An independent evaluation of an architectural-based scoring system for renal cell carcinoma risk-stratification

Clear cell renal cell carcinoma (ccRCC), the most common RCC type, is characterized by poor prognosis and heterogeneous morphology. Recently, a prognostic ccRCC architectural grading system was proposed. Here, we evaluated the prognostic value of this architectural score. FFPE tissues were obtained from the population-based Netherlands Cohort Study. Architectural score was analyzed from digital slides (n = 333) by two independent researchers. Clinicopathological data were collected from Dutch databases. Kaplan-Meier analyses and Cox proportional hazard models were used to evaluate the association between architectural score and cause-specific survival (CSS), and models were compared with ROC curves, AICs and c-statistics. Architectural score was associated with all clinicopathological variables, besides age at diagnosis and sex, and was associated with CSS in univariate analyses (HR_{score 3}=2.70, HR_{score 4}=4.06, both P < 0.001). However, in multivariate analyses (including age at diagnosis, sex, WHO/ISUP grade, TNM stage and tumor size), architectural score did not predict survival. Model fit was similar for models containing the architectural score and the 2022 WHO/ISUP grade (c-statistics: 0.69-0.73). Nevertheless, the architectural score was possibly found to be a significant predictor of survival in lower-grade WHO/ISUP tumors in post hoc analyses (P = 0.008), but not in higher grade WHO/ISUP tumors. Although architectural scoring could be used to identify patients with a poor CSS, these effects are not independent of other established prognostic factors. However, architectural scoring might identify a high-risk subgroup that are low-grade WHO/ISUP tumors and might further refine ccRCC risk-stratification, yet, more research is needed to confirm this exploratory finding in larger cohorts.