Vasiliy Igorevich Pustovoyt

INTRODUCTION. Contamination of injection and infusion preparations in the vials with fragments of a punctured stopper (coring) poses direct risks of microembolism, inflammatory reactions in patients, and dosing errors. No risk minimization strategy for coring of a punctured stopper currently exists in the Russian Federation. AIM. This study aimed to determine the ways to minimize contamination risk of an injectable drug with particles of a vial stopper punctured by a needle. DISCUSSION . Standardized puncturing technique for vial stoppers was found to be a fundamental preventative measure: puncturing at an angle of 45–60º, followed by a right angle, with the bevel facing upwards, significantly reduces the risk of a fragmented stopper. Using small-gauge needles (≥20G) and restricting the number of punctures (as few as possible and at least 0.75 mm apart) also minimized coring. Filter needles that trap the resulting particles were substantiated as a reliable technological solution, as well as choosing closed-loop transfer systems and prefilled single-use syringes that prevents any stopper manipulations. The important elements of the strategy include a thorough visual inspection of each dose against a dark background under intense light and training medical personnel to puncture the vials using simulation techniques. CONCLUSIONS. Based on the references, the strategy offered to minimize the risks associated with the vial stopper fragmentation at injection and infusion integrates technological, methodological, and organizational solutions. For practical implementation, it is recommended to standardize the puncturing technique of the vial stopper; to train medical personnel in puncturing techniques that reduce the risk of forming stopper particles; to visually inspect each dose of the drug before administration; and to use specific medical devices that prevent manipulations with the vial stopper.

Модели машинного обучения для стратификации тяжести черепно-мозговой травмы на основе временной динамики интерлейкина-6

Objective. To assess the diagnostic significance of the UCHL1 protein in mine-blast traumatic brain injury (TBI), to determine its correlation with the severity of brain damage, and to compare its use with conventional diagnostic methods. Material and methods. The study included 253 servicemen with mine-blast TBI of varying severity: mild (n=53), moderate (n=50), and severe (n=150). The serum level of UCHL1 was measured 1, 3, 6, and 12 hours after injury using an enzyme-linked immunosorbent assay. TBI severity was assessed using the Glasgow Coma Scale and neuroimaging data. Statistical processing included analysis of variance (ANOVA), Kruskal—Wallis test, Spearman correlation analysis, logistic regression, and ROC analysis. Results. UCHL1 levels statistically significantly correlated with TBI severity (ρ=0.66, p<0.001). The levels were significantly higher in patients with severe TBI compared with those with mild or moderate TBI. Analysis of the ROC curve showed high diagnostic sensitivity of UCHL1 for severe TBI (AUC=0.83). However, the temporal change in UCHL1 demonstrated limited predictive value, highlighting the need for a multi-biomarker approach. Conclusion. UCHL1 is a promising biomarker for the early diagnosis and stratification of patients with mine-blast TBI. However, its limited change variability reduces its value for monitoring. The combined use of UCHL1 with other biomarkers (GFAP, NSE, and MAPt) and neuroimaging can improve the accuracy of diagnosis and outcome prediction, especially during the prehospital stage of medical care.

Introduction: Traumatic Brain Injury (TBI) is a frequent cause of long‑term disability in service members who have been exposed to blast, and clinicians often lack reliable tools to quantify the actual burden of damage early after injury. Among available candidates, S100B is a calcium‑binding protein of astroglial origin that appears in the bloodstream when the blood–brain barrier is disrupted, and neuronal structures are injured. Its diagnostic and prognostic properties in the specific setting of blast‑related closed TBI, however, remain incompletely defined. The present study, therefore, focuses on military personnel with blast‑related closed TBI and examines the diagnostic and prognostic significance of serum S100 by tracking its concentration at several post‑injury time points across predefined severity groups. Methods: A prospective cohort of military patients with blast‑related closed TBI was examined, and serum S100 concentrations were measured at 1, 3, 6, and 12 hours after injury. Patients were stratified into mild, moderate, and severe TBI categories, and S100 dynamics were analysed in relation to these severity groups using standard descriptive statistics, group‑comparison tests, ROC analysis, correlation analysis, and K‑means clustering, as detailed in the Methods section. Results: Serum S100 levels peaked at 6 hours after trauma and showed a moderate correlation with TBI severity (r = 0.65, p < 0.001). Statistically significant differences in S100 concentrations were found between severity groups at all investigated time points (p < 0.001), and ROC analysis yielded an AUC of 0.81 for severity discri-mination. Diagnostic performance indices (AUC, optimal cutoff, sensitivity, specificity, positive and negative predictive values, and overall accuracy with 95% confidence intervals) are summarised in the Results, and K‑means clustering revealed distinct patient subgroups with different temporal S100 profiles, indicating the feasibility of more personalized risk stratification and management. Discussion: These findings support the use of S100 as a clinically relevant biomarker for early grading of blast‑related closed TBI and short‑term outcome prediction. Incorporating S100 into multimodal assessment algorithms may improve risk stratification and guide timely therapeutic decisions in affected patients. Conclusion: S100 can be considered a useful biomarker for early grading of brain injury severity and outcome prediction in blast‑related closed TBI. Integration of S100 into multimodal diagnostic algorithms has the potential to improve stratification of patients, inform therapeutic decision‑making, and enhance outcomes in military and civilian populations exposed to blast trauma.

