times but temporarily increased treatment‐eligible patients.

DISCUSSION

Replacing positron emission tomography (PET) or cerebrospinal fluid with BBM‐based diagnostics may increase treatment eligibility because of lower costs, driving higher demand for testing.

Highlights

• A dynamic simulation models Alzheimer's diagnostic workflows in Japan.

• Blood biomarker (BBM) tests reduce diagnostic wait times for Alzheimer's in Japan.

• Implementing BBM tests improves access to Alzheimer's diagnostics.

• Study quantifies demand for diagnostic testing based on costs and accessibility.

• Testing costs impact the number of treatment‐eligible Alzheimer's patients.

1. INTRODUCTION

In recent years, several disease‐modifying treatments (DMTs) have been developed to address mild cognitive impairment (MCI) and mild dementia associated with Alzheimer's disease (AD). ¹ , ² , ³ Among these, lecanemab received approval in Japan on September 23, 2023, and became available for clinical use on December 20, 2023. Before initiating lecanemab treatment, amyloid‐β pathology should be confirmed, in addition to a diagnosis of MCI or mild dementia.

Despite these advances, challenges remain in the current diagnostic infrastructure. The Japanese Ministry of Health, Labour and Welfare has issued an Optimal Use Guideline (OUG) for lecanemab, outlining prescription criteria related to facility requirements and patient eligibility. ⁴ In addition to ensuring the safe and effective use of novel pharmaceuticals, OUGs consider the economic impact of high‐cost drugs on the healthcare insurance system. ⁵ While not legally binding, these guidelines carry substantial influence, making adherence critical in clinical practice. According to the OUG, patients seeking lecanemab treatment must consult specialists and undergo diagnostic testing, with either positron emission tomography (PET) or cerebrospinal fluid (CSF) analysis, to confirm AD pathology through amyloid‐β detection. However, prior research has identified prolonged wait times for these diagnostic tests in Japan because of a shortage of specialists. ⁶ Clinicians report delays of several months at certain medical institutions before patients can consult a specialist. Additionally, PET scans are costly and available at a limited number of facilities, primarily in urban areas. ⁶ , ⁷ This disparity imposes a greater burden on patients in rural regions when accessing treatment opportunities. While CSF tests are more accessible and less expensive than PET scans, they are invasive and pose risks of adverse effects. ⁸ Combined with the shortage of specialists, these factors limit timely diagnosis and treatment access.

RESEARCH IN CONTEXT

1. Systematic review: We reviewed literature from traditional (e.g., PubMed) sources to assess challenges associated with diagnostic tests for Alzheimer's disease required for prescribing disease‐modifying drugs, such as lecanemab. Prior studies have highlighted issues with prolonged wait times, access disparities, and high costs of traditional modalities such as positron emission tomography (PET) scanning and cerebrospinal fluid (CSF) analysis. Recently, blood biomarker (BBM) tests have also been evaluated for their potential to streamline workflows and increase testing capacities.

2. Interpretation: This study shows that integrating BBM tests into diagnostic workflows increases the proportion of treatment‐eligible patients, primarily due to reduced costs, which raises the demand for testing. Using dynamic simulations, we provide new insights into demand dynamics influenced by costs and other factors, complementing existing research.

3. Future directions: Future research should explore the scalability of BBM test implementation across diverse healthcare settings and examine its impact on infrastructure, including magnetic resonance imaging (MRI) monitoring, drug infusion capacity, treatment costs, and overall benefits, to ensure healthcare sustainability.

BBM tests have emerged as a promising alternative for detecting AD pathology. ⁹ , ¹⁰ , ¹¹ Blood tests are widely accessible, less dependent on specialists, and suitable for primary care. ¹⁰ Two clinical pathways for BBM use have been proposed: as triage tools before confirmatory tests or as confirmatory diagnostics. ¹⁰ The Global CEO Initiative on Alzheimer's Disease BBM Workgroup recommends ≥90% sensitivity and ≥85% specificity for triage and ≥90% sensitivity and specificity for confirmatory tests, ¹¹ with the World Health Organization stating that BBM tests should have ≥90% sensitivity and specificity to have diagnostic value. ¹² Integrating BBM tests into practice is expected to streamline diagnostics and expand testing capacity by allowing general practitioners (GPs) and other healthcare providers to conduct them, meeting the rising demand. Additionally, blood tests are less costly and less invasive than PET or CSF tests, ⁹ , ¹⁰ , ¹¹ potentially increasing patient demand. While implementation challenges persist, ⁹ incorporating BBM tests into clinical workflows could enhance treatment accessibility for eligible patients. Currently, BBM is recommended as a triage tool before conducting PET or CSF tests in specialist settings in the Japanese guidelines related to BBM. ¹³ However, given BBM's diagnostic performance, it is anticipated that it could eventually be used as a substantially equivalent alternative to PET or CSF. Furthermore, considering BBM's versatility, its use as a triage tool in primary care settings may also be explored.

