Objective:

To identify independent risk factors for postcontrast acute kidney injury (PC-AKI) in patients with postoperative AKI (PO-AKI) following acute Stanford type A aortic dissection (ATAAD), and to develop a prediction model for clinical use.

Approach:

• Study Design: This retrospective cohort study enrolled 604 PO-AKI patients from January 2014 to December 2024 at Nanjing Drum Tower Hospital who underwent postoperative contrast-enhanced CTA.
• Model Development: Three variable-selection strategies (backward stepwise AIC, LASSO, and XGBoost-SHAP) were used; a multivariable logistic regression model was constructed and validated through bootstrap resampling.

Key Findings:

• PC-AKI incidence was 9.8% (59/604), with significant recovery-dependent stratification: 3.5% in fully recovered PO-AKI vs. 52.5% in unrecovered PO-AKI.
• Independent predictors included PO-AKI stage 3 (OR = 3.144, 95% CI: 1.41–7.06) and unrecovered PO-AKI before first CTA (OR = 25.212, 95% CI: 12.57–53.49).
• The model exhibited good discrimination (AUC=0.848, 95% CI: 0.78–0.91) and calibration (Brier=0.057).
• PC-AKI was associated with prolonged ICU stay (RR = 1.521, 95% CI: 1.19–1.97) and incomplete renal recovery at discharge (OR = 2.554, 95% CI: 1.30–4.86).

Interpretation:

Dynamic PO-AKI recovery and advanced AKI stage are associated with PC-AKI risk in post-ATAAD patients.

Limitations:

• External validation of the prediction model is needed before clinical deployment to ensure its applicability in broader settings.

Conclusion:

The internally validated model may aid individualized risk stratification before contrast procedures in this high-risk subgroup.