1563P - Tumor-intrinsic subtypes of esophageal adenocarcinoma associate cellular phenotypes with responses to therapy

Background

Effective treatment of esophageal adenocarcinoma (EAC) is hampered by a high degree of mesenchymal plasticity that contributes to acquired therapy resistance and disease recurrence. Many trials are or have been investigating new treatments, such as additional epidermal growth factor signaling inhibition, to improve patient outcome. However, patient selection methods are limited. The existence of a tissue-level molecular subtype that could be used to stratify patients has remained elusive, in large part due to an abundance of non-tumor cells confounding bulk gene expression data.

Methods

We profiled gene expression from 186 esophageal cancer samples and applied non-negative matrix factorization to identify tumor-intrinsic features. Subgroups were discovered using consensus clustering of the tumor-intrinsic gene expression signatures and single-nucleus RNA sequencing analysis.

Results

Two subtypes were identified; intestinal-like (IL) and mesenchymal-like (ML). The latter enriched following neoadjuvant treatment and during disease progression. In a separate cohort of RNA-sequenced metastatic EAC, the ML subtype was found to predict poor response to treatment. Conversely, IL subtype cancer cells were highly dependent on epidermal growth factor signalling, and sensitive to receptor tyrosine kinase inhibition. Mechanisms behind the dependence on epidermal growth factor signalling are currently elucidated using comprehensive analyses.

Conclusions

Targeting of epidermal growth factor signaling is currently only available for HER2-positive patients, about 15% of all cases. Based on our tumor-intrinsic subtypes, a larger fraction of EAC patients seem to benefit from epidermal growth factor inhibition. Therefore, the ability to assign subtype labels in the clinic will be an important step forward to identify patients who may benefit from additional epidermal growth factor inhibition.

Clinical trial identification

NCT05188313.

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

KWF.

Disclosure

H.W.M. van Laarhoven: Financial Interests, Institutional, Advisory Board: BMS, MSD; Financial Interests, Personal, Invited Speaker: Lilly; Financial Interests, Institutional, Writing Engagement: Nordic Pharma; Financial Interests, Institutional, Invited Speaker: Servier; Financial Interests, Institutional, Research Grant, REPEAT study: Bayer; Financial Interests, Institutional, Research Grant: BMS, Philips; Financial Interests, Institutional, Local PI, FRACTION study: BMS; Financial Interests, Institutional, Research Grant, ACTION study: Celgene; Financial Interests, Institutional, Research Grant, DECO study: Janssen; Financial Interests, Institutional, Local PI, RAINFALL study: Lilly; Financial Interests, Institutional, Local PI, KEYNOTE 062 and KEYNOTE 181 study: Merck/MSD; Financial Interests, Institutional, Research Grant, SOX study: Nordic Pharma; Financial Interests, Institutional, Research Grant, TRAP study, PERFECT study; local PI of JACOB study: Roche; Financial Interests, Institutional, Research Grant, LyRICX study: Servier; Financial Interests, Institutional, Research Grant, TAPESTRY study: Merck; Financial Interests, Institutional, Research Grant, Research money and investigational product: Incyte; Non-Financial Interests, Institutional, Product Samples, For all clinical study mentioned, study medication is provided: See 'research funding'. M.F. Bijlsma: Financial Interests, Institutional, Other, Consultant: Servier; Financial Interests, Institutional, Funding: Celgene, Lead Pharma. All other authors have declared no conflicts of interest.