A deep-learning workflow to predict upper tract urothelial carcinoma protein-based subtypes from H&E slides supporting the prioritization of patients for molecular testing.

Upper tract urothelial carcinoma (UTUC) is a rare and aggressive, yet understudied, urothelial carcinoma (UC). The more frequent UC of the bladder comprises several molecular subtypes, associated with different targeted therapies and overlapping with protein-based subtypes. However, if and how these findings extend to UTUC remains unclear. Artificial intelligence-based approaches could help elucidate UTUC's biology and extend access to targeted treatments to a wider patient audience. Here, UTUC protein-based subtypes were identified, and a deep-learning (DL) workflow was developed to predict them directly from routine histopathological H&E slides. Protein-based subtypes in a retrospective cohort of 163 invasive tumors were assigned by hierarchical clustering of the immunohistochemical expression of three luminal (FOXA1, GATA3, and CK20) and three basal (CD44, CK5, and CK14) markers. Cluster analysis identified distinctive luminal (N = 80) and basal (N = 42) subtypes. The luminal subtype mostly included pushing, papillary tumors, whereas the basal subtype diffusely infiltrating, non-papillary tumors. DL model building relied on a transfer-learning approach by fine-tuning a pre-trained ResNet50. Classification performance was measured via three-fold repeated cross-validation. A mean area under the receiver operating characteristic curve of 0.83 (95% CI: 0.67-0.99), 0.8 (95% CI: 0.62-0.99), and 0.81 (95% CI: 0.65-0.96) was reached in the three repetitions. High-confidence DL-based predicted subtypes showed significant associations (p < 0.001) with morphological features, i.e. tumor type, histological subtypes, and infiltration type. Furthermore, a significant association was found with programmed cell death ligand 1 (PD-L1) combined positive score (p < 0.001) and FGFR3 mutational status (p = 0.002), with high-confidence basal predictions containing a higher proportion of PD-L1 positive samples and high-confidence luminal predictions a higher proportion of FGFR3-mutated samples. Testing of the DL model on an independent cohort highlighted the importance to accommodate histological subtypes. Taken together, our DL workflow can predict protein-based UTUC subtypes, associated with the presence of targetable alterations, directly from H&E slides.

PITX2 as a Sensitive and Specific Marker of Midgut Neuroendocrine Tumors: Results from a Cohort of 1157 Primary Neuroendocrine Neoplasms.

As neuroendocrine tumors (NETs) often present as metastatic lesions, immunohistochemical assignment to a site of origin is one of the most important tasks in their pathologic assessment. Because a fraction of NETs eludes the typical expression profiles of their primary localization, additional sensitive and specific markers are required to improve diagnostic certainty. We investigated the expression of the transcription factor Pituitary Homeobox 2 (PITX2) in a large-scale cohort of 909 NET and 248 neuroendocrine carcinomas (NEC) according to the immunoreactive score (IRS) and correlated PITX2 expression groups with general tumor groups and primary localization. PITX2 expression (all expression groups) was highly sensitive (98.1%) for midgut-derived NET, but not perfectly specific, as non-midgut NET (especially pulmonary/duodenal) were quite frequently weak or moderately positive. The specificity rose to 99.5% for a midgut origin of NET if only a strong PITX2 expression was considered, which was found in only 0.5% (one pancreatic/one pulmonary) of non-midgut NET. In metastases of midgut-derived NET, PITX2 was expressed in all cases (87.5% strong, 12.5% moderate), whereas CDX2 was negative or only weakly expressed in 31.3% of the metastases. In NEC, a fraction of cases (14%) showed a weak or moderate PITX2 expression, which was not associated with a specific tumor localization. Our study independently validates PITX2 as a very sensitive and specific immunohistochemical marker of midgut-derived NET in a very large collective of neuroendocrine neoplasms. Therefore, our data argue toward implementation into diagnostic panels applied for NET as a firstline midgut marker.

