Far Eastern Memorial Hospital

Gastric cancer (GC) remains a global health burden. While international guidelines share consensus, variations exist in disputed issues. The 2025 Taiwan Consensus and Management Guidelines for Gastric Cancer had just been released. We compared the key recommendations with established international guidelines. A multidisciplinary taskforce addressed key questions and recommendations using modified Delphi method and evidence-based approaches. The comparative review aimed to elucidate similarities and differences among guidelines from Taiwan, Japan, South Korea, China, Europe, and the US. Guidelines converge on absolute endoscopic resection criteria for early GC but differ in extended indications and perioperative approaches for locally advanced disease. Heterogeneity exists in biomarker assessment protocols, cutoff thresholds, and companion diagnostics. For metastatic disease, consensus exists on anti-HER2, anti-VEGF, immunotherapy, and biomarker-driven strategies, though oligo-metastatic definitions and intraperitoneal chemotherapy indications remain controversial. Optimal GC management requires integrating global evidence with regional contexts. The comparative review addresses heterogeneity among international guidelines, which helps harmonize management strategies as therapeutic standards evolve.

Albumin, a highly biocompatible protein, has been utilized in the construction of anti-cancer drugs for the treatment of various cancers. However, albumin-based nanomedicines still face several clinical challenges, including low stability in the bloodstream, non-specific targeting ability, and lack of controlled release capability. To address these limitations, we developed a multiprotein-based nanomedicine for targeted chemotherapy of non-small cell lung cancer (NSCLC). This nanomedicine was assembled using a heart-shaped albumin and two Y-shaped anti-CD44 and anti-CD133 antibodies. To enhance stability and improve therapeutic efficacy, the multiproteins were crosslinked using disulfide bond linkers and loaded with paclitaxel and ceramide. This resulted in nanomedicines that exhibited responsiveness to glutathione (GSH) and demonstrated inhibition of tumor cancer cells and cancer stem cells (CSCs). Our experimental results indicated that the inclusion of anti-CD44 and anti-CD133 antibodies enhanced the targeting capability of nanomedicines towards cancer cells and CSCs in an in vitro study and the accumulation in both normoxic regions and hypoxic niches in in vivo tumor xenografts. The multiprotein-based nanomedicines also demonstrated GSH-dependent drug release behavior, induced apoptosis in cancer cells and CSCs, inhibited cell migration, and effectively suppressed NSCLC tumor growth. Overall, our findings presented a novel nanostructure created from differently shaped proteins for application in drug delivery. This multiprotein-based nanomedicines showed promising potential in addressing the limitations of albumin-based nanomedicines and may offer improved therapeutic outcomes for cancer treatment.

Bronchiectasis has traditionally been characterized as a neutrophil-driven disease, yet emerging evidence suggested inflammatory heterogeneities. The prognostic significance of elevated serum immunoglobulin E (IgE) in patients without peripheral eosinophilia remains unclear. We conducted a multicenter prospective cohort study between 2017 and 2020 across 16 institutions in Taiwan. Individuals with bronchiectasis but without allergic bronchopulmonary aspergillosis were included. Patients were stratified by baseline absolute eosinophil count (cutoff 300 /uL) and serum IgE level (≤ 100, 100–500, > 500 IU/mL). The primary endpoint was severe exacerbations resulting in hospitalization at one year. Secondary endpoints included all-cause mortality, distribution of sputum pathogen, imaging pattern, and lung function. A total of 579 individuals were enrolled. Nontuberculous mycobacteria (10.7

Most existing LLM diagnoses are evaluated on static, single-turn settings where complete patient information is provided upfront, an oversimplification of real clinical practice. We study active diagnosis: the real-life clinical process of starting from initial observation, ordering tests, interpreting results, and updating a differential diagnosis across multiple turns. Through systematic analysis, we identify three recurring failure modes in current LLMs — ungrounded test ordering, unreliable diagnostic update, and degraded multi-turn coherence — that together reveal a core deficit: existing medical training data teaches models to reason from complete information but not to act under evolving, partial evidence. To address this gap, we introduce MedAction, a tree-structured distillation pipeline that synthesizes diverse and high-quality multi-turn diagnostic trajectories via LLM–environment interaction. We propose two knowledge-graph-grounded metrics to filter trajectory quality: Disease Trajectory Consistency (DTC), which tracks whether the model's hypothesis converges toward the correct diagnosis, and Reasoning–Action Consistency (RAC), which verifies that belief updates are driven by gathered evidence. Using this pipeline, we construct MedAction-32K, a dataset of 32,681 trajectories from 2,896 PMC cases. Fine-tuning an 8B model on MedAction-32K achieves state-of-the-art performance among open-source models on both MedR-Bench and our curated MedAction-300-Hard benchmark, pushing the edge for open-source medical LLMs.