547P - Artificial Intelligence-powered analysis of the tumor immune microenvironment in primary and metastatic colorectal cancer

Background

The Immunoscore has been recently proposed for incorporation in the European Society for Medical Oncology (ESMO) Clinical Practice Guidelines for gastrointestinal cancers and the WHO classification of tumors of the digestive system. Additionally, responses to immunotherapy vary depending on the burden and location of metastases, with liver metastases portending worse prognosis. Herein, we constructed a deep learning (DL) model to explore immune infiltration status in primary tumors and metastatic specimens from colorectal cancer (CRC).

Methods

A fully automated multi-block DL model, named Deep-ICP, was developed, incorporating tissue classification (tumor vs. non-tumor including artifacts), cell detection, and Tumor Infiltrating Lymphocytes (TILs) classification. This model was employed on over 10 million image tiles (135 μm x 135 μm) derived from full-face H&E stained images of 2213 CRC patients from the Dana-Farber Cancer Institute.

Results

The Deep-ICP model achieved a tumor tissue classification accuracy of 0.96, with precision and recall of 0.95. In the study cohort, 70% (n=1560) had only primary tumors, and 30% (n=653) had metastatic tumors. The TIL density (per mm²) was significantly higher in primary sites, with a median of 668 (Q1=468, Q3=927), compared to metastatic tumors, which had a median of 528 (Q1=358, Q3=751). Among metastatic sites, liver (n=297) metastases exhibited the lowest TIL density with a median of 478 compared to other sites, including bowel (n=1559), lung (n=123), and others (n=233; p-adjust < 0.0001). In a subset of patients with available mismatch repair (MMR) NGS data, 9% (n=89) harbored a deficient phenotype, while 91% (n=855) had a proficient phenotype. TILs were significantly higher in the MMR deficient subgroup independent of specimen site (p-adjust < 0.0001).

Conclusions

Our findings suggest that clinical implementation of the Immunoscore may require threshold adjustments based on specimen site and MMR status in CRC. Furthermore, the lower immune infiltration observed in liver metastases could potentially explain the documented poor response to immunotherapy in such tumors, as reported in the literature.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Norwegian Cancer Society.

Disclosure

A. Helland: Financial Interests, Institutional, Advisory Board, Advisory boards: Janssen, Takeda, AstraZeneca, Abbvie, Roche, BMS, Pfizer, MSD, Bayer, Lilly, Medicover; Financial Interests, Institutional, Invited Speaker, talks at meetings: AstraZeneca, Roche, Abbvie, Pfizer; Financial Interests, Institutional, Coordinating PI, BMS provides drug to patients in an investigator initiated clinical trial: BMS; Financial Interests, Institutional, Coordinating PI, Ultimovacs provides drug and funds for investigator initiated clinical trial: Ultimovacs; Financial Interests, Institutional, Coordinating PI, AstraZeneca provides drug and funds for investigator initiated clinical trial: AstraZeneca; Financial Interests, Institutional, Coordinating PI, Roche provides drug and funds for investigator initiated clinical trial: Roche; Financial Interests, Institutional, Coordinating PI, Novartis provides drug and funds for clinical trial: Novartis; Financial Interests, Institutional, Coordinating PI, Eli Lilly provides drug and funds for clinical stduy: Eli Lilly; Financial Interests, Institutional, Coordinating PI, Incyte provides drug and funds for clinical stduy: Incyte; Financial Interests, Institutional, Coordinating PI, Illumina provides assays for patients in a clinical trial: Illumina; Financial Interests, Institutional, Coordinating PI, GSK provides drug and funds for investigator initiated clinical trial: GSK; Non-Financial Interests, Other, Board member in the patient organisation until 2022. Provides advice and gives talks.: The lung cancer patients organisation. D.J. Kwiatkowski: Non-Financial Interests, Institutional, Funding: Genentech, AADI, Revolution Medicines; Non-Financial Interests, Institutional, Advisory Role: Genentech, AADI, Expertconnect, Guidepoint, Bridgebio, Slingshot Insights, William Blair, MEDACorp, Radyus Research. All other authors have declared no conflicts of interest.