Routine Pathology Research Articles

Spatial transcriptomics and the anatomical pathologist: Molecular meets morphology.

In recent years anatomical pathology has been revolutionised by the incorporation of molecular findings into routine diagnostic practice, and in some diseases the presence of specific molecular alterations are nowessential for diagnosis. Spatial transcriptomics describes a group of technologies that provide up to transcriptome-wide expression profiling while preserving the spatial origin of the data, with many of these technologies able to provide these data using a single tissue section. Spatial transcriptomics allows expression profiling of highly specific areas within a tissue section potentially to subcellular resolution, and allows correlation of expression data with morphology, tissue type and location relative to other structures. While largely still research laboratory-based, several spatial transcriptomics methods have now achieved compatibility with formalin-fixed paraffin-embedded tissue (FFPE), allowing their use in diagnostic tissue samples, and with further development potentially leading to their incorporation in routine anatomical pathology practice. This mini review provides an overview of spatial transcriptomics methods, with an emphasis on platforms compatible with FFPE tissue, approaches to assess the data and potential applications in anatomical pathology practice.

Digital image analysis and artificial intelligence in pathology diagnostics-the Swiss view.

Digital pathology (DP) is increasingly entering routine clinical pathology diagnostics. As digitization of the routine caseload advances, implementation of digital image analysis algorithms and artificial intelligence tools becomes not only attainable, but also desirable in daily sign out. The Swiss Digital Pathology Consortium (SDiPath) has initiated aDelphi process to generate best-practice recommendations for various phases of the process of digitization in pathology for the local Swiss environment, encompassing the following four topics: i)scanners, quality assurance, and validation of scans; ii)integration of scanners and systems into the pathology laboratory information system; iii)the digital workflow; and iv)digital image analysis (DIA)/artificial intelligence (AI). The current article focuses on the DIA-/AI-related recommendations generated and agreed upon by the working group and further verified by the Delphi process among the members of SDiPath. Importantly, they include the view and the currently perceived needs of practicing pathologists from multiple academic and cantonal hospitals as well as private practices.

Morphological subtypes of colorectal low-grade intraepithelial neoplasia: diagnostic reproducibility, frequency and clinical impact

AimsSpecial histomorphological subtypes of colorectal low-grade intraepithelial neoplasia (LGIN) with variable prognostic impact were recently described in patients with inflammatory bowel disease (IBD) referred to as non-conventional dysplasia. However, they...

Artifact Augmentation for Enhanced Tissue Detection in Microscope Scanner Systems.

As the field of routine pathology transitions into the digital realm, there is a surging demand for the full automation of microscope scanners, aiming to expedite the process of digitizing tissue samples, and consequently, enhancing the efficiency of case diagnoses. The key to achieving seamless automatic imaging lies in the precise detection and segmentation of tissue sample regions on the glass slides. State-of-the-art approaches for this task lean heavily on deep learning techniques, particularly U-Net convolutional neural networks. However, since samples can be highly diverse and prepared in various ways, it is almost impossible to be fully prepared for and cover every scenario with training data. We propose a data augmentation step that allows artificially modifying the training data by extending some artifact features of the available data to the rest of the dataset. This procedure can be used to generate images that can be considered synthetic. These artifacts could include felt pen markings, speckles of dirt, residual bubbles in covering glue, or stains. The proposed approach achieved a 1-6% improvement for these samples according to the F1 Score metric.

Immunoglobulin light chain transcript detection by ultrasensitive RNA in situ hybridization for B-cell lymphoma diagnosis.

