Computer-assisted mitotic count using a deep learning–based algorithm improves interobserver reproducibility and accuracy

The mitotic count (MC) is an important histological parameter for prognostication of malignant neoplasms. However, it has inter- and intraobserver discrepancies due to difficulties in selecting the region of interest (MC-ROI) and in identifying or classifying mitotic figures (MFs). Recent progress in the field of artificial intelligence has allowed the development of high-performance algorithms that may improve standardization of the MC. As algorithmic predictions are not flawless, computer-assisted review by pathologists may ensure reliability. In the present study, we compared partial (MC-ROI preselection) and full (additional visualization of MF candidates and display of algorithmic confidence values) computer-assisted MC analysis to the routine (unaided) MC analysis by 23 pathologists for whole-slide images of 50 canine cutaneous mast cell tumors (ccMCTs). Algorithmic predictions aimed to assist pathologists in detecting mitotic hotspot locations, reducing omission of MFs, and improving classification against imposters. The interobserver consistency for the MC significantly increased with computer assistance (interobserver correlation coefficient, ICC = 0.92) compared to the unaided approach (ICC = 0.70). Classification into prognostic stratifications had a higher accuracy with computer assistance. The algorithmically preselected hotspot MC-ROIs had a consistently higher MCs than the manually selected MC-ROIs. Compared to a ground truth (developed with immunohistochemistry for phosphohistone H3), pathologist performance in detecting individual MF was augmented when using computer assistance (F1-score of 0.68 increased to 0.79) with a reduction in false negatives by 38%. The results of this study demonstrate that computer assistance may lead to more reproducible and accurate MCs in ccMCTs.

Proposed Diagnostic Criteria and Classification of Canine Mast Cell Neoplasms: A Consensus Proposal

Mast cell neoplasms are one of the most frequently diagnosed malignancies in dogs. The clinical picture, course, and prognosis vary substantially among patients, depending on the anatomic site, grade and stage of the disease. The most frequently involved organ is the skin, followed by hematopoietic organs (lymph nodes, spleen, liver, and bone marrow) and mucosal sites of the oral cavity and the gastrointestinal tract. In cutaneous mast cell tumors, several grading and staging systems have been introduced. However, no comprehensive classification and no widely accepted diagnostic criteria have been proposed to date. To address these open issues and points we organized a Working Conference on canine mast cell neoplasms in Vienna in 2019. The outcomes of this meeting are summarized in this article. The proposed classification includes cutaneous mast cell tumors and their sub-variants defined by grading- and staging results, mucosal mast cell tumors, extracutaneous/extramucosal mast cell tumors without skin involvement, and mast cell leukemia (MCL). For each of these entities, diagnostic criteria are proposed. Moreover, we have refined grading and staging criteria for mast cell neoplasms in dogs based on consensus discussion. The criteria and classification proposed in this article should greatly facilitate diagnostic evaluation and prognostication in dogs with mast cell neoplasms and should thereby support management of these patients in daily practice and the conduct of clinical trials.

Canine mast cell tumors: utility of stereologic tools in cytology

Quantitative morphologic parameters assessed in cytologic samples of canine cutaneous mast cell tumors (ccMCTs) may assist with surgical planning and prognostication. Robust cutoffs can be defined, with high reproducibility, for parameters such as the nuclear area (NA). The NA may be determined by morphometry (image analysis, NAI) or by stereology, such as the 2D-nucleator method (NAN); stereologic techniques have not been applied to cytologic specimens of ccMCT, to our knowledge. We retrospectively selected routine cytology smears from 51 ccMCT cases and screened them to determine the percentage of neoplastic mast cells with indistinct nuclear borders; this was repeated after the slides were restained with H&E. The NAI and the NAN were estimated in 100 mast cells per animal in H&E-stained slides. All nuclei were visible in H&E smears, and unbiased quantification was feasible. The NAN was similar to NAI, but less time-consuming. Both the NAN and NAI determined by cytology differed in histologic low- and high-grade ccMCTs, and in histologic grade I plus II versus grade III ccMCTs. Stereologic parameters such as the NAN could be considered as complementary techniques for the cytologic evaluation of ccMCTs.

Immunohistochemical and molecular profiling of CD 117, Oct-4, and Sox-2 in canine cutaneous mast cell tumor of the crossbred dogs in Faculty of Veterinary Science, Chulalongkorn University, Bangkok, Thailand

Background and Aim: CD 117 (c-KIT) internal tandem duplication (ITD), octamer-binding transcription factor 4 (Oct-4), and sex-determining region Y-box 2 (Sox-2) may govern the oncogenicity and aggressiveness of canine cutaneous mast cell tumor (MCT) in the crossbred dogs. Thus, a comprehension of this matter may help us establishing a novel platform to treat the disease in those dogs. However, evidence has lacked so far. Thus, this study aimed to survey CD 117 ITD, Oct-4, and Sox-2 expressions and their relations to the 2-tier grading in a group of Thai crossbreed dogs. The study was done using polymerase chain reaction (PCR), Reverse transcription PCR (RT-PCR), and immunohistochemistry. Materials and Methods: Thirty-three MCT specimens graded by the 2-tier histopathology grading were collected from the crossbred and purebred dogs. CD 117 ITD was detected by conventional PCR and immunohistochemistry. While, Oct-4 and Sox-2 expression levels were determined at the protein and mRNA levels by immunohistochemistry and RT-PCR, respectively. The expression magnitude of each parameter was then related to the grades and breeds. Results: About 60.61% of specimens were low grade, while 39.39% were high grade. CD 117 ITD was not detected in all specimens. A significant increase of Oct-4 expression was found in the high-grade, crossbred dogs. Meanwhile, Sox-2 expressions were increased both in the purebred and crossbred dogs with high-grade MCT. Conclusion: The study finding has indicated that the level of Sox-2 expression may be a useful tumorigenic and prognostic biomarker because it correlates to the 2-tier grades but not dog breeds.

