BMP2/BMP4 colorectal cancer susceptibility loci in northern and southern European populations.
Fernandez-Rozadilla C., Palles C., Carvajal-Carmona L., Peterlongo P., Nici C., Veneroni S., Pinheiro M., Teixeira MR., Moreno V., Lamas M-J., Baiget M., Lopez-Fernandez LA., Gonzalez D., Brea-Fernandez A., Clofent J., Bujanda L., Bessa X., Andreu M., Xicola R., Llor X., Jover R., EPICOLON Consortium None., Castells A., Castellvi-Bel S., Carracedo A., Tomlinson I., Ruiz-Ponte C.
Genome-wide association studies have successfully identified 20 colorectal cancer susceptibility loci. Amongst these, four of the signals are defined by tagging single nucleotide polymorphisms (SNPs) on regions 14q22.2 (rs4444235 and rs1957636) and 20p12.3 (rs961253 and rs4813802). These markers are located close to two of the genes involved in bone morphogenetic protein (BMP) signaling (BMP4 and BMP2, respectively). By investigating these four SNPs in an initial cohort of Spanish origin, we found substantial evidence that minor allele frequencies (MAFs) may be different in northern and southern European populations. Therefore, we genotyped three additional southern European cohorts comprising a total of 2028 cases and 4273 controls. The meta-analysis results show that only one of the association signals (rs961253) is effectively replicated in the southern European populations, despite adequate power to detect all four. The other three SNPs (rs4444235, rs1957636 and rs4813802) presented discordant results in MAFs and linkage disequilibrium patterns between northern and southern European cohorts. We hypothesize that this lack of replication could be the result of differential tagging of the functional variant in both sets of populations. Were this true, it would have complex consequences in both our ability to understand the nature of the real causative variants, as well as for further study designs.