Wikipedia provides a good introduction to breast cancer
Advice from the National Cancer Institute
Basic information to keep in mind: a normal individual carries two copies of all genes located on chromosomes 1 - 22 (the autosomes). So, for example, everyone has two copies of the BRCA1 gene. Along the length of each gene, the DNA base at most positions is the same for everyone. However, at certain positions, variations occur. Each of these variations can be of no consequence (they are benign), or perhaps they have been statistically shown to slightly increase risk for a disease like breast cancer, or they can lead some or most of the time to the disease if the person lives long enough (these are causal mutations). For some variants, it just isn't known if they are of significance; these are termed variants of unknown significance ("VUS").
In general, the causal mutations that are recognized to significantly increase breast cancer risk within families are quite rare; these include mutations in the ATM, BARD1, BRCA1, BRCA2, BRIP1, CDH1, CHEK2, MLH1, MRE11A, MSH2, MSH6, MUTYH, NBN, PALB2, PMS1, PMS2, PTEN, RAD50, RAD51C, STK11 and TP53 genes. All together, these rare mutations probably account for a small (2 to 5%) percentage of breast cancer cases. Therefore, the most effective manner to assess familial breast cancer risk is not via SNP testing, but rather through identifying (if possible) the specific causative mutations within one's family and then testing relatives for those variations.
Variants may also lower one's risk for breast cancer. Recently (2014), SNP rs140068132 was found in Latin women ("indigenous americans") and was associated with both lower breast density (as seen on mammograms) and with a 40-60% lower risk for breast cancer.
A free risk "breast cancer risk evaluator" calculator is available online courtesy of the US National Cancer Institute. A more detailed, though not peer-reviewed, version of this calculator is also available.
2009 review: Breast cancer susceptibility: current knowledge and implications for genetic counselling [PMID 19092773]
If using family history to estimate breast cancer risk is not possible, one alternative is to simply tally up the number of non-causal breast cancer risk alleles that have some small increased risk each, in order to estimate an overall risk score. At least one study [PMID 17341484] has reported that overall risk can be (somewhat) estimated from looking at risk alleles that by themselves would each present too small a risk to look at independently. The SNPs considered to have risk alleles for breast cancer were primarily from the genes already mentioned, and included:
- BRCA1 gene SNPs:
- BRCA2 gene SNPs:
- ATM gene SNPs:
As for how to calculate the (estimated!) risk, this report [PMID 17341484] provides two different algorithms. The first is to simply count the number of risk alleles for these 25 SNPs, i.e. the # of rarer alleles for each SNP. (A homozygote for a rare allele ups the count by 2.) The odds ratio associated with increasing numbers of risk alleles is reported as follows:
0: 1.00 (by definition)
1: 1.46 (CI: 0.89 - 2.40)
2: 1.39 (CI: 0.86 - 2.25)
3: 1.75 (CI: 1.09 - 2.80)
4: 1.56 (CI: 0.95 - 2.55)
5: 1.31 (CI: 0.76 - 2.24)
6: 1.84 (CI: 1.04 - 3.26)
7: 2.10 (CI: 1.06 - 4.16)
8: 4.02 (CI: 1.56 - 10.38)
9 or more: 8.04 (CI: 1.89 - 34.26)
The second way to calculate overall risk for breast cancer reported by these authors involved counting the total of only the alleles that had relative frequencies (in their control population) of less than 10%. The simplest way to do this is to count the alleles for the same 25 SNPs presented above, but ignoring the 4 that have minor allele frequencies of over 10%, which are SNPs rs16942, rs1799966, rs144848, and rs1042522. Relative risk reported for this total count was:
0: 1.00 (by definition)
1: 1.24 (CI: 0.99 - 1.54)
2: 1.51 (CI: 1.10 - 2.09)
3: 2.90 (CI: 1.69 - 4.97)
It should be noted that having several minor (risk) alleles is quite common; in fact, 97% of women who did not have breast cancer had 0, 1, or 2 minor risk alleles, and even 1.4% of these women had 3 risk alleles (vs. 4% of women with breast cancer who had 3 risk alleles, for example).
Furthermore, the use of SNPs to predict breast cancer risk has been found to be less accurate than traditional methods, according to a 2008 study [PMID 18612136], and even when performed in combination with traditional methods, of little additive value. On the other hand, in the absence of family information and physical data, genetic data may be easier to collect and analyse.
Some SNPs have been reported to influence breast cancer progression, though not the odds of having it; these include:
As for the good news, some SNPs are associated with lower risk for breast cancer; these include:
- rs3218536 in the DNA repair gene XRCC2
- rs1045485 in CASP8, which is associated with apoptosis (cell death)
Other genes that may affect the risk of breast cancer include:
- rs3803662, risk allele T
- rs889312, risk allele A
- rs3817198, risk allele T
Several SNPs have been linked to poorer survival odds in patients diagnosed with breast cancer, irrespective of whether they affect the odds of having it:
- rs713041 and rs757229, in the antioxidant GPX4 gene
- rs438034, in the centromere protein F CENPF gene
Aside from risk of disease, there are SNPs that play a role in how patients respond to drugs used to treat breast cancer.
- [PMID 17115111] CYP2D6 metabolism is an independent predictor of Breast cancer outcome in post-menopausal women receiving Tamoxifen for early breast cancer. Determination of CYP2D6 genotype may be of value in selecting adjuvant hormonal therapy and it appears CYP2D6 inhibitors should be avoided in tamoxifen-treated women.