Evaluation of B-cell clonality can be challenging in the interpretation of lymphoid infiltrates on tissue sections. Clonality testing based on IG gene rearrangements analysis by PCR (IG-PCR) is the gold standard. Alternatively, B-cell clonality can be assessed by the recognition of immunoglobulin light chain (IgLC) restriction, by immunohistochemistry (IHC), chromogenic in situ hybridization (ISH) or flow cytometry (FC). IG-PCR requires molecular facilities, and FC requires cell suspensions, both not widely available in routine pathology units. This study evaluates the performance of B-cell clonality detection by IgLC-RNAscope® (RNAsc) in a group of 216 formalin-fixed, paraffin-embedded samples including 185 non-Hodgkin B-cell lymphomas, 11 Hodgkin lymphomas (HL) and 20 reactive samples. IgLC-RNAsc, performed in parallel with FC in 53 cases, demonstrated better performances (93% vs 83%), particularly in diffuse large B-cell lymphoma (98% vs 71%) and follicular lymphoma (93% vs 83%) diagnosis. IgLC-RNAsc was also superior to IHC and ISH especially in samples with limited tumor cell content, where IG-PCR was not informative. Performed for the first time on mediastinal lymphomas, IgLC-RNAsc identified monotypic IgLC transcripts in 69% of primary mediastinal large B-cell lymphoma (PMBCL) and 67% of mediastinal gray zone lymphomas (MGZL). IGK/L double-negative cells were detected in 1 PMBCL, 2 MGZL, and all classical HL, while monotypic IgLC expression appeared to be a hallmark in nodular lymphocyte-predominant HL. IgLC-RNAsc demonstrates to be a powerful tool in B-cell lymphoma diagnosis, above all in challenging cases with limited tumor cell content, ensuring in situ investigations on mechanisms of Ig regulation across lymphoma entities.

Evaluating Different Methods for Determining Erythrocyte Sedimentation Rate in Free-Living Blanding's Turtles (Emydoidea blandingii).

Characterizing host factors affecting individual health can be important in the conservation of many chelonians, including the Blanding's turtle (Emydoidea blandingii). However, many routine clinical pathology tests do not reliably detect differences in health status in reptiles. The erythrocyte sedimentation rate (ESR) has been shown to characterize inflammation in many species, including tortoises, and may be useful in assessing Blanding's turtles. The purpose of our study was to evaluate two methods of ESR measurement in free-ranging Blanding's turtles. Sixty Blanding's turtles captured at two sites in Illinois received a complete physical examination, and blood was collected for determination of packed cell volume, total solids, and ESR with both microhematocrit tubes and commercial kits (Winpette). Method agreement was assessed with Passing-Bablok regression. Associations between ESR and demographic, environmental, and health factors were determined by linear regression. The microhematocrit tube method proportionately overestimated ESR compared with the Winpette. With the use of both measurement methods, ESR was significantly different between study sites, higher in adult Blanding's turtles, and higher in females that were gravid compared with those that were not. ESR was also negatively associated with packed cell volume. ESR values in Blanding's turtles are comparable to other reptiles, and although microhematocrit tube and Winpette results are not interchangeable, their associations with biological predictors are similar, and either method has the potential to be used with separate reference intervals to assess turtles for inflammatory conditions.

Carbon footprint evaluation of routine anatomic pathology practices using eco-audit: Current status and mitigation strategies

Because of global warming, integrating an ecological dimension to the practice of pathology is crucial, as already performed in other medical specialties. The aim of the study was to estimate the greenhouse gas emissions (GHE) of one biopsy/surgical specimen hematoxylin-phloxine-saffron (HPS) slide, one frozen section examination, and one immunohistochemistry (IHC) slide, and to propose environmentally friendly recommendations.The different steps of the pathological procedures, from sample transport to slide/block destruction, were considered for the GHE evaluation. Consumables and reagents used at each step were noted, as well as the electricity consumption of the premises/equipment, specific wastes, and staff travel. The GHE analyses were performed according to the eco-audit method.The GHE attributable to one HPS slide of a biopsy, one HPS slide of a surgical specimen, one frozen section examination, and one IHC slide were 0.589 kgCO2e (CO2 equivalent), 0.618 kgCO2e, 1.481 kgCO2e, and 0.363 kgCO2e, respectively. For a one-year activity in the pathology department, the GHE for HPS was estimated at 91,897 kgCO2e, equivalent to 422,321 km of thermal car ride. The main GHE sources were staff travel for one biopsy HPS slide (34.2 %) and one surgical specimen HPS slide (32.6 %), materials (39.1 %) and staff travel (27.2 %) for one frozen section examination, and staff travel (28.2 %) and reagents and their packaging (44.9 %) for one IHC slide.The present study highlighted the most significant sources of GHE for routine practices in a pathology department, underlining that, from pathologists to policy makers, each actor has a responsibility to implement sustainable changes.

