Bingshan Li

Importance: Hypertension is among the most important modifiable risk factors for Alzheimer’s disease and related dementias (ADRD), but whether different antihypertensive drug classes confer different cognitive risk, particularly through differential modulation of angiotensin II receptor signaling, remains uncertain. Prior pharmacoepidemiologic evidence is based largely on a single Medicare cohort, and external validation across distinct healthcare data environments is needed. Objective: To externally replicate and extend prior findings across two structurally independent healthcare data sources, evaluating whether predominant use of antihypertensive medications that stimulate versus inhibit angiotensin II type 2 and type 4 receptor signaling is associated with risk of incident ADRD. Design, Setting, and Participants: Retrospective new-user, active-comparator cohort study with propensity-score matching, conducted in parallel across two independent data sources: the Vanderbilt University Medical Center Synthetic Derivative (VUMC SD; deidentified academic medical center electronic health record through February 2025) and the Merative ^™ MarketScan ^® Medicare Database (2015–2023). Adults aged 65 years or older with hypertension and at least 365 cumulative days of antihypertensive exposure were included after a 365-day blanking period. Exposures: Predominant use (≥80% of cumulative exposure days) of receptor-stimulating antihypertensive medications (angiotensin II receptor blockers, dihydropyridine calcium channel blockers, thiazide diuretics) vs receptor-inhibiting antihypertensive medications (angiotensin-converting enzyme inhibitors, β-blockers, nondihydropyridine calcium channel blockers); a third group of nonusers comprised participants without predominant use of either class. Main Outcomes and Measures: Time to first incident ADRD diagnosis was identified using ICD-9 and ICD-10 codes consistent with the CMS Chronic Conditions Data Warehouse definition. Adjusted hazard ratios (HRs) and 95% CIs were estimated using Cox proportional hazards models with time-dependent covariates. Results: The Medicare cohort included 1,596,034 beneficiaries (mean [SD] age, 75.0 [8.0] years; 55% female), and the VUMC SD cohort included 80,267 patients (mean [SD] age, 72.3 [5.8] years; 57% female). Over approximately 7 years of follow-up, ADRD incidence was lower among predominant users of stimulating vs. inhibiting medications in both cohorts (Medicare: 6.7% vs. 8.2%; VUMC SD: 4.9% vs. 5.7%). After adjustment, predominant use of stimulating medications was associated with lower ADRD risk versus inhibiting medications in both cohorts (Medicare: HR, 0.92; 95% CI, 0.90–0.93; P < .001; VUMC SD: HR, 0.92; 95% CI, 0.85–0.99; P = .03). Established cardiovascular and neuropsychiatric risk factors (atrial fibrillation, heart failure, chronic kidney disease, depression) showed expected associations with ADRD risk in both cohorts, supporting internal validity. Conclusions and Relevance: Across two structurally independent data sources, the predominant use of antihypertensive medications that stimulate angiotensin II type 2 and type 4 receptor signaling was associated with a modestly lower risk of incident ADRD compared with predominantly inhibiting agents. By replicating in two distinct data environments, these findings extend prior single-source pharmacoepidemiologic evidence and support evaluation of receptor-targeted antihypertensive strategies for cognitive outcomes in randomized comparative-effectiveness trials.

Genome-wide association studies (GWAS) have identified over 200 genetic risk loci for breast cancer, yet their target genes remain largely unknown. We conduct multi-ancestry transcriptome-wide association studies (TWAS) to discover potential breast cancer susceptibility genes. We develop ancestry-specific genetic models to predict levels of gene expression, alternative splicing, and 3' UTR alternative polyadenylation using genomic and transcriptomic data from 652 normal female tissue samples and apply these models to GWAS data of 178,534 cases and 248,300 controls for association analyses. We identify 290 genes associated with breast cancer risk, including 103 previously unreported and 46 not located at known GWAS loci, and 39 genes show distinct associations with breast cancer risk by estrogen-receptor status. Single-cell RNA sequencing and in vitro experiment data provide additional functional evidence for 169 genes. These genes are enriched in pathways implicated in breast carcinogenesis. Our study uncovers insights into breast cancer genetics and biology.

Background: Head and neck squamous cell carcinoma (HNSCC) presents significant diagnostic and therapeutic challenges due to its rising incidence and poor prognosis. Liquid biopsy, utilizing biofluids such as blood and saliva, offers a noninvasive approach to detect circulating tumor DNA (ctDNA), circulating tumor RNA (ctRNA), and small extracellular vesicles (sEVs) containing DNA (sEV-DNA) and RNA (sEV-RNA), which reflect tumor biology. However, comprehensive comparisons of these liquid biopsy markers are limited. Methods: Paired plasma and saliva samples were collected from HNSCC patients and healthy controls. We analyzed ctDNA and sEV-DNA from both plasma and saliva for tumor fraction (TF), copy number variations (CNVs), and methylation patterns. Additionally, we profiled small RNAs from saliva-derived ctRNA and sEV-RNA to explore their potential as biomarkers for disease monitoring. Results: Saliva sEV-DNA exhibited significantly higher TF compared to plasma sEV-DNA in HNSCC patients (mean, 0.0744 vs 0.0253, P < 0.001), suggesting superior tumor DNA capture in saliva. CNV analysis revealed shared and distinct genomic alterations across different sample types. Aberrant methylation patterns were observed more frequently in sEV-DNA than in ctDNA across both saliva and plasma samples. Small RNA profiling demonstrated broader coverage of differentially expressed genes in saliva sEV-RNA, enhancing its biomarker potential. Conclusions: Our findings suggest that saliva-derived sEVs hold promise as noninvasive biomarkers for HNSCC. Saliva-based liquid biopsies, particularly sEV-DNA and sEV-RNA, provide a comprehensive molecular portrait of HNSCC that may facilitate the development of personalized medicine strategies.

