The computational methods presently united by the concept of artificial intelligence began to form almost at the time of the emergence of mathematics. In recent decades, artificial intelligence has gained tremendous momentum and has become actively used in various fields, including pharmacy.

The use of artificial intelligence in the life cycle of a medicinal product is the topic of this interview with Andrey V. POGREBNYAK, Doctor of Chemistry and Candidate of Pharmaceutical Sciences, Head of the Information Technology Department and Professor of the Department of Physical and Colloid Chemistry of the Pyatigorsk Medical and Pharmaceutical Institute (Branch of the Volgograd State Medical University).

Scientific relevance. Currently, machine learning (ML) methods are widely used in the research and development of new pharmaceuticals. ML methods are particularly important for assessing the safety of pharmacologically active substances early in the research process because such safety assessments significantly reduce the risk of obtaining negative results in the future.

Aim. This study aimed to review the main information and prediction resources that can be used for the assessment of the safety of pharmacologically active substances in silico.

Discussion. Novel ML methods can identify the most likely molecular targets for a specific compound to interact with, based on structure–activity relationship analysis. In addition, ML methods can be used to search for potential therapeutic and adverse effects, as well as to study acute and specific toxicity, metabolism, and other pharmacodynamic, pharmacokinetic, and toxicological characteristics of investigational substances. Obtained at early stages of research, this information helps to prioritise areas for experimental testing of biological activity, as well as to identify compounds with a low probability of producing adverse and toxic effects. This review describes free online ML-based information and prediction resources for assessing the safety of pharmacologically active substances using their structural formulas. Special attention is paid to the Russian computational products presented on the Way2Drug platform (https://www.way2drug.com/dr/).

Conclusions. Contemporary approaches to the assessment of pharmacologically active substances in silico based on structure–activity relationship analysis using ML methods provide information about various safety characteristics and allow developers to select the most promising candidates for further in-depth preclinical and clinical studies.

Scientific relevance. Studies of the toxicological and pharmacokinetic properties of medicinal compounds are a crucial stage of preclinical research; unsatisfactory results may invalidate further drug development. Therefore, the development of in silico methods for a preliminary pre-experimental assessment of toxicological and pharmacokinetic properties is a relevant and crucial task.

Aim. The study aimed to review current approaches to in silico prediction of the absorption, distribution, metabolism, excretion, and toxicity (ADMET) parameters of pharmacologically active compounds, in particular, the most important toxicological and pharmacokinetic parameters, and to present the results of the authors’ own research in this area.

Discussion. According to the review of models for predicting the toxicological properties of chemical compounds (acute toxicity, carcinogenicity, mutagenicity, genotoxicity, endocrine toxicity, cytotoxicity, cardiotoxicity, hepatotoxicity, and immunotoxicity), the accuracy of predictions ranged from 74.0% to 98.0%. According to the review of models for predicting the pharmacokinetic properties of chemical compounds (gastrointestinal absorption; oral bioavailability; volume of distribution; total, renal, and hepatic clearance; and half-life), the coefficient of determination for the predictions ranged from 0.265 to 0.920. The literature review showed that the most widely used methods for in silico assessment of the ADMET parameters of pharmacologically active compounds included the random forest method and the support vector machines method. The authors compared the literature data with the results they obtained by modelling 12 toxicological and pharmacokinetic properties of chemical compounds using the consensus method in the IT Microcosm system and artificial neural networks. IT Microcosm outperformed the models described in the literature in terms of predicting 2 toxicological properties, including carcinogenicity and blood–brain barrier penetration (the prediction accuracy reached 93.4%). Neural network models were superior in predicting 4 toxicological properties, including acute toxicity, carcinogenicity, genotoxicity, and blood–brain barrier penetration (the prediction accuracy reached 93.8%). In addition, neural network models were better in predicting 3 pharmacokinetic properties, including gastrointestinal absorption, volume of distribution, and hepatic clearance (the coefficient of determination reached 0.825).

Conclusions. The data obtained suggest that artificial neural networks are the most promising and practically significant direction for the development of in silico systems for predicting the ADMET characteristics of new medicinal products.

Scientific relevance. Medication adherence is an important condition for effective and safe treatment. The adherence of patients to prescriptions is tracked by assessing their condition, counting the pills taken, and using other indirect methods. Digital technologies can help healthcare providers improve their patients’ medication adherence.

Aim. The authors aimed to review literature describing the medication adherence impact on treatment effectiveness, as well as digital solutions accompanying pharmacotherapy.

Discussion. Poor adherence to treatment is a significant risk factor for patients. The most common examples of poor adherence are omissions and delays in the timing of doses. Compared with classical daily dosing, individualised regimens significantly increase the risk of adherence errors. Significant consequences of non-adherence include exacerbation of the disease, insufficient effectiveness of treatment, adverse drug reactions, and drug resistance. Promising hardware and software approaches to supporting medication adherence include innovative technological solutions (pillboxes, bottles with electronic reminder systems, digital pills, and smart medication adherence monitoring systems), mobile apps, and chatbots.

