Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in context

Summary: Background: Ductal Carcinoma In Situ (DCIS) can progress to ipsilateral invasive breast cancer (IBC) but over 75% of DCIS lesions do not progress if untreated. Currently, DCIS that might progress to IBC cannot reliably be identified. Therefore, most patients with DCIS undergo treatment res...

Full description

Saved in:
Bibliographic Details
Main Authors: Shannon Doyle, Esther H. Lips, Eric Marcus, Lennart Mulder, Yat-Hee Liu, Francesco Dal Canton, Timo Kootstra, Maartje M. van Seijen, Ihssane Bouybayoune, Elinor J. Sawyer, Alastair M. Thompson, Sarah E. Pinder, Clara I. Sánchez, Jonas Teuwen, Jelle Wesseling, Jos Jonkers, Jacco van Rheenen, Marjanka Schmidt, Lodewyk F.A. Wessels, Proteeti Bhattacharjee, Alastair Thompson, Serena Nik-Zainal, Helen Davies, Andrew Futreal, Nicholas Navin, E. Shelley Hwang, Fariba Behbod, Daniel Rea, Hilary Stobart, Deborah Collyar, Donna Pinto, Ellen Verschuur, Marja van Oirsouw
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:EBioMedicine
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S235239642500194X
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850184228758618112
author Shannon Doyle
Esther H. Lips
Eric Marcus
Lennart Mulder
Yat-Hee Liu
Francesco Dal Canton
Timo Kootstra
Maartje M. van Seijen
Ihssane Bouybayoune
Elinor J. Sawyer
Alastair M. Thompson
Sarah E. Pinder
Clara I. Sánchez
Jonas Teuwen
Jelle Wesseling
Jelle Wesseling
Jos Jonkers
Jacco van Rheenen
Esther H. Lips
Marjanka Schmidt
Lodewyk F.A. Wessels
Proteeti Bhattacharjee
Alastair Thompson
Serena Nik-Zainal
Helen Davies
Elinor J. Sawyer
Andrew Futreal
Nicholas Navin
E. Shelley Hwang
Fariba Behbod
Daniel Rea
Hilary Stobart
Deborah Collyar
Donna Pinto
Ellen Verschuur
Marja van Oirsouw
author_facet Shannon Doyle
Esther H. Lips
Eric Marcus
Lennart Mulder
Yat-Hee Liu
Francesco Dal Canton
Timo Kootstra
Maartje M. van Seijen
Ihssane Bouybayoune
Elinor J. Sawyer
Alastair M. Thompson
Sarah E. Pinder
Clara I. Sánchez
Jonas Teuwen
Jelle Wesseling
Jelle Wesseling
Jos Jonkers
Jacco van Rheenen
Esther H. Lips
Marjanka Schmidt
Lodewyk F.A. Wessels
Proteeti Bhattacharjee
Alastair Thompson
Serena Nik-Zainal
Helen Davies
Elinor J. Sawyer
Andrew Futreal
Nicholas Navin
E. Shelley Hwang
Fariba Behbod
Daniel Rea
Hilary Stobart
Deborah Collyar
Donna Pinto
Ellen Verschuur
Marja van Oirsouw
author_sort Shannon Doyle
collection DOAJ
description Summary: Background: Ductal Carcinoma In Situ (DCIS) can progress to ipsilateral invasive breast cancer (IBC) but over 75% of DCIS lesions do not progress if untreated. Currently, DCIS that might progress to IBC cannot reliably be identified. Therefore, most patients with DCIS undergo treatment resembling IBC. To facilitate identification of low-risk DCIS, we developed deep learning models using histology whole-slide images (WSIs) and clinico-pathological data. Methods: We predicted invasive recurrence in patients with primary, pure DCIS treated with breast-conserving surgery using clinical Cox proportional hazards models and deep learning. Deep learning models were trained end-to-end with only WSIs or in combination with clinical data (integrative). We employed nested k-fold cross-validation (k = 5) on a Dutch multicentre dataset (n = 558). Models were also tested on the UK-based Sloane dataset (n = 94). Findings: Evaluated over 20 years on the Dutch dataset, deep learning models using only WSIs effectively stratified patients into low-risk (no recurrence) and high-risk (invasive recurrence) groups (negative predictive value (NPV) = 0.79 (95% CI: 0.74–0.83); hazard ratio (HR) = 4.48 (95% CI: 3.41–5.88, p < 0.0001); area under the receiver operating characteristic curve (AUC) = 0.75 (95% CI: 0.70–0.79)). Integrative models achieved similar results with slightly enhanced hazard ratios compared to the image-only models (NPV = 0.77 (95% CI 0.73–0.82); HR = 4.85 (95% CI 3.65–6.45, p < 0.0001); AUC = 0.75 (95% CI 0.7–0.79)). In contrast, clinical models were borderline significant (NPV = 0.64 (95% CI 0.59–0.69); HR = 1.37 (95% CI 1.03–1.81, p = 0.041); AUC = 0.57 (95% CI 0.52–0.62)). Furthermore, external validation of the models was unsuccessful, limited by the small size and low number of cases (22/94) in our external dataset, WSI quality, as well as the lack of well-annotated datasets that allow robust validation. Interpretation: Deep learning models using routinely processed WSIs hold promise for DCIS risk stratification, while the benefits of integrating clinical data merit further investigation. Obtaining a larger, high-quality external multicentre dataset would be highly valuable, as successful generalisation of these models could demonstrate their potential to reduce overtreatment in DCIS by enabling active surveillance for women at low risk. Funding: Cancer Research UK, the Dutch Cancer Society (KWF), and the Dutch Ministry of Health, Welfare and Sport.
