Deep learning (DL) methods are increasingly applied to digitized hematoxylin and eosin (H E) slides in breast cancer to assist in staging and prognosis. One emerging application is the assessment of axillary lymph node (ALN) status, either by detecting metastases on lymph node slides or by predicting nodal involvement from primary tumor histology. This narrative review summarizes recent progress in these two complementary areas. Literature was identified through a focused, non-systematic search of PubMed, Scopus, and Web of Science using terms related to breast cancer, lymph node metastasis, and deep learning on H E slides. Prediction studies showed variable discrimination. In independent or external evaluations, reported AUCs ranged from approximately 0.50 to 0.83, with multimodal models generally outperforming image-only models. However, performance fell across institutions and favorable luminal cohorts. Endpoint definitions were inconsistent, and several studies did not state whether isolated tumor cells (ITCs), micrometastases, or only macrometastases constituted a positive node. In contrast, DL models for detecting metastases on lymph node slides achieved near-perfect sensitivity and are now entering clinical use, reducing immunohistochemistry workload and review time without compromising diagnostic accuracy. Overall, deep learning on histopathology slides holds promise for improving axillary staging efficiency and precision, with metastasis detection already clinically viable and predictive modeling from primary tumor morphology representing the next frontier for research and validation.
CIC::NUTM1-rearranged sarcoma is a rare molecular variant of CIC-rearranged undifferentiated round cell sarcoma (URCS), with fewer than 20 cases reported to date. We present eight additional cases from two institutions to expand the clinicopathologic spectrum of this emerging variant. Our cohort comprised eight patients (five females, three males) with a median age of 8 years (range 2–15). Tumors arose predominantly in axial sites, including paraspinal soft tissue (n = 4), head and neck soft tissue (n = 2), temporal lobe meninges (n = 1), and distal extremity (right index finger; n = 1). Histologically, tumors showed sheets of round cells with variable rhabdoid or spindled features within myxoid or chondromyxoid stroma, frequently with necrosis. Immunohistochemically, tumors expressed NUT (6/6), CD99 (7/7), and WT1 (3/5); DUX4 was negative in all tested cases (0/6). RNA sequencing confirmed CIC::NUTM1 fusions involving CIC exons 16–18 and NUTM1 exons 5–6. Among five patients with follow-up, three died of disease (12, 28, and 34 months), one was alive with disease, and one was alive without disease. Pooled survival analysis of 21 total cases demonstrated a median overall survival of 17 months and median progression free survival of 9 months, without statistically significant survival difference compared to deep-seated CIC::DUX4 sarcoma. CIC::NUTM1 sarcoma is an aggressive malignancy with a predilection for pediatric patients and axial sites. Given its morphologic and immunohistochemical variability, diagnosis requires high index of suspicion. An immunopanel of NUT, CD99, WT1, and DUX4 serves as a valuable screening tool in the appropriate histologic context.
Folliculin (FLCN)-mutated renal tumors, in the setting of Birt-Hogg-Dubé (BHD) syndrome or sporadically, typically show low-grade eosinophilic morphology and follow an indolent clinical course; however, rare high-grade variants with more aggressive behavior remain poorly defined and underrecognized. To better characterize this unusual subset, we identified six high-grade renal cell carcinomas (RCCs) and performed integrated clinicopathologic, immunohistochemical, molecular, and clinical follow-up analyses. All tumors demonstrated infiltrative growth, extensive necrosis, variable desmoplastic stroma, and high-grade nuclear features. They exhibited papillary/tubulopapillary and solid growth patterns, mixed eosinophilic and clear cell cytology, focal biphasic morphology, and prominent nucleoli, overlapping morphologically with high-grade papillary RCC, FH-deficient RCC, and TFE3- or TFEB-rearranged RCC. Targeted DNA sequencing identified FLCN frameshift mutations in all cases, including four germline mutations, one likely germline mutation, and one somatic mutation, without additional alterations characteristic of other established RCC subtypes. In keeping with the FLCN-mutated phenotype, all tumors extensively and strongly expressed GPNMB. Clinically, three patients had features consistent with BHD syndrome; one additional patient lacked typical pulmonary and cutaneous findings but was confirmed to carry a germline mutation; another could not be definitively classified as having BHD and was considered suspected; and the remaining one had no clinical evidence of BHD and was considered likely sporadic. At last follow-up, three patients were free of recurrence or metastasis at 10, 25, and 7 months; the remaining three developed distant metastases, including two who died of disease at 17 and 26 months postoperatively and one who remained alive with metastatic disease at 85 months. Our findings expand the morphologic spectrum of FLCN-mutated renal tumors and the genetic landscape of papillary-patterned RCCs, raising awareness of this aggressive subset and supporting consideration of FLCN testing in morphologically overlapping high-grade papillary tumors.
