Hyperthermia is a promising strategy for chronic wound management, with efficacy critically dependent on precise temperature control. Current strategies largely rely on two discrete modes: high-temperature bactericidal ablation or mild hyperthermia to promote tissue regeneration. Enabling hyperthermia to autonomously match the thermal requirements of each wound-healing stage remains challenging. Herein, we developed a biomimetic thermal-feedback modality switchable (TFMS) hydrogel, obtained by copolymerizing N-isopropylacrylamide with sulfobetaine vinylimidazolium (SBVI) and loading glucose oxidase (GOx) together with EGCG-Fe³⁺-modified Bi₂Te₃. Tuning SBVI content precisely programs the volume phase transition temperature of hydrogel, enabling thermostatic photothermal antibacterial activity within a biologically safe window. GOx mediates the dissociation of the metal-polyphenol network during wound repair, thereby flexibly self-limiting the photothermal temperature and activating a thermoelectric effect at mild temperature to compensate for loss of intrinsic electric field in diabetic wounds. TFMS hydrogel enables stage-adaptive repair regulation of infected diabetic wounds through progressive regulation. Compared with commercial bioactive Flamigel, it significantly accelerates wound closure and improves wound repair quality in diabetic mice. The TFMS hydrogel expands the design of dynamic thermostatic hyperthermia in self-adaptive wound dressings, which can promote adaptability and user-friendly wound care, and has great application potential in chronic wound care.

Type 2 diabetes mellitus (T2DM) is characterized by progressive metabolic reprogramming in which lipid metabolism imbalance links nutrient excess to insulin resistance, organelle stress, and tissue dysfunction. Although mitochondria, the endoplasmic reticulum (ER), and lipid droplets (LDs) each have established roles in metabolic homeostasis, how their coordinated interactions govern lipid remodeling in T2DM remains incompletely defined, owing in part to the technical difficulty of resolving these highly dynamic contact sites in space and time. In this Review, we integrate current evidence on mito-ER-LD interactions in T2DM, spanning advances in omics, imaging, and biochemical approaches, and summarize their structural organization, regulatory machinery, tissue-specific remodeling, and therapeutic modulation. Emerging data identify this tri-organelle network as a central metabolic interface that couples phospholipid synthesis and trafficking, calcium flux, mitochondrial dynamics, LD biogenesis and turnover, fatty-acid oxidation, autophagy, and stress signalling. Disruption of this network, including mitochondria-associated ER membrane (MAM) instability, altered balance between cytoplasmic and peridroplet mitochondria, and pathological LD expansion, reprograms lipid handling across liver, adipose tissue, skeletal muscle, and pancreatic β cells, thereby promoting lipotoxicity, oxidative and ER stress, mitochondrial dysfunction, β-cell failure, and systemic insulin resistance. Notably, glucose-lowering drugs and lifestyle interventions may exert part of their metabolic benefit by restoring organelle crosstalk and re-establishing lipid homeostasis. Together, this Review positions mito-ER-LD contact sites not simply as structural interfaces, but as core nodes that regulate lipid metabolic reprogramming in T2DM, thereby providing a mechanistic framework for targeting organelle crosstalk in metabolic disease.

Chronological age is widely used in perioperative risk assessment, but does not adequately capture interindividual differences in physiological reserve. Recently developed phenotypic biological aging measures quantify biological age using clinical biomarkers and may better reflect frailty and surgical vulnerability. This study compared the predictive value of biological and chronological age for functional and perioperative outcomes in patients undergoing total knee arthroplasty (TKA).

We retrospectively analyzed 173 consecutive adults who had osteoarthritis undergoing TKA. Biological aging metrics were calculated from nine standard blood biomarkers using a previously validated algorithm. Outcomes included total and limb skeletal muscle mass, 10-meter gait time, and overall and symptomatic perioperative complications. Associations were assessed using multivariable linear and logistic regression models adjusted for sex, body mass index (BMI), body weight, frailty status using the 11-item modified frailty index, diabetes mellitus, and hypertension. Predictive performance was evaluated using adjusted R² and area under the curve (AUC) with 95% confidence intervals.

