In a large group of diffuse large-B-cell lymphomas (DLBCL), DNA microarrays identified three patterns of gene expression that were correlated with the likelihood of survival after chemotherapy. Individual genes within these patterns formed molecular signatures that had an even stronger correlation with survival after chemotherapy. The predictive power of the molecular signatures was independent of the international prognostic index.
From 12,000 to 17 genes
DNA microarrays reflect the activity of tens of thousands of genes in a sample of tissue. This study of the most common lymphoma in adults focused on more than 12,000 genes expressed by lymphoid tissue and found 17 that were strongly related to the outcome. This work is an example of how microarray technology is leading to clinically useful insights into the molecular genetics of cancer.
"We're able to reliably predict the survival of these patients using data from a small number of genes, indicating that this technique should be entirely manageable for routine use," said National Cancer Institute (NCI) investigator Elaine M. Hurt, Ph.D., one of the authors on the study.
DLBCL is the most common type of non-Hodgkin's lymphoma in adults. Standard chemotherapy for the disease is effective in only 40 percent of patients. Profiling gene expression in patients' tumors may help clinicians decide which patients are suitable candidates for standard therapy and which should consider other options for treatment.
Discovery relied on microarrays
The discovery of the predictive genes relied on DNA microarray technology, which allows researchers to determine which genes are active within cells. Microarrays, also known as gene chips, are glass slides that have been coated with thousands of spots of DNA, each representing a different gene. When a gene is active in a cell, it produces RNA copies known as transcripts. To measure the activity of genes, researchers use the RNA transcripts to make a fluorescent gene probe. When these gene probes are allowed to bind to their corresponding DNA spot on the chip, those spots on the chip light up. Scientists use the pattern and intensity of light emitted to determine the activity of each of the chip's thousands of genes.
For this study, researchers used the Lymphochip, a specialized microarray containing 12,000 DNA spots representing genes expressed in normal and malignant lymphoid cells. Developed as part of the NCI's Cancer Genome Anatomy Project, the Lymphochip is particularly useful for finding differences in gene expression among lymphoid cancers.
The researchers profiled gene expression in 240 tumor biopsies from european and american patients with DLBCL and identified more than 600 genes whose expression varied significantly between patients who had responded well to treatment and those whose response was poor. These genes highlight aspects of the tumors that affected response to therapy, including how fast tumor cells were dividing and from what type of normal lymphocyte (a type of white blood cell) the tumor originated. Many of the predictive genes suggest that a patient's immune response to the tumor is important for achieving a cure with chemotherapy.
Focusing on genes where the difference in expression was most dramatic between the two groups of patients, researchers narrowed the key genes down to 17. From these genes, the investigators created a formula that could be used to predict survival following chemotherapy. This predictor classified the patients into four groups of equal size. The five-year survival rates for these groups were 73 percent, 71 percent, 34 percent, and 15 percent.
Currently, physicians rely on the International Prognostic Index (IPI) to evaluate patients with DLBCL. This predictive index is based on clinical factors including age, stage of the tumor, and the presence of disease that has spread outside the point of origin. While useful for some purposes, Hurt and her collegues noted that the IPI has not been successful in identifying the best candidates for alternate therapies. "Based on variations in gene expression, we can now do a better job of predicting patient outcomes," she said.
As an example, Hurt explained that 32 of the 240 patients in this study were classified in the group with the poorest prognosis according to the IPI. Of these, four were in fact cured by standard chemotherapy. Gene expression profiling successfully identified each of these.
To identify drugs for subgroups of patients
For those that don't respond to chemotherapy, alternatives are available. "For half of the patients with diffuse large B-cell lymphoma, conventional chemotherapy appears to be a reasonable option, but for patients in the poor-risk group, we have to consider other therapies," Hurt said. One possibility for some patients would be a bone marrow transplant. There are also numerous clinical trials for which these patients may be eligible.
Trials designed to correlate clinical results with molecular data will allow researchers to identify drugs that are effective in subgroups of cancer patients, an approach that has already proven effective in finding new agents to treat breast cancer and leukemia. Hurt said gene profiling will make it possible to obtain more information from clinical trials in the future.
This research was sponsored by NCI as part of the Lymphoma/Leukemia Molecular Profiling Project and the NCI Director's Challenge. The participating institutions included these:
USA: University of Nebraska Medical Center, Omaha; the Southwest Oncology Group; and the NCI Center for Cancer Research, Bethesda, Md.
Canada: the British Columbia Cancer Agency, Vancouver
Norway: the Norwegian Radium Hospital, Oslo
Germany: the University of Wuerzburg
Spain: the University of Barcelona
Published in NEJM 20 June. Andreas Rosenwald, M.D., George Wright, Ph.D., Wing C. Chan, M.D., Joseph M. Connors, M.D., Elias Campo, M.D., Richard I. Fisher, M.D., Randy D. Gascoyne, M.D., H. Konrad Muller-Hermelink, M.D., Erlend B. Smeland, M.D., Ph.D., Jena M. Giltnane, B.S., Elaine M. Hurt, Ph.D. NIH, Hong Zhao, M.S., Lauren Averett, B.A., Liming Yang, Ph.D., Wyndham H. Wilson, M.D., Ph.D., Elaine S. Jaffe, M.D., Richard Simon, D.Sc., Richard D. Klausner, M.D., John Powell, M.S., Patricia L. Duffey, R.N., Dan L. Longo, M.D., Timothy C. Greiner, M.D., Dennis D. Weisenburger, M.D., Warren G. Sanger, Ph.D., Bhavana J. Dave, Ph.D., James C. Lynch, Ph.D., Julie Vose, M.D., James O. Armitage, M.D., Emilio Montserrat, M.D., Armando López-Guillermo, M.D., Thomas M. Grogan, M.D., Thomas P. Miller, M.D., Michel LeBlanc, Ph.D., German Ott, M.D., Stein Kvaloy, M.D., Ph.D., Jan Delabie, M.D., Ph.D., Harald Holte, M.D., Ph.D., Peter Krajci, M.D., Ph.D., Trond Stokke, Ph.D., Louis M. Staudt, M.D., Ph.D., for the Lymphoma/Leukemia Molecular Profiling Project.
For information on clinical trials, go to http://cancer.gov/clinical_trials/
18th UICC International Cancer Congress: www.oslo2002.org
NCI Press Office 301-496-6641
NIH/National Cancer Institute