Although substantial wait times for patients seeking DMTs due to the limited capacity of current testing infrastructure have been projected, ⁶ the balance between supply and demand for testing remains insufficiently explored. Implementing BBM tests is expected to increase demand and expand patient access to treatment. While this shift could benefit individuals and the nation, it may also increase national healthcare expenditures and strain medical resources. Therefore, understanding both the benefits and burdens associated with changes in testing infrastructure and developing healthcare systems to address these issues are essential.

This study projected wait times for diagnostic testing and estimated the number of patients eligible for lecanemab treatment, along with the medical costs associated with testing, across various diagnostic workflows, including the current flow and those integrating BBM tests. A dynamic simulation model was employed to assess interactions between key variables. For example, wait times can influence testing demand, and changes in demand can further affect wait times. An online survey was conducted to estimate willingness‐to‐pay (WTP) for testing and evaluate intangible costs using conjoint analysis.

2. METHODS

2.1. Study overview

This study employed a simulation model to project the status of diagnostic testing for lecanemab treatment among the Japanese population aged 50 years and older, examining wait times and the number of treatment‐eligible patients across all regions. The model incorporated multiple testing workflows and accounted for testing demand. In estimating demand, we assumed that the decision to undergo testing was influenced by both direct (tangible) costs and intangible factors, such as wait and travel times to testing locations. These intangible factors were converted into monetary values using conjoint analysis, ¹⁴ , ¹⁵ resulting in intangible costs. Individuals seeking testing were defined as those whose WTP exceeded the total costs of testing (Table 1).

TABLE 1.

Cost items included in assessing the demand for the diagnostic testing.

Category	Tangible costs	Intangible costs a
Testing costs b	Out‐of‐pocket expenses associated with the testing	Wait time, test sensitivity, travel time to the test location, test duration (time commitment), radiation exposure, physician explanation of test results using images, side effects
Travel costs	TranspoTransportation costs	Travel time

^a Intangible costs represent monetary values of intangible factors converted using conjoint analysis.

^b Testing costs do not include the fees for consultations, including cognitive assessment, with general practitioners.

The conjoint analysis was conducted using SAS software (version 9.4, SAS Institute Inc., Cary, NC, USA), while the dynamic simulation was implemented in Python (version 3.11.9, Python Software Foundation).

2.2. Estimation of WTP and intangible costs

An online survey estimated WTP and intangible costs associated with diagnostic testing among the Japanese population with varying cognitive conditions (see Appendix Method A1 for details). The survey collected data on WTP, preferences for diagnostic test attributes (for conjoint analysis), self‐reported and informant‐reported cognitive complaints, and attributes including demographic characteristics and health‐related information.

For the WTP survey, respondents were randomly assigned to one of three cognitive condition roles: cognitively normal, MCI, or mild dementia. Each respondent received a scenario describing their assigned condition, the treatment effect of medication, and the probability of being eligible for treatment. Using an open‐ended format, respondents indicated their WTP for diagnostic testing to determine treatment eligibility (see Appendix Method A2 for details).

The intangible costs were estimated using conjoint analysis (see Appendix Method A3 for details). Conjoint analysis is a well‐established, survey‐based statistical technique used to assess how individuals value different attributes, including features, functions, and benefits that define a product or service. ¹⁴ , ¹⁵ Seven attributes influencing demand for diagnostic testing were identified based on expert opinion (authors: N.K. and E.M.): wait time for treatment initiation, test sensitivity, travel time to the test location, test duration, radiation exposure, physician explanation of test results using images, and side effects. Additionally, out‐of‐pocket expenses for the test were included as a monetary attribute. Preference scores for each attribute level were estimated using a logistic regression model. The highest preference level for each attribute was assigned a score of 0, indicating the strongest preference, while lower preference levels were assigned negative scores, reflecting their relative disutility compared to the reference level. These scores were then converted into monetary values in Japanese Yen (JPY; USD 1.00 = approximately JPY 153.72 as of November 2024) by comparing them with the monetary attribute.

