MP47-09: The added value of histological subtype in the prediction of oncologic outcomes in patients with non-metastatic papillary renal cell carcinoma

Giuseppe Rosiello, Gianfranco Baiamonte*, Giuseppe Fallara, Giuseppe Basile, Roberta Lucianò, Maurizio Colecchia, Alberto Martini, Daniele Cignoli, Luigi Nocera, Federico Belladelli, Gianmarco Colandrea, Daniela Canibus, Chiara Re, Giacomo Musso, Francesco Cei, Alberto Briganti, Roberto Bertini, Andrea Necchi, Daniele Raggi, Milan, Italy, Pierre I. Karakiewicz, Montreal, Canada, Francesco Montorsi, Andrea Salonia, Alessandro Larcher, Umberto Capitanio, Milan, Italy

Introduction: The accuracy of VENUSS prognostic model to assess oncologic outcomes in patients harbouring papillary renal cell carcinoma (pRCC) has been previously validated. However, papillary subtypes (type I vs. type II) are historically associated with different survival outcomes. We analyzed the added value of histological subtype in predicting oncologic outcomes in patients with non-metastatic pRCC.

Methods: Within a prospectively maintained database, patients surgically treated for non-metastatic pRCC from 2000 to 2021 at a single centre were included. Primary endpoints were clinical progression (CP) and cancer-specific mortality (CSM). Kaplan-Meier plots estimated 5-year CP and CSM. Multivariable Cox-regression models (MVA) investigated predictors of CP and CSM. Concordance index (c-index) assessed the performance of the models. Bootstrapping and calibration plots were used to compare predicted probabilities at 5 years. Finally, decision-curve analyses (DCA) were applied to determine the net benefit of the derived models over the threshold probabilities.

Results: Out of 403 pRCC patients, 227 (56%) were pRCC type I and 176 (44%) type II. According to the VENUSS classification, 277 (69%) vs. 88 (22%) vs. 38 (9%) patients were classified as low- vs. intermediate- vs. high-risk. Within a median follow-up of 60 months, 37 (9%) patients experienced CP and 25 (6.2%) died due to RCC. 5-year CP and CSM resulted 1.6% (95%CI 0-3.3%) and 2.1% (95%CI 0-4%) in low-risk, 9.6% (95%CI 1.1-17%) and 6.4% (95%CI 0-13%) in intermediate-risk, 65% (95%CI 41-80%) and 57% (95%CI 29-74%) in high-risk patients, respectively. At MVA, pRCC type II emerged as an independent predictor of CP (HR 2.2, 95%CI 1.2-4.2; p=0.01), and CSM (HR 1.7, 95%CI 1.1-2.3; p=0.01). Using bootstrapping for internal validation, c-index at 5-year for the VENUSS score was 84% and 82%, for CP and CSM, respectively. However, when papillary subtype was incorporated, the prognostic model significantly increased accuracy (91% for 5-year CP and 90% for CSM). DCA demonstrated a net benefit in the use of the VENUSS score, although favoring the prognostic model which included also histological subtype.

Conclusions: We confirmed the clinical utility of the VENUSS prognostic model in case of pRCC. However, since patients harbouring pRCC type II are at significant higher risk of unfavorable oncologic outcomes, this histological subtype classification should be incorporated, as well.

Source of Funding: no