Blood test identified patients with lung cancer.

TORONTO -- Gene expression profiling of peripheral blood lymphocytes successfully identified lung cancer patients with an overall accuracy of 87% in a cross-sectional study of 230 subjects.

The new test, which is in the early development stage, had a sensitivity of 85% and a specificity of 87%, Dr. Anil Vachani said at an international conference of the American Thoracic Society. The study included 140 subjects with lung cancer and 90 without.

"What we have shown in a preliminary fashion in collaboration with researchers from the Wistar Institute is that with a 24-gene signature we had 87% accuracy for distinguishing between cancer and controls," Dr. Vachani of the University of Pennsylvania, Philadelphia, said at a press conference.

Lung cancer remains a very difficult cancer to diagnose, with many patients having lung nodules detected incidentally on chest x-ray or CT scan. The next step in diagnosis is problematic, as needle biopsy of the chest is difficult and bronchoscopy requires anesthesia and carries risks of bleeding and atelectasis. The nodule is often surgically removed instead, and a common outcome is that the nodule turns out to be benign and the patient did not have cancer at all, Dr. Vachani said.

Clearly, a blood test that could be used when patients are found to have lung nodules would be useful, and many research groups have worked on this. Most have focused on finding a signature protein, such as prostate-specific antigen, secreted by tumor cells into the bloodstream.

Unfortunately, lung cancer is a much more heterogeneous cancer than some others, and no single protein has been found to identify all types of lung cancer, Dr. Vachani said.

"Instead, our approach has involved profiling of gene expression in peripheral blood mononuclear cells that are involved in tumor immunity," he said.

The approach involves isolating lymphocytes from peripheral blood and performing global gene expression profiling. "We measure the 20,000 genes found in these cells, and using advanced statistical algorithms we identified genes that are differentially expressed between patients and controls. We then go back to see if we can validate whether genes that are differentially expressed can actually predict which patients have cancer and which do not."

Although the results of his study were good, they need to get better, Dr. Vachani said. "The technology is in early development, and the study population will be expanded. External validation studies need to be done to see if the results in Pennsylvania can be replicated elsewhere, which has been problematic in some gene expression studies," he said.

The expected role of this test would be as a second test following the identification of lung nodules on imaging studies. Patients found to be at high risk for cancer could go on to a more extensive work-up, while those at low risk might not need any further work-up, or perhaps minimal imaging studies in the future, he said.

"Also, as lung cancer screening with low-dose CT scans starts to take off, we anticipate that many more patients will be diagnosed with lung nodules and will need follow-up," he said.

The study was funded by the Pennsylvania Department of Health.

BY NANCY WALSH

New York Bureau