AgingResearch.NewsJoin free
BMC Cardiovascular Disorders

Proteins improved cardiovascular risk prediction in metabolic syndrome

A blood-protein model better distinguished who later developed cardiovascular events, scoring 0.730 versus 0.673 for a clinical model in a validation group.

Paper by Yanlin Chen, Wei Pan, Haichu Wen, 4 more authorsJie Du · Capital Medical University
Figure 1 from BMC Cardiovascular Disorders
Open the figure at full size
Figure 1. Chen et al.
Cohort study of 12,308 people with metabolic syndrome, against the clinical PREVENT modelBiomarkers

Limit: Participants were healthier than the broader population, potentially biasing the model’s performance

Edited by Boris Djordjevic · AgingResearch.News

Paper
Follow

In an observational study, researchers analyzed 2,911 blood proteins in 12,308 UK Biobank participants with metabolic syndrome, a cluster of conditions that raise cardiovascular risk. They recorded 1,899 major adverse cardiovascular events over a median 14-year follow-up.

The researchers built a protein-based prediction model in a training group and tested it in a validation group. Its C-statistic, a measure of how well a model distinguishes risk, was 0.730, compared with 0.673 for a clinical model called PREVENT. Among participants PREVENT classified as intermediate risk, the protein score correctly reclassified 64% of those who developed an event.

Why it matters

Cardiovascular disease can limit health in later life. Better risk estimates may help guide personalized prevention for people with metabolic syndrome.

Caveats

Validation used another group from UK Biobank with metabolic syndrome, not an outside population. The observational study did not test whether using the score prevented cardiovascular events.

From the paper

Participants

This study included 12,308 individuals with MetS who were free of MACEs at baseline and for whom plasma proteomics data was available.

Results, Baseline characteristics of the study population

Follow-up

Over a median 14-year follow-up, 1,899 incident MACEs occurred.

Results

Result

… genetically predicted higher levels of advanced glycosylation end-product-specific receptor (AGER; OR = 0.754, 95% CI: 0.667–0.853, P = 6.53 × 10⁻⁶) were associated with a decreased risk of HF …

Results, Causal evaluation of MACE-predictive proteins in MetS via cis-pQTLs and Mndelian randomization

Limitation

First, the proteomics prediction models were mainly constructed using data from individuals of White ethnicity within the UKB, and lack an independent external validation cohort, which limits the generalizability of the results.

Discussion

Quoted word for word from the full text on link.springer.com.

Comments

Be the first to comment.

More on Machine learning