Development and validation of deep learning classifiers to detect Epstein-Barr virus and microsatellite instability status in gastric cancer: a retrospective multicentre cohort study

Muti, Hannah Sophie; Heij, Lara; Keller, Gisela; Kohlruss, Meike; Langer, Rupert; Dislich, Bastian; Cheong, Jae‐Ho; Kim, Young Woo; Kim, Hyunki; Kook, Myeong–Cherl; Cunningham, David; Allum, William; Langley, Ruth E.; Nankivell, Matthew; Quirke, Philip; Hayden, Jeremy D.; West, Nicholas P.; Irvine, Andrew J; Yoshikawa, Takaki; Oshima, Takashi; Huss, Ralf; Grosser, Bianca; Roviello, Franco; D’Ignazio, Alessia; Quaas, Alexander; Alakus, Hakan; Tan, Xiuxiang; Pearson, Alexander T.; Luedde, Tom; Ebert, Matthias; Jäger, Dirk; Trautwein, Christian; Gaisa, Nadine T.; Grabsch, Heike; Kather, Jakob Nikolas

doi:10.1016/s2589-7500(21)00133-3

Cited by 79 publications

(64 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…To predict LNM status and dMMR status, we used our own open‐source algorithm [10] that has been extensively validated in colorectal cancer [35], bladder cancer [36], and gastric cancer [37]. In brief, we trained a ShuffleNet [38] network model with transfer learning for end‐to‐end prediction of LNM status.…”

Section: Methodsmentioning

confidence: 99%

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

Brockmoeller

Echle

Laleh

et al. 2021

The Journal of Pathology

Self Cite

View full text Add to dashboard Cite

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.

show abstract

Section: Methodsmentioning

confidence: 99%

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

Brockmoeller

Echle

Laleh

et al. 2021

The Journal of Pathology

Self Cite

View full text Add to dashboard Cite

show abstract

“…In addition, we evaluated the baseline performance of CNN and ViT on subtyping of gastric cancer. 32,45 When trained on the TCGA-GASTRIC cohort (N=191 patients, Suppl. Figure 1C ) and tested on the BERN cohort (N=249 patients, Suppl.…”

Section: Resultsmentioning

confidence: 99%

Adversarial attacks and adversarial robustness in computational pathology

Laleh

Truhn

Veldhuizen

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Artificial Intelligence (AI) can support diagnostic workflows in oncology by aiding diagnosis and providing biomarkers. AI applications are therefore expected to evolve from academic prototypes to commercial products in the coming years. However, AI applications are vulnerable to adversarial attacks, such as malicious interference with test data aiming to cause misclassifications. Therefore, it is essential for the use of AI-based diagnostic devices to secure them against such attacks before widespread use. Unfortunately, no resistant systems exist in computational pathology so far. To address this problem, we investigate the susceptibility of convolutional neural networks (CNNs) to multiple types of white- and black-box attacks. We demonstrate that both attacks can easily confuse CNNs in clinically relevant pathology tasks and impair classification performance. Classical adversarially robust training and dual batch normalization (DBN) are possible mitigation strategies but require precise knowledge of the type of attack used in the inference. We demonstrate that vision transformers (ViTs) perform equally well compared to CNNs at baseline and are orders of magnitude more robust to different types of white-box and black-box attacks. At a mechanistic level, we show that this is associated with a more robust latent representation of clinically relevant categories in ViTs compared to CNNs. Our results are in line with previous theoretical studies. We show that ViTs are robust learners in computational pathology. This implies that large-scale rollout of AI models in computational pathology should rely on ViTs rather than CNN-based classifiers to provide inherent protection against adversaries.

show abstract

“…Previous studies have shown that both TMB and MSI could be used as biomarkers to predict prognosis after immunotherapy in a variety of tumors (47)(48)(49)(50). As an emerging and promising biomarker for tumor prediction and an important potential biomarker for immune checkpoint inhibitors, TMB and MSI may synergistically open up a new perspective for precision immunotherapy (33,(51)(52)(53)(54). This study further revealed that the expression level of ANXA2P2 has relevance with TMB and MSI in various tumors, indicating that the expression level of ANXA2P2 would impact the TMB and MSI in many tumors, thus affecting the patient's response to immune checkpoint inhibition therapy.…”

Section: Discussionmentioning

confidence: 99%

ANXA2P2: A Potential Immunological and Prognostic Signature in Ovarian Serous Cystadenocarcinoma via Pan-Carcinoma Synthesis

Zhang

Chen

2022

Front. Oncol.

View full text Add to dashboard Cite

BackgroundAlthough the effect of pseudogene ANXA2P2 on some tumors has been reported in a few literatures, the therapeutic potential and prognostic value of ANXA2P2 in ovarian serous cystadenocarcinoma (OV) have not been elucidated.MethodsThe correlation for ANXA2P2 expression patterns to prognostic characteristics, tumor immune microenvironment, immune cell infiltration level, tumor mutation burden (TMB), tumor microsatellite instability (MSI), drug sensitivity, and pathway function enrichment were investigated in pan-carcinoma via TCGA and GTEx databases. Subsequently, the role of ANXA2P2 expression levels in the pathway enrichments and prognosis prediction in OV were further explored using weighted correlation network analysis (WGCNA) analysis, gene mutation analysis, and risk-independent prognostic analysis.ResultsANXA2P2 was frequently overexpressed in a variety of tumors compared with normal tissues. The correlation analysis for prognostic characteristics, tumor immune microenvironment, immune cell infiltration level, TMB, MSI, drug sensitivity, and pathway function enrichment revealed that ANXA2P2 expression patterns might deal a significant impact on the pathogenesis, development, and prognosis of various tumors. Then, GSVA, GSEA, WGCNA, gene mutation, and independent prognostic analysis for OV have indicated that high expression in ANXA2P2 could be mostly enriched in TNF-α signaling-via-NF-κB, epithelial-mesenchymal transition, apical junction, IL-6-JAK STAT3 signaling, etc., which were also proved to act as crucial factors on tumorigenesis, development, invasion, and metastasis. The mutation of TP53 (94%), TTN (24%), and CSMD3 (9%) in the biological process of tumor had been confirmed by relevant studies. Finally, the independent prognostic analysis demonstrated that ANXA2P2 expression in OV contributes greatly to the dependability of 3- and 5-year survival prediction.ConclusionIn summary, our findings might provide a helpful foundation for prospective explorative researches, afford new strategies for the clinical treatment, deal prognosis prediction, and give new hope for OV patients.

show abstract

Development and validation of deep learning classifiers to detect Epstein-Barr virus and microsatellite instability status in gastric cancer: a retrospective multicentre cohort study

Cited by 79 publications

References 40 publications

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

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

Adversarial attacks and adversarial robustness in computational pathology

ANXA2P2: A Potential Immunological and Prognostic Signature in Ovarian Serous Cystadenocarcinoma via Pan-Carcinoma Synthesis

Contact Info

Product

Resources

About