Transformer-based biomarker prediction from colorectal cancer histology: A large-scale multicentric study

Wagner SJ, Reisenbüchler D, West NP, Niehues JM, Zhu J, Foersch S, Veldhuizen GP, Quirke P, Grabsch HI, van den Brandt PA, Hutchins GG, Richman SD, Yuan T, Langer R, Jenniskens JC, Offermans K, Mueller W, Gray R, Gruber SB, Greenson JK, Rennert G, Bonner JD, Schmolze D, Jonnagaddala J, Hawkins NJ, Ward RL, Morton D, Seymour M, Magill L, Nowak M, Hay J, Koelzer VH, Church DN, Church D, Domingo E, Edwards J, Glimelius B, Gogenur I, Harkin A, Hay J, Iveson T, Jaeger E, Kelly C, Kerr R, Maka N, Morgan H, Oien K, Orange C, Palles C, Roxburgh C, Sansom O, Saunders M, Tomlinson I, Matek C, Geppert CI, Peng C, Zhi C, Ouyang X, James JA, Loughrey MB, Salto-Tellez M, Brenner H, Hoffmeister M, Truhn D, Schnabel JA, Boxberg M, Peng T, Kather JN (2023)

Publication Type: Journal article

Publication year: 2023

Journal

Book Volume: 41

Pages Range: 1650-1661.e4

Journal Issue: 9

DOI: 10.1016/j.ccell.2023.08.002

Abstract

Deep learning (DL) can accelerate the prediction of prognostic biomarkers from routine pathology slides in colorectal cancer (CRC). However, current approaches rely on convolutional neural networks (CNNs) and have mostly been validated on small patient cohorts. Here, we develop a new transformer-based pipeline for end-to-end biomarker prediction from pathology slides by combining a pre-trained transformer encoder with a transformer network for patch aggregation. Our transformer-based approach substantially improves the performance, generalizability, data efficiency, and interpretability as compared with current state-of-the-art algorithms. After training and evaluating on a large multicenter cohort of over 13,000 patients from 16 colorectal cancer cohorts, we achieve a sensitivity of 0.99 with a negative predictive value of over 0.99 for prediction of microsatellite instability (MSI) on surgical resection specimens. We demonstrate that resection specimen-only training reaches clinical-grade performance on endoscopic biopsy tissue, solving a long-standing diagnostic problem.

Authors with CRIS profile

Additional Organisation(s)

Related research project(s)

Involved external institutions

LMU Klinikum Germany (DE) Universitätsspital Zürich (USZ) Switzerland (CH) Universitätsmedizin der Johannes Gutenberg-Universität Mainz Germany (DE) Technische Universität Dresden Germany (DE) University of Leeds United Kingdom (GB) Maastricht University Netherlands (NL) Deutsches Krebsforschungszentrum (DKFZ) Germany (DE) Johannes Kepler Universität (JKU) Linz Austria (AT) University of Oxford United Kingdom (GB) City of Hope Medical Center United States (USA) (US) Technion - Israel Institute of Technology Israel (IL) University of New South Wales (UNSW) Australia (AU) University Hospitals Birmingham NHS Foundation Trust United Kingdom (GB) St James's University Hospital United Kingdom (GB) University of Birmingham United Kingdom (GB) University of Glasgow United Kingdom (GB) Jinggangshan University (JGSU) / 井冈山大学 China (CN) Guangzhou Medical University (GMU) / 广州医科大学 China (CN) Queen's University United Kingdom (GB) Universitätsklinikum Aachen (UKA) Germany (DE) Technische Universität München (TUM) Germany (DE)

How to cite

APA:

Wagner, S.J., Reisenbüchler, D., West, N.P., Niehues, J.M., Zhu, J., Foersch, S.,... Kather, J.N. (2023). Transformer-based biomarker prediction from colorectal cancer histology: A large-scale multicentric study. Cancer Cell, 41(9), 1650-1661.e4. https://dx.doi.org/10.1016/j.ccell.2023.08.002

MLA:

Wagner, Sophia J., et al. "Transformer-based biomarker prediction from colorectal cancer histology: A large-scale multicentric study." Cancer Cell 41.9 (2023): 1650-1661.e4.

BibTeX: Download