Computational Modeling to Predict Conservative Treatment Outcome for Patients with Plaque Erosion: An OCT-Based Patient-Specific FSI Modeling Study

Image-based computational models have been used for vulnerable plaque progression and rupture predictions, and good results have been reported. However, mechanisms and predictions for plaque erosion are under-investigated. Patient-specific fluid-structure interaction (FSI) models based on optical coherence tomography (OCT) follow-up data from patients with plaque erosion and who received conservative antithrombotic treatment (using medication, no stenting) to identify risk factors that could be used to predict the treatment outcome. OCT and angiography data were obtained from 10 patients who received conservative antithrombotic treatment. Five participants had worse outcomes (WOG, stenosis severity ≥ 70% at one-year follow-up), while the other five had better outcomes (BOG, stenosis severity < 70% at one-year follow-up). Patient-specific 3D FSI models were constructed to obtain morphological and biomechanical risk factor values (a total of nine risk factors) for comparison and prediction. A logistic regression model was used to identify optimal predictors with the best treatment outcome prediction accuracies. Our results indicated that the combination of wall shear stress (WSS), lipid percent, and thrombus burden was the best group predictor according to the mean area under the curve (AUC) of 0.96 (90% confidence interval = (0.85, 1.00)). WSS was the best single predictor with mean AUC = 0.70 (90% confidence interval = (0.20, 1.00)). Thrombus burden was the only risk factor showing statistically significant group difference, suggesting its crucial role in the outcomes of conservative anti-thrombotic therapy. This pilot study indicated that integrating morphological and biomechanical risk factors could improve treatment outcome prediction accuracy in patients with plaque erosion compared to predictions using single predictors. Large-scale patient studies are needed to further validate our findings.