Digital Accessible Remote Olfactory Mediated Health Assessments forPreclinical AD

Abstract Over 6.9 million Americans are currently affected by Alzheimer's disease (AD) with an economic burden estimated at > $345 billion/year in 2023 that is projected to double over the next several decades. Alzheimer's disease (AD) pathology emerges up to two decades prior to the onset of memory symptoms. The desire to test disease modifying therapies for dementias at preclinical stages of the disease is growing. The olfactory bulb and the entorhinal cortex are early sites of tau pathology, and different tests of odor identification have consistently shown a deficit in the MCI and dementia stages of AD. However, these olfactory tests were not designed to be self-administered, increasing the burden on research staff and participants. In Phase II, we validated the AROMHA Brain Health Test, a web-based app paired with smell cards using proprietary naturalistic odors for at-home self administration of odor identification, odor memory, and odor discrimination tests. We built on our POEM algorithm that incorporates results from all 3 tests by discovering a second novel outcome that predicts cognitive impairment using machine learning. This new outcome, OPID18noguess, associates with hippocampal volume loss in pre-MCI and MCI patients. The self-administered, validated AROMHA Brain Health Test has potential to scale this promising at-home screen for preclinical AD to generate large datasets for machine learning, and make recruitment more efficient for preclinical AD trials. Building from these promising results in Phase II, in Phase IIB, we propose to enhance the multilingual AROMHA Brain Health Test by adding accounts and create the foundation for returning results in real time (Aim 1). In Aim 2, we will collect longitudinal olfactory data from the participants recruited in Phase II from the clinical cohort of Mass. ADRC, which includes deeply phenotyped patients who are cognitively normal (CN), with subjective cognitive complaints (SCC) or with MCI. In addition, we will recruit deeply phenotyped CN, SCC, MCI participants from two additional ADRCs with diverse populations. We will model olfactory performance to predict the presence of blood-based, imaging, and CSF biomarkers in pre-MCI patients using both conventional and data-driven machine learning techniques to determine sensitivity and specificity of olfactory outcomes related to biomarkers of aging and early AD. In Aim 3, we will build on our testing in 25 different states and Puerto Rico during Aim 3, by specifically recruiting from mobile medical units in underserved populations in Chicago and in 55+ communities. Successful completion of these aims will contribute to a data package that includes longitudinal smell testing in multiple patient cohorts with a deep phenotyping of AD biomarkers for consideration by the FDA. Used alone or together with other non-invasive measures, a remote test to identify cognitively normal people harboring olfactory dysfunction - and possible AD pathology - who are at risk of developing AD may be an important tool to efficiently recruit participants for clinical trials testing therapies for preclinical AD and follow their disease progression during the preclinical phase of the disease.