Search Results (81,810)

Aim: To evaluate the diagnostic value of conventional ultrasound (CUS) and contrast-enhanced ultrasound (CEUS) features in differentiating benign from malignant Bethesda III/IV thyroid nodules, and to identify independent predictors of malignancy. Methods: We retrospectively analyzed 164 surgically confirmed Bethesda III/IV thyroid nodules. CUS and CEUS features were evaluated by two experienced radiologists blinded to pathological outcomes. Univariate analysis compared features between benign and malignant groups. Multivariate logistic regression was used to identify independent predictors. Diagnostic models were constructed based on CUS alone, CEUS alone, and their combination, with performance evaluated using receiver operating characteristic (ROC) curve analysis. The area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for each model. Results: The malignancy rate was 48.8% (80/164). Multivariate analysis identified microcalcifications (OR = 4.815, p < 0.001), aspect ratio >1 (OR = 2.499, p = 0.028), and irregular shape (OR = 2.465, p = 0.035) as independent risk factors, while older age (OR = 0.926 per year, p < 0.001) was protective. The CUS model achieved an AUC of 0.815 with high sensitivity (91.3%) and NPV (87.7%). The CEUS model performed poorly (AUC = 0.609). The combined model (AUC = 0.823) showed no significant improvement over CUS alone (p > 0.05). Physician subjective diagnosis based on CEUS TI-RADS yielded an AUC of 0.775. Conclusions: Conventional ultrasound features provide good diagnostic value for Bethesda III/IV nodules, with high sensitivity and NPV suitable for clinical screening. The addition of CEUS offered limited incremental benefit in this specific population, suggesting that the diagnostic value of CEUS for differentiating benign from malignant cytologically indeterminate thyroid nodules (ITNs) may be limited. Full article

Background: Tumour volume percentage (TVP) is considered an important pathological parameter, particularly in prostate cancer, representing the ratio of tumour volume to the total gland, and it can be used to measure the quantity of malignancy. Previous reports have already demonstrated that an elevated tumour volume percentage is associated with unfavourable factors, including extraprostatic extension, positive surgical margins, and lymph node metastasis. The independent value of TVP, especially in high-risk prostate cancer treated by radical prostatectomy, remains an area of active research, despite established prognostic factors such as PSA, ISUP grade, and TNM stage. Materials and Methods: We retrospectively analyzed the records of 159 high-risk prostate cancer patients who underwent open or laparoscopic radical prostatectomy between January 2016 and January 2025 at the Clinic of Urology of Targu Mures. High-risk patients were defined as those with ISUP grade 4–5 or PSA >20 ng/mL or clinical stage ≥T2c or stage cT3–4 and/or lymph node metastasis. Tumour volume percentage was calculated from the final pathology result and was determined as the proportion of prostate cancer volume relative to the total prostate volume. Clinical and pathological features, including PSA, ISUP grade, TNM stage, surgical margin, and lymph node involvement, were reported. To assess TVP as an indicator of tumour aggressiveness, univariate and multivariate regression analyses were performed. A p-value <0.05 was considered statistically significant. Results: A total of 159 high-risk prostate cancer patients (100%), with a median age of 66 years, who underwent open or laparoscopic radical prostatectomy were included. The median tumour volume percentage was 7.6%, and the median prostate volume was 43.8 cc. On univariate analysis, patients with extraprostatic extension (p < 0.001), positive surgical margins (p = 0.005), a higher ISUP grade (p < 0.001), and lymph node metastasis (p = 0.006) exhibited higher TVP compared to their counterparts. A significant correlation was also observed between TVP and the number of positive biopsy cores (p < 0.001), a higher PSA value (p = 0.005), and a younger age (p = 0.041). Conversely, no correlation was identified between TVP and perioperative factors such as hospital stay, surgery duration, ICU days, type of approach, or positive urine culture. Two regression models on multivariate analyses were performed with TVP as the dependent variable. In the continuous variable model (Adjusted R² = 0.43, p < 0.001), independent predictors of higher TVP were the number of positive biopsy cores (B = 0.54, p < 0.001), the number of positive lymph nodes (B = 2.59, p < 0.001), and surgical margin dimension (B = 1.19, p < 0.001). Age, PSA, and perioperative variables showed no significant correlation with TVP on multivariate analysis. In the categorical regression model (Adjusted R² = 0.438), statistical significance was confirmed (F-test, p < 0.001). Independent predictors of increased tumour volume percentage included ISUP grade 5 in the effect-coded model (B = +6.60, 95% CI: 0.96–12.25, p = 0.022), and pathological TNM stage pT4 (B = +24.70, 95% CI: 17.69–31.70, p < 0.001). ROC analysis showed limited-to-moderate discrimination for positive surgical margins (AUC = 0.655; 95% CI 0.565–0.744; p = 0.001) and stronger discrimination for pN1 (AUC = 0.793; 95% CI 0.650–0.936; p = 0.002). The Youden-derived cut-offs were 4.90% for positive surgical margins and 5.77% for lymph-node metastasis. Conclusions: Tumour volume percentage is significantly associated with several adverse pathological features in high-risk prostate cancer. Rather than a standalone biomarker, its association with adverse pathological features underscores its potential role in risk stratification models, and the incorporation into pathology reports and prognostic nomograms may improve clinical decision-making. Full article

