Therapeutic hypothermia (TH) is widely used as a neuroprotective strategy, but variability in bedside implementation may limit its optimal application. This study aimed to develop a standardized TH management protocol for patients with neurocritical illness and evaluate the clinical outcomes associated with its implementation.

An evidence-based TH protocol was developed through a two-round Delphi consensus process. A subsequent observational before-and-after study included 244 patients with neurocritical illness receiving TH. The historical control group (n = 120) received routine care, whereas the TH protocol group (n = 124) was managed according to the standardized protocol. No single primary endpoint was prespecified for the final comparative analysis. Outcomes were interpreted as exploratory and included temperature-management process indicators, individual TH-related complications requiring intervention, and neurological outcomes assessed using the Glasgow Outcome Scale.

Protocol implementation was associated with more favorable temperature-management process indicators, including earlier TH initiation, a shorter recorded time to target temperature, fewer unplanned treatment interruptions, a longer controlled rewarming phase, and greater adherence to rewarming criteria. In exploratory analyses, the TH protocol group had lower incidences of selected TH-related complications and more favorable Glasgow Outcome Scale outcomes at discharge than the historical control group.

Implementation of a consensus-derived TH management protocol was associated with more consistent temperature-management processes, fewer selected TH-related complications, and more favorable short-term neurological outcomes. These findings support prospective evaluation of protocolized TH management in larger multicenter cohorts.

Despite the 2012 Berlin definition standardizing acute respiratory distress syndrome (ARDS) diagnosis, pharmacological therapies have consistently failed in phase 3 trials, partly reflecting the syndrome’s profound biological heterogeneity. We synthesize evidence on ARDS subphenotyping to evaluate prospects for precision medicine. This narrative review was informed by targeted searches of PubMed/MEDLINE, Embase, and Web of Science from database inception through February 2026, supplemented by an updated PubMed/MEDLINE search through September 2026. Representative clinical trials, multicenter cohorts, mechanistic studies, and methodological reports were selected to support a critical synthesis of ARDS subphenotyping, its therapeutic implications, and barriers to clinical translation. Hyperinflammatory and hypoinflammatory subphenotypes—originally identified using plasma biomarkers (IL-6, IL-8, and soluble tumor necrosis factor receptor 1) in the ARMA and ALVEOLI cohorts—demonstrate reproducible differences in mortality, with worse outcomes in the hyperinflammatory subphenotype, and have shown potential treatment-effect heterogeneity in secondary analyses. Approximately 30%–35% of patients exhibit the hyperinflammatory profile, characterized by greater vasopressor requirements and organ failure. Radiological phenotypes (focal versus non-focal) on computed tomography imaging have been associated with differences in lung recruitability and physiological responses to ventilatory interventions, although the LIVE trial did not demonstrate an overall survival benefit from morphology-guided ventilation. Recent multi-omics studies (transcriptomic, proteomic, and metabolomic) reveal distinct endotypes involving dysregulated innate immunity, endothelial activation, and aberrant coagulation. Machine learning algorithms, particularly unsupervised clustering and deep learning-based computed tomography radiomics, have been investigated for more rapid phenotyping, but reported performance varies by task, cohort, and validation design. However, prospective, phenotype-guided randomized trials remain limited. Subphenotype stratification offers a mechanistically grounded framework for re-evaluating therapies that were ineffective in unselected populations and for investigating phenotype-informed ventilatory strategies. Clinical implementation requires standardized, bedside-compatible biomarker panels and definitive trials demonstrating that phenotype-specific interventions improve patient-centered outcomes.

Pegylated interferon alpha (PegIFNα) has been associated with altered lipid metabolism, but long-term lipid trajectories in chronic hepatitis B remain poorly characterized. We evaluated lipid profiles and virological parameters during 144 weeks of PegIFNα-based therapy.

This single-center retrospective cohort study included 129 patients with chronic hepatitis B who completed 144 weeks of PegIFNα monotherapy, PegIFNα plus tenofovir alafenamide, or PegIFNα plus entecavir. Serum lipids, hepatitis B surface antigen (HBsAg), hepatitis B e antigen, and hepatitis B virus DNA were assessed at baseline and weeks 24, 48, 72, 96, and 144. Within-patient changes were analyzed with paired tests. Associations between early lipid changes and on-treatment HBsAg seroclearance at week 144 were examined with logistic regression.

From baseline to week 144, mean high-density lipoprotein cholesterol (HDL-C) increased by 0.10 ± 0.25 mmol/L (95% confidence interval, 0.06 to 0.15; P < 0.001) and mean total cholesterol decreased by 0.15 ± 0.41 mmol/L (95% confidence interval, −0.22 to −0.08; P < 0.001). Changes in triglycerides, low-density lipoprotein cholesterol, and the triglyceride-to-HDL-C ratio were not significant. HDL-C declined early and later recovered above baseline. At week 144, 37 patients (28.7%) achieved on-treatment HBsAg seroclearance (≤ 0.05 IU/mL). Of the 34 patients with HBsAg seroclearance at week 96, 13 had detectable HBsAg again at week 144. Early lipid changes were not associated with this endpoint.