Low-grade fibromyxoid sarcoma (LGFMS) arising at acral sites is exceedingly rare, and its clinicopathologic features and clinical outcomes remain poorly characterized. Herein, we report a cohort of ten cases of acral LGFMS. The cohort included six males and four females, with a mean age at diagnosis of 23 years (median, 14 years; range, 2–67 years). Tumors arose in the foot (n = 7) or hand (n = 3). All cases were superficially located and exhibited classic morphologic features of LGFMS. Diffuse MUC4 expression was observed in all nine cases tested. Fluorescence in situ hybridization (FISH) and next-generation sequencing (NGS) demonstrated FUS gene rearrangements in all five cases analyzed. Clinical follow-up was available for six patients, with a mean duration of 37 months (range, 6–91 months). One patient developed a local recurrence and subsequently underwent digital amputation. No patients have developed metastases to date. Recognition of this rare, deceptively bland neoplasm in acral locations is critical, given its potential for local recurrence and metastatic spread.
Langerhans cell sarcoma (LCS), malignant indeterminate dendritic cell tumor (IDCT), and interdigitating dendritic cell sarcoma (IDCS) are rare malignant neoplasms that originate from antigen-presenting conventional dendritic cells (cDCs). These tumors, collectively termed cDC sarcoma in this study, have been reported only rarely in association with myeloid malignancies. We collected five cases of cDC sarcoma coexisting with acute myeloid leukemia (AML) or myelodysplastic neoplasm with increased blasts (MDS-IB), conducted clinicopathologic and genetic analyses, and compared these with four reported cases. Among the nine cases, histological subtypes included LCS (n = 6), malignant IDCT (n = 2), and IDCS (n = 1), whereas associated myeloid neoplasms comprised AML (n = 8) and MDS-IB2 (n = 1). Clinically, five patients developed cDC sarcoma and AML/MDS-IB nearly simultaneously, whereas four developed AML after cDC sarcoma. Six patients died or required palliative care within 2 years, whereas three achieved complete remission or long-term survival. Nodal involvement predominated (8/9 cases). Immunohistochemically, cDC sarcomas frequently expressed myeloid markers (CD33 5/5; CD34 2/8). Conversely, myeloid blasts expressed histiocytic/monocytic-associated markers including CD68 and lysozyme in 3/5 cases, whereas S-100 and CD1a were focally expressed in only one case. Genetically, next-generation sequencing revealed shared mutations—including those in the RAS–MAPK pathway—in four of five paired cases. Further, one case harbored a shared KMT2A rearrangement, which supports a common clonal origin. These results indicate a close biological relationship between cDC sarcoma and myeloid neoplasms and highlight the importance of recognizing concomitant myeloid disease in patients with cDC sarcoma.
Artificial intelligence (AI) is rapidly transforming histopathology, with applications ranging from workflow optimisation and quality assurance to tumour diagnosis, grading, biomarker assessment and estimation of prognosis. While numerous AI algorithms have demonstrated promising analytical and clinical performance, pathology laboratories are increasingly adopting commercially available AI systems with regulatory-approval rather than developing their own algorithms. Existing guidance largely focuses on AI development, validation and regulatory approval, with comparatively little practical direction on the local verification, governance and ongoing assurance required for safe routine clinical implementation. This paper proposes a practical framework for the clinical implementation of AI specifically within pathology laboratories. Rather than addressing AI development, it focuses on the responsibilities of laboratories adopting established AI systems into clinical practice. The framework distinguishes AI applications according to their intended clinical function, recognising that diagnostic applications, biomarker evaluation, workflow optimisation and generative AI applications require different implementation, verification, governance and quality assurance strategies. It further distinguishes algorithm validation, local verification and continuous assurance as complementary stages of implementation and advocates a function-based, risk-proportionate approach integrated within existing laboratory quality management systems. Practical recommendations are provided for workflow integration, interoperability, human oversight, user competency, performance monitoring, incident management, software updates and proportionate re-verification throughout the AI operational lifecycle. By extending implementation beyond regulatory approval, this guidance complements existing AI development and regulatory frameworks rather than replacing them. It provides a practical governance framework for pathology laboratories, professional organisations, accreditation bodies, and healthcare providers to support the safe, standardised, and sustainable integration of AI into routine histopathology while maintaining diagnostic quality, patient safety, and clinical governance.