Background and Objectives: Pro-inflammatory diet and high-fat diet (HFD) often coexist in real-world, but their combined impact on the gut–liver axis and potential nutritional countermeasures remain insufficiently studied. This study aimed to evaluate a pro-inflammatory–high-fat composite dietary pattern on the intestine and liver in the population, and to further evaluate the protective potential of astaxanthin (ATX) in complementary experimental systems. Methods: Data from the NHANES 2005–2010 were used to construct four composite exposure groups based on the dietary inflammation index (DII) and energy from fat. Survey-weighted regression analyses were performed to examine associations with systemic inflammation and liver injury. Interaction and C-reactive protein (CRP)-mediated effect analyses were conducted. Fifty SD rats were randomly divided into control group, model group induced by HFD combined with inflammatory factors, and low-, medium-, and high-dose Haematococcus pluvialis (HP) intervention groups. Serum lipids, liver enzymes, liver and colon pathology, and inflammatory and oxidative markers were measured in rats. In an in vitro organ-on-chip barrier model, the effect of ATX was observed when colonic barrier damage was induced using palmitic acid and lipopolysaccharides. Results: The high DII combined with HFD showed the largest increases in CRP, liver enzymes, and fatty liver index. A synergistic interaction was observed between DII and HFD, with CRP mediating approximately 20% of the effect. In rat model, HP-derived ATX improved the lipid profile, attenuated hepatic steatosis and oxidative damage, and reduced colonic pro-inflammatory cytokines, while restoration of tight junction proteins was limited. In colon organoid model, ATX showed limited efficacy in improving inflammation and barrier function. Conclusions: The pro-inflammatory–high-fat dietary pattern synergistically exacerbates gut–liver dysfunction. HP-derived ATX alleviates metabolic and inflammation-induced enterohepatic comorbidity, but its effect on repairing barrier structure is limited. Full article

Background/Objectives: Diabetic peripheral neuropathy (DPN) and Charcot foot (CF) represent progressive and disabling neuropathic complications of type 2 diabetes mellitus (T2DM). Circulating microRNAs and inflammatory cytokines may reflect underlying molecular alterations associated with disease progression and offer potential value for discriminating between stages of diabetic neuropathic complications. This study aimed to evaluate circulating miRNA expression profiles and inflammatory cytokine biomarkers in T2DM patients with and without neuropathic complications and to assess their potential non-invasive utility as combined biomarkers for differentiating disease stages and identifying molecular patterns associated with progression from T2DM to DPN and CF. Methods: The study included the following four groups: healthy controls, T2DM patients without complications, T2DM patients with DPN, and T2DM patients with CF. Expression profiles of five miRNAs (miR-19b-3p, miR-451a, miR-199a-3p, miR-146a-5p, and miR-93-5p) were quantified using qPCR. Inflammatory cytokine biomarkers including NLRP3, TNF-α, NF-κB, IL-1β, caspase-3, and Serpin E2 were measured using ELISA assays. Results: Distinct expression patterns of both miRNAs and inflammatory cytokine biomarkers were observed across diabetic neuropathy stages. Several miRNAs demonstrated significant dysregulation in DPN and CF compared with T2DM patients without complications. Correlation analyses revealed stage-specific patterns of interaction between inflammatory cytokines and miRNAs, indicating coordinated molecular alterations across different stages of diabetic neuropathic complications. Conclusions: These findings suggest that combining circulating miRNA and inflammatory marker profiles may improve the discrimination of CF from other diabetic neuropathic stages and may support clinical assessment when conventional diagnostic methods remain unclear. However, prospective longitudinal studies are required to determine their value for risk prediction and disease progression. Full article

