This article was originally posted June 12.
Researchers from Washington University's Genome Institute and Siteman Cancer Center have used next-generation sequencing in a clinical trial to identify genomic pathways that may predict response and resistance to aromatase inhibitor therapy in women with estrogen-positive breast cancer.
The results of the study were published this week in Nature. The sequencing was done in conjunction with the American College of Surgeons Oncology Group's clinical trial Z1031, which began in 2006 and aims to evaluate how well the aromatase inhibitors exemestane, letrozole, and anastrozole work in treating postmenopausal women who are undergoing surgery for stage II or stage III breast cancer.
The Wash U team now plans to start a separate clinical trial to test whether the genomic markers identified in the study can be used to guide treatment of ER+ breast cancer patients and lead to better outcomes.
The Nature study "gives us a clear path forward," Elaine Mardis, a senior author of the study and co-director of Wash U's Genome Institute, told Clinical Sequencing News.
Broadly speaking, the team found that mutations within the TP53 signaling pathway were associated with resistance, while mutations within the MAP3K1 pathway were associated with response.
While a genomic "predictor" based on these mutations is still being refined in additional patients, Mardis said the idea is that it "could allow you to evaluate the genome up front and then determine whether the patient should or should not be treated with aromatase inhibitor therapy."
The Wash U team presented preliminary results from the study at the annual American Association of Cancer Research meeting in Orlando, Fla., last year (CSN 4/5/2011).
Since then, they have expanded the study from evaluating 50 tumor/normal pairs to the 77 that were reported in this week's Nature study. Additionally, the team is using a similar approach on another ACOSOG trial, Z1041, which is studying trastuzumab plus chemotherapy in HER2-positive breast cancer.
Of the 77 patients in the current study, the team sequenced the whole genomes of 46 patients and the whole exomes of 31. Twenty-nine patients were resistant to aromatase inhibitor therapy, while 48 responded.
Sequencing was done on the Illumina HiSeq 2000 platform to a minimum of 30-fold coverage for the tumor samples and 25-fold haploid coverage for the normal samples.
The researchers identified 18 genes that were significantly mutated, including known cancer genes such as PIK3CA, TP53, MAP3K1, and GATA3. They also identified five genes that have previously been linked to hematopoietic disorders and three genes not previously implicated in cancer.
From a broad perspective, the genomes of the patients that responded to therapy were less mutated than those that were resistant to therapy. Looking at the 46 tumors for which whole-genome sequencing was done, the resistant tumors carried 1.62 mutations per megabase, a mutation rate nearly twice that of the responsive tumors at 0.862 mutations per megabase. The resistant tumors also had more structural variants than the tumors that responded.
Following an initial analysis of those 77 patients, the team examined select genes in an additional 240 cases.
They also integrated clinical features — such as levels of a tumor proliferation marker known as Ki67 and whether the tumor was considered luminal A or luminal B — into their analysis to further support their genomic findings.
"This is an overarching theme," said Mardis. Next-gen sequencing "is not a standalone [tool], but in combination with common aspects of medicine it allows doctors to do a better job of understanding the patient's disease."
Aside from looking at significantly mutated genes, the researchers also performed pathway analysis in order to tease apart differences between the resistant and responsive tumors.
The most significant pathway related to tumor resistance was the TP53 signaling pathway, while the most significant pathway related to aromatase inhibitor response was the MAP3K1 pathway.
Additionally, pathway analysis helped "expand the reach of the genes we identified" to include even those that were only found in a few percentage of the patients, said Mardis.
"It's important to understand the whole mutational spectrum," Mardis said. Because cancer is so heterogeneous and many different genes are mutated in each patient, there will often be genes that are mutated in only one or two percent of a patient population. So while the genes themselves don't cross the threshold of being statistically significant, they may still play a role in the tumor biology and disease etiology and progression.
Pathway analysis revealed that several genes with low-frequency mutations were subunits of complexes — for instance the condensin complex, which had four mutations across three genes; and the MRN complex, which had four mutations across four genes. Both complexes affect genome stability and chromosome structure and organization, and an MRN protein is also important for DNA repair. Mutations in these complexes have been associated with breast cancer.
While 71 percent of the tumors contained significantly mutated genes in significant pathways, an additional 21 percent contained only non-significantly mutated genes in these same pathways.
Using an algorithm called PARADIGM (Pathway Representation and Analysis by Direct Reference on Graphical Models) developed by Joshua Stuart's lab at the University of California, Santa Cruz, the team was able to incorporate transcriptional activity into the analysis.
Looking at transcriptional activity in relationship to Ki67 allowed them to identify key "hub" genes in resistant tumors, including MYC, FYN, and MAP kinases. "Targeting these hubs in resistant tumors could produce therapeutic advances," the authors wrote.
The analysis also lent more strength to the MAP3K1 pathway — showing that eight samples with mutations in the MAP3K1 substrate MAP2K4 "were sufficient to derive a reliable pathway-based gene signature in PARADIGM that aligns with MAP3K1."
Pathway analysis "allows us to layer on detail … and get the full picture of tumor biology," Mardis said.
The next step is to incorporate these findings into a clinical trial where genomic data will be used to guide treatment for women with ER+ breast cancer. Mardis said that the team is using a recent $4 million grant from the Susan G. Komen for the Cure foundation to begin such a trial.
The goal is to see whether incorporating whole-genome sequencing can help improve survival and overall outcomes. The researchers plan to look at the whole genome rather than create a targeted gene panel because the detailed pathway analysis will be important for putting specific mutations in the context of other genomic data and clinical features, Mardis said.
Additionally, while the team was able to conduct pathway analysis for patients who received whole-genome sequencing as well as those who received whole-exome sequencing, they only looked at structural variations from the whole-genome sequencing data.
"The accrual of large numbers of patients and the use of comprehensive sequencing and gene expression approaches will be required because of the extreme genomic heterogeneity documented by this investigation," the authors wrote.
Mardis said around 100 women will likely be enrolled in the next clinical trial. These women will have their tumor and normal genomes sequenced at initial diagnosis.
Based on the genomic data along with other clinical features, participants will either receive aromatase inhibitor therapy or will be recommended for surgery followed by chemotherapy and potentially a targeted therapy.
Mardis said the team is aiming for a four-week turnaround time from the time a sample is taken to the initial stratification.
For women who are predicted to be resistant to aromatase inhibitors, their genomes will be further mined for other druggable mutations to help guide the use of a targeted therapy after surgery.
In the Nature study, for instance, the team identified a number of mutations within kinase domains that are druggable, including a mutation in ERBB2, which is sensitive to the lung cancer drug gefitinib.
One potential outcome is that patients who are resistant to aromatase inhibitors will have mutations that indicate response to a drug that is approved for another type of cancer.
Mardis said it is not yet clear whether those patients would be able to receive that drug. Potentially a pharmaceutical company would be willing to supply it in the context of a clinical trial, she said, but those details have not yet been worked out.
She added that this issue underscores the current thinking among many cancer researchers that therapies should be approved based on the mutations they target rather than for treating cancer that originates from a specific tissue type.