Development of a deep-learning model tailored for HER2 detection in breast cancer to aid pathologists in interpreting HER2-low cases.

AIMS: Over 50% of breast cancer cases are "Human epidermal growth factor receptor 2 (HER2) low breast cancer (BC)", characterized by HER2 immunohistochemistry (IHC) scores of 1+ or 2+ alongside no amplification on fluorescence in situ hybridization (FISH) testing. The development of new anti-HER2 antibody-drug conjugates (ADCs) for treating HER2-low breast cancers illustrates the importance of accurately assessing HER2 status, particularly HER2-low breast cancer. In this study we evaluated the performance of a deep-learning (DL) model for the assessment of HER2, including an assessment of the causes of discordances of HER2-Null between a pathologist and the DL model. We specifically focussed on aligning the DL model rules with the ASCO/CAP guidelines, including stained cells' staining intensity and completeness of membrane staining. METHODS AND RESULTS: We trained a DL model on a multicentric cohort of breast cancer cases with HER2-IHC scores (n = 299). The model was validated on two independent multicentric validation cohorts (n = 369 and n = 92), with all cases reviewed by three senior breast pathologists. All cases underwent a thorough review by three senior breast pathologists, with the ground truth determined by a majority consensus on the final HER2 score among the pathologists. In total, 760 breast cancer cases were utilized throughout the training and validation phases of the study. The model's concordance with the ground truth (ICC = 0.77 [0.68-0.83]; Fisher P = 1.32e-10) is higher than the average agreement among the three senior pathologists (ICC = 0.45 [0.17-0.65]; Fisher P = 2e-3). In the two validation cohorts, the DL model identifies 95% [93% - 98%] and 97% [91% - 100%] of HER2-low and HER2-positive tumours, respectively. Discordant results were characterized by morphological features such as extended fibrosis, a high number of tumour-infiltrating lymphocytes, and necrosis, whilst some artefacts such as nonspecific background cytoplasmic stain in the cytoplasm of tumour cells also cause discrepancy. CONCLUSION: Deep learning can support pathologists' interpretation of difficult HER2-low cases. Morphological variables and some specific artefacts can cause discrepant HER2-scores between the pathologist and the DL model.

Deep learning predicts patients outcome and mutations from digitized histology slides in gastrointestinal stromal tumor.

Risk assessment of gastrointestinal stromal tumor (GIST) according to the AFIP/Miettinen classification and mutational profiling are major tools for patient management. However, the AFIP/Miettinen classification depends heavily on mitotic counts, which is laborious and sometimes inconsistent between pathologists. It has also been shown to be imperfect in stratifying patients. Molecular testing is costly and time-consuming, therefore, not systematically performed in all countries. New methods to improve risk and molecular predictions are hence crucial to improve the tailoring of adjuvant therapy. We have built deep learning (DL) models on digitized HES-stained whole slide images (WSI) to predict patients' outcome and mutations. Models were trained with a cohort of 1233 GIST and validated on an independent cohort of 286 GIST. DL models yielded comparable results to the Miettinen classification for relapse-free-survival prediction in localized GIST without adjuvant Imatinib (C-index=0.83 in cross-validation and 0.72 for independent testing). DL splitted Miettinen intermediate risk GIST into high/low-risk groups (p value = 0.002 in the training set and p value = 0.29 in the testing set). DL models achieved an area under the receiver operating characteristic curve (AUC) of 0.81, 0.91, and 0.71 for predicting mutations in KIT, PDGFRA and wild type, respectively, in cross-validation and 0.76, 0.90, and 0.55 in independent testing. Notably, PDGFRA exon18 D842V mutation, which is resistant to Imatinib, was predicted with an AUC of 0.87 and 0.90 in cross-validation and independent testing, respectively. Additionally, novel histological criteria predictive of patients' outcome and mutations were identified by reviewing the tiles selected by the models. As a proof of concept, our study showed the possibility of implementing DL with digitized WSI and may represent a reproducible way to improve tailoring therapy and precision medicine for patients with GIST.

Federated learning for predicting histological response to neoadjuvant chemotherapy in triple-negative breast cancer.

Triple-negative breast cancer (TNBC) is a rare cancer, characterized by high metastatic potential and poor prognosis, and has limited treatment options. The current standard of care in nonmetastatic settings is neoadjuvant chemotherapy (NACT), but treatment efficacy varies substantially across patients. This heterogeneity is still poorly understood, partly due to the paucity of curated TNBC data. Here we investigate the use of machine learning (ML) leveraging whole-slide images and clinical information to predict, at diagnosis, the histological response to NACT for early TNBC women patients. To overcome the biases of small-scale studies while respecting data privacy, we conducted a multicentric TNBC study using federated learning, in which patient data remain secured behind hospitals' firewalls. We show that local ML models relying on whole-slide images can predict response to NACT but that collaborative training of ML models further improves performance, on par with the best current approaches in which ML models are trained using time-consuming expert annotations. Our ML model is interpretable and is sensitive to specific histological patterns. This proof of concept study, in which federated learning is applied to real-world datasets, paves the way for future biomarker discovery using unprecedentedly large datasets.

