Predicting perioperative risk of persistent paraplegia from preoperative factors in patients who undergo elective, open thoracoabdominal aortic aneurysm repair.

BACKGROUND

Spinal cord injury is a risk for patients undergoing thoracoabdominal aortic aneurysm (TAAA) repair, the most substantial deficit being persistent paraplegia (PP). Adequately estimating patients' risk of PP is vital for patient counseling and postoperative risk stratification.

METHODS

Patients who underwent open, elective TAAA repair (1986-2024; N = 2827) at a single practice were categorized into 4 groups: those with PP (n = 68) at discharge or death, those with persistent paraparesis (n = 55) at discharge or death, those with transient paraplegia or paraparesis (n = 132), and those without any spinal cord deficit (SCD, n = 2572); 4 cohorts were created comprising combinations of these groups. Several models were evaluated, including logistic regression and machine-learning approaches, to better understand predictors of PP and other SCDs in key cohorts.

RESULTS

Regarding prediction of PP, the gradient boosting machine was best (C = 0.79 [0.75-0.81]); however, the multivariable logistic regression model was similar (0.78 [0.74-0.81]) and was converted into a nomogram to highlight the predictive value of diabetes (odds ratio, 2.48; P = .009), previous percutaneous coronary intervention (OR, 3.04; P < .001), chronic kidney disease (OR, 2.48; P < .001), tobacco use (OR, 13.25; P = .01), symptomatic aortic disease (OR, 1.96; P = .01), and Crawford extent of TAAA repair (II: OR, 4.00; P < .001). Predictors of PP and SCD in key cohorts were similar.

CONCLUSIONS

For patients undergoing elective, open TAAA repair, the risk of PP can be estimated from preoperative characteristics and explained by an intuitive nomogram to aid personalized preoperative counseling, with greater-risk patients prioritized perioperatively for rigorous blood pressure monitoring, increased cerebrospinal fluid drainage, or a lower threshold for transfusion.