Search Results (716)

(1) Background and Objectives: Parkinson’s disease’s (PD) underlying pathophysiology still remains incompletely understood, with Braak’s hypothesis of ASyn pathology propagation being the most widely accepted. Recently, a novel model has been introduced, proposing two distinct ASyn propagation pathways: a bottom-up trajectory termed Body-first PD, and a central nervous system (CNS)-initiated pathway termed Brain-first PD. This distinction introduces new perspectives in the PD literature landscape regarding diagnosis, prognostic factors and patient management. This study set out to systematically synthesize the current literature comparing Brain-first and Body-first PD, with a focus on clinical characteristics and disease progression, diagnostic biomarkers, and management approaches. (2) Materials and Methods: A systematic literature search was conducted in March 2025 using PubMed, Cochrane Library, DOAJ and Google Scholar. Human observational, diagnostic, and interventional studies published between 2019 and March 2025, including patients with de novo or early PD, were eligible. Pre-motor REM sleep behavioral disorder (RBD) was used as the primary differentiation criterion. Risk of bias was evaluated using the Joanna Briggs Institute (JBI) critical appraisal checklists. Results were synthesized using a narrative approach. (3) Results: Sixteen studies comprising 2107 PD patients met the inclusion criteria. Body-first PD was associated with a higher non-motor symptom (NMS) burden, faster disease progression, and a higher prevalence of cognitive impairment. Additionally, Body-first PD patients exhibited more widespread and symmetrical neurodegeneration, along with electrophysiological and metabolic differences. Distinct biomarker and microbiome profiles were also observed between subtypes. No eligible studies addressing management approaches were identified. (4) Conclusions: In conclusion, the available evidence suggests that Brain-first and Body-first PD may represent two distinct pathophysiological entities, a proposal with great significance for the diagnosis, prognosis and management of PD patients. However, the predominantly cross-sectional nature of the current literature limits causal inference. Future longitudinal and interventional studies are required to clarify the potential clinical implications of this subtype classification theory. Full article

Expansion of the global aluminum industry has produced severe landscape degradation through surface bauxite mining and massive accumulation of bauxite residue (red mud). This study explores a circular economy strategy for rehabilitating post-bauxite mining Oxisols in the Tayan Bauxite Mining Cluster (0°02′01″ S, 110°06′07″ E), West Kalimantan, Indonesia, by applying red mud blended with composted cow manure as soil ameliorants. A 6-month field trial using a Randomised Block Design (RBD) (20 plots) assessed the effects of graded red mud–manure combinations on foliar nutrient uptake and vegetative growth of Neolamarckia cadamba (Jabon). Treatment C (1.0 kg red mud + 13.5 kg manure) optimised soil chemical conditions most effectively, yielding the highest foliar macronutrient concentrations per plant (N: 3.47%, P: 0.57%, K: 1.63%) and the greatest height increment (84.99 cm, a 36.9% gain over the unamended control at six months). Higher doses (Treatment D) produced diminishing returns, pointing to an inferred sodium-induced inhibitory threshold based on observed growth responses. These findings indicate that integrating a source of sodium from red mud with organic matter shows strong potential as an effective soil ameliorant for post-mining rehabilitation. The approach supports SDG (Sustainable Development Goals) 12 (Responsible Consumption and Production) by converting industrial waste into a productive soil input, and SDG 15 (Life on Land) by facilitating ecological recovery on degraded mine landscapes. Full article

Multi-body separation of flight vehicles is challenged by potential collisions that critically affect dynamic stability. This study develops a numerical method for simulating coupled aerodynamics, kinematics, and collision dynamics. Building upon a conventional computational fluid dynamics/rigid body dynamics (CFD/RBD) framework, the proposed approach integrates a collision dynamics model based on impulse–momentum theory and Coulomb’s friction law, together with a parallelized collision detection algorithm employing edge-face bounding boxes. A loosely coupled staggered solution scheme is established to effectively overcome the limitation of overset mesh in handling colliding bodies. The method is validated through store separation and rigid sphere collision, confirming its capability in resolving aerodynamic/kinematic coupling and normal/tangential collision responses. Application to a cluster munition separation case reveals shell behaviors at distinct initial velocities and identifies a critical safety boundary when the initial shell separation velocity reaches 3.25 times the projectile velocity, defining kinematic and aerodynamic threshold criteria for collision-free separation. Quantitative error analysis shows that the velocity and angular velocity errors from the aerodynamic approximation remain below 2.5% of the collision-induced increments, confirming the method’s engineering accuracy. Flowfield analysis shows that lower velocities result in severe shock interference and collision, whereas higher velocities enable rapid clearance, aerodynamic recovery, and clean separation. Full article

