The association of endoplasmic reticulum aminopeptidase 2 (ERAP2) with ankylosing spondylitis (AS) was recently described in the large International Genetics of AS Consortium Immunochip study. 1 Variants in ERAP2 have also been associated with inflammatory bowel disease, psoriasis, acute anterior uveitis and birdshot chorioretinopathy. 2–5 Subsequent investigation demonstrated an association of ERAP2 with AS which was present when one conditioned on one of the two independent haplotypes of ERAP1 associated with AS or when HLA-B27-negative patients were analysed separately. 1 These two analyses provide analogous evidence for the association of ERAP2 with AS in HLA-B27-negative cases because of the genetic interaction between HLA-B27 and the AS-associated ERAP1 variants in AS cases. ERAP1 and ERAP2 are located on chromosome 5q15 in the opposite orientation. The locus is challenging to analyse because of the strong linkage disequilibrium (LD) across the locus and the epistasis between ERAP1 and HLA-B alleles associated with AS. We therefore sought to investigate the association of ERAP2 with AS in HLA-B27-positive patients. 6 This is of clinical importance because functional studies have demonstrated that the strongly AS-protective variant rs2248374 causes a functional ERAP2 protein knockout, because its G allele causes a loss of ERAP2 protein expression. 5 7 There is also a variant of ERAP2 which changes its enzyme catalytic activity and specificity (rs2549782, K392A 8). Because this is in almost complete LD with rs2248374 (1000 Genomes D 0 =1.00, r 2 =0.90), it is almost never translated in vivo. Further, the very strong LD between these markers means that analysis of rs2549782 for association would yield results almost identical to the results for rs2248374 presented below. Therefore, it is of relevance to determine whether the association of ERAP2 with HLA-B27-negative disease is also found in HLA-B27-positive cases, since ERAP inhibition may offer a novel therapeutic for AS. 9 We studied European Immunochip HLA-B27-positive patients (n=7772), HLA-B*27-positive controls (n=1204) and unse-lected European controls (n=13,578). HLA-B*27-positive status was taken as an imputed allele dosage of 0.6 or greater in the HLA-B*27 tag SNP rs116488202 that has 99% sensitivity and 99% specificity for HLA-B*27 in Europeans. 1 We used rs2248374 as a marker of the associated ERAP2 haplotype. We then constructed a range of logistic regression models that controlled for ERAP1 haplotype association and included HLA-B27 controls and unselected controls and also included the first four eigenvectors to control for any potential population stratification. Analyses including only HLA-B27-positive controls are somewhat underpowered relative to the unselected control cohort but were included for illustrative purposes. Analyses were performed using R and the glm() function. The results show that when using only HLA-B*27-positive patients with AS and either HLA-B*27-positive or unselected controls there is significant association at the ERAP2 locus (6.03×10 −4 and 3.54×10 −9 respectively; table 1 and figure 1).