Analysis of Specific Transcriptional Regulators as Early Predictors of Independent Prognostic Relevance in Resected Colorectal Cancer

Purpose: Prognostic studies on transcription factors acting at specific promoter elements have never been done so far. However, in tumors with long necessary follow-up, such as colorectal cancer, early-risk predictors would be needed. The invasion-related gene u-PAR is regulated via an activator protein 2 (AP-2)/Sp1 (−152/−135) and an AP-1 binding promoter motif (−190/−171), mediating u-PAR induction by K-Ras and Src. The present study was done to give first evidence for early prognostic relevance of transcription factors differentially bound to the u-PAR promoter, and their molecular inducers, in colorectal cancer. Experimental Design: Tumor/normal tissues of 92 prospectively followed (median = 26.3 months) patients were analyzed for Src activity/protein, K-ras mutations, and transcription factor binding to both u-PAR promoter motifs ( in vivo gel shift, kinase assay, and PCR). Results: Kaplan-Meier/Mantel-Cox analysis showed a significant correlation among elevated Sp1/Sp3 binding to region −152/−135 ( P = 0.002 and P = 0.006), the combinations of Sp1/AP-2 and Sp1/AP-1 binding to both motifs ( P = 0.010 and P = 0.005), and Sp1 binding/high Src protein in tumors ( P < 0.001), with poor survival. Survival decreased with the number of bound transcription factors to both motifs, with binding of three factors defining a high-risk group ( P = 0.021). In multivariate analysis, elevated Sp1 binding, combinations of Sp1/AP-2 binding and Sp1/AP-1 binding, or Sp1 binding/high Src were independent prognostic variables; u-PAR expression itself being not yet prognostic. A first molecular staging model (CART) was defined, providing novel early high-risk groups (mean survival time as low as for non-curatively resected patients) from these variables. Conclusions: This study defines transcription factors acting at specific promoter elements of an invasion-related gene, mediating specific signaling, as novel, independent, early predictors of prognosis in colorectal cancer.