= Discovery stage. (24.37%, 2023)
= Translation stage. (39.50%, 2023)
= Clinically available. (36.13%, 2023)
MSACL 2023 : Ford

MSACL 2023 Abstract

Self-Classified Topic Area(s): Precision Medicine > Cases in Clinical MS > Cases of Unmet Clinical Needs

Podium Presentation in Steinbeck 3 on Thursday at 17:10 (Chair: Christopher Chouinard / William Perry)

Feasibility Analysis of iEndoscope for Real-Time Data Driven Pathology Using Novel MS and Optical Technology

Lauren Ford1, Vadzim Chalau1, Ajantheny Naguleswaran1, James McKenzie2, Sam Mason1, Eftychios Manoli2, Robert Goldin3, Zoltan Takats2, Daniel Elson1, James Kinross1
1. Surgery and Cancer, Imperial College London, London, UK 2. Metabolism, digestion and reproduction, Imperial College London, UK 3. Clinical Pathology, Imperial College Healthcare Trust, London, UK

Lauren Ford, BSc (Hons), PhD (Presenter)
Imperial College London

Presenter Bio: I am an early career researcher and have a background in materials chemistry, having studied for a PhD between the School of Chemistry and the School of Design at the University of Leeds I have experience in polymer technology, physical adsorption theory and purification. I am interested in using these skills to aid detection of disease using mass spectrometry detection. Since joining Imperial in 2019 I have been working as a post-doctoral research associate in the department of Surgery and Cancer, working on the iEndoscope project. This project utilised ambient ionisation mass spectrometry and allowed me to gain critical experience of ambient MS for early cancer detection.

Abstract

Background:
There is a significant demand for technologies that provide real time histological data during endoluminal therapeutic procedures for complex polyps and polyp cancers.1 Current strategies based on deep learning of white light imaging data fail to provide the endoscopist with data on resection margin status, tissue subtype or tumour heterogeneity. Moreover, standalone optical biopsy approaches fail to provide molecular insights for treatment or surveillance stratification. This analysis compares two novel translational methodologies; optical biopsy and chemical histology using Rapid Evaporative Ionisation Mass Spectrometry (REIMS) for their diagnostic yield and use in near real time detection of colonic pathology.

Objectives of study:
To compare the diagnostic accuracy two novel techniques for detection of colorectal cancer. Diagnostic accuracy, sensitivity and specificity was determined for: non-destructive optical biopsy and REIMS for the detection of colorectal cancer and adenomas.

Methods:
A prospective, observational feasibility analysis of elective colorectal resections a single institution (Imperial College London). Ex-vivo REIMS database and models were built using tissue samples from 161 patients, producing 1013 spectra from tumour (n = 346) adenoma (n= 247) and normal mucosa (n= 420). REIMS data were acquired using a Waters Xevo G2XS ToF mass spectrometer modified for use in surgery and fitted with a REIMS source. Data were acquired negative ionisation mode in the m/z range of 50-1200. Multivariate statistical analysis was performed using Python in house-built scripts. Logistic regression (LR) and linear discriminant analysis (LDA) classification with leave one patient out cross validation (LOOCV) was performed. Recursive Feature Elimination (RFE) was used to refine features of interest for classification. These models were assessed for the in-vivo colorectal cases.

29 Tissue samples from 20 patients (normal n=9, tumour n=7 and adenoma n=12) were assessed using bimodal optical diagnostic system, combined Diffuse Reflectance Scattering (DRS) and Laser Induced Fluorescence (LIFS). DRS Spectra were registered with fiber-to-fiber distances 0.50, 0.80, 1.60, 2.80 mm and LIFS spectra were registered at excitations wavelengths 375 and 405 nm. The sample set was then sectioned for analysis with REIMS for localised assessment of heterogenous tissue with the consequential section being histologically validated using H&E staining.

Results:
Ex-vivo reference models
Current REIMS ex-vivo model displays a classification accuracy of 96.7% for three-way classification (Sensitivity 94.2% (polyp), 94.6% (Tumour) and specificity 98.9% (polyp), 97.2 (Tumour)) with molecular species of interest including Phosphatidylethanolamine (PE), Phosphatidic acid (PA) Triglyceride (TG) and Fatty acids (FA).

For optical diagnostics DRS spectra registered at 2.8 mm fibre-to-fibre separation provided optimal three-way classification accuracy (89%) between Normal, Tumour and Polyp tissue types. LIFS spectra registered at an excitation wavelength of 375 nm provided better classification compared to excitation of 405 nm across all three tissue types (Sensitivity 83.3% (Polyp) and 87.7% (Tumour), specificity 93.6% (Polyp) and 90.9% (Tumour)). Combining DRS and LIFS had an effect on the overall classification accuracy. The same tissues were then analysed with REIMS for spatially resolved molecular verification using Mass Spectrometry (MS). The spatially resolved nature of REIMS imaging allowed us to assess the degree of tissue heterogeneity by assessing spatial variation in the data and applying models to compare to corresponding histopathology information.

In-vivo MS analysis
REIMS in-vivo data was collected from 7 Patients (8 distinct lesions – resulting in 84 individual mass spectra) undergoing colonoscopy or TAMIS procedures which required the use of a diathermy procedure. Any patients under the age of 18, or patients suffering from inflammatory bowel disease or a hereditary polyposis syndrome were excluded from the study. The in-vivo diagnostic accuracy of REIMS using the ex-vivo models was shown to be 73% for three-way classification (Sensitivity: 85% (Polyp), 68% (Tumour), 75% (Normal)).

Conclusions:
Novel methodology compared herein show that different spectroscopic techniques are valuable for identifying tissue of unknown histopathological status with high diagnostic accuracy. Optical technology gives information on the tissue type using non-destructive methods and hence does not alter existing clinical pathways. Mass spectrometry data can provide in-depth molecular information both in a surgical environment and can be translated into images displaying tumour heterogeneity giving molecular insight.
Both technologies can be integrated into a standard colonoscope and are now being assessed as part of a prospective clinical trial. Future work will assess the feasibility of combining the two methodologies herein.

References:
1) Mason, S. E., Poynter, L., Takats, Z., Darzi, A., & Kinross, J. M. (2019). Optical Technologies for Endoscopic Real-Time Histologic Assessment of Colorectal Polyps: A Meta-Analysis. American Journal of Gastroenterology, 114(8), 1219–1230. https://doi.org/10.14309/ajg.0000000000000156.


Financial Disclosure

DescriptionY/NSource
Grantsno
SalaryyesImperial College London
Board Memberno
Stockno
Expensesno
IP Royaltyno

Planning to mention or discuss specific products or technology of the company(ies) listed above:

yes