Background: Drug-induced agranulocytosis is a potentially life-threatening adverse reaction. Genome-wide association studies (GWASs) in ethnic Chinese people in Taiwan and Hong Kong have shown an association between agranulocytosis induced by antithyroid drugs and the HLA alleles HLA-B*38:02 and HLA-DRB1*08:03. We aimed to identify genetic variants associated with antithyroid drug-induced agranulocytosis in a white European population. Methods: We did a GWAS in 234 European adults with any non-chemotherapy drug-induced agranulocytosis (absolute neutrophil count ≤0·5 × 109/L [≤500/μL]) and 5170 population controls. 39 of the 234 patients had agranulocytosis that was induced by antithyroid drugs (thiamazole [methimazole], carbimazole, or propylthiouracil). After imputation and HLA allele prediction, 9 380 034 single nucleotide polymorphisms (SNPs) and 180 HLA alleles were tested for association. The genome-wide significance threshold was p<5 × 10-8. Findings: Agranulocytosis induced by non-chemotherapy drugs in general was significantly associated with the HLA region on chromosome 6, with odds ratios (ORs) of 3·24 (95% CI 2·31-4·55, p=1·20 × 10-11) for HLA-B*27:05 and 3·57 (2·61-4·90, p=2·32 × 10-15) for the top SNP (rs114291795). Drug-specific analysis showed that the association with HLA-B*27:05 was largely driven by cases induced by antithyroid drugs. In a multiple logistic regression model, the OR for HLA-B*27:05 was 7·30 (3·81-13·96) when antithyroid drug-induced agranulocytosis was compared with population controls (p=1·91 × 10-9) and 16·91 (3·44-83·17) when compared with a small group of hyperthyroid controls (p=5·04 × 10-4). Three SNPs were strongly associated with antithyroid drug-induced agranulocytosis: rs652888 (OR 4·73, 95% CI 3·00-7·44, p=1·92 × 10-11) and rs199564443 (17·42, 7·38-41·12, p=7·04 × 10-11), which were independent of HLA-B*27:05, and rs1071816 (5·27, 3·06-9·10, p=2·35 × 10-9) which was in moderate linkage disequilibrium with HLA-B*27:05. In heterozygous carriers of all three SNPs, the predicted probability of antithyroid drug-induced agranulocytosis was about 30% (OR 753, 95% CI 105-6812). To avoid one case of agranulocytosis, based on the possible risk reduction if all three SNPs are genotyped and carriers are treated or monitored differently from non-carriers, roughly 238 patients would need to be genotyped. Interpretation: In white European people, antithyroid drug-induced agranulocytosis was associated with HLA-B*27:05 and with other SNPs on chromosome 6. In the future, carriers of these variants could be placed under intensified monitoring or offered alternative treatment for hyperthyroidism. Funding: Swedish Research Council, Swedish Heart and Lung Foundation, Clinical Research Support at Uppsala University, German Federal Institute for Drugs and Medical Devices, Carlos III Spanish Health Institute, European Regional Development Fund, UK National Institute for Health Research, The Selander's Foundation, Thuréus Foundation, European Commission, and Science for Life Laboratory.

Additional Metadata
Persistent URL,
Journal The Lancet Diabetes and Endocrinology
Hallberg, P, Eriksson, N, Ibanez, L, Bondon-Guitton, E, Kreutz, R, Carvajal, A, … Stephens, C. (2016). Genetic variants associated with antithyroid drug-induced agranulocytosis: A genome-wide association study in a European population. The Lancet Diabetes and Endocrinology, 4(6), 507–516. doi:10.1016/S2213-8587(16)00113-3