Modern Pathology最新文献

Tubo-ovarian carcinosarcomas (OCS) are uncommon, aggressive tumors. Recent literature in uterine carcinosarcomas (UCS) has shown prognostic differences by the molecular classification established by The Cancer Genome Atlas (TCGA). Our aim was to delineate the molecular subtypes within OCS and associated clinicopathologic, immunohistochemical, and additional molecular features. A total of 57 OCS were identified at our institution. Overall median follow-up was 32.3 months, and 5-year survival rates were 71% (Stage I/II), 42% (Stage III), and 17% (stage IV). Fifty-one tumors (89%) were p53-abnormal (p53abn) molecular subtype. Five (9%) were no specific molecular subtype (NSMP) and all 5 of these tumors harbored canonical mutations in KRAS (codon 12). We also identified the first reported primary POLE-mutated OCS in a patient with Lynch Syndrome; this case was assigned as double-classifier POLE-mutated/mismatch repair deficient molecular subtype. No tumors were single-classifier mismatch repair deficient or POLE-mutated molecular subtype. Compared to p53abn tumors, KRAS-mutated tumors occurred in younger women at lower stages, but did recur in 2/5 (40%) patients. They always showed endometrioid rather than high-grade serous morphology, and were usually ER, PR and WT1 negative. Three KRAS-mutated tumors also had at least focal mesonephric-like histology. Although rare, p53 wild-type tumors represent a small subset of OCS that show distinct clinical and histologic features and are largely driven by KRAS mutations.

Despite promising results in using deep learning to infer genetic features from histological whole-slide images (WSIs), no prior studies have specifically applied these methods to lung adenocarcinomas from subjects who have never smoked tobacco (NS-LUAD) - a molecularly and histologically distinct subset of lung cancer. Existing models have focused on LUAD from predominantly smoker populations, with limited molecular scope and variable performance. Here, we propose a customized deep convolutional neural network based on ResNet50 architecture, optimized for multilabel classification for NS-LUAD, enabling simultaneous prediction of 16 molecular alterations from a single H&E-stained WSI. Key architectural modifications included a simplified two-layer residual block without bottleneck layers, selective shortcut connections, and a sigmoid-based classification head for independent prediction of each alteration, designed to reduce computational complexity while maintaining predictive accuracy. The model was trained and evaluated on 495 WSIs from the Sherlock-Lung study (70% training with 10% internal test set for 10-fold cross-validation, and 30% held-out validation set for final evaluation). For the held-out validation data, our model achieved high areas under the receiver operating characteristic curve [AUROC] values =0.84-0.93 for detecting 11 features: EGFR, KRAS, TP53, RBM10 mutations, MDM2 amplification, kataegis, CDKN2A deletion, ALK fusion, whole-genome doubling, and EGFR hotspot mutations (p.L858R and p.E746_A750del). Performance was low to moderate for tumor mutational burden (AUROC=0.67), APOBEC mutational signature (AUROC=0.57), and KRAS hotspot mutations (p.G12C: AUROC=0.74, p.G12V: AUROC=0.55, p.G12D: AUROC=0.43). Compared to results from established architectures such as Inception-v3 on the same WSIs, our model demonstrated significantly improved performance for most features. With further optimization, our model could support triaging for molecular testing and inform precision treatment strategies for NS-LUAD patients.

