There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Ron Handels |
| Institution: | Maastricht University |
| Department: | Psychiatry and Neuropsychology |
| Country: | |
| Proposed Analysis: | INTRODUCTION Health-economic models are important to estimate the impact of new interventions in Alzheimer's disease (AD). They are used both to estimate patient-and society-relevant outcomes, as well as to extrapolate short-term trial outcomes to lifetime impact. Such models aggregate data from different sources into a single mathematical description of AD disease progression (in terms of pathology and symptoms), quality of life, care use and mortality. Longitudinal data on the natural course of AD is an important input for health-economics models. A number of models have been developed with a variety of methods regarding natural progression and implementation of treatment effect. Studying these models is important to understand the implications of different methods on model results. The International PharmacoEconomic Collaboration on Alzheimer's Disease (IPECAD) has cross-compared models in a workshop in 2020 (Handels et al. 2022; accepted for publication in Alzheimer's & Dementia). We wish to replicate this workshop and focus on comparing methods for implementing treatment effect into decision models. For this we wish to use the outcomes of a hypothetical AD drug treatment randomized controlled trial. Our aim is twofold. First, we aim to further develop the IPECAD open-source decision-analytic model by analyzing and using the NACC data on AD natural disease progression, mainly in the predementia stage of AD, in terms of amyloid, tau, cognitive and functional performance. Second, we aim to describe the natural progression over a short-term period by mimicing/emulating data typically obtained from AD drug treatment randomized trials. METHODS For the first aim we will select those from ADNI with amyloid positivity in pre-clinical and prodromal stage of AD. We will fit a mixed GLM model to 4 outcomes: amyloid, tau, cognition, function. The model characteristics (link, family) will be determined on best fit. The exact outcomes (e.g., CSF based amyloid or PET-based amyloid, cognitive (composite) test) will be decided based on best model fit and availability of data. In addition individuals will be categorized as tau positive/negative; cognitive status (pre-clinical/prodromal) and functional status (mild/moderate/severe functional disability) and survival analysis will be performed to estimate transitions between these disease states, which aligns with the IPECAD open-source model. For the second aim we will select outcomes from ADNI that overlap to outcomes commonly obtained in AD drug trials. The following outcomes are anticipated: age, sex, ethnicity, education, use of AChEI, ApoE e4 carrier, CDR (item scores, sum of boxes and global score), MMSE (item scores and total score), ADAS-Cog14 (item scores and total score), ADCOMS iARDS, ADL (outcome to be determined), PET amyloid (outcome to be determined), CSF AB42, CSF phosphorylated tau, CSF total tau. We anticipate selecting individuals with age 50-85, amyloid positivity and objective memory impairment, and use their data from the first point they adhere to this onwards. Next, we will estimate the variance-covariance matrix on these outcomes longitudinally for 18 months. We will then use the variance-covariance matrix to generate synthetic data (mimic/emulate the data). Next, we will apply a hypothetical treatment effect to half of the emulated data. We will make this synthetic data available to health-economic modeling groups we invite for our workshop and ask them to implement the hypothetical treatment based on the data they receive (the ADNI data itself will never be shared with the model participants (about 10-20 researchers from academic or industry setting), only emulated data based on ADNI summary statistics). We will then discuss the difference in how workshop participants have implemented the treatment effect. |
| Additional Investigators | |
| Investigator's Name: | Linus Jonsson |
| Proposed Analysis: | same as current application |
| Investigator's Name: | Will Herring |
| Proposed Analysis: | same as current application |
| Investigator's Name: | Anders Gustavsson |
| Proposed Analysis: | same as current application |
| Investigator's Name: | Ashley Tate |
| Proposed Analysis: | same as current application |
| Investigator's Name: | Colin Green |
| Proposed Analysis: | same as current application |
| Investigator's Name: | Anders Skoldunger |
| Proposed Analysis: | same as current application |
| Investigator's Name: | Anders Wimo |
| Proposed Analysis: | same as current application |
| Investigator's Name: | Bengt Winblad |
| Proposed Analysis: | same as current application |