Tailoring pretext tasks to improve self-supervised learning in histopathologic subtype classification of lung adenocarcinomas

Lung adenocarcinoma (LUAD) is a morphologically heterogeneous disease with five predominant histologic subtypes. Fully supervised convolutional neural networks can improve the accuracy and reduce the subjectivity of LUAD histologic subtyping using hematoxylin and eosin (H&E)-stained whole slide images (WSIs). However, developing supervised models with good prediction accuracy usually requires extensive manual data annotation, which is time-consuming and labor-intensive. This work proposes three self-supervised learning (SSL) pretext tasks to reduce labeling effort. These tasks not only leverage the multi-resolution nature of the H&E WSIs but also explicitly consider the relevance to the downstream task of classifying the LUAD histologic subtypes. Two tasks involve predicting the spatial relationship between tiles cropped from lower and higher magnification WSIs. We hypothesize that these tasks induce the model to learn to distinguish different tissue structures presented in the images, thus benefiting the downstream classification. The third task involves predicting the eosin stain from the hematoxylin stain, inducing the model to learn cytoplasmic features relevant to LUAD subtypes. The effectiveness of the three proposed SSL tasks and their ensemble was demonstrated by comparison with other state-of-the-art pretraining and SSL methods using three publicly available datasets. Our work can be extended to any other cancer type where tissue architectural information is important. The model could be used to expedite and complement the process of routine pathology diagnosis tasks. The code is available at https://github.com/rina-ding/ssl_luad_classification.

Prognostic role of tumor budding in oral squamous cell carcinoma: a retrospective study

Background: Tumor budding (TB) is defined as a single cell or a cluster of up to five tumor cells at the invasion front of all oral squamous cell carcinoma (OSCC) cases. Tumor budding is considered a marker of many important events in oral carcinoma, including the epithelial-mesenchymal transition, invasion, metastasis, and for further prognosis. Methods: A total of 30 clinically and histopathologically diagnosed cases of OSCC were retrieved from the archival of the Department of Oral Pathology and Microbiology. After assessing the sections for the presence of TB, it is further categorized into high-intensity TB and low-intensity TB. Results: We found a strong association between lymph node metastasis, lymph node metastasis, and histopathological grading in the current study. There was no association between TB and a patient's survival rate. Conclusions: This study emphasizes the significance of tumor budding, its relevance to prognosis, and reproducibility, as well as the significance of its evaluation and incorporation into routine surgical pathology practice in the management of OSCC. As a result, we believe that tumor budding is a crucial factor in determining tumor behavior.

Artificial Intelligence–Based Tool for Tumor Detection and Quantitative Tissue Analysis in Colorectal Specimens

Digital pathology adoption allows for applying computational algorithms to routine pathology tasks. Our study aimed to develop a clinical-grade artificial intelligence (AI) tool for precise multiclass tissue segmentation in colorectal specimens (resections and biopsies) and clinically validate the tool for tumor detection in biopsy specimens. The training data set included 241 precisely manually annotated whole-slide images (WSIs) from multiple institutions. The algorithm was trained for semantic segmentation of 11 tissue classes with an additional module for biopsy WSI classification. Six case cohorts from 5 pathology departments (4 countries) were used for formal and clinical validation, digitized by 4 different scanning systems. The developed algorithm showed high precision of segmentation of different tissue classes in colorectal specimens with composite multiclass Dice score of up to 0.895 and pixel-wise tumor detection specificity and sensitivity of up to 0.958 and 0.987, respectively. In the clinical validation study on multiple external cohorts, the AI tool reached sensitivity of 1.0 and specificity of up to 0.969 for tumor detection in biopsy WSI. The AI tool analyzes most biopsy cases in less than 1 minute, allowing effective integration into clinical routine. We developed and extensively validated a highly accurate, clinical-grade tool for assistive diagnostic processing of colorectal specimens. This tool allows for quantitative deciphering of colorectal cancer tissue for development of prognostic and predictive biomarkers and personalization of oncologic care. This study is a foundation for a SemiCOL computational challenge. We open-source multiple manually annotated and weakly labeled test data sets, representing a significant contribution to the colorectal cancer computational pathology field.