format Article
id doaj-art-a1f2ed74e43a4f328748f8ae52dcc0a5
institution OA Journals
issn 2352-3964
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series EBioMedicine
spelling doaj-art-a1f2ed74e43a4f328748f8ae52dcc0a52025-08-20T02:17:05ZengElsevierEBioMedicine2352-39642025-06-0111610575010.1016/j.ebiom.2025.105750Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in contextShannon Doyle0Esther H. Lips1Eric Marcus2Lennart Mulder3Yat-Hee Liu4Francesco Dal Canton5Timo Kootstra6Maartje M. van Seijen7Ihssane Bouybayoune8Elinor J. Sawyer9Alastair M. Thompson10Sarah E. Pinder11Clara I. Sánchez12Jonas Teuwen13Jelle Wesseling14Jelle WesselingJos JonkersJacco van RheenenEsther H. LipsMarjanka SchmidtLodewyk F.A. WesselsProteeti BhattacharjeeAlastair ThompsonSerena Nik-ZainalHelen DaviesElinor J. SawyerAndrew FutrealNicholas NavinE. Shelley HwangFariba BehbodDaniel ReaHilary StobartDeborah CollyarDonna PintoEllen VerschuurMarja van OirsouwDivision of Radiation Oncology, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, the Netherlands; Informatics Institute, University of Amsterdam, Amsterdam, the NetherlandsDivision of Molecular Pathology, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, the NetherlandsDivision of Radiation Oncology, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, the Netherlands; Informatics Institute, University of Amsterdam, Amsterdam, the NetherlandsDivision of Molecular Pathology, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, the NetherlandsDivision of Molecular Pathology, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, the NetherlandsDepartment of Pathology, Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam, the NetherlandsDepartment of Pathology, Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam, the NetherlandsDepartment of Pathology, Leiden University Medical Center, Leiden, the NetherlandsSchool of Cancer &amp; Pharmaceutical Sciences, King's College London, UKSchool of Cancer &amp; Pharmaceutical Sciences, King's College London, UKDepartment of Surgery, Baylor College of Medicine, Houston, TX, USASchool of Cancer &amp; Pharmaceutical Sciences, King's College London, UKInformatics Institute, University of Amsterdam, Amsterdam, the NetherlandsDivision of Radiation Oncology, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, the Netherlands; Department of Medical Imaging, Radboud University Nijmegen, Nijmegen, the NetherlandsDivision of Molecular Pathology, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, the Netherlands; Department of Pathology, Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam, the Netherlands; Department of Pathology, Leiden University Medical Center, Leiden, the Netherlands; Corresponding author. Plesmanlaan 121, Room D2.038, Amsterdam 1066 CX, the Netherlands.Summary: Background: Ductal Carcinoma In Situ (DCIS) can progress to ipsilateral invasive breast cancer (IBC) but over 75% of DCIS lesions do not progress if untreated. Currently, DCIS that might progress to IBC cannot reliably be identified. Therefore, most patients with DCIS undergo treatment resembling IBC. To facilitate identification of low-risk DCIS, we developed deep learning models using histology whole-slide images (WSIs) and clinico-pathological data. Methods: We predicted invasive recurrence in patients with primary, pure DCIS treated with breast-conserving surgery using clinical Cox proportional hazards models and deep learning. Deep learning models were trained end-to-end with only WSIs or in combination with clinical data (integrative). We employed nested k-fold cross-validation (k = 5) on a Dutch multicentre dataset (n = 558). Models were also tested on the UK-based Sloane dataset (n = 94). Findings: Evaluated over 20 years on the Dutch dataset, deep learning models using only WSIs effectively stratified patients into low-risk (no recurrence) and high-risk (invasive recurrence) groups (negative predictive value (NPV) = 0.79 (95% CI: 0.74–0.83); hazard ratio (HR) = 4.48 (95% CI: 3.41–5.88, p < 0.0001); area under the receiver operating characteristic curve (AUC) = 0.75 (95% CI: 0.70–0.79)). Integrative models achieved similar results with slightly enhanced hazard ratios compared to the image-only models (NPV = 0.77 (95% CI 0.73–0.82); HR = 4.85 (95% CI 3.65–6.45, p < 0.0001); AUC = 0.75 (95% CI 0.7–0.79)). In contrast, clinical models were borderline significant (NPV = 0.64 (95% CI 0.59–0.69); HR = 1.37 (95% CI 1.03–1.81, p = 0.041); AUC = 0.57 (95% CI 0.52–0.62)). Furthermore, external validation of the models was unsuccessful, limited by the small size and low number of cases (22/94) in our external dataset, WSI quality, as well as the lack of well-annotated datasets that allow robust validation. Interpretation: Deep learning models using routinely processed WSIs hold promise for DCIS risk stratification, while the benefits of integrating clinical data merit further investigation. Obtaining a larger, high-quality external multicentre dataset would be highly valuable, as successful generalisation of these models could demonstrate their potential to reduce overtreatment in DCIS by enabling active surveillance for women at low risk. Funding: Cancer Research UK, the Dutch Cancer Society (KWF), and the Dutch Ministry of Health, Welfare and Sport.http://www.sciencedirect.com/science/article/pii/S235239642500194XDeep learningDuctal carcinoma in situMultiomic integrationRisk prediction
