Institute of Cancer Research (ICR) Background
Digital pathology is an emerging field. It uses sophisticated computing tools and artificial intelligence (AI) to diagnose disease and guide treatments faster and more easily. It offers particular promise for developing novel ways to understand cancer.
The Computational Pathology and Integrative Genomics Team at The Institute of Cancer Research, London, has already created a method to automate tumour-infiltrating lymphocyte (TIL) scoring in several cancer types, by scanning routine pathological slides into digital images and using an AI algorithm to analyse the images automatically at the single-cell level. They are currently adapting and trialling this algorithm for use as a novel biomarker.
Triple-Negative Breast Cancers (TNBC) are among the most aggressive forms of breast cancer but there are too few targeted treatments or biomarkers for patient stratification, or to guide immunotherapeutic development. Manual estimation of tumour-infiltrating lymphocyte (TIL) score, as proposed by the International Immuno-Oncology Biomarker Working Group on Breast Cancer, has emerged as a promising prognostic and predictive biomarker in TNBC.
Currently, the main prognostic factors in early-stage TNBC represent anatomic tumour burden; the host microenvironment is not usually considered in the prognostic assessment. The TIL method, based on clinical routine diagnostic histology samples, has been shown to be a reproducible biomarker that could be fairly straightforward to implement in standard clinical pathology globally. TILs are immune cells that protect the patient against their cancer.
If successful, the application of AI-TIL as a biomarker will guide therapeutic decision-making, for example recommending chemotherapy de-escalation for the subgroup of node-negative TNBC patients with high TIL levels (which represent relatively good prognosis).
Technology Overview
An algorithm has been developed to enable the AI-driven measurement of TIL scores from digitised images of routine pathology samples. A project is underway in collaboration with The Royal Marsden NHS Foundation Trust, The Institute of Cancer Research’s (ICR’s) hospital partner. If successful, it will deliver:
A sample processing framework for digitalisation and AI deployment.
A standardization protocol for AI input, tailoring and testing the AI algorithm for the analysis of TNBC samples in the training set.
Validation of the AI algorithm.
Further Details
AbdulJabbar K et al., Geospatial immune variability illuminates differential evolution of lung adenocarcinoma. Nat Med. 2020 Jul;1062-1054:(7)26. doi: 10.1038/s-0900-020-41591x
Yuan Y. Modelling the spatial heterogeneity and molecular correlates of lymphocytic infiltration in triple-negative breast cancer.J R Soc Interface. 2015;12(103):20141153.doi:10.1098/rsif.2014.1153EndFragment
Benefits
Reproducible new prognostic biomarker with potential in multiple tumour types and particular promise in triple-negative breast cancer (TNBC).
Pioneering use of AI in digital pathology.
The fully automated measurement from digitised slide samples is quicker, more accurate and more cost-effective than traditional estimation.
Evaluation of immune-checkpoint-based therapy could potentially be extended to other high-TIL TNBC subgroups and the overall method could be extended to other tumour types.
Opportunity
The ICR is seeking a partner to continue the commercial development of AI-TIL: an image-based biomarker that quantifies tumour-infiltrating lymphocyte (TIL) from histopathological samples using proprietary AI technologies. It has already shown particular promise as a predictor of response to immunotherapy in triple-negative breast cancer.
The team is working towards CE-marking AI-TIL as a class C IVD medical device and is now seeking commercial partners to further develop and scale the method to be incorporated in large clinical trials and everyday diagnostic histopathology practice. The team is also seeking partners to commercialize AI-TIL for use in R&D and pharmaceutical drug development.