Alternative promoters and splicing create multiple functionally distinct isoforms of oestrogen receptor alpha in breast cancer and healthy tissues.

Oestrogen receptor alpha (ER) is involved in cell growth and proliferation and functions as a transcription factor, a transcriptional coregulator, and in cytoplasmic signalling. It affects, for example, bone, endometrium, ovaries and mammary epithelium. It is a key biomarker in clinical management of breast cancer, where it is used as a prognostic and treatment-predictive factor, and a therapeutical target. Several ER isoforms have been described, but transcript annotation in public databases is incomplete and inconsistent, and functional differences are not well understood. We have analysed short- and long-read RNA sequencing data from breast tumours, breast cancer cell lines, and normal tissues to create a comprehensive annotation of ER transcripts and combined it with experimental studies of full-length protein and six alternative isoforms. The isoforms have varying transcription factor activity, subcellular localisation, and response to the ER-targeting drugs tamoxifen and fulvestrant. Antibodies differ in ability to detect alternative isoforms, which raises concerns for the interpretation of ER-status in routine pathology. Future work should investigate the effects of alternative isoforms on patient survival and therapy response. An accurate annotation of ER isoforms will aid in interpretation of clinical data and inform functional studies to improve our understanding of the ER in health and disease.

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.

Anaplastic large cell lymphomas with equivocal DUSP22 FISH results: recommendations for clinical reporting and diagnostic evaluation

Anaplastic large cell lymphoma (ALCL), one of the most common T-cell lymphomas, shows unifying pathological features but is clinically and genetically heterogeneous. One genetic subgroup, characterized by recurrent DUSP22 rearrangements (R), has distinct morphologic, immunophenotypic, and molecular features and can be identified in routine pathology practice using a breakapart (BAP) fluorescence in situ hybridization (FISH) probe. However, some cases show equivocal BAP-FISH findings (BAP-FISHEQ) and the features of these cases are poorly understood. Here, we sought to characterize DUSP22 BAP-FISHEQ ALCLs further. First, we applied an immunohistochemistry (IHC) algorithm using TIA1, pSTAT3Y705, and LEF1, which can predict DUSP22-R with high accuracy. Among 37 BAP-FISHEQ ALCLs, 18 (49%) were IHC-algorithm positive (IHCPOS), 8 (21%) were IHC-algorithm negative (IHCNEG), and 11 (30%) were IHCEQ. In 32 BAP-FISHEQ cases, we also applied a dual-color, dual-fusion (D-FISH) probe for t(6;7)(p25.3;q32.3), which accounts for 45% of DUSP22-R ALCLs. Among BAP-FISHEQ cases, D-FISH was positive in 10/18 IHCPOS cases (56%), 0/9 IHCEQ cases (0%), and 0/5 IHCNEG cases (0%). Median survival in BAP-FISHEQ ALCLs was 105 months, intermediate between BAP-FISHPOS ALCLs (median survival not reached) and BAP-FISHNEG ALCLs (19 months). Thus, DUSP22 BAP-FISHEQ ALCLs are clinicopathologically heterogeneous, likely due to an admixture of cases with an unbalanced DUSP22-R and cases with focal deletions without rearrangement. For clinical reporting, we recommend that DUSP22 BAP-FISHEQ ALCLs be reported as equivocal, and not be grouped with BAP-FISHPOS ALCLs. Clinical adoption of an IHC algorithm, possibly supplemented by t(6; 7) D-FISH, could facilitate genetic subtyping in about two-thirds of BAP-FISHEQ ALCLs.

