Probing Biological Systems using Reflectance and Fluorescence Spectroscopy.

Abstract

Pancreatic adenocarcinoma is a leading cause of cancer death with a five-year survival rate of only 5%. Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA), the current diagnostic standard, cannot reliably rule out malignancy and is insensitive to distinguishing adenocarcinoma from chronic pancreatitis (inflammation). To investigate the ability of multi-modal optical spectroscopy to detect signals from human pancreatic tissue, a clinically-compatible instrument was developed for rapid, quantitative reflectance and fluorescence spectroscopy in tissues, including fluorescence lifetime sensing. Reflectance and fluorescence spectra and time-resolved fluorescence decay curves were successfully measured for the first time from freshly excised human pancreatic tissues and in vivo human pancreatic cancer xenografts in mice.

For the first time, pancreatic tissue classification algorithms using optical spectroscopy data were developed. A total of 96 fluorescence and 96 reflectance spectra were considered from 50 sites (adenocarcinoma, chronic pancreatitis, and normal tissues) on 9 patients. The SpARC (Spectral Areas and Ratios Classifier) and PCA (principal component analysis) algorithms employed linear discriminant analysis on classification variables extracted from optical data. Maximum sensitivity, specificity, NPV, and PPV (85%, 89%, 92%, and 80%, respectively for the SpARC, and 91%, 90%, 95%, 83%, respectively for the PCA algorithm) for correctly identifying adenocarcinoma were achieved employing both reflectance and fluorescence spectra. Inclusion of time-resolved fluorescence data in the PCA algorithm further improved the distinction between pancreatitis and normal tissues in a limited data set. Importantly, the sensitivity of both algorithms far exceeds reported EUS-FNA sensitivity (54%) at distinguishing adenocarcinoma from chronic pancreatitis. The developed algorithms show promise for rapid automated pancreatic tissue classification using multi-modal optical spectroscopy and could be employed in a clinical setting.

The possibility of applying optical spectroscopy to evaluate tissue engineered devices was also investigated. Tissue engineered constructs are functional biologic devices employed for grafting wounds or replacing diseased tissue. Non-invasive methods are required to assess the viability of these engineered constructs. Monte Carlo simulations and multi-modal optical spectroscopy were coupled to assess porcine articular cartilage and oral mucosa constructs for the first time. The developed methods would be safe for clinical human use as they employ endogenous contrast for non-invasive quantitative assessment.