LETTER Hypothesis-free analysis of

wide extent of association with Crohn’s disease susceptibility We read with interest the recent paper by Plantinga et al showing a role for the threo-nine to alanine (T300A)-variant of ATG16L1 in regulating the expression of ATG16L1 and cytokine production in response to a NOD2-ligand.1 ATG16L1, notably T300A (rs2241880) in exon 9, was first implicated in Crohn’s disease (CD) susceptibility in a non-synonymous single nucleotide poly-morphism (SNP) study, followed by a tagging SNP study and regression analysis.2 These authors were unable to demonstrate differences in ATG16L1 expression based on T300A-genotype or intestinal inflammation. ATG16L1 consists of a N-terminal Atg5-interacting domain, a coiled-coil domain (involved in self-dimerisation, containing the T300A variant) and seven WD-domains (putatively interacting with NOD2) at the C-terminus. In ATG16L1-deficient and hypomorphic mice, canonical and bacteria-induced autophagy, Paneth-cell homeostasis and interleukin-1b secretion were dependent on ATG16L1.3 4 By contrast, studies focusing on T300A have shown conflicting results.1 5 6 The association signal at other susceptibility loci (eg, NOD2 and IL23R) consists of common and rare variants.7 To date, sequencing studies have not demonstrated rare non-synonymous variants within the 18 exons of ATG16L1.2 7 In order to assess the SNP association signal without any bias towards function (either coding, splicing, expression or autophagy), we performed a hypothesis-free analysis of the ATG16L1 association signal in two independent case-control cohorts. In the first cohort, 38 SNPs spanning the ATG16L1 gene were imputed for 1735 subjects (800 CD/935 HC (Healthy Controls): both from the NIDDK cohort).8 In the second cohort, 27 SNPs of the ATG16l1 locus were directly genotyped o