A 3-dimensional histology computer model of malignant melanoma and its implications for digital pathology

BackgroundHistorically, cancer diagnoses have been made by pathologists using two-dimensional histological slides. However, with the advent of digital pathology and artificial intelligence, slides are being digitised, providing new opportunities to integrate their information. Since nature is 3-dimensional (3D), it seems intuitive to digitally reassemble the 3D structure for diagnosis. ObjectiveTo develop the first human-3D-melanoma-histology-model with full data and code availability. Further, to evaluate the 3D-simulation together with experienced pathologists in the field and discuss the implications of digital 3D-models for the future of digital pathology. MethodsA malignant melanoma of the skin was digitised via 3 µm cuts by a slide scanner; an open-source software was then leveraged to construct the 3D model. A total of nine pathologists from four different countries with at least 10 years of experience in the histologic diagnosis of melanoma tested the model and discussed their experiences as well as implications for future pathology. ResultsWe successfully constructed and tested the first 3D-model of human melanoma. Based on testing, 88.9% of pathologists believe that the technology is likely to enter routine pathology within the next 10 years; advantages include a better reflectance of anatomy, 3D assessment of symmetry and the opportunity to simultaneously evaluate different tissue levels at the same time; limitations include the high consumption of tissue and a yet inferior resolution due to computational limitations. Conclusions3D-histology-models are promising for digital pathology of cancer and melanoma specifically, however, there are yet limitations which need to be carefully addressed.

Deep Learning–Enabled Diagnosis of Liver Adenocarcinoma

Diagnosis of adenocarcinoma in the liver is a frequent scenario in routine pathology and has a critical impact on clinical decision making. However, rendering a correct diagnosis can be challenging, and often requires the integration of clinical, radiologic, and immunohistochemical information. We present a deep learning model (HEPNET) to distinguish intrahepatic cholangiocarcinoma from colorectal liver metastasis, as the most frequent primary and secondary forms of liver adenocarcinoma, with clinical grade accuracy using H&E-stained whole-slide images. HEPNET was trained on 714,589 image tiles from 456 patients who were randomly selected in a stratified manner from a pool of 571 patients who underwent surgical resection or biopsy at Heidelberg University Hospital. Model performance was evaluated on a hold-out internal test set comprising 115 patients and externally validated on 159 patients recruited at Mainz University Hospital. On the hold-out internal test set, HEPNET achieved an area under the receiver operating characteristic curve of 0.994 (95% CI, 0.989-1.000) and an accuracy of 96.522% (95% CI, 94.521%-98.694%) at the patient level. Validation on the external test set yielded an area under the receiver operating characteristic curve of 0.997 (95% CI, 0.995-1.000), corresponding to an accuracy of 98.113% (95% CI, 96.907%-100.000%). HEPNET surpassed the performance of 6 pathology experts with different levels of experience in a reader study of 50 patients (P= .0005), boosted the performance of resident pathologists to the level of senior pathologists, and reduced potential downstream analyses. We provided a ready-to-use tool with clinical grade performance that may facilitate routine pathology by rendering a definitive diagnosis and guiding ancillary testing. The incorporation of HEPNET into pathology laboratories may optimize the diagnostic workflow, complemented by test-related labor and cost savings.

Application of the ViroKey® SQ FLEX assay for detection of cytomegalovirus antiviral resistance

BackgroundCytomegalovirus (CMV) is a viral infection which establishes lifelong latency, often reactivating and causing disease in immunosuppressed individuals, including haematopoietic stem cell transplant (HSCT) recipients. Treatment can be problematic due to antiviral resistance which substantially increases the risk of patient mortality. Diagnostic testing capabilities for CMV antiviral resistance in Australia and elsewhere have traditionally relied on gene-specific Sanger sequencing approaches, however, are now being superseded by next generation sequencing protocols. ObjectiveProvide a snapshot of local mutations and explore the feasibility of the ViroKeyࣨ® SQ FLEX Genotyping Assay (Vela Diagnostics Pty Ltd) by examining sequencing success. MethodPerformed sequencing on adult (n = 38) and paediatric (n = 81) plasma samples, over a large range of viral loads (above and below the assay recommended threshold of ≥1,000 International Units (IU)/mL; noting most of our paediatric samples have loads <1,000 IU/mL). ResultsEleven test runs (including three repeat runs; 14 to 15 samples per run) were conducted, and four runs were deemed valid. The overall individual sample success rate for the four evaluable test runs was 71.2% (42/59 samples); 80.4% (37/46) samples ≥1,000 IU/mL were valid. Ten clinically important antiviral resistance mutations were detected, the most common being A594V in the UL97 gene, found in 6 (5%) samples. ConclusionsA range of technical issues were experienced, however with improvement this platform could be a useful addition to routine pathology workflows, providing timely antiviral resistance results for patients undergoing HSCT.

