Radiomics-based imaging tool to monitor tumor-lymphocyte infiltration and outcome in cancer patients treated by anti-PD-1/PD-L1
AssigneesInstitut National de la Sante et de la Recherche Medicale, Universite Paris-Saclay
InventorsCharles Ferte, Elaine Johanna Limkin, Roger Sun, Eric Deutsch
Life of this patent
Estimating the term…
Abstract
The present invention proposes a radiomics-based biomarker for detecting the presence and the density of tumor infiltrating CD8 T-cells in a solid tumor without having to use any biopsy of said tumor. The invention also proposes to use this information to assess the immune phenotype of said solid tumor. In a particular embodiment, the invention proposes to prognose the survival and/or the treatment efficiency of cancer patients treated with immunotherapy such as anti-PD-1/PD-L1 monotherapy.
Claims14
A method for evaluating the spatial density of a CD8+ T cells infiltrate in a solid tumor, said method comprising:
- a) specifying two regions of interest (ROI) in a radiological image of a region of a tumoral tissue of a solid tumor, said radiological image including a plurality of voxels, wherein the two ROI are:
- a solid tumor volume and
- a peripheral ring volume created around the solid tumor margin;
- b) extracting a set of features from said two ROI, wherein the set of features comprises:
- Min Value of the tumor as conventional variable,
- at least three Gray-level Run Length Matrix (GLRLM) variables selected from:
- GLRLM short-run high gray-level emphasis (GLRLM_SRHGE) of the solid tumor,
- GLRLM short-run low gray-level emphasis (GLRLM_SRLGE) of the peripheral ring volume,
- GLRLM low gray-level run emphasis (GLRLM_LGRE) of the peripheral ring volume, and
- GLRLM long-run low gray-level emphasis (GLRLM_LRLGE) of the peripheral ring volume;
- acquisition marker kilovoltage peak (kVp), and
- location of adenopathy (VOI_Adenopathy) and location of head and neck cancer (VOI_head_and_neck), wherein the values of VOI Adenopathy and VOI head and neck are null when the target tumors are not from adenopathy or head-and-neck;
- and
- c) providing the extracted features to a trained machine learning predictor which calculates a quantitative score of the spatial density of the CD8+ T cells infiltrate in the solid tumor.
A method for predicting the outcome or the efficiency of an anti-cancer treatment for a cancer patient with a tumor, said method comprising:
- a) specifying two regions of interest (ROI) in a radiological image of a region of a tumoral tissue of a solid tumor, said radiological image including a plurality of voxels, wherein the two ROI are:
- a solid tumor volume and
- a peripheral ring volume created around the solid tumor margin;
- b) extracting a set of features from said two ROI, wherein the set of features comprises:
- Min Value of the tumor as conventional variable,
- at least three Gray-level Run Length Matrix (GLRLM) variables selected from
- GLRLM short-run high gray-level emphasis (GLRLM_SRHGE) of the solid tumor,
- GLRLM short-run low gray-level emphasis (GLRLM_SRLGE) of the peripheral ring volume,
- GLRLM low gray-level run emphasis (GLRLM_LGRE) of the peripheral ring volume, and
- GLRLM long-run low gray-level emphasis (GLRLM_LRLGE) of the peripheral ring volume;
- acquisition marker kilovoltage peak (kVp), and
- location of adenopathy (VOI_Adenopathy) and location of head and neck cancer (VOI_head_and_neck), wherein the values of VOI Adenopathy and VOI head and neck are null when the target tumors are not from adenopathy or head-and-neck;
- and
- c) providing the extracted features to a trained machine learning predictor which calculates a quantitative score of the spatial density of the CD8+ T cells infiltrate in the solid tumor.
Dependencies are inferred from the claim text.