Understanding how clinical symptoms relate to immune responses during major variant transitions remains important for informing post-pandemic surveillance and vaccination strategies. This study compared symptom patterns and SARS-CoV-2-specific anti-RBD IgM and anti-S1 IgG antibody responses among vaccinated individuals infected during the pre-Omicron and Omicron-dominant periods, representing a key phase in the evolution of SARS-CoV-2 population immunity. A retrospective analysis of 216 confirmed COVID-19 cases was performed by evaluating 11 predefined symptoms together with anti-RBD IgM and anti-S1 IgG levels measured at Day-14 after symptom onset, corresponding to the period when humoral antibody responses are detectable following SARS-CoV-2 infection. Participants with breakthrough infection during the Omicron-dominant period reported fewer symptoms overall compared to the pre-Omicron period, with a median of three versus four symptoms, respectively. Cough was the most common symptom during the Omicron period (82.1%), followed by sore throat (81.4%) and fever (78.6%). In contrast, loss of taste or smell was significantly more frequent in the pre-Omicron period (64.8% versus 22.9%, p < 0.05). IgG levels were significantly higher during the Omicron period than during the pre-Omicron period, increasing by 42.3%, reflecting enhanced antibody responses likely driven by repeated exposure. A consistent association between cough and elevated IgG levels was observed in both periods (p < 0.05), suggesting an association between symptom presentation and the magnitude of the early humoral response. These findings suggest that while clinical symptom profiles evolved across a major SARS-CoV-2 variant transition, certain symptom–antibody relationships remained consistent. Such associations may provide insight into how clinical manifestations relate to immune responses in populations with pre-existing immunity and may support interpretation of symptomatic infection during ongoing SARS-CoV-2 circulation. Full article

The continued evolution of SARS-CoV-2 has enabled an escape from most monoclonal antibodies, yet a subset of broadly neutralizing antibodies targeting three newly identified super-conserved RBD epitopes—SCORE-A, SCORE-B, and SCORE-C—retains remarkable activity against even the most recent JN.1-derived sublineages. Here, we employed an integrated computational framework combining conformational dynamics, mutational scanning, MM-GBSA binding energetics, and frustration profiling to dissect the molecular mechanisms by which XGI antibodies achieve broad neutralization and resistance to immune escape. Structural analysis revealed that all three SCORE epitopes share a common architecture: a highly conserved, minimally frustrated core that provides stable anchoring, flanked by peripheral regions that accommodate antibody-specific variations. Conformational dynamics showed that SCORE-A antibodies (XGI-183) rigidify the lateral epitope while leaving the RBM partially mobile; SCORE-B antibodies (XGI-198, XGI-203) clamp the RBM apex, directly blocking ACE2; and SCORE-C antibodies (XGI-171) allosterically loosen the RBM loop, impairing receptor engagement indirectly. Mutational scanning identified a hierarchical hotspot organization where primary hotspots (e.g., K356, T500, Y380, T385) are evolutionarily constrained and minimally frustrated, while secondary hotspots (e.g., V503, Y508, S383) are neutrally frustrated and represent the principal sites of immune-driven mutations. MM-GBSA decomposition revealed that van der Waals-driven hydrophobic packing dominates binding, with electrostatic interactions providing auxiliary stabilization. Critically, frustration analysis demonstrated that immune escape hotspots reside precisely in zones of neutral frustration—“energetic playgrounds” that permit mutational exploration without destabilizing the RBD—while minimally frustrated cores are evolutionarily locked. The comparative analysis of conformational versus mutational frustration distributions revealed a unifying principle: aligned neutral frustration yields permissive, escape-prone interfaces; decoupling enables the targeting of constrained cores; and the convergence of minimal frustration in both distributions creates invulnerable interfaces. These findings establish that broad neutralization arises not from ultra-high-affinity anchors but from strategic energy distribution across rigid, evolutionarily informed interfaces, providing a roadmap for designing next-generation therapeutics that target the invulnerable cores of viral surface proteins. Full article