Biological age acceleration was associated with slower gait time (β = 0.48 seconds/year, P < 0.001; R² = 0.19), outperforming chronological age (R² = 0.05). In adjusted logistic regressions, biological age acceleration was associated with increased odds of overall perioperative complications, including asymptomatic deep vein thrombosis detected by routine ultrasonography (odds ratio [OR] = 1.20 per one-year increase, 95% confidence interval [CI] 1.10 to 1.30, P < 0.001), with good discrimination (AUC = 0.77). This association persisted for symptomatic complications (OR = 1.11, 95% CI 1.04 to 1.19, P = 0.003). Chronological age was not associated with complications (P = 0.34). Daily alcohol consumption and smoking were modifiable correlates of accelerated biological aging.

Biological aging metrics better reflected perioperative vulnerability than chronological age in patients undergoing TKA and may support individualized risk stratification and lifestyle optimization.

To investigate the association between serum extracellular vesicle-derived neuropilin-1 (EV NRP1) and endothelial dysfunction in patients with type 2 diabetes mellitus (T2DM).

In this cross-sectional study, 164 hospitalized patients with type 2 diabetes mellitus underwent Endo PAT testing and were classified into endothelial dysfunction and normal endothelial function groups according to the reactive hyperemia index (RHI). Serum extracellular vesicles were isolated using a precipitation-based kit. Serum extracellular vesicle-derived neuropilin-1 and serum neuropilin-1 concentrations were measured using a flow cytometric multiplex assay.

EV NRP1 levels were significantly lower in the endothelial dysfunction group than in the normal endothelial function group (207.69 ± 142.97 vs. 271.89 ± 174.07 ng/mL, P = 0.015), and serum NRP1 showed a similar decreasing trend (206.10 ± 48.35 vs. 223.22 ± 49.03 ng/mL, P = 0.030). EV NRP1, rather than serum NRP1, was positively correlated with RHI and LnRHI. Higher EV NRP1 was independently associated with lower odds of endothelial dysfunction (OR = 0.627, 95% CI 0.417-0.944; P = 0.025).

EV NRP1 levels are reduced and independently associated with endothelial dysfunction in patients with T2DM, supporting its potential as a biomarker for diabetic endothelial dysfunction.

Diabetic peripheral neuropathy (DPN) is associated with impaired skeletal muscle function, particularly in the lower limbs. However, microvascular function in lower leg skeletal muscles among individuals with type 1 diabetes mellitus (T1DM), stratified by DPN and neuropathic pain status, remains poorly understood. This study examined blood oxygen level-dependent (BOLD) responses to ischemia-reperfusion in the tibialis anterior (TA), soleus (SO), and medial gastrocnemius (GM) muscles.

Participants included individuals with T1DM and painful DPN (T1DM + PDPN; n = 19), T1DM with painless DPN (T1DM + DPN; n = 18), T1DM without DPN (T1DM-DPN; n = 15), and age-matched healthy controls (n = 20). While supine in a Magnetic Resonance scanner, participants underwent 5 min of thigh cuff occlusion followed by reperfusion, during which BOLD imaging captured peak and time-to-peak (TTP) in TA, SO, and GM.

Across muscles, peak BOLD was lower in T1DM + PDPN (105.0% of baseline) compared to both T1DM-DPN (107.9%; p = 0.025) and controls (108.4%; p < 0.01), and lower in T1DM + DPN (105.8%) compared to controls (p < 0.01). No significant group differences were found in TTP. Additional analyses revealed that individuals with confirmed small fiber neuropathy exhibited both a diminished (105.3%) and delayed (42.1 s) BOLD response across muscles compared to controls (31.5 s; p < 0.01).

