Warning: Genome-wide association studies can be misleading. An example in hepatology

Abstract

Galichon et al.1 have found that the single-nucleotide polymorphism (SNP) rs2880301 is not a true sequence polymorphism, but rather a pseudovariant originating from a region of homology between the TPTE2 gene and sequences on the Y chromosome. We performed follow-up analysis of this SNP and concur that TPTE2 should no longer be a candidate gene for hepatocellular carcinoma (HCC) disease risk. Based on the observations of Galichon et al., we examined the genomic sequences surrounding the position of rs2880301, which is reported in the National Center for Biotechnology Information (NCBI) dbSNP2 database as an SNP in TPTE2 on chromosome 13.3 We performed a BLAST (Basic Local Alignment Search Tool) analysis similar to Galichon et al. and also showed that the region of homology between chromosomes 13 and Y extends for more than 100 kilobases, with 91.5% identity overall with nucleotide differences reported as SNPs representing differences on the two chromosomes. However, similar analysis of rs2551677, which also distinguishes HCC and cirrhosis in our study,4 confirms that it is a bona fide SNP on chromosome 2. We agree with Galichon et al. of the need to validate results from genome-wide association studies, especially those performed with high-density microarrays. Although genome-wide association studies provide a powerful means to identify disease-associated loci, they should be treated as exploratory. Definitive results depend on the validation of candidate disease-associated variants by independent, detailed investigations such as that performed by Galichon et al. Robert J. Clifford Ph.D.*, Kenneth H. Buetow Ph.D.*, * Laboratory of Population Genetics, National Cancer Institute, Bethesda, MD.