Disease Progression and DMT Effect

The natural history of AD progression for patients receiving usual care was modeled using AD ACE based on disease equations developed from longitudinal patient-level ADNI data for early AD [5] and published AHEAD equations for more severe stages of AD [8, 9]. Clinical Dementia Rating scale Sum of Boxes (CDR-SB) thresholds were used to determine patients’ disease severity at baseline and over time in AD ACE (i.e., MCI due to AD < 4.5, mild AD ≥ 4.5 to < 9.5, moderate AD ≥ 9.5 to < 16, and severe AD ≥ 16) [13].

The treatment effect was modeled on the basis of the key assumption that the effect of a DMT on the clinical outcomes is correlated with the amyloid PET level as a surrogate endpoint [14, 15]. In AD ACE the relationships between biomarkers of disease and clinical outcomes are based on correlations mainly observed in the ADNI data, and disease equations are evaluated repeatedly at subsequent time intervals every 6 months to estimate the AD disease trajectory of patients. An anti-amyloid DMT can be potentially modeled in AD ACE by imposing effects on estimated amyloid PET standardized uptake value ratio (SUVr) outcomes of a simulated patient. In this study, a calibration process was applied to adjust the predicted measures of amyloid PET SUVr at each time interval to slow down the progression of disease and achieve 1-, 3-, and 5-year average delay in onset of AD. Amyloid PET is a predictor in all AD ACE disease equations; therefore, any calibrated reduction in amyloid PET SUVr at a given time interval impacts the prediction of amyloid PET SUVr and other modeled AD biomarkers and scales at later time intervals and consequently the time to onset of AD. Calibration is the process of determining or adjusting parameter values in a model by constraining model output to replicate empirical data within an acceptable range. In this analysis, calibration was performed by comparing model output from different reductions in amyloid PET SUVr in subsequent time intervals to identify the parameter sets that best correspond to desired average delay in onset of AD.

The calibration process was focused on patients with amyloid-β-positive (Aβ+) MCI who were eligible to receive the hypothetical DMT. The AD ACE model was run to calibrate the effects of a hypothetical amyloid-targeting treatment to achieve 1-, 3-, and 5-year delays in onset of AD. A lifetime simulation of 2000 sampled ADNI patients was used during the treatment effect calibration process where treatment discontinuation was not allowed. The ADNI sample population included 526 Aβ+ patients ages 60 or older who received a clinical diagnosis of MCI, which was defined by a score > 24 on the Mini-Mental State Examinations (MMSE) and a global Clinical Dementia Rating scale of 0.5 at baseline. The accumulation of amyloid-β in the brain was measured by PET imaging with ¹⁸F-AV-45 (florbetapir), and patients with a baseline mean cortical standardized uptake value ratio (mcSUVR) ≥ 1.1 were considered as Aβ+. To inform the burden-of-illness analysis, the analysis was focused on patients with MCI. Scenarios were run separately for usual care and each of the hypothetical DMT effects and uptake scenarios by sampling 2000 profiles from a cohort of 826 ADNI patients ages 60 or older with MCI. Under the 100% uptake scenario, the hypothetical DMT was initiated immediately for patients who were Aβ+ at baseline and with a delay for those who became Aβ+ by crossing the ¹⁸F-AV-45 cutoff value of mcSUVR ≥ 1.1 or progressed to mild AD defined as a CDR-SB ≥ 4.5 during the simulation runs. The DMT was terminated when a patient’s AD reached moderate severity, as measured by a CDR-SB score ≥ 9.5. Other types of treatment discontinuation were not considered in this analysis (e.g., patient decision, adverse events). This economic analysis assumed that the hypothetical DMT will become available in 2022 for eligible patients, hence patients could not access treatment prior to 2022 and a treatment time rule was applied to incident patients entering the model from 2000 to 2021. Each DMT scenario was run 22 times and the timing to DMT initiation lagged from zero to 21 years. For example, patients with incident MCI in 2005 had a minimum time lag of 17 years before treatment initiation, whereas the time lag was at least 7 years for the incident patients in 2015.