Search

A single previous study has demonstrated significant association of psoriasis with copy number of β-defensin genes, using DNA from psoriasis cases and controls from Nijmegen and Erlangen. In this study, we attempted to replicate that finding in larger new cohorts from Erlangen (N=2,017) and Michigan (N=5,412), using improved methods for β-defensin copy number determination based on the paralog ratio test, and enhanced methods of analysis and association testing implemented in the CNVtools resource. We demonstrate that the association with psoriasis found in the discovery sample is maintained after applying improved typing and analysis methods (P=5.5 × 10(-4), odds ratio (OR)=1.25). We also find that the association is replicated in 2,616 cases and 2,526 controls from Michigan, although at reduced significance (P=0.014), but not in new samples from Erlangen (1,396 cases and 621 controls, P=0.38). Meta-analysis across all cohorts suggests a nominally significant association (P=6.6 × 10(-3)/2 × 10(-4)) with an effect size (OR=1.081) much lower than found in the discovery study (OR=1.32). This reduced effect size and significance on replication is consistent with a genuine but weak association.

Recently, a deletion of two late cornified envelope (LCE) genes within the epidermal differentiation complex on chromosome 1 was shown to be overrepresented in 1,426 psoriasis vulgaris (PsV) patients of European ancestry. In this study, we report a confirmation of this finding in 1,354 PsV patients and 937 control individuals of German origin. We found an allele frequency of the deletion of 70.9% in PsV patients and of 64.9% in control individuals (chi(2)=17.44, P=2.97 x 10(-5), odds ratio (95% confidence interval)=1.31 (1.15-1.48)). The overall copy number of the two LCE genes had no influence on the age of onset, but we observed a dosage effect at the genotype level. There was no evidence of statistically significant interaction with copy number of the beta-defensin cluster on 8p23.1 or with an IL-23R pathway variant in a combined data set of German and Dutch individuals, whereas evidence for interaction with the PSORS1 risk allele in German individuals was marginal and did not remain significant after correction for multiple testing. Our study confirms the recently published finding that the deletion of the two LCE genes is a susceptibility factor for PsV with dosage effect, while, because of power limitation, no final conclusion regarding interaction with other PsV risk factors can be made at this stage.