Medicine and Pharmacology

Abstract: Tumor budding is a prognostic marker associated with epithelial-mesenchymal transition and tumor microenvironment in various tumors. The programmed cell death protein 1 (PD-1) pathway functions as a crucial immune checkpoint that regulates lymphocyte activity within the tumor microenvironment. The expression patterns of PD-1 ligands may significantly influence the potential success of therapeutic blockade of this pathway. This study aims to compare the expression of programmed cell death ligand 1 (PD-L1) with clinical-pathological features using different scoring systems and to evaluate its prognostic value in patients with non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors. A total of 361 cases of NSCLC were included in the study. A statistically significant association was found between patients with a negative combined positive score (CPS) or immune cell score (ICS) and those with an advanced clinical stage and poor tumor differentiation. A high tumor budding score correlated with positive PD-L1 expression across all three scoring systems. Overall survey (OS) revealed that PD-L1 positivity with ICS is a significant independent factor associated with a favorable prognosis. Kaplan-Meier analysis showed that CPS+ or ICS+ cases had significantly longer OS compared to CPS- or ICS- patients. In the immune checkpoint inhibitor group, no significant OS difference was found between positive and negative cases in TPS, CPS, and ICS. In conclusion, our study demonstrated a significant correlation between high tumor budding in NSCLC and PD-L1 expression in tumor and immune cells. We showed that CPS or ICS is a better predictor of prognosis in NSCLC.

Abstract: Chronic exposure to ethanol (ETH) and e-cigarette vapor activates the P2X7 receptor (P2X7R) in human pulmonary alveolar epithelial cells (hPAEpiCs), promoting extracellular vesicle (EV) release. Proteomic analysis of this EVs identified enrichment of urokinase-type plasminogen activator (uPA), a key extracellular matrix remodeling enzyme. Treatment of hPAEpiCs with ETH, acetaldehyde (ALD), and e-cigarette vapor, with or without nicotine, induced NLRP3 inflammasome activation, IL-1β secretion, and increased incorporation of uPA into EVs. Exposure of human brain microvascular endothelial cells (hBMVECs) to this EVs enhanced plasmin conversion, activated MMP-9 and TGF-β1, resulting in reduced transendothelial electrical resistance (TEER), indicating blood-brain barrier (BBB) dysfunction. Pharmacological inhibition of P2X7R using ‘A804598’ significantly reduced EV-associated uPA release and attenuated endothelial injury. In vivo, chronic ETH vapor exposure increased expression of uPA, uPAR, MMP-2, and MMP-9 while decreasing SERPIN1 and TIMP1 in murine brain microvessels, changes associated with BBB disruption and IgG leakage into brain parenchyma. P2X7R inhibition normalized these effects. Collectively, these findings identify EV-associated uPA as a mediator of lung-to-brain signaling and BBB injury following ETH and e-cigarette exposure.

Abstract: Bladder cancer surveillance remains challenging due to the limited sensitivity of con-ventional diagnostic methods for detecting biologically aggressive disease, particularly in low-grade urothelial carcinomas (LGUC). This study evaluated the diagnostic per-formance and biological significance of the urinary DNA methylation assay Bladder EpiCheck™ in patients undergoing histopathological evaluation for suspected urothe-lial carcinoma. A retrospective cohort of 134 patients with paired histopathology, urine cytology and Bladder EpiCheck™ results was analyzed. Bladder EpiCheck™ showed excellent performance for high-grade urothelial carcinoma (HGUC) detection, with 91.5% sensitivity, 81.6% specificity and an AUC of 0.876. Methylation scores in-creased with tumor grade and identified two molecular subgroups within LGUC. Methylation-positive LGUCs displayed HGUC-like epigenetic profiles, increased pro-liferation, more frequent MTAP loss and higher recurrence/progression risk, whereas methylation-negative LGUCs resembled benign urothelium and had more favorable outcomes. These findings indicate that urinary DNA methylation provides clinically relevant biological information beyond histological grading and may improve LGUC risk stratification and personalized surveillance.