Whole Slide Imaging Technology and Its Applications: Current and Emerging Perspectives.

Background. Whole slide imaging (WSI) represents a paradigm shift in pathology, serving as a necessary first step for a wide array of digital tools to enter the field. It utilizes virtual microscopy wherein glass slides are converted into digital slides and are viewed by pathologists by automated image analysis. Its impact on pathology workflow, reproducibility, dissemination of educational material, expansion of service to underprivileged areas, and institutional collaboration exemplifies a significant innovative movement. The recent US Food and Drug Administration approval to WSI for its use in primary surgical pathology diagnosis has opened opportunities for wider application of this technology in routine practice. Main Text. The ongoing technological advances in digital scanners, image visualization methods, and the integration of artificial intelligence-derived algorithms with these systems provide avenues to exploit its applications. Its benefits are innumerable such as ease of access through the internet, avoidance of physical storage space, and no risk of deterioration of staining quality or breakage of slides to name a few. Although the benefits of WSI to pathology practices are many, the complexities of implementation remain an obstacle to widespread adoption. Some barriers including the high cost, technical glitches, and most importantly professional hesitation to adopt a new technology have hindered its use in routine pathology. Conclusions. In this review, we summarize the technical aspects of WSI, its applications in diagnostic pathology, training, and research along with future perspectives. It also highlights improved understanding of the current challenges to implementation, as well as the benefits and successes of the technology. WSI provides a golden opportunity for pathologists to guide its evolution, standardization, and implementation to better acquaint them with the key aspects of this technology and its judicial use. Also, implementation of routine digital pathology is an extra step requiring resources which (currently) does not usually result increased efficiency or payment.

Intraoperative Pathology Assessment May Lead to Overtreatment of the Axilla in Clinically Node-Negative Breast Cancer Patients Undergoing Upfront Mastectomy.

Randomized trials have established the safety of observation or axillary radiation (AxRT) as an alternative to axillary lymph node dissection (ALND) in patients with limited nodal disease who undergo upfront surgery. Variability remains in axillary management strategies in cN0 patients undergoing mastectomy found to have one to two positive sentinel lymph nodes (SLNs). We examined the impact of intraoperative pathology assessment in axillary management in a national cohort of AMAROS-eligible mastectomy patients. The National Cancer Database was used to identify AMAROS-eligible cT1-2N0 breast cancer patients undergoing upfront mastectomy and SLN biopsy (SLNB) and found to have one to two positive SLNs, from 2018 to 2019. We constructed a variable defining intraoperative pathology as 'not performed/not acted on' if ALND was either not performed or performed at a later date than SLNB, or 'performed/acted on' if SLNB and ALND were completed on the same day. Adjusted multivariable analysis examined predictors of treatment with both ALND and AxRT. Overall, 8222 patients with cT1-2N0 disease underwent upfront mastectomy and had one to two positive SLNs. Intraoperative pathology was performed/acted on in 3057 (37.2%) patients. These patients were significantly more likely to have both ALND and AxRT than those without intraoperative pathology (41.0% vs. 4.9%; p<0.001). On multivariate analysis, the strongest predictor of receiving both ALND and AxRT was use of intraoperative pathology (odds ratio 8.99, 95% confidence interval 7.70-10.5; p<0.001). We advocate that consideration should be made for omission of routine intraoperative pathology in mastectomy patients likely to be recommended postmastectomy radiation to minimize axillary overtreatment with both ALND and AxRT in appropriate patients.

