Search

Simple Summary We presented a machine learning approach for accurate quantification of nuclear morphometrics and differential diagnosis of primary intestinal T-cell lymphomas. The human interpretable machine learning approach can be easily applied to other lymphomas and potentially even broader disease categories. This approach not only brings deeper insights into lymphoma phenotypes, but also paves the way for future discoveries concerning their relationship with disease classification and outcome. Abstract The aim of this study was to investigate the feasibility of using machine learning techniques based on morphological features in classifying two subtypes of primary intestinal T-cell lymphomas (PITLs) defined according to the WHO criteria: monomorphic epitheliotropic intestinal T-cell lymphoma (MEITL) versus intestinal T-cell lymphoma, not otherwise specified (ITCL-NOS), which is considered a major challenge for pathological diagnosis. A total of 40 histopathological whole-slide images (WSIs) from 40 surgically resected PITL cases were used as the dataset for model training and testing. A deep neural network was trained to detect and segment the nuclei of lymphocytes. Quantitative nuclear morphometrics were further computed from these predicted contours. A decision-tree-based machine learning algorithm, XGBoost, was then trained to classify PITL cases into two disease subtypes using these nuclear morphometric features. The deep neural network achieved an average precision of 0.881 in the cell segmentation work. In terms of classifying MEITL versus ITCL-NOS, the XGBoost model achieved an area under receiver operating characteristic curve (AUC) of 0.966. Our research demonstrated an accurate, human-interpretable approach to using machine learning algorithms for reducing the high dimensionality of image features and classifying T cell lymphomas that present challenges in morphologic diagnosis. The quantitative nuclear morphometric features may lead to further discoveries concerning the relationship between cellular phenotype and disease status.

In fine-needle aspirations (FNA) of thyroid, Hürthle cells can be found in a broad spectrum of lesions, ranging from non-neoplastic conditions to aggressive malignant tumors. Recognize them morphologically, frequently represents a challenging for an adequately diagnosis and are associated with a significant interobserver variability. Although the limitations of the morphologic diagnosis still exist, the interpretation of the context where the cells appear and the recent advances in the molecular knowledge of Hürthle cells tumors are contributing for a more precise diagnosis. This review aims to describe the cytology aspects of all Hürthle cells neoplastic and non-neoplastic thyroid lesions, focusing on the differential diagnosis and reporting according to The Bethesda System for Reporting Thyroid Cytology (TBSRTC). New entities according to the latest World Health Organization (WHO) classification are included, as well as an update of the current molecular data.

Acute myeloid leukemia (AML) is defined by the presence of ≥ 20% myeloblasts in the blood or bone marrow. Spontaneous remission (SR) of AML is a rare event, with few cases described in the literature. SR is generally associated with recovery from an infectious or immunologic process, and more recently possibly with clonal hematopoiesis. We review the literature and assess the trends associated with SR, and report a new case of a 58-year-old man with a morphologic diagnosis of AML associated with a severe gastrointestinal (GI) tract infection. The patient had an NF1 variant that was previously unreported in AML as the only clonal abnormality. After treatment of the infection, the increased blast population subsided with no leukemia-directed therapy, and the patient has remained in a continuous, spontaneous complete remission for > 2 years.

For decades, histopathology with routine hematoxylin and eosin staining has been and remains the gold standard for reaching a morphologic diagnosis in tissue samples from humans and veterinary species. However, within the past decade, there has been exponential growth in advanced techniques for in situ tissue biomarker imaging that bridge the divide between anatomic and molecular pathology. It is now possible to simultaneously observe localization and expression magnitude of multiple protein, nucleic acid, and molecular targets in tissue sections and apply machine learning to synthesize vast, image-derived datasets. As these technologies become more sophisticated and widely available, a team-science approach involving subspecialists with medical, engineering, and physics backgrounds is critical to upholding quality and validity in studies generating these data. The purpose of this manuscript is to detail the scientific premise, tools and training, quality control, and data collection and analysis considerations needed for the most prominent advanced imaging technologies currently applied in tissue sections: immunofluorescence, in situ hybridization, laser capture microdissection, matrix-assisted laser desorption ionization imaging mass spectrometry, and spectroscopic/optical methods. We conclude with a brief overview of future directions for ex vivo and in vivo imaging techniques.

