By Monica Heger
This story was originally published on August 6
Using patient samples from an ongoing clinical trial of hormone therapy in breast cancer, researchers from Washington University's Genome Center have sequenced 50 tumor/normal pairs from breast cancer patients with the goal of identifying genomic signatures in the patients who respond to the therapy as well as in those who did not respond.
Elaine Mardis, co-director of the Genome Center, presented early results of the sequencing study — as well as preliminary results from a separate study comparing the sequence of primary and metastatic breast cancer tumors — last week at the Next Generation Sequencing and Genomic Medicine Applications Summit in Burlingame, Calif.
The hormone therapy study, which found that tumors from patients resistant to treatment were more highly mutated than those who responded, showed the power of combining a sequencing study with a clinical trial, Mardis said. The "unbiased whole-genome sequencing in a clinical setting is revealing genomic signatures that are predictive of aromatase inhibitor response," she said. "Ultimately, the clinical application of those profiles will have profound impacts on treatment."
The American College of Surgeons Oncology Group is leading the trial, which has enrolled a total of 377 patients with the luminal B subtype of breast cancer to test aromatase inhibitor therapy, a type of hormone therapy that blocks the production of estrogen.
"Ideally, a genomic profile of aromatase inhibitor responsiveness will enable us to ascertain which patients will respond to the therapy," Mardis said at the meeting. "The goal is to develop an assay" that could be used early on to guide treatment.
The Wash U team sequenced 50 samples, 25 from patients who responded to the chemotherapy treatment, and 25 from patients who did not respond. The DNA was taken from biopsies removed prior to treatment. In all cases the team used a paired-end sequencing strategy on the Illumina Genome Analyzer with 100 base pair reads to an average 30-fold coverage.
The genomes resistant to therapy showed more point mutations and structural variations than the tumors that responded to therapy. Twenty-eight genes were mutated in both the responsive and non-responsive patients, while 24 genes were uniquely mutated in the resistant group, and eight genes were uniquely mutated in the treatment-responsive group.
One pathway the team focused on in particular was the PIK3CA pathway, a signaling pathway involved in cell proliferation. The genomes of patients who were responsive to therapy had more mutations within that pathway than the non-responsive patients, in both the coding and non-coding regions.
"By sequencing the entire genes, we identified mutations outside of the common" coding sites, Mardis said, which they would not have identified had they sequenced just the exome.
The fact that that patients who responded to aromatase inhibitors were also more likely to have mutations in the PIK3CA pathway is particularly interesting because there are currently drugs being developed that target that pathway, suggesting a "compelling possibility for a combination therapy with aromatase inhibitors," she added.
So far, the team has not found mutational patterns specific to the therapy-resistant tumors, but Mardis said they have only just begun to analyze the results.
Sequencing patient samples from ongoing clinical trials could prove to be an effective approach to cancer genome sequencing because so much information about the patient is already known, making it easier to determine relevant mutations, Mardis said.
For instance, the clinical collaborators can "tell us what the important clinical questions are" and they can provide "high-quality samples to address those questions" from patients who have already given informed consent. They also have access to hundreds of patients from the same type or subtype of cancer, Mardis said, so the team can go back and validate its results in the other patients.
She said the team's next step is to finish analyzing the sequence data and then they will go back and screen all the patients from the clinical trial for the recurrent mutations revealed by the sequencing study.
"We'll compare genomic signatures of responsive and resistant genomes," she said, and ultimately "devise a predictive assay and algorithm to predict the response phenotype."
In addition, tumors that have been surgically removed from patients in the clinical trial following chemotherapy are being banked, and Mardis said that she and her colleagues now have patient permission to sequence them, which could provide insight into the effects of chemotherapy on tumor genomes. "A lot of interesting things could come of this; it will just take some digging to get it all out," she said.
Lynda Chin, scientific director at the Belfer Institute for Applied Cancer Science, who also presented at this week's meeting, said that combining sequencing with clinical trials could help make sense of the enormous amount of data generated by sequencing.
"Just because your gene of interest shows up as relevant doesn't mean it really is relevant. That's where clinical validation comes in," she said. However, she also stressed that it would be important to validate the findings in a completely separate population as well.
Mardis also presented some early results from a study of metastatic and primary tumors. Earlier this year, her team demonstrated that a subset of mutations in a primary breast cancer tumor were also present in a xenograft tumor and a metastatic tumor (IS 4/20/2010).
Following up on that study, the researchers have sequenced two additional primary tumors from basal-like breast cancers that went on to metastasize. In one sample, they sequenced a primary tumor, matched normal, and metastatic tumors in the liver and lung.
They found 34 shared mutations across the three tumor samples, 25 mutations unique to the liver metastatic tumor, and 10 unique to the lung metastatic tumor. In the second sample, aside from the primary tumor and matched normal sample, they also sequenced metastatic tumors from the adrenal gland, liver, and spinal cord, and have just started to analyze the results from those samples.