Extracting structured information from unstructured histopathology reports using generative pre-trained transformer 4 (GPT-4).

Deep learning applied to whole-slide histopathology images (WSIs) has the potential to enhance precision oncology and alleviate the workload of experts. However, developing these models necessitates large amounts of data with ground truth labels, which can be both time-consuming and expensive to obtain. Pathology reports are typically unstructured or poorly structured texts, and efforts to implement structured reporting templates have been unsuccessful, as these efforts lead to perceived extra workload. In this study, we hypothesised that large language models (LLMs), such as the generative pre-trained transformer 4 (GPT-4), can extract structured data from unstructured plain language reports using a zero-shot approach without requiring any re-training. We tested this hypothesis by utilising GPT-4 to extract information from histopathological reports, focusing on two extensive sets of pathology reports for colorectal cancer and glioblastoma. We found a high concordance between LLM-generated structured data and human-generated structured data. Consequently, LLMs could potentially be employed routinely to extract ground truth data for machine learning from unstructured pathology reports in the future. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

End-to-end prognostication in colorectal cancer by deep learning: a retrospective, multicentre study.

BACKGROUND: Precise prognosis prediction in patients with colorectal cancer (ie, forecasting survival) is pivotal for individualised treatment and care. Histopathological tissue slides of colorectal cancer specimens contain rich prognostically relevant information. However, existing studies do not have multicentre external validation with real-world sample processing protocols, and algorithms are not yet widely used in clinical routine. METHODS: In this retrospective, multicentre study, we collected tissue samples from four groups of patients with resected colorectal cancer from Australia, Germany, and the USA. We developed and externally validated a deep learning-based prognostic-stratification system for automatic prediction of overall and cancer-specific survival in patients with resected colorectal cancer. We used the model-predicted risk scores to stratify patients into different risk groups and compared survival outcomes between these groups. Additionally, we evaluated the prognostic value of these risk groups after adjusting for established prognostic variables. FINDINGS: We trained and validated our model on a total of 4428 patients. We found that patients could be divided into high-risk and low-risk groups on the basis of the deep learning-based risk score. On the internal test set, the group with a high-risk score had a worse prognosis than the group with a low-risk score, as reflected by a hazard ratio (HR) of 4·50 (95% CI 3·33-6·09) for overall survival and 8·35 (5·06-13·78) for disease-specific survival (DSS). We found consistent performance across three large external test sets. In a test set of 1395 patients, the high-risk group had a lower DSS than the low-risk group, with an HR of 3·08 (2·44-3·89). In two additional test sets, the HRs for DSS were 2·23 (1·23-4·04) and 3·07 (1·78-5·3). We showed that the prognostic value of the deep learning-based risk score is independent of established clinical risk factors. INTERPRETATION: Our findings indicate that attention-based self-supervised deep learning can robustly offer a prognosis on clinical outcomes in patients with colorectal cancer, generalising across different populations and serving as a potentially new prognostic tool in clinical decision making for colorectal cancer management. We release all source codes and trained models under an open-source licence, allowing other researchers to reuse and build upon our work. FUNDING: The German Federal Ministry of Health, the Max-Eder-Programme of German Cancer Aid, the German Federal Ministry of Education and Research, the German Academic Exchange Service, and the EU.

Encrypted federated learning for secure decentralized collaboration in cancer image analysis.

Artificial intelligence (AI) has a multitude of applications in cancer research and oncology. However, the training of AI systems is impeded by the limited availability of large datasets due to data protection requirements and other regulatory obstacles. Federated and swarm learning represent possible solutions to this problem by collaboratively training AI models while avoiding data transfer. However, in these decentralized methods, weight updates are still transferred to the aggregation server for merging the models. This leaves the possibility for a breach of data privacy, for example by model inversion or membership inference attacks by untrusted servers. Somewhat-homomorphically-encrypted federated learning (SHEFL) is a solution to this problem because only encrypted weights are transferred, and model updates are performed in the encrypted space. Here, we demonstrate the first successful implementation of SHEFL in a range of clinically relevant tasks in cancer image analysis on multicentric datasets in radiology and histopathology. We show that SHEFL enables the training of AI models which outperform locally trained models and perform on par with models which are centrally trained. In the future, SHEFL can enable multiple institutions to co-train AI models without forsaking data governance and without ever transmitting any decryptable data to untrusted servers.