European armillarioid fungi (Armillaria and Desarmillaria) are white-rot fungi with soilborne pathogenic potential on woody hosts and clear relevance to forest pathology. This structured narrative review synthesizes the reported chemical composition of European armillarioid taxa across wild basidiocarps, infected host material, and laboratory-derived matrices, with emphasis on metabolites and biochemical traits that can be interpreted in a forestry context. The available evidence is taxonomically and methodologically uneven and is dominated by A. mellea and A. ostoyae, whereas infected host material remains less well characterized than culture-derived and basidiocarp-derived material. Protoilludene-derived aryl esters provide the clearest compound-resolved evidence for antifungal and phytotoxic activity across controlled assay systems, but these data derive from distinct experimental formats that are not directly comparable as a single potency scale, while lipid profiling offers the strongest support for chemistry-based diagnosis from infected material. Overall, current evidence supports forestry interpretation, diagnostic profiling, and selective screening of culture-derived metabolites, but practical application remains constrained by matrix dependence, taxonomic gaps, and the limited transferability of in vitro findings to forest conditions. Full article

Vertical transmission of plant pathogenic viruses is an important component of viral persistence, survival, and spread in agricultural production systems. This type of transmission is of considerable economic significance as it can cause major crop losses by serving as the initial focus of infection for future epidemics. Vertical transmission occurs when a virus is passed on to offspring either by direct invasion of the developing seed embryo from infected mother plants or through infected pollen grains after fertilization. We have recently demonstrated via high-throughput sequencing that mature seeds of the agriculturally important forage crop alfalfa (Medicago sativa L.) are associated with a broad range of viruses, some of which could potentially spread over long distances via seed. With the exception of the alfalfa mosaic virus, little is currently known about viral transmission through alfalfa pollen and its subsequent impact on the disease epidemiology of the crop. The objective of this study was to screen pollen from diverse alfalfa genotypes for pathogenic viruses and assess their risk of transmission. The pollen was collected from alfalfa genotypes selected for fungal disease resistance and agronomic performance in the USDA ARS pre-breeding program in Prosser, WA. Full article

Background: Among gynecological malignancies, ovarian cancer (OC) remains a leading cause of mortality worldwide, often characterized by the highest fatality-to-case ratio due to its asymptomatic progression and late-stage detection. Despite substantial investigation, the root cause of disease development and pathology remains unknown. Early detection is critical for improving OC prognosis. Unfortunately, because of the lack of identifiable symptoms in the early stages, the disease is frequently detected late. As a result, regular check-ups, being aware of risk factors, and paying attention to unusual symptoms can all help discover OC early. Apolipoproteins (APOs) and Annexins (ANXs) have recently been linked to OC. Aim: We conducted a cutting-edge bioinformatics investigation to find novel therapeutic targets and precise biomarkers linked to OC against APO and ANX. Methods: We started by compiling the ANX and APO families via HUGO Gene Nomenclature Committee (HGNC) homepage. Next, we accessed GEPIA2 to compare the relative messenger RNA (mRNA) expression levels of all ANX and APO family members across the cancer genome atlas (TCGA)-OC cohort and matched normal and GTEx data. Prognostic analysis of all significantly expressed ANXs and APOs was performed via Kaplan–Meier (KM) plotter. cBioPortal was used for mutational analysis of prognostic ANXs and APOs. Finally, we ran functional enrichment, molecular docking, and molecular dynamics (MD) simulation analyses. Results: Overall, the results suggest that ANXA2 and its related genetic changes represent potential focal points for precision oncology, offering a computational rationale for the development of target-driven therapeutic interventions in OC. Conclusions: Molecular docking and MD simulation analyses identified curcumin as a potential inhibitor of ANXA2, demonstrating stable binding affinity and structural conservation throughout the simulation period. These computational findings characterize curcumin as a promising candidate for targeting ANXA2 in OC, warranting further experimental validation to confirm its therapeutic efficacy. Full article

