Tissue Memory Score captures proximal tubular identity loss and independently stratifies survival in clear cell renal...

In clear cell renal cell carcinoma (ccRCC), stage dominates prognosis, and grading of tubular dedifferentiation depends on observer judgment. We evaluated the Tissue Memory Score (TMS), a continuous measure of retained proximal tubular identity, for independent prognosis, biological validity, platform robustness, and replication.

TMS scores a tumor's proximity to the kidney centroid in a locked principal-component space of 2,694 healthy GTEx samples; higher values denote greater kidney identity. The prespecified cohort was TCGA-KIRC (n = 540); a rank-based variant was transferred between platforms. Multivariable Cox regression adjusted for stage, purity, and mutational burden tested the overall-survival association, with a disease-specific analysis. Biological validity used ClearCode34, renal markers, ISUP grade, and BAP1, PBRM1, and SETD2 status. External validation used CPTAC-3 and E-MTAB-1980, pooled by random-effects meta-analysis.

TMS was independently associated with overall survival (HR 0.595 per SD; P = 6.6 × 10^-5) and improved discrimination over stage (paired ΔC-index 0.0189). The rank-based score reproduced it (Spearman ρ = 0.976), and the association was stronger for disease-specific survival (hazard ratio [HR] 0.43). TMS was higher in ClearCode34 ccA than ccB tumors (Cohen d = 0.93), fell with grade, and separated BAP1-mutant (lower) from PBRM1-mutant (higher) tumors. It replicated in E-MTAB-1980 (HR 0.51; 0.63 stage-adjusted) and was concordant in CPTAC-3 (HR 0.78). In stage II-III disease, 5-year survival differed by 17.3 points by TMS tertile (adjusted HR 0.56).

TMS is a platform-robust, biologically anchored measure of proximal tubular identity loss that replicated externally and warrants prospective evaluation for post-nephrectomy ccRCC risk stratification.