Needle in a haystack or elephant in the room? Identifying germline predisposition syndromes in the setting of a new myeloid malignancy diagnosis.

Myeloid malignancies associated with germline predisposition syndromes account for up to 10% of myeloid neoplasms. They are classified into three categories by the proposed 5th Edition of the World Health Organization Classification of Hematolymphoid Tumors: (1) neoplasms with germline predisposition without a pre-existing platelet disorder or organ dysfunction, (2) neoplasms with germline predisposition and pre-existing platelet disorder, or (3) neoplasms with germline predisposition and potential organ dysfunction. Recognizing these entities is critical because patients and affected family members benefit from interfacing with hematologists who specialize in these disorders and can facilitate tailored treatment strategies. However, identification of these syndromes in routine pathology practice is often challenging, as characteristic findings associated with these diagnoses at baseline are frequently absent, nonspecific, or impossible to evaluate in the setting of a myeloid malignancy. Here we review the formally classified germline predisposition syndromes associated with myeloid malignancies and summarize practical recommendations for pathologists evaluating a new myeloid malignancy diagnosis. Our intent is to empower clinicians to better screen for germline disorders in this common clinical setting. Recognizing when to suspect a germline predisposition syndrome, pursue additional ancillary testing, and ultimately recommend referral to a cancer predisposition clinic or hematology specialist, will ensure optimal patient care and expedite research to improve outcomes for these individuals.

010 Enhanced outcome prediction in cutaneous squamous cell carcinoma using deep-learning and computational histopathology (cSCCnet)

Abstract Cutaneous squamous cell carcinoma (cSCC) is a common skin cancer and leads to a significant disease burden. Although metastases occur in &amp;lt; 5% of patients, effective treatments are limited and 5-year survival is poor, with mortality rates equivalent to those of melanoma in some regions of the world. Despite recent advances in understanding the molecular pathogenesis of cSCC, patient risk stratification is limited and based on clinicopathological staging, with few validated prognostic biomarkers available to guide management [de Jong E, Lammerts MUPA, Genders RE, Bowes Bavinck JN. Update of advanced cutaneous squamous cell carcinoma. J Eur Acad Dermatol Venereol 2022; 36(Suppl. 1):6–10]. In this study, we aimed to use artificial intelligence technology to develop a deep learning model using digitized whole-slide images of cSCC to quantify the risk of metastasis. Routine haematoxylin and eosin-stained slides of 256 primary cSCCs from immunocompetent patients at four collaborating centres were scanned. The area of interest encompassing tumour and peritumoral stroma was annotated, and 512 × 512 pixel tiles were generated from each slide. High-quality tiles (containing sufficient tumour content) were pooled together, with patient-level labels inherited by tiles. We trained multiple models using an 80:10:10 split for training, validation and testing, and the model achieving the highest tile- and slide-level validation accuracy was identified (ResNet50). The final model, namely, cSCCnet, achieved 88% sensitivity, 73% specificity, 64% positive predictive value and 92% negative predictive values on the independent test set. Blinded review of the top predictive tiles by three histopathologists suggested that the predictive features identified by cSCCnet were distinct from histopathological features commonly used by clinicians to estimate metastatic risk. We have demonstrated that it is possible to train a deep learning model using digital pathology to predict the risk of metastasis. cSCCnet surpassed the published performance of currently used staging systems, as well as commercially available gene panel testing (de Jong et al.; Wysong A, Newman JG, Covington KR et al. Validation of a 40-gene expression profile test to predict metastatic risk in localized high-risk cutaneous squamous cell carcinoma. J Am Acad Dermatol 2021; 84:361–9). The morphological features and patterns identified by cSCCnet are different from established histopathological high-risk criteria, indicating that this technology can enhance current practice. Although further validation in a prospective cohort is pending, this is an important step toward creating a risk stratification tool that can be incorporated into routine pathology workflows, significantly improving patient outcomes for this common cancer.