Session: Parallel session 7 - Club jeune SFSM
Development of an Esogastric Cancer Improved Diagnosis with SpiderMass Technology
Introduction
With more than 1 million cases each year, esogastric cancer is the fifth most often-diagnosed cancer worldwide. Real-time pathological analysis faces challenges, especially in certain histological subtypes such as PCC. The SpiderMass technology, an ambient mass spectrometry-based approach, was used to improve diagnosis and prognosis precision.
Methodology
A cohort of 158 tissue samples from 107 patients, of which 77 are healthy and 84 are adenocarcinomas, with 55 non-PCC and 29 PCC subtypes were sectioned on a cryostat. Three sections were collected, one (5 µm) for the H&E staining, the second (20 µm) for SpiderMass analysis and the last one (12 µm) for the MALDI-MSI. The SpiderMass analysis was performed on a Xevo in both polarities. The data were processed using an homemade ML pipeline recently published in Cell Reports Medicine (on GBM tissues).
Results
125 EC sections were analyzed using SpiderMass and showed a different and specific molecular profile for each type of tissue. The SpiderMass data were then used to build classification models for typing and subtyping the EC using ML. More than 92% of accuracy was obtained for typing and sub-typing models in both ion mode. The models were further validated in blind on an extra cohort of 33 tissues and only 3 were misclassified. Furthermore, our interest extended to the identification of lipids linked to the distinction between cancerous and healthy esogastric tissues. Our goal was to enhance our understanding of the biological mechanisms responsible for this differentiation. The use of a supervised approach like the LIME algorithm and an unsupervised statistical analysis allowed us to generate a comprehensive list of confident biomarkers. In total, we identified 50 reliable biomarkers, these biomarkers being associated with tumor or healthy tissue.
Conclusion
Developing a fast intraoperative diagnosis and prognosis based on MS for esogastric cancer with better performances than the gold standard.
With more than 1 million cases each year, esogastric cancer is the fifth most often-diagnosed cancer worldwide. Real-time pathological analysis faces challenges, especially in certain histological subtypes such as PCC. The SpiderMass technology, an ambient mass spectrometry-based approach, was used to improve diagnosis and prognosis precision.
Methodology
A cohort of 158 tissue samples from 107 patients, of which 77 are healthy and 84 are adenocarcinomas, with 55 non-PCC and 29 PCC subtypes were sectioned on a cryostat. Three sections were collected, one (5 µm) for the H&E staining, the second (20 µm) for SpiderMass analysis and the last one (12 µm) for the MALDI-MSI. The SpiderMass analysis was performed on a Xevo in both polarities. The data were processed using an homemade ML pipeline recently published in Cell Reports Medicine (on GBM tissues).
Results
125 EC sections were analyzed using SpiderMass and showed a different and specific molecular profile for each type of tissue. The SpiderMass data were then used to build classification models for typing and subtyping the EC using ML. More than 92% of accuracy was obtained for typing and sub-typing models in both ion mode. The models were further validated in blind on an extra cohort of 33 tissues and only 3 were misclassified. Furthermore, our interest extended to the identification of lipids linked to the distinction between cancerous and healthy esogastric tissues. Our goal was to enhance our understanding of the biological mechanisms responsible for this differentiation. The use of a supervised approach like the LIME algorithm and an unsupervised statistical analysis allowed us to generate a comprehensive list of confident biomarkers. In total, we identified 50 reliable biomarkers, these biomarkers being associated with tumor or healthy tissue.
Conclusion
Developing a fast intraoperative diagnosis and prognosis based on MS for esogastric cancer with better performances than the gold standard.