CD15 is a risk predictor and a novel target in clear cell renal cell carcinoma.

INTRODUCTION: Tumor cells use adhesion molecules like CD15 or sialylCD15 (sCD15) for metastatic spreading. We analyzed the expression of CD15 and sCD15 in clear cell renal cell carcinoma (ccRCC) regarding prognosis. METHODS: A tissue microarray containing tissue specimens of 763 patients with ccRCC was immunohistochemically stained for CD15 and sCD15, their expression quantified using digital image analysis and the impact on patients' survival analyzed. The cell lines 769p and 786o were stimulated with CD15 or control antibody in vitro and the effects on pathways activating AP-1 and tumor cell migration examined. RESULTS: ccRCC showed a broad range of CD15 and sCD15 expression. A high CD15 expression was significantly associated with favorable outcome (p<0.01) and low-grade tumor differentiation (p<0.001), whereas sCD15 had no significant prognostic value. Tumors with synchronous distant metastasis had a significantly lower CD15 expression compared to tumors without any (p<0.001) or with metachronous metastasis (p<0.01). Tumor cell migration was significantly reduced after CD15 stimulation in vitro, but there were no major effects on activating pathways of AP-1. CONCLUSION: CD15, but not sCD15, qualifies as a biomarker for risk stratification and as an interesting novel target in ccRCC. Moreover, the data indicates a contribution of CD15 to metachronous metastasis. Further research is warranted to decipher the intracellular pathways of CD15 signaling in ccRCC in order to characterize the CD15 effects on ccRCC more precisely.

[Multistain deep learning as a prognostic and predictive biomarker in colorectal cancer]. / Multistain Deep Learning zur Vorhersage von Prognose und Therapieansprechen im kolorektalen Karzinom.

The tumor immune microenvironment (TIME) plays a crucial prognostic and predictive role in solid malignancies such as colorectal cancer (CRC). Nevertheless, scoring systems based on TIME such as the Immunoscore (IS) are rarely used in clinical practice. Among other reasons, this might be due to the additional time required for manual quantification of tumor-associated immune cells or costs associated with proprietary/commercial solutions. To address these issues, we developed a multistain deep learning model (MSDLM) and trained, validated, and tested it on immunohistochemical image data of different immune cell subtypes from over 1000 patients with CRC. Our model showed high prognostic accuracy and outperformed other clinical, molecular, and immune cell-based parameters. It might also be used for therapy response prediction in rectal cancer patients undergoing neoadjuvant therapy. Leveraging artificial intelligence interpretability/explainability methods, we ascertained that the MSDLM's predictions align with recognized antitumor immune response patterns. Consequently, the AImmunoscore (AIS) could emerge as a potential TIME-based decision-making tool for clinicians.

High expression of insulinoma-associated protein 1 (INSM1) distinguishes colorectal mixed and pure neuroendocrine carcinomas from conventional adenocarcinomas with diffuse expression of synaptophysin.

