ANN ARBOR, MICH—Gene-expression profiles can be used to predict risk in patients with non–small-cell lung cancer (NSCLC), new research has shown. Currently, about 25% to 30% of patients have stage I disease and are treated with surgical intervention alone. Of these, 35% to 50% will suffer a recurrence within five years. David G. Beer, PhD, and colleagues found that gene profiles can identify stage I patients who may benefit from additional therapy.[1]

“This type of research involves the collaboration of many people who have expertise in different areas, such as cancer biology, bioinformatics, surgery, pathology, and oncology,” said Dr. Beer, a Professor of Surgery and Radiation Oncology at the University of Michigan. Using oligonucleotide microarrays, which can simultaneously analyze the expression of thousands of genes, the investigators generated gene-expression profiles for 86 primary lung adenocarcinomas—67 stage I tumors and 19 stage III tumors—as well as for 10 noncancerous lung samples. Hierarchical clustering with the resulting 4,966 genes produced three clusters of tumors.

Cluster 1 contained all of the noncancerous samples and the greatest percentage of well-differentiated tumors (42.8%). Cluster 3 contained the highest percentages of both poorly differentiated and stage III tumors (47.6% and 42.8%, respectively). However, 11 stage I tumors were present in Cluster 3, suggesting a common gene-expression profile for some stage I and stage III tumors. Cluster 2 was an intermediate group with both well- and poorly differentiated tumors.

Once the clustering was completed, the investigators found that 967 of 4,966 genes differed significantly between stage I and stage III tumors. To verify the microarray expression, mRNA from 20 samples was examined with probes for three arbitrarily selected genes: insulin-like growth factor binding protein 3 (IGFBP3), cystatin C, and lactate dehydrogenase-A (LDH-A). IGFBP3 and LDH-A mRNA levels increased from stage I to stage III tumors and were higher than those in normal lung samples.

A risk index of 50 genes that had the best overall association with survival was developed. Cross-validation of this index was performed using gene-expression data from a separate, independent sample of 84 lung tumors. New samples were assigned risk status based on cutoff points identical to those in the original samples. Among 62 stage I tumors, high- and low-risk groups that differed significantly in survival were identified.

Further research will be needed before a risk index can be used clinically, said Dr. Beer, but he is hopeful about the outcome of such research. “Using the expression profile of resected tumors to help identify appropriate patient therapy or those patients who require additional therapy may very well be something we might see in clinical medicine.” He added that this will also depend on whether relatively low-cost, rapid, and accurate methods for gene measurement can be developed.

Dr. Beer said that he and his colleagues will soon be undertaking an additional, larger study “to expand this research and, hopefully, refine the risk index for the most predictive genes in identifying high-risk stage I lung adenocarcinoma patients.”

—Gale Jurasek

References
1. Beer DG, Kardia SLR, Huang C-C, et al. Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med. 2002;8:816-824.

Return to table of contents