Cutaneous mast cell tumor in a captive Bush dog (Speothos venaticus): pathological and immunophenotypical aspects - case report

ABSTRACT A senile male captive bush dog (Speothos venaticus) presented a small perianal cutaneous nodule. Histologically, there was an ulcerated round cell tumor composed of well differentiated mast cells with abundant intracytoplasmic purple Giemsa-positive granules, with a diffuse eosinophilic infiltrate. Immunohistochemistry revealed that 30% of the neoplastic cells were positive for Kit in the cytoplasm and cell membrane, and all neoplastic cells were negative for MAC and CD3. Less than 10% of the neoplastic cells were positive for Ki67. At necropsy other primary tumors were identified in this animal, including an intestinal adenoma, an adrenal cortex adenoma and a testicular interstitial cell tumor.

Galectin-3 immunolabelling correlates with BCL2 expression in canine cutaneous mast cell tumours

Mast cell tumour (MCT) is the most frequent skin neoplasm in dogs. These tumours are characterised by variable behaviour and clinical presentation that make prognosis an important and challenging task in the veterinary practice. Galectin-3 (Gal-3) is known to influence several biological processes that are important in the cancer context and has been described as a prognostic marker for several human cancers. The aim of the present work was to characterise Gal-3 immunolabelling in canine cutaneous MCTs and to investigate its value as a prognostic marker for the disease. Thirty-four random cases of canine cutaneous MCT that were surgically treated with wide margins were included in this study. Gal-3 expression was evaluated using immunohistochemistry and the results were compared with the expression of apoptosis-related proteins, Ki67 index, histopathological grades, mortality due to the disease and post-surgical survival. The majority of the MCTs (65.8%) were positive for Gal-3. Gal-3 immunolabelling was variable among the samples (2.7%–86.8% of the neoplastic cells). The protein was located in the cytoplasm or in the cytoplasm and the nucleus. Gal-3 positivity was correlated with BCL2 expression (P < 0.001; r = 0.604), but not with Ki67 and BAX. No significant differences were detected between histological grades or in the survival analysis. Gal-3 expression correlates with BCL2 expression in MCTs. Although an efficient marker for several human neoplasms, the results presented herein suggest that Gal-3 immunolabelling is not an independent prognostic indicator for this disease.

Computer-Assisted Mitotic Count Using a Deep Learning-based Algorithm Improves Inter-Observer Reproducibility and Accuracy in canine cutaneous mast cell tumors

The mitotic count (MC) is an important histological parameter for prognostication of malignant neoplasms. However, it has inter- and intra-observer discrepancies due to difficulties in selecting the region of interest (MC-ROI) and in identifying/classifying mitotic figures (MFs). Recent progress in the field of artificial intelligence has allowed the development of high-performance algorithms that may improve standardization of the MC. As algorithmic predictions are not flawless, the computer-assisted review by pathologists may ensure reliability. In the present study we have compared partial (MC-ROI preselection) and full (additional visualization of MF candidate proposal and display of algorithmic confidence values) computer-assisted MC analysis to the routine (unaided) MC analysis by 23 pathologists for whole slide images of 50 canine cutaneous mast cell tumors (ccMCTs). Algorithmic predictions aimed to assist pathologists in detecting mitotic hotspot locations, reducing omission of MF and improving classification against imposters. The inter-observer consistency for the MC significantly increased with computer assistance (interobserver correlation coefficient, ICC = 0.92) compared to the unaided approach (ICC = 0.70). Classification into prognostic stratifications had a higher accuracy with computer assistance. The algorithmically preselected MC-ROIs had a consistently higher MCs than the manually selected MC-ROIs. Compared to a ground truth (developed with immunohistochemistry for phosphohistone H3), pathologist performance in detecting individual MF was augmented when using computer assistance (F1-score of 0.68 increased to 0.79) with a reduction in false negatives by 38%. The results of this study prove that computer assistance may lead to a more reproducible and accurate MCs in ccMCTs.

Value, Limitations, and Recommendations for Grading of Canine Cutaneous Mast Cell Tumors: A Consensus of the Oncology-Pathology Working Group

One of the primary objectives of the Oncology Pathology Working Group (OPWG), a joint initiative of the Veterinary Cancer Society and the American College of Veterinary Pathologists, is for oncologists and pathologists to collaboratively generate consensus documents to standardize aspects and provide guidelines for oncologic pathology. Consensus is established through review of relevant peer-reviewed literature relative to a subgroup’s particular focus. In this article, the authors provide a critical review of the current literature for grading of canine cutaneous mast cell tumors, suggest guidelines for reporting, and provide recommendations for its clinical interpretation. The article mainly focuses on histologic grading, but relevant information on mitotic count and cytological grading are also discussed. This document represents the opinions of the working group and the authors but does not constitute a formal endorsement by the American College of Veterinary Pathologists or the Veterinary Cancer Society.