2.3. Simulation

An agent‐based model was used for the dynamic simulation. This model is particularly useful for studying how patient behavior and clinician decision‐making interact to influence healthcare outcomes. ¹⁶ , ¹⁷ Agent‐based modeling has been widely applied in healthcare research, facilitating simulations of complex scenarios, including infection spread, intervention efficacy, and resource allocation. ¹⁸ , ¹⁹ In this study, the number of subjects remained constant throughout the simulation, with their attributes unchanged.

The simulation (Figure 1 and Appendix Method A4) was conducted under the following scenarios (Table 2): (A) the current testing flow, (B) implementing BBM as a triage tool by specialists, (C) implementing BBM as a confirmatory diagnostic tool by specialists, and (D) implementing BBM for triage by GPs and as a confirmatory diagnostic tool by specialists. In Scenario D, two BBM tests with different specificities were assumed: 85% and 90% for triage and confirmatory diagnosis, respectively. For this simulation, we assumed that the first specialist visit would be the bottleneck of the testing process, based on the current clinical context and available data. Theoretically, when a bottleneck exists at the initial stage of a process, a waiting queue forms at that stage and should not form at subsequent stages. Therefore, we did not explicitly model the subsequent processes, including the capacities for PET or CSF testing. The simulation began on the launch date of lecanemab in Japan (December 20, 2023), defined as the index date, and continued for 1000 days from that point.

FIGURE 1.

Outline of the simulation. Illustrates the simulation process for modeling diagnostic pathways for Alzheimer's disease. It depicts individuals aged 50 years and older with cognitive complaints, their progression through diagnostic steps, including WTP thresholds, and eventual classification as treatment‐eligible based on diagnostic outcomes such as PET, CSF, or BBM tests. Detailed information about the simulation is presented in Appendix Method A4. Note: ^*For example, the intangible cost is JPY 0 for a wait time of one month and JPY 2141 for a wait time of 3.8 months. In Scenario C, costs excluding wait time and travel amount to JPY 12,439 for individuals with an out‐of‐pocket medical cost burden of 10%. Assuming an individual whose WTP is JPY 13,000 and travel costs are JPY 0, the individual is willing to take the test when the wait time is one month (total costs: JPY 12,439 < WTP: JPY 13,000) but is not willing to take the test when the wait time is 3.8 months (total costs: JPY 14,580 > WTP: JPY 13,000). ^†The number of individuals with accepted appointments is 4.5 times that of those initiating diagnostic testing per specialist (Appendix Methods A4B #14). BBM, blood biomarker test; CSF, cerebrospinal fluid test; GP, general practitioner; JPY, Japanese Yen; MRI, magnetic resonance imaging test; PET, positron emission tomography test; WTP, willingness‐to‐pay.

TABLE 2.

Testing conditions by general practitioners (A) and specialists (B) for each scenario in the simulation.

Description	Scenario A	Scenario B	Scenario C	Scenario D
A.
Items	Consultation Cognitive assessment (Se: 82%, Sp: 73%)	Consultation Cognitive assessment (Se: 82%, Sp: 73%)	Consultation Cognitive assessment (Se: 82%, Sp: 73%)	Consultation Cognitive assessment (Se: 82%, Sp: 73%) BBM triage (Se: 90%, Sp: 85%)
B.
Monthly number of individuals who undergo the testing per specialist a	43.9	50.5	68.7	64.6
Items
1st visit	Cognitive assessment (Se: 82%, Sp: 73%) General blood test	Cognitive assessment (Se: 82%, Sp: 73%) General blood test	Cognitive assessment (Se: 82%, Sp: 73%) General blood test	Cognitive assessment (Se: 82%, Sp: 73%) General blood test
2nd visit	MRI	MRI BBM triage (Se: 90%, Sp: 85%)	MRI BBM confirmatory (Se: 90%, Sp: 90%)	MRI BBM confirmatory (Se: 90%, Sp: 90%)
3rd visit	Diagnostic disclosure	Diagnostic disclosure	Diagnostic disclosure	Diagnostic disclosure
4th visit	PET/CSF b (Se: 90%, Sp: 90%)	PET/CSF b (Se: 90%, Sp: 90%)	N/A	N/A

Note: Scenario A: the current testing flow, Scenario B: implementing BBM as a triage tool by specialists, Scenario C: implementing BBM as a confirmatory diagnostic tool by specialists, Scenario D: implementing BBM for triage by GPs and as a confirmatory diagnostic tool by specialists.

