Cross-site validation of allopurinol SCAR prevention remains difficult because genotype, kidney function, and severe adverse-event data are sensitive and rarely pooled at patient level.

Hypothesis A federated validation framework using homomorphically encrypted pooled sufficient statistics for HLA-B*58:01 status, CKD stage, starting dose, and SCAR outcomes will preserve discrimination and clinically useful calibration of allopurinol prevention rules compared with conventional centralized validation.

Rationale

• HLA-B*58:01 is the strongest established inherited risk signal for allopurinol SCAR.
• CKD and dose strategy likely modify preventable risk at the bedside.
• Decentralized cohorts need privacy-preserving methods because genotype-linked adverse-event registries are hard to share.

Testable design

• Multi-center retrospective/prospective gout cohorts with site-local outcome adjudication for DRESS/SJS/TEN.
• Compare centralized patient-level validation vs encrypted site-level sufficient-statistic aggregation.
• Primary metrics: AUROC, calibration slope, expected calibration error, decision-curve net benefit.
• Secondary analysis: subgroup performance by CKD stage and starting-dose tier.

Falsification criteria The hypothesis fails if encrypted aggregation materially degrades discrimination or calibration versus patient-level validation, or if CKD-adjusted screening adds no incremental value over HLA-B*58:01 alone.

Limitations SCAR is rare, so event scarcity may widen uncertainty. Privacy-preserving success would not eliminate the need for high-quality phenotype adjudication at each site.

References

1. Hung SI, Chung WH, Liou LB, et al. HLA-B*5801 allele as a genetic marker for severe cutaneous adverse reactions caused by allopurinol. Proc Natl Acad Sci U S A. 2005;102(11):4134-4139. DOI: 10.1073/pnas.0409500102
2. Hershfield MS, Callaghan JT, Tassaneeyakul W, et al. CPIC guidelines for HLA-B genotype and allopurinol dosing. Clin Pharmacol Ther. 2013;93(2):153-158. DOI: 10.1038/clpt.2012.209
3. FitzGerald JD, Dalbeth N, Mikuls T, et al. 2020 ACR Guideline for the Management of Gout. Arthritis Care Res (Hoboken). 2020;72(6):744-760. DOI: 10.1002/acr.24180