Background: Non-invasive approaches to brain tumour detection and diagnosis are limited by the absence of clinically validated circulating biomarkers. This study utilised a miniaturised tissue perfusion model to maintain human brain tumour tissue ex vivo with the aim of identifying tissue-derived proteins with potential biomarker utility. Methods: 55 tumour samples from 11 different brain tumours (glioblastoma n = 4, low-grade glioma n = 4, brain metastases n = 3) were micro-dissected and maintained ex vivo on a continuous-flow perfusion device for 168 h. Proteomic analysis of tumour effluent was performed by reversed-phase capillary liquid chromatography-mass spectrometry. Two candidate proteins—extracellular matrix protein 1 (ECM1) and cathepsin D—were quantified using ELISA. Results: All tumour subtypes retained tissue viability over 168 h of perfusion. Proteomic profiling identified 90 tissue-derived proteins in the tumour effluent. Many proteins corresponded to previously described cancer biomarkers such as glial fibrillary acidic protein (GFAP) while others, including Serpin A12 and collapsin response mediator protein-2 (CRMP2), had not yet been described in a brain tumour context. ELISA confirmed significantly higher ECM1 levels in high-grade glioma effluent compared with low-grade glioma (p = 0.0407), whereas cathepsin D levels did not differ significantly between tumour types. Conclusions: The ex vivo perfusion model effectively preserved primary and metastatic human brain tumour tissue and enabled direct characterisation of tumour-secreted proteins. The proteins identified here warrant further validation as tumour biomarkers in patient serum or cerebrospinal fluid. Full article

Background: Severe aortic stenosis (AS) increases left ventricular afterload and disrupts ventricular–vascular coupling. Transcatheter aortic valve replacement (TAVR) promptly relieves valvular obstruction, but its immediate effects on blood pressure-dependent arterial load and ventricular–vascular interactions are not fully clarified. Estimated pulse wave velocity (ePWV), derived from age and mean arterial pressure, is a convenient surrogate of global arterial load. The study aimed to assess ePWV before and after TAVR and its relationship with ventricular function and inflammatory biomarkers. Methods: In this retrospective observational study, 100 elderly patients with severe AS undergoing TAVR underwent detailed clinical, laboratory, and echocardiographic assessments before and after the procedure. Arterial stiffness was quantified using ePWV, while left ventricular geometry and systolic function were evaluated by standard echocardiography. Post-procedural reassessment was performed at hospital discharge (median 8 days after TAVR). Results: TAVR led to a modest but significant reduction in ePWV (from 12.79 ± 1.54 to 12.39 ± 1.54 m/s, p < 0.01) and improvement in left ventricular ejection fraction (LVEF) (from 44.89 ± 9.2% to 46.7 ± 7.95%, p < 0.01). Higher baseline ePWV correlated with unfavorable left ventricular remodeling and systolic dysfunction, and post-procedural ePWV remained linked to right ventricular performance. Before TAVR, ePWV and LVEF were both associated with inflammatory biomarkers, relationships that disappeared after intervention. Conclusions: Overall, ePWV functioned as an integrated measure of ventricular–vascular interaction and global hemodynamic load, though its interpretation post-TAVR requires caution due to direct blood pressure dependence and confounding by acute procedural inflammation. Full article

Background/Objectives: Schools, as structured social environments, are important settings for shaping lifelong eating habits, and teachers play a mediating role in nutrition education. This study aimed to assess nutrition knowledge among public school teachers, examine their perceptions of the school food environment, and identify factors associated with knowledge scores. Methods: A stratified cross-sectional survey was conducted among 736 teachers from 12 public schools during the fall of 2025. A structured questionnaire based on the Knowledge–Attitudes–Practice model was used to evaluate nutrition knowledge, teaching practices, perceived school food environment, and teachers’ observations of student food-related behaviors. Group differences were examined using t-tests and ANOVA, and multivariable logistic regression was applied to identify factors associated with low nutrition knowledge. Results: The mean knowledge score was 6.26 ± 2.64 out of 12, with 23.6% of teachers classified as having low knowledge, 59.9% satisfactory, and 16.4% good. Primary school teachers scored significantly higher than subject teachers (7.27 vs. 5.64; p < 0.001). Regular conduct of nutrition classes was associated with lower odds of low knowledge (adjusted OR 0.10, 95% CI 0.05–0.23, p < 0.001). A sensitivity analysis using continuous knowledge scores confirmed this pattern, with the absence of nutrition teaching predicting a 1.40-point reduction in scores (95% CI −1.86 to −0.94, p < 0.001). Teachers rated school meal quality moderately high, and frequently observed student refusal of canteen food and purchase of sweets or fast food outside school. Conclusions: This study found that public school teachers in Karaganda, Kazakhstan showed satisfactory overall nutrition knowledge, with gaps in applied understanding and common dietary myths. Primary teachers and those actively teaching nutrition had higher knowledge scores, indicating an association between pedagogical engagement and content knowledge. To help optimize the school food environment, educational policies would benefit from the development of cross-curricular instructional materials fitted specifically for secondary school subject teachers. Full article