Pancreatic ductal adenocarcinoma (PDAC) is highly resistant to immune therapies. Limited biomarkers, such as mismatch repair proteins, have been used to identify those who may respond to immunotherapy. We identified a subset of aggressive PDACs (approximately 25%) carrying mutations in COMPASS-like complex genes (CLCG) which can be used as new biomarkers for targeted immunotherapy. In this study, we compared the immune microenvironment of PDACs harboring CLCG mutations with matched wild type PDACs using multiplex fluorescent immunohistochemistry (mfIHC) and computational imaging techniques. We observed CLCG-mutant PDACs were infiltrated with fewer CD4+ T cells and antigen-presenting cells (APCs), but elevated immune checkpoint TIGIT expression on CD4+ T cells and APCs. There was no difference in the expressions of other immune checkpoints, such as PD-L1 and TIM3. More CD4+ T cells near epithelial cells (tumor cells) and APCs expressed TIGIT in CLCG-mutant PDACs. Additionally, CLCG-mutant PDACs displayed a malfunctional immune cell crosstalk. Single cell RNA sequencing data confirmed the elevated TIGIT expression on CD4+ T cells and increased exhausted CD4+ T cells in CLCG-low PDACs. These findings uncovered the unique underlying mechanisms of immune suppression in CLCG-deficient PDACs and identified CLCG as potential biomarkers to identify those who may benefit from TIGIT-targeting immunotherapies.

Solid pseudopapillary neoplasm (SPN) of the pancreas is a rare tumor with generally indolent behavior, and surgical resection yields a 5-year survival rate exceeding 95%. However, 5-10% of patients develop metastases, underscoring the need for reliable prognostic biomarkers to identify individuals at higher metastatic risk and to optimize postoperative management. In this study, we performed DIA-MS-based proteomics profiling on resected primary tumors from 59 SPN patients to identify proteins differentially expressed between metastatic and non-metastatic cases. A candidate protein, MCM4, was further examined for potential pathogenic functions in vitro. An independent multicenter cohort of 255 patients was subsequently analyzed for MCM4 positivity by immunohistochemistry (IHC). Prognostic performance for postoperative metastasis was evaluated using Kaplan-Meier analysis, Cox regression, and time-dependent receiver operating characteristic (ROC) curves, with histopathologic invasion defined as neural, vascular, or peripancreatic invasion. MCM4 protein was aberrantly upregulated in tumors from patients who developed metastasis, and functional assays demonstrated a pro-proliferative role of MCM4. In the validation cohort, MCM4-positive tumors (MCM4 index ˃ 3%) exhibited a higher Ki-67 index (P=0.0006). Patients with MCM4-positive tumors had a higher incidence of metastasis (20.6% vs 4.1%, P=0.0019) and significantly shorter metastasis-free survival (MFS; log-rank P<0.0001). In univariate Cox regression analysis, MCM4 positivity was significantly associated with reduced MFS (HR=12.78, 95% CI: 3.01-54.31, P=0.0006). In multivariate Cox regression analysis, MCM4 positivity remained an independent prognostic biomarker for shorter postoperative MFS after adjustment for histopathologic invasion (HR=11.55, 95% CI: 2.76-48.37, P=0.0008). Time-dependent ROC analyses demonstrated that MCM4 positivity achieved AUCs of 0.817 (95% CI: 0.763-0.864) at 3 years and 0.777 (95% CI: 0.720-0.828) at 5 years for postoperative metastasis assessment. Taken together, these findings identify MCM4 positivity as an independent prognostic biomarker for postoperative metastasis risk assessment in SPN.

HER2 gene amplification and heterogeneity influence anti-HER2 therapy efficacy, but quantitative assessment of HER2 expression heterogeneity remains scarce. This study evaluates HER2 protein heterogeneity and its effect on neoadjuvant therapy (NAT) outcomes in HER2-positive breast cancer (BC). In total, 295 HER2-positive BC patients treated with neoadjuvant anti-HER2 therapy were included. HER2 expression was quantified by the percentage of HER2-stained cells across intensities, and heterogeneity was assessed using quadratic entropy (QE). The overall pCR rate was 51.86%. Patients with HER2 heterogeneity showed a 24.5% pCR rate, significantly lower than the 62.5% in patients with HER2 homogeneity (P < 0.001). HER2 heterogeneity QE (HER2-HQE) correlated with the percentage of HER2 IHC 3+ tumor cells and hormone receptor (HR) status (P < 0.001). Residual cancer burden (RCB) was significantly associated with HR status, tumor stage, lymph node stage, HER2 IHC 3+ cell percentage, and HQE grade (P ≤ 0.017). No correlation was found with age, histological grade, or Ki67 index. In the HER2 IHC 3+ ≥95% subgroup, neither pCR rate nor RCB grade after NAT was associated with traditional clinicopathological parameters, even HER2-HQE and HR status (P≥0.154). AJCC anatomical stage significantly improved post-NAT for low/intermediate HER2-HQE subgroups (P < 0.001), but not for high HQE cases. In summary, HER2 heterogeneity and HR status are key predictors of anti-HER2 NAT outcomes. Higher HER2-HQE is associated with increased RCB. Quantifying HER2 heterogeneity via HQE and percentage of HER2 IHC 3+ may help optimize therapeutic strategies in clinical practice.