Abbreviations: BBM, blood biomarker test; CSF, cerebrospinal fluid test; MRI, magnetic resonance imaging test; N/A, not available; PET, positron emission tomography test; Se, sensitivity; Sp, specificity.

^a Monthly number of individuals who undergo testing per specialist for Scenario A was calculated by dividing the average number of patients per month by the average total number of dementia specialists and physicians with relevant clinical experience per facility, based on a national survey. ²⁰ . We assumed that patients who tested positive would continue to visit until their fifth visit, when confirmatory test results would be provided for Scenario A. For Scenarios B–D, the numbers were calculated based on the figure for Scenario A according to procedures in each scenario. Calculation details are presented in Appendix Method A4B (#10–#13).

^b Proportion of patients undergoing PET versus CSF test (PET:CSF) was assumed as 54.2:45.8 based on the MDV database (December 2023–May 2024) analysis.

The outputs generated for each scenario included the following: (1) Wait time: Wait time was defined as the period from when an individual makes an appointment with a specialist to the day of their first specialist visit. The mean wait time per municipality (in days) was projected and visualized graphically. Additionally, the maximum mean wait time during the simulation period and the number of days from the index date to reach this peak wait time were estimated. (2) Lecanemab treatment‐eligible patients: Treatment‐eligible patients were defined as individuals diagnosed as amyloid‐β positive through a confirmatory test. The number of such patients was calculated by multiplying the number of those who underwent the confirmatory test by the proportion of individuals who were amyloid‐β positive in each population. The cumulative number of these patients was presented in a graph, and annual estimates were provided for 2023–2025. (3) Medical costs associated with testing: Yearly medical costs associated with testing across Japan (2023–2025) were estimated as: (number of individuals who underwent the test) × (medical costs for the test, including procedures associated with the test). Medical costs represented total healthcare expenditures, not just out‐of‐pocket expenses (Appendix Method A5). Costs for GP consultations (excluding BBM tests in Scenario D), which were excluded to assess the demand, were also calculated separately to provide a comprehensive cost estimate.

3. RESULTS

3.1. WTP and intangible costs

The total number of respondents was 3302 (Table 3). The distribution of respondents by age, gender, and living area closely matched that of the Japanese population (Appendix Table A1). The percentages of respondents with self‐reported or informant‐reported cognitive complaints are presented in Appendix Table A2.

TABLE 3.

Willingness‐to‐pay for each assumed condition.

Statistic	CN	MCI	Mild DM
Respondents (N)	1101	1117	1084
WTP (JPY)
Mean	16,507	21,228	22,810
SD	49,209	77,157	57,209
Min	0	0	0
Max	1,000,000	1,000,000	1,000,000
25%tile	3000	3000	3000
50%tile	5000	5000	5000
75%tile	10,000	10,000	10,000

Note: US dollar 1.00 corresponded to approximately JPY 153.72 as of November 2024.

Abbreviations: CN, cognitive normal; DM, dementia; JPY, Japanese Yen; Max, maximum; MCI, mild cognitive impairment; Min, minimum; N, number; SD, standard deviation; WTP, willingness‐to‐pay.

Mean WTP varied depending on the assumed severity of the cognitive condition. Among the conditions, mild dementia had the highest mean WTP, at JPY 22,810 (standard deviation: JPY 57,209) (Table 3). However, the minimum, maximum, and quartile values remained consistent across all conditions.

The intangible cost for wait time before treatment initiation was notably high, estimated at JPY 3120 and JPY 6179 for 6 months and 1 year, respectively, compared to 1 month (Appendix Table A3). Excluding wait time, intangible costs were estimated at JPY 2335, JPY 10,104, and JPY 5295 for PET, CSF, and BBM, respectively (Appendix Table A4). For BBM, intangible costs were attributed solely to the absence of a physician's explanation of the test results using images. The summary of costs, along with the breakdowns of tangible and intangible costs for each scenario, is presented in Tables 4A, 4B, and 4C, respectively.

TABLE 4A.

Testing costs for each scenario: summary.

		Intangible costs (JPY)
Scenario	Tangible costs a (JPY)	Total excluding wait time for treatment initiation	Wait time for treatment initiation
A	17,427–52,282 b	5893 b	0–3234 c
B	19,529–58,588 b	5893 b , d	0–2604 c
C	5003–15,009	5295	0–3814 c
D	5823–17,469	5295	0–2141 c

Note: US dollar 1.00 corresponded to approximately JPY 153.72 as of November 2024.

