Desarrollo de un modelo de predicción de riesgo de trombosis ventricular izquierda en pacientes con SICA

To develop a risk prediction model for left ventricular thrombus (LVT) formation in patients with acute ST-segment elevation myocardial infarction (STEMI).

We performed a retrospective analysis of patients with STEMI in our hospital between October 2017 to October 2020. According to transthoracic echocardiography, these patients were included in the LVT group (n = 50) or no-LVT group (n = 130). Clinical data were collected from both groups. The comparison between groups, single-factor logistic regression analysis, Lasso regression analysis and multi-factor logistic regression analysis were performed successively to screen the risk factors and to establish risk prediction models. After evaluation and internal verification, we obtained an optimal risk prediction model. A nomogram was constructed to visualize the optimal model.

The risk prediction model contained seven variables including MCV (OR = 1.251, 95% CI = 1.021-1.531, P = 0.030), D-Dimer (OR = 9.798, 95% CI = 2.630-36.503, P = 0.001), CRP (OR = 1.033, 95% CI = 1.011-1.055, P = 0.003), LVEF (OR = 0.903, 95% CI = 0.819-0.995, P = 0.040), A wave velocity (OR = 0.044, 95% CI = 0.002-0.906, P = 0.043), pericardial effusion (OR = 16.926, 95% CI = 2.767-103.522, P = 0.002) and anterior wall infarction (OR = 12.275, 95% CI = 2.136-70.548, P = 0.005). The C-index value was 0.966 and the area under the ROC curve was 96.6% (95% CI = 0.9399-0.9914). Simple cross-validation, k-fold, leave-one-out, and bootstrap analyses were used to internally verify the model. The accuracy of the model were 0.926, 0.9, 0.9, and 0.899, and Kappa values were 0.807, 0.744, 0.744, and 0.739. The area under the ROC curve was 94.7%, as verified by bootstrapping.

MCV, D-dimer level, CRP level, LVEF, A-wave velocity, pericardial effusion, and anterior wall infarction were independently related to the occurrence of LVT in STEMI at the acute stage. The multivariate logistic regression risk prediction model developed in this study demonstrates good discrimination and calibration, enabling preliminary risk stratification for LVT in patients with acute STEMI.