Download Relationship Between Tumor Gene Expression and

Relationship between tumor gene expression and recurrence in an observational
cohort of patients with stage II/III colon cancer treated with surgery only:
Quantitative RT-PCR assay of 375 genes in fixed paraffin-embedded (FPE) tissue
I. Lavery, J. Hammel, J. W. Cowens, K. M. Clark-Langone, J. Hackett, F. Baehner, G.
Yothers, M. J. O'Connell, N. Wolmark
Cleveland Clinic, Cleveland, Ohio; Genomic Health, Inc., Redwood City, CA; ; NSABP
Biostatistical Center, Univ. of Pittsburgh, Pittsburgh, PA; National Surgical Adjuvant
Breast & Bowel Project, Pittsburgh, PA.
Background: High throughput RT-PCR, which has been used to identify genes
associated with colon cancer recurrence in NSABP C-01/C-02 (Proc ASCO, 2006), was
carried out in an observational cohort of patients treated with surgery alone at CCF. We
are reporting genes whose expression predicts colon cancer recurrence in both of these
populations.
Methods: RNA was extracted from three manually micro-dissected 10-micron sections
of FPE tumor tissue obtained at the initial diagnosis from patients treated at CCF between
1981 and 2000. RNA expression was quantified for 375 cancer-related and reference
genes using RT-PCR.
Results: Blocks from 765 patients (504 stage II and 261 stage III) were evaluable. Of 105
genes significantly associated with recurrence-free interval in the CCF study (Cox
proportional hazards regression), 59 were also significant in the NSABP study. The
relative risk of recurrence (hazard) associated with expression levels of these 59 genes
was similar in both studies with and without adjustment for possible confounders. In both
studies, the range of individual gene expression was associated with up to an 11 fold
difference in risk. Biological pathways represented include proliferation, cell cycle,
stromal response, early response, and invasion.
Conclusions: Quantitative RT-PCR of FPE colon cancer tissue applied to two large
independent populations has identified 59 genes (at most 2 expected by chance) that are
prognostic for stage II and III colon cancer. A multi-gene prognostic algorithm will be
developed based largely on a subset of these 59 genes and will be validated in an
independent study.