Abbreviations: BBM, blood biomarker; CSF, cerebrospinal fluid; JPY, Japanese Yen; PET, positron emission tomography.

^a Shown with a range of 10% to 30% coinsurance

^b Weighted average of PET and CSF costs, assuming 54.2% and 45.8% of individuals undergo PET and CSF, respectively, with the CSF cost being a weighted average of 20% outpatient and 80% inpatient cases (see B for breakdown of tangible costs and C for intangible costs, for PET, CSF outpatient, and CSF inpatient)

^c Maximum costs were calculated based on the maximum wait time from the dynamic simulation.

^d Additional intangible cost for BBM triage was set to 0 because the intangible cost for BBM only includes costs associated with the absence of a physician's explanation of test results using images. For PET, an explanation using images is provided, and for CSF, this cost is included in the intangible costs.

TABLE 4B.

Breakdown of tangible costs.

	Tangible costs (JPY) a
	Scenario A			Scenario B
Item	PET	CSF (OP)	CSF (IP)	PET	CSF (OP)	CSF (IP)	Scenario C	Scenario D
By GPs
BBM test (GPs)	‐	‐	‐	‐	‐	‐	‐	8200
By specialists
1st visit
First visit fees	2880	2880	2880	2880	2880	2880	2880	2880
Cognitive function test	800	800	800	800	800	800	800	800
General blood test	2720	2720	2720	2720	2720	2720	2720	2720
2nd visit
Subsequent visit fees	730	730	730	730	730	730	730	730
MRI test	19,000	19,000	19,000	19,000	19,000	19,000	19,000	19,000
BBM test (by specialists)	‐	‐	‐	21,020	21,020	21,020	21,020	21,020
3rd visit
First visit fees	2880	2880	2880	2880	2880	2880	2880	2880
4th visit
First visit fees	2880	2880		2880	2880
Inpatient fees			13,460			13,460
PET test	239,430			239,430
CSF test		15,760	15,760		15,760	15,760
5th visit
First visit fees			2880			2880
Subsequent visit fees	730	730		730	730

Note: Costs, including medical costs for GP consultation, are presented in Appendix Method A5. US dollar 1.00 corresponded to approximately JPY 153.72 as of November 2024.

Abbreviations: BBM, blood biomarker test; CSF, cerebrospinal fluid test; GP, general practitioner; IP, inpatient; JPY, Japanese Yen; MRI. magnetic resonance imaging; OP, outpatient; PET, positron emission tomography test.

^a Shown as the full amount (100% of costs).

TABLE 4C.

Breakdown of intangible costs excluding wait time.

	Intangible cost (JPY)
	Scenarios A and B			Scenarios C and D
Item	PET	CSF (OP)	CSF (IP)	BBM
Test sensitivity	0	0	0	0
Test duration (time commitment)	335	335	2949	0
Radiation exposure	2000	0	0	0
Physician's explanation of test results using images	0	5295	5295	5295
Side effects (headache lasting 3 days).	0	2383	2383	0

Note: Detailed information, including the level of each attribute for each test method, is presented in Appendix Table A4. US dollar 1.00 corresponded to approximately JPY 153.72 as of November 2024.

Abbreviations: BBM, blood biomarker test; CSF, cerebrospinal fluid test; GP, general practitioner; IP, inpatient; JPY, Japanese Yen; OP, outpatient; PET, positron emission tomography test.

3.2. Wait times

Among individuals whose WTP exceeded tangible costs (excluding travel costs), the estimated number of individuals diagnosed as positive by a GP based on cognitive assessment was as follows: 980,218 for Scenario A, 954,268 for Scenario B, 1,778,373 for Scenario C, and 1,777,134 for Scenario D. In Scenario D, among these individuals, 1,068,413 were diagnosed as positive based on BBM triage. These individuals were included in the simulation to project wait times for specialist visits for diagnostic testing.

In Scenario A, which assumes the current testing flow, the maximum mean wait time was projected to be 6.4 months, peaking at 113 days after the index date (Figure 2A). In Scenarios B and D, where the BBM test was included for triage, the maximum mean wait times decreased to 5.4 months (94 days after the index date) and 4.7 months (85 days after the index date), respectively, with Scenario D having the shortest wait time among all scenarios. In Scenario C, where the BBM test was included only as a confirmatory test, the maximum mean wait time increased to 7.8 months (131 days after the index date).

