Romain Barus

Purpose Pharmacovigilance Centres (PVCs) provide independent drug information to healthcare professionals in addition to their pharmacovigilance activities. However, their impact on clinical decision-making has been poorly studied. This study assessed the impact of a French PVC on clinical management in a tertiary university hospital (Toulouse University Hospital). Methods We conducted a retrospective observational study including all drug-related enquiries submitted to the Toulouse PVC during 2024. Impact was evaluated according to three predefined criteria: whether the enquiry was submitted before or after a clinical management decision had been made, concordance between PVC recommendations and subsequent clinical management, and documentation of the advice in the electronic patient record. Based on these criteria, impact was classified as certain, probable, or uncertain for enquiries submitted before clinical management, and as confirmatory (validating an already implemented decision) or uncertain for enquiries submitted after clinical management had already been initiated. Results A total of 128 enquiries were analysed. Most concerned adverse drug reactions (111/128), involved hospitalised patients (100/128), and originated from a broad range of clinical departments, mainly geriatrics (14/128), gastroenterology (11/128), neurology (11/128), infectious diseases (10/128) and internal medicine (10/128). Overall, clinical management was concordant with PVC recommendations in 87% of cases. Among the 92 pre-decision enquiries, PVC impact was classified as certain in 62% and probable in 24% of cases. Among the 36 post-decision enquiries, PVC advice was confirmatory in 44% of cases. The enquiries also generated 50 pharmacovigilance reports, including one potential safety signal. Conclusion PVCs support both clinical decision-making and drug safety surveillance.

The increasing use of tumor necrosis factor (TNF) inhibitors poses new challenges in terms of drug iatrogenicity and diagnosis. In this study, we focus on the description of the TNF-inhibitors-induced vasculitis. We conducted a retrospective descriptive study using the French national pharmacovigilance database to study cases of vasculitis attributable to TNF-inhibitor treatment. A total of 164 cases were analyzed, patients' mean age was 49.1 years, and 61.4% were women. The indications for TNF-inhibitor therapy were Crohn's disease (27%), rheumatoid arthritis (27%) and ankylosing spondylitis (21%). The median time to onset of vasculitis was 390 days [120-1095]. Clinical manifestations included cutaneous manifestations (88%), mainly purpura, rheumatic manifestations (27%), general symptoms (15%), renal manifestations (14%) and gastrointestinal and neurological manifestations (8% each). Skin biopsy was the most important diagnostic test, with 67.7% of cases showing vasculitis. Our study found mainly nonspecific vasculitis (54.9%), IgA vasculitis (28%) and ANCA vasculitis (7.9%). A possible association was observed between IgA vasculitis and TNF-inhibitors treatment for Crohn's disease, compared with other indications. Our study, based on national pharmacovigilance data, provides a better characterization of TNF-inhibitor-induced vasculitis. Resolution of vasculitis upon TNF inhibitor discontinuation, followed by recurrence upon reintroduction, confirms the drug's causative role and suggests that treatment should be stopped if possible. Clinicians should exercise heightened vigilance for symptom onset within the first 500 days of therapy. The atypical features of certain vasculitis, particularly from an epidemiological perspective given the unusual age at presentation, should prompt consideration of an iatrogenic etiology, including potential anti-TNF involvement.

Aim: Machine learning (ML) applications in pharmacovigilance remain limited and underexplored. Using data from the French National pharmacovigilance database (FNPV), this proof‐of‐concept study aimed to assess the feasibility of using a ML algorithm—eXtreme Gradient Boosting (XGBoost)—combined with SHapley Additive exPlanations (SHAP) analysis, to classify a well‐defined drug–adverse reaction pair (warfarin‐associated gastrointestinal bleeding) and identify risk factors for reporting this adverse reaction. Methods: We extracted individual case safety reports (ICSRs) involving warfarin from the FNPV. Following data preprocessing, the dataset was randomly split into a training set (75%) and an independent test set (25%). The XGBoost algorithm was trained to classify reports as either gastrointestinal bleeding or non‐gastrointestinal bleeding adverse reactions. Model performance was assessed on the test set using multiple metrics: accuracy, recall, precision, F1‐score and the area under the receiver operating characteristic curve (AUC‐ROC). Feature importance and directional effects were interpreted using SHAP values. Results: A total of 2025 ICSRs involving warfarin were extracted from the FNPV. After preprocessing, 1045 ICSRs were retained. On the test set, the XGBoost model demonstrated an accuracy of 0.68, recall of 0.80, precision of 0.68, F1‐score of 0.74 and an AUC‐ROC of 0.716. SHAP analysis revealed the four most influential features in predicting gastrointestinal bleeding reporting: age, gastrointestinal bleeding risk, antiplatelet and history of renal dysfunction. Conclusion: We demonstrated the feasibility of using an XGBoost algorithm combined with SHAP analysis to pharmacovigilance data for the classification of adverse drug reactions and identify predictive features of reporting, following post hoc model interpretation by pharmacologists.

