Coronary disease & ACSDiagnostic or prognostic study
Prediction of coronary artery disease using urinary proteomics
- Diagnostic accuracy study
- discovery and validation cohorts
Classifier developed by support vector machine in a discovery cohort of 36 coronary artery disease (CAD) and 36 matched controls
Authors' conclusion
A novel urinary proteomic classifier related to collagen metabolism, lipids, and inflammation showed potential for the risk prediction of CAD.
Study context
- Population
- Participants with urinary proteome analysed by capillary electrophoresis–mass spectrometry; validation cohort of individuals who progressed to CAD and controls
- Participants
- 965 participants
- Comparison
- Novel urinary proteomic classifier vs CAD238 and ACSP75 classifiers
- Outcome
- CAD prediction (AUC)
- Follow-up
- 8 years
Performance
As reported in the abstract
CAD prediction, validation cohort (115 CAD, 778 controls)
| Test or model | AUC |
|---|---|
| Novel classifier | 0.82 (95% CI 0.78-0.87) |
| CAD238 | 0.71 (95% CI 0.66-0.77) |
| ACSP75 | 0.53 (95% CI 0.47-0.60) |
Further findings
| Outcome | HR (95% CI) | P value | |
|---|---|---|---|
| CAD risk per 1-SD increment in novel classifier (multivariable Cox) | 1.54 (1.26–1.89) | P < 0.0001 | |
| 0.512Hazard ratio (log scale) |
| Outcome | net reclassification index (95% CI) | P value |
|---|---|---|
| Risk reclassification on top of Framingham or SCORE2 (n = 737 with scores) — Novel classifier + Framingham vs Framingham | 0.61 (0.25–0.95) | P = 0.001 |
| Risk reclassification on top of Framingham or SCORE2 (n = 737 with scores) — Novel classifier + SCORE2 vs SCORE2 | 0.64 (0.28–0.98) | P = 0.001 |
AUC of novel classifier added to CAD238 and ACSP75
- AUC improved
- 0.84 (95% CI 0.80-0.89)
Original publication
Prediction of coronary artery disease using urinary proteomics. Eur J Prev Cardiol. Published 21 March 2023. doi:10.1093/eurjpc/zwad087
Summarised from the published abstract; every number is checked against it