Establishment and validation of a nomogram for predicting the surgical difficulty of lateral retroperitoneal laparoscopic adrenalectomy.

BACKGROUND

Lateral retroperitoneal laparoscopic adrenalectomy (LRLA) is widely performed for the resection of adrenal disorders, but when larger and more malignant lesions are involved, the difficulty of LRLA increases. We aimed to develop and evaluate a predictive model for the surgical difficulty of LRLA.

METHODS

A retrospective, observational, single-center study was performed involving all consecutive cases of unilateral RLA for adrenal disease from 2012.01 to 2021.12. Cases were randomly divided into training and validation cohorts (split ratio =7:3), then the least absolute shrinkage and selection operator (LASSO) regression was applied to reduce data dimension and select predictors. Multivariate logistic regression followed to develop the prediction nomogram for the surgical difficulty of LRLA. Finally, receiver operating characteristic (ROC) curve, calibration curve plot and decision curve analysis (DCA) were used to evaluate the nomogram's discrimination, calibration, and clinical usefulness, respectively.

RESULTS

A total of 621 cases were enrolled with a median age of 53 years and a median tumor diameter of 1.7 cm. After LASSO regression analysis, surgeon's experience, tumor diameter, resection procedure, histological type, patient's sex and body mass index (BMI) were identified to establish the nomogram. The model displayed good discrimination with area under the curve (AUC) in both the training cohort (0.754, 95% CI: 0.701-0.806) and validation cohort (0.742, 95% CI: 0.646-0.838). Additionally, excellent calibration curves were revealed for surgical difficulty evaluation in both the training cohort (P=0.999) and validation cohort (P=0.444). DCA results indicated the prediction model was clinically useful.

CONCLUSIONS

Our novel and effective predictive model can be used to assess the individual surgical difficulty of LRLA. By stratifying patients at risk of having a difficult LRLA for adrenal disease, the model could contribute to improvements in perioperative strategy and therapy.