Abstract: Purpose: Some sudden infant death syndrome (SIDS) and sudden unexpected infant death (SUID) cases remain unexplained after medicolegal investigation. The Metabolic Vulnerability Index (MVI) was proposed as a specimen-aware framework in which the total score summarizes measured multidomain findings and archetype classification organizes patterns of domain involvement. This review proposes a post-MVI framework for organizing candidate non-metabolic vulnerabilities after Stage 1 MVI classification. Methods: Literature on molecular autopsy, cardiac conduction-system pathology, concealed cardiomyopathy, epilepsy/SUDEP, autonomic or respiratory developmental disorders, and postmortem specimen constraints was synthesized into a staged eligibility framework, tiered escalation pathway, and pilot-implementation elements. Results: The framework distinguishes MVI-positive, confirmed low-MVI, and specimen-limited evaluations before routing eligible cases into a tiered pathway. Tier 1 prioritizes targeted cardiac channelopathy testing because of DNA robustness, family actionability, and relevance to autopsy-negative sudden death. Conditional Tier 2 addresses cardiac, neurologic, autonomic/respiratory-control, and structural mechanisms when supported by history, autopsy context, specimen readiness, or predefined triggers. Tier 3 is reserved for unresolved cases when research-level genomic expansion or specialized assessment is feasible and authorized. Stage 1 classification preserves indeterminate or not-evaluable results rather than treating missing findings as normal and separates exposure-context abstraction from score-eligible analytic findings and research-only measurements. Conclusions: The staged MVI/post-MVI model organizes unexplained SIDS/SUID cases into specimen-aware research profiles that integrate abnormal domain patterns, domain evaluability, exposure context, and post-MVI findings. These profiles can generate testable hypotheses about the biologic or physiologic pathways underlying candidate vulnerability. Pilot studies can establish the feasibility and reproducibility of domain collection; subsequent adequately powered studies can determine whether recurring patterns distinguish biologically meaningful subgroups within heterogeneous SIDS/SUID. The resulting classifications remain research constructs and do not constitute diagnoses or individual cause-of-death determinations.

Abstract: As a common benign bone tumor, the pathological diagnosis of chondrogenic tumors needs to accurately analyze the calcification pattern of the cartilage matrix, key spatial topology information, and other important indicators. However, chondrogenic tumors is a rare disease, doctors lack sufficient reference data and experience in diagnosis. At the same time, complex spatial structural information also increases the difficulty of diagnosis, which leads to inconsistencies in the diagnosis’s results of different doctors. In contrast, with the development of technology, artificial intelligence (AI) with its fast, accurate, and robust characteristics can effectively improve the efficiency and accuracy of diagnosis. However, the application of AI in this field is still unrecognized, so it is urgent to develop a model that can help doctors in diagnosis to improve accuracy and efficiency. In this study, we propose XChondNet, an explainable spatial-context-aware synergistic deep feature fusion model for WSI-based chondrogenic tumor classification. The model proposes a fusion mechanism of pathological and positional features so that the model can effectively perceive spatial structural information and a parallel classifier mechanism based on potential coding, which can effectively solve the problem of class imbalance in chondrogenic tumors data. We evaluated the XChondNet model on our chondrogenic tumors dataset and the experimental results verified its effectiveness in the classification of the chondrogenic tumors subtype. In the test phase, compared to the other advanced models, the XChondNet model achieved state-of-the-art accuracy (ACC), area under the curve (AUC), F1 score and Recall rate. And the XChondNet model explicitly reflects the consistency with pathologists’ concerns, which improves the interpretability of AI.

