Volume 256, Issue 3 p. 269-281
Original Article

Deep learning identifies inflamed fat as a risk factor for lymph node metastasis in early colorectal cancer

Scarlet Brockmoeller

Corresponding Author

Scarlet Brockmoeller

Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK

Correspondence to: S Brockmoeller or JN Kather, Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, LS9 7TF, UK. E-mail: [email protected] or

[email protected]

Search for more papers by this author
Amelie Echle

Amelie Echle

Department of Medicine III, University Hospital RWTH Aachen, Aachen, Germany

Search for more papers by this author
Narmin Ghaffari Laleh

Narmin Ghaffari Laleh

Department of Medicine III, University Hospital RWTH Aachen, Aachen, Germany

Search for more papers by this author
Susanne Eiholm

Susanne Eiholm

Department of Pathology, Zealand University Hospital, University of Copenhagen, Roskilde, Denmark

Search for more papers by this author
Marie Louise Malmstrøm

Marie Louise Malmstrøm

Department of Surgery, Nordsjællands Hospital, Hillerod, Denmark

Search for more papers by this author
Tine Plato Kuhlmann

Tine Plato Kuhlmann

Department of Pathology, Herlev University Hospital, Copenhagen, Denmark

Search for more papers by this author
Katarina Levic

Katarina Levic

Department of Surgery, Herlev University Hospital, Copenhagen, Denmark

Search for more papers by this author
Heike Irmgard Grabsch

Heike Irmgard Grabsch

Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK

Department of Pathology, GROW School for Oncology and Developmental Biology, Maastricht University Medical Center+, Maastricht, The Netherlands

Search for more papers by this author
Nicholas P West

Nicholas P West

Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK

Search for more papers by this author
Oliver Lester Saldanha

Oliver Lester Saldanha

Department of Medicine III, University Hospital RWTH Aachen, Aachen, Germany

Search for more papers by this author
Katerina Kouvidi

Katerina Kouvidi

Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK

Search for more papers by this author
Aurora Bono

Aurora Bono

Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK

Search for more papers by this author
Lara R Heij

Lara R Heij

Department of Pathology, GROW School for Oncology and Developmental Biology, Maastricht University Medical Center+, Maastricht, The Netherlands

Institute of Pathology, University Hospital RWTH Aachen, Aachen, Germany

Department of Surgery and Transplantation, University Hospital RWTH Aachen, Aachen, Germany

Search for more papers by this author
Titus J Brinker

Titus J Brinker

Digital Biomarkers for Oncology Group, National Center for Tumour Diseases (NCT), German Cancer Research Center (DKFZ), Heidelberg, Germany

Search for more papers by this author
Ismayil Gögenür

Ismayil Gögenür

Department of Surgery, Zealand University Hospital, University of Copenhagen, Køge, Denmark

Gastrounit – Surgical Division, Center for Surgical Research, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark

Search for more papers by this author
Philip Quirke

Philip Quirke

Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK

Search for more papers by this author
Jakob Nikolas Kather

Corresponding Author

Jakob Nikolas Kather

Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK

Department of Medicine III, University Hospital RWTH Aachen, Aachen, Germany

Medical Oncology, National Center of Tumour Diseases (NCT), University Hospital Heidelberg, Heidelberg, Germany

Correspondence to: S Brockmoeller or JN Kather, Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, LS9 7TF, UK. E-mail: [email protected] or

[email protected]

Search for more papers by this author
First published: 05 November 2021
Citations: 24

Conflict of interest statement: JNK reports consulting activities for Owkin, France and Panakeia, UK. PQ has had paid roles in the English NHS bowel cancer screening programme over the course of this study. No other potential conflicts of interest were declared.

Abstract

The spread of early-stage (T1 and T2) adenocarcinomas to locoregional lymph nodes is a key event in disease progression of colorectal cancer (CRC). The cellular mechanisms behind this event are not completely understood and existing predictive biomarkers are imperfect. Here, we used an end-to-end deep learning algorithm to identify risk factors for lymph node metastasis (LNM) status in digitized histopathology slides of the primary CRC and its surrounding tissue. In two large population-based cohorts, we show that this system can predict the presence of more than one LNM in pT2 CRC patients with an area under the receiver operating curve (AUROC) of 0.733 (0.67–0.758) and patients with any LNM with an AUROC of 0.711 (0.597–0.797). Similarly, in pT1 CRC patients, the presence of more than one LNM or any LNM was predictable with an AUROC of 0.733 (0.644–0.778) and 0.567 (0.542–0.597), respectively. Based on these findings, we used the deep learning system to guide human pathology experts towards highly predictive regions for LNM in the whole slide images. This hybrid human observer and deep learning approach identified inflamed adipose tissue as the highest predictive feature for LNM presence. Our study is a first proof of concept that artificial intelligence (AI) systems may be able to discover potentially new biological mechanisms in cancer progression. Our deep learning algorithm is publicly available and can be used for biomarker discovery in any disease setting. © 2021 The Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Data availability statement

Raw data for the pT1/ pT2 cohorts are stored and administered by the early colorectal cancer study group. The corresponding author of this study is not involved in data-sharing decisions of the early colorectal cancer study group. All other data supporting the findings of this study are available from the corresponding author upon reasonable request.