Multi-Cohort Transcriptomic and Single-Cell Analyses Identify Six Directionally Concordant Blood-Transcriptomic Signals Associated with Low Bone Mineral Density and Alzheimer’s Disease
Open Access
Abstract
Background/Objectives: Alzheimer’s disease (AD) and low bone mineral density (BMD) frequently coexist, but whether they show reproducible concordant blood-transcriptomic signals remains unclear. We conducted a retrospective computational study to identify cross-phenotype expression patterns and explore their biological context. Methods: One low-versus-high-BMD monocyte cohort and two AD whole-blood cohorts were used to derive six directionally concordant candidate genes. Repeated nested cross-validation provided internal predictive characterization, followed by independent AD evaluation in GSE140829 and exploratory osteoporosis evaluation in GSE56116. Gene–immune analyses, leukocyte-composition sensitivity analyses, Mendelian randomization, WGCNA, donor-aware single-cell analysis, virtual knockout, and direction-constrained compound prioritization were used as complementary exploratory analyses. Results: The candidates were FOXO3, IL1R2, WWP1, PDPK1, DPP8, and TSNAX. In GSE140829, all six retained their discovery directions, but only IL1R2 passed candidate-level FDR correction; the six-gene and robust-core scores showed modest discrimination (AUCs 0.601 and 0.610). Adjustment for inferred leukocyte composition attenuated the IL1R2 effect, but it remained FDR-supported using both CIBERSORT-derived (FDR = 0.00483) and xCell2-derived (FDR = 0.0140) composition covariates. Mendelian randomization yielded no FDR-supported causal estimate. Network, single-cell, virtual-knockout, and compound-prioritization results were exploratory and hypothesis-generating. Conclusions: Six blood-transcriptomic signals showed directional concordance between low BMD and AD, with IL1R2 providing the strongest independent AD evidence after leukocyte-composition sensitivity analyses. These findings do not establish a disease-specific shared causal mechanism, a clinically useful biomarker, or a therapeutic strategy. Prospective, protein-level, and functional validation is required before clinical or therapeutic application can be considered.Keywords
This publication has 44 references indexed in Scilit:
- Integrating genetic data with biological insight: A practical guide to cis-Mendelian randomizationThe American Journal of Human Genetics, 2026
- Bidirectional association between cognitive impairment and bone mineral density reduction in aging populations: A systematic review and meta-analysis of osteoporosis, osteopenia, and bone mineral contentBone, 2025
- Causal effect of blood osteocalcin on the risk of Alzheimer’s disease and the mediating role of energy metabolismTranslational Psychiatry, 2024
- scTenifoldKnk: An efficient virtual knockout tool for gene function predictions via single-cell gene regulatory network perturbationPatterns, 2022
- RNA-Seq Signatures Normalized by mRNA Abundance Allow Absolute Deconvolution of Human Immune Cell TypesCell Reports, 2019
- A framework for the investigation of pleiotropy in two‐sample summary data Mendelian randomizationStatistics in Medicine, 2017
- Robust enumeration of cell subsets from tissue expression profilesNature Methods, 2015
- Soluble interleukin-1 receptor type II, IL-18 and caspase-1 in mild cognitive impairment and severe Alzheimer's diseaseNeurochemistry International, 2005
- Adjusting for multiple testing—when and how?Journal of Clinical Epidemiology, 2001
- Soluble interleukin-1 receptor type II levels are elevated in cerebrospinal fluid in Alzheimer's disease patientsBrain Research, 1999