Abstract: Background: Esophageal squamous cell carcinoma (ESCC) accounts for over 80% of esophageal cancer cases globally. Accurate preoperative prediction of lymph node metastasis (LNM) is critical for treatment decision-making, yet current clinical staging frequently underestimates nodal disease burden. Pathomics — the high-throughput quantitative analysis of histopathological images — offers an opportunity to capture tumor microenvironmental patterns invisible to routine microscopy. We aimed to develop and validate a pathomics-clinical integrated model for predicting LNM in resectable ESCC. Methods: A total of 477 consecutive ESCC patients who underwent radical esophagectomy were retrospectively enrolled. After excluding 61 patients with missing data, 416 patients were included in the final analysis (LNM-positive: 217, 52.2%). H&E-stained whole-slide images were acquired using a KFBIO KF-PRO-400-HI scanner at 20× magnification (pixel size 0.5 μm/px). Nuclear morphological features were extracted via QuPath hematoxylin optical density (OD)-based detection, yielding 17 pathomic features selected by LASSO-logistic regression. A combined clinical-pathomics nomogram was constructed and internally validated via 10-fold cross-validation (out-of-fold, OOF), with apparent (resubstitution) performance reported alongside to quantify optimism. Secondary analyses examined survival stratification within node-negative (pN0) patients and across whole-cohort risk groups. The final nomogram was additionally applied, without refitting, to an independent external cohort of 101 ESCC patients from a separate institution to assess transportability. Results: The combined nomogram achieved an out-of-fold (cross-validated) AUC of 0.821 (95% CI 0.779–0.858), compared with 0.791 for the clinical model and 0.762 for the pathomics-alone model. In multivariable analysis, the pathomics score remained an independent predictor of LNM (adjusted OR 2.47, 95% CI 1.80–3.39; P < 0.001), with significant incremental value over clinical variables alone (likelihood-ratio χ² = 35.5, P = 2.56e-09). Risk stratification by nomogram score tertiles showed a clear gradient of LNM prevalence: 15.8% (low-risk), 55.1% (intermediate-risk), and 85.6% (high-risk) (Cochran-Armitage trend P < 0.001). In the pN0 subgroup (n = 199), patients with high pathomics risk had significantly worse overall survival (3-year OS: 77.2% vs. 94.3%; log-rank P = 0.005) and progression-free survival (3-year PFS: 75.5% vs. 88.9%; log-rank P = 0.007). Externally, in an independent cohort of 101 patients from a separate institution (LNM+ = 25, 24.8%), the combined model achieved an AUC of 0.774 (95% CI 0.663–0.870), the clinical model 0.801, and the pathomics-alone model 0.685, confirming partial transportability of the clinical component while the pathomics signature showed limited generalizability across centers. Conclusions: The integration of quantitative pathomic features with conventional clinicopathological variables provides robust predictive power for LNM in ESCC and identifies a subset of node-negative patients at elevated risk of poor outcomes. These findings support the potential utility of pathomics as a complementary tool for preoperative risk stratification in ESCC. External validation in an independent cohort confirmed partial transportability of the clinical component but highlighted limited generalizability of the pathomics signature, underscoring the need for multi-center harmonization before clinical deployment.

Abstract: Background: Programmed death-ligand 1 (PD-L1) contributes to immune escape and is a therapeutically relevant target in aggressive B-cell lymphomas. However, its distribution and clinicopathologic significance across primary mediastinal aggressive large B-cell lymphomas remain incompletely defined. Methods: We retrospectively reviewed 53 cases diagnosed between 2010 and 2025, including diffuse large B-cell lymphoma (DLBCL), primary mediastinal large B-cell lymphoma (PMBCL), mediastinal gray zone lymphoma (MGZL), and T-cell/histiocyte-rich large B-cell lymphoma (THRLBCL). PD-L1 immunohistochemistry (clone 22C3) was performed on formalin-fixed, paraffin-embedded tissue. Tumor cells and intratumoral inflammatory cells were assessed together. Membranous staining was scored as 0, 1+, 2+, or 3+, and the proportion of cells in each category was recorded. Results: PD-L1 staining was present in 40/53 cases (75.5%). Mean PD-L1-positive percentages were similar in DLBCL, PMBCL, and MGZL (38%, 36%, and 44%, respectively; p = 0.939), and no tested cutoff separated the diagnostic groups. In contrast, MGZL showed a higher mean 3+ component (31.11%) and more cases with 3+ staining in at least 50% of evaluated cells (44% vs. 4.3% in DLBCL and 4.7% in PMBCL; p = 0.007). PD-L1 extent was not associated with age, tumor size, LDH, beta-2 microglobulin, Ann Arbor stage, SUVmax, treatment response, or overall survival. Conclusions: PD-L1 expression is common in mediastinal aggressive large B-cell lymphomas. MGZL is distinguished by a strong-staining pattern, whereas combined PD-L1 extent was not prognostic in this cohort.

