Background/Aims: Various expanded criteria (EC) for liver transplantation (LT) in patients with hepatocellular carcinoma (HCC) have been proposed to avoid the narrow nature of the Milan criteria (MC). To investigate which EC predicts more favorable outcomes in terms of overall survival (OS) and recurrence-free survival (RFS), we conducted a network meta-analysis (NMA). Methods: A database search was conducted on PubMed, Embase, and the Cochrane Library, to identify studies comparing OS and RFS between patients within the MC and those exceeding the MC but within the EC. Hazard ratios (HRs) were pooled using a random-effects NMA and validated in an in-house cohort of 1,008 LT recipients. Results: Among 22,466 articles identified, 35 studies with 45 pairwise comparisons were included in the NMA along with 8 different EC. The University of California San Francisco (HR, 1.43; 95% CI, 1.19–1.71), Up-to-Seven (HR, 1.50; 95% CI, 1.15–1.97), and Hangzhou criteria (HR, 1.69; 95% CI, 1.11–2.57) showed inferior OS to the MC. The MC ranked highest for both OS and RFS, followed by Metroticket 2.0 for OS and the Asan criteria for RFS. In the validation cohort, both Metroticket 2.0 and AFP model yielded more favorable HCC-specific mortality than other EC. Conclusions Several EC, of which those of Metroticket 2.0 were the best, yielded comparable outcomes to the MC. AFP-based EC such as Metroticket 2.0 and AFP model appeared to be useful in both the NMA and the validation cohort, suggesting a potential role in identifying selected low-risk patients beyond the MC.

Vaccines are highly effective, but rare or delayed adverse events following immunization (AEFIs) require post-licensure surveillance beyond clinical trials. Korea lacks a comprehensive, active, database-based framework, yet key assets exist: nationwide claims databases (National Health Insurance Service/Health Insurance Review and Assessment Service), the national immunization registry (Korea Disease Control and Prevention Agency’s Immunization Registry Information System) for National Immunization Program (NIP) and non-NIP vaccines, and increasingly standardized hospital electronic health records.We propose a federated, code to data architecture with data linkages between these data.Implementation should adopt a common data model (CDM), standardized case definitions, latency accounting, and transparent public reporting under strong privacy governance. Major challenges include multi step administrative approvals for data linkage, incomplete capture of adult non-NIP vaccinations, heterogeneous hospital data structures, and strict data protection constraints. Strategic priorities are to streamline statutory and administrative processes for public health use, mandate or enable claims-based capture of adult vaccinations, enhance CDM based interoperability, and develop secure hubs for aggregated outputs. With these measures, Korea will be well positioned to establish a scalable active surveillance system capable of detecting rare AEFIs, supporting transparent and evidence-based communication, and ensuring equitable injury compensation grounded in domestic data.

Purpose: This study evaluated the clinical outcomes and prognostic factors in patients with borderline resectable pancreatic cancer (BRPC) and locally advanced pancreatic cancer (LAPC) treated with upfront FOLFIRINOX followed by local-regional therapy (LRT), surgical resection (SR), and radiotherapy (RT). We aimed to identify specific patient subgroups for which RT may serve as a reasonable alternative to SR for local tumor control. Methods: We retrospectively analyzed 116 patients (SR group, n= 70; RT group, n= 46) at a single center between 2015 and 2020. Survival outcomes were compared based on LRT modalities, focusing on identifying subgroups in which RT provided an efficacy comparable to that of SR. Results: Among 116 patients, the SR group achieved a significantly higher 5-year overall survival (OS) than the RT group (27.1% vs. 8.7%, P< 0.0001), despite similar progression-free survival (P= 0.23). Significant prognostic factors for OS included carbohydrate antigen 19-9 (CA19-9) response in BRPC (P= 0.02) and radiologic partial response in LAPC (P= 0.05). Subgroup analysis revealed that, while SR provided a survival advantage in CA19-9 responders, no significant difference in OS was observed between SR and RT in CA19-9 non-responders (P= 0.37). Conclusion Although surgery remains the gold standard, RT may be considered a justifiable local alternative for CA19-9 non-responders and surgically ineligible patients with LAPC, yielding comparable outcomes in these specific, biologically unfavorable subgroups.

