Surgical outcomes and postoperative hemorrhage risk of percutaneous nephrolithotomy (PCNL) for deer horn shaped stones analyzed by Lasso regression.

OBJECTIVE

To predict surgical outcomes and postoperative hemorrhage risk for percutaneous nephrolithotomy (PCNL) in cases of staghorn-shaped stones using lasso regression.

METHODS

We collected data from 104 patients with staghorn-shaped stones treated with PCNL between January 2019 and December 2022 at the Department of Urology Surgery, the Third People's Hospital of Gansu Province. Medical history, stone-related parameters, and lab test data were collected. Patients were categorized into stone clearance or residual groups based on postoperative stone status, and bleeding or non-bleeding groups based on post-surgery blood transfusion. The lasso model's predictive ability for post-PCNL Stone Free Rate (SFR) and hemorrhage risk was evaluated using ROC curves. The lasso model's predictive performance for post-PCNL SFR was compared to the S.T.O.N.E. score.

RESULTS

Overall stone clearance rate was 59.29%. The lasso model identified hypertension history, calyx count at stone location, prior calyx surgeries, age, operation duration, and pre-op creatinine level as SFR predictors. The AUC of lasso model (0.867) significantly surpassed the S.T.O.N.E. model (0.748) (P=0.006) in predicting post-PCNL SFR. In addition, the AUC of lasso model in predicting the risk of postoperative bleeding was 0.779, suggesting an ability in the prediction of bleeding occurrence.

CONCLUSION

A predictive model utilizing lasso algorithm was successfully established. It effectively predicts stone clearance rate and bleeding risk after PCNL for staghorn shaped kidney stones.