FIGURE 2.

Wait times for diagnostic testing (A), cumulative numbers of treatment‐eligible patients (B), and annual medical costs throughout the testing process (C) based on the simulation. (A) Presents the mean estimated wait times per municipality for diagnostic testing under four scenarios modeled in the simulation. The number of subjects was assumed to remain constant throughout the simulation, with their attributes remaining unchanged. (B) Shows the cumulative number of patients eligible for lecanemab treatment, categorized by four testing scenarios, over the simulation period from December 2023 through 2025. For 2023, the annual number included only December 20–31, reflecting the index date. The number of subjects was assumed to remain constant throughout the simulation, with their attributes remaining unchanged. (C) Illustrates the estimated annual medical costs throughout the diagnostic testing process, separated by medical costs associated with testing and GP consultation, under four scenarios. For 2023, the annual number included only December 20–31, reflecting the index date. The number of subjects was assumed to remain constant throughout the simulation, with their attributes remaining unchanged. GP, general practitioner; JPY, Japanese Yen. US dollar 1.00 corresponded to approximately JPY 153.72 as of November 2024.

3.3. Number of treatment‐eligible patients

The estimated number of lecanemab treatment‐eligible patients by year was as follows: 14,554 in 2023 (from the launch date, December 20), 183,376 in 2024, and 32,301 in 2025 for Scenario A. The number was slightly higher in 2023 (15,075) but consistently lower in 2024 (171,007) and 2025 (25,274) for Scenario B than for Scenario A. In Scenario D, the number of treatment‐eligible patients increased rapidly in 2023, reaching 34,275, the highest among all scenarios for that year, followed by a slower rate of increase in subsequent years. Scenario C had a lower number of eligible patients than Scenario D in 2023 (23,331) but higher in 2024 (326,493 vs. 339,294) and 2025 (34,062 vs. 68,732), resulting in the largest cumulative number of treatment‐eligible patients among the scenarios in 2024 and subsequent years (Figure 2B).

3.4. Medical costs associated with testing

Medical costs associated with testing per individual who completed the full testing process were JPY 174,273, JPY 195,293, JPY 50,030, and JPY 58,230 for Scenarios A, B, C, and D, respectively (Appendix Method A5). Total costs across the country, accounting for the number of individuals who underwent each test, were highest for Scenario A in all years: JPY 4.44 billion in 2023, JPY 55.89 billion in 2024, and JPY 9.84 billion in 2025, followed by Scenario B (Figure 2C). In 2023 and 2024, costs for Scenario C were lower than those for Scenario D. However, in 2025, costs for Scenario D were estimated to be the lowest among all scenarios, followed by Scenario C. Despite higher costs for GP consultations and a larger number of patients undergoing testing in Scenarios C and D than in Scenario A, total costs, including both testing‐related expenses and additional GP consultation costs, were projected to remain the highest for Scenario A in 2024 and 2025.

4. DISCUSSION

In this study, we projected the diagnostic testing landscape for lecanemab treatment in Japan under various testing workflows. The simulation targeted the entire Japanese population instead of only the patient population for lecanemab. The projected maximum mean wait time was approximately 6 months under the current testing flow but decreased with the implementation of BBM tests for triage. The number of treatment‐eligible patients increased considerably when BBM was used for confirmatory testing. We utilized a dynamic simulation model that integrated demand for testing while accounting for interactions between wait times and demand fluctuations. Demand was modeled based on WTP and costs derived from an online survey. Previous studies have highlighted challenges in the current healthcare system related to diagnostic testing during the introduction of DMTs for AD ⁶ , ²¹ , ²² , ²³ and have explored potential improvements with BBM implementation. ²⁴ , ²⁵ However, these studies have not incorporated demand for testing into their simulations. The National Institute for Health and Care Excellence guidance indicates that the introduction of lecanemab may significantly increase the number of patients seeking diagnosis, placing considerable strain on diagnostic and testing resources, such as PET scans. This strain could impact treatment capacity for other diseases. ²⁶ Therefore, it is essential to project the number of individuals seeking treatment and the corresponding resource requirements to enable effective planning and equitable care. We assume that demand is influenced by both tangible and intangible costs, leading to variations across test types and conditions, including wait times. By incorporating these factors into the simulation, we aimed to provide more reasonable demand estimates for each scenario. Given that predicted demand significantly affects projections such as wait times, integrating demand into our simulation offers valuable insights into healthcare system dynamics.