These findings suggest that skeletal muscle microvascular function is compromised in individuals with T1DM, particularly those with DPN, neuropathic pain, and small fiber neuropathy. This reduced perfusion capacity may contribute to functional impairments in lower limb musculature, highlighting the need for further investigation into underlying mechanisms and potential therapeutic strategies.

Single-cell RNA sequencing reveals cellular heterogeneity, yet computational methods struggle to balance performance with biological interpretability. Embedded topic models provide interpretable cell representations, but may learn overly similar topics, resulting in redundancy and incomplete capture of biological variation. Single-cell foundation models create opportunities to harness external biological knowledge for guiding model embeddings. Here, we present scE²TM, an external knowledge-guided embedded topic model for interpretable scRNA-seq analysis. scE²TM implements embedding clustering regularization where each topic is encouraged to represent a distinct group of genes, enabling it to capture unique biological information. We show that across 20 datasets, scE²TM outperforms seven state-of-the-art methods in clustering performance. We perform an interpretability benchmark to show that scE²TM topics exhibit greater diversity and stronger consistency with biological pathways. When modelling interferon-stimulated peripheral blood mononuclear cells, we find that scE²TM simulates topic perturbations that shift control cells toward stimulated states, recapitulating experimental interferon responses. When tested on a melanoma dataset, scE²TM identifies malignant-specific topics and extrapolates them to unseen patient data, highlighting melanoma-associated gene programs linked to patient survival.

The clinical utility of measurable residual disease monitoring based on next-generation sequencing in T-cell acute lymphoblastic leukemia remains incompletely defined. Here we show the prognostic value of tracking clonal T-cell receptor gene rearrangements by next-generation sequencing in pediatric patients with T-cell acute lymphoblastic leukemia. Among the 101 patients with trackable clones, measurable residual disease levels at the end of consolidation strongly stratify 4-year event-free survival: 81.6% ± 5.6% for disease levels below 0.0001%, 61.9% ± 11.8% for 0.0001-0.1%, and 45.1% ± 12.1% for levels ≥ 0.1% (P = 0.004). Both T-cell receptor beta and gamma/delta rearrangements demonstrate prognostic significance. Notably, patients with multiparameter flow cytometry-negative but next-generation sequencing-positive measurable residual disease at the end of consolidation have significantly lower 4-year event-free survival compared to those achieving double-negative status (60.8% ± 9.4% vs. 84.9% ± 5.3%, P = 0.038). In conclusion, measurable residual disease monitoring by next-generation sequencing provides superior sensitivity and important clinical value, offering a potential strategy for precise risk stratification in T-cell acute lymphoblastic leukemia.

In approximately one third of patients, resected head and neck squamous cell carcinoma will recur. We postulated that the induction of tumor neoantigen-specific T cell responses could prevent relapse. To this end, we developed TG4050, an individualized neoantigen therapeutic vaccine encoding up to 30 patient-specific predicted tumor neoantigens delivered by a Modified Vaccinia Ankara viral vector. We tested adjuvant TG4050 as single agent in a randomized phase I trial comparing treatment with TG4050 immediately after standard of care adjuvant therapy versus watchful waiting and treatment with TG4050 after recurrence (NCT04183166). The primary endpoint was safety, secondary endpoints included feasibility and efficacy, and immunogenicity was exploratory. TG4050 was well tolerated. Of 16 evaluable patients randomized the immediate treatment arm, none relapsed after a median follow-up of 30 months, while 3 of 16 relapsed in the control arm. T cell responses to vaccine neoantigens were detected in 73.3% of patients treated with TG4050 immediately, with a median of 3 neoantigens per responder. These responses were maintained throughout treatment and persisted for over one year after the last dose. Vaccine neoantigen-specific CD8⁺ T cells had an effector phenotype, displayed high expression of cytotoxic and tissue-resident markers, were polyclonal and comprised both de novo responses and amplification of pre-existing tumor-infiltrating T cell clones. Together, these translational data are consistent with the hypothesis in which single-agent delivery of TG4050 induces long-lasting tumor neoantigen-specific cytotoxic T cell responses that can prevent tumor recurrence.