- [PMID 18294295] rs12762549 and rs11045585 can be used to predict the risk for docetaxel-induced leukopenia/neutropenia
- CYP2D6 poor metabolizers may not benefit from the use of tamoxifen as significantly as others for the treatment of breast cancer, prompting the recommended relabeling of this drug
The following SNPs have also been linked to breast cancer
The Icelandic company deCODE launched a $1625 BreastCancer™ test in October 2008 that determines the genotypes for 7 known SNPs that have been linked to female breast cancer. Note that this test is not a direct-to-consumer test; it is only available through a licensed physician. The 7 SNPs assayed are:
- rs13387042, on chromosome 2q35 (risk allele A) [PMID 17529974]
- rs4415084, on ch 5p12, near the MRPS30 gene
- rs889312, on ch 5q11 near the MAP3K1 gene
- rs13281615, on ch 8q24
- rs1219648, on ch 10q26 near the FGFR2 gene
- rs3817198, on ch 11p15 near the LSP1 gene
- rs3803662, on 16q12 near the TNRC9/TOX3 gene
From deCode: "The deCODE BreastCancer™ test reports the subject’s measured genetic risk of breast cancer relative to the average population and the lifetime risk of being diagnosed with breast cancer. The relative risk of breast cancer determined by the test can vary from 0.45 for subjects who are homozygous (2 copies) for the protective variants at all 7 markers, to 3.77 for subjects who are homozygous for risk variants at all markers. About 42% of the female population has genotype combinations of the tested markers that confer an increased relative risk (>1) of breast cancer. Ten percent of women in the general population have genotypes that confer more than a 40% increase in the relative risk of breast cancer, about 5% would be considered high risk with a 1.66-fold or greater risk (corresponding to a lifetime risk of 20% or greater), and approximately 1% of women have an average 3-fold risk (37% lifetime risk)."
omim summarizes the latest research
[PMID 16768547] Breast Cancer Our findings suggest that BARD1 Cys557Ser is an ancient variant that confers risk of single and multiple primary breast cancers, and this risk extends to carriers of the BRCA2 999del5 mutation.
The relative risk of breast cancer associated with PALB2 mutations was estimated to be 2.3 ( two-fold ). This is similar to the increase in risk seen with the CHEK2, ATM and BRIP1 genes â€“ all reported by the same research group since 2002. PALB2 is the first low-risk gene protein product found that interacts with BRCA2 â€“ all of the other low-risk genes, CHEK2, ATM and BRIP1, are linked to BRCA1. Together, these genes are thought to account for around two per cent of all breast cancer cases. 
This article pools the data from several different studies to identify 5 snps which consistently demonstrate relevance to breast cancer
- [PMID 16306136]
- [PMID 17018785]
- [PMID 17018785]
- [PMID 16606630]
- [PMID 17216494] rs361525
The mannose-binding lectin (MBL2) haplotype and breast cancer: an association study in African American and Caucasian women.
This study presents preliminary evidence that common genetic variants in the 3' UTR of MBL2 might influence the risk for breast cancer in African American women, probably in interaction with the 5' secretor haplotypes that are associated with high concentrations of mannose binding lectin. [PMID 17071626]
Breast Cancer Gene Mutations More Common Than Thought provides a nice summary of this research paper
In contrast to the discussion of the SNPs underlying familial cases of breast cancer, this paper discusses genetic risk factors underlying sporadic cases [PMID 16835078].
[PMID 17908970] The AA Genotype of the Regulatory BCL2 Promoter Polymorphism ( 938C>A) Is Associated with a Favorable Outcome in Lymph Node Negative Invasive [breast cancer] Patients. Bachmann HS, Otterbach F, Callies R, NÃ¼ckel H, Bau M, Schmid KW, Siffert W, Kimmig R.
Authors' Affiliations: Institutes of Pharmacogenetics and Pathology and Neuropathology.
PURPOSE: Expression of the antiapoptotic and antiproliferative protein Bcl-2 has been repeatedly shown to be associated with better clinical outcome in breast cancer. We recently showed a novel regulatory (-938C>A) single-nucleotide polymorphism (SNP) in the inhibitory P2 BCL2 gene promoter generating significantly different BCL2 promoter activities. EXPERIMENTAL DESIGN: Paraffin-embedded neoplastic and nonneoplastic tissues from 274 patients (161 still alive after a follow-up period of at least 80 months) with primary unilateral invasive breast carcinoma were investigated. Bcl-2 expression of tumor cells was shown by immunohistochemistry; nonneoplastic tissues were used for genotyping. Both the Bcl-2 expression and the (-938C>A) genotypes were correlated with the patients' survival. RESULTS: Kaplan-Meier curves revealed a significant association of the AA genotype with increased survival (P = 0.030) in lymph node-negative breast cancer patients, whereas no genotype effect could be observed in lymph node-positive cases. Ten-year survival rates were 88.6% for the AA genotype, 78.4% for the AC genotype, and 65.8% for the CC genotype. Multivariable Cox regression identified the BCL2 (-938CC) genotype as an independent prognostic factor for cancer-related death in lymph node-negative breast carcinoma patients (hazard ratio, 3.59; P = 0.032). Immunohistochemical Bcl-2 expression was significantly associated with the clinical outcome of lymph node-positive but not of lymph node-negative breast cancer patients. In lymph node-negative cases, the (-938C>A) SNP was both significantly related with the immunohistochemically determined level of Bcl-2 expression (P = 0.044) and the survival of patients with Bcl-2-expressing carcinomas (P = 0.006). CONCLUSIONS: These results suggest the (-938C>A) polymorphism as a survival prognosticator as well as indicator of a high-risk group within patients with lymph node-negative breast cancer.