Glucocorticoid-centered drug spectrum of non-jaw osteonecrosis, a FAERS pharmacovigilance study with independent validation in Canada Vigilance and JADER
Year: 2026
Type: article
Study design: Observational Study
Abstract: Objective: To distinguish jaw misclassification and litigation-related reporting from non-jaw osteonecrosis signals and define their reproducible spectrum. Methods: We analyzed 20,690,895 deduplicated Food and Drug Administration Adverse Event Reporting System (FAERS) reports (2004Q1-2026Q2), using report-level adjudication and four disproportionality criteria. A descriptive jaw-propensity index (JPI) measured the share of each drug's osteonecrosis reports coded as "jaw." Reporter restrictions and comparisons with Canada Vigilance and the Japanese Adverse Drug Event Report (JADER) database examined reporting bias. These external databases recorded less solicitation. Additional analyses included glucocorticoid stratification, empirical null calibration with unmasking, indication restriction of every surviving non-glucocorticoid drug, time-to-onset modeling, and machine learning. Estimates are reported with 95% confidence intervals. Results: Of 47 signals, 27 passed jaw exclusion (JPI < 0.50) and reporter restriction; the final core was unchanged across thresholds of 0.30-0.70. Tenofovir disoproxil's reporting odds ratio (ROR) fell from 157.9 (153.4-162.6) to 6.33 (5.38-7.45) among health professionals; 84.5% of cases were lawyer-sourced. Canada Vigilance (6.24, 2.00-19.4) and JADER (10.5, 6.60-16.8) were consistent with a litigation-inflated FAERS estimate. Glucocorticoids were replicated in both databases (class RORs 10.3, 54.4). Drug JPI rankings were concordant (r = 0.985, 0.960). All 34 negative controls had ROR < 1 (median 0.16); unmasking rescaled other estimates approximately 2.35-fold. Within-indication analyses detected no residual excess for mycophenolate (0.89) or voxelotor (0.97). Asparaginase combined a within-leukemia residual signal (7.14, 5.10-10.0) with mechanistic evidence and JADER replication (75.7). Conclusion: The reproducible spectrum is glucocorticoid-centered, with the asparaginase class the best-supported residual and three drugs unresolved. Jaw misclassification, reporting context, and indication explain much of the apparent spectrum. Spontaneous reports cannot establish causation; these adjudications require cohort confirmation. The JPI and cross-jurisdiction comparisons may inform analyses of other misclassification-prone endpoints.
Source: Frontiers in Pharmacology
Language: English
Subfield: Toxicology
Domain: Life Sciences
Open Access status: gold
Estimated APC paid: $3,801
OpenAlex ID: W7223155815