The literature is highly conflicting regarding the nature and behavior of pancreatic ductal adenocarcinomas (PDAC) arising from intraductal papillary mucinous neoplasm (IPMN). We investigated and contrasted the clinicopathological features of PDACs associated with IPMN (I-PDAC, n = 52 - only bona-fide and ≥ 1 cm IPMNs along with ordinary pancreatobiliary type invasive adenocarcinomas were included in this group-), versus without (NI-PDAC, n = 390) versus those associated with other IPMN mimicker cysts (pseudo-IPMN-associated, ps-PDAC, n = 39) including retention cysts. In I-PDAC, the average size of the IPMN was 4.05 cm, and 36.5
The FET(FUS/EWSR1)::TFCP2 fusion defines a distinct molecular subtype of spindle cell/sclerosing rhabdomyosarcoma (RMS), typically involving the mandible and maxilla. These tumors demonstrate myogenic differentiation and are clinically aggressive. In addition to the characteristic fusion, they often show features of homologous recombination deficiency (HRD), genomic instability, and expression of ALK and TERT truncated variants. This study presents a cohort of 25 patients with FET::TFCP2 fusion, representing the largest single-center series to date. While most tumors involve bones, 8 cases are soft tissue tumors, including the cheek (2 cases), epididymis, bladder, abdominal wall, neck, temporal part, and scalp. Genomically, approximately 40
Although most dysplastic lesions in inflammatory bowel disease (IBD) resemble sporadic adenomas, several less well-recognized morphologic variants have recently been described. These variants can be diagnostically challenging because of their subtle cytologic atypia and/or overlap with regenerative changes, often resulting in misclassification as benign, reactive, or indefinite for dysplasia. Furthermore, the use of different classification systems has led to inconsistent terminology and definitions across the literature and in clinical practice. Nevertheless, accurate recognition of these lesions is essential, as many are endoscopically invisible or flat and are associated with advanced neoplasia on follow-up, as well as synchronous and/or metachronous neoplasia, frequently arising within the same colonic segment. Accordingly, this review summarizes the evolution of the classification of IBD-associated dysplasia culminating in the 6th edition of the World Health Organization (WHO) Classification of Digestive System Tumours and discusses how the WHO framework relates to previously proposed classification systems. It further clarifies the terminology and diagnostic criteria of various dysplastic entities recognized in the WHO classification. In addition, the review presents a practical approach to the recognition and classification of these dysplastic variants and provides a standardized reporting framework to facilitate clinical management, including recommendations for lesions that exhibit partial histologic features but do not meet the criteria for a definitive diagnosis of dysplasia. Finally, the review provides an updated overview of the clinicopathologic features of these dysplastic variants and discusses ongoing controversies and unresolved questions in the field.
The American Joint Committee on Cancer (AJCC) staging system for MCC defines pT1–pT3 disease by tumor size and pT4 disease as disease invading deep structures, including skeletal muscle. However, the prognostic significance of skeletal muscle invasion in MCC remains incompletely characterized, especially in the head and neck (H N), where the dermis and subcutis are often thin and even small tumors may invade skeletal muscle. We evaluated the clinicopathologic significance of skeletal muscle invasion in H N and non-H N MCC. MCCs diagnosed at a tertiary cancer center during 2002–2025 were stratified by anatomic site and presence of skeletal muscle invasion. Clinicopathologic features, Merkel cell polyomavirus (MCPyV) status, treatment parameters, and clinical outcomes were analyzed. The study included 65 cases: 23 H N skeletal-muscle-invasive, 18 H N non-skeletal-muscle-invasive, 4 non-H N skeletal-muscle-invasive, and 20 non-H N non-skeletal-muscle-invasive MCCs. Muscle-invasive tumors were significantly larger and thicker than non-muscle-invasive tumors and required more surgical procedures to achieve negative margins. In the H N, skeletal-muscle-invasive tumors had higher rates of vascular invasion and distant metastasis (3-year rate: 31
The diagnostic of hepatocellular adenoma (HCA) and hepatocellular carcinoma (HCC) has shifted from a classical morphological evaluation to a complex and precise morpho-molecular analysis that represents a core focus of the updated 6th edition of the WHO classification of digestive system tumours. Four molecular pathways across five distinct subtypes have been identified for HCA, with clinical implications. When specific immunohistochemical and molecular techniques are applied, less than 5
We report a case of a 28-year-old man who presented with a slowly enlarging intramuscular mass in left posterior neck. Imaging demonstrated a relatively well-circumscribed mass with focal scattered calcifications. Histologic examination revealed a lobulated neoplasm composed of monotonous round to ovoid to spindle shaped tumor cells arranged in large nodules, loose fascicles, and solid nests within a sclerotic fibrous stromal background. Focal metaplastic bone formation and rare calcification were identified. No significant cytologic atypia, increased mitotic activity or necrosis was identified. The tumor showed immunophenotypic evidence of rhabdomyoblastic differentiation. Molecular studies identified a novel EWSR1::ATF7 fusion and methylation profiling yielded no matching classes. Overall, the findings favor classification as a fusion-associated spindle cell rhabdomyosarcoma. However, additional studies involving a larger number of cases will be required to confirm its classification as such and to better define its biologic behavior.
