The apolipoprotein E gene (APOE) is the strongest genetic risk factor for late-onset Alzheimer’s disease (AD), yet the expression of APOE is not clearly understood. For example, it is unclear whether AD patients have elevated or decreased APOE expression or why the correlation levels of APOE RNA and the ApoE protein differ across studies. Likewise, APOE has a single CpG island (CGI) that overlaps with its 3’-exon, and this CGI’s effect is unknown. We previously reported that the APOE CGI is highly methylated in human postmortem brain (PMB) and that this methylation is altered in AD frontal lobe. In this study, we comprehensively characterized APOE RNA transcripts and correlated levels of RNA expression with DNA methylation levels across the APOE CGI. We discovered the presence of APOE circular RNA (circRNA) and found that circRNA and full-length mRNA each constitute approximately one third of the total APOE RNA, with truncated mRNAs likely constituting some of the missing fraction. All APOE RNA species demonstrated significantly higher expression in AD frontal lobe than in control frontal lobe. Furthermore, we observed a negative correlation between the levels of total APOE RNA and DNA methylation at the APOE CGI in the frontal lobe. When stratified by disease status, this correlation was strengthened in controls but not in AD. Our findings suggest a possible modified mechanism of gene action for APOE in AD that involves not only the protein isoforms but also an epigenetically regulated transcriptional program driven by DNA methylation in the APOE CGI.

Background: Cancer screening is a complex process involving multiple steps and levels of influence (e.g., patient, provider, facility, health care system, community, or neighborhood). We describe the design, methods, and research agenda of the Population-based Research to Optimize the Screening Process (PROSPR II) consortium. PROSPR II Research Centers (PRC), and the Coordinating Center aim to identify opportunities to improve screening processes and reduce disparities through investigation of factors affecting cervical, colorectal, and lung cancer screening in U.S. community health care settings. Methods: We collected multilevel, longitudinal cervical, colorectal, and lung cancer screening process data from clinical and administrative sources on >9 million racially and ethnically diverse individuals across 10 heterogeneous health care systems with cohorts beginning January 1, 2010. To facilitate comparisons across organ types and highlight data breadth, we calculated frequencies of multilevel characteristics and volumes of screening and diagnostic tests/procedures and abnormalities. Results: Variations in patient, provider, and facility characteristics reflected the PROSPR II health care systems and differing target populations. PRCs identified incident diagnoses of invasive cancers, in situ cancers, and precancers (invasive: 372 cervical, 24,131 colorectal, 11,205 lung; in situ: 911 colorectal, 32 lung; precancers: 13,838 cervical, 554,499 colorectal). Conclusions: PROSPR II's research agenda aims to advance: (i) conceptualization and measurement of the cancer screening process, its multilevel factors, and quality; (ii) knowledge of cancer disparities; and (iii) evaluation of the COVID-19 pandemic's initial impacts on cancer screening. We invite researchers to collaborate with PROSPR II investigators. Impact: PROSPR II is a valuable data resource for cancer screening researchers.

The application of natural language processing and machine learning (ML) in electronic health records (EHRs) may help reduce dementia underdiagnosis, but models that are not designed to reflect minority populations may instead perpetuate underdiagnosis. To improve the identification of undiagnosed dementia, particularly in Black Americans (BAs), we developed support vector machine (SVM) ML models to assign dementia risk scores based on features identified in unstructured EHR data (via latent Dirichlet allocation and stable topic extraction in n = 1 M notes) and structured EHR data. We hypothesized that separate models would show differentiation between racial groups, so the models were fit separately for BAs (n = 5 K with dementia ICD codes, n = 5 K without) and White Americans (WAs; n = 5 K with codes, n = 5 K without). To validate our method, scores were generated for separate samples of BAs (n = 10 K) and WAs (n = 10 K) without dementia codes, and the EHRs of 1.2 K of these patients were reviewed by dementia experts. All subjects were age 65+ and drawn from the VA, which meant that the samples were disproportionately male. A strong positive relationship was observed between SVM-generated risk scores and undiagnosed dementia. BAs were more likely than WAs to have undiagnosed dementia per chart review, both overall (15.3% vs. 9.5%) and among Veterans with >90th percentile cutoff scores (25.6% vs. 15.3%). With chart reviews as the reference standard and varied cutoff scores, the BA model performed slightly better than the WA model (AUC = 0.86 with negative predictive value [NPV] = 0.98, positive predictive value [PPV] = 0.26, sensitivity = 0.61, specificity = 0.92 and accuracy = 0.91 at >90th percentile cutoff vs. AUC = 0.77 with NPV = 0.98, PPV = 0.15, sensitivity = 0.43, specificity = 0.91 and accuracy = 0.89 at >90th). Our findings suggest that race-specific ML models can help identify BAs who may have undiagnosed dementia. Future studies should examine model generalizability in settings with more females and test whether incorporating these models into clinical settings increases the referral of undiagnosed BAs to specialists.