Identification of high-risk patients with clear cell renal cell carcinoma based on interphase-FISH. Br J Cancer. 2014;110:2537-2543. [PMID: 24667645 PMCID: PMC4021511 DOI: 10.1038/bjc.2014.159]

A self-supervised vision transformer to predict survival from histopathology in renal cell carcinoma

PURPOSE

To develop and validate an interpretable deep learning model to predict overall and disease-specific survival (OS/DSS) in clear cell renal cell carcinoma (ccRCC).

METHODS

Digitised haematoxylin and eosin-stained slides from The Cancer Genome Atlas were used as a training set for a vision transformer (ViT) to extract image features with a self-supervised model called DINO (self-distillation with no labels). Extracted features were used in Cox regression models to prognosticate OS and DSS. Kaplan-Meier for univariable evaluation and Cox regression analyses for multivariable evaluation of the DINO-ViT risk groups were performed for prediction of OS and DSS. For validation, a cohort from a tertiary care centre was used.

RESULTS

A significant risk stratification was achieved in univariable analysis for OS and DSS in the training (n = 443, log rank test, p < 0.01) and validation set (n = 266, p < 0.01). In multivariable analysis, including age, metastatic status, tumour size and grading, the DINO-ViT risk stratification was a significant predictor for OS (hazard ratio [HR] 3.03; 95%-confidence interval [95%-CI] 2.11-4.35; p < 0.01) and DSS (HR 4.90; 95%-CI 2.78-8.64; p < 0.01) in the training set but only for DSS in the validation set (HR 2.31; 95%-CI 1.15-4.65; p = 0.02). DINO-ViT visualisation showed that features were mainly extracted from nuclei, cytoplasm, and peritumoural stroma, demonstrating good interpretability.

CONCLUSION

The DINO-ViT can identify high-risk patients using histological images of ccRCC. This model might improve individual risk-adapted renal cancer therapy in the future.

[Biomarkers for renal cell tumours]

During the last three decades, renal tumours have become increasingly well differentiated on the basis of their histopathological and molecular features. This subtyping has increasingly impacted clinical practice because more therapeutic options are available in organ-confined and metastatic renal cell tumours. The knowledge of the underlying molecular alterations is essential to develop molecular targeted therapies and to select the most effective systemic therapy for each patient. This manuscript gives an overview of the molecular differentiation on the one hand, and on diagnostic, prognostic and predictive biomarkers on the other hand.

Combination drug regimens for metastatic clear cell renal cell carcinoma

Renal cell carcinomas (RCC) make up about 90% of kidney cancers, of which 80% are of the clear cell subtype. About 20% of patients are already metastatic at the time of diagnosis. Initial treatment is often cytoreductive nephrectomy, but systemic therapy is required for advanced RCC. Single agent targeted therapies are moderately toxic and only somewhat effective, leading to development of immunotherapies and combination therapies. This review identifies limitations of monotherapies for metastatic renal cell carcinoma, discusses recent advances in combination therapies, and highlights therapeutic options under development. The goal behind combining various modalities of systemic therapy is to potentiate a synergistic antitumor effect. However, combining targeted therapies may cause increased toxicity. The initial attempts to create therapeutic combinations based on inhibition of the vascular endothelial growth factor or mammalian target of rapamycin pathways were largely unsuccessful in achieving a profile of increased synergy without increased toxicity. To date, five combination therapies have been approved by the U.S. Food and Drug Administration, with the most recently approved therapies being a combination of checkpoint inhibition plus targeted therapy. Several other combination therapies are under development, including some in the phase 3 stage. The new wave of combination therapies for metastatic RCC has the potential to increase response rates and improve survival outcomes while maintaining tolerable side effect profiles.

Metastatic risk stratification of clear cell renal cell carcinoma patients based on genomic aberrations

Prognostic markers for the definition of the individual metastatic risk in renal cell carcinoma are still missing. The aim of our study was to establish a total number of specific aberrations (TNSA) genetic score as a new prognostic test for metastatic risk evaluation. Fluorescence in situ hybridization (FISH) was performed on isolated cell nuclei of 100 ccRCCs (50 M1/50 M0) and 100 FFPE sections (second cohort, 32 M1/68 M0). For each chromosomal region (1q21.3, 7q36.3, 9p21.3p24.1, 20q11.21q13.32) cut-off values were determined by receiver-operator curve (ROC)-curve analysis. TNSA was calculated based on the dichotomized specific CNVs. The prognostic significance of CNVs was proven by Cox and logistic regression. TNSA was the best predictor of metastasis and recurrence free survival in both cohorts. We derived an algorithm for risk stratification by combining TNSA and T-category, which increased the prognostic accuracy to 87% (specificity = 86%, sensitivity = 88%). This model divides patients into two risk groups with significantly different RFS, CSS, and OS (P = 3.8×10^-5 , P = 5×10^-6 and P = 3.57×10^-8 respectively). The genetic risk model was superior to Leibovich score and was able to identify patients with metachronous metastatic spread which were incorrectly classified as "low" or "intermediate risk." We present a new tool for individual risk stratification by combining genetic alterations with clinico-pathologic parameters. Interphase FISH proves to be a dependable method for prognostic evaluation in primary tumor tissue on isolated cell nuclei as well as on FFPE sections. Especially in organ-confined tumors the genetic score seems to be an important tool to identify patients at high risk for metastatic disease.