Introduction. The development of parenteral depot formulations of drugs with modified release is a critical area in the treatment of chronic diseases such as cancer, schizophrenia, and diabetes. These formulations improve treatment adherence, maintain stable plasma concentrations, and reduce the frequency of injections. However, their development is fraught with technological and pharmacokinetic challenges, including nonlinear release of the active ingredient and high interindividual variability. Aim. To systematically analyze and summarize current methodological approaches to mathematical modeling of the pharmacokinetics of parenteral long-acting drug formulations. Materials and methods. A targeted literature search was conducted in PubMed, Google Scholar, and Scopus for the period 2015–2025 using Boolean operators and combinations of relevant keywords. The review includes data on clinically used long-acting parenteral formulations as well as on polymeric carriers (PLGA, hydrogels, in situ gels, and hybrid systems). To illustrate instrumental approaches, we consider compartmental PK models, PBPK models, and models based on machine learning methods. Results and discussion. The characteristic features of pharmacokinetic profiles of parenteral depot formulations are summarized, including the phases of initial burst, lag phase, controlled release, and late decay, as well as their relationship with the physicochemical properties of the carrier and the drug substance. The variability of the profile is shown to be associated with a combination of technological parameters (particle size, polymer composition and architecture), injection site and route of administration, and individual patient characteristics. The capabilities of nonlinear mixed-effects models, PBPK approaches, and machine learning-based models are demonstrated for describing inter- and intraindividual variability, performing simulations of dosing regimens, and supporting in silico optimization of formulation development. A basic system of ordinary differential equations (ODEs) is proposed that reflects sequential release from different depot fractions followed by distribution and elimination of the drug. Conclusion. Pharmacokinetic modeling of parenteral long-acting formulations is a key instrument within the MIDD (Model-Informed Drug Development) concept, enabling integration of mechanistic knowledge on release, absorption, and distribution with clinical data. The proposed basic ODE model structure can serve as a methodological framework for developing and adapting models for specific drug products, thereby supporting optimization of formulation composition and dosing regimens and reducing the extent of costly in vivo studies.

High physical activity often results in complex and ambiguous proteomic changes. This study aimed to analyze possible changes in the composition and content of proteins in plasma samples of elite athletes from sports with various activity levels and intensity. We performed proteomic profiling of blood plasma from 93 elite athletes with over eight years of continuous professional experience. High-resolution tandem mass spectra obtained in the data-independent acquisition mode were analyzed using the PLGS algorithm against a library of known protein sequences, and the PowerNovo algorithm for de novo protein sequencing. Combining the two protein identification strategies improved the completeness of the analysis and expanded the number of identified proteins. We identified changes in levels of circulating proteins that distinguish the high-intensity group from other groups by proteins involved in immune response, lipid transport and metabolism, and oxygen and iron transport. Changes in protein levels in biological samples of professional athletes may be associated with the training load intensity and type of sport.

Acute radiation syndrome (ARS) is one of the most severe systemic conditions resulting from exposure to high doses of ionizing radiation and remains among the most diagnostically challenging forms ...