Purpose: Preterm birth remains a leading cause of long-term neurodevelopmental impairment, yet early evaluations frequently underestimate subsequent deficits. This study examined longitudinal neurodevelopmental trajectories across gestational age groups and identified predictors of developmental delay. Methods: A retrospective cohort of 532 preterm children, stratified by gestational age, was followed from the neonatal period to early preschool age. Neurodevelopment was assessed using the Korean version of the Bayley Scales of Infant and Toddler Development, Third Edition at 8–12 months (n=481), 13–24 months (n=118), and 25–42 months (n=100). Longitudinal trajectories were analyzed using general linear models, and predictors of developmental delay were identified through multivariable logistic regression. Results: During the first year, motor scores differed significantly across gestational age groups, with extremely preterm infants showing the lowest values. By the third to fourth years of life, cognitive and language scores diverged markedly, with extremely preterm children exhibiting the steepest decline and additional deficits in motor and adaptive behavior domains. Lower gestational age remained an independent predictor of both cognitive and language delay at early preschool age, while no independent predictors were identified for motor, social-emotional, or adaptive behavior outcomes. Conclusion Neurodevelopmental outcomes in preterm children follow dynamic, domain-specific trajectories influenced by gestational age and developmental timing. Motor delays are most evident in infancy, whereas cognitive and language impairments emerge by early preschool age. Gestational age remains a consistent predictor of later delay, emphasizing the need for longitudinal, gestational age–stratified monitoring and early, targeted intervention.

Objective: To evaluate the accuracy and reasoning capabilities of large multimodal language models compared with those of neuroradiology subspecialty-trained radiologists in neuroradiology case interpretation. Materials and Methods: This experimental study used custom-made 401 radiologic quizzes derived from articles published in RadioGraphics covering neuroradiology and head and neck topics (October 2020 to February 2024). We prompted the GPT-4 Turbo with Vision (GPT-4V), GPT-4 Omni, Gemini Flash, and Claude models to provide the top three differential diagnoses with a rationale and describe examination characteristics such as imaging modality, sequence, use of contrast, image plane, and body part. The temperature was adjusted to 0 and 1 (T1). Two neuroradiologists answered the same questions.The accuracies of the large language models (LLMs) and the neuroradiologists were compared using generalized estimating equations. Three neuroradiologists assessed the rationale provided by the LLMs for their differential diagnoses using four-point scales, separately for specific lesion locations and imaging findings, and evaluated the presence of hallucinations and the overall acceptability of the responses. Results: Top-3 accuracy (i.e., correct answers present among top-3 differential diagnoses) of LLMs ranged from 29.9% (120 of 401) to 49.4% (198 of 401, obtained with GPT-4V in the T1 setting), while radiologists achieved 80.3% (322 of 401) and 68.3% (274 of 401), respectively (P < 0.001). Regarding the rationale for differential diagnoses, GPT-4V (T1) accurately identified both the specific lesion location and imaging findings in 30.7% (123 of 401) and 12.9% (16 of 124) of cases without textual clinical history. Hallucinations occurred in 4.5% (18 of 401), and only 29.4% (118 of 401) of the LLM-generated analyses were deemed acceptable. GPT-4V (T1) demonstrated high accuracy in identifying the imaging modality (97.4% [800 of 821]) and scanned body parts (92.2% [756 of 820]). Conclusion LLMs remarkably underperformed compared with neuroradiologists and showed unsatisfactory reasoning for their differential diagnoses, with performance declining further in cases without textual input of clinical history. These findings highlight the limitations of current multimodal LLMs in neuroradiological interpretation and their reliance on text input.

