The analysis drew on 24,117,949 adverse-event reports available in FAERS during the study period. Records were identified under the generic names givinostat and givinostat hydrochloride and the trade name Duvyzat, and only reports assigning givinostat the primary-suspect role were included. Investigators removed duplicate records using the database case identifier and coded events with Medical Dictionary for Regulatory Activities preferred terms. The resulting series was highly concentrated geographically, with 306 reports, or 95.3%, coming from the United States. Patients were predominantly male: 308 of 321 cases, or 96.0%, were male, while one was female and 12 had missing sex. Ages were broadly distributed, but 72 reports, or 22.4%, lacked age information, and outcome data were absent in 257 cases, or 80.1%. These features constrain subgroup analysis and make the distinction between a genuinely non-serious event and an event for which no serious outcome was documented especially important.

At the system-organ-class level, investigations had the strongest reporting-odds signal, with a reporting odds ratio of 5.54 and a 95% confidence interval of 4.71 to 6.53. Gastrointestinal disorders followed with a reporting odds ratio of 4.09, with a 95% confidence interval of 3.48 to 4.80, while blood and lymphatic system disorders had a ratio of 3.43, with a 95% confidence interval of 2.51 to 4.71. The Bayesian information components for these three categories had lower 95% confidence bounds above zero, meeting the study’s signal threshold. Investigators also applied the proportional reporting ratio, Bayesian Confidence Propagation Neural Network and Multi-item Gamma Poisson Shrinker. A preferred term was considered a signal if it met the criteria of at least three of the four methods, allowing signal detection to depend on consistency across frequentist and Bayesian approaches. Twenty preferred-term signals were identified by all four methods, including platelet count decreased, thrombocytopenia, diarrhoea, nausea and increased blood triglycerides. These statistics describe reporting disproportionality relative to the FAERS background; they are not incidence rates or direct measures of a drug’s clinical risk.

At the preferred-term level, blood triglycerides increased produced the strongest signal by reporting odds ratio: 341.3, with a 95% confidence interval of 253.18 to 460.10. It appeared in 47 reports, representing 14.64% of the givinostat series. Platelet count decreased was the most frequently reported term, occurring in 96 reports, or 29.91%, and had a reporting odds ratio of 91.7, with a 95% confidence interval of 73.97 to 113.68. Diarrhoea occurred in 62 reports, or 19.31%, with a reporting odds ratio of 8.29 and a 95% confidence interval of 6.39 to 10.75. Abdominal discomfort appeared in 31 reports, or 9.66%, and thrombocytopenia in 30, or 9.35%. Nausea and vomiting were reported in 27 and 23 cases, respectively. The contrast between frequency and disproportionality was notable: decreased platelet count was the most common reported event, whereas increased triglycerides had the largest relative reporting signal. Product dose omission issue and off-label use were also identified but excluded as operational or recording-related signals rather than pharmacological safety events.

The investigators analyzed platelet count decreased and thrombocytopenia separately because MedDRA distinguishes a reduced platelet-count laboratory finding, categorized under investigations, from a clinician-coded diagnosis of thrombocytopenia, categorized under blood and lymphatic system disorders. When both terms appeared in one report, the case was retained in both term-specific analyses and the overlap was quantified; the terms were not combined into a composite endpoint. Seriousness was based solely on FAERS outcome codes for death, hospitalization, life-threatening events or other serious outcomes. Sixty-four reports, or 19.9% of the series, contained at least one such coded outcome. The other 257 reports, or 80.1%, were operationally grouped as having no documented serious outcome. They were not confirmed non-serious events, and some reports could contain more than one serious outcome, so total outcome codes can exceed the number of reports.

Unadjusted analyses suggested different seriousness patterns for laboratory and clinical platelet terms. Among the 30 reports containing thrombocytopenia, 53.33% had a documented serious outcome and 46.67% did not, whereas among the 96 reports containing platelet count decreased, 14.58% had a documented serious outcome and 85.42% did not. These are proportions within each preferred-term group, not the proportion of all serious reports in which each event occurred. Pyrexia also showed a higher documented-seriousness proportion within its own group, at 63.64%, while blood triglycerides increased had a documented-seriousness proportion of 14.89%. After a likelihood ratio analysis with Benjamini–Hochberg false-discovery-rate correction, platelet count decreased was the only preferred term that differed significantly from the overall preferred-term-entry baseline of 30.07% documented serious outcomes, with an adjusted p value of 0.024. Thrombocytopenia had an adjusted p value of 0.071, blood triglycerides increased had 0.151 and pyrexia had 0.132, meaning those nominally significant unadjusted differences did not survive correction. Their estimates remained imprecise because of the small number of reports.