Background: A major challenge in antiviral development is the identification of novel virus–host interactions while ensuring therapeutic efficacy and safety. These challenges have renewed interest in phytochemicals derived from medicinal plants as alternative antiviral agents. Objectives: In this study, we investigated the antioxidant, antiviral, and immunomodulatory properties of a Mediterranean Urtica dioica extract (UdE) against SARS-CoV-2 using chemical, biochemical, and in vitro approaches. Methods: The physicochemical properties of UdE were characterized using microtiter assays and HPLC analysis. Cytocompatibility was evaluated in HEK293T, Vero E6, Caco-2, and Calu-3 cell lines while antioxidant activity was assessed using both chemical and cell-based assays. Antiviral activity was evaluated by assessing inhibition of SARS-CoV-2 receptor binding domain (RBD)–ACE2 interaction using ELISA, inhibition of SARS-CoV-2 main protease (Mpro) activity via FRET assay and inhibition of viral entry using SARS-CoV-2 S1 pseudovirus neutralization assay. Results: UdE (100 µg/mL) inhibited RBD–ACE2 binding by 94% and suppressed Mpro activity by 74%, while reducing moderate but significant inhibition of pseudovirus entry (33.6%) at 300 µg/mL dose level in ACE2 expressing HEK293T cells. Immunomodulatory analysis revealed significant suppression of IL-1β and IL-6 production, accompanied by increased TNF-α and IL-8 levels. Conclusions: Collectively, these findings highlight that UdE exhibits multi-target in vitro antioxidant, antiviral, and immunomodulatory activity against SARS-CoV-2; therefore, UdE represents a promising bioactive extract for the management of SARS-CoV-2 infection. Full article

Background/Objective: This study is a cross-sectional investigation of long-term immune responses measured at different time intervals after COVID-19 infections, vaccinations, or combined exposure. The focus is on immune reactivity against recombinant spike (S) and nucleocapsid (N) protein antigens. Materials and Methods: Serum antibody levels were assessed up to four to four and a half years after infection or immunization, including virus-specific immunoglobulin G (IgG), IgA and IgM antibodies, as well as neutralizing antibodies against the S-protein. Cellular immunity was assessed by analyzing peripheral blood mononuclear cells (PBMC; n = 43 in first cohort, n = 32 in second cohort), including T-helper memory and cytotoxic subsets, and cytokine production after in vitro stimulation with recombinant SARS-CoV-2 proteins. A multiplex cytokine assay was used to analyze effector and regulatory immune responses. Results: Virus-specific IgG antibodies persisted for years after exposure to SARS-CoV-2, with IgG against the receptor-binding domain (RBD) correlating most strongly with neutralizing activity. Vaccinated individuals demonstrated higher IgA responses, whereas antibodies to the N-protein were associated with previous infection. No IgM antibodies were detected in any subjects, suggesting an immune response based on memory rather than ongoing infection. PBMCs from individuals with a history of both COVID-19 exposure and vaccination exhibited enhanced responsiveness, characterized by increased frequencies of memory T cells compared to vaccination alone. Stimulating with the S-protein induces higher cytokine production, including IFN-gamma, TNF-alfa, and IL-12(p70), compared with stimulation by the N-protein. Cytokines such as IL-10 and TGF-beta are also elevated, suggesting immune regulation rather than persistent inflammation. Conclusions: SARS-CoV-2 infection and vaccination are associated with persistent humoral and cellular immune responses detectable several years after exposure. Individuals with hybrid immunity exhibit broader and functionally enhanced immune reactivity, indicating more robust long-term immune memory. Future studies should focus on the long-term consequences of hybrid immunity and optimize other vaccine strategies, including recombinant antigen vaccines. Full article

