Laboratory Investigation最新文献

Surgical pathology reports provide essential diagnostic information critical for cancer staging, treatment planning, and cancer registry documentation. However, their writing styles and formats vary widely, reflecting each pathologist's stylistic choices, institutional norms, and inherited practices from residency training. When performing large-scale data analysis, this unstructured nature and variability across tumor types and institutions pose significant hurdles for automated data extraction. To overcome these challenges, we present a consensus-driven, reasoning-based framework that adapts multiple locally deployed large language models (LLMs) to extract both standard diagnostic variables (site, laterality, histology, stage, grade, and behavior) and organ-specific biomarkers. Each LLM generates structured outputs, accompanied by justifications, which are subsequently evaluated for accuracy and coherence by three separate reasoning models (DeepSeek-R1-large, Qwen3-32B, and QWQ-32B). Final consensus values are determined through aggregation. Board-certified pathologists conducted expert validation. This framework was applied to over 6,100 pathology reports from The Cancer Genome Atlas (TCGA) spanning 10 organ systems and 510 reports from Moffitt Cancer Center. For the TCGA dataset, automated evaluation demonstrated mean accuracy of 84.9%±7.3%, with histology (88%), site (87%), stage and behavior (84%) showing the highest extraction accuracy averaged across all models. Expert review of randomly selected 138 reports confirmed high agreement for behavior (100.0%), histology (99%), grade (96%), and site (95%) in the TCGA dataset, with slightly lower performance for stage (89%) and laterality (88%). In Moffitt Cancer Center reports (brain, breast, and lung), accuracy remained high (88.2%±7.2%), with behavior (99%), histology (97%), laterality (96%), grade (94%), and site (93%) achieving strong agreement. Biomarker extraction achieved 70.6%±7.9% overall accuracy, with TP53 (84%) on brain tumor, Ki-67 (68%) on breast cancer, and ROS1 (82%) on lung cancer showing highest accuracy. Inter-evaluator agreement analysis revealed high concordance (correlations ≥ 0.93) across the three evaluation models. Statistical analyses revealed significant main effects of model type (F=1716.82, p<0.001), variable (F=3236.68, p<0.001), and organ system (F=1946.43, p<0.001), as well as model × variable × organ interactions (F=24.74, p<0.001), emphasizing the role of clinical context in model performance. These results highlight the potential of stratified, multi-organ evaluation frameworks with multi-evaluator consensus in LLM benchmarking for clinical applications. Overall, this consensus-based approach demonstrates that locally deployed LLMs can provide a transparent, accurate, and auditable solution for integration into real-world pathology workflows such as synoptic reporting and cancer registry abstraction.

Intestinal epithelial barrier loss has been suggested as a pathogenic factor in Hirschsprung-associated enterocolitis (HAEC) and intestinal dysfunction in children with Hirschsprung disease (HD), but there is a paucity of comprehensive studies on the tight junction proteins that regulate paracellular permeability in this population. This case-control study aimed to determine if colonic epithelial tight junction protein expression is altered in children with HD. We use quantitative immunofluorescence microscopy to assess the expression of tight junction proteins (claudins 1, 2, 3, 4, 7, 15; ZO-1, ZO-2, occludin) in 29 children with HD who underwent surgical reconstruction and 16 controls who underwent transmural colorectal surgical resection for other etiologies between January 2015 and December 2021. We found that the expression of claudin-2, claudin-15, and occludin was reduced in both ganglionic and aganglionic colon specimens from children with HD compared to controls. Expression of other tight junction proteins did not differ between the groups. Together with previous studies, these data suggest that decreased expression of paracellular Na⁺ and water channel-forming claudin-2 and claudin-15 may limit lumenal hydration, enhance fecal stasis, and promote dysbiosis. Conversely, occludin downregulation can increase paracellular macromolecular flux, but also limit epithelial sensitivity to apoptotic stimuli. Together, these changes in expression of claudins 2 and 15, as well as occludin, may promote intestinal dysfunction and contribute to HAEC pathogenesis following surgical reconstruction.

All knee joint tissues undergo aging- and osteoarthritis (OA)-associated changes, but our understanding of the knee as an organ, and the tissue crosstalk in homeostasis, aging and OA is limited. The emergence of molecular profiling imaging technologies now enables comprehensive profiling of joint tissues to address these knowledge gaps. Successful application of these novel technologies requires a precise clinical diagnosis, and a rigorous and consistent definition of tissue-specific variables, including documentation of the regions of interest selected for macroscopic, histological, cellular, and molecular characterization. Macroscopic and histological scoring systems represent a benchmark for the interpretation of cellular and molecular analyses. Thus, standardizing these systems is essential to minimize experimental variability. Currently, most joint tissues lack a universally accepted scoring system, and various histological features are selected and quantified using different methods, limiting comparability and reproducibility across studies. Here, we review current methods, discuss limitations, and propose new approaches based on features that should be consistently evaluated across tissue types to overcome these caveats.

Triple-negative breast cancer (TNBC) represents 10-15% of all breast cancer cases, predominantly affecting younger women. Due to the absence of hormone receptors and HER2 expression, TNBC lacks effective targeted therapies, resulting in poor prognosis, with 5-year survival rates ranging from 91% for localized disease to 12% for metastatic disease. Fascin, encoded by the FSCN1 gene, is overexpressed in 88% of TNBC cases and promotes tumor invasion, metastasis, and chemotherapy resistance. This study explored fascin as a prognostic marker and therapeutic target by analyzing its expression in the largest TNBC cohort to date and assessing the effects of imipramine, an antidepressant that acts as a fascin inhibitor, on TNBC and luminal breast cancer cell lines. In a retrospective cohort of 145 TNBC patients, fascin expression in tumor cells and stromal myofibroblasts correlated with high histological grade, Ki67 >30%, and BCL-2 overexpression in myofibroblasts, as well as with higher chemotherapeutic response rates in the surgical setting. Fascin expression in stromal myofibroblasts has been identified as an independent predictive marker of chemotherapeutic response and as a prognostic factor for improved survival in patients undergoing neoadjuvant chemotherapy. In vitro, imipramine significantly reduced FSCN1 expression and impaired cell migration in TNBC (MDA-MB-231) and luminal (MCF7) cell lines, with stronger effects on TNBC. These findings highlight the dual role of fascin in tumor cells and the tumor microenvironment (TME), and reinforce its potential as a biomarker for personalized TNBC therapies. Ongoing clinical trials, including HITCLIF, are exploring the efficacy of imipramine in patients with fascin-overexpressing cancers, paving the way for targeted treatment strategies.