Abstract
Aberrant cell proliferation is a fundamental feature of cancer. It is thus not unexpected that many studies have been devoted to the exploration of tumour cell proliferation as a potential indicator of outcome. Indeed, in most malignancies, high expression of proliferation markers and more accurately, increased expression of proliferation-related genes have been strongly associated with poor outcome. In colorectal cancer (CRC), however, discordant results have been reported and the prognostic significance of cell proliferation has not been demonstrated in this type of cancer. As these results were mostly based on the subjective assessment of a single proliferation marker, this work set out to evaluate the association between the proliferative activity of CRCs and their malignant potential using an objective microarray-based multi-gene proliferation signature.
In the first step, a gene proliferation signature (OPS) was derived from analysis of a CRC cell line model. Ten CRC cell lines were harvested under semi- and fully-confluent conditions to obtain RNA from two stages that differed in their proliferative activity. Gene expression profiles of the two growth stages were analyzed on oligonucleotide arrays and a OPS was identified by gene ontology analysis of differentially expressed genes. In the second step, the performance of the signature to classify patients into prognostic groups was examined using two independent cohorts of primary CRC tumours (cohort A: 73 tumours in stages I-IV, cohort B: 55 tumours in stages I-II). Further, the signature was applied to a population of liver metastases to establish its association with the malignant potential of CRC. Finally, the capacity of the signature to detect clinically distinct CRC populations was compared with that of the proliferation marker Ki-67 in a classic immunohistochemical approach.
The OPS consisted of 38 mitotic cell cycle genes, which were over-expressed in actively cycling cells relative to growth-inhibited cells in vitro. Intriguingly, a reduced OPS expression was associated with (i) the presence of lymph node or distant metastasis in cohort A, (ii) an increased risk of recurrence, and (iii) shorter overall and recurrence-free survival in both cohorts of primary CRCs (p<0.05). While the association between the OPS and clinical outcome was not independent of the disease stage in cohort A, reduced OPS expression was an independent predictor of outcome in cohort B. Importantly, adjuvant therapy had no impact on this association. Furthermore, OPS expression was reduced in CRC liver metastases, confirming that decreased proliferation is an indicator of the malignant potential in CRC. While reduced proliferation in liver metastases was also observed by Ki-67 immunostaining, the classic proliferation marker was not able to stratify primary CRCs into
different prognostic groups.
To the best of our knowledge, this study is the first to report an association between the reduced expression of a multi-gene proliferation signature and poor outcome in cancer. This finding contradicts the long-held belief that increased proliferation is an indicator of tumour malignancy. In contrast to many other cancers, reduced proliferation appears to be a significant component of a biological signature associated with malignant potential of CRC tumours. Investigating the reasons why CRC differs from other cancer types may provide insights into important underlying biological mechanisms. If this association can be verified in larger cohorts, the GPS may have important clinical implication for identification of high-risk early stage CRC patients.