An artificial intelligence model predicts the survival of solid tumour patients from imaging and clinical data.

BACKGROUND: The need for developing new biomarkers is increasing with the emergence of many targeted therapies. Artificial Intelligence (AI) algorithms have shown great promise in the medical imaging field to build predictive models. We developed a prognostic model for solid tumour patients using AI on multimodal data. PATIENTS AND METHODS: Our retrospective study included examinations of patients with seven different cancer types performed between 2003 and 2017 in 17 different hospitals. Radiologists annotated all metastases on baseline computed tomography (CT) and ultrasound (US) images. Imaging features were extracted using AI models and used along with the patients' and treatments' metadata. A Cox regression was fitted to predict prognosis. Performance was assessed on a left-out test set with 1000 bootstraps. RESULTS: The model was built on 436 patients and tested on 196 patients (mean age 59, IQR: 51-6, 411 men out of 616 patients). On the whole, 1147 US images were annotated with lesions delineation, and 632 thorax-abdomen-pelvis CTs (total of 301,975 slices) were fully annotated with a total of 9516 lesions. The developed model reaches an average concordance index of 0.71 (0.67-0.76, 95% CI). Using the median predicted risk as a threshold value, the model is able to significantly (log-rank test P value < 0.001) isolate high-risk patients from low-risk patients (respective median OS of 11 and 31 months) with a hazard ratio of 3.5 (2.4-5.2, 95% CI). CONCLUSION: AI was able to extract prognostic features from imaging data, and along with clinical data, allows an accurate stratification of patients' prognoses.

Comprehensive Molecular and Pathologic Evaluation of Transitional Mesothelioma Assisted by Deep Learning Approach: A Multi-Institutional Study of the International Mesothelioma Panel from the MESOPATH Reference Center.

INTRODUCTION: Histologic subtypes of malignant pleural mesothelioma are a major prognostic indicator and decision denominator for all therapeutic strategies. In an ambiguous case, a rare transitional mesothelioma (TM) pattern may be diagnosed by pathologists either as epithelioid mesothelioma (EM), biphasic mesothelioma (BM), or sarcomatoid mesothelioma (SM). This study aimed to better characterize the TM subtype from a histological, immunohistochemical, and molecular standpoint. Deep learning of pathologic slides was applied to this cohort. METHODS: A random selection of 49 representative digitalized sections from surgical biopsies of TM was reviewed by 16 panelists. We evaluated BAP1 expression and CDKN2A (p16) homozygous deletion. We conducted a comprehensive, integrated, transcriptomic analysis. An unsupervised deep learning algorithm was trained to classify tumors. RESULTS: The 16 panelists recorded 784 diagnoses on the 49 cases. Even though a Kappa value of 0.42 is moderate, the presence of a TM component was diagnosed in 51%. In 49% of the histological evaluation, the reviewers classified the lesion as EM in 53%, SM in 33%, or BM in 14%. Median survival was 6.7 months. Loss of BAP1 observed in 44% was less frequent in TM than in EM and BM. p16 homozygous deletion was higher in TM (73%), followed by BM (63%) and SM (46%). RNA sequencing unsupervised clustering analysis revealed that TM grouped together and were closer to SM than to EM. Deep learning analysis achieved 94% accuracy for TM identification. CONCLUSION: These results revealed that the TM pattern should be classified as non-EM or at minimum as a subgroup of the SM type.

Deep learning-based classification of mesothelioma improves prediction of patient outcome.

Malignant mesothelioma (MM) is an aggressive cancer primarily diagnosed on the basis of histological criteria1. The 2015 World Health Organization classification subdivides mesothelioma tumors into three histological types: epithelioid, biphasic and sarcomatoid MM. MM is a highly complex and heterogeneous disease, rendering its diagnosis and histological typing difficult and leading to suboptimal patient care and decisions regarding treatment modalities2. Here we have developed a new approach-based on deep convolutional neural networks-called MesoNet to accurately predict the overall survival of mesothelioma patients from whole-slide digitized images, without any pathologist-provided locally annotated regions. We validated MesoNet on both an internal validation cohort from the French MESOBANK and an independent cohort from The Cancer Genome Atlas (TCGA). We also demonstrated that the model was more accurate in predicting patient survival than using current pathology practices. Furthermore, unlike classical black-box deep learning methods, MesoNet identified regions contributing to patient outcome prediction. Strikingly, we found that these regions are mainly located in the stroma and are histological features associated with inflammation, cellular diversity and vacuolization. These findings suggest that deep learning models can identify new features predictive of patient survival and potentially lead to new biomarker discoveries.