Background/Objectives: Our previous study demonstrated that while the third SARS-CoV-2 booster effectively enhanced immunity against the Delta subvariant, its protection declined over time. This study aimed to evaluate and compare the humoral and cellular immune responses, as well as reactogenicity, of the mRNA-1273 vaccine administered as a fourth booster in healthy Thai adults previously vaccinated with two doses of CoronaVac (CV) followed by a third dose of either AZD1222 (AZ) or BNT162b2 (BNT). Methods: Participants received a single 100 µg (0.5 mL) intramuscular dose of mRNA-1273. Blood samples were collected at baseline (D0), D14, D90, and D180 to assess anti-RBD IgG, conduct a surrogate virus neutralization test (sVNT) against the Delta and Omicron variants, and assess IFN-γ levels and reactogenicity. Results: Both 2CV/AZ- and 2CV/BNT-primed groups exhibited comparable local and systemic reactogenicity. The fourth mRNA-1273 dose markedly increased Delta variant inhibition within 14 days in both groups and remained at high levels at Days 90 and 180. sVNT inhibition against Omicron rose similarly in both groups at Day 14; it declined sharply by Days 90 and 180, with the 2CV/AZ-primed group showing significantly lower levels than the 2CV/BNT-primed group. Baseline anti-RBD IgG levels were lower in the 2CV/AZ group (p = 0.003) but surpassed those of the 2CV/BNT group by Day 14, with no significant differences at later time points. IFN-γ responses followed a similar pattern to anti-RBD IgG Conclusions: A heterologous fourth mRNA-1273 booster in both 2CV/AZ- and 2CV/BNT-primed groups effectively enhances B-cell and T-cell responses against SARS-CoV-2. However, emerging variants such as Omicron may still pose challenges. The trial was registered with the Thai Clinical Trials Registry: the name of the registry: “The comparison of immune response to the 4th dose booster with mRNA-1273 COVID-19 vaccine in individuals who had received 2 doses of CoronaVac and booster with ChAdOx-1 or BNT162b2 COVID-19 vaccine”, TCTR20220205002 on 5 February 2022. Full article

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) enters host cells when the spike receptor-binding domain (RBD) engages angiotensin-converting enzyme 2 (ACE2). Cannabinoid scaffolds have recently been reported to bind S1/RBD, block spike-mediated membrane fusion, and modulate host inflammatory pathways, making them attractive candidates for entry inhibition. Here, we applied an integrated computational pipeline to prioritize cannabis-derived compounds as interfacial blockers of the RBD–ACE2 complex across variants. Eleven phytocannabinoids were docked into the wild-type (WT) RBD–ACE2 interface, identifying three cavities, with ligands preferentially occupying pocket 1. Complexes were subjected to triplicate 200 ns all-atom molecular dynamics (MD) simulations for WT, Delta, and Omicron BA.1 RBD–ACE2. Binding energetics were quantified using molecular mechanics/generalized Born surface area (MM/GBSA) and solvated interaction energy (SIE), and per-residue contributions were analyzed together with solvent-accessible surface area (SASA) and residue interaction networks. Among all compounds, cannabidiol (CBD) and cannabinol (CBN) were the only ligands that remained stably bound in pocket 1 for all variants. CBN showed the most favorable ligand–complex binding in WT, whereas CBD preserved favorable binding in Omicron BA.1 despite reduced interface burial, indicating that van der Waals/electrostatic complementarity and solvation, rather than surface coverage alone, govern affinity. Both ligands weakened modeled RBD–ACE2 binding by perturbing hot-spot residues centered on Y505 or N501Y in RBD and E37, A387, and R393 in ACE2. Overall, our results highlight CBD and CBN as tractable, variant-spanning interface disruptors and illustrate how MD-based free-energy calculations can support computational drug discovery against evolving viral protein–protein interfaces. Full article

Background and Objectives: Sleep-related breathing disorders (SRBDs) are common but often underdiagnosed in children. Early identification is essential, as untreated pediatric SRBDs can lead to cognitive, metabolic, and cardiovascular complications. This study aimed to estimate the prevalence of children at risk for SRBDs, defined as those screening positive based on Pediatric Sleep Questionnaire (PSQ) scores, and to analyze the association with potential risk factors in the general pediatric population of Lithuania. Materials and Methods: This cross-sectional study included 1929 children aged 2–17 years. Parents completed a questionnaire covering demographics, health status, and the PSQ. A validated Lithuanian version of the PSQ was used, with ≥8 (40%) positive responses indicating suspected SRBDs. Results: Overall, 14.9% of children were suspected of having SRBDs, with the highest prevalence among those aged 7–11 years (17.5%), followed by 2–6 years (14.9%) and ≥12 years (12.6%) (p = 0.032). In preschoolers (2–6 years), the strongest predictors were attention deficit hyperactivity disorder (ADHD; p < 0.001, OR 4.456, 95% CI 1.992–9.968) and allergic rhinitis (p < 0.001, OR 2.268, 95% CI 1.433–3.591). In children aged 7–11 years, endocrine diseases showed the strongest association (p < 0.001, OR 13.366, 95% CI 2.914–61.313), followed by ADHD (p = 0.001, OR 5.693, 95% CI 1.981–16.363) and adenotonsillar hypertrophy (p < 0.001, OR 3.079, 95% CI 1.839–5.156). In adolescents (≥12 years), SRBDs were primarily associated with ADHD (p < 0.001, OR 17.513, 95% CI 9.597–31.961) and endocrine diseases (p < 0.001, OR 6.214, 95% CI 2.965–13.020), while obesity remained significant (p < 0.001, OR 3.400, 95% CI 2.106–5.489). Conclusions: Approximately 15% of Lithuanian children were at risk for SRBDs. Risk factors differed by age: adenotonsillar hypertrophy in school-aged children, allergic rhinitis in preschoolers, and obesity in adolescents, whereas ADHD was associated across age groups. Full article

