Epidemiology
Adding Lp(a) and three other markers to the GRACE score lifts in-hospital MACE prediction AUC from 0.81 to 0.86 in 647 heart attack patients (Int J Cardiol Cardiovasc Risk Prev 2024)
Original title: Enhanced predictive performance of the GRACE risk score by incorporating lipoprotein(a) for major adverse cardiac events in acute myocardial infarction patients undergoing PCI
In 647 adult patients admitted with acute myocardial infarction between June 2016 and September 2019, the authors sought perioperative indicators to enhance the GRACE risk score's prediction of in-hospital major adverse cardiovascular events (MACE). A union model combining Lp(a) with serum uric acid, fasting blood glucose, and haemoglobin, built via stepwise and multivariate logistic regression and validated internally with bootstrap resampling, achieved an area under the receiver operating curve of 0.86, significantly better discrimination than the GRACE score alone (AUC 0.81, P < 0.05). Calibration plots showed better agreement between predicted and observed outcomes for the union model than GRACE alone, and decision curve analysis confirmed superior clinical applicability. Lp(a) enhances GRACE score prediction of in-hospital MACE after myocardial infarction, though the authors note it performs best when combined with other readily available laboratory markers rather than used alone.
Original abstract
Background: As scientific research advances, the landscape of detection indicators and methodologies evolves continuously. Our current study aimed to identify some novel perioperative indicators that can enhance the predictive accuracy of the Global Registry of Acute Coronary Events (GRACE) score for the in-hospital major adverse cardiovascular events (MACEs) in patients with acute myocardial infarction.
Methods: A total of 647 adult patients with AMI admitted to the emergency department were consecutively enrolled in the retrospective research starting from June 2016 to September 2019. The endpoint was in-hospital MACE. Stepwise regression analysis and multivariate logistic regression were performed to select the indicators for the union model established by nomogram. Bootstrap with 1000 replicates was chosen as the internal validation of the union model. The area under the receiver operating curve (AUC) and calibration plot were used to evaluate the discrimination and calibration. Decision curve analysis (DCA) was performed to evaluate the clinical sufficiency of the nomogram. Akaike's information criterion (AIC) and Bayesian Information Criterion (BIC) were used to evaluate the goodness of fit.
Results: Lipoprotein(a) combined with serum uric acid, fasting blood glucose, and hemoglobin could improve the GRACE risk score. The AUC of the union model was 0.86, which indicated a better discriminative ability than the GRACE risk score alone (AUC, 0.81; P < 0.05). The calibration plots of the union model showed favorable consistency between the prediction of the model and actual observations, which was better than the GRACE risk score. DCA plots suggested that the union model had better clinical applicability than the GRACE risk score.
Conclusion: Lipoprotein(a) has shown promise in augmenting the predictive capability of the GRACE risk score, however, it may be beneficial to integrate it with other commonly used indicators.
Summary written by lp-a.org from the published abstract; figures as published. Page updated 18 August 2026. Methods.