When an imaging test shows a suspicious mass in the lungs of a smoker, the next step is a bronchoscopy. A thin tube with a camera is passed into the person's airway to look for abnormalities and collect tissue for biopsy. But in many cases the results do not reveal whether the person has cancer.

At this point, a physician may decide, based on the available evidence, to pursue more invasive testing or to monitor a patient - a "watch and wait" approach. Unfortunately, there are no validated diagnostic tests or biological markers that reliably distinguish between smokers with and without the disease.

Researchers in Boston hope to change that. They have developed a genomic test that analyzes the activity of 80 genes in cells from a patient's airway. The airway cells, which can be collected during bronchoscopy, appear normal under the microscope, but they may have abnormal patterns of gene activity associated with lung cancer.

The test is based on the idea that toxins in cigarette smoke alter the behavior of genes throughout the respiratory tract - not just in the tumor - and that these changes have diagnostic value.

"The goal of this research was to identify gene expression patterns that can distinguish smokers with and without lung cancer," said Dr. Avrum Spira of Boston University School of Medicine, who co-led the project.

Preliminary results reported last year showed that the genomic test and bronchoscopy together were a better predictor of lung cancer than either test alone. There were even hints that the combination may improve the detection of small lesions or early-stage cancers - the ones more likely to respond to therapies and most often missed by bronchoscopy.

Dr. Spira discussed the test last month at the annual meeting of the NCI-sponsored Early Detection Research Network. A theme of the meeting was the need for biological markers with clinical relevance, as the lung cancer genomic test aims to be. EDRN and Dr. Spira are discussing ways to move this test towards clinical validation.

"This test addresses a serious dilemma in the clinic," said co-investigator Dr. Marc Lenburg of Boston University. "It's a real-world problem."

The need for better diagnostic tools is clear. Bronchoscopy detects stage 1 lung cancer in only 15 to 20 percent of cases, so physicians must move on to more invasive and more expensive tests, including exploratory surgeries, to determine if the lesions they see are cancer, according to Dr. Jerome Brody, who directs the Boston University Pulmonary Center and helped develop the genomic test.

As with all biomarker candidates, the 80-gene signature is experimental and must be validated in a large, well-designed study. A trial is being planned to evaluate the genomic test in 300 patients in the United States and Europe.

The validation study will also explore whether the genomic test can help physicians identify the subtype of lung cancer a person has and the underlying genetic flaws. This information could help physicians identify the most appropriate therapies for patients.

All three researchers have financial interests in a company called Allegro Diagnostics, which has raised funds to validate and extend this study and to make the test available commercially.

The researchers describe their approach as a "clinicogenomic" prediction model. In a study published online yesterday in Cancer Prevention Research, they say that the clinical and genomic information is both complementary and synergistic.

"What's new here is the combining of the genomic data with clinical information to develop a statistically valid predictor of lung cancer," said Dr. Brody, who was not an author of the most recent study. "This is one of only a few studies in any cancer that has attempted to combine the two in a statistically valid fashion."

The prediction model may expedite more invasive testing and appropriate therapies for smokers with lung cancer and reduce invasive diagnostic procedures for individuals without lung cancer, the authors conclude.

—Edward R. Winstead