Background: Electromyography (EMG), video-polysomnography (vPSG), and wrist actigraphy are each used to develop diagnostic algorithms for rapid-eye-movement sleep behavior disorder (RBD). However, the extent to which they capture overlapping versus distinct motor phenomena remains unknown. We evaluated the respective contributions of actigraphy, EMG and vPSG to the measurement of REM sleep motor activity. Methods: Seventeen adults with RBD (Mount Sinai n = 9; Stanford n = 8) and eight control participants from an open Newcastle dataset underwent vPSG and concomitant wrist actigraphy. Flexor digitorum superficialis EMG activity and video-detected movements were manually scored in 3 s mini epochs. Actigraphy was quantified using an acceleration-magnitude-based activity count model. Statistical and agreement analyses were performed to assess the motor events captured by all three, any two, or by each modality independently during REM sleep. Results: In participants with RBD, actigraphy-derived movement load was significantly higher during REM sleep than during non-REM stages, a pattern not observed in control participants. REM movement load was also higher in RBD participants compared to controls, although this difference did not remain significant after correction for multiple comparisons. Across 12,941 3 s mini epochs, EMG, actigraphy, and video detected 1703, 1613, and 811 motor events, of which 413 were detected concurrently by all three modalities. Pairwise agreement was moderate and increased from EMG–actigraphy (κ = 0.27 ± 0.10) to actigraphy–video (κ = 0.41 ± 0.12) and EMG–video (κ = 0.45 ± 0.15). Of EMG-detected events, 49.0% were also detected by actigraphy; of actigraphy-detected events, 37.2% were detected by EMG and 34.9% by video. Actigraphy activity counts were highest for events detected by all three modalities and lowest for actigraphy-only events. Conclusions: Actigraphy-measured REM-related motor activity was elevated in RBD but not in controls. EMG, actigraphy, and video captured partially overlapping motor events in RBD patients, with actigraphy showing the highest sensitivity and manually scored video the lowest. Full article

In this study, the phenotypic consequences of naturally occurring single nucleotide polymorphisms (SNPs) in the Middle East respiratory syndrome coronavirus (MERS-CoV) Spike protein were investigated. The impact of Spike mutations on the syncytia formation and neutralisation of contemporary MERS-CoV strains is not currently well understood. Mutations were identified by aligning 584 MERS-CoV Spike sequences from either human clinical isolates collected between 2012 and 2024 or from a clinical isolate that had been passaged in human cells. Fifteen SNPs of interest occurring in the N-terminal domain (NTD), receptor binding domain (RBD) and adjacent to the S1/S2 cleavage site were selected for further characterisation based on their location in the Spike protein, frequency and identification in previous studies. A contemporary clade B, lineage 5 wildtype Spike sequence, obtained from a human MERS-CoV clinical isolate, was used as the backbone in this study. The mutations of interest were introduced to the wildtype backbone to generate Spike variants. Spike variants were characterised via cell–cell fusion assays, and a lentiviral pseudotyping system was used to investigate the impact of these Spike mutations on neutralisation. The I529T, E536K and L745F mutations were shown to increase fusion and syncytia formation. The L411F, T424I, L506F, L745F and T746K mutations were found to increase resistance to neutralisation by pooled patient sera. This study has identified novel naturally occurring Spike mutations that resulted in phenotypic differences in the syncytia formation and neutralisation of contemporary MERS-CoV strains. Continued investigation of the phenotypic consequences of MERS-CoV Spike mutations is essential for assessing the risk to public health, especially given the pandemic potential of this virus. Full article

