Session: Parallel session 4 - Spatial OMICs and MS Imaging

MS Imaging-Based Prediction of Immune Cell and Bacterial strains Population Ratio : Towards in vivo Immunoscores and Bacterioscores for Cancer Diagnosis and Prognosis

Introduction

Patient prognosis is predictable through measurement of immune cells infiltration in TME. It’s achieved post-operatively by IHC which is time-consuming. In addition, there are growing evidence of microbiota importance in cancer. So, we investigated the development of an immunoscoring and bacterioscoring based on SpiderMass-MSI which could be performed in real-time during the surgery, to predict the distribution of each cell or bacteria type within pixels.

Methodology

The SpiderMass enables real-time in vivo analysis with no invasiveness by using mid-IR to excite water molecules. Different immune cell populations were analyzed using SpiderMass after cell sorting. Classification models were built from cell population MS spectra. A LGBM model was built to predict the ratio of each cells in each pixel of the GBM tissues image. The same strategy was used to create bacterioscoring using 6 bacterials strains and using it on esogastric tissues.

Results

The analysis of various immune cell populations reveals distinct molecular profiles. It enable the development of a classification model through supervised machine learning approache. Immunoscoring is then conducted by leveraging LGBM from GBM image data, enabling the prediction of cell population ratios in each pixel. These ratios correlate with patient's overall survival, allowing the description of a prognosis score by evaluating the M1 versus M2 immune cell ratios. The bacterioscore was developed in the same way, but this time on oesogastric cancer tissue. Indeed, we were able to see that the predicted ratio of three bacterial strain were significantly different according to the cancerous or healthy character of the tissue. Indeed, L. gasseri strain were predicted to have a higher abundance in healthy tissue in contrary to L. lactis and S. bovis.

Conclusion

This breakthrough enables real-time diagnosis and prognosis assessment for patients through the direct integration of SpiderMass technology during surgery.