NEW YORK CITY—In response to the lack of effective tools to diagnose lung cancer at an early stage, researchers from Boston University have developed a novel biomarker that uses gene-expression profiling of airway epithelial cells for early detection of lung cancer.
Analysis of the biomarker in airway cells yields 95% sensitivity and a 95% negative predictive value for Stage 1 lung cancer when combined with cytopathology of lower airway cells obtained at bronchoscopy, said Avrum Spira, MD, Director of the Bioinformatics and Systems Biology Program at Boston University Medical Center and Assistant Professor of Medicine, Pathology, and Laboratory Medicine at the Pulmonary Center there, speaking here at the Chemotherapy Foundation Symposium.
“Given that cigarette smoke creates a field of injury throughout the airway, we sought to determine if gene expression in histologically normal larger airway epithelial cells obtained at bronchoscopy from smokers with suspicion of lung cancer could be used as a lung cancer marker.”
Although the biomarker was used for diagnosis in this study, Dr. Spira said that this biomarker might one day lead to individualized, targeted therapy for lung cancer.
He explained that the airway field-of-injury hypothesis holds that when an individual smokes, the epithelial cells that line the entire respiratory tract, from trachea and main stem bronchi to the alveolar epithelial cells, are affected at a gene-expression level by that exposure.
“We also believe there is variability in how an individual's epithelial cell system responds to and is damaged by that exposure, and that variability is associated with the risks for developing tobacco-induced lung cancer.”
The researchers developed a relatively noninvasive tool to identify those 10% to 25% of smokers at risk for developing lung cancer simply by collecting the cells from the major airways at the time of bronchoscopy and looking at their gene-expression patterns.
The tool was developed in the “Airway Transcriptome Project,” now five years old, carried out in association with Affymetrix, the company that makes the microarray technology.
The project first identified the “normal” airway transcriptome—the set of genes expressed at baseline in the bronchial epithelial cells of healthy never-smoking volunteers. The researchers then used the assay to define how the transcriptome is affected by smoking and then defined how the transcriptome changes when smoking is discontinued.
The fourth step is identifying the gene-expression profile that can distinguish smokers with and without lung cancer, Dr. Spira said.
“We did this in a very clinically relevant way, to tackle problems seen week in and week out in pulmonary practices—smokers referred by primary care physicians because of a suspicion of lung cancer based on an abnormality seen on a chest x-ray or a CT scan.”
If the initial fiber optic bronchoscopy is diagnostic, the physician can stage and treat the patient. “But more often than not, the bronchoscopy is nondiagnostic and we're stuck with a clinical dilemma—what do we do next to this patient,” he said. “Do we take them to surgery and be definitive in terms of what the diagnosis is—with the risks and costs of a major operation—or employ a watch-and-wait strategy, with the risk that the disease will spread beyond the lung during the interval of waiting?”
Intermediate options, such as computed tomography-guided biopsies, are not as invasive as surgery but are not as definitively diagnostic, he said.
In the study described at the meeting, bronchoscopists did routine washings, brushings, and biopsies, but before pulling the scope out of the patient's nose or mouth the bronchoscopists stopped in the main stem bronchus and collected the histologically normal epithelial cells that line that site.
Those cells were then run on a microarray to determine the pattern of gene expression.
“That will tell us something about the likelihood that that individual has lung cancer, and the information could also be used to guide how aggressive to be in the next stages of that patient's workup should the bronchoscopy be nondiagnostic.”
The study included 152 individuals from four different medical centers in the United States and Europe from January 2003 to May 2005. All were undergoing bronchoscopy for suspicion of lung cancer, during which airway epithelium was collected and run through an Affymetrix HG-U133A microassay that measures approximately 22,000 human genes.
The patients were followed post-bronchoscopy until a final diagnosis was made. “Of the 152 subjects, 129 had high-quality microarray data available, of which 60 had lung cancer and 69 did not,” he said.
The 129 patients were randomly assigned to a training set of 77 samples on which the researchers built an 80-gene expression model capable of distinguishing smokers with and without lung cancer. That 80-gene model was then tested on an independent set of 52 samples, and was found to have an accuracy rate of 83% (80% sensitivity, 84% specificity).
The results have been reproduced in two other large microarray data sets: Spira A et al: Nat Med 2007; 13, 361–366, and those findings have just recently been validated in a prospective series of 35 samples, predicting the class of 80% of the patients, he said.
“Of the 18 [of 35], the combined gene expression and bronchoscopy study is able to accurately diagnose 95 percent of those cases.”
The assay is not meant to replace bronchoscopy but to be synergistic with it. Learning how the cancer-specific molecular field of injury reflects information about specific oncogenic pathways may potentially allow personalized genomic approaches to chemoprophylaxis and therapy, Dr. Spira said.
Still Need Bronchoscopy
Session co-moderator Roman Perez-Soler, MD, Associate Director of Clinical Oncology at New York University Medical Center, said after the presentation that he thought the biomarker was very new and interesting but that it still remains to be seen how it can be applied, because bronchoscopy is still needed. “And the accuracy or the predictive power of such a test remains to be seen,” he said.
Dr. Perez-Soler speculated that the test might be easier to do clinically if it could be done on a sample from the oral mucosa, which is also exposed to the tobacco.
“Obviously, in the context of a lot of these genetic technologies, [this study] shows that some of these technologies actually will have a use in the future, to better diagnose or to assess risk or prognosis,” he said. “[But] all the biomarkers we are studying are in prospective, randomized trials, and it's going to take a while to know what we should be using to select our patients.”