EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Multiple instance learning-based prediction of programmed death-ligand 1 (PD-L1) expression from hematoxylin and eosin (H&E)-stained histopathological images in breast cancer

Zhen Da, Heng Yang, Bianba Zhaxi, Kaixiang Sun, Guohui Bai, Chao Wang, Feiyan Wang, Weijun Pan, Rui Du

PeerJ · 2025

DOI: 10.7717/peerj.19201 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PeerJ, 2025. Verfasst von Zhen Da und 8 weiteren Autor:innen. Beteiligte Einrichtung: Department of Pathology, People's Hospital of Xizang Autonomous Region, Lhasa, Xizang, China. Hauptthemen laut MEDLINE: B7-H1 Antigen, Breast Neoplasms, Deep Learning, Biomarkers, Tumor. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Xizang Natural Science Foundation, 2022 Major Science and Technology Innovation R&D Project of Chongqing Municipality.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Programmed death-ligand 1 (PD-L1) is an important biomarker increasingly used as a predictive marker in breast cancer immunotherapy. Immunohistochemical quantification remains the standard method for assessment. However, it presents challenges related to time, cost, and reliability. Hematoxylin and eosin (H&E) staining is a routine method in cancer pathology, known for its accessibility and consistently reliability. Deep learning has shown the potential in predicting biomarkers in cancer histopathology. This study employs a weakly supervised multiple instance learning (MIL) approach to predict PD-L1 expression from H&E-stained images using deep learning techniques. In the internal test set, the TransMIL method achieved an area under the curve (AUC) of 0.833, and in an independent external test set, it achieved an AUC of 0.799. Additionally, since RNA sequencing results indicate a threshold that allows for the separation of H&E pathology images, we further validated our approach using the public TCGA-TNBC dataset, achieving an AUC of 0.721. These findings demonstrates that the Transformer-based TransMIL model can effectively capture highly heterogeneous features within the MIL framework, exhibiting strong cross-center generalization capabilities. Our study highlights that appropriate deep learning techniques can enable effective PD-L1 prediction even with limited data, and across diverse regions and centers. This not only underscores the significant potential of deep learning in pathological artificial intelligence (AI) but also provides valuable insights for the rational and efficient allocation of medical resources.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

B7-H1 AntigenBreast NeoplasmsDeep LearningBiomarkers, TumorHumansFemaleHematoxylinEosine Yellowish-(YS)Staining and LabelingReproducibility of ResultsImmunohistochemistryMultiple-Instance Learning Algorithms

Stichwörter der Autor:innen

Breast cancer diagnosisImmunohistochemistry (IHC)Multiple instance learning (MIL)PD-L1 expressionWeakly supervised learning

Substanzen

B7-H1 AntigenCD274 protein, humanHematoxylinEosine Yellowish-(YS)Biomarkers, Tumor

Autor:innen und Institutionen

  • Department of Pathology, People's Hospital of Xizang Autonomous Region, Lhasa, Xizang, China.
  • Department of Pathology, Guangdong Women and Children Hospital, Guangzhou, Guangdong, China.
  • Kunyuan Fangqing Medical Technology Co., LTD, Guangzhou, Guangdong, China.
  • Jinfeng Laboratory, Chongqing, China.
  • Department of General Surgery, People's Hospital of Xizang Autonomous Region, Lhasa, Xizang, China.
  • Department of Pathology, Nanfang Hospital and Basic Medical College, Southern Medical University, Guangzhou, Guangdong, China.

Förderung

  • Xizang Natural Science Foundation
  • 2022 Major Science and Technology Innovation R&D Project of Chongqing Municipality

Zitieren

APA

Da, Z., Yang, H., Zhaxi, B., Sun, K., Bai, G., Wang, C., Wang, F., Pan, W., & Du, R. (2025). Multiple instance learning-based prediction of programmed death-ligand 1 (PD-L1) expression from hematoxylin and eosin (H&E)-stained histopathological images in breast cancer. PeerJ. https://doi.org/10.7717/peerj.19201

Vancouver

Da Z, Yang H, Zhaxi B, Sun K, Bai G, Wang C, et al. Multiple instance learning-based prediction of programmed death-ligand 1 (PD-L1) expression from hematoxylin and eosin (H&E)-stained histopathological images in breast cancer. PeerJ. 2025. doi:10.7717/peerj.19201

BibTeX
@article{Da2025,
  author = {Da, Zhen and Yang, Heng and Zhaxi, Bianba and Sun, Kaixiang and Bai, Guohui and Wang, Chao and Wang, Feiyan and Pan, Weijun and Du, Rui},
  title = {Multiple instance learning-based prediction of programmed death-ligand 1 (PD-L1) expression from hematoxylin and eosin (H&E)-stained histopathological images in breast cancer},
  journal = {PeerJ},
  year = {2025},
  doi = {10.7717/peerj.19201}
}
Kostenlos starten →