Abstract
<title>Abstract</title> <p>Purpose To evaluate the clinical value of the stone volume–density index (SVDI), derived from the product of stone volume and mean Hounsfield unit (HU), in predicting treatment failure and complication development following percutaneous nephrolithotomy (PNL). Methods Data from 336 patients who underwent conventional PNL were retrospectively analyzed. Patients were stratified into two separate cohorts based on stone-free status (n = 250 success vs. n = 86 failure) and complication status according to the modified Clavien–Dindo classification (n = 276 no complication vs. n = 60 complication). Stone size, area, volume, mean HU, and SVDI were measured on preoperative non-contrast computed tomography (NCCT). Areas under the curve (AUC) were calculated using receiver operating characteristic (ROC) analyses, and independent predictors were identified through multivariate logistic regression. Results The overall stone-free rate was 74.4%, and the complication rate was 17.9%. For predicting failure, SVDI yielded the highest AUC (0.771; cut-off ≥ 3817, sensitivity 83.7%, specificity 60.4%). SVDI demonstrated significantly superior discriminative ability compared with stone size (AUC = 0.728; p = 0.044) and mean HU (AUC = 0.592; p < 0.001, although the difference did not reach statistical significance against stone volume (AUC = 0.753; p = 0.184) or stone area (AUC = 0.750; p = 0.128). On multivariate analysis with log-transformed SVDI, multiple calyceal involvement (OR = 3.02), secondary case status (OR = 3.06), and log(SVDI) (OR = 1.82; p < 0.001) were independent predictors of failure. No stone parameter achieved adequate discriminative power for complications (all AUC < 0.60); only tubeless procedure with intraoperative double-J stent placement was an independent protective factor (OR = 0.17; p = 0.004). Conclusions SVDI is a single composite parameter that demonstrated the highest discriminative performance among all stone parameters for predicting PNL failure, with statistically significant superiority over stone size and mean HU. The predictive value of stone parameters for complications remains limited.</p>