A new prognostic marker has shown potential to predict recurrence in certain colon cancer patients

Apr 29 2004

A new prognostic marker has shown potential to predict recurrence in certain colon cancer patients, according to a study published in the May 1st issue of the Journal of Clinical Oncology.

In the study, a team of researchers led by Veridex, LLC, a Johnson & Johnson company, used gene expression profiling to identify the new marker – a 23-gene signature – which may predict disease recurrence in patients initially diagnosed with Dukes' B colon cancer.

The study marks the first significant application of genomics in searching for molecular markers for colon cancer prognosis and is also the subject of an editorial in the Journal.

While the standard of care for patients with Dukes' B, or Stage II, colon cancer – surgical resection of the diseased tissue – is considered highly effective, up to 20-30 percent of these patients can later develop recurrence and die from the disease. Patients classified as Dukes' B are thought to only have local disease, however, the incidence of relapse indicates existence of undetected disseminated disease and points to the need for more accurate methods for predicting recurrence.

"The key to increasing survival of these patients lies in the ability to accurately identify which of them is at high risk of relapse," says David Atkins, Ph.D., General Manager, Molecular Diagnostics for Veridex and principal investigator on the study. "The hypothesis that we're testing is that all of the information we need to characterize the disease can be found in the primary tumor."

The study involved the analysis of RNA samples from banked tumors from 74 Dukes' B colon cancer patients. Each sample was analyzed using several highly parallel methods for analysis of gene expression. Thirty-one of these patients developed tumor relapse in less than three years, while 43 patients remained disease-free for more than three years after surgery. Armed with this information, two approaches were used to identify the genetic markers that can best discriminate between these two groups of patients, ultimately reaching the final 23-gene combination.

The signature was then validated in 36 independent patients, accurately predicting 13 of 18 relapse patients and 15 of 18 disease-free patients, for an overall performance accuracy of 78 percent – a rate unprecedented for this type of test in colon cancer, says Atkins.

Additional studies will be needed, Atkins cautions, but if validated this prognostic marker could have broad implications, including, potentially providing guidance in recommending adjuvant chemotherapy for certain Dukes' B patients. Furthermore, the ability to stratify this patient class into different risk groups could be a useful tool for aiding the development of new, targeted therapies for the disease.

"It is well known that cancer is not a homogeneous disease; it can be broken down into still more categories or stages than are currently standard," adds Atkins. "The development of a prognostic assay to test for this genetic signature should allow physicians to better identify patients with a poorer prognosis and plan for more appropriate, even more aggressive, therapies. The data from this study shows that such an assay is highly feasible."