Raman Imaging Technique for Prostate Cancer Prognostication: A Feasibility Study
AH Uihlein, AJ Drauch, JS Maier, JK Cohen, PR Olson, JS Silverman. Allegheny General Hospital, Pittsburgh, PA; ChemImage Corp., Pittsburgh, PA
Background: Raman Molecular Imaging (RMI), a technique used primarily in chemical analysis, is based on the scattering of laser light. RMI yields an image of a sample wherein each pixel of the image is the Raman spectrum (RS) of the sample at the corresponding location. The RS generates data reflecting the local chemical environment of the sample at each location. RMI has a spatial resolving power of 250 nm and can potentially provide qualitative and quantitative image information based on molecular composition and morphology. We have previously applied RMI as a reagentless tissue imaging method for the analysis of pathology specimens. In this study, we investigated the application of RMI on radical prostatectomies specimens with Gleason Score 7 prostate cancer (GS7 PCa), as a tool for prognostication, since the majority of GS7 PCa will be surgically cured, but a significant number (20%) will progress.
Design: Our study evaluated whether RMI can be used to differentiate GS7 PCa that subsequently progress (PROG) from those that have no evidence of disease (NED) after 5 years. Raman images were generated from unstained sections of 10 PROG patients and 10 NED patients. Measurements derived from RMI of tissue samples were analyzed using multivariate techniques established in analytical spectroscopy. We applied Mahalanobis Distance Analysis (MD), a chemometric method that measures distance between data sets in RS, for our study.
Results: Establishing PROG and NED as two different classes and performing Principal Component Analysis (PCA) of the 68 spectra derived from RMI, allows calculation of MD between the two classes. This evaluation shows a clear distinction between PROG and NED data. A two-sample z-test applied to the PCA scores has a p value of less than 0.01, confirming a significant difference between the two groups of GS7 PCa patients (progressive disease vs. no evidence of disease after 5 years). All cases of GS7 cancer were confirmed based on histopathologic findings.
Conclusions: RMI shows potential to distinguish between GS7 PCa that will eventually progress and those with no evidence of disease on follow-up, thus complementing current prognostic methodologies. RMI represents a novel analytical technique that may be able to assist in predicting tumor behavior and the need for adjuvant treatment.
Monday, March 9, 2009 9:30 AM
Poster Session I Stowell-Orbison/Autopsy Award # 248, Monday Morning