News Release

Different gene-expression predictors
of breast cancer agree, UNC study shows

CHAPEL HILL -- Breast cancer researchers at the University of North Carolina at Chapel Hill have identified a number of activity patterns in the genes of individual tumors that make them biologically different from others. These findings could provide valuable clinical information such as how likely the tumors are to be invasive, how well they might respond to different treatments and how likely they are to recur or spread.

Currently, doctors treating patients with breast cancer make treatment decisions and predictions based largely on the location and size of the tumor and if the cancer has spread, or metastasized, to lymph nodes and distant sites of the body.

But not all patients who are similar in terms of these clinical indicators get the same benefits from treatment.

These new findings could remedy that situation. Such differences in gene activity may be used as biomarkers to identify which treatments can be individually matched.

Over the past five years, gene expression profiles have been identified that appear to be predictive for cancer patients, especially for breast cancer patients. But these tests show very little overlap in their gene lists, and thus it is not known just how distinct these different assays might be.

According to Dr. Charles M. Perou, assistant professor of genetics and pathology at the UNC School of Medicine and a member of the UNC Lineberger Comprehensive Cancer Center, some of the predictive assays are available commercially and others are under study in clinical trials in which treatment decisions, including whether or not to use chemotherapy, are being made based on them.

"An important and unanswered question, however, is whether these assays actually disagree or agree concerning outcome predictions for the individual patient," Perou said. "I think this is a very important point because if they disagree then it becomes difficult to determine which to use and when, and which are more robust and helpful."

To compare the individual predictions made by these different genomic tests, Perou and his colleagues at UNC and at The Netherlands Cancer Institute in Amsterdam, The Netherlands, studied the concordance of five different predictors that were all applied to a single data set of 295 tumor samples for which patient survival data was available - relapse-free survival and overall survival.

Writing in the Aug. 10 issue of the New England Journal of Medicine, the researchers note that four predictors showed "significant agreement" in their outcome predictions on individual breast cancer patients, despite having little gene overlap. Of the three predictors showing the greatest concordance, two were the main assays that are commercially available and being used to guide clinical trials.

"If one assay said this patient was going to do poorly, then so did the other two," Perou said, noting that although the two commercial assays overlapped each other only by one gene, they were in 80 percent agreement with each other.

"This is good news for breast cancer patients. It means that different groups have independently arrived at tests which agree with each other and that they all do add information not provided by existing clinical tests," Perou said.

For example, several of the predictors in this study appear to predict the likelihood of breast cancer recurrence in various populations of women with node-negative disease.

Such information would be useful for identifying women who are unlikely to experience recurrence and, thus, potentially unlikely to benefit from chemotherapy.

"We find our results encouraging and interpret them to mean that although different gene sets are being used, they are each tracking a common set of biological characteristics that are present across different breast cancers and are making similar outcome predictions," Perou said.

UNC co-authors along with Perou are Cheng Fan, research associate at the Lineberger Comprehensive Cancer Center; Daniel S. Oh, doctoral student in the department of genetics; and Dr. Andrew B. Nobel, associate professor in the department of statistics and operations research. Collaborators from The Netherlands Cancer Institute are Drs. Lodewyk Wessels, Britta Weigelt, Dimitry S. A. Nuyten, and Laura J. van't Veer.

The research was supported by funds from the National Cancer Institute Breast SPORE Program, the Breast Cancer Research Foundation, the Dutch Cancer Society, the Dutch National Genomics Initiative and the National Science Foundation.

-30-

Contact Perou at 919-843-5740 or cperou@med.unc.edu
School of Medicine contact: Les Lang, 919-843-9687 or llang@med.unc.edu or Stephanie Crayton, 919-966-2860 or scrayton@unch.unc.edu
UNC Lineberger contact: Dianne Shaw, 919-966-4752 or dgs@med.unc.edu