Complementary to synaptophysin and chromogranin A, insulinoma-associated protein 1 (INSM1) has emerged as a sensitive marker for the diagnosis of neuroendocrine neoplasms. Since there are no comparative data regarding INSM1 expression in conventional colorectal adenocarcinomas (CRCs) and colorectal mixed adenoneuroendocrine carcinomas/neuroendocrine carcinomas (MANECs/NECs), we examined INSM1 in a large cohort of conventional CRCs and MANECs/NECs. In conventional CRC, we put a special focus on conventional CRC with diffuse expression of synaptophysin, which carry the risk of being misinterpreted as a MANEC or a NEC. We investigated INSM1 according to the immunoreactive score in our main cohort of 1,033 conventional CRCs and 21 MANECs/NECs in comparison to the expression of synaptophysin and chromogranin A and correlated the results with clinicopathological parameters and patient survival. All MANECs/NECs expressed INSM1, usually showing high or moderate expression (57% high, 34% moderate, and 9% low), which distinguished them from conventional CRCs, which were usually INSM1 negative or low, even if they diffusely expressed synaptophysin. High expression of INSM1 was not observed in conventional CRCs. Chromogranin A was negative/low in most conventional CRCs (99%), but also in most MANECs/NECs (66%). Comparable results were observed in our independent validation cohorts of conventional CRC (n = 274) and MANEC/NEC (n = 19). Similar to synaptophysin, INSM1 expression had no prognostic relevance in conventional CRCs, while true MANEC/NEC showed a highly impaired survival in univariate and multivariate analyses (e.g. disease-specific survival: p < 0.001). MANECs/NECs are a highly aggressive variant of colorectal cancer, which must be reliably identified. High expression of INSM1 distinguishes MANEC/NEC from conventional CRCs with diffuse expression of the standard neuroendocrine marker synaptophysin, which do not share the same dismal prognosis. Therefore, high INSM1 expression is a highly specific/sensitive marker that is supportive for the diagnosis of true colorectal MANEC/NEC.

Transformer-based biomarker prediction from colorectal cancer histology: A large-scale multicentric study.

Deep learning (DL) can accelerate the prediction of prognostic biomarkers from routine pathology slides in colorectal cancer (CRC). However, current approaches rely on convolutional neural networks (CNNs) and have mostly been validated on small patient cohorts. Here, we develop a new transformer-based pipeline for end-to-end biomarker prediction from pathology slides by combining a pre-trained transformer encoder with a transformer network for patch aggregation. Our transformer-based approach substantially improves the performance, generalizability, data efficiency, and interpretability as compared with current state-of-the-art algorithms. After training and evaluating on a large multicenter cohort of over 13,000 patients from 16 colorectal cancer cohorts, we achieve a sensitivity of 0.99 with a negative predictive value of over 0.99 for prediction of microsatellite instability (MSI) on surgical resection specimens. We demonstrate that resection specimen-only training reaches clinical-grade performance on endoscopic biopsy tissue, solving a long-standing diagnostic problem.

Denoising diffusion probabilistic models for 3D medical image generation.

Recent advances in computer vision have shown promising results in image generation. Diffusion probabilistic models have generated realistic images from textual input, as demonstrated by DALL-E 2, Imagen, and Stable Diffusion. However, their use in medicine, where imaging data typically comprises three-dimensional volumes, has not been systematically evaluated. Synthetic images may play a crucial role in privacy-preserving artificial intelligence and can also be used to augment small datasets. We show that diffusion probabilistic models can synthesize high-quality medical data for magnetic resonance imaging (MRI) and computed tomography (CT). For quantitative evaluation, two radiologists rated the quality of the synthesized images regarding "realistic image appearance", "anatomical correctness", and "consistency between slices". Furthermore, we demonstrate that synthetic images can be used in self-supervised pre-training and improve the performance of breast segmentation models when data is scarce (Dice scores, 0.91 [without synthetic data], 0.95 [with synthetic data]).

Generalizable biomarker prediction from cancer pathology slides with self-supervised deep learning: A retrospective multi-centric study.

