2008 Articles

Discovering interactions among BRCA1 and other candidate genes associated with sporadic breast cancer

Lo, Shaw-Hwa; Chernoff, Herman; Cong, Lei; Ding, Yuejing; Zheng, Tian

Analysis of a subset of case-control sporadic breast cancer data, [from the National Cancer Institute's Cancer Genetic Markers of Susceptibility (CGEMS) initiative], focusing on 18 breast cancer-related genes with 304 SNPs, indicates that there are many interesting interactions that form two- and three-way networks in which BRCA1 plays a dominant and central role. The apparent interactions of BRCA1 with many other genes suggests the conjecture that BRCA1 serves as a protective gene and that some mutations in it or in related genes may prevent it from carrying out this protective function even if the patients are not carriers of known cancer-predisposing BRCA1 mutations. The method of analysis features the evaluation of the effect of a gene by averaging the effects of the SNPs covered by that gene. Marginal methods that test one gene at a time fail to show any effect. That may be related to the fact that each of these 18 genes adds very little to the risk of cancer. Analysis that relates the ratio of interactions to the maximum of the first-order effects discovers significant gene pairs and triplets.
Breast cancer (MIM 114480) has complex causes. Known predisposition genes explain <15% of the breast cancer cases. It is generally believed that most sporadic breast cancers are triggered by unknown combined effects, possibly because of a large number of genes and other risk factors, each adding a small risk toward cancer etiology. Progress in seeking breast cancer genes other than BRCA1 and BRCA2 has been slow and limited because the individual risk due to each gene is small. This difficulty may be partly due to the fact that current methods rely largely on marginal information from genes studied one at a time and ignore potentially valuable information because of the interaction among multiple loci. Because each responsible gene may have a small marginal effect in causing disease, it is likely that such methods will fail to capture many responsible genes by studying a dataset where the disease may be due to a variety of different sources. The possible presence of many genes responsible for different subgroups of cancer patients may reduce the power of current methods to detect genes partly responsible for some forms of breast cancer. It is believed that methods effective in extracting interactive information from data should be developed.
What should be done when marginal effects are too weak to be detected? Our methods use interactive information from multiple sites as well as marginal information, They provide power to detect interactive genes. To test this claim and to demonstrate the practical value of these methods in real applications, we apply them to an important study: a subset of a large dataset collected from a case-control sporadic breast cancer study, focusing on gene–gene-based analysis. This partial dataset comprises 18 genes with 304 SNP markers. The application results in a number of scientific findings.
The message of this article is fourfold. First, if marginal methods fail, more powerful methods that take into account interactive information can be used effectively. We apply our proposed methods to this dataset to illustrate the detection of the interactions between genes. We point out that in our findings, none of the 18 selected genes show any detectable marginal effects that are significantly higher than those generated by random fluctuations. In other words, all of the 18 genes would be missed if only marginal methods were used.
Second, we demonstrate how to carry out a gene-based analysis by treating each gene as a basic unit while incorporating relevant information from all SNPs within that gene. Two summary test scores are proposed to quantify the strength of interactions for each pair of genes. The pairwise interactions can be extended easily. We also provide results using third-order interactions.
Third, to establish statistical significance, we generate a large number of permutations of the dependent variable (case or control) to see how the measures of interaction for the real data compare with those from the many permutations.
Finally, when these procedures are applied to the data, they lead to a number of interesting findings. It is shown that there are a substantial number of significant interactions that form a network in which BRCA1 plays a dominant role. The interactions of BRCA1 with many of the other genes suggests the conjecture that BRCA1 serves as a protective gene and that some mutations in it or in related genes may prevent it from carrying out the protective function.