Light chain proximal tubulopathy (LCPT) is a rare kidney disorder associated with monoclonal gammopathy, traditionally defined by crystalline inclusions in proximal tubular epithelial cells. A poorly recognized variant, fibrillary LCPT, is characterized by large non-amyloid fibrillar aggregates but has remained inconsistently classified, leading to diagnostic uncertainty. We aimed to clarify its clinicopathologic features and refine its position within the LCPT spectrum. We retrospectively analyzed 41 biopsy-proven LCPT cases, reclassified by ultrastructural features into crystalline LCPT (n = 22), fibrillary LCPT (n = 11), amyloid LCPT (n = 2), and LCPT with lysosomal indigestion (n = 6). Fibrillary LCPT showed irregular cytoplasmic fibrils measuring 6-18 nm in diameter, arranged in intersecting, fishbone-like, or compact bundle-like patterns, confined to proximal tubular epithelial cells. These fibrils were Congo red-negative and not apparent on light microscopy. Conventional immunofluorescence on frozen tissue was uniformly negative in fibrillary LCPT, and pronase-digested paraffin immunofluorescence detected κ restriction in only 7/11 cases; immunoelectron microscopy confirmed κ light chain labeling in all paraffin IF-negative cases, yielding 100% diagnostic sensitivity. Clinically, most fibrillary LCPT cases were associated with monoclonal gammopathy of renal significance (MGRS, 72.7%), with the remainder linked to multiple myeloma (27.3%). This subtype was commonly associated with Fanconi syndrome (FS) and reduced estimated glomerular filtration rate (eGFR), with a subset of cases developing acute kidney injury. Following clone-directed therapy, a majority of patients achieved stable renal function and 85% showed improvement in tubular dysfunction, although proteinuria reduction was less pronounced than in crystalline LCPT. Fibrillary LCPT represents a distinct, non-amyloid entity characterized by Congo red-negative fibrils, κ light chain restriction, and frequent association with FS and MGRS. We propose a refined four-tier ultrastructural classification comprising four LCPT subtypes (crystalline LCPT, fibrillary LCPT, amyloid LCPT, and LCPT with lysosomal indigestion) to improve diagnostic precision and guide management.

Accurate assessment of visceral pleural invasion (VPI) is essential for staging and prognostication in non-small cell lung cancer (NSCLC), yet distinguishing elastin-rich pleural layers on routine hematoxylin and eosin (H&E) sections remains a diagnostic challenge. To overcome the cost and workflow delays associated with special elastic stains, we developed a deep learning pipeline that generates a virtual elastin stain, termed synthetic eosin-based elastin fluorescence (EBEF), directly from standard brightfield H&E slides. A key innovation of this study was the use of intrinsic eosin fluorescence from the same H&E section to create a perfectly co-registered, high-fidelity ground truth for training a conditional generative adversarial network, eliminating the spatial mismatches common in multi-slide approaches. In a multi-institutional validation, supplementing H&E review with synthetic EBEF significantly improved pathologists' diagnostic accuracy for VPI compared with H&E alone (P < 0.0001). Notably, the pre-analytical factors that optimized model performance, including thinner tissue sections (1-3 μm) and high-resolution scanning, also enhanced the perceptual contrast of elastin for pathologists, demonstrating a strong synergy between computational and conventional diagnostic optimization. This study establishes and validates a robust framework for high-fidelity virtual staining that improves diagnostic accuracy and provides a scalable pathway for integrating deep learning-based tools into routine digital pathology. The proposed approach offers a practical and cost-effective alternative to ancillary special stains in NSCLC evaluation.