Abstract: Digital pathology is transforming diagnostic workflows through whole-slide imaging, digital archiving, AI-assisted analysis, and remote consultation. Existing literature only partly addresses access to pathology records and cross-institutional sharing within a holistic governance framework. This study proposes a Self-Sovereign Identity (SSI)-based conceptual governance model that treats digital pathology archives as long-term clinical memory with patient-managed access. Eighty publicly available complaints from the Şikayetvar platform were examined using thematic content analysis. National Health Service (NHS) Written Complaints data and academic, regulatory, and technical materials were examined as complementary evidence. The main problems were missing or inaccessible pathology results (66.3%) and records remaining unavailable despite notifications that results were ready (55.0%). Synthesizing these findings with complementary sources identified 11 governance requirements for secure, interoperable, auditable, and patient-centered digital pathology management. Developed through Design Science Research, the model integrates verifiable credentials, a patient digital wallet, purpose- and time-limited authorization, dynamic consent and access revocation, secure off-chain storage, and blockchain-based integrity and auditing. A scenario-based comparison of traditional, centralized digital, and SSI-based workflows indicates the model’s potential to address access, verifiable sharing, selective disclosure, portability, and auditability holistically. Establishing traceability among patient complaints, governance requirements, and model components extends digital pathology research beyond diagnostic technologies.

Abstract: Mosaic RASopathies are rare conditions caused by postzygotic mutations in RAS/MAPK signaling pathway genes arising during embryonic development. They are mainly characterized by a combination of epidermal nevi with extra-cutaneous manifestations. Choosing between different entities can be challenging because of their substantial clinical and genetic overlap. Our aim was to demonstrate the importance of genetic background in accurate diagnosis and further prognostic assessment of disease progression by analyzing genotype–phenotype correlations within this clinical group. The study enrolled 18 patients (10 males, 8 females) aged from 1 month to 21 years with congenital nevi of craniofacial localization and variable extracutaneous abnormalities. DNA isolated from peripheral blood lymphocytes and affected tissues (nevus, tumor) was analyzed using a targeted next-generation sequencing (NGS) panel covering RAS/MAPK pathway genes, followed by Sanger sequencing. Eight patients were diagnosed with Schimmelpenning–Feuerstein–Mims syndrome caused by postzygotic mutations in the HRAS (n=4) or KRAS (n=4) genes. Seven patients were clinically diagnosed with phacomatosis pigmentokeratotica, and mosaic mutations were identified in the HRAS (n=4), KRAS (n=2), or BRAF (n=1) genes. Two patients with similar phenotypes carried somatic mutations in the PIK3CA and FGFR2 genes, respectively, and in one case no mutation was detected in the affected skin. We compared the spectrum of extracutaneous anomalies depending on the mutation in a specific gene by combining our data with data from other clinical cases of mosaic RASopathies described in the literature. In summary, 53 cases with HRAS postzygotic mutations and 41 cases with KRAS mutations were involved in the analysis of genotype–phenotype correlations. Hypophosphatemic rickets and malignant tumors were associated with mutations in the HRAS gene, while ophthalmological, neurological, cardiovascular, and renal disorders were more common in KRAS-mutant patients. Thus, genetic testing of affected tissue is essential for accurate diagnosis of these mosaic disorders. Taking into account clinical manifestations and the identified genotype–phenotype correlations may further assist in prognosis and patient management.

Abstract: A model can discriminate well in the cohort that trained it and still fail where some of its inputs are never measured. That failure is rarely separated into missing measurements and a different population. We pooled ten cycles of the National Health and Nutrition Examination Survey (NHANES, 1999–2018; n = 24,027 adults) linked to its Mortality File, built a 17-feature inflammation–anemia–cardiorenal panel from uremic pathophysiology, and benchmarked five machine-learning models on a 10-year composite of cardiovascular, kidney, or diabetes-related death or dialysis. XGBoost discriminated best internally (AUROC 0.909; 95 % CI 0.900–0.918). Three nested feature sets were then carried to the UCI Chronic Kidney Disease cohort (n = 400). The full panel transported poorly (external AUROC 0.598; gap 0.311). Dropping transferrin saturation, absent from UCI, cost nothing internally and raised external AUROC to 0.712. Restricting to the six features UCI genuinely supplies cost more internally (0.872) but narrowed the gap furthest (external 0.796; gap 0.077). Because that panel requests nothing UCI lacks, roughly three-quarters of the original gap tracks feature availability and one-quarter reflects case-mix and outcome definition. Feature availability in the deployment setting belongs in model selection, not the limitations paragraph.

