The GLP-1 Organ Atlas

How to read this, and how not to

Source. FDA Adverse Event Reporting System (FAERS), public quarterly files aggregated by year, 2018–2025. Semaglutide is captured from 2018; tirzepatide exists only from its 2022 approval, so its earlier years are sparse and shown greyed. The 2026 quarters are not yet included.

Cohort. Reports in which the drug is the primary suspect, deduplicated by report id. Brand and generic names are mapped to ingredient (semaglutide: Ozempic, Wegovy, Rybelsus; tirzepatide: Mounjaro, Zepbound, a GIP/GLP-1 dual agonist).

Organ mapping. Each report's MedDRA Preferred Terms are mapped to organs by a curated keyword scheme, an approximation of MedDRA System Organ Class, not the licensed hierarchy. A report can touch several organs. Three categories appear as panels, not on the body: metabolic/glycaemic, general/systemic, and administration or product-use events. Thyroid, skin and musculoskeletal appear in the ranked list only (no dedicated 3D mesh here).

Disproportionality. ROR = reporting odds ratio for an organ in this drug versus all other drugs in FAERS, with a 95% confidence interval; the background is estimated from a 30% sample. The year scrubber holds that background fixed and varies only the drug's own reports, so it shows how the drug's signal emerged over time. ROR is a signal of disproportionate reporting, not an incidence, a risk, or proof of causation.

Limitations. FAERS has no denominator; reporting is voluntary and shaped by media attention, litigation, and heavy off-label weight-loss use. Duplicates and misclassification remain. Not for clinical decision-making.

Anatomy. 3D meshes adapted from BodyParts3D, © The Database Center for Life Science, licensed CC BY 4.0; geometry simplified for the browser. Packaging adapted from human-atlas.