Background/Objectives: Traditional histopathologic grading of renal cell carcinoma (RCC) is subjective, is poorly reproducible, and fails to predict responses to modern targeted agents or immunotherapies. In the era of precision oncology, molecular pathology offers objective tools for individualized management. We aimed to characterize genomic alterations in clear-cell RCC (ccRCC) with venous tumor thrombus and to develop pathology-driven panels for personalized prognostic stratification, with exploratory assessment of their potential to predict therapeutic response. Methods: Formalin-fixed paraffin-embedded pT1 ccRCC samples with and without thrombus underwent whole-exome sequencing. Distinct somatic mutations and copy number variations were incorporated into multigene panels. External assessment was performed in TCGA and PAWG cohorts, assessing survival outcomes and therapeutic biomarkers including homologous recombination deficiency (HRD), tumor mutational burden (TMB), and microsatellite instability (MSI). Results: Thrombus cases showed unique genomic heterogeneity compared with matched controls. Three multigene panels were constructed. Across external datasets, including a 354-patient TCGA-KIRC ccRCC cohort, the panels provided consistent molecular stratification signals for overall, disease-specific, and progression-free survival, complementing established pathological risk factors. They were significantly associated with established therapy-related genomic biomarkers, including HRD, TMB, and MSI, showing high sensitivity and negative predictive value in identifying patients unlikely to harbor these biomarker-positive profiles. These findings support the panels’ prognostic utility, with exploratory evidence for their potential in therapy response prediction. Conclusions: Small ccRCC with thrombus harbors distinct molecular pathological features. The proposed pathology-driven panels, compatible with FFPE tissue, represent pathology-compatible genomic tools that may support modern precision pathology by improving molecular risk stratification and informing exploratory therapeutic biomarker assessment. Full article

Tuft cells (TCs) are rare chemosensory epithelial cells that regulate mucosal homeostasis in multiple organs, but their role in salivary gland (SG) biology remains poorly defined. This study aimed to define TC structure in mice submandibular glands (SMGs), determine how TC loss affects gland organization and function, and evaluate whether TC abundance in human minor SGs is associated with Sjögren’s disease (SjD) features. Specifically, TC ultrastructure and ductal localization were characterized in female and male C57BL/6J mouse SMGs by transmission electron microscopy and immunostaining. Wild-type and C57BL/6J-Pou2f3^-/- (TC-deficient mouse strain) SMGs were analyzed by histology and bulk RNA-seq, and salivary function was assessed by saliva flow and proteomics. Human minor SG biopsies from SjD and non-Sjögren sicca (nSjD) patients were analyzed by immunostaining and Poisson regression. In mice SMGs, TCs showed conserved ultrastructural features and localization in both sexes. TC loss was associated with marked sex-biased transcriptome remodeling, morphological disruption, and altered saliva quantity and quality. In humans, TC counts differed between nSjD and SjD groups and were associated with salivary flow, serologic status, and histopathologic features. These findings support a role for TCs in SG epithelial integrity and suggest TC abundance as a candidate biomarker of SG dysfunction. Full article