Abstract: Background: Unexplained cytopenia presents a common diagnostic challenge in clinical haematology, reflecting either benign conditions or early manifestations of myeloid neoplasms. Although genetic testing is increasingly becoming the standard reference for diagnosis, an initial cellular assessment of the bone marrow is often the first decisive diagnostic step. Methods: In this retrospective cross-sectional study, we analysed 557 patients with cytopenia who underwent initial diagnostic evaluation. The diagnostic performance of the initial cellular assessment defined in this study as morphological and ancillary studies excluding genetic analyses was evaluated against a genetically supported reference diagnosis established by a multidisciplinary haematology tumour board. We examined the extent to which the initial cellular diagnosis can reliably detect or rule out myeloid neoplasia. The analysis was performed for the entire cohort and stratified by specific myeloid neoplasm entities. Results: In the overall cohort, the initial cellular assessment showed high diagnostic performance (sensitivity, 0.864; specificity, 0.934). The misclassification analysis revealed more false negatives than false positives (31 vs. 21), with no significant asymmetry. In the entity-specific analysis, a sensitivity of 1.0 was observed for acute myeloid leukaemia, with a reduced specificity (0.400). Myelodysplastic neoplasms exhibited a higher rate of false-negative findings, whereas myelodysplastic/myeloproliferative overlap neoplasms were characterised by very high sensitivity (0.982) and limited specificity (0.632). A comparable diagnostic pattern was observed for myelofibrotic neoplasms. Conclusion: Initial cellular assessment demonstrates high overall diagnostic performance in cytopenia but shows important variation across individual entities. These findings highlight the need to consider entity-specific limitations and the role of genetic confirmation in the diagnostic work-up of specific entities.

Abstract: Background/Objectives: Recent World Health Organization (WHO) classifications of gliomas emphasize integrated molecular diagnosis, improving diagnostic precision but increasing the need for immunohistochemistry and genetic testing. As a result, increasing costs and a decrease in diagnostic rates in developing countries have become issues. Consequently, there is growing interest in the development of image-based approaches that can extract biological information from routine histopathological slides. In this study, we investigated whether nuclear morphometric features derived from hematoxylin–eosin (HE) images of tumors can predict the molecular subtype, pathological tumor type, WHO grade, and prognosis of gliomas using machine learning (ML). Methods: A total of 157 specimens were analyzed. All HE slides were scanned and 15 regions of interest (ROIs) were chosen. About 960 features of cell nuclei in the ROIs were analyzed using support vector machine and random forest models to predict molecular alterations (isocitrate dehydrogenase 1 [IDH-1], alpha thalassemia/mental retardation syndrome X-linked, p53, and the 1p/19q codeletion), pathological diagnosis, WHO grade, and prognosis. Results: The tumor type classification and prognosis were predicted with 96% and 94% respectively. IDH-1 mutation status was predicted with 95% accuracy. Feature importance analysis indicated that parameters associated with nuclear shape, particularly orientation, eccentricity, and solidity, contributed substantially to the prediction models. Conclusions: Quantitative nuclear morphometric analysis of HE images using ML enables accurate prediction of the molecular characteristics and prognosis of gliomas. This approach may complement conventional diagnostics, and demonstrates the potential to reduce diagnostic costs. Furthermore, with the incorporation of additional genetic analyses, this approach may lead to personalized medicine in the future.

Abstract: Background: Reliable molecular biomarkers that complement cytomorphological assessment for early identification of high-risk oral epithelial lesions remain limited. Although p53 immunocytochemistry (ICC) is widely used as a surrogate marker of TP53 abnormalities, p53 protein expression does not always reflect TP53 transcriptional activity. This study investigated the relationships between p53 ICC, TP53 mRNA expression, and p53 immunohistochemical (IHC) staining patterns in oral high-grade squamous intraepithelial lesions (OHSIL) and oral squamous cell carcinoma (SCC). Methods: A total of 504 liquid-based cytology (LBC) specimens were classified according to the Bethesda System for Reporting Oral Cytology as NILM (n = 394), OLSIL (n = 72), OHSIL (n = 18), and SCC (n = 20). p53 ICC was performed on cytological specimens, TP53 mRNA expression was quantified by quantitative real-time PCR using residual LBC samples, and corresponding biopsy specimens were evaluated for p53 IHC staining patterns. Associations between p53 ICC and clinicopathological variables were also analyzed. Results: The p53 labeling index significantly increased with cytological severity, and TP53 mRNA expression was significantly higher in OHSIL and SCC than in NILM and OLSIL. In OHSIL and SCC, lesions showing a null-type p53 IHC staining pattern showed significantly lower TP53 mRNA expression than lesions with non-null staining patterns. No significant associations between p53 ICC and clinicopathological variables were observed in OHSIL. In contrast, p53 ICC positivity in SCC was significantly associated with the tumor size, depth of invasion, and p53 IHC staining patterns. Conclusions: Negative p53 ICC may result from either sampling limitations or null-type TP53 expression. TP53 mRNA analysis provides complementary molecular information that distinguishes false-negative p53 ICC caused by sampling limitations from biologically reduced TP53 expression. Integrating cytomorphology, p53 ICC, histopathology, and TP53 mRNA analysis may improve diagnostic accuracy and risk stratification of high-risk oral epithelial lesions.

