We read with great interest the article by Weiss et al 1 reporting genetic associations of rs632111 (fucosyltransferase 2; FUT2), rs8176693 (ABO) and rs889512 (chymotrypsinogen B2; CTRB2) with lipase levels. Weiss et al also claimed that the variants at the FUT2 and ABO loci were associated with chronic pancreatitis (CP). No association with CP was observed for the CTRB2 locus. Elevated lipase levels are a diagnostic criterion for acute pancreatitis and might mirror subclinical pancreatic injury in patients without severe complaints. Hence, variants associated with elevated serum lipase levels might also be associated with CP risk. In a recent genome-wide association study, genetic variants of CP risk were identified in PRSS1 and CLDN2-MORC4.
2A large European replication study refined these associations to alcohol-related CP. 3 However, no associations were revealed at FUT2 and ABO in the former genome-wide association study.
2Given the relatively moderate association of genetic variants with CP in the paper by Weiss et al, we analysed the above-mentioned FUT2 and ABO single nucleotide polymorphism (SNPs) regarding association with CP in a German cohort of 1458 cases (non-alcohol-related CP n=584; alcohol-related CP n=874) and 5133 controls derived from the KORA study and patients with alcohol dependence (GESGA (-) consortium) according to DSM-IV criteria to replicate the finding. Controls included 1488 individuals with alcohol consumption of >60 g/day and 1915 individuals with alcohol consumption of <20 g/day.All individuals were genotyped using Illumina SNP-chip technology. Briefly, data was filtered using Plink 1.9 at an individual-and SNP-wise call-rate >0.99 for relatedness ( pi-hat <0.185), minor allele frequency >0.01 and HardyWeinberg disequilibrium with p value >10 −6 . Imputation at 1000 Genomes reference panel ( phase 1, release 3, software SHAPEIT V.2+IMPUTE2.3.0) was performed with 279 188 high-quality SNPs available in all cohorts. Analyses were conducted with R applying logistic regression with the first three principal components of the SNP data included as covariates to account for possible population stratification. We analysed additive, recessive and dominant models of inheritance and subgroup and interaction analysis regarding alcohol consumption status.Our analyses revealed no significance for rs632111 and rs8176693 in statistical models reported previously (table 1).1 For rs632111, the same (non-significant) direction of effect was observed, while for rs8176693, the effect direction was reverse. Interaction and subgroup analysis revealed significant interaction effects of rs632111 with alcohol consumption ( p value 0.04, OR 0.82, 95% CI 0.68 to 0.99) and for rs8176693 for the subgroup of alcohol-dependent individuals ( p value 0.04, OR 0.78, 95% CI 0.62 to 0.98); however, these associations would not withstand correction for multiple testing.Next, we screened whether there were stronger associations in the vicinity of the reported SNPs. Here, the two regions did not reveal any convincin...