External validation of a tumor-derived 5-gene prognostic signature for recurrence (R) of stages I/II colon and stage I rectal cancer following potentially curative resection.

Abstract

437 Background: Optimizing post-operative clinical management for early-stage (I/II) colorectal cancer (CRC) patients (pts) is a significant unmet medical need. We hypothesized that a molecular prognostic test using primary CRC tissue would better predict the chances of tumor R within 36 months (mo) than current NCCN Clinical Practice Guidelines (NCCN). Methods: Pts had tumor R by 36 mo (n=46) or confirmed non-recurrence (NR) for ≥36 mo (n=69) after surgery; none had received neoadjuvant or adjuvant therapy. Archival formalin-fixed paraffin-embedded primary adenocarcinoma tissues (median storage 7 years; range 4-15) obtained at initial surgical resection with curative intent were retrieved for 86 stage I/II (pT1-4 pN0 M0) colon cancer and 29 stage I (pT1-2, pN0 M0) rectal cancer pts from 2 US and 2 European sites that were different from those previously used to generate the molecular test. Tumor gene expression was assessed by qRT-PCR with custom 384- well TaqMan Low Density Arrays (Applied Biosystems) using RNA that had satisfied a set of rigorous quality control parameters. Results: For stages I/II CRC (n=115), the dichotomous rule correctly classified 32/46 R and 38/69 NR pts: sensitivity (S) 0.70, specificity (SP) 0.55. High-risk pts had a significantly higher probability of R by 36 mo than low-risk pts: PPV 0.51, NPV 0.73; hazard ratio (HR) 2.06 (95% CI: 1.10 to 3.86; p=0.020). NCCN (V.1.2011) was not able to differentiate 36-mo R vs NR in this population: S 0.72, SP 0.42, PPV 0.45, NPV 0.69; HR 1.38 (95% CI: 0.73 to 2.63; p=0.315). The SP of the molecular test was significantly greater than that for NCCN (p=0.05). For stage I pts (n=29; 13R, 16NR), the prognostic accuracy of the test (0.79; 23/29) surpassed that for NCCN (0.55; 16/29). Conclusions: External validation of a 5-gene prognostic rule confirmed its ability to differentiate early stage CRC pts at high risk vs low risk for R within 36 mo after surgery better than current NCCN Guidelines, especially for stage I. The improved specificity and high sensitivity of the molecular test confirm its potential utility for optimizing post-op clinical management of early stage CRC. [Table: see text]