Deep learning to predict breast cancer sentinel lymph node status on INSEMA histological images

Abstract

Background: Sentinel lymph node (SLN) status is a clinically important prognostic biomarker in breast cancer and is used to guide therapy, especially for hormone receptor-positive, HER2-negative cases. However, invasive lymph node staging is increasingly omitted before therapy, and studies such as the randomised Intergroup Sentinel Mamma (INSEMA) trial address the potential for further de-escalation of axillary surgery. Therefore, it would be helpful to accurately predict the pretherapeutic sentinel status using medical images.

Methods: Using a ResNet 50 architecture pretrained on ImageNet and a previously successful strategy, we trained deep learning (DL)-based image analysis algorithms to predict sentinel status on hematoxylin/eosin-stained images of predominantly luminal, primary breast tumours from the INSEMA trial and three additional, independent cohorts (The Cancer Genome Atlas (TCGA) and cohorts from the University hospitals of Mannheim and Regensburg), and compared their performance with that of a logistic regression using clinical data only. Performance on an INSEMA hold-out set was investigated in a blinded manner.

Results: None of the generated image analysis algorithms yielded significantly better than random areas under the receiver operating characteristic curves on the test sets, including the hold-out test set from INSEMA. In contrast, the logistic regression fitted on the Mannheim cohort retained a better than random performance on INSEMA and Regensburg. Including the image analysis model output in the logistic regression did not improve performance further on INSEMA.

Conclusions: Employing DL-based image analysis on histological slides, we could not predict SLN status for unseen cases in the INSEMA trial and other predominantly luminal cohorts.

Keywords: Breast cancer; Deep learning; Digital biomarker; Lymph node status; Sentinel.

Conflict of interest statement

Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Titus Josef Brinker would like to disclose that he is the owner of Smart Health Heidelberg GmbH (Handschuhsheimer Landstr. 9/1, 69120 Heidelberg, Germany, https://smarthealth.de) which develops teledermatology mobile apps, outside of the submitted work. DK received honoraria from Merck Sharp & Dohme, Pfizer, Medupdate, Onkowissen, best practice Onkologie, ESO, ESMO and Gilead as well as research funding from Merck KGaA, outside of the submitted work. Sibylle Loibl reports grants and other from Abbvie, other from Amgen, grants and other from AstraZeneca, other from BMS, grants and other from Celgene, grants, non-financial support and other from Daiichi-Sankyo, other from Eirgenix, other from Eisai Europe Ltd, other from GSK, grants, non-financial support and other from Immunomedics/Gilead, other from Lilly, other from Merck, grants from Molecular Health, grants, non-financial support and other from Novartis, grants, non-financial support and other from Pfizer, other from Pierre Fabre, other from Relay Therapeutics, grants, non-financial support and other from Roche, other from Sanofi, non-financial support and other from Seagen, other from Olema Pharmaceutics, other from VMscope GmbH, outside the submitted work. In addition, Sibylle Loibl has a patent EP21152186.9 pending, a patent EP19808852.8 pending, and a patent EP14153692.0 pending. Valentina Nekljudova declares to be GBG Forschungs GmbH employee. GBG Forschungs GmbH received funding for research grants from Abbvie, Amgen, AstraZeneca, BMS, Daiichi-Sankyo, Gilead, Molecular Health, Novartis, Pfizer and Roche (paid to the institution); other (non-financial/medical writing) from Daiichi-Sankyo, Gilead, Novartis, Pfizer, Roche, and Seagen (paid to the institution). GBG Forschungs GmbH has licensing fees from VMscope GmbH. In addition, GBG Forschungs GmbH has a patent EP21152186.9 pending, a patent EP19808852.8 pending, and a patent EP14153692.0 pending. All other authors have no conflicts of interest to disclose.

Copyright © 2023 The Authors. Published by Elsevier Ltd.. All rights reserved.