Conclusions. Digital solutions to support pharmacotherapy help improve patients’ adherence to their dosing regimens and individualise their treatment. Further research is needed to select the most promising areas and develop novel digital technologies.

Scientific relevance. Favipiravir is an antiviral RNA polymerase inhibitor used to treat COVID-19. An adverse drug reaction associated with the use of favipiravir is renal disorder.

Aim. This study aimed to investigate favipiravir nephrotoxicity by assessing its effects on the integrity of a monolayer formed by renal proximal tubular epithelial cells (RPTECs).

Materials and methods. This study focused on an RPTEC monolayer culture that was seeded at a density of 6×10⁴ cells/cm² on plates with membrane inserts with 0.4 μm pores. Favipiravir was added to the plate wells at a concentration of 5, 10, or 15 μg/mL. The nephrotoxicity evaluation relied on measuring the transepithelial electrical resistance (TEER) of the RPTEC monolayer. A TEER value of 120–140 Ω×cm² was considered an indication of nephrotoxicity.

Results. RPTEC incubation with favipiravir led to a dose-dependent decrease in the TEER values. However, the TEER values after 6 days of incubation ranged within 250–280 Ω×cm² and were above the critical threshold

of 120–140 Ω×cm².

Conclusions. The results of this study indicate that favipiravir has no pronounced effect on the TEER of the RPTEC monolayer.

Scientific relevance. Vancomycin and linezolid are the antibacterial agents of choice for severe infections caused by multidrug-resistant pathogens, including methicillin-resistant Staphylococcus aureus (MRSA). However, few studies have been conducted in Russia to analyse the safety of these medicinal products.

Aim. The study aimed to compare the safety of vancomycin and linezolid using the Moscow segment of the Russian Federal Service for Surveillance in Healthcare’s database for adverse drug reaction (ADR) reports.

Materials and methods. The study used information from the spontaneous reporting database for 2018–2022, which contained 147 ADR reports for vancomycin (122 reports) and linezolid (25 reports). The authors analysed the ADR distribution and assessed the statistical significance of the identified differences by sex, weight, and age of patients, conditions of medical care, route of administration, single dose, daily dose, therapy duration, ICD-10 codes, ADR severity, and ADR outcome.

Results. The distribution of adverse reactions to vancomycin and linezolid by patient age was relatively uniform. Outpatient linezolid was associated with a significantly higher rate of ADRs (3 of 5 reports) than outpatient vancomycin (21 of 129 reports; p=0.0408). For ADR severity, 5 of 20 ADRs reported with linezolid required hospitalisation or prolongation of hospitalisation—considerably more than with vancomycin (16 of 94 reports; p=0.528). The average single dose of vancomycin (794 mg) was higher than that of linezolid (467 mg; p=0.007); the same was noted for average daily doses (1273 mg vs 998 mg; p=0.3664). The mean duration of treatment with linezolid before ADR onset was 5.26 days, which was significantly longer than the mean duration of treatment with vancomycin (2.44 days; p=0.0053). Oral linezolid was associated with a significantly higher ADR rate (4 of 19 cases) than oral vancomycin (5 of 96 cases; p=0.0027).

Conclusions. The ADRs observed with vancomycin and linezolid were predictable and class-specific. According to the results of the ADR report analysis, adverse reactions to vancomycin and linezolid were associated with different factors. Similar results of the literature analysis confirmed this conclusion. However, according to the results of the linear regression analysis, none of the factors considered in this study had a statistically significant influence on the probability of developing an adverse reaction to vancomycin or linezolid.

Scientific relevance. The main cause of cardiovascular pathologies is atherosclerosis, which is secondary to lipid metabolism disorders, in particular, the accumulation of low-density lipoprotein (LDL) cholesterol. Dyslipidaemia treatment with the largest evidence base predominantly includes statins in combination therapy, but their use is limited by into lerance in some patients. Alternatively, the treatment may include proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors.

Aim. The study aimed to analyse the applicability of PCSK9 inhibitors in patients with statin intolerance.

Discussion. According to the literature analysis, the most common presentation of statin intolerance is statin-associated muscle symptoms. The pathogenesis of statin-associated adverse events is mainly mediated by HMGCoA reductase inhibition, treatment effects on cellular and subcellular processes and skeletal muscles, and patients’ genetic makeup. The mechanism of action of PCSK9 inhibitors is entirely different and involves binding and inactivation of the PCSK9 protein, which lowers blood LDL cholesterol levels. PCSK9 inhibitors have been associated with some adverse drug reactions, most notably immunogenicity; however, PCSK9 inhibitors effectively reduce LDL levels even if patients develop antibodies.