Deep learning (DL) can predict microsatellite instability (MSI) from routine histopathology slides of colorectal cancer (CRC). However, it is unclear whether DL can also predict other biomarkers with high performance and whether DL predictions generalize to external patient populations. Here, we acquire CRC tissue samples from two large multi-centric studies. We systematically compare six different state-of-the-art DL architectures to predict biomarkers from pathology slides, including MSI and mutations in BRAF, KRAS, NRAS, and PIK3CA. Using a large external validation cohort to provide a realistic evaluation setting, we show that models using self-supervised, attention-based multiple-instance learning consistently outperform previous approaches while offering explainable visualizations of the indicative regions and morphologies. While the prediction of MSI and BRAF mutations reaches a clinical-grade performance, mutation prediction of PIK3CA, KRAS, and NRAS was clinically insufficient.

An overview and a roadmap for artificial intelligence in hematology and oncology.

BACKGROUND: Artificial intelligence (AI) is influencing our society on many levels and has broad implications for the future practice of hematology and oncology. However, for many medical professionals and researchers, it often remains unclear what AI can and cannot do, and what are promising areas for a sensible application of AI in hematology and oncology. Finally, the limits and perils of using AI in oncology are not obvious to many healthcare professionals. METHODS: In this article, we provide an expert-based consensus statement by the joint Working Group on "Artificial Intelligence in Hematology and Oncology" by the German Society of Hematology and Oncology (DGHO), the German Association for Medical Informatics, Biometry and Epidemiology (GMDS), and the Special Interest Group Digital Health of the German Informatics Society (GI). We provide a conceptual framework for AI in hematology and oncology. RESULTS: First, we propose a technological definition, which we deliberately set in a narrow frame to mainly include the technical developments of the last ten years. Second, we present a taxonomy of clinically relevant AI systems, structured according to the type of clinical data they are used to analyze. Third, we show an overview of potential applications, including clinical, research, and educational environments with a focus on hematology and oncology. CONCLUSION: Thus, this article provides a point of reference for hematologists and oncologists, and at the same time sets forth a framework for the further development and clinical deployment of AI in hematology and oncology in the future.

Multistain deep learning for prediction of prognosis and therapy response in colorectal cancer.

Although it has long been known that the immune cell composition has a strong prognostic and predictive value in colorectal cancer (CRC), scoring systems such as the immunoscore (IS) or quantification of intraepithelial lymphocytes are only slowly being adopted into clinical routine use and have their limitations. To address this we established and evaluated a multistain deep learning model (MSDLM) utilizing artificial intelligence (AI) to determine the AImmunoscore (AIS) in more than 1,000 patients with CRC. Our model had high prognostic capabilities and outperformed other clinical, molecular and immune cell-based parameters. It could also be used to predict the response to neoadjuvant therapy in patients with rectal cancer. Using an explainable AI approach, we confirmed that the MSDLM's decisions were based on established cellular patterns of anti-tumor immunity. Hence, the AIS could provide clinicians with a valuable decision-making tool based on the tumor immune microenvironment.

Direct prediction of genetic aberrations from pathology images in gastric cancer with swarm learning.

BACKGROUND: Computational pathology uses deep learning (DL) to extract biomarkers from routine pathology slides. Large multicentric datasets improve performance, but such datasets are scarce for gastric cancer. This limitation could be overcome by Swarm Learning (SL). METHODS: Here, we report the results of a multicentric retrospective study of SL for prediction of molecular biomarkers in gastric cancer. We collected tissue samples with known microsatellite instability (MSI) and Epstein-Barr Virus (EBV) status from four patient cohorts from Switzerland, Germany, the UK and the USA, storing each dataset on a physically separate computer. RESULTS: On an external validation cohort, the SL-based classifier reached an area under the receiver operating curve (AUROC) of 0.8092 (± 0.0132) for MSI prediction and 0.8372 (± 0.0179) for EBV prediction. The centralized model, which was trained on all datasets on a single computer, reached a similar performance. CONCLUSIONS: Our findings demonstrate the feasibility of SL-based molecular biomarkers in gastric cancer. In the future, SL could be used for collaborative training and, thus, improve the performance of these biomarkers. This may ultimately result in clinical-grade performance and generalizability.

pT3 colorectal cancer revisited: a multicentric study on the histological depth of invasion in more than 1000 pT3 carcinomas-proposal for a new pT3a/pT3b subclassification.