Abstract: Aging and glycative stress are major contributors to skeletal fragility, yet they affect bone through partially overlapping and distinct mechanisms. Aging is characterized by progressive deterioration of bone mass, microarchitecture, and remodeling balance, driven by hallmarks such as cellular senescence, mitochondrial dysfunction, oxidative stress, chronic low-grade inflammation, impaired autophagy, and altered intercellular communication. These processes disrupt the function of osteocytes, osteoblasts, and osteoclasts, leading to reduced bone formation, increased bone resorption, and impaired adaptation to mechanical loading. In parallel, glycative stress results from the accumulation of advanced glycation end-products (AGEs) and advanced glycoxidation end-products, which accumulate during aging and are accelerated in metabolic disorders such as type 2 diabetes mellitus. Unlike aging, glycative stress predominantly compromises bone quality rather than bone mass by altering collagen cross-linking, matrix mechanics, and cellular signaling through activation of the AGE–RAGE axis. Both conditions converge on shared pathways involving oxidative stress, inflammation, mitochondrial dysfunction, autophagy impairment, senescence, and defective mechanotransduction, ultimately reducing osteocyte viability and disrupting bone remodeling. Glycative stress additionally impairs bone mechanosensitivity by modifying extracellular matrix properties and altering key signaling networks, including Wnt/β-catenin, connexin 43, and PTH1R-dependent pathways. Emerging evidence identifies NLRP3 inflammasome activation, ferroptosis, and metabolic reprogramming as important downstream mediators linking aging, inflammation, and glycation-induced skeletal deterioration. Understanding the convergent and divergent mechanisms underlying bone aging and glycative stress may facilitate the development of targeted therapeutic strategies combining anti-resorptive, anabolic, senolytic, antiglycative, and mechanoprotective approaches to reduce fracture risk and preserve skeletal health.

Abstract: Background. Routine laboratory tests provide multidimensional information about systemic physiology, but their interpretation is usually analyte-specific and may not capture cumulative abnormality burden across biological systems. A laboratory-derived Restorative Index (RI) framework has been formulated to transform normalized laboratory deviations into a bounded 0–100 score, where higher values indicate lower laboratory abnormality burden. This study applied the global and domain-specific RI framework to de-identified longitudinal clinical laboratory data and evaluated its computability, longitudinal behavior, domain decomposition, and exploratory predictability. Methods. This retrospective application study analyzed de-identified longitudinal laboratory and treatment/dose data. Eligible numeric laboratory values were transformed into normalized abnormality distances relative to reference intervals. Subject-date panels with at least five RI-eligible analytes were used to compute global RI. Domain-specific RIs were computed for predefined biological domains. Subject-date panels were classified as baseline, intermediate, follow-up, or single-record observations. Exploratory predictive models included regularized linear models, Bayesian regression, tree-based ensemble models, boosting models, support-vector regression, neural networks, and domain-to-global models. Cross-validation used subject-level grouping where repeated observations were present. Results. The RI framework was applied to 2,001 eligible subject-date panels. Baseline RI was available in 902 subjects, follow-up RI in 551 subjects, and intermediate RI in 306 subjects, contributing 548 intermediate RI rows. Median global RI increased from 59.66 at baseline to 61.72 at follow-up, with a median change of +2.52 and mean change of +3.28 RI points. Using a 5-point threshold, 227 subjects improved, 167 remained stable, and 157 declined. Global RI prediction was feasible but modest; the strongest global model predicted final RI from the latest known pre-follow-up RI using ExtraTrees, with cross-validated R²=0.417, MAE = 12.24, and RMSE = 15.23. Domain-specific prediction was stronger for renal/uric intermediate RI (R²=0.598) and hematology/CBC intermediate RI (R²=0.570). Same-date global RI could be partially estimated from domain-specific RI features, with the best model achieving R²=0.639, whereas future final global RI prediction from domain RI dynamics was weaker (R²=0.237). Conclusion. The RI framework was applicable to heterogeneous longitudinal laboratory data and provided a computable, trackable, and biologically decomposable summary of laboratory abnormality burden. Domain-specific RI improved interpretability and selected-domain predictability. These findings support RI as a retrospective laboratory-informatics framework, but RI changes should not be interpreted as treatment efficacy and require prospective external validation against clinical outcomes.