Time-to-onset analysis suggested that gastrointestinal and platelet-related events could emerge at different stages after treatment began, although only 124 of the 321 reports, or 38.6%, had complete and plausible dates for calculating onset. Among non-serious diarrhoea reports, the median time to onset was 3.5 days, with an interquartile range of 0.25 to 18.25 days. Non-serious platelet count decreased had a median onset of 15.5 days, with an interquartile range of 14.25 to 32.5 days, while thrombocytopenia had a median onset of 17 days and an interquartile range of 2 to 24 days. Across evaluable adverse events, 50.79% were reported within the first 30 days. Serious platelet count decreased had a descriptive median onset of 63 days, but that group contained only four reports, compared with six non-serious reports, and the comparison was not adjusted for multiple testing. The authors therefore did not treat the timing difference as evidence that serious platelet changes consistently occur later.

Regression models did not identify age, weight or corticosteroid use as independent predictors of a documented serious outcome. The complete-case analysis included 185 reports, of which 51, or 27.6%, had a documented serious outcome. Reports involving patients aged 18 or older had a serious-outcome proportion of 35.6%, compared with 23.8% among those younger than 18, but the adjusted odds ratio was 1.64, with a 95% confidence interval of 0.79 to 3.37 and an adjusted p value of 0.182. Patients weighing more than 45 kilograms had an adjusted odds ratio of 1.21, with a 95% confidence interval of 0.60 to 2.45, while reported corticosteroid use had an adjusted odds ratio of 1.66, with a 95% confidence interval of 0.87 to 3.22. No preferred term significantly modified the reporting profile across age, weight or time-to-onset subgroups after correction. These null findings do not prove that the factors have no association: the sample was small, confidence intervals were wide and the severe missingness reduced statistical power.

In an exploratory active-comparator analysis, givinostat had a reporting odds ratio of 91.70, with a 95% confidence interval of 73.97 to 113.68, for platelet count decreased and 23.76, with a 95% confidence interval of 16.48 to 34.24, for thrombocytopenia. Vamorolone had a positive disproportionality signal for platelet count decreased, with a reporting odds ratio of 2.39 and a 95% confidence interval of 1.38 to 4.12, but not for thrombocytopenia, for which its ratio was 0.17, with a 95% confidence interval of 0.02 to 1.19. Stable reporting-odds-ratio estimates were unavailable for prednisone and deflazacort because of insufficient reports, so the analysis did not establish that givinostat had stronger platelet-related signals than either drug. Comparator results were also not head-to-head risk estimates. The exceptionally large triglyceride signal may reflect givinostat, underlying disease activity, concomitant glucocorticoid-associated metabolic effects, more intensive lipid surveillance or a combination of these factors, rather than givinostat alone.

Overall, the FAERS analysis points to platelet-related and metabolic events as the clearest reporting signals following givinostat’s launch, with platelet count decreased particularly frequent and increased blood triglycerides particularly disproportionate in the database. The adjusted seriousness result adds a nuance: a laboratory finding of decreased platelets was common but had a lower documented-seriousness proportion than the overall preferred-term-entry baseline, while clinically coded thrombocytopenia appeared more serious but did not remain significant after multiple-testing correction. Missing exposure denominators, under-reporting, stimulated reporting, incomplete outcome fields, heavy use of concomitant corticosteroids and the predominance of 2024 reports prevent the study from estimating how often these events occur or whether givinostat caused them. The authors identified platelet and lipid monitoring during early treatment as targets for structured follow-up, but the signals require validation in registries or electronic health records with more complete exposure, laboratory and clinical-outcome data before they can be treated as definitive safety estimates.

Subject of Research: Retrospective pharmacovigilance analysis of givinostat adverse-event reports for Duchenne muscular dystrophy

Article Title: A disproportionality analysis of adverse event reports for givinostat in the FDA adverse event reporting system (FAERS) database

Article References: Chen, L., Zhang, H., Wang, Y., Feng, X., Hou, Y., & Su, Q. (2026). A disproportionality analysis of adverse event reports for givinostat in the FDA adverse event reporting system (FAERS) database. Frontiers in Pharmacology, 17, Article 1954821. https://doi.org/10.3389/fphar.2026.1954821

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1954821

Keywords: givinostat, Duchenne muscular dystrophy, FAERS, pharmacovigilance, thrombocytopenia, platelet count decreased, hypertriglyceridemia, adverse-event reporting, disproportionality analysis, drug safety, time to onset, disproportionality