BACKGROUND: Pathological TNM staging (pTNM) is the strongest prognosticator in colorectal carcinoma (CRC) and the foundation of its post-operative clinical management. Tumours that invade pericolic/perirectal adipose tissue generally fall into the pT3 category without further subdivision. METHODS: The histological depth of invasion into the pericolic/perirectal fat was digitally and conventionally measured in a training cohort of 950 CRCs (Munich). We biostatistically calculated the optimal cut-off to stratify pT3 CRCs into novel pT3a (≤3 mm)/pT3b (>3 mm) subgroups, which were then validated in two independent cohorts (447 CRCs, Bayreuth/542 CRCs, Mainz). RESULTS: Compared to pT3a tumours, pT3b CRCs showed significantly worse disease-specific survival, including in pN0 vs pN+ and colonic vs. rectal cancers (DSS: P < 0.001, respectively, pooled analysis of all cohorts). Furthermore, the pT3a/pT3b subclassification remained an independent predictor of survival in multivariate analyses (e.g. DSS: P < 0.001, hazard ratio: 4.41 for pT3b, pooled analysis of all cohorts). While pT2/pT3a CRCs showed similar survival characteristics, pT3b cancers remained a distinct subgroup with dismal survival. DISCUSSION: The delineation of pT3a/pT3b subcategories of CRC based on the histological depth of adipose tissue invasion adds valuable prognostic information to the current pT3 classification and implementation into current staging practices of CRC should be considered.

Senescence-Associated Molecules and Tumor-Immune-Interactions as Prognostic Biomarkers in Colorectal Cancer.

Background and Aims: The initiation of cellular senescence in response to protumorigenic stimuli counteracts malignant progression in (pre)malignant cells. Besides arresting proliferation, cells entering this terminal differentiation state adopt a characteristic senescence-associated secretory phenotype (SASP) which initiates alterations to their microenvironment and effects immunosurveillance of tumorous lesions. However, some effects mediated by senescent cells contribute to disease progression. Currently, the exploration of senescent cells' impact on the tumor microenvironment and the evaluation of senescence as possible target in colorectal cancer (CRC) therapy demand reliable detection of cellular senescence in vivo. Therefore, specific immunohistochemical biomarkers are required. Our aim is to analyze the clinical implications of senescence detection in colorectal carcinoma and to investigate the interactions of senescent tumor cells and their immune microenvironment in vitro and in vivo. Methods: Senescence was induced in CRC cell lines by low-dose-etoposide treatment and confirmed by Senescence-associated ß-galactosidase (SA-ß-GAL) staining and fluorescence activated cell sorting (FACS) analysis. Co-cultures of senescent cells and immune cells were established. Multiple cell viability assays, electron microscopy and live cell imaging were conducted. Immunohistochemical (IHC) markers of senescence and immune cell subtypes were studied in a cohort of CRC patients by analyzing a tissue micro array (TMA) and performing digital image analysis. Results were compared to disease-specific survival (DSS) and progression-free survival (PFS). Results: Varying expression of senescence markers in tumor cells was associated with in- or decreased survival of CRC patients. Proximity analysis of p21-positive senescent tumor cells and cytotoxic T cells revealed a significantly better prognosis for patients in which these cell types have the possibility to directly interact. In vitro, NK-92 cells (mimicking natural killer T cells) or TALL-104 cells (mimicking both cytotoxic T cells and natural killer T cells) led to dose-dependent specific cytotoxicity in >75 % of the senescent CRC cells but <20 % of the proliferating control CRC cells. This immune cell-mediated senolysis seems to be facilitated via direct cell-cell contact inducing apoptosis and granule exocytosis. Conclusion: Counteracting tumorigenesis, cellular senescence is of significant relevance in CRC. We show the dual role of senescence bearing both beneficial and malignancy-promoting potential in vivo. Absence as well as exceeding expression of senescence markers are associated with bad prognosis in CRC. The antitumorigenic potential of senescence induction is determined by tumor micromilieu and immune cell-mediated elimination of senescent cells.

