While microarray-based gene expression profiling has yielded promising results for diagnosis of certain cancers, the motley assembly of array technologies is hardly ready for the prime time clinical setting: There is as yet no standard protocol or platform for conducting these experiments that could be reproducibly and consistently repeated in the clinic, and in most cases researchers haven’t even agreed on the expression profiles that signal particular cancers.
In an effort to address this problem, Affymetrix and the Whitehead Institute Center for Genome Research recently announced a collaboration in which Whitehead researcher Todd Golub will lead a team of researchers in determining standardized ways to profile common cancers using Affymetrix gene chips. The 18-month project will focus on lymphoma and prostate cancer, according to Affymetrix.
For Golub, also a researcher in the pediatric oncology department at Boston's Dana-Farber Cancer Institute, the project follows several years of work, including 14 published papers, aiming to define gene expression profiles for various cancers using microarrays.
But for Affymetrix, a further objective is to ensure the standardization problem for clinical research is solved in its favor. This project “will help allow us to drive the gene chip platform from a very strong position in research into a clinical environment,” said Lisa Cowell, Affymetrix’s vice president of business development. “To be able to position our platform for the clinical application — that’s our goal.”
Cowell outlined the four main “prongs” of the project: The first or validation part, involves “being able to look at how you prepare your clinical samples prior to analysis.” The second part involves standardizing the data analysis. “At the Whitehead, they have done a lot of research in this area, looking at how you can standardize around the statistical science of analyzing expression data,” Cowell said. A third prong is the validation of some of the earlier research in microarray-based classification of cancers through follow-up studies, and finally, the study aims to determine the variables that would be involved if expression-based analysis were to be taken into clinical trials.
For the validation of earlier research, the group is likely to revisit the task of determining predictive gene expression profiles in diffuse large B-cell lymphoma. In two separate Nature papers, Golub’s group and one led by Ash Alizedah from Pat Brown’s lab at Stanford, found two different sets of gene expression markers for this disease. Golub used Affymetrix chips and Alizedah used cDNA arrays, however, so this project’s further validation of Golub’s results with Affymetrix arrays may not resolve this difference.
The Whitehead is planning to collaborate with clinical institutions in this work, although the identity of these instititutions was not disclosed at press time. Likely candidates are Dana-Farber, Golub's research home, and Brigham and Women's Hospital, Harvard Medical School, as researchers from the latter institution co-authored a recent paper with Golub, “Gene Expression Correlates of Clinical Prostate Cancer Behavior,” in the March 2002 issue of Cancer Cell.
Cancer Collaboration Cacophony?
This collaboration arrives amid other similar efforts. The International Genomics Consortium has recently begun assembling partners for its Expression Project in Oncology, expO, which has the stated goal of collecting 10,000 tumor tissue samples spanning a large number of cancer types, as well as 1,000 normal tissues, from various academic institutions over the next three years, then performing gene expression analyses on the samples, and making the data freely available to the public. Genetic security firm First Genetic Trust recently signed on as the bank for this project’s data, and 18 academic cancer centers — including Johns Hopkins and the MD Anderson Cancer Center — have also agreed to participate.
Additionally, John Quackenbush's group at TIGR has been analyzing data from hundreds of microarray experiments in different cancers, in an effort to come up with widely applicable predictive gene expression markers. Affymetrix, which has supported the expO effort in the past, does not see this project as clashing with the others. “They’re all part of a growing effort to look at standardization of expression profiles for a clinical application,” said Cowell. “What we have chosen to do with the Whitehead focuses in on prostate cancer and lymphoma. There are quite specific aims within the Whitehead research agreement.”