Autofluorescence spectroscopic differentiation between normal and cancerous colorectal tissues by means of a two-peak ratio algorithm.

We conducted an ex vivo study with light-induced autofluorescence spectroscopy to distinguish between normal and cancerous colorectal tissues. A total of 20 normal and 20 cancerous tissue samples were obtained from 20 patients undergoing oncologic surgery of the colon and rectum. The tissue diagnosis was confirmed histologically. After excitation, the autofluorescence spectra of each sample were measured and recorded. To determine the most appropriate excitation wavelength, we analyzed the autofluorescence spectra at excitation wavelengths of 280 to 400 nm, in 10-nm increments. To distinguish between normal and cancerous tissues, we first calculated the "two-peak ratio" (the ratio of the integrated intensity of the first peak +/- 6 nm to the integrated intensity of the second peak +/- 6 nm) of the autofluorescence spectra. By applying receiver operating characteristic (ROC) curves and calculating the areas under the ROC curves (AUC), we found 330 and 340 nm to be the most appropriate excitation wavelengths to distinguish between cancerous and normal colorectal tissue. When approximate thresholds were selected, autofluorescence spectroscopy with 330 nm excitation yielded a sensitivity of 85%, a specificity of 90%, and a positive predictive value of 89% for detection of cancerous tissue. The ex vivo autofluorescence study and two-peak ratio algorithm developed herein may be very useful for developing an algorithm for in vivo diagnosis of colorectal cancer.