Genome-wide association study identifies 14 novel risk alleles associated with basal cell carcinoma

Abstract

Basal cell carcinoma (BCC) is the most common cancer worldwide with an annual incidence of 2.8 million cases in the United States alone. Previous studies have demonstrated an association between 21 distinct genetic loci and BCC risk. Here, we report the results of a two-stage genome-wide association study of BCC, totalling 17,187 cases and 287,054 controls. We confirm 17 previously reported loci and identify 14 new susceptibility loci reaching genome-wide significance (P<5 × 10(-8), logistic regression). These newly associated SNPs lie within predicted keratinocyte regulatory elements and in expression quantitative trait loci; furthermore, we identify candidate genes and non-coding RNAs involved in telomere maintenance, immune regulation and tumour progression, providing deeper insight into the pathogenesis of BCC.

Figure 1. Manhattan plot of stage 1 GWAS analysis of BCC from 23andMe data set.

Total stage 1 GWAS analysis included 12,945 cases and 274,252 controls. Loci with smallest P<10⁻⁶ (28 total, logistic regression) are labelled with the name of the nearest gene. Positions with P<5 × 10⁻⁸ (genome-wide significance) are shown in red. In stage 1, ten novel BCC susceptibility loci reached genome-wide significance after adjusting for genomic control, all of which are labelled in the figure with asterisks: from left to right, 3p13 (FOXP1), 3q28 (LPP), 6p21.32 (HLA-DQA2), 6p21.33 (HLA-B), 7p12.3 (TNS3), 7q22.1 (CUX1), 8q21.13 (ZBTB10), 9p22.2 (near BNC2), 19p13.3 (PLIN3), 21q22.3 (LINC00111). Four additional novel susceptibility loci—6p21.3 (NEU1), 10q24.3 (OBFC1), 6q27 (MIR3939), 6p22.3(CASC15)—were genome-wide significant in the overall meta-analysis (Table 2) and thus are not labelled in the figure.

Figure 2. Gene expression analysis for novel BCC susceptibility loci.

Processed microarray expression data were obtained from Gene Expression Omnibus (GSE53462, blue, and GSE7553, orange). Transcript levels in BCC samples were compared to levels in normal skin controls via Geo2R. Three genes—FOXP1, TNS3 and CASC15—were significantly upregulated in BCC relative to normal skin (P<0.05, linear models for microarray analysis) in both data sets.