Probing Field Cancerization in the Gastrointestinal Tract Using a Hybrid Raman and Partial Wave Spectroscopy Microscope

TL;DR

This paper introduces a hybrid microscope combining Raman and partial wave spectroscopy to detect early cancerous changes in the gastrointestinal tract of mice.

Contribution

The novel use of a hybrid RS-PWS microscope improves detection of field cancerization in macroscopically normal tissues.

Findings

Raman spectroscopy detects increased amino acid and decreased lipid/carotenoid signals in tumor-adjacent tissues.

Combining Raman and partial wave spectroscopy improves classification accuracy by up to 9% in mouse models.

Transcriptomic analysis confirms a correlation between optical findings and gene expression changes.

Abstract

Field cancerization
(FC) refers to spatially distributed premalignant
tissue changes that lead to the appearance of local malignancy, and
its detection can improve cancer screening. In this work, we employ
combined Raman and partial wave spectroscopy (RS-PWS) to detect FC
in gastroesophageal (L2-IL1B) and intestinal (Villin-Cre, Apcfl/wt) tumor mouse models. Using a hybrid RS-PWS microscope,
we acquire both molecular and morphological information from macroscopically
normal tumor-adjacent tissue and investigate the individual and combined
performance of each modality. For data analysis, we use partial least-squares
discriminant analysis (PLS-DA). In the normal tissue of L2-IL1B mice,
we demonstrate a statistically significant increase (p < 0.001) in Raman band intensities associated with free amino
acids and a decrease in bands associated with lipids (p < 0.005) and carotenoids (p < 0.001) compared
to healthy controls. Similarly, in the normal mucosa of Villin-Cre,
Apcfl/wt mice, the intensities of RS bands associated with
amino acids increase significantly (p < 0.05)
compared to controls, while the intensities of lipid-associated bands
decrease significantly (p < 0.05). Transcriptomic
profiling using RNA-sequencing analysis on these samples identified
a significant correlation between gene expression and optical findings.
Moreover, we demonstrate that combining RS and PWS data further improves
the significance of our classification results. When macroscopically
normal tumor-adjacent tissue is compared with tissue from healthy
controls, we observe that PWS increases the R
2 of RS results by ∼9% in L2-IL1B mice and ∼5%
in Villin-Cre, Apcfl/wt mice. Combining molecular RS with
structural PWS information enhances the ability to detect precancerous
changes and provides insights into tissue alterations during cancer
development.

Figures