Post a new comment on this article
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Why should this posting be reviewed?
See also Guidelines for Comments and Corrections.
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.close
Caution in Conclusion
Posted by alisond on 10 Aug 2010 at 08:23 GMT
We note that this study reports the same SNP, rs4784227, that we found to exhibit the strongest evidence in our study of >27,000 subjects of European, Asian and African ancestry. (Udler MS. et al. Fine scale mapping of the breast cancer 16q12 locus. Hum. Mol. Genet. 19, 2507-15, doi: 10.1093/hmg/ddq122). However, we are concerned by the interpretation of this present study: that rs4784227 is functionally related to breast cancer risk.
The SNPs examined here were selected from HapMap phase II (http://hapmap.ncbi.nlm.ni...), but this is an incomplete catalogue of common variants. In our study we identified 26 highly correlated, candidates for causative variant via our own re-sequencing coupled with the 1000 Genomes Project (http://www.1000genomes.or...). Only seven of these 26 are present in HapMap phase II.
Although SNP rs4784227 gives the strongest signal among those examined, this does not imply that the others can be definitively excluded. Of our 26 original candidates, we were able to exclude 12 as being at least 100 times less likely than the best. Within the remaining 14 candidates, although rs4784227 had the highest likelihood, nine others were, statistically, only marginally less likely to be causative. Four of these are almost perfectly correlated (r2 ≥ 0.96) in both Europeans and Asians and are thus unlikely to be separable by epidemiological studies alone. Notably, five of the remaining 14 candidates are tightly clustered in open chromatin in the evolutionarily conserved gap between TOX3 and LOC643714 and three of these are also predicted to affect transcription factor binding sites. Although SNP rs4784227 lies in a predicted C/EBP binding site, it falls outside this conserved region of open chromatin.
In most post-GWAS fine scale mapping studies to date, evidence clearly identifying directly causative variants has been elusive - hampered by strong linkage disequilibrium and small effect sizes. Where there are multiple, tightly linked candidates, exemplifying that any one is functional does not prove its causality. Any conclusions should be cautious until combined epidemiological and functional studies can determine the exact variant, or variants, directly responsible for breast cancer risk at this important locus.
Alison M. Dunning
Miriam S. Udler
Douglas F. Easton
We read with interest the comment by Dr. Dunning et al on our paper regarding the association of rs4784227 (16q12.1) with breast cancer risk in Asians and in vitro functional characterization of this SNP (Long J et al. Identification of a functional genetic variant at 16q12.1 for breast cancer risk: Results from the Asia Breast Cancer Consortium. PLoS Genet. 2010;6:e1001002). As described in the paper, our study consists of a multi-stage GWAS and a series of in vitro functional analyses that specifically evaluated the potential functionality of rs4784227. The DNA constructs with the T allele for the luciferase reporter assays were generated by site-directed mutagenesis from the DNA constructs with the C allele, and the sequences of oligonucleotide probes used in the electrophoretic mobility shift assay were identical except for the rs4784227 site. To minimize potential laboratory assay errors and chance findings, we used four cell lines for the luciferase reporter assays and three cell lines for the electrophoretic mobility shift assay. These in vitro experiments have shown consistently that the T allele of rs4784227 reduced luciferase activities and altered DNA-nuclear protein interactions compared to the C allele (Figure 3), suggesting that rs4784227 may be functionally significant.
As described in the paper, three SNPs at 16q12.1 (rs8051542, rs12443621, and rs3803662) were reported from previous GWAS conducted in women of European ancestry (Easton DF et al. Genome-wide association study identifies novel breast cancer susceptibility loci. Nature. 2007 Jun 28;447:1087-93). However, the SNP (rs4784227) identified in our GWAS is not in LD with rs12443621 (r2=0.07) and in weak LD with rs3803662 (r2=0.14) and rs8051542 (r2 = 0.37) in Asians (Table 3 and Figure 2). Furthermore, the association of breast cancer with rs4784227 remained highly statistically significant after adjusting for all three previously reported SNPs (Table 4). In fact, in Asian women, rs12443621 was not associated with breast cancer risk, and the association of breast cancer risk with rs8051542 and rs3803662 was weaker than the one observed for rs4784227 (Table 4). Adjusting for rs4784227 substantially attenuated the association with rs3803662 and completely eliminated the association with rs8051542. Similarly, among European-American women, rs4784227 was found to be associated with breast cancer risk even after adjusting for rs12443621 or rs8051542 at P < 0.05 (Table 5). SNPs rs3803662 and rs4784227 are highly correlated in European descendants (r2=0.86). Including these two SNPs in the same model eliminated the association with rs3803662 but did not alter the risk estimate for rs4784227, although the statistical test was no longer significant (Table 5). As reported in our paper, rs12443621 and rs8051542 were weakly associated with breast cancer risk in European-Americans. Evidently, of the SNPs evaluated in this locus, rs4784227 showed the strongest and most robust association with breast cancer risk in both Asians and European-Americans.
In an extensive search for the causal variant directly responsible for the observed associations at 16q12.1, Udler et al have identified 14 variants that are all strong candidates for being causally important for breast cancer risk (Udler MS et al. Fine scale mapping of the breast cancer 16q12 locus. Hum Mol Genet. 2010 Jun 15;19:2507-15). The authors cautioned that although rs4784227 showed the strongest signal as the potential causal variant for breast cancer, the other 13 remaining candidates could not be excluded due to the LD structure in this region. The authors called for larger epidemiological studies among women of Asian and African ancestry to further evaluate this important locus. It is possible that multiple causal variants may exist at this locus. Nevertheless, results from Udler et al’s study support our finding suggesting that rs4784227 is an important candidate for future epidemiological studies and functional characterization.