Deep phenotyping of Alzheimer’s disease leveraging electronic medical records identifies sex-specific clinical associations

Abstract

Alzheimer’s Disease (AD) is a neurodegenerative disorder that is still not fully understood. Sex modifies AD vulnerability, but the reasons for this are largely unknown. We utilize two independent electronic medical record (EMR) systems across 44,288 patients to perform deep clinical phenotyping and network analysis to gain insight into clinical characteristics and sex-specific clinical associations in AD. Embeddings and network representation of patient diagnoses demonstrate greater comorbidity interactions in AD in comparison to matched controls. Enrichment analysis identifies multiple known and new diagnostic, medication, and lab result associations across the whole cohort and in a sex-stratified analysis. With this data-driven method of phenotyping, we can represent AD complexity and generate hypotheses of clinical factors that can be followed-up for further diagnostic and predictive analyses, mechanistic understanding, or drug repurposing and therapeutic approaches.

Utilizing Electronic Medical Records Identifies Similarities and Differences in Alzheimer’s Disease Comorbidities Between Racialized Populations

“…First, we asked whether patients' phenotypic profiles differed by AD status, identified R&E, and UC location using low-dimensional embedding (UMAP). For AD status, significant differences were found between patients with AD and controls along both UMAP components at UCSF and in the validation cohort, consistent with Tang et al 7 This suggests that representing patients using ICD diagnoses mapped to phenotypes yield similar results to using ICD diagnoses directly, providing support for the use of phenotypes to investigate patient comorbidities. Next, while we found numerous differences between patients' phenotypic profiles based on identified R&E at UCSF, many of these differences were not corroborated in the validation cohort, with the exception that Black-identified patients' phenotypic profiles were found to be significantly different from Latine-and White-identified patients' phenotypic profiles along the first UMAP component.…”

“…This suggests that AD networks are relatively more connected, with many shared phenotypes amongst patients. Degree and neighborhood connectivity were also found to be higher for patients with AD relative to controls in Tang et al 7 For AD networks representing Asian-identified patients specifically, closeness centrality was higher, while average shortest path length and eccentricity were lower relative to matched control networks at UCSF and in the validation cohort; closeness centrality was also found to be higher for AD relative to controls in Tang et al This suggests that phenotypes in Asian-identified patients with AD are more closely connected to one another relative to controls. Finally, for AD networks representing Latine-identified patients at UCSF and in the UC-wide validation cohort, stress was found to be higher relative to matched control networks; stress was also found to be significantly different in Tang et al This suggests that phenotypes are relatively more connected to one another for Latine-identified patients with AD.…”

“…Patient cohorts were selected based on a similar method to Tang et al 7 , with some modifications. Patients with AD were identified from the UCSF OMOP-based EMR database in September and October 2021, which contained clinical data from over 5 million patients from January 1, 1982, to January 7, 2021.…”

“…This provides a wealth of data that can be utilized to characterize complex and heterogeneous diseases like AD from many individuals and institutions. Indeed, EMR data have been leveraged to perform sex-stratified deep phenotyping of AD, including the analysis of sex-stratified AD networks, as well as the exploration of comorbidities, medications, and lab values that may be differentially associated in individuals with AD depending on their identified sex 7 . In another study, EMR-derived disease networks in racialized populations revealed differences in disease trajectories, demonstrating how networkbased analyses can identify differences in health outcomes between racialized populations (we use the term 'racialized' to emphasize the process by which people are assigned racial categories; while some definitions of 'racialized' may exclude White-identified individuals, they are included in our usage) 8,9 .…”

Neuroimmune contributions to Alzheimer’s disease: a focus on human data

“…While men diagnosed with AD seem to progress to death quicker, women with increased tau pathology can exhibit greater cognitive resilience [ 202 , 203 ]. The underlying causes for sex differences in prevalence and vulnerability to AD are currently subject to a lively debate in the field [ 204 ], with an increasing amount of studies trying to delineate sex-specific clinical associations [ 205 ]. Among women, changes in hormonal exposure—such as early sudden, surgical menopause—are associated with more rapid cognitive decline and a greater burden of neuritic amyloid plaque on autopsy [ 206 ].…”