# Forecasting response to pharmacological interventions in Alzheimer's and related dementias based on medical record and brain MRI characteristics: A precision medicine approach

> **NIH NIH R21** · JOHNS HOPKINS UNIVERSITY · 2020 · $450,313

## Abstract

Summary: Toward our long-term goal of developing a methodological framework for predicting response to
pharmacological treatments for Alzheimer’s disease (AD) and related dementias in clinical settings, we
propose a proof of concept study to investigate clinical and neuroanatomical features related to response to
treatment with two classes of medications that are currently in wide use: cholinesterase inhibitors (CEI) and
serotonin specific reuptake inhibitors (SSRI). These treatments are focused on alleviating symptoms and
extending patient autonomy. CEIs have shown efficacy in improving activities of daily living and are more cost-
effective than supportive care. SSRIs are a first choice for the pharmacological treatment of neuropsychiatric
symptoms (NPS) that affect ~97% of AD patients. However, response to CEI/SSRI varies between patients,
with only half responding to these drugs. Moreover, one-third of patients discontinue therapy because of
adverse effects. Given the AD pandemic, with 16M patients expected in the US by 2050, the targeted use of
CEI/SSRI to patients who will benefit from treatment is a public health priority. Although predictive markers to
identify those who will respond to CEI/SSRI have been highly anticipated, there, as yet, no established such
markers. Several clinical or genetic markers have been proposed as response predictors, but none have been
validated. Studying neuroanatomical features of the brain, which can be detected through MRI (= brain
phenotype), has succeeded in the identification of subgroups of AD related to clinical, cognitive, and
neuropsychiatric characteristics. However, little is known about whether brain phenotyping predicts response to
pharmacological interventions. Available results from pilot studies with small and highly selected samples point
to the necessity of studying larger, more inclusive cohorts. Since individualized prediction cannot be made from
a single feature, as each feature weakly correlates with outcomes, we hypothesize that patient stratification
that combines brain phenotype and clinical characteristics, will be highly accurate at individualized prediction.
In this study, we will utilize pre-existing resources to identify brain phenotypes and clinical features that might
predict response to CEI/SSRI treatment. First, data extracted from a large pool of electronic medical records
acquired through clinical practice will be used to identify brain phenotypes and clinical features related to a
response to CEI/SSRI. Second, our automated brain features extraction pipeline, which is robust to the great
heterogeneity in diseased brains will be used to quantify brain phenotypes. We propose Aim 1 to assess the
feasibility of brain phenotyping and Aim 2 to assess the feasibility of a supervised clustering approach based
on brain phenotype and clinical features, in predicting a response to CEI/SSRI. The proposed feasibility testing
based on analysis of data from a real world clinical care pa...

## Key facts

- **NIH application ID:** 10109413
- **Project number:** 1R21AG070404-01
- **Recipient organization:** JOHNS HOPKINS UNIVERSITY
- **Principal Investigator:** Kenichi Oishi
- **Activity code:** R21 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2020
- **Award amount:** $450,313
- **Award type:** 1
- **Project period:** 2020-09-15 → 2022-08-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10109413

## Citation

> US National Institutes of Health, RePORTER application 10109413, Forecasting response to pharmacological interventions in Alzheimer's and related dementias based on medical record and brain MRI characteristics: A precision medicine approach (1R21AG070404-01). Retrieved via AI Analytics 2026-05-22 from https://api.ai-analytics.org/grant/nih/10109413. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*