The wait time projected in this study is shorter than that reported in a 2019 simulation study. ⁶ The previous study assumed that individuals aged 50 years and older without a diagnosis of MCI or dementia were eligible for annual cognitive screening, with assumed rates of consent to screening and seeking specialist evaluation. This resulted in approximately 3.9 million individuals seeking specialist evaluations, with a maximum projected wait time of approximately 15 months. In contrast, in the scenario without the BBM test, we estimated approximately one million individuals seeking diagnostic testing and being diagnosed positive by a GP based on a simulation incorporating WTP. Differences in the number of individuals seeking specialist evaluations, along with differing assumptions about medical resources, appear to contribute significantly to the variation in projected wait times between studies.

We evaluated the validity of our model in terms of projected wait times and the inclusion of intangible costs. The longest mean wait time projected under the scenario assuming the current testing flow was 6.4 months. Although publicly available data on wait times for diagnostic testing for lecanemab treatment in Japan are limited, interviews with specialists indicated that some medical institutions currently have wait times of several months. Thus, our projected wait times are considered reasonable. We assumed that long wait times or distant testing locations could deter individuals from undergoing testing. Therefore, including these deterrents as intangible costs and comparing total costs with WTP to determine whether individuals would undergo testing is considered appropriate, as it reflects real‐world decision‐making processes. We used an agent‐based model to appropriately reflect how individual decisions are influenced by changes in wait times and how these decisions, in turn, affect overall demand. This approach enables the dynamic simulation of interactions between individual behavior and system‐level factors such as wait time.

The beneficial effects of implementing BBM tests were primarily attributed to their lower costs compared to currently utilized diagnostic modalities. These lower costs arose from several factors. First, the cost of a BBM test was approximately one‐tenth that of a PET test, leading to a substantial cost reduction. Second, we assumed that BBM confirmatory tests could be performed at the second specialist visit alongside the MRI, thereby eliminating the need for two additional specialist visits and further reducing costs. This also reduced the total number of specialist visits, increasing specialist capacity (measured as the “monthly number of individuals who undergo the testing per specialist” in Table 2B) and decreasing wait times associated with intangible costs. Third, the BBM test requires less time than CSF and is not associated with the risk of headache as a side effect, leading to lower intangible costs than the CSF test. These cost reductions led to a greater number of individuals seeking diagnostic testing, thereby increasing the estimated number of treatment‐eligible patients. Notably, when BBM triage was not implemented, the increase in the number of individuals seeking diagnostic testing canceled the effect on decreasing wait times; instead, the wait times increased. Despite this increase in testing demand, total medical costs for diagnostic testing across the country are estimated to decrease under scenarios incorporating BBM. Even when including the costs of GP consultations, total costs are not expected to exceed those of the current testing flow scenario, except in Scenario D in 2023. These findings suggest that implementing BBM tests may expand access to treatment without increasing national healthcare expenditures. However, these costs represent only a small fraction of the total expense associated with lecanemab treatment, which is currently estimated at approximately JPY 3 million per person per year. Therefore, the financial implications of increased treatment eligibility must be carefully considered when implementing BBM tests. In such cases, the benefits of treatment should also be evaluated to ensure that expanded access is balanced against overall healthcare costs. We should also consider some challenges in implementing BBM in clinical practice, including the absence of consistent recommendations for cognitive assessment in primary care—which poses a major barrier to the adoption of BBM testing—the need to educate healthcare professionals on interpreting BBM results across diverse populations, and limited evidence supporting BBM interpretation in different groups. ⁹

Implementing a BBM triage test is projected to shorten wait times but not increase the number of treatment‐eligible patients. When BBM is used only as a triage test in addition to the current testing flow, wait times are projected to decrease. However, the number of treatment‐eligible patients is estimated to be slightly higher for a limited period before decreasing because of the higher costs than those associated with the current testing flow. When BBM is used for confirmatory testing, incorporating BBM for triage as well is projected to shorten wait times and lead to an earlier increase in treatment‐eligible patients. However, the number of treatment‐eligible patients when BBM is used for confirmatory testing alone is eventually estimated to surpass that when BBM is used for both triage and confirmation. These findings indicate that testing flows incorporating BBM triage by GPs could reduce wait times and improve access to treatment in settings with high demand for testing and limited medical resources. The projected wait time under the current testing flow—6 months—appears relatively long, considering that lecanemab is indicated for patients with MCI or mild dementia but becomes unsuitable as the condition progresses. Moreover, its treatment effect includes delaying the progression to dementia by approximately 2–3 years in patients with MCI and delaying the progression to moderate dementia by 1.5 years in those with mild dementia. ²⁷ During extended wait times, implementing BBM triage could provide significant value by allowing patients to gain earlier access to treatment, potentially maximizing the therapeutic benefits of lecanemab. Notably, we assumed that the decision to undergo triage was determined by the WTP for testing, based on the probability of being eligible for treatment before triage and the costs associated with both triage and confirmatory tests. If WTP were assessed for individuals diagnosed as positive through triage, the probability of being eligible for treatment would increase, likely resulting in a higher WTP than that before triage. Therefore, if the decision to undergo triage is made without considering the costs of confirmatory diagnosis and more individuals undergo triage, the number of treatment‐eligible patients may increase.