Genome-wide association studies (GWASs) have identified approximately 100 loci for lung cancer, but potential causal genes remain largely unknown. To address this, we conducted a lung tissue-specific transcriptome-wide association study (TWAS). Gene expression prediction models were constructed using data of adjacent normal lung tissues from our Vanderbilt Thoracic Biorepository (N=314) and normal lung tissues from the GTEx (N=466) and then applied to our lung cancer GWAS meta-analysis (55,174 cases and 1,294,174 controls). We identified 109 unique risk genes for lung cancer and its histological subtypes. Of them, 71 unique genes were novel discoveries, and 13 unique genes reside in novel loci. Smoking-conditional analysis revealed that 52 unique genes are unrelated to smoking behavior. Seven unique genes showed cell-type-specific colocalization within potential risk cell types, including the alveolar type I and II, dendritic, and natural killer cells. Seventeen unique genes are targeted of 58 drugs that have been approved or in Phase II or III trials. In addition, 22 unique potential causal genes were supported by both Mendelian randomization and colocalization. Functional validation identified three genes through in vitro knockdown experiments. Our study identified new lung cancer candidate risk genes and offered insights into lung cancer biology and future translational utilities.

Drug repurposing can efficiently identify promising therapeutic targets using existing data; however, current approaches have important limitations. There is a particular need for high-throughput approaches that are both versatile and rigorous. As such, we developed a flexible, high-throughput, Mendelian randomization (MR)-based drug repurposing pipeline with three stages: 1) MR-based protein target identification, 2) MR-based validation and prioritization, and 3) drug target mapping. This pipeline can be applied to a broad range of binary and continuous traits and incorporates quality control measures such as testing for heterogeneity, horizontal pleiotropy, and Bayesian colocalization. In Stage 1, the pipeline conducts MR analyses to identify proteins with putative causal effects on a specified trait or condition. In Stage 2, targets with significant associations in Stage 1 are evaluated using MR for either the same outcome in an external cohort or a related outcome. Targets with a consistent direction of association in Stages 1 and 2 are then assessed in Stage 3, which queries DGIdb, a database of druggable therapeutic targets, to identify repurposing candidates. To demonstrate the utility and flexibility of this pipeline, we applied it to atherosclerotic cardiovascular disease. Using UKB-PPP cis-pQTLs as instruments for 2,923 circulating proteins, we identified 72 proteins associated with LDL-C and 75 with triglyceride levels from the GLGC (Stage 1). Of these, 18 lipid-associated targets were also associated with coronary artery disease (Stage 2). Drug target mapping identified 5 proteins targeted by approved drugs, highlighting potential repurposing opportunities (Stage 3).

Genetics-informed drug repurposing holds promise in translating genetic findings into therapeutics. In this study, we developed a Genetics Informed Network-based Drug Repurposing via in silico Perturbation (GIN-DRIP) framework and applied the framework to repurpose drugs for type-2 diabetes (T2D). In GIN-DRIP for T2D, it integrates multi-level omics data to translate T2D Genome-Wide Association Study (GWAS) signals into a genetics-informed network that simultaneously encodes gene importance scores and a directional effect (up/down) of risk genes for T2D; it then bases on the GIN to perform signature matching with drug perturbation experiments to identify drugs that can counteract the effect of T2D risk alleles. We identified telmisartan that belongs to the Angiotensin Receptor Blockers (ARB) drug category as a candidate, and validated the ARB antihypertensive drugs’ potential therapeutic effects on T2D in Vanderbilt University Medical Center (VUMC)’s Electronic Health Records (EHR) data with over 3 million patients.

Polygenic risk score (PRS) models effectively predict breast cancer (BC) risk in European-ancestry women but have limited accuracy for African-ancestry women, particularly for aggressive subtypes. We developed PRS models for overall BC, estrogen receptor (ER)-positive, ER-negative and triple-negative BC (TNBC) in African-ancestry women using data from the African Ancestry Breast Cancer Genetics consortium (17,391 cases and 18,800 controls). We applied several PRS methods and integrated information across ancestries and BC subtypes. The best models for overall, ER-positive, ER-negative and TNBC showed an area under the receiving operating curve of 0.612, 0.621, 0.611 and 0.639, respectively, and maintained predictive accuracy in external validation studies with area under the receiving operating curves of 0.612, 0.640, 0.605 and 0.652. We further introduce a parsimonious 162-variant PRS for TNBC with comparable accuracy (0.626). These findings demonstrate markedly improved PRS accuracy for BC risk prediction in African-ancestry women. Using these PRS models for screening will help promote more equitable cancer prevention efforts.

Susceptibility transcription factors (TF) whose DNA bindings are altered by genetic variants regulating colorectal cancer (CRC) risk genes remain poorly defined. Using generalized linear mixed models, we analyze 218 TF ChIP-Seq datasets alongside GWAS data from 100,204 CRC cases and 154,587 controls of East Asian and European ancestries. We identify 51 TFs and TF-cofactor interactions, including VDR-cofactors, as key regulators of CRC risk. Integrating these TF insights with transcriptome-wide association studies (TWAS), we further evaluate associations between genetically predicted gene expression, alternative splicing, and alternative polyadenylation with CRC risk, using RNA-seq data from 364 Asian-ancestry and 707 European-ancestry individuals. Multi-ancestry TWAS identify 222 risk genes, including 95 novel genes and 48 potentially druggable targets. Single-cell analysis provides additional functional evidence supporting ~45% of these genes, and experimental validation confirms oncogenic roles for RHPN2, IRS2, and TXN. Our findings elucidate key TF–gene regulatory networks and uncover novel CRC risk genes.