Post-marketing safety signal profile of mavacamten in the FDA adverse event reporting system (FAERS): a disproportionality analysis

Figures

• 

  FIGURE 1. Data extraction and study flow diagram. Flow diagram showing the identification of mavacamten-associated reports from the FDA Adverse Event Reporting System (FAERS) quarterly open data files, 2022 Q2 to 2025 Q2 (13 quarters; 386 quarterly data files). After file audit and extraction, 4,514,490 unique cases were retained following global deduplication (ROW_NUMBER() OVER (PARTITION BY caseid ORDER BY receivedate DESC, version DESC)). Mavacamten reports were identified using the drug names “MAVACAMTEN” and “CAMZYOS” with primary suspect or secondary suspect role codes (2,933 cases); the remaining 4,511,557 reports formed the background. A total of 955 unique preferred terms were evaluated. Applying the primary criterion (a ≥ 3, ROR 95% CI lower bound > 1, and IC025 > 0) yielded 80 primary signals (68 clinical, 9 administrative/product-use, 3 ambiguous) and 875 non-signals; 45 signals were concordant across all four algorithms. Exploratory adjusted ROR models were fitted for all 80 signals. ROR, reporting odds ratio; PRR, proportional reporting ratio; IC, information component; EBGM, empirical Bayes geometric mean; aROR, adjusted reporting odds ratio.

• 

  FIGURE 2. Volcano plot of the 80 primary mavacamten-associated disproportionality signals. Each point represents one preferred term meeting the primary signal criterion (a ≥ 3, ROR 95% CI lower bound > 1, and IC025 > 0). The x-axis shows the effect size on the log2 (ROR) scale, with a vertical reference line at log2 (ROR) = 0 (ROR = 1, no association). The y-axis shows IC025, the lower 95% credibility bound of the information component, with a horizontal reference line at IC025 = 0. Colors distinguish clinical signals with a ≥ 5 reports (red), clinical signals with a = 3–4 reports (blue), administrative/product-use PTs (orange, n = 9), and ambiguous PTs (teal, n = 3); 80 points are plotted in total (68 clinical). Selected PTs (top 10 by ROR and top 10 by report count) are labeled. ROR, reporting odds ratio; IC, information component; PT, preferred term.

• 

  FIGURE 3. Forest plot of 24 prioritized clinical signals associated with mavacamten. Forest plot showing 24 clinical preferred terms with at least five reports that met the primary signal criterion, prioritized by biological plausibility (cardiac, neurological, and vascular events) rather than by ROR ranking alone. Points represent the ROR and horizontal lines the 95% confidence intervals on a logarithmic scale; the vertical reference line indicates ROR = 1.0. Estimates based on small report counts (e.g., atrial flutter, a = 5) have wide confidence intervals and are displayed for completeness rather than emphasis. ROR, reporting odds ratio; CI, confidence interval; PT, preferred term.

• 

  FIGURE 4. Exploratory comparison of crude and adjusted reporting odds ratios for ten selected preferred terms. Paired comparison plot showing the crude ROR and the multivariable logistic regression-adjusted ROR (aROR) for ten selected mavacamten-associated preferred terms (ejection fraction decreased, atrial fibrillation, cardiac failure, dyspnoea, fatigue, syncope, dizziness, palpitations, chest pain, and ventricular tachycardia) on a logarithmic scale. Adjusted models included age category (unknown as a separate category), sex, report year (centered at 2023), and healthcare-professional reporter status as covariates. Model-level covariate missingness was 42.4% for age and 16.3% for sex, and HCM severity was unavailable; the adjusted analysis is therefore exploratory. Crude and adjusted estimates were closely concordant for all displayed PTs, supporting the robustness of the primary disproportionality findings. ROR, reporting odds ratio; aROR, adjusted reporting odds ratio; CI, confidence interval; PT, preferred term.