DeepRePath: Identifying the Prognostic Features of Early-Stage Lung Adenocarcinoma Using Multi-Scale Pathology Images and Deep Convolutional Neural Networks

Abstract

The prognosis of patients with lung adenocarcinoma (LUAD), especially early-stage LUAD, is dependent on clinicopathological features. However, its predictive utility is limited. In this study, we developed and trained a DeepRePath model based on a deep convolutional neural network (CNN) using multi-scale pathology images to predict the prognosis of patients with early-stage LUAD. DeepRePath was pre-trained with 1067 hematoxylin and eosin-stained whole-slide images of LUAD from the Cancer Genome Atlas. DeepRePath was further trained and validated using two separate CNNs and multi-scale pathology images of 393 resected lung cancer specimens from patients with stage I and II LUAD. Of the 393 patients, 95 patients developed recurrence after surgical resection. The DeepRePath model showed average area under the curve (AUC) scores of 0.77 and 0.76 in cohort I and cohort II (external validation set), respectively. Owing to low performance, DeepRePath cannot be used as an automated tool in a clinical setting. When gradient-weighted class activation mapping was used, DeepRePath indicated the association between atypical nuclei, discohesive tumor cells, and tumor necrosis in pathology images showing recurrence. Despite the limitations associated with a relatively small number of patients, the DeepRePath model based on CNNs with transfer learning could predict recurrence after the curative resection of early-stage LUAD using multi-scale pathology images.

Tumor Budding as a Marker for Poor Prognosis and Epithelial–Mesenchymal Transition in Lung Cancer: A Systematic Review and Meta-Analysis

“…However, in a previous study, 30% of patients at those stages showed disease recurrence ( 6 , 32 ). Actually, clinicians have some difficulty deciding on chemotherapy to prevent recurrence in stage I patients ( 6 ). We suggest that TB may help identify patients at high risk of recurrence and offer them adjuvant chemotherapy, especially in cases where they otherwise may not receive necessary treatment.…”

“…Moreover, adjuvant chemotherapy for lung cancer is limited to patients with lung cancer beyond stage I. However, in a previous study, 30% of patients at those stages showed disease recurrence ( 6 , 32 ). Actually, clinicians have some difficulty deciding on chemotherapy to prevent recurrence in stage I patients ( 6 ).…”

“…This might be because the TB-YN method is simple and more uniform than the TB-1HPF method. However, considering that TB imitators, like macrophages, tangentially sectioned tumor glands, or apoptotic tumor cells, can be interpreted as high-grade TB using the TB-YN method, dividing TB as present or absent could still result in an error ( 6 , 10 , 12 ). Therefore, standardization of the TB assessment method is urgently needed to predict a precise prognosis.…”

“…Despite advancements in the present treatment, most patients commonly experience recurrence and still have a poor prognosis ( 5 ). Adjuvant chemotherapy is essential for some patients with resected lung cancer to attain improved clinical outcomes; however, it is unclear which patients benefit from adjuvant chemotherapy ( 6 ). Therefore, it is important to predict an accurate prognosis.…”

Deep learning can predict survival directly from histology in clear cell renal cell carcinoma

“…These mutations apparently lead to morphological changes which are detectable by artificial intelligence techniques, especially convolutional neural networks (CNNs). Such works have laid the ground for the rising number of studies investigating AI for the prediction of oncological outcomes [ 15 , 16 ].…”