Histological tumor typing in the age of molecular profiling

Clinical applications of graph neural networks in computational histopathology: A review

Pathological examination is the optimal approach for diagnosing cancer, and with the advancement of digital imaging technologies, it has spurred the emergence of computational histopathology. The objective of computational histopathology is to assist in clinical tasks through image processing and analysis techniques. In the early stages, the technique involved analyzing histopathology images by extracting mathematical features, but the performance of these models was unsatisfactory. With the development of artificial intelligence (AI) technologies, traditional machine learning methods were applied in this field. Although the performance of the models improved, there were issues such as poor model generalization and tedious manual feature extraction. Subsequently, the introduction of deep learning techniques effectively addressed these problems. However, models based on traditional convolutional architectures could not adequately capture the contextual information and deep biological features in histopathology images. Due to the special structure of graphs, they are highly suitable for feature extraction in tissue histopathology images and have achieved promising performance in numerous studies. In this article, we review existing graph-based methods in computational histopathology and propose a novel and more comprehensive graph construction approach. Additionally, we categorize the methods and techniques in computational histopathology according to different learning paradigms. We summarize the common clinical applications of graph-based methods in computational histopathology. Furthermore, we discuss the core concepts in this field and highlight the current challenges and future research directions.

NXPH4 Used as a New Prognostic and Immunotherapeutic Marker for Muscle-Invasive Bladder Cancer

Background

One of the most common malignant tumors of the urinary system is muscle-invasive bladder cancer (MIBC). With the increased use of immunotherapy, its importance in the field of cancer is becoming abundantly evident. This study classifies MIBC according to GSVA score from the perspective of the GSEA immune gene set.

Methods

This study integrated the sequencing and clinical data of MIBC patients in TCGA and GEO databases, then scored the data using the GSVA algorithm, the CNMF algorithm was implemented to divide the subtypes of GEO and TCGA datasets, respectively, and finally screened and determined the key pathways in combination with clinical data. Simultaneously, LASSO Cox regression model was constructed based on key pathway genes to assess the model's predictive ability (ROC) and describe the immune landscape differences between high- and low-risk groups; key genes were further analyzed and verified in patient tissues.

Results

404 TCGA and 297 GEO datasets were divided into C1-3 groups (TCGA-C1:120/C2:152/C3:132; GEO- C1:112/C2:101/C3:84), of which TCGA-C2 (n = 152) subtype and GEO-C1 (n = 112) subtype had the worst prognosis. LASSO Cox regression model with ROC (train set = 0.718, test set = 0.667) could be constructed. When combined with the Cancer Immunome Atlas database, it was found that patients with high-risk scores were more sensitive to PD-1 inhibitor and PD-1 inhibitor combined with CTLA-4. NXPH4, as a key gene, plays a role in MIBC with tissue validation results show that nxph4 is highly expressed in tumor.

Conclusion

The immune gene score of MIBC data in TCGA and GEO databases was successfully evaluated using GSVA in this research. The lasso Cox expression model was successfully constructed by screening immune genes, the high-risk group had a worse prognosis and higher sensitivity to immunotherapy, PD-1 inhibitors or PD-1 combined with CTLA-4 inhibitors can be preferentially used in high-risk patients who are sensitive to immunotherapy, and NXPH4 may be a molecular target to adjust the effect of immunotherapy.

The Role of the Pathologist in the Next-Generation Era of Tumor Molecular Characterization

Current pathology practice is being shaped by the increasing complexity of modern medicine, in particular of precision oncology, and major technological advances. In the "next-generation technologies era", the pathologist has become the person responsible for the integration and interpretation of morphologic and molecular information and for the delivery of critical answers to diagnostic, prognostic and predictive queries, acquiring a prominent position in the molecular tumor boards.

Supervised non-negative matrix factorization methods for MALDI imaging applications

Motivation

Non-negative matrix factorization (NMF) is a common tool for obtaining low-rank approximations of non-negative data matrices and has been widely used in machine learning, e.g. for supporting feature extraction in high-dimensional classification tasks. In its classical form, NMF is an unsupervised method, i.e. the class labels of the training data are not used when computing the NMF. However, incorporating the classification labels into the NMF algorithms allows to specifically guide them toward the extraction of data patterns relevant for discriminating the respective classes. This approach is particularly suited for the analysis of mass spectrometry imaging (MSI) data in clinical applications, such as tumor typing and classification, which are among the most challenging tasks in pathology. Thus, we investigate algorithms for extracting tumor-specific spectral patterns from MSI data by NMF methods.

Results

In this article, we incorporate a priori class labels into the NMF cost functional by adding appropriate supervised penalty terms. Numerical experiments on a MALDI imaging dataset confirm that the novel supervised NMF methods lead to significantly better classification accuracy and stability as compared with other standard approaches.

Availability and implementaton

https://gitlab.informatik.uni-bremen.de/digipath/Supervised_NMF_Methods_for_MALDI.git

Supplementary information

Supplementary data are available at Bioinformatics online.

Genotyping of colorectal cancer for cancer precision medicine: Results from the IPH Center for Molecular Pathology

Cancer precision medicine has opened up new avenues for the treatment of colorectal cancer (CRC). To fully realize its potential, high-throughput sequencing platforms that allow genotyping beyond KRAS need to be implemented and require performance assessment. We comprehensively analyzed first-year data of 202 consecutive formalin-fixed paraffin embedded (FFPE) CRC samples for which prospective genotyping at our institution was requested. Deep targeted genotyping was done using a semiconductor-based sequencing platform and a self-designed panel of 30 CRC-related genes. Additionally, microsatellite status (MS) was determined. Ninety-seven percent of tumor samples were suitable for sequencing and in 88% MS could be assessed. The minimal drop-out rates of 6 and 25 cases, respectively were due to too low amounts or heavy degradation of DNA. Of 557 nonsynonymous mutations, 90 (16%) have not been described in COSMIC at the time of data query. Forty-three cases (22%) had double- or triple mutations affecting a single gene. Sixty-four percent had genetic alterations influencing oncological therapy. Eight percent of patients (MSI phenotype: 6%; mutated POLE: 2%) were potentially eligible for treatment with immune checkpoint inhibitors. Of 56% of KRASwt CRC that potentially qualified for anti-EGFR treatment, 30% presented with mutations in BRAF/NRAS. Mutated PIK3CA was detected in 21%. In conclusion, we here present real-life routine diagnostics data that not only demonstrate the robustness and feasibility of deep targeted sequencing and MS-analysis of FFPE CRC samples but also contribute to the understanding of CRC genetics. Most importantly, in more than half of the patients our approach enabled the selection of the best treatment currently available. © 2016 Wiley Periodicals, Inc.