Abstract 30P
Background
Immune checkpoint inhibitors treatments have proven their efficiency against Non-Small Cell Lung Cancer (NSCLC). However a large number of patients (around 45%) don’t respond to immunotherapies and despite the existing biomarkers (PDL1, CD8) and patient selection practices, there is not currently any relevant predictive biomarker of this response. The recent advances in Deep learning image analysis - namely Multiple Instance Learning - allow to train a model to predict survival for patients receiving immunotherapy and to confront visual results to known biomarkers.
Methods
2220 H&E histology images were gathered from 456 NSCLC patients. The 36-month follow-up with vital status is retrospectively documented (68.5% alive, and 31.5% deceased). The slides are first tiled into patches and histology phenotypes are extracted using K-means on VGG-16 feature extractor. The phenotypes are then fed into siamese networks and aggregated into a single patient representation through an Attention layer predicting the vital status (alive / deceased). The Attention map is finally used to extract the most important phenotypes driving the model decision of immunotherapy response. The result of the attention maps is compared with the CD8 and PDL1 biomarker stainings.
Results
We find that a relevant number of histologically phenotypes used for the analysis is 4 and that, based on tiles coming from each of these four phenotypes the survival prediction can be performed with an AUC of 0.75. Qualitative analysis shows that the model decision for predicting response to immunotherapy from H&E stained slide not only rely on PDL1 biomarker but also how immune cells infiltrate the tumoral PDL1 area.
Conclusions
In this study we use a new deep learning approach enabling us to discover new histological phenotypes that predict the response to immunotherapy in NSCLC. Our first results are encouraging and demonstrate the AI capability for discovering new discriminative phenotypes. Our future work will be focused on inspecting more granular phenotypes.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Tribun Health.
Disclosure
All authors have declared no conflicts of interest.
Resources from the same session
2P - SOT102, a novel CLDN18.2-targeting antibody-drug conjugate for gastric and pancreatic cancer with a wide range of the tumor target expression
Presenter: Radek Spisek
Session: Cocktail & Poster Display session
Resources:
Abstract
3P - Correlation between antibody-drug conjugate (ADC) targetable antigen expression and occurrence of interstitial lung disease (ILD)
Presenter: Solange Peters
Session: Cocktail & Poster Display session
Resources:
Abstract
4P - Optimizing utilization of antibody-drug conjugates in NSCLC by identification of subsets using RNA sequencing
Presenter: Edwin Lin
Session: Cocktail & Poster Display session
Resources:
Abstract
5P - Tisotumab vedotin (TV) dose schedule optimization in non-cervical populations
Presenter: Jenna Voellinger
Session: Cocktail & Poster Display session
Resources:
Abstract
6P - RC48-ADC for metastatic salivary duct carcinoma with HER2 expression: A single-center retrospective study
Presenter: jun Cao
Session: Cocktail & Poster Display session
Resources:
Abstract
7P - Global ring study determining reproducibility & comparability of CLDN18 testing assays in gastric cancer
Presenter: Bharat Jasani
Session: Cocktail & Poster Display session
Resources:
Abstract
8P - Targeting antiapoptotic Bcl-2 proteins with highly specific BH3 mimetics in solid tumors
Presenter: Benjamin Sobol
Session: Cocktail & Poster Display session
Resources:
Abstract
9P - An in vivo model for therapeutic antibody efficacy evaluation: The chicken embryo’s CAM-based assay
Presenter: Yan Wang
Session: Cocktail & Poster Display session
Resources:
Abstract
11P - Successful transfer and prolonged persistence of engineered lymphocytes with T-cell receptor targeting NY-ESO-1
Presenter: Johnathan Arnon
Session: Cocktail & Poster Display session
Resources:
Abstract
12P - CRISPR/Cas9-induced knock-out of DGKαζ in TAG-72 CAR-T cells improves function and persistence in ovarian cancer
Presenter: Vera Evtimov
Session: Cocktail & Poster Display session
Resources:
Abstract