Endothelial dysfunction, chronic inflammation, immune dysregulation, oxidative stress, mitochondrial dysfunction, and metabolic disturbances collectively contribute to cardiovascular diseases (CVDs) associated with blood stasis patterns. Xuefu Zhuyu Decoction (XFZYD) is widely used clinically for the management of CVDs. Based on serum-exposed prototype profiling in rats, two pharmacology-driven core component sets of XFZYD were defined as the core set for the promotion of blood circulation and the elimination of blood stasis (CPBEB; HSYA, GRo, FA, β-ECD, AMY, ALB, PF) and the core set for the regulation of qi and the relief of pain (CRQRP; LIQ, NR, NAR, ROF, HSD, NHP, LTG, NRG, ISL, FNT, NOB, PD, SSa). CPBEB primarily targets vascular pathology by regulating endothelial dysfunction with dyslipidemia-driven arterial lipid deposition. Mechanistically, CPBEB is associated with improved endothelial function, reduced plaque instability, attenuated chronic inflammation and oxidative stress, normalized lipid and bile acid metabolism, and decreased thrombosis. CRQRP primarily modulates vascular tone and systemic energy metabolism. These effects are linked to enhanced AMPK/SIRT1-driven antioxidant defenses and mitochondrial homeostasis, increased NO/cGMP signaling, coordinated crosstalk among the TLR4/NF-κB, JAK/STAT, NLRP3, and PPAR pathways, and remodeling of the gut microbiota–immune network. In summary, this review integrates modern analytical approaches with network pharmacology and the literature evidence to clarify the material basis underlying XFZYD’s therapeutic effects in CVDs, thereby supporting the modernization and internationalization of traditional Chinese medicine. Full article

Background and Objectives: Dentofacial asymmetries are common in patients with malocclusions, while mild body postural asymmetries are frequently reported in otherwise healthy individuals. However, their interrelationship remains insufficiently investigated in adults without diagnosed spinal disorders. This study aimed to evaluate the association between dentofacial and body postural asymmetries in adults with malocclusions. Materials and Methods: A clinical cross-sectional observational study was conducted on 102 adults (18–45 years) with malocclusions and no spinal pathology. Standardized clinical morphometric examinations assessed dentofacial asymmetries (horizontal and vertical planes), dental parameters (dental midlines deviation and occlusal plane inclination), and body postural asymmetries (head, shoulder, trunk, pelvic, and lower limb alignment). Asymmetries were recorded using predefined clinical thresholds. Statistical analyses included the Wilcoxon signed-rank test, Pearson chi-square test, and Spearman’s rank correlation coefficient. Results: Dentofacial asymmetries were identified in both planes and occurred more frequently on the left side. Horizontal facial asymmetries were most common at the cheek (74.5%), nostril (66.7%), and mandibular angle levels (57.9%), and were influenced by sex, age, facial growth pattern, and facial profile (p ≤ 0.05). Mandibular dental midline asymmetry was present in 55.8% of patients. Body postural asymmetries were also frequent, particularly unilateral (60.8%) or anterior (55.9%) head inclination and shoulder asymmetries (54.9%), with a predominance on the left side and associations with age, body mass index, and postural attitude (p ≤ 0.05). Correlations were identified among facial asymmetries and among body postural asymmetries (p ≤ 0.01), indicating a bilateral distribution pattern. Additionally, right-sided facial asymmetries showed significant positive associations with right-sided body postural asymmetries (ρ = 0.197–0.229; p ≤ 0.05). Conclusions: Dentofacial and body postural asymmetries have been identified in adults with malocclusions and presented side-specific associations regarding the patterns of asymmetry. Full article

Background: Predicting genomic alterations from routine hematoxylin and eosin (H&E) whole-slide images (WSIs) may help triage molecular testing. Methods: We retrospectively enrolled 437 patients at Osaka Metropolitan University Hospital across 26 cancers, matched with clinical gene-panel data. We curated 1023 binary endpoints across SNV, CNV, and SV categories. We extracted slide embeddings from five pathology foundation models (Prism, GigaPath, Feather, Chief, and Titan) using a unified feature extraction pipeline and benchmarked them using a lightweight downstream Multi-Layer Perceptron (MLP) classifier. Using the best-performing patch feature system, we trained a multi-instance learning model to assess incremental benefit. Results: Titan achieved the highest and most stable transfer performance, with a median endpoint-wise Area Under the Receiver Operating Characteristic curve (AUROC) of 0.77 in the slide benchmarking; at the patch-level, prediction of APC_SNV reached an AUROC of 0.916, and prediction of KRAS_SNV reached an AUROC of 0.811 on the held-out test set. Conclusions: In a heterogeneous clinical gene-panel setting, pathology foundation models can provide strong baseline genomic-prediction signals without additional fine-tuning. We propose a practical, deployment-oriented two-stage workflow: rapid slide-embedding screening to prioritize robust representations and candidate endpoints, followed by patch-level training for high-value tasks where additional performance gains and interpretable regions are clinically worthwhile. Full article