Loss of expression of TTF1 and PAX8 in primary thyroid malignancies raises concerns for metastatic disease in addition to a high-grade primary tumor. Precise distinction is critical, as treatment strategies differ. Herein, we present the case of a 74-year-old man with a right-sided neck swelling, and change in voice for 2 months. Clinically, a 6 cm restricted mobility mass involving thyroid was detected, which on CT scan showed abutment of trachea and esophagus, along with right cervical and upper-mediastinal lymphadenopathy. Biopsies from right thyroid lobe and right supraclavicular node showed an infiltrative carcinoma in nests and cords, a distinct squamoid appearance, focal rhabdoid morphology, extensive necrosis, and a desmoplastic stroma. Overt anaplasia was absent. Immunohistochemically, the tumor was positive for cytokeratins; negative for PAX8, TTF1, CK20, p63, p40, CDX2, SOX10, AR and synaptophysin; p53 showed wild-type staining. Further evaluation revealed retained SMARCB1 and complete loss of SMARCA4 (BRG1) expression in tumor cells. Lack of thyroid lineage markers prompted molecular profiling, and NGS revealed Pathogenic mutations in exon 3 of NRAS gene and exon 11 of BRAF gene (non BRAF v600E), consistent with a follicular thyroid carcinoma profile. Thus, a diagnosis of anaplastic thyroid carcinoma with SMARCA4 loss; likely arising from dedifferentiation of a follicular thyroid carcinoma, was made. The patient is on palliative chemotherapy, with progressive disease. SMARCA4-deficient carcinomas occur at various sites, de novo or as dedifferentiated carcinomas; hitherto unreported in thyroid. This case highlights the importance of integrating morphology, immunohistochemistry, and molecular testing in aggressive thyroid-associated carcinomas, with implications for diagnosis, prognostication, and targeted therapy.
Spatial immune cell profiling using multiplex immunofluorescence was examined to characterize the inflammatory cell infiltrates and to possibly aid in the distinction of acute cellular rejection (ACR) and lymphocytic bronchiolitis (LB) from its mimickers including bronchus-associated lymphoid tissue (BALT) and infection in lung allograft tissue. Lung allograft biopsies with ACR and/or LB were identified from an institutional pathology lung transplant database (2012–2024). Surgical lung specimens, transbronchial biopsies, and autopsy lungs were used for BALT and infection, respectively. Multiplex immunofluorescence using the PhenoCode Signature Immune Profile Human Protein Panel (CD3e, CD8, CD20, CD68, FoxP3, pankeratin) was performed. The study included tissue specimens with ACR (n = 13), LB (n = 10), BALT (n = 12), and infection (n = 13). The percentage of CD3 + T-cells, CD8 + T-cells, and CD68 + macrophages was higher and the percentage of CD20 + B-cells was lower in ACR, LB, and infection than BALT (p ≤ 0.02, all). The percentage of FoxP3 + CD3 + T-cells was higher in ACR and BALT than infection (p ≤ 0.02, all). In conclusion, ACR and LB are predominantly comprised of T-cells, however B-cells, macrophages, and FoxP3 + CD3 + T-cells are also present. The immune cell composition of ACR and LB is distinct from that of BALT but similar to infection.
