To assess the ability of the biomarkers neuron-specific enolase (NSE), tau, neurofilament light chain (NFL), and glial fibrillary acidic protein (GFAP) to predict postoperative cognitive dysfunction (POCD) at discharge in patients who underwent cardiac surgery.

Post hoc analyses (with tests being prespecified before data analyses) from a randomized clinical trial.

Single-center study from a primary heart center in Denmark.

Adult patients undergoing elective or subacute on-pump coronary artery bypass grafting and/or aortic valve replacement.

Blood was collected before induction of anesthesia, after 24 hours, after 48 hours, and at discharge from the surgical ward. The International Study of Postoperative Cognitive Dysfunction test battery was applied to diagnose POCD at discharge and after three months. Linear mixed models of covariance were used to assess whether repeated measurements of biomarker levels were associated with POCD. Receiver operating characteristic (ROC) curves were applied to assess the predictive value of each biomarker measurement for POCD.

A total of 168 patients had biomarkers measured at baseline, and 47 (28%) fulfilled the POCD criteria at discharge. Patients with POCD at discharge had significantly higher levels of tau (p = 0.02) and GFAP (p = 0.01) from baseline to discharge. The biomarker measurements achieving the highest area under the ROC curve for prediction of POCD at discharge were NFL measured at discharge (AUC, 0.64; 95% confidence interval [CI], 0.54-0.73), GFAP measured 48 hours after induction (AUC, 0.64; 95% CI, 0.55-0.73), and GFAP measured at discharge (AUC, 0.64; 95% CI, 0.54-0.74), corresponding to a moderate predictive ability.

Postoperative serum levels of tau and GFAP were significantly elevated in cardiac surgery patients with POCD at discharge, however, the biomarkers achieved only modest predictive abilities for POCD at discharge. Postoperative levels of NSE were not associated with POCD at discharge.