Prediction of mechanical complications post-acute myocardial infarction in individuals with type 2 diabetes mellitus

Abstract

Background: Acute myocardial infarction (AMI) patients with type 2 diabetes mellitus (T2DM) represent a unique population characterized by poorer prognoses, which may be further exacerbated by mechanical complications. This study aims to develop a predictive model to identify high-risk individuals within this populations.

Methods: This study enrolled AMI patients with T2DM and categorized them into complication and control groups. The mechanical complications were defined as papillary muscle rupture (with or without acute mitral regurgitation), ventricular septal defect, left ventricular pseudoaneurysm or aneurysm (with or without thrombus) and free wall rupture. Characteristics were selected using relaxed least absolute shrinkage and selection operator (LASSO) logistic regression, multivariate logistic regression and random forest model. Selected variables were utilized to construct a nomogram to predict the possibility of mechanical complications.

Results: A total of 2,816 patients were enrolled, with 191 individuals classified into the complication group. Baseline analysis identified 31 factors exhibiting potential differences, which were subsequently employed for LASSO-logistic regression, multivariate logistic regression and random forest model. After comprehensive evaluation, nine variables emerged as predictive factors for mechanical complications, including gender, pulmonary hypertension, ST-segment elevation myocardial infarction, body mass index, N-terminal pro-brain natriuretic peptide, creatine kinase, left ventricle ejection fraction and hemoglobin A1c, which were used to construct a reliable nomogram. The complication group also showed higher in-hospital mortality rates compared to controls, alerting the worse prognosis of these populations.

Conclusion: This study identified nine factors upon admission that may be associated with mechanical complications during the hospitalization. A nomogram was developed based on these factors for clinical application. T2DM patients should emphasize glucose control, which may offer benefits following the onset of AMI.

Keywords: acute myocardial infarction; hemoglobin A1c; mechanical complications; prediction model; type 2 diabetes mellitus.

Figure 1

Correlation heatmap between selected variables and the incidence of complications. LYMPH, lymphocyte count; AST, aspartate aminotransferase; CK-MB, creatine kinase isoenzymes MB; CK, creatine kinase; hs-cTnT, high-sensitivity cardiac troponin T; LDH, lactate dehydrogenase; NEU, neutrophil count; NLR, neutrophil-to-lymphocyte ratio; WBC, white blood cell; HDL, high-density lipoprotein; PH, pulmonary hypertension; FDP, fibrin degradation products; BUN, blood urea nitrogen; NT-ProBNP, N-terminal pro-brain natriuretic peptide; STEMI, ST-segment elevation myocardial infarction; HbA1c, hemoglobin A1c; PCT, procalcitonin; LVEF, left ventricle ejection fraction; A/G, albumin/globulin ratio; BMI, body mass index; TG, triglycerides.

Figure 2

Results of LASSO-logistic regression. (a) Cross-validation plot of LASSO-logistic regression. (b) Selection process of LASSO-logistic regression model by cross-validation method. BMI, body mass index; STEMI, ST-segment elevation myocardial infarction; HbA1c, hemoglobin A1c; WBC, white blood cell; NEU, neutrophil count; LYMPH, lymphocyte count; NLR, neutrophil-to-lymphocyte ratio; PCT, procalcitonin; AST, aspartate aminotransferase; A/G, albumin/globulin ratio; BUN, blood urea nitrogen; HDL, high-density lipoprotein; TG, triglycerides; NT-ProBNP, N-terminal pro-brain natriuretic peptide; hs-cTnT, high-sensitivity cardiac troponin T; CK-MB, creatine kinase isoenzymes MB; CK, creatine kinase; LDH, lactate dehydrogenase; FDP, fibrin degradation products; LVEF, left ventricle ejection fraction; PH, pulmonary hypertension.

Figure 3

Results of random forest model. (a) Mean decrease accuracy and mean decrease Gini index of all variables. (b) Arithmetic mean value between mean decrease accuracy and mean decrease Gini index of all variables. LVEF, left ventricle ejection fraction; NT-ProBNP, N-terminal pro-brain natriuretic peptide; CK, creatine kinase; BMI, body mass index; PCT, procalcitonin; TG, triglycerides; LYMPH, lymphocyte count; HbA1c, hemoglobin A1c; PH, pulmonary hypertension; STEMI, ST-segment elevation myocardial infarction.

Figure 4

Nomogram for the prediction of mechanical complications in acute myocardial infarction patients with type 2 diabetes mellitus. PH, pulmonary hypertension; STEMI, ST-segment elevation myocardial infarction; BMI, body mass index; NT-ProBNP, N-terminal pro-brain natriuretic peptide (reference range for healthy populations: 0–125 ng/mL; diagnosis range for acute heart failure in people less than 50 years old: >450 ng/mL; diagnosis range for chronic heart failure: >2,000 ng/mL; upper detection limit: 35,000 pg/mL); CK, creatine kinase (reference range: 40–200 U/L); LVEF, left ventricle ejection fraction (reference range: 52–75%); HbA1c, hemoglobin A1c (reference range: 4.0–6.0%).

Figure 5

Evaluation of the nomogram. The receiver operating characteristic curve (a), decision curve (b) and calibration curve (c) of the nomogram. AUC, area under curve.