Prognostic implications of the high sensitivity C-reactive protein to albumin ratio for major cardiovascular adverse events in patients undergoing noncardiac surgery

The prognostic significance of the C-reactive protein-to-albumin ratio (CAR) in patients undergoing noncardiac surgery has not been conclusively determined. This study aimed to investigate the correlation and predictive capacity of the CAR for major perioperative adverse cardiovascular events (MACEs) in noncardiac surgical patients. Individuals who underwent noncardiac surgical procedures were identified within the perioperative medicine database (INSPIRE 1.1) and stratified into tertiles on the basis of the CAR. The associations between the CAR and the risk of MACEs (all-cause death, perioperative myocardial infarction, and acute heart failure) were analysed via Cox proportional hazards regression analysis, and restricted cubic splines were subsequently used to examine the shape of the associations. Additionally, the CAR's incremental predictive value for MACEs was evaluated using the C statistic, continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI), in order to quantify the improvement in predictive performance achieved by adding the CAR to the Revised Cardiac Risk Index (RCRI). The feature significance and predictive models were developed via the Boruta algorithm and machine learning techniques. A total of 7,366 patients were included. There was a significant difference in the perioperative MACE risk among the three groups (tertiles 1 to 3: 0.24% vs. 0.69% vs. 1.63%; log-rank P < 0.001). Compared with patients in the lowest tertile, those with the highest CAR had a significantly increased risk of MACEs (hazard ratio [HR] 4.43, 95% confidence interval [CI] 1.53-12.86, P = 0.006), and an L-shaped association was observed. For each SD (standard deviation) increase in the CAR, the risk of perioperative MACE increased by 1.22 times (HR 1.22, 95% CI 1.06-1.41). Furthermore, incorporating the CAR into the Revised Cardiac Risk Index (RCRI) or baseline risk models significantly enhanced the ability of the CAR to predict perioperative MACEs. The CatBoost algorithm model showed the best performance (AUC = 0.865). The CAR was associated with an increased risk of perioperative MACEs in patients undergoing noncardiac surgery, indicating its potential as a valuable and reliable prognostic tool for evaluating the risk of perioperative MACEs.