Abstract: Background Bloodstream infection (BSI) complicates 20–50% of ICU sepsis cases and independently increases mortality. Conventional biomarkers — procalcitonin and C-reactive protein — lose discriminative power amid the high background inflammation of critical illness. Cell population data (CPD) from automated hematology analyzers offer morphological neutrophil metrics at no added cost. We evaluated NE-WY (neutrophil side scatter width) and NE-SFL (neutrophil fluorescence intensity) for confirmed BSI identification in ICU patients. Methods This STARD 2015-compliant prospective study enrolled 72 adult ICU patients at Dr. Soetomo General Academic Hospital, Surabaya. The primary analysis compared G1 (Sepsis-3 criteria, NHSN LCBI-confirmed BSI; n=26) against G2 (SOFA < 2, negative culture; n=21). NE-WY and NE-SFL were measured on Sysmex XN-3000; MDW on Beckman Coulter DxH 900. ROC analysis with DeLong AUC comparison assessed diagnostic accuracy. A pre-specified sensitivity analysis included G1 versus G2 plus 25 suspected-sepsis patients with non-diagnostic cultures (n=72). Results NE-WY was higher in G1 (median 789.5 [IQR 758.8–867.3]) than G2 (686.0 [IQR 650.0–747.0]; p=0.001, r=0.47). At cutoff 766, NE-WY AUC was 0.775 (95%CI 0.632–0.917): sensitivity 73.1% (95%CI 53.9–86.3%), specificity 85.7% (95%CI 65.3–95.0%), PPV 86.4%, NPV 72.0%, and accuracy 78.7%. NE-WY outperformed MDW (AUC 0.533; DeLong p=0.047). In the sensitivity analysis (n=72), NE-WY AUC fell to 0.620 (95%CI 0.485–0.756), indicating that performance depends on comparator composition. Conclusions NE-WY identifies confirmed BSI against non-septic ICU controls with AUC=0.775 and specificity 85.7%, outperforming MDW. Performance falls to AUC=0.620 against a mixed comparator that includes suspected-sepsis patients. Future studies should compare culture-confirmed BSI against culture-negative Sepsis-3 patients to isolate the bacteraemia-specific signal.

Abstract: Endometrial cancer is one of the most rapidly increasing gynaecological malignancies worldwide. The clinically adapted molecular classification of endometrial carcinoma, derived from The Cancer Genome Atlas, comprises four major subtypes: POLE-mutated, mismatch repair deficiency, p53-abnormal expression, and no specific molecular profile. Its clinical implementation has improved prognostic stratification, risk assessment, and treatment decision-making in patients with endometrial carcinoma. However, current workflows rely on immunohistochemistry and targeted sequencing, which increase costs, turnaround times, and infrastructure requirements, thereby limiting their universal adoption in routine practice. In this context, recent advances in artificial intelligence (AI), particularly deep learning models capable of predicting molecular features directly from H&E-stained whole-slide images, have emerged as promising tools for precision oncology. In addition to reproducing established molecular classification, these approaches may reveal previously unrecognised biomarker-defined histologic patterns that are difficult to detect using conventional methods. This article synthesises the current evidence on AI-based molecular classification in endometrial carcinoma from a pathologist-centred perspective, with an emphasis on the biological rationale, methodological limitations, and future directions for clinical translation.