The extent of residual disease after neoadjuvant chemotherapy (NAC) in breast cancer (BC) patients holds prognostic value. However, current practices for reporting post-NAC BC specimens according to the ypTNM classification vary. This study aimed to map these practices and provide recommendations for standardization. A survey was developed and globally circulated to pathologists with a special interest in BC through personal networks and working group mailing lists. The survey included general questions about tumor diameter assessment, as well as graphical scenarios presenting different distributions of tumor cells. We did not provide definitions mentioned in reporting guidelines to capture unbiased current real-world practices. A total of 208 pathologists from 35 countries completed the survey. Almost all responding pathologists (97.1%) report the ypTNM in daily practice. Despite self-reported strict adherence to the 8^th edition of the international ypTNM classification, we found substantial variation in practice concerning the application of this staging system, particularly in cases with an uneven distribution of scattered residual disease. Notably, 57.2% of respondents reported measuring the largest 'continuous cluster of tumor cells', but the interpretation of this definition varied widely. This international survey identifies the challenges and the practice heterogeneity in the current application of the ypTNM staging system, which hampers the value of ypTNM reporting in daily practice. To enhance reproducibility and to provide more reliable post-NAC risk stratification, we recommend adopting standardized reporting with clearer pattern-based definitions of the ypTNM guidelines, supplemented with the elements of the Residual Cancer Burden system.

Hepatoid lung carcinomas, similar to hepatoid carcinomas of other sites, are defined as extrahepatic tumors exhibiting divergent hepatocellular differentiation. Uniquely, hepatoid carcinomas of lung origin are reported to commonly express only HepPar1 - a hepatocellular marker, which recognizes mitochondrial enzyme carbamoyl-phosphate synthetase-1 (CPS1). Recently, HepPar1/CPS1 was found to accumulate in lung adenocarcinomas (LUAD) harboring STK11 mutations, presumably as a genotype-associated metabolic adaptation. The impact of these insights on the concept of hepatoid lung carcinoma has not been explored. Here, we performed a detailed clinicopathologic and genomic analysis on carcinomas prospectively regarded as hepatoid with isolated HepPar1 expression (n=17). We found that while robustly positive for HepPar1, these tumors were entirely negative for an extended panel of other hepatocellular markers (AFP, Arginase1, Glypican3, Albumin-ISH). Morphologically, tumors exhibited solid-trabecular architecture with expanded granular-vacuolated-clear cytoplasm, thus evoking hepatoid morphology; however, focal-to-moderate intracytoplasmic mucin was consistently present and hepatoid resemblance was variable. Pneumocytic markers (TTF1, Napsin A) were entirely negative (except for cytoplasmic TTF1), commonly leading to diagnostic challenges at metastatic sites. Remarkably, next-generation sequencing revealed invariable STK11 mutations/loss (P<0.00001 vs unselected LUAD, n>2.5K). Patient survival was dismal (median: 5.8 vs 25 months for stage-matched LUAD, P=0.0002). Tumors harbored high mitochondrial content by electron microscopy and other methods. For comparison, we reviewed conventional, predominantly acinar LUAD with HepPar1 expression (n=22), and found that they also lacked any other hepatocellular markers, had invariable STK11 mutations/loss, increased granular cytoplasm, lower TTF1 and poor prognosis. We conclude that isolated HepPar1 expression in LUAD reflects mitochondrial adaptation to STK11 mutations rather than bona fide hepatocellular differentiation, and that HepPar1-expressing solid and granular adenocarcinomas (SAGA) represent an undifferentiated (solid, TTF1-negative) variant in this spectrum of tumors. Recognition of these tumors is warranted due to their exceptionally aggressive behavior, distinct pathogenomic features and common association with diagnostic challenges.