The FDA Oncologic Drugs Advisory Committee (ODAC) highlighted that the risk-benefit profile for patients with low PD-L1 expression remains unclear in front-line treatment of HER2-negative gastric and gastroesophageal junction (GC/GEJ) cancer with PD-(L)1 therapy. The first PD-1/CTLA-4 bispecific antibody (BsAb), cadonilimab, has been approved, but it is unknown if it can extend the PD-(L)1 regimen to lower PD-L1 patients.

To indirectly compare the survival outcomes of cadonilimab versus PD-(L)1 therapy in first-line HER2-negative gastric and gastroesophageal junction cancer, with particular focus on patients with low PD-L1 expression, using a network meta-analysis.

Randomized controlled trials with ICIs as first-line treatment were searched in public databases and FDA ODAC up to December 31, 2024. A network meta-analysis was conducted to evaluate the efficacy and safety of approved ICI-based treatments to refine the optimal approach for front-line GC/GEJ.

A total of 7127 patients from eight Phase III trials were included. All ICIs-based therapies significantly improved overall survival (OS) compared to chemotherapy. In indirect comparisons, cadonilimab plus XELOX was associated with numerically more favorable OS (HR = 0.78, 95% CI: 0.62-0.98) compared with PD-(L)1 inhibitors. Numerical trends toward longer survival were observed across subgroups, including patients with PD-L1 < 10 (HR = 0.77, 95% CI: 0.58-1.01) and liver metastasis (HR = 0.67, 95% CI: 0.49-0.93), both historically considered immunotherapy-insensitive populations. All ICIs improved PFS compared to chemotherapy. Cadonilimab was associated with numerically better PFS versus PD-(L)1 inhibitors in indirect comparisons (HR = 0.70, 95% CI: 0.57-0.86). The safety analysis showed no significant increase in severe adverse events.

In this indirect comparison, cadonilimab showed more favorable survival outcomes versus PD-(L)1 inhibitors in first-line GC/GEJ. Subgroup benefits were observed in liver metastasis, while the PD-L1 < 10 subgroup showed numerical trends. These results highlight the importance of selecting appropriate ICIs to optimize outcomes.

We evaluated the impact of incorporating systematic fourth-arm traction and articulating-needle drivers on the perioperative and oncological outcomes of Senhance robotic radical prostatectomy.

We conducted a retrospective single-center analysis of 30 consecutive patients who underwent Senhance robotic radical prostatectomy procedures between October 2024 and December 2025. The patients were stratified into an initial cohort (cases 1-15), which underwent surgery using a three-arm robotic configuration and conventional straight needle drivers, and a refined cohort (cases 16-30), in which continuous fourth-arm traction and articulating-needle drivers were used. The primary endpoints were operative time and positive surgical margin rate.

The median operative time decreased significantly, from 260 min in the initial cohort to 200 min in the refined cohort (p < 0.01). This figure was further reduced, from 244 to 194 min (p < 0.01), after all cases involving lymphadenectomy were excluded. The positive surgical margin rate showed a non-significant trend toward reduction from 33% to 7% (p = 0.17). Early urinary continence was also evaluated as an initial functional outcome, with no significant difference between the two cohorts. A learning curve analysis demonstrated a progressive improvement in operative efficiency (R² = 0.57, p < 0.01). No major complications (Clavien-Dindo grades ≥ III) occurred in either cohort.

The refined protocol was associated with shorter operative time and a non-significant trend toward a lower positive surgical margin rate. These findings should be interpreted cautiously, as our refinement of the surgical protocol overlapped with the learning curve of the surgeon who performed the procedures.