Background:
The directed study of the functional proteome in colorectal cancer (CRC) has identified critical protein markers and signaling pathways. However, the prognostic relevance of many of these proteins remains unclear.
Methods:
We determined the prognostic implications of the functional proteome on 263 CRC tumor samples from patients treated at MD Anderson Cancer Center (MDACC) and 462 patients from The Cancer Genome Atlas (TCGA) to identify patterns of protein expression that drive tumorigenesis. 163 validated proteins were analyzed by RPPA. Unsupervised hierarchical clustering of the tumor proteins from MDACC was performed. Clustering was validated using RPPA data from TCGA CRC. Cox regression was used to identify predictors of tumor recurrence.
Results:
Clustering revealed dichotomization, with Subtype A notable for a high EMT protein signature while Subtype B was notable for high Akt/TSC/mTOR pathway components. Survival data was only available for the MDACC cohort and was used to evaluate prognostic relevance of these protein signatures. Group B demonstrated worse relapse-free survival (HR 2.11, 95% CI 1.04–4.27, p=0.039), though there was no difference in known genomic drivers between the two proteomic groups. Proteomic grouping and stage were significant predictors of recurrence on multivariate analysis. 8 proteins were found to be significant predictors of tumor recurrence on multivariate analysis: Collagen VI, FOXO3a, INPP4B, LcK, phospho-PEA15, phospho-PRAS40, Rad51, phospho-S6.
Conclusion:
CRC can be classified into distinct subtypes by proteomic features independent of common oncogenic driver mutations. Proteomic analysis has identified key biomarkers with prognostic importance, however these findings require further validation in an independent cohort.