An international research team, led by investigators at the Mayo Clinic and National Institutes of Health's Pharmacogenomics Research Network, has identified a set of SNPs with promise as a tool to guide more personalized preventive breast cancer therapy using selective estrogen receptor modulators.
The team published the finding in Cancer Discovery last month, along with results of functional validation studies showing that the SNPs — in and near the ZNF423 and CTSO genes — are indeed implicated in SERM-associated pathways, adding biological plausibility to their potential predictive role.
According to lead author Richard Weinshilboum, members of the group are now hoping to begin follow-up clinical studies to establish how well the markers might be able to distinguish women who are more likely to benefit from preventive therapy with SERMs from those who are less likely to benefit.
Weinshilboum told PGx Reporter this week that the team is simultaneously planning to look more deeply into one of the biological implications of the initial study — that these potentially predictive SNPs are involved in the estrogen-dependent expression of BRCA1 — to see if the finding could inform new treatment strategies for the disease.
Despite recommendations earlier this year by the US Preventive Services Task Force that doctors consider preventive therapy for all women deemed to be at heightened risk for breast cancer, Weinshilboum said the safety profiles of these drugs deter many from acting on this guidance.
"Clearly these drugs are effective," he said. "However there are rare, but life-threatening risks, particularly with tamoxifen, of deep vein thrombosis and increased risk of endometrial carcinoma. Because of this, even though highly effective, these drugs are not widely prescribed for prevention."
The hope, Weinshilboum explained, is that a better understanding of which women actually benefit from treatment will improve the risk-benefit ratio for these drugs and potentially increase the number of women who are prescribed preventive therapy.
In the study, Weinshilboum and his colleagues analyzed samples from two breast cancer prevention trials, the National Surgical Adjuvant Breast and Bowel Project P-1 trial of tamoxifen, and the NSABP P-2 trial comparing tamoxifen and raloxifene.
The researchers looked at 592 subjects from the trials who developed breast cancer despite five years of treatment with a SERM and 1,171 matched controls who did not develop cancer.
Collaborators at the RIKEN Center for Genomic Medicine in Japan genotyped the samples, and the team analyzed the results to determine any SNPs with the potential to distinguish cases from controls.
The team then narrowed their findings to a group of SNPs with the lowest P values on chromosome 16, chromosome 13, and chromosome 4. Even though these markers did not reach statistical criteria for genome-wide significance — usually required in GWAS analysis to insure a signal is not due to chance — the team decided to move forward with functional analysis of the best candidates anyway using a model system made up of 300 lymphoblastoid cell lines.
"We had used [cohorts] that included the majority of samples worldwide from women in preventive SERM trials, so it was clear we would not find a larger replication study and getting DNA from smaller studies wasn’t going to be possible either. We felt responsibility to pursue the possibility that there might be something here," Weinshilboum said.
In this gambit, the researchers found that two SNPs on chromosome 16 associated with decreased risk of cancer despite SERM treatment mapped to an intron of the ZNF423 gene, while the most significant of the SNPs on chromosome 4 was located near the CTSO gene.
Using the model cell line system, the researchers then tried to determine whether the ZNF423 SNPs or the CTSO SNP might play a role in pathways involved in the development of breast cancer and its inhibition by SERM drugs.
According to the study authors, both genes appeared to be induced by estrogen exposure in a SNP-dependent manner. More importantly, this estrogen-dependent induction of both ZNF423 and CTSO also influenced downstream expression of BRCA1.
"It has been known for years that estrogen can induce BRCA1, and now we think the two candidates that came out of the GWAS might be part of that induction … so we plan to study that," Weinshilboum said.
In an editorial accompanying the study in Cancer Discovery this month, Stephen Chanock and Mitchell Machiela of the National Cancer Institute pointed out that the use of lymphoblastoid cell lines — rather than breast cancer cell lines — is a questionable move, despite the promising results in this case.
Weinshilboum said his group has had success using the same cell lines for functional validation of GWAS results in previous PGx studies of other drugs like aromatase inhibitors.
"Functional validation gives us biological plausibility, and if it fits with the way the drug works, that’s a powerful argument," he said.
Of course, the team's results will have to be investigated further to determine whether they may lead to a clinically useful method for better classifying women according to their likelihood of responding to preventive therapy.
Weinshilboum said his team has already begun planning studies to follow up on the finding, both of whether the SNPs are a powerful-enough response predictor, and of how the connection between these markers and BRCA1 might inform new treatment options.
"We are engaged in discussion about potential clinical trials to test whether these markers would make us better able to classify women based on their response," Weinshilboum said.