SESSION TYPE: Lung Cancer Posters I
PRESENTED ON: Wednesday, October 24, 2012 at 01:30 PM - 02:30 PM
PURPOSE: Lung cancer is the number one cause of cancer-related death in the United States and the world. Raman spectroscopy could provide a real-time, non-invasive methodology for identifying lung lesions earlier, and more accurately, than existing methods. In a previous study, a Raman spectroscopy model identified normal from cancerous lung tissue using the peak ratios 722/756, 1160/1177, 1253/1342, and 1450/1662 cm-1. Adenocarcinoma was distinguished from squamous cell carcinoma using the following peaks and peak ratios: 813, 833, 939, 957, 722/756, 1253/1342, 1450/1662, and 1533/1541 cm-1. Each Raman peak (and corresponding ratio) corresponds to the concentration of a specific biomolecule, such as proteins, lipids, and amino acids. The purpose of this work is to validate this classification methodology in a naive set of tissue samples.
METHODS: Paired normal and cancerous lung tissue specimens were retrieved from storage in a -80°C tissue bank. Samples were thawed at room temperature for 15 minutes, and a Renishaw inVia Raman microscope with 785 nm excitation wavelength and 50x objective was used for measurement. 8 accumulations of 20 seconds each were averaged for each spectrum. Approximately 14 spectra were measured on each tissue. Spectra were processed with spike elimination, background subtraction, vector normalization, and Whitaker smoothing. The level of fluorescence interference was identified in each spectrum. The previously developed discriminant function analysis classification algorithms were applied to identify normal lung from lung cancer and adenocarcinoma from squamous cell carcinoma.
RESULTS: Following removal of outlying data, 594 new spectra were available for analysis. Cancer was identified with 75.3% sensitivity and 68.6% specificity. Among 82 spectra exhibiting minimum fluorescence interference, sensitivity and specificity were 97.4% and 76.5%. Squamous cell carcinoma and adenocarcinoma were identified with 61.2% accuracy, increasing to 72% with only low-fluorescence data.
CONCLUSIONS: The previously developed algorithm identified lung cancer with high sensitivity and specificity, especially in data with minimal fluorescence interference.
CLINICAL IMPLICATIONS: Raman spectroscopy has potential for minimally invasive diagnostics, however fluorescence interference must first be addressed and minimized.
DISCLOSURE: The following authors have nothing to disclose: Rachel Kast, Sally Yurgelevic, Katrina Nelson, George Divine, Laila Poisson, Brian Lace, Gregory Auner, Michael Simoff
Raman spectroscopy is a technique being investigated for optical biopsyWayne State University, Detroit, MI