Blood Proteins Linked to Heart Imaging Signatures and Disease Risk in Large UK Biobank Study

Proteins circulating in the blood show measurable ties to the structure and function of the heart and vasculature as seen on cardiac scans, and some of these proteins may influence disease risk, according to a study published in Cardiovascular Diabetology. Combining cardiovascular magnetic resonance imaging, plasma proteomics and genetic analysis in UK Biobank participants, the researchers mapped how imaging-defined cardiovascular phenotypes relate to a broad range of diseases and identified candidate proteins that could underpin those links.

The observational arm drew on longitudinal data from 68,566 UK Biobank participants with a median follow-up of 3.82 to 3.98 years. The team evaluated associations between 14 cardiovascular magnetic resonance metrics and the risks of 41 cardiovascular and systemic diseases, as well as all-cause and cause-specific mortality. They identified 88 significant cross-system associations, supporting the idea that cardiovascular structural and functional phenotypes captured by imaging are relevant to disease well beyond the heart itself.

To probe the biology behind these imaging phenotypes, the researchers turned to a proteomic sub-cohort of 7,892 participants. Multivariable-adjusted regression identified 681 unique proteins significantly associated with cardiovascular magnetic resonance phenotypes. The resulting biological landscape was dominated by immune-inflammatory processes and extracellular matrix remodeling, pointing to inflammation and tissue restructuring as central themes in cardiovascular remodeling as reflected in the blood protein profile.

Using machine learning approaches, the team distilled the CMR-associated plasma proteins into parsimonious protein-CMR models comprising 26 to 57 proteins. These models showed moderate to strong predictive performance for the imaging metrics, with Pearson correlation coefficients ranging from 0.43 to 0.72, and demonstrated potential utility for predicting and stratifying the risk of incident cardiovascular and other systemic morbidities and mortality.

On the genetic side, Mendelian randomization analysis prioritized 222 potential causal protein-CMR pairs whose effect directions were consistent with the observational analyses. The analysis highlighted proteins associated with multiple imaging phenotypes, including ISLR2, EPHA4 and NRCAM. Colocalization analysis, which tests whether two associations share the same causal genetic variants, further prioritized 6 pairs with moderate-to-strong evidence of shared causal variants, strengthening the case that these protein-imaging links are not merely correlational.

The researchers then applied two-step Mendelian randomization to ask whether any imaging phenotype might sit on the causal pathway between proteins and disease. This suggested that left ventricular mass potentially mediates the genetically supported associations of five upstream proteins with atrial fibrillation. As with all Mendelian randomization work, the findings are hypothesis-generating: the approach relies on genetic assumptions and indicates potential causal directions that require confirmation in other study designs.

To explore translational relevance, the team integrated single-cell transcriptomic annotations and a druggability assessment of the prioritized proteins. This layer of analysis indicates which of the candidate proteins act in specific cell types and which might be reachable with existing or future drugs, though the study itself did not test any therapeutic intervention.

The study is an early-access, peer-reviewed accepted article carrying a permanent DOI, and its version is subject to further edits before the final Version of Record. The mortality and disease endpoints reflect associations observed within that window.

Overall, the work offers an integrated framework connecting circulating proteomic signatures to CMR-defined cardiovascular remodeling and to cardiovascular and systemic disease risk. By triangulating observational imaging associations, machine-learning protein models, genetic causal inference and cellular annotation, it provides a map of candidate proteins and pathways for future studies of cardiovascular remodeling and multi-system disease.

Subject of Research: Proteomic and genetic determinants of cardiovascular magnetic resonance imaging phenotypes and their links to cardiovascular and systemic disease risk

Article Title: Proteomic signatures of cardiovascular imaging phenotypes and links with cardiovascular and systemic outcomes: a prospective cohort and Mendelian randomization study

Article References: Xu, F., Liu, Y., Li, M., Ning, J., Yang, H., Ma, Z., Liu, Y., Ji, C., Chang, Q., Zhao, Y., Hu, J., Sun, F., & Xia, Y. (2026). Proteomic signatures of cardiovascular imaging phenotypes and links with cardiovascular and systemic outcomes: a prospective cohort and Mendelian randomization study. Cardiovascular Diabetology. https://doi.org/10.1186/s12933-026-03416-5

Image Credits: AI Generated

DOI: 10.1186/s12933-026-03416-5

Keywords: UK Biobank, proteomics, cardiovascular magnetic resonance, Mendelian randomization, atrial fibrillation, left ventricular mass, extracellular matrix, immune-inflammatory processes, colocalization, single-cell transcriptomics, cardiovascular remodeling, mortality risk