spellingShingle Shannon Doyle
Esther H. Lips
Eric Marcus
Lennart Mulder
Yat-Hee Liu
Francesco Dal Canton
Timo Kootstra
Maartje M. van Seijen
Ihssane Bouybayoune
Elinor J. Sawyer
Alastair M. Thompson
Sarah E. Pinder
Clara I. Sánchez
Jonas Teuwen
Jelle Wesseling
Jelle Wesseling
Jos Jonkers
Jacco van Rheenen
Esther H. Lips
Marjanka Schmidt
Lodewyk F.A. Wessels
Proteeti Bhattacharjee
Alastair Thompson
Serena Nik-Zainal
Helen Davies
Elinor J. Sawyer
Andrew Futreal
Nicholas Navin
E. Shelley Hwang
Fariba Behbod
Daniel Rea
Hilary Stobart
Deborah Collyar
Donna Pinto
Ellen Verschuur
Marja van Oirsouw
Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in context
EBioMedicine
Deep learning
Ductal carcinoma in situ
Multiomic integration
Risk prediction
title Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in context
title_full Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in context
title_fullStr Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in context
title_full_unstemmed Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in context
title_short Deep learning for predicting invasive recurrence of ductal carcinoma in situ: leveraging histopathology images and clinical featuresResearch in context
title_sort deep learning for predicting invasive recurrence of ductal carcinoma in situ leveraging histopathology images and clinical featuresresearch in context
topic Deep learning
Ductal carcinoma in situ
Multiomic integration
Risk prediction
url http://www.sciencedirect.com/science/article/pii/S235239642500194X
work_keys_str_mv AT shannondoyle deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT estherhlips deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT ericmarcus deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT lennartmulder deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT yatheeliu deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT francescodalcanton deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT timokootstra deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT maartjemvanseijen deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT ihssanebouybayoune deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT elinorjsawyer deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT alastairmthompson deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT sarahepinder deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT claraisanchez deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT jonasteuwen deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT jellewesseling deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT jellewesseling deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT josjonkers deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT jaccovanrheenen deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT estherhlips deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT marjankaschmidt deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT lodewykfawessels deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT proteetibhattacharjee deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT alastairthompson deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT serenanikzainal deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT helendavies deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT elinorjsawyer deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT andrewfutreal deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT nicholasnavin deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT eshelleyhwang deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT faribabehbod deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT danielrea deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT hilarystobart deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT deborahcollyar deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT donnapinto deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT ellenverschuur deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext
AT marjavanoirsouw deeplearningforpredictinginvasiverecurrenceofductalcarcinomainsituleveraginghistopathologyimagesandclinicalfeaturesresearchincontext