As SARS-CoV-2 continues to evolve with increased transmissibility and immune evasion, the need for vaccines that provide broader and more durable protection has become increasingly urgent. The extensive research spurred by the pandemic has accelerated the development of diverse vaccine platforms, including mRNA, DNA, virus-like particles (VLPs), recombinant proteins, and mosaic mono- and polyvalent vaccines. While several of these platforms have reached regulatory approval and widespread clinical employment, others remain under evaluation or in various stages of clinical development. These vaccines have significantly reduced infection rates, severe disease, and hospitalizations, particularly among high-risk group. Nevertheless, the ongoing emergence of novel variants and subvariants has challenged the efficacy of both existing and newly developed vaccines. This evolving landscape underscores the urgent need for a universal SARS-CoV-2 vaccine platform capable of providing comprehensive and long-lasting immunity. In this review, we evaluate current and emerging strategies for SARS-CoV-2 universal vaccine development, with a focus on antigen design, breadth of immune protection, and clinical feasibility. Attention is given to various universal vaccine platforms such as the mosaic polyvalent spike construct, multi-epitope vaccines targeting the receptor-binding domain (RBD), and approaches centered on the conserved S2 subunit of the spike protein. We also discuss strategies leveraging additional conserved viral proteins and T helper (Th) and cytotoxic T lymphocyte (CTL) epitopes from across coronaviruses. By highlighting the advances in these areas, this review provides a framework to guide the rational design of next-generation universal vaccines capable of delivering broad and durable protection against SARS-CoV-2 variants. Full article

Pediatric obstructive sleep apnea (OSA) is a prevalent and underdiagnosed condition associated with significant neurocognitive, behavioral, and systemic consequences. Sleep-related breathing disorders (SRBDs) in children range from primary snoring to OSA, with even mild forms increasingly linked to adverse outcomes. Given their frequent contact with pediatric patients, pediatric dentists and orthodontists are uniquely positioned to contribute to early identification and management within a multidisciplinary framework. Objectives: This narrative review aimed to summarize and critically appraise current evidence to clarify the clinical role, scope of practice, and responsibilities of pediatric dentists and orthodontists within the multidisciplinary management of pediatric obstructive sleep apnea. A comprehensive literature search was conducted in PubMed, Scopus, Web of Science, and EMBASE up to 1 November 2025. Review articles addressing the involvement of pediatric dentists and orthodontists in pediatric OSA were included. No restriction was applied to language or publication year. Two authors independently performed study selection and data extraction. The methodological quality and data extraction of the studies were structured according to the SANRA scale. Ten studies were deemed suitable for inclusion in the current review. After examination of the full texts, the available evidence was filtered into specific clinical domains aimed at clarifying the role of the pediatric dentist and orthodontist in the management of pediatric obstructive sleep apnea (OSA). Qualitative thematic analysis of the included studies identified three main areas in which pediatric dentists and orthodontists contribute to the management of pediatric OSA. The first area involves screening through recognition of clinical signs and symptoms, use of validated questionnaires, and identification of craniofacial and occlusal features associated with increased airway risk. The second area concerns participation in the diagnostic–therapeutic pathway and multidisciplinary care, including timely referral, clinical documentation, and collaboration with pediatricians, otolaryngologists, and sleep specialists. The third area relates to orthodontic treatments such as rapid maxillary expansion and mandibular advancement appliances, which may provide adjunctive benefits in selected patients, although current evidence is limited by heterogeneity and growth-related confounding factors. Pediatric dentists and orthodontists play a pivotal yet complementary role in the management of pediatric OSA. In particular, all the involved specialists are encouraged to actively participate in the screening process, interdisciplinary communication, and diagnostic and therapeutic decision-making processes. Full article

This paper proposes a Stein Variational Gradient Descent (SVGD) solution framework for Model Predictive Control (MPC) to enhance robustness in bipedal locomotion. While conventional simplified rigid-body dynamics (SRBD) based MPC with deterministic solvers such as the active-set method performs reliably during straight-line walking, its performance degrades significantly in agile maneuvers, where modeling simplifications become more pronounced. Simulations on the 43-DoF QingLoong humanoid demonstrate that SVGD-MPC exhibits improved robustness in dynamic locomotion tasks, including high-angular-velocity turning and circular walking. However, conventional MPC fails to maintain stability. Rather than relying on a single nominal solution, SVGD-MPC maintains an ensemble of candidate control trajectories that evolve jointly within the MPC optimization process. This ensemble-based formulation reduces sensitivity to modeling inaccuracies and allows the controller to retain feasible control candidates when the assumptions underlying the SRBD model are violated. The method sacrifices marginal tracking precision to maintain viability under model inadequacy. Moreover, our framework establishes a paradigm for robust locomotion control where the traditional optimization method fails to work. Full article