Conclusions. Therefore, PCSK9 inhibitors are a safe, well-tolerated, and effec tive therapeutic strategy for hyperlipidaemia in patients with statin intolerance.

Scientific relevance. Valproic acid (VPA) is a psychotropic medicinal product, which may be associated with serious adverse drug reactions (ADRs). While pharmacogenetics and pharmacometabolomics can significantly affect the safety of valproates, there are no unified approaches to predicting, preventing, and correcting VPA-induced ADRs.

Aim. This study aimed to collate the results of national and international studies on toxic VPA metabolites and to develop a novel personalised approach to assessing the safety and risks of valproate therapy in real-world clinical practice.

Discussion. This study analysed national and international publications reflecting the results of preclinical and clinical studies on toxic VPA metabolites submitted to e-Library, PubMed, Scopus, and Google Scholar in 2012–2022. The inclusion criteria were full-text original articles, systematic reviews, meta-analyses, Cochrane reviews, and clinical cases in Russian or English. According to the analysis results, VPA has 20 studied toxic metabolites, which result from hepatic VPA metabolism involving P-oxidation, acetylation (β-oxidation), and glucuronidation enzymes. The functional activity of these enzymes is genetically determined and associated with heterozygous or homozygous carriage of non-functional/low-function single-nucleotide variant alleles in genes encoding these enzymes. The safety of VPA and its compounds can be improved by transferring the results of preclinical and clinical studies into real-world clinical practice using pharmacogenetics-informed pharmacometabolomics. Pharmacogenetics-informed pharmacometabolomics is a novel and personalised approach that helps, based on pharmacogenetic profiling, identify patients at high risk of VPA-induced ADRs, individually select starting and target doses of VPA and its compounds, determine the timing and frequency for therapeutic drug monitoring and monitoring toxic VPA metabolites in biological fluids (blood, saliva, and urine), and select a strategy for the prevention and correction of VPA-induced ADRs, taking into account patients’ individual pharmacometabolic profiles.

Conclusions. The quality of medical care for patients with neurological diseases and mental disorders will improve with proper monitoring of VPA-induced ADRs by all entities involved in the medicinal product life cycle; active involvement of neurologists and psychiatrists in the prediction, prevention, and monitoring of the safety of valproate treatment; and inclusion of specific sections on practical pharmacogenetics-informed pharmacometabolomics and pharmacovigilance in the professional training curricula for neurologists and psychiatrists.

Scientific relevance. On 6 December 2022, an updated version of the Rules for Good Pharmacovigilance Practice of the Eurasian Economic Union (EAEU GVP) came into force. The greatest changes were made to the requirements for pharmacovigilance documents, particularly the risk management plan (RMP). In practice, the changed EAEU GVP has resulted in multiple errors, creating the need to thoroughly analyse their structure and causes and to develop recommendations for their prevention.

Aim. This study aimed to identify, analyse, and collate inconsistencies between the information submitted by marketing authorisation holders in their RPMs and the updated EAEU GVP requirements.

Materials and methods. The Scientific Centre for Expert Evaluation of Medicinal Products analysed 50 RMPs received after 6 December 2022 as part of registration dossiers aimed to support marketing authorisation applications and/or align the registration dossiers with the EAEU requirements.

Results. The errors made by applicants when preparing RMPs were categorised according to their influence on the interpretation of a medicinal product’s safety profile. The errors leading to incorrect safety profile interpretations were considered type 1 errors (63% of the cases). The errors affecting the perception of the RMP but not the interpretation of the safety profile (e.g., grammatical errors, notes and comments by applicants, incorrect translation of terms) were deemed type 2 errors (37% of the cases). The majority of EAEU GVP noncompliance cases were detected in Part II of the RMP, the section providing the most information on the safety profile of a medicinal product.

Conclusions. There are several ways to improve the quality of RMP preparation. The information included in the RMP should be compared with the information provided in the registration dossier. The RMP should be incorporated into the integrated pharmaceutical quality system according to the requirements of good practices.

A responsible employee of the marketing authorisation holder’s quality assurance system should control the final RMP version. Employees of pharmacovigilance departments should receive regular training.

This article is the Russian translation of the Plain Language Summary (PLS) of the Cochrane Review previously published in the Cochrane Database of Systematic Reviews. Original publication: Schmidt AF, Carter J-PL, Pearce LS, Wilkins JT, Overington JP, Hingorani AD, Casas J. PCSK9 monoclonal antibodies for the primary and secon dary prevention of cardiovascular disease. Cochrane Database of Systematic Reviews. 2020, Issue 10. Art. No.: CD011748. https://doi.org/10.1002/14651858.CD011748.pub3