While this study provides valuable insights, several considerations should be noted when interpreting the results. The outcomes are based on simulations and represent projections rather than actual conditions. These results are influenced by the parameters and models used. Additionally, relocations may alter individuals’ proximity to medical institutions, which the simulation does not account for. The study assumes the current state of healthcare resources but does not fully reflect the OUG. While facility locations capable of conducting tests are based on actual data, the distribution of specialists and other essential diagnostic resources—including the number of available slots for PET or CSF tests—was not explicitly modeled, even though these are important factors in the testing process. As described in the Section 1, we deliberately made this decision because, with the first specialist visit being the bottleneck of the process, the capacities for these tests were not expected to be bottlenecks in practice.

The Web survey used to estimate WTP and intangible costs also has limitations. First, variability in respondents’ comprehension of the questions and sincerity in their answers could affect results. Second, the respondents were likely more familiar with the Internet, as they had registered for online surveys, which may have influenced their responses. Third, although symptoms of MCI and AD and the expected effects of lecanemab treatment were explained, the level of public awareness about lecanemab during the survey may have influenced responses. Fourth, WTP surveys can entail certain issues, including the possibility that the WTP reported by survey respondents may differ from that of individuals actually facing the need for treatment. Additionally, it stated that WTP does not necessarily align with revealed WTP. Despite these limitations, incorporating the WTP derived from the survey may have enabled us to better account for the demand of the target population in the simulation. After implementing BBM, further confirmatory studies will be needed to validate the actual demand in real‐world settings.

In conclusion, the dynamic simulation demonstrates that the number of treatment‐eligible patients is significantly influenced by costs, encompassing both tangible and intangible components. The findings suggest that utilizing BBM tests for confirmation instead of PET or CSF can increase the number of treatment‐eligible patients, primarily because of the reduced costs associated with testing. Additionally, implementing BBM for triage, particularly in GP settings, shows potential for reducing wait times.

CONFLICT OF INTEREST STATEMENT

A.I. reports receiving research funding from Eisai Co., Ltd.; and advisory fees from Eli Lilly Japan K.K., Novartis Pharma K.K., Janssen Pharmaceutical K.K., Ono Pharmaceutical Co., Ltd., and Chugai Pharmaceutical Co., Ltd. N.K. reports receiving a grant from the Japan Society for the Promotion of Science, and honoraria from Sumitomo Pharma Co., Ltd., Daiichi Sankyo, Eisai Co., Ltd., Otsuka Pharmaceutical Co., Ltd., PDRadiopharma Inc., Kyowa Kirin Co., Ltd., Kowa Company, Ltd., Eli Lilly, and Chugai Pharmaceutical Co., Ltd. T.I., K.S., C.K., M.T., Y.S., and M.A. are employees of Eisai Co., Ltd. A.C., T.T., and K.I. are employees of Milliman, Inc., which provides services to multiple pharmaceutical and medical device companies. T.A. and E.M. report no conflicts of interest. Author disclosures are available in the supporting information.

CONSENT STATEMENT

This study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects. The study protocol received approval from the Eisai Research Ethics Committee (Approval No: REP‐2023‐0908). Additionally, the protocol was pre‐registered with the University Hospital Medical Information Network Clinical Trials Registry (ID: UMIN000053721). Informed consent was obtained from all Web survey participants before they began the questionnaire.

Supporting information

Supporting Information

Supporting Information

Igarashi A, Kimura N, Ataka T, et al. Simulation of Alzheimer's diagnostic flows with blood biomarker test options in Japan. Alzheimer's Dement. 2025;11:e70157. 10.1002/trc2.70157