Purpose: Neoadjuvant treatment (NAT) is now the standard for borderline resectable pancreatic cancer (RPC) and is being considered for RPC. Early recurrence after curative surgery in RPC is often seen as a treatment failure, prompting considerations for NAT. Our goal was to develop an artificial intelligence (AI)-based predictive model utilizing preoperatively available factors to forecast early recurrences of resected RPC. Methods: This study included 469 patients who underwent surgery for RPC between 2011 and 2019. Clinicopathologic and oncologic data were retrospectively reviewed. Preoperative variables, including laboratory data and imaging findings, were collected. Early recurrence was defined as recurrence occurring within a year after surgery. Deep neural networks were then used to select variables by assessing their importance. A new model predicting early recurrence of RPC was subsequently developed. Results: Of the patients evaluated, 199 (42.4%) experienced early recurrence. The predictive model included 14 preoperative variables: CA 19-9, preoperative pancreatitis, serum albumin, platelet count, lymphocyte count, the American Society of Anesthesiologists physical status classification, tumor size, monocyte count, age, body mass index, CRP, hemoglobin, WBC count, and CEA. The area under the curve for the model was 0.786 in the training set and 0.734 in the test set. Conclusion We developed an AI-based model to predict the early recurrence of RPC using preoperative parameters. By identifying patients at risk of early recurrence, optimal individualized treatments such as NAT can be considered. Future prospective studies are crucial to establish clear indications for NAT in RPC.

Purpose: This study aimed to report the real-world outcomes of intravesical gemcitabine for bacillus Calmette–Guérin (BCG)–unresponsive, high-risk, non–muscle-invasive bladder cancer (HR-NMIBC) in Korean patients who were unable or unwilling to undergo radical cystectomy (RC). Materials and Methods: This retrospective study included 131 patients (median age, 69 years; 88.5% men) treated with intravesical gemcitabine for BCG-unresponsive HR-NMIBC at nine centers between May 2019 and April 2022. The primary endpoint was 1-year recurrence-free survival (RFS). The secondary endpoints included factors influencing RFS, progression-free survival (PFS), cystectomy- free survival, cancer-specific survival (CSS), overall survival (OS), and safety. Survival analysis was performed using the Kaplan-Meier method, and risk factors for recurrence were assessed using Cox regression models. Results: Patients were followed up for a median duration of 25 months, with carcinoma in situ (CIS) in 41.9% of the patients. The 1-year and 2-year RFS rates were 68% and 42%, while the 1-year and 2-year PFS rates were 87% and 77%, respectively. No significant factors influencing RFS were identified. Seventeen patients underwent RC during a median follow-up of 16 months, with the condition in three patients progressing to muscle-invasive disease on final pathological analysis. The 2-year CSS and OS rates were 98% and 97%, respectively. Intravesical gemcitabine was well-tolerated, with only seven patients (5.3%) unable to complete the full induction course. Conclusion Our research highlights the potential of intravesical gemcitabine as a viable bladder-sparing treatment option for BCG-unresponsive HR-NMIBC, providing real-world evidence on its safety, efficacy, and tolerability.

The pathogenesis of endotoxin-induced acute lung injury is fundamentally driven by dysregulated innate immune responses, where macrophage-mediated cytokine surges and subsequent signaling cascades trigger neutrophil infiltration and tissue damage.This study investigated whether curcumin modulates inflammatory signaling pathways and attenuates lung injury in experimental endotoxemia. Using endotoxin-stimulated murine macrophages and a mouse model of intratracheal lipopolysaccharide challenge, inflammatory cytokine production, mitogen-activated protein kinase activation, pulmonary edema, neutrophil accumulation, histopathologic injury, and short-term survival were assessed. Curcumin suppressed endotoxin-induced tumor necrosis factor- production and selectively inhibited ERK1/2 and JNK phosphorylation without affecting p38 signaling in macrophages. In vivo, curcumin reduced pulmonary cytokine levels, neutrophil infiltration, lung edema, and histologic injury, and was associated with improved survival following severe endotoxin exposure. These findings indicate that curcumin attenuates acute lung injury by selectively modulating intracellular inflammatory signaling pathways, supporting the concept that targeted inhibition of specific kinase cascades may mitigate inflammatory lung damage without broad immune suppression.