Les microhémorragies cérébrales (MHC), petites lésions dont la prévalence augmenteavec l’âge sont associées à un risque augmenté de troubles cognitifs. Les mécanismes parlesquels les MHC pourraient altérer les fonctions cognitives ainsi que l’effet du sexe sur lesMHC ne sont pas bien élucidés. Le premier objectif de ce travail de thèse était d’étudier lesmécanismes mis en place après la formation d’une MHC corticale, chez des souris mâles etfemelles. Le deuxième objectif était de déterminer les effets de l’atorvastatine, médicamentconnu pour ses effets pléïotropes, sur la cognition et sur les mécanismes induits par la MHC.Six semaines après la formation d’une MHC induite par injection stéréotaxique de collagénasedans le cortex de souris C57BL/6J, des effets différents en fonction du sexe d’ordre cellulaire,protéique, génomique et métabolique étaient observés dans le cortex et l’hippocampe des sourismâles et femelles, conduisant à l’hypothèse d’une altération des réseaux corticohippocampiques,à l’origine de troubles de la cognition chez les mâles. L’atorvastatinecorrigeait les modifications induites par la MHC en fonction du sexe et améliorait les troublesde la cognition chez les mâles. Au total, cette étude a montré, à partir d’un travail précis portantsur l’impact cognitif d’une MHC, que la prise en compte du sexe dans un protocole précliniques’avère indispensable pour mieux élucider les mécanismes physiopathologiques de la maladiemodélisée dans le but de développer des traitements les plus adaptés possibles.

Background: Drug–drug interactions (DDI) are an important cause of adverse drug reactions (ADRs). Could large language models (LLMs) serve as valuable tools for pharmacovigilance specialists in detecting DDIs that lead to ADR notifications? Objective: To compare the performance of three LLMs (ChatGPT, Gemini, and Claude) in detecting and explaining clinically significant DDIs that have led to an ADR. Design: Observational cross-sectional study. Methods: We used the French National Pharmacovigilance Database to randomly extract Individual Case Safety Reports (ICSRs) of ADRs with DDI (positive controls) and ICSRs of ADRs without DDI (negative controls) registered in 2022. Interaction cases were classified by difficulty level (level-1 DDI being the easiest and level-2 DDI being the most difficult). We give each LLM (ChatGPT, Gemini, and Claude) the same prompt and case summary. Sensitivity, specificity, and F -measure were calculated for each LLM in detecting DDIs in the case summaries. Results: We assessed 82 ICSRs with DDIs and 22 ICSRs without DDIs. Among ICSRs with DDIs, 37 involved level-1 DDIs, and 45 involved level-2 DDIs. Correct responses were more frequent for level-1 DDIs than for level-2 DDIs. Regardless of difficulty level, ChatGPT detected 99% of DDI cases, and Claude and Gemini detected 95%. The percentage of correct answers to all DDI-related questions was 66% for ChatGPT, 68% for Claude, and 33% for Gemini. ChatGPT and Claude produced comparable results and outperformed Gemini ( F -measure between 0.83 and 0.85 for ChatGPT and Claude and 0.63–0.68 for Gemini) to detect drugs involved in DDI. All exhibited low specificity (ChatGPT 0.68, Claude 0.64, and Gemini 0.36) and reported nonexistent DDIs for negative controls. Conclusion: LLMs can detect DDIs leading to pharmacovigilance cases, but cannot reliably exclude DDIs in cases without interactions. Pharmacologists are crucial for assessing whether a DDI is implicated in an ADR.

Ondansetron is an antiemetic drug (AED) used to prevent and treat nausea and vomiting. The summary of product characteristics reports a rare risk of transient blindness primarily during IV injections, notably with the concomitant use of chemotherapeutic agents. We aimed to refine the characterization of ondansetron-induced blindness. We performed a descriptive and a case/non-case analysis using VigiBase®. Cases were defined as reports of adverse drug reactions (ADRs) related to blindness: amaurosis, amaurosis fugax, blindness. Non-cases were all other recorded reactions. Reporting risk of blindness-related ADRs was assessed using a disproportionality analysis and expressed as Reporting Odds Ratios (ROR). 138,315 ADRs were reported with AEDs, including 136 blindness-related ADRs, among them 44 (32.4%) with ondansetron. For ondansetron users, blindness-related ADRs occurred mainly on the first day. Out of the 25 patients with known outcomes, 18 (72.0%) were recovering or had recovered, 7 (28.0%) patients had not recovered There were no statistical differences in the number of cases for IV or oral users and for users or not of chemotherapeutic agents. Compared with other AEDs, ondansetron was associated with an increase in the reporting risk of blindness-related ADRs (ROR = 4.00 [2.79-5.72], p < 0.001). Rarely blindness can occur following intravenous or oral administration of ondansetron.

The association between dental problems and sublingual/buccal buprenorphine is unclear. We conducted an analysis of dental adverse drug reactions reported with sublingual/buccal buprenorphine in VigiBase®, the pharmacovigilance database of the World Health Organization. We performed disproportionality analyses to compare the reporting rates of dental problems with sublingual/buccal buprenorphine, compared to other buprenorphine formulations and methadone. Significant signals were considered if the lower boundary of the 95% confidence interval of the Reporting Odds Ratio (ROR) was > 1; cases were ≥ 3 and p-value p < 0.005) and compared to methadone (ROR = 6.02; 95% CI [3.21-11.30]; p < 0.005). Overreporting of dental problems was consistent in sensitivity analyses, except in Europe compared with other buprenorphine formulations and with methadone. Sublingual/buccal buprenorphine might increase the risk of reporting dental problems. However, these results do not modify the benefits of sublingual/buccal buprenorphine in the treatment of opioid use disorders.