Swarm learning for decentralized artificial intelligence in cancer histopathology.

Artificial intelligence (AI) can predict the presence of molecular alterations directly from routine histopathology slides. However, training robust AI systems requires large datasets for which data collection faces practical, ethical and legal obstacles. These obstacles could be overcome with swarm learning (SL), in which partners jointly train AI models while avoiding data transfer and monopolistic data governance. Here, we demonstrate the successful use of SL in large, multicentric datasets of gigapixel histopathology images from over 5,000 patients. We show that AI models trained using SL can predict BRAF mutational status and microsatellite instability directly from hematoxylin and eosin (H&E)-stained pathology slides of colorectal cancer. We trained AI models on three patient cohorts from Northern Ireland, Germany and the United States, and validated the prediction performance in two independent datasets from the United Kingdom. Our data show that SL-trained AI models outperform most locally trained models, and perform on par with models that are trained on the merged datasets. In addition, we show that SL-based AI models are data efficient. In the future, SL can be used to train distributed AI models for any histopathology image analysis task, eliminating the need for data transfer.

The future of artificial intelligence in digital pathology - results of a survey across stakeholder groups.

AIMS: Artificial intelligence (AI) provides a powerful tool to extract information from digitised histopathology whole slide images. During the last 5 years, academic and commercial actors have developed new technical solutions for a diverse set of tasks, including tissue segmentation, cell detection, mutation prediction, prognostication and prediction of treatment response. In the light of limited overall resources, it is presently unclear for researchers, practitioners and policymakers which of these topics are stable enough for clinical use in the near future and which topics are still experimental, but worth investing time and effort into. METHODS AND RESULTS: To identify potentially promising applications of AI in pathology, we performed an anonymous online survey of 75 computational pathology domain experts from academia and industry. Participants enrolled in 2021 were queried about their subjective opinion on promising and appealing subfields of computational pathology with a focus upon solid tumours. The results of this survey indicate that the prediction of treatment response directly from routine pathology slides is regarded as the most promising future application. This item was ranked highest in the overall analysis and in subgroups by age and professional background. Furthermore, prediction of genetic alterations, gene expression and survival directly from routine pathology images scored consistently high throughout subgroups. CONCLUSIONS: Together, these data demonstrate a possible direction for the development of computational pathology systems in clinical, academic and industrial research in the near future.

Artificial intelligence to identify genetic alterations in conventional histopathology.

Precision oncology relies on the identification of targetable molecular alterations in tumor tissues. In many tumor types, a limited set of molecular tests is currently part of standard diagnostic workflows. However, universal testing for all targetable alterations, especially rare ones, is limited by the cost and availability of molecular assays. From 2017 to 2021, multiple studies have shown that artificial intelligence (AI) methods can predict the probability of specific genetic alterations directly from conventional hematoxylin and eosin (H&E) tissue slides. Although these methods are currently less accurate than gold standard testing (e.g. immunohistochemistry, polymerase chain reaction or next-generation sequencing), they could be used as pre-screening tools to reduce the workload of genetic analyses. In this systematic literature review, we summarize the state of the art in predicting molecular alterations from H&E using AI. We found that AI methods perform reasonably well across multiple tumor types, although few algorithms have been broadly validated. In addition, we found that genetic alterations in FGFR, IDH, PIK3CA, BRAF, TP53, and DNA repair pathways are predictable from H&E in multiple tumor types, while many other genetic alterations have rarely been investigated or were only poorly predictable. Finally, we discuss the next steps for the implementation of AI-based surrogate tests in diagnostic workflows. © 2022 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.