Abstract: Diffuse large B-cell lymphoma (DLBCL) is an aggressive lymphoma characterized by a diffuse proliferation of large neoplastic B lymphocytes. DLBCL is histologically and clinically heterogeneous. Skeletonization reduces binary objects to 1-pixel wide curved line representations without changing the essential structure of the image. This is useful for feature extraction and represents an object’s topology. This study used the skeletonization function to analyze the topological heterogeneity of DLBCL in a series of 153 patients, including 110 patients with DLBCL and 43 patients with reactive lymphoid tissue. In comparison with reactive tissue, DLBCL was characterized by lower skeletonization: 5665.20 ± 1012.82 vs 6341.16 ± 548.23, respectively (P < 0.001). Within the DLBCL diagnostic category, high skeletonization correlated with poor overall survival (hazard risk = 2.5, P = 0.003). High skeletonization also correlated with higher EBER positivity and lower CD5, E2F1, BCL2, ISY1, and TNFAIP8 protein levels (all P values < 0.05). Gene expression was available in a subset of 30 cases, including 18 cases with high skeletonization and 12 cases with low skeletonization. The analysis showed that 166 immuno-oncology genes were upregulated in the high group and only 8 genes were upregulated in the low skeletonization group. In conclusion, DLBCL is characterized by lower skeletonization than reactive lymphoid tissue. In DLBCL, high skeletonization is associated with poor prognosis and enrichment of immuno-oncology markers.

Abstract: The analysis of non-blood biological fluids, including cerebrospinal fluid (CSF), serous effusions, and synovial fluid, plays a central role in laboratory medicine by providing essential diagnostic and prognostic information for neurological, infectious, inflammatory, and neoplastic diseases. However, the interpretation of these specimens remains challenging because it requires the integration of heterogeneous biochemical, cytological, microbiological, molecular, and clinical data, often in the absence of standardized analytical workflows. Artificial intelligence (AI), particularly Machine Learning (ML) and Deep Learning (DL), is emerging as a powerful approach for extracting clinically relevant information from complex multidimensional datasets beyond the capabilities of conventional analytical methods. AI-driven Clinical Decision Support Systems (CDSS) can integrate laboratory findings with clinical, demographic, imaging, and multi-omics data, supporting diagnostic interpretation, patient stratification, and personalized clinical decision-making. At the same time, the convergence of AI with proteomics, metabolomics, metagenomics, and other omics technologies is accelerating biomarker discovery and advancing precision laboratory medicine. Current evidence indicates different levels of maturity across biological fluids. AI-assisted interpretation of CSF biomarkers and digital cytology of serous effusions currently show the strongest clinical evidence, whereas applications involving synovial fluid and integrated multi-omics remain largely exploratory. Although important technical, methodological, and regulatory challenges still limit widespread clinical implementation, AI has the potential to improve diagnostic accuracy, reduce interpretative variability, and support more integrated diagnostic workflows. This mini-review summarizes current and emerging AI applications in non-blood biological fluid analysis, with particular emphasis on biomarker discovery, CDSS, multi-omics integration, current evidence, existing limitations, and future perspectives for precision laboratory medicine.

Abstract: Background. Asthma pathophysiology spans immune mechanisms, epithelial biology, environmental exposures, microbiome, metabolism, neuroimmune circuits, and comorbidities, but the recent review literature is highly compartmentalized by domain. A systems-level representation of the current consensus on human asthma biology is lacking. Methods. A scoping review was conducted following PRISMA-ScR guidelines. Five databases (PubMed/MEDLINE, Scopus, Web of Science, Embase, Cochrane) were searched with seventeen thematic queries for reviews and systematic reviews published 2021–2026. After screening and full-text assessment, 251 human-evidence sources were included. Entities (nodes) and mechanistic connections (edges) were extracted, normalized, and classified by role (pathogenic, protective, mixed, therapeutic), biological level, and asthma subtype. Nodes were filtered by a pre-specified priority classification; the final network included Priority A nodes reported in ≥3 independent studies and all edges connecting included nodes reported in at least one study. Results. The network comprises 265 nodes and 1,632 edges, organized around 30 high-degree hubs and ten thematic clusters. Edges were 88.7% pathogenic, 9.7% protective, 1.5% mixed, and 0.2% therapeutic. The network is dominated by an epithelium-to-outcome axis rather than the canonical T2/non-T2 dichotomy; environmental, microbiome, metabolic, and comorbidity clusters are structurally integrated into core disease mechanism. Conclusions. The literature-derived network provides a reusable scaffold for systems-level research on asthma, operationalizes the treatable-traits framework, and makes explicit the integration of comorbidities and environmental determinants into core disease mechanism.