Preferential picking of SNPs was conducted under the pairwise tagging option, with a minimum allele frequency of 25% and a high Illumina design score. The algorithm was set to select tags that would cover the Caucasian HapMap panel with an r2 of 0·8 or greater [11]. Furthermore, for both genes one additional custom SNP was selected on the basis of previously published association studies or presumed functionality. The following

RXDX-106 in vitro SNPs were genotyped in the IL1B gene; rs1143627 (tag), rs1143634 (tag), rs1143643 (tag) and rs1799916 (custom); IL1RN: rs11677397 (custom), rs2637988 (tag), rs408392 (tag), rs397211 (tag). DNA was extracted from whole blood samples and SNP typing was conducted using a custom Illumina goldengate bead SNP assay in accordance with the manufacturer’s recommendations (Illumina Inc., San Diego, CA, USA). Serum and BALF levels of IL-1β and IL-1Ra were determined using a multiplex suspension bead array system according to the manufacturer’s protocol (Bio-Rad Laboratories, Hercules, CA, USA). Data analysis was performed find more using the Bioplex 100 system and Bioplex Manager software version 4·1 (Bio-Rad Laboratories). The lower limit of detection was 0·3 pg/ml for IL-1β and 2·2 pg/ml for IL-1Ra. Because the variation in BALF retrieval in healthy controls was not significantly

different from retrieval in IPF patients, we did not correct for that. Genotype frequencies were tested for Hardy–Weinberg equilibrium (http://ihg2.helmholtz-muenchen.de/ihg/snps.html). Genotype and allele frequencies in the IPF group were compared with the control population using the χ2 test. Haplotypes and linkage disequilibrium (LD) were calculated (Haploview 4·1; Broad Institute of MIT and Harvard, Cambridge, MA, USA). Serum and BALF data were expressed as median and IQR. Differences in serum or BALF

concentrations between patients and controls were analysed using a Mann–Whitney U-test. For analysis of correlation, log-transformation was used to Amobarbital reach near-normal distribution. The correlation between cytokines in BALF and clinical data was assessed using Pearson’s correlation coefficients. The differences between cytokine levels in different genotypes were assessed with the Kruskal–Wallis test. Statistical analysis was performed using spss version 15·0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 5·0 (GraphPad Software, Inc., San Diego, CA, USA). Statistical significance was considered at a value of P < 0·05. Serum levels of IL-1β in IPF patients were increased significantly compared to healthy controls, while serum levels of IL-1Ra were decreased (Table 1). Furthermore, BALF levels of both IL-1β and IL-1Ra were increased significantly in IPF patients compared to healthy controls.