HDL-C shows an early decline followed by recovery above baseline, whereas total cholesterol decreases modestly, with no significant net changes in triglycerides, low-density lipoprotein cholesterol, or the triglyceride-to-HDL-C ratio. Early lipid changes are not significantly associated with on-treatment HBsAg seroclearance. The durability of HBsAg loss requires post-treatment follow-up.

N-acetyltransferase 10 (NAT10), a major RNA acetyltransferase mediating N4-acetylcytidine modification, has been implicated in cancer progression. We examined its cancer-type-specific associations with clinical outcomes, genomic features, and tumor immune characteristics.

We analyzed public TCGA and GTEx datasets using UCSC Xena and complementary resources, including GEPIA2, HPA, cBioPortal, TIMER2.0, STRING, LinkedOmics, and CancerSEA. Analyses included NAT10 expression, pathological stage, survival, genomic alterations, immune-cell infiltration, stemness, and molecular pathways. Exploratory immune-checkpoint-blockade (ICB) survival analyses used Kaplan–Meier Plotter cohorts receiving anti-PD-1 (n = 411), anti-PD-L1 (n = 459), or anti-CTLA-4 (n = 112).

NAT10 expression was elevated in multiple cancer types and was associated with adverse survival in selected malignancies. Higher NAT10 expression correlated positively with estimated myeloid-derived suppressor cell and cancer-associated fibroblast infiltration and negatively with several T-cell and natural killer-cell populations. NAT10-associated genes were enriched in RNA-processing pathways. In exploratory ICB cohorts, higher NAT10 expression was nominally associated with longer overall survival following anti-PD-1 therapy (hazard ratio (HR) = 0.68, 95% confidence interval (CI): 0.52–0.90) and anti-PD-L1 therapy (HR = 0.72, 95% CI: 0.56–0.92), but not anti-CTLA-4 therapy (HR = 0.66, 95% CI: 0.39–1.11).

NAT10 shows cancer-type-dependent prognostic, molecular, and immune associations. Its associations with survival after ICB are exploratory and require independent clinical and mechanistic validation.

Extensive burn injuries are associated with a profound systemic response, the severity of which increases with the extent of tissue damage. Beyond local destruction, extensive burns induce a sustained immune-inflammatory disturbance that alters immune regulation and contributes to secondary organ failure and infectious complications. These mechanisms are major determinants of adverse outcomes in critically ill patients with burns. In this context, therapeutic strategies targeting immune dysregulation have gained increasing attention. Immunomodulatory approaches aim to limit excessive inflammatory signaling while maintaining effective host defense. Among extracorporeal blood purification techniques, hemoadsorption has emerged as a potential adjunctive therapy designed to reduce circulating inflammatory mediators, particularly cytokines, and to attenuate systemic inflammatory burden. Rather than providing another general overview of extracorporeal blood purification, this review focuses specifically on the cellular and mechanistic basis of burn-induced immune dysregulation and offers a stratified, critical appraisal of the experimental and clinical evidence for hemoadsorption in burns, together with current safety considerations and unresolved questions that should guide future research. Experimental studies and early clinical observations suggest that hemoadsorption may contribute to hemodynamic stabilization and modulation of inflammation in critically ill patients, including those with severe burns. However, clinical evidence in this specific population remains scarce, and its impact on patient-centered outcomes has not yet been established. Although available data suggest that this technique is generally well tolerated, the non-selective removal of circulating molecules underscores the need for careful patient selection and close monitoring. Future investigations should focus on defining the appropriate indications, optimal timing, and treatment duration of hemoadsorption for burn care. Comparative studies with other immunomodulatory strategies are required to clarify its role within the therapeutic armamentarium for burn-induced immune dysfunction.

We compared logistic regression, random forest, and extreme gradient boosting (XGBoost) for classifying current colorectal cancer screening status among U.S. adults aged 45–75 years.

We analyzed 2024 Behavioral Risk Factor Surveillance System data for 227,647 adults aged 45–75 years with determinable screening status. Data were split into stratified training and held-out test sets before imputation. Missing predictors were handled with 20 multiply imputed datasets; imputation and preprocessing were fitted in training data and applied to held-out data. Model performance was assessed using five-fold cross-validation and the held-out test set.

In the test set, areas under the receiver operating characteristic curve were 0.771 for logistic regression, 0.778 for random forest, and 0.781 for XGBoost; corresponding areas under the precision-recall curve were 0.884, 0.888, and 0.890. Brier scores were 0.152, 0.153, and 0.149, respectively. At model-specific Youden thresholds, sensitivity ranged from 75.5% to 78.6% and specificity from 62.5% to 66.3%.

XGBoost has the highest discrimination and lowest Brier score, but differences among models are small. The models classify current screening status in this analytic cohort; external validation is needed before use in screening outreach.