The increasing role of molecular testing in thyroid pathology, from the pre-operative stratification of indeterminate fine-needle aspiration (FNA) categories to the management of advanced carcinomas, has prompted the development of institution-specific custom NGS solutions as a locally deployable alternative to centralized tests. The aim of the present study is to validate a new custom NGS panel dedicated to the molecular analysis of routine thyroid pathology specimens. The panel was evaluated on thyroid cancer cell lines, 124 FFPE thyroid tissue specimens, and 43 FNA samples. Nexthyro 2.0 achieved a high sequencing success rate on both histological (100
Categorizing HER2-negative invasive breast carcinomas into low, ultralow, and null tiers continues to challenge diagnostic reproducibility, largely due to spatial heterogeneity. This study evaluated inter-pathologist agreement and spatial intratumoral concordance across multi-site specimens using a highest-tier approach in treatment-naive invasive breast carcinomas (IBC). This retrospective study analyzed 125 primary IBC excisions (evaluated via two separate topographical blocks), 78 matched core needle biopsies, and 58 lymph node metastases (LNM). Two blinded pathologists independently reviewed slides stained with Ventana 4B5. Multi-block discrepancies were managed via a highest-tier approach (null < ultralow < low). Inter-pathologist reproducibility was substantial across all specimen types (weighted Kappa: 0.697–0.744, p < 0.001). Intra-tumor block agreement was moderate (Kappa = 0.580). The final highest-tier excision status yielded substantial concordance with core biopsies (Kappa = 0.773) but moderate concordance with LNM (Kappa = 0.486), primarily due to expression downgrading in nodal metastases (15/22 ultralow primary cases reverted to null). Pooling ultralow/low cases revealed a significant association with Grade 3 tumors (p = 0.049). HER2 status showed no independent correlation with overall survival (p > 0.05). Diagnostic discrepancies in minimal HER2 expression result from a combination of intrinsic spatial heterogeneity and the subjective challenges of inter-observer interpretation. Evaluating a second tumor block, combined with confirmation by a second pathologist for HER2-ultralow cases, effectively compensates for this variation, improving diagnostic consistency and the selection of candidates for novel targeted therapies.
Locally advanced prostate cancer (PCa) is associated with a high recurrence rate even after curative treatment. We aimed to develop a precise risk model by integrating the status of intraductal carcinoma of the prostate (IDC-P) and the International Society of Urological Pathology Grade Group (ISUP GG) with deep learning (DL)-based grading to predict clinical recurrence after intensity-modulated radiation therapy (IMRT). Furthermore, we identified key pathological features associated with IDC-P. We retrospectively analyzed 165 patients treated with high-dose IMRT for high- and very high-risk PCa at two institutions. Pathological features were extracted from hematoxylin and eosin–stained specimens from 100 cases using a DL-based model, from which an expert pathologist selected 10 cancer-related features. The area under the curve (AUC) values derived from split-sample validation for predicting clinical recurrence using ISUP GG and IDC-P status were 0.732 and 0.735, respectively, and 0.789 for their combination. Integrating 10 key features further improved the AUC to 0.850, with robust performance in external validation (AUC = 0.833). Kaplan–Meier analysis showed a significant difference (p < 0.05) in clinical recurrence between high- and low-risk prediction groups. Additionally, we identified four DL-derived pathological features that are significantly associated with IDC-P (p < 0.05). Incorporation of ISUP GG, IDC-P status, and DL-derived pathological features improves the prediction of clinical recurrence after high-dose IMRT in high- and very high-risk PCa. Our findings underscore the clinical significance of IDC-P and support optimized treatment strategies.
Intrahepatic cholangiocarcinoma (iCCA) is histologically classified into the small duct type (SDT) and large duct type (LDT), although their morphological differential diagnosis can be difficult. We aimed to identify new immunohistochemical markers to classify SDT and LDT with high sensitivity and specificity. We compared the gene expression profiles of SDT and LDT using cases obtained from our institution (n = 8) and The Cancer Genome Atlas database (n = 28). We selected the candidate molecules that were up-regulated in LDT, were minimally or not expressed in SDT, and had specific and available antibodies suitable for immunohistochemistry. We finally selected ST6 N-acetylgalactosaminide alpha-2,6-sialyltransferase 1 (ST6GalNAc1), Breast carcinoma amplified sequence-1 (BCAS-1), and Olfactomedin 4 (OLFM4), which were expressed in 100
Extensive myocardial disarray is a characteristic histopathological feature of hypertrophic cardiomyopathy (HCM), but its occurrence in hearts with normal cardiac mass remains poorly characterized. We reviewed 2,823 consecutive forensic autopsies and identified 1,387 hearts with normal cardiac weight. Ten cases (0.72
