Article Impact Level: HIGH Data Quality: STRONG Summary of European Heart Journal, ehaf595. https://doi.org/10.1093/eurheartj/ehaf595 Dr. Rick H. J. A. Volleberg et al.
Points
- Identifying high-risk coronary plaques from complex OCT images is a time-consuming and specialized task that limits its widespread clinical use for predicting future adverse cardiovascular events.
- Researchers evaluated an AI algorithm against an expert core lab in 414 post-myocardial infarction patients to identify vulnerable plaques and predict poor outcomes over two years.
- AI-identified plaques were significantly associated with adverse events, whereas plaques identified by the expert lab showed a non-significant association with the primary outcome of death, MI, or revascularization.
- AI analysis of the entire imaged coronary artery demonstrated a significantly stronger ability to predict future cardiovascular events, outperforming the assessment of only a single predefined lesion.
- This AI-driven OCT analysis provides a standardized tool for rapid risk stratification, enabling preventative treatments by comprehensively mapping all vulnerable spots within a patient’s coronary arteries.
Summary
A new study evaluated an artificial intelligence algorithm (OCT-AID) for identifying coronary thin-cap fibroatheromas (TCFA) on optical coherence tomography (OCT) images and its ability to predict clinical outcomes. The PECTUS-AI study was a secondary analysis of 438 patients with myocardial infarction, of whom 414 were included in the final analysis. OCT images of all non-culprit lesions were analyzed for TCFA by both the AI algorithm (AI-TCFA) and an independent core laboratory (CL-TCFA), which served as the gold standard for manual interpretation.
The primary outcome was a composite of all-cause death, non-fatal myocardial infarction, or unplanned revascularization at two years. AI-TCFA was identified in 143 patients (34.5%) and CL-TCFA in 124 patients (30.0%). The presence of AI-TCFA within the target lesion was significantly associated with the primary outcome (hazard ratio [HR] 1.99, 95% confidence interval [CI] 1.02–3.90, P = .04). In contrast, the association for CL-TCFA was not statistically significant (HR 1.67, 95% CI: 0.84–3.30, P = .14).
Furthermore, when the AI algorithm evaluated the complete OCT pullback rather than just the target lesion, it demonstrated a substantially stronger association with the primary outcome (HR 5.50, 95% CI: 1.94–15.62, P < .001). This whole-vessel analysis also yielded a high negative predictive value of 97.6% (95% CI: 94.0%–99.3%). The findings suggest that AI-based OCT analysis offers a standardized, rapid, and prognostically superior alternative to manual interpretation, particularly when assessing entire coronary segments for risk stratification.
Link to the article: https://academic.oup.com/eurheartj/advance-article/doi/10.1093/eurheartj/ehaf595/8244402
References Volleberg, R. H. J. A., Luttikholt, T. J., Van Der Waerden, R. G. A., Cancian, P., Van Der Zande, J. L., Gu, X., Mol, J.-Q., Roleder, T., Prokop, M., Sánchez, C. I., Van Ginneken, B., Išgum, I., Saitta, S., Thannhauser, J., & Van Royen, N. (2025). Artificial intelligence-based identification of thin-cap fibroatheromas and clinical outcomes: The PECTUS-AI study. European Heart Journal, ehaf595. https://doi.org/10.1093/eurheartj/ehaf595