Soft tissue sarcomas represent a heterogeneous group of rare malignancies exhibiting mesenchymal differentiation with an overall incidence of around 5/100,000/year. Rarity and morphologic heterogeneity significantly affect diagnostic accuracy; therefore, expertise can be achieved only through access to large number of cases. Soft tissue sarcomas are currently classified on the basis of the 2013 WHO classification of soft tissue tumors that integrate conventional morphology with immunohistochemistry and molecular genetics. The morphologic diagnosis of sarcoma relies on the evaluation as well as the integration of four main features: the shape of the neoplastic cells; the pattern of growth; the quality of the background; the architecture of the vascular network. Immunohistochemical characterization plays a key role in the diagnostic workup of soft tissue sarcomas. The majority of classic differentiation markers tend to show good sensitivity, however, associated with rather limited specificity, making interpretation in context with morphology mandatory. Molecular genetics is increasingly used for diagnostic purposes to distinguish specific subtypes of sarcomas, to support diagnosis in non-canonical clinical presentations and also to distinguish true sarcomas from benign mimickers. With many exceptions, histologic typing does not provide sufficient information for predicting the clinical course of the disease and, therefore, grading systems based on histological parameters were introduced to provide a more accurate estimation of the degree of malignancy of tumors. The three-tiered system devised by the French Federation of Cancer Centers Sarcoma Group (FNCLCC) systems is widely adopted; however, several limitations exist that have led to the development of prognostic nomograms that incorporate the specific histotype as one of the relevant parameters.

High-grade neuroendocrine neoplasms (World Health Organization [WHO] G3) of the pancreas include both well-differentiated neuroendocrine tumor (WD-NET) and poorly differentiated neuroendocrine carcinoma (PD-NEC). According to the WHO classification scheme, the diagnosis of this group of tumors is based on both the histopathology of the tumor and the assessment of proliferation fraction. However, the former can be challenging due to the lack of well-defined histologic criteria, and the latter alone (ie, >20 mitoses/10 high-power fields or Ki67>20%) may not sufficiently distinguish WD-NETs from PD-NECs. Given the considerable differences in treatment strategies and clinical outcome, additional practical modalities are required to facilitate the accurate diagnosis of high-grade pancreatic neuroendocrine neoplasms. We examined 33 cases of WHO G3 neuroendocrine neoplasms of the pancreas and attempted to classify them into WD-NET, small cell PD-NEC (PD-NEC-SCC), and large cell PD-NEC (PD-NEC-LCC) or to designate them as "ambiguous" when an uncertain diagnosis was rendered by any of the observers or there was any disagreement in classification among the 3 observers. To simplify the interpretation, both PD-NEC-SCC and PD-NEC-LCC were considered together as PD-NECs in the final analysis. The initial approach was to assess microscopically a single morphologically challenging hematoxylin and eosin section from each case without the knowledge of Ki67 values, performed independently by 3 pathologists to assess the degree of diagnostic concordance, and then evaluate immunohistochemical staining for surrogate biomarkers of known genotypes of WD-NET and PD-NEC, respectively, and, lastly, complete a clinicopathologic review to establish a final definitive classification. Loss of DAXX or ATRX protein expression defined WD-NET, and abnormal p53, Rb, SMAD4 expression signified PD-NEC. When the chosen section displayed an element of WD histopathology, or other tumor sections contained WHO G1/G2 components, or there had been a prior established diagnosis of a primary WD-NET, the final diagnosis was rendered as a WD-NET with high-grade (G3) progression. If a component of conventional adenocarcinoma was present (in slides not seen in the initial review), the diagnosis was established as a combined adenocarcinoma and PD-NEC. All 3 pathologists agreed on the morphologic classification of 33% of the cases (6 WD-NET, 3 PD-NEC-SCC, and 2 PD-NEC-LCC), were conflicted on 2 cases between PD-NEC-SCC and PD-NEC-LCC, and disagreed or were uncertain on the classification for the remaining 20 cases (61%), which were therefore categorized as ambiguous. In the group of cases in which all pathologists agreed on the classification, the 6 WD-NET cases had either loss of DAXX or ATRX or had evidence of a WD-NET based on additional or prior pathology slides. The 7 PD-NEC cases had abnormal expression of p53, Rb, and/or SMAD4 or a coexisting adenocarcinoma. In the ambiguous group (n=20), 14 cases were established as WD-NETs, based upon loss of DAXX or ATRX in 7 cases and additional pathology evidence of high-grade progression from WD-NET in the other 7 cases; 5 cases were established as PD-NEC based upon abnormal expression of p53, Rb, and/or SMAD4; 1 case remained undetermined with normal expression of all markers and no evidence of entity-defining histologic findings in other slides. On the basis of the final pathologic classifications, the disease-specific survival was 75 and 11 months for the WD-NET and PD-NEC groups, respectively. Thus, we conclude that morphologic diagnosis of high-grade pancreatic neuroendocrine neoplasms is challenging, especially when limited pathologic materials are available, and necessitates better defined criteria. The analysis of both additional sections and prior material, along with an immunohistochemical evaluation, can facilitate accurate diagnosis in the majority of cases and guide the appropriate clinical management and prognosis.