AI-generated research brief — verify at source
GRANT Score Validated Prospectively for RCC Recurrence and Survival Risk
In a prospective validation analysis of 714 patients enrolled in the phase III PROSPER trial, the GRANT prognostic score stratified renal cell carcinoma patients into meaningfully distinct survival groups, with unfavorable-risk patients facing roughly three times the hazard of relapse (HR 0.36, 95% CI 0.27–0.48) and four times the hazard of death (HR 0.25, 95% CI 0.15–0.42) compared to favorable-risk patients. Discriminatory performance was modest overall but substantially stronger in patients with nonclear cell histology, where the C-index reached 0.74 for both endpoints.
What Was Studied
Researchers sought to externally validate the GRANT score — a guideline-endorsed prognostic model incorporating tumor grade, patient age, lymph node status, and tumor size — in a prospective, randomized trial setting. Robust external validation is essential before prognostic tools can be reliably used to guide post-surgical risk stratification for patients with renal cell carcinoma (RCC).
How It Was Studied
This analysis used patient-level data from the EA8143 PROSPER trial, a phase III randomized study evaluating perioperative nivolumab in surgically treated RCC, registered on ClinicalTrials.gov (NCT03055013). A total of 714 patients were classified into two GRANT risk groups: favorable (0–1 risk factors) and unfavorable (2–4 risk factors). The primary outcomes examined were relapse-free survival (RFS) and overall survival (OS), estimated using the Kaplan-Meier method. Model discrimination was quantified using Harrell’s C-index, and subgroup analyses were performed by histological subtype, notably separating clear cell from nonclear cell RCC.
What Was Observed
- Favorable-risk patients had a substantially longer median RFS compared to unfavorable-risk patients (61.1 vs. 36.9 months), corresponding to about a 64% lower hazard of relapse in the favorable group (HR 0.36, 95% CI 0.27–0.48). This difference was highly statistically significant, meaning it is very unlikely to be due to chance alone (P < .001).
- The survival benefit was even more pronounced for OS: median OS was not yet reached in the favorable-risk group, while unfavorable-risk patients showed markedly worse outcomes, reflecting approximately a 75% lower hazard of death in the favorable group (HR 0.25, 95% CI 0.15–0.42, P < .001).
- Overall model discrimination was modest across the full cohort, with C-index values of 0.63 for RFS and 0.66 for OS — levels that indicate moderate but imperfect ability to rank-order individual patient risk. These values are above chance (0.50) but fall short of strong discrimination (typically above 0.70).
- In the nonclear cell RCC subgroup, discriminatory performance improved meaningfully, reaching a C-index of 0.74 for both RFS and OS, with hazard ratios suggesting approximately 87% lower risk of relapse (HR 0.13, 95% CI 0.05–0.33) and 86% lower risk of death (HR 0.14, 95% CI 0.04–0.50) in the favorable group — though the wide confidence intervals for OS reflect smaller subgroup numbers and reduced precision.
Why This Matters
External prospective validation in an independent trial population is a critical step in establishing that a prognostic tool performs consistently beyond its development cohort. The GRANT score’s confirmation across both survival endpoints strengthens its standing among guideline-recommended tools for post-nephrectomy RCC risk stratification. The notably stronger performance in nonclear cell RCC is a clinically and scientifically relevant finding, as this histological subgroup is frequently underrepresented in prognostic model research and has fewer validated risk tools available.
How to Read This Result
While the prospective design and sizable cohort enhance confidence in these findings, the GRANT score’s modest overall C-index suggests it captures meaningful prognostic signal but does not finely discriminate individual patient risk, particularly in the more heterogeneous clear cell population.
Limitations
The abstract does not explicitly report study limitations.