Accumulating evidence indicates that epigenetic and post-transcriptional mechanisms interact to shape stress vulnerability and the adaptive capacity of the central nervous system (CNS). This review aimed to identify molecular markers with potential prognostic value for stress-induced CNS disorders. We analyzed 93 publications (2008–2025) identified in PubMed, Scopus, Web of Science Core Collection, and the Cochrane Library, including 80 original experimental and clinical studies, as well as 13 reviews and meta-analyses addressing epigenetic regulation, hypothalamic–pituitary–adrenal (HPA) axis function, CNS remodeling, and therapeutic or environmental modulation in stress-exposed models and clinical cohorts with stress-related disorders. Across studies, altered methylation of NR3C1, FKBP5, and BDNF, reduced hippocampal histone acetylation, and shifts in microRNA profiles (miR-16, miR-124, miR-132, miR-135a, miR-34c) were repeatedly associated with HPA axis dysregulation, limbic system remodeling, and phenotypes relevant to PTSD and depression. Evidence further suggests that at least some of these signatures show partial reversibility, with modulation reported after pharmacological interventions (e.g., SSRIs, histone deacetylase inhibitors, FKBP51 inhibitors, ketamine) and non-pharmacological approaches (e.g., physical activity, social support) in animal models and, to a lesser extent, in clinical and observational studies. We conclude that targeted modulation of specific epigenetic and post-transcriptional pathways supports the development of candidate biomarkers and may inform stratified prevention and treatment strategies for stress-induced CNS disorders, while acknowledging that further validation in large, well-characterized cohorts is required.

INTRODUCTION. Current methods of handling medicine regulatory documents are associated with high time cost (40-60% of labour hours), frequent documentation errors, and limited data interoperability. Neural network technologies have enabled the enhanced document preparation and a transition to full automation of the registration dossier life cycle. AIM. This study aimed to evaluate the possibility of using artificial intelligence (AI) systems in preparing a drug registration dossier. DISCUSSION. Natural language processing (NLP) models demonstrate high efficiency for the regulatory documentation. Named entity recognition (NER) systems with 89–96% entity extraction accuracy rate reduces the processing (preparation and quality review) time for documents within electronic Common Technical Document (eCTD) by 64%, but face limitations in interpreting morphologically complex terms and require annotated datasets. Without additional fine-tuning, generative models such as GPT-4, are prone to generating inaccurate facts when used in the Retrieval-Augmented Generation (RAG) architecture. Predictive systems based on graph neural networks and XGBoost ensembles demonstrate high accuracy (ROC AUC up to 0.88) when predicting drug approval; however, they cannot interpret decisions and data systematic biases. Developing document-centric platforms with NLP reduces the dossier preparation time by 60%, still, implementing an automated procedure for generating dossier sections requires an expert verification. CONCLUSIONS. The concept of integrated AI systems proves its effectiveness by reducing the document handling time by manufacturers and increasing the accuracy of decisions, which in turn speeds up the market launch of medicinal products. The prospects of introducing digital technologies are associated with overcoming definitions differences through unified ontologies. Practical implementation requires the development of unified standards for the validation of AI algorithms and adaptive systems.  

BACKGROUND: Common consequences of spinal cord injury in athletes include reduced basal metabolic rate, loss of muscle mass, impaired thermoregulation, and the development of chronic subclinical inflammation. Evidence-based recommendations for nutritional support in this group of athletes are currently lacking. AIM: This work aimed to characterize the plasma proteomic and metabolomic profiles of athletes from the Russian national wheelchair basketball team with spinal cord injury in order to develop personalized nutritional support recommendations. METHODS: Mass spectrometry analysis of blood plasma was performed in nine athletes from the Russian national wheelchair basketball team (mean age 37.9 ± 5.8 years, spinal cord injury at Th3–L1). A total of 61 proteins were identified and 40 endogenous metabolites were quantitatively measured. RESULTS: Activation of the acute-phase response was revealed (an increase in the levels of ITIH4, SERPINA3, AHSG, and ORM2 proteins by 2.3–9.6-fold), as well as a fivefold increase in apolipoprotein C-III concentration (57.78 ± 30.39 μM). Marked reductions were detected in glucogenic amino acids (alanine, serine, threonine decreased by 84%–96%), urea cycle components (arginine, ornithine decreased by 78%–88%), and Krebs cycle substrates (citrate and lactate decreased by 89%–99%). A sixfold increase in d-hydroxylysine concentration (54.022 ± 13.599 μM) indicates intensive collagen catabolism. CONCLUSION: Wheelchair basketball players exhibit a specific metabolic phenotype characterized by deficiency of amino acids involved in energy metabolism, dyslipidemia, and activation of inflammatory processes. To correct these alterations, personalized nutritional support is recommended, including increased protein intake to 2.0–2.2 g/kg, additional supplementation with L-citrulline (6–8 g/day), taurine (2–3 g/day), collagen synthesis substrates, and omega-3 polyunsaturated fatty acids (3–4 g/day).