Model Overview

A modeling framework was developed to estimate the economic burden of AD in the USA assuming that an amyloid-targeting treatment will become available in 2022; this hypothetical DMT was compared to the usual care under different scenarios based on delay in the onset of AD (1, 3, and 5 years) and DMT uptake among eligible patients (25%, 50%, and 100%). The model consisted of three modules—epidemiology, disease progression, and burden of illness (Fig. 1)—and assessed the annual burden of illness of AD from 2020 to 2050 (aka analysis interval) under different scenarios. The epidemiology module used the annual incidence rates of MCI due to AD by age group with the projected US population estimates by age group over 60 years old and reported on the total number of new subjects with MCI due to AD each year from 2000 to 2050 (Fig. 1a). The disease progression module leveraged the AD Archimedes Condition Event (AD ACE) disease simulator to track disease progression of incident subjects with MCI due to AD over their lifetime (Fig. 1b).

Model structure. AD  Alzheimer’s disease, MCI  mild cognitive impairment

The AD ACE is a patient-level simulation model that predicts the natural history of individuals from a preclinical disease state through the severe AD stage and estimates potential effects of a DMT on disease progression. The model simulates disease progression based on changes in the underlying AD biomarkers (e.g., measures of Aβ and tau levels) and their connections to clinical presentation of AD, which are measured by various patient-level scales of cognition, behavior, function, and dependence. In early AD, disease progression is measured by interconnected predictive equations derived from longitudinal assessments of clinical and biomarker data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) [5]. Biomarkers extracted from the ADNI longitudinal data set were cerebrospinal fluid (CSF) proteins (Aβ_1–42 and total-tau) linked to abnormal brain deposits, fluorodeoxyglucose (FDG)–positron emission tomography (PET) linked to reduced brain cell metabolic activity, and one magnetic resonance imaging (MRI) measurement of hippocampal volume linked to brain shrinkage. Full details on the AD ACE model structure, data inputs, and predictive equations have been previously published [6, 7].

As patients progress to more severe stages of AD that the ADNI study does not effectively represent, the AD ACE switches to Assessment of Health Economics in Alzheimer’s Disease II (AHEAD) equations for cognition and behavioral scales to make the model more representative and accurate across all stages of AD [8, 9]. Additionally, the model captured transitions to/between community and institutional care settings as patients progressed to more severe stages of AD. Full details on the AD ACE model structure and equations have been previously published [7, 10–12]. The AD ACE was recently used to study the potential long-term health and economic outcomes of lecanemab in patients with early AD.

The disease progression module reported the annual proportions of patients with MCI due to AD, and with mild, moderate, or severe AD; proportions of deceased patients; proportions of patients in community or residential care; as well as the mean annual total costs of community or residential care per patient over lifetime, including direct medical and non-medical costs for healthcare resource use and indirect costs for caregiving. The outcomes from the epidemiology module (i.e., annual MCI incidence) and disease progression module (i.e., annual AD disease stage, mortality, and costs) were used to estimate the total number of Americans with MCI due to AD and mild, moderate, or severe AD in community or residential care settings and their associated costs each year, and subsequently compute the clinical and economic burden of AD during the analysis interval in the burden of illness module (Fig. 1c). A pre-analysis interval from 2000 to 2019 was incorporated in the burden of illness module to gradually accumulate the number of alive patients as they progress in their disease to achieve an accurate estimate of patients with MCI and AD during the analysis interval. The pre-analysis interval was initiated in 2000 as almost all patients with incident MCI from this year were dead before the start of the analysis interval and with minimal impact on burden of illness analyses. The model-predicted AD prevalence in 2020 closely matched the estimates by the Alzheimer’s Association (i.e., 6.1 million individuals in 2020) [1].