Construction and validation of nomogram model for predicting the risk of ventricular arrhythmia after emergency PCI in patients with acute myocardial infarction

Abstract

Objective: To make predictions about the risk of MVA (Malignant Ventricular Arrhythmia) after primary PCI (Percutaneous Coronary Intervention) in patients with AMI (Acute Myocardial Infarction) through constructing and validating the Nomogram model.

Methods: 311 AMI patients who suffered from emergency PCI in Hefei Second People's Hospital from January 2020 to May 2023 were selected as the training set; 253 patients suffering from the same symptom in Hefei First People's Hospital during the same period were selected as the validation set. Risk factors were further screened by means of multivariate logistic and stepwise regression. The nomogram model was constructed, and then validated by using C-index, ROC curve, decision curve and calibration curve.

Results: Multivariate logistic analysis revealed that urea, systolic pressure, hypertension, Killip class II-IV, as well as LVEF (Left Ventricular Ejection Fraction) were all unrelated hazards for MVA after emergency PCI for AMI (P<0.05); a risk prediction nomogram model was constructed. The C-index was calculated to evaluate the predictive ability of the model. Result showed that the index of the training and the validation set was 0.783 (95% CI: 0.726-0.84) and 0.717 (95% CI: 0.65-0.784) respectively, which suggested that the model discriminated well. Meanwhile, other tools including ROC curve, calibration curve and decision curve also proved that this nomogram plays an effective role in forecasting the risk for MVA after PCI in AMI patients.

Conclusions: The study successfully built the nomogram model and made predictions for the development of MVA after PCI in AMI patients.

Keywords: AMI (acute myocardial infarction); PCI (percutaneous coronary intervention); predictive model; ventricular arrhythmia.

Figure 1

Screening of risk factors for ventricular arrhythmia after emergency PCI in patients with AMI based on stepwise regression. (A) Random forest variable importance ranking; SBP: Systolic blood pressure; LVEF: Left ventricular ejection fraction; (B) Consistency index of prognostic models constructed by different clinical factors on training set and validation set; 1. Urea; 2. SBP:Systolic blood pressure; 3. LVEF:Left ventricular ejection fraction; 4. Killip Score II-IV; 5. Hypertension; 6. Urea/SBP; 7. Urea/SBP/LVEF; 8. Urea/SBP/LVEF/Killip Score II-IV; 9. Urea/SBP/LVEF/Killip Score II-IV/Hypertension.

Figure 2

Forest plot for multivariate logistic regression analysis. SBP: Systolic blood pressure. LVEF: Left ventricular ejection fraction.

Figure 3

Nomogram of ventricular arrhythmia after PCI in AMI patients. SBP: Systolic blood pressure; LVEF: Left ventricular ejection fraction; *: P < 0.05.

Figure 4

ROC curve was used to assess the predictive efficacy of the model for predicting the risk of ventricular arrhythmia after PCI in AMI patients. (A) Training set ROC curve; (B) Validation set ROC curve.

Figure 5

Calibration curve. (A) Training set calibration curve; (B) Validation set calibration curve.

Figure 6

Clinical decision curve analysis for nomogram models. (A) Training Set DCA; (B) Validation Set DCA. The DCA was plotted with the probability of high-risk threshold as abscissa and the net benefit rate as ordinate, where the probability of high-risk threshold was set to (0, 1), the solid black line represents the net benefit rate without MVA in all patients, the solid gray line represents the net benefit rate with MVA in all patients, the red curve represents the decision curve of the nomogram model of the training set, and the blue sky represents the decision curve of the validation set