Issue of
February 9, 2000

Lymphochip demonstrates power of gene profiling to diagnose human cancers

BY KRISTIN WEIDENBACH

Stanford researchers studying thousands of genes on a workbench the size of a penny have found that a particular type of B-cell lymphoma is actually two diseases rather than one. The experiments mark one of the first clinical applications of the so-called DNA chip and demonstrate the enormous power of gene profiling of human cancers.

The researchers used a specially designed DNA chip that they named the Lymphochip. They made the chip by spotting more than 12,000 genes in a grid pattern onto a small piece of glass. Most of the genes that they chose for the Lymphochip are expressed in B-lymphocytes, one of the predominant classes of white blood cells. When the researchers added DNA from healthy B cells to the chip, they could see which genes were switched on and which were lying dormant. By following the activity of the genes they could build a biological picture of the gene pattern of a normal B cell.

Researchers in the laboratory of Patrick Brown, MD, PhD, a Howard Hughes investigator in the biochemistry department, and their colleagues in the laboratory of Louis Staudt, MD, PhD, at the National Cancer Institute, then added DNA from three different B-lymphocyte cancers to the Lymphochip. According to Ash Alizadeh, a sixth year MD/PhD student in Brown's lab, chronic lymphocytic leukemia (CLL) is the most common adult leukemia, and follicular lymphoma (FL) and diffuse large B cell lymphoma (DLBCL) are the most common kinds of non-Hodgkin's lymphoma. A pathologist can easily diagnose these different leukemias, and researchers in the Brown lab were also easily able to tell them apart via their gene patterns.

The researchers decided to launch their study with B-cell cancers because so much is known about the normal B cell. "Lymphoma may be a relatively non-prevalent cancer, accounting for less than 10 percent of all malignancies," said Alizadeh. "But the bells and whistles of the cells from which lymphomas are derived are pretty well known. Since normal B cells have been so extensively studied, B cell cancers are a very attractive model for studying cancer," he said.

The experiments confirmed that the three different cancers could be molecularly diagnosed; each B-cell disease had a distinctive gene pattern or signature. But the researchers also detected that the DLBCL appeared to split into two subgroups.

Alizadeh said physicians had long suspected that DLBCL contains two separate diseases because people diagnosed with the disease usually fall into a high or low risk group based on clinical indicators such as their age and the extent and location of their disease. But pathologists, looking for changes in the patients' B cells, have found it difficult to classify two different diseases. Being able to conclusively assign patients to a high or low risk group could allow physicians to better tailor treatment for each patient. For example, low-risk patients could potentially be spared an invasive procedure such as bone marrow transplantation, says Alizadeh.

When Alizadeh and colleagues scrutinized the samples of DLBCL DNA grouped by their gene expression patterns and compared them to the long-term survival of the 40 patients in the study, they found one pattern correlated with a high chance of survival beyond five years and the other pattern correlated with a low chance of survival. The researchers had discovered specific gene profiles indicative of each type of
DLBCL. They named the two groups germinal center B cell-like DLBCL and activated B cell-like DLBCL, according to the stage of maturity of a normal B cell that the two malignant cell types were most similar to.

Of the 40 patients in the DLBCL study, 24 were in the low clinical risk group and the remainder were at high clinical risk. But using their new classification system, Alizadeh and his colleagues were able to further subtype the 24 low-risk patients. Six of the 10 patients who had the activated B cell-like DLBCL had died, whereas only three of the 14 patients who had the germinal center B cell-like DLBCL had died, suggesting that patients with germinal center B cell-like DLBCL have a better chance for survival.

"This study demonstrates that a genomic view of gene expression in cancer can bring clarity to previously muddy diagnostic categories," the authors write in the report of their study, which was published in the February 3 issue of Nature. Alizadeh is a co-lead author of the study with Michael Eisen, PhD, now at the University of California, Berkeley, who contributed to the study when he was a research fellow in the laboratory of David Botstein, PhD, professor of genetics. Brown and Staudt are the senior authors.

Other Stanford authors include Botstein; Lisheng Lu, PhD, a postdoctoral fellow in Brown's lab; Ron Levy, MD, professor of medicine (oncology); Izidore Lossos, MD, a clinical fellow in Levy's lab; Gavin Sherlock, PhD, a research fellow in Bostein's lab; David Lewis, MD, associate professor of pediatrics (immunology and transplantation biology); Robert Tibshirani, PhD, professor of health research and policy (biostatistics); and Roger Warnke, MD, professor of pathology.

The research at Stanford was supported by the National Cancer Institute, the Howard Hughes Medical Institute, the Alfred E. Sloan Foundation and Stanford's Medical Scientist Training Program. SR