(Chest. 2004;125:104S.)
© 2004
American College of Chest Physicians
Class Prediction of Lung Nodule Gene Expression Profiles*
Kristin L. Walter, MD;
Alain C. Borczuk, MD;
Ligun Q. Wang, MS;
Adel M. Assaad, MD;
J.H.M. Austin, MD, FCCP;
Gregory D. Pearson, PhD, MD;
Maria C. Shiau, MD and
Charles A. Powell, MD
* From the Columbia University College of Physicians & Surgeons, New York, NY.
Correspondence to: Charles A. Powell, MD, Division of Pulmonary Allergy, and Critical Care Medicine, Columbia University, Rm P&S 8–512, 630 W 168th St, Box 91, New York, NY 10032; e-mail: cap6{at}columbia.edu
 |
Introduction
|
|---|
Gene expression profiling is a powerful tool that may improve the methods for risk stratification and treatment optimization in patients with lung cancer. Our long-term goal is to utilize gene expression profiling prospectively in the diagnosis and treatment planning for patients with lung carcinoma. To investigate the feasibility of this approach, we designed an experiment to examine the gene expression profiles of specimens obtained from CT scan-guided biopsy or endobronchial brushing of undiagnosed pulmonary nodules.
Our hypothesis was that the gene expression profiles derived from these tissues would correlate closely with the histopathologic diagnosis of the pulmonary nodule.
|
Materials and Methods
|
|---|
The subjects of the study were 24 patients undergoing biopsies of undiagnosed pulmonary nodules. After a biopsy of a lung nodule was performed and specimens were obtained for pathology, residual cells were placed into a buffer for RNA extraction. Specimens were processed using the modified Eberwine protocol for analysis (U95Av2 array; Affymetrix; Santa Clara, CA), which contains probes for approximately 12,000 genes. We performed a class prediction analysis of the test samples using computer software (GeneSpring, version 5.0; Silicon Genetics; Redwood City, CA), using 300 predictor genes and 3 nearest neighbors. Our training set consisted of 71 microdissected samples, which included non-small cell lung cancer (NSCLC), small cell lung cancer, carcinoid, mesothelioma, normal lung tissue, and pleural tissue. Our gene expression profile prediction categories included NSCLC, small cell lung cancer, and nonneoplastic lung.
|
Results
|
|---|
To validate the protocols, we first determined the correlation between the gene profiles of tumor aspirates and microdissected tumors. Needle aspiration biopsies were performed on four resected lung cancers, and the gene profiles were compared with those generated from the same microdissected specimen. The correlation (ie, the r value) ranged from 0.66 to 0.84. We next examined the correlation of gene expression profiles generated from the standard protocol with those generated from the modified Eberwine protocol. We found that r = 0.82 when comparing the gene profiles using 100 ng RNA processed using a modified Eberwine protocol with profiles obtained by standard amplification using 4 µg RNA from the same tumor. A class prediction analysis of the gene profiles from the clinical biopsy specimens accurately predicted the histopathology in 17 of 24 instances. Specimens with incorrect classifications included three nonmalignant lung tissue samples predicted to be NSCLC, two adenocarcinomas predicted to be nonmalignant lung tissue, one small cell lung cancer predicted to be an NSCLC, and one mixed small cell lung cancer/NSCLC predicted to be a small cell lung cancer.
|
Conclusions
|
|---|
We have shown that gene expression profiles obtained from residual tissue after CT scan-guided biopsy or endobronchial brushing were approximately 71% accurate in predicting the pathology diagnosis. Therefore, a prospective analysis of the gene expression profiles of pulmonary nodules is feasible and may ultimately provide clinically relevant diagnostic and prognostic information that could improve the management of patients with lung carcinoma.
|
Footnotes
|
|---|
Abbreviation: NSCLC = non-small cell lung cancer
TOP
Introduction
Materials and Methods
