By Monica Heger
This article was originally published Dec. 2.
The University of Michigan's Center for Translational Pathology has completed a pilot study demonstrating how it plans to use next-generation sequencing of its cancer patients to guide the next course of treatment, including matching patients to available clinical trials.
In the study, published last week in Science Translational Medicine, the Michigan team, led by Arul Chinnaiyan, used a comprehensive sequencing strategy, including low-pass whole-genome sequencing of tumor DNA, exome sequencing of tumor and matched normal DNA, and transcriptome sequencing of the tumor, on two xenograft samples of patients with prostate cancer and two biopsied patient samples with colorectal and metastatic melanoma.
Dubbed Mi-OncoSeq, the protocol includes not only sequencing, but also a multidisciplinary sequencing tumor board that helps with clinical interpretation of the sequencing data and recommends the next course of action for the patient.
The study illustrated that the approach is feasible to do in a clinical setting, with results delivered to the board for deliberation within 24 days of biopsy, and at a reasonable cost of around $3,600.
It also demonstrated the importance of using an integrative sequencing approach, rather than focusing on a handful of known cancer genes, and highlighted a major challenge in personalized cancer care — the fact that current oncology clinical trials are defined by the tissue in which the cancer originated rather than a tumor's mutational status.
Chinnaiyan said the team is now in the process of scaling up the study. Twenty cancer patients have already progressed through the Mi-OncoSeq protocol, and he is aiming to sequence 100 patients over the next year.
The lab is also seeking CLIA and CAP certification for its sequencing strategy, which Chinnaiyan expects to have by May 2012. Currently, all sequencing is being done on the Illumina HiSeq, with clinically actionable results validated by Sanger sequencing in a CLIA setting. Chinnaiyan said he would aim to have the protocol CLIA-certified on both the HiSeq and the MiSeq, with the benchtop instrument used mainly to validate results from the HiSeq.
Going forward, Chinnaiyan said that the team would likely use a similar protocol to the one described in the Science Translational Medicine study, except that it may discontinue the low-pass whole-genome sequencing.
That step was included in order to help call structural variants, but Chinnaiyan said that the exome sequencing and transcriptome sequencing seem to be sufficient.
"We're finding the structural rearrangements and copy number alterations in the exome" sequencing data, he told Clinical Sequencing News. The exome and transcriptome sequencing data are "the most rich," and "the complementary information from the whole-genome sequencing has not helped as much as we thought," he added.
Dropping the whole-genome sequencing would also help reduce costs. Chinnaiyan noted that during the course of the pilot, the reagent cost per patient fell from around $5,000 to around $3,600, and he expects that the costs will continue to fall.
An Unbiased Approach
The Michigan team announced its Mi-OncoSeq program earlier this year (CSN 8/31/2011). The published study demonstrates that the protocol can find actionable mutations in patients in a clinically relevant turnaround time of three to four weeks, and at a reasonable cost.
In addition, the study illustrates the advantages of using an unbiased and integrative sequencing approach for clinical cancer sequencing.
One patient was a 46-year old man diagnosed with colorectal cancer in the spring of 2009, and later developed metastases in the liver. After standard therapy and a phase I trial with an Aurora kinase B inhibitor failed to slow down disease progression he enrolled in Mi-OncoSeq.
Sequencing and analysis was completed within 24 days of biopsy, and identified an activating mutation to the gene NRAS, which is not a commonly mutated gene in colorectal cancer. However, according to the authors, the mutations are "biologically similar" to activating mutations in KRAS, which is mutated in 35 percent to 40 percent of colorectal cancer patients, but was not mutated in this patient.
The NRAS mutation "wouldn't have normally been looked for in colon cancer," since KRAS is more common, said Chinnaiyan. But, because the team used an unbiased sequencing approach, they were able to pick that up.
Additionally, the NRAS mutations point to MEK inhibitors as a potential therapy option.
Including transcriptome sequencing as part of the protocol has been even more valuable, said Chinnaiyan. It's "clearly allowing us to identify gene arrangements and fusions that would not have been picked up by a focused strategy," he said. "And we believe many of these things are clinically actionable targets," such as kinases for which there are drugs or pathways that might be actionable.
Chinnaiyan added that especially for the additional patients that he has sequenced since the study, transcriptome sequencing has been invaluable at pinpointing gene fusions and rearrangements that would not have been seen with other methods.
Other groups have also found that transcriptome sequencing is key to identifying targetable mutations in cancer and figuring out which pathways are driving the cancer. For instance, researchers at Washington University's Genome Institute have found RNA-seq data to be critical for helping to guide treatment of cancer patients. The team has found that often only 40 percent of a patient's mutations are actually expressed (CSN 10/5/2011).
Not Enough Drugs
The Mi-OncoSeq study also demonstrated that the protocol could identify potential drug targets for cancer patients.
One major hurdle going forward, however, is that while clinical trials exist for many of the candidate druggable mutations, these studies are not usually open to patients with tumor types beyond those targeted in the trial.
Additionally, the team is uncovering what appear to be good drug candidates, such as kinases, but for which no compound currently exists.
"We don't have as many drugs as we'd like," said Chinnaiyan. "Even if we see a driver mutation, often we're not positioned to have a drug available."
In the case of the patient with the NRAS mutation, there are ongoing clinical trials of Raf inhibitors for colorectal cancer patients with KRAS or BRAF mutations, but they "fail to include patients with NRAS mutations" despite the similarity of those mutations to KRAS mutations, according to the authors.
Additionally, the same patient also had an amplification of CDK8 which, in combination with his NRAS mutation, suggested that a CDK inhibitor administered with MEK or PI3K inhibitors would be a potential therapeutic option. However, while there is currently a phase I trial for such a combination, the study is limited to breast cancer.
Similarly, the second patient described in the study, a 48-year old woman with metastatic melanoma, had an activating mutation in HRAS, which the board identified as potentially actionable. The mutation was surprising because it had not previously been identified in melanoma, but is similar to NRAS mutations, which are present in about 15 percent of malignant melanoma cases.
Currently there are no available clinical trials for this patient's tumor type and mutational status, but the board identified an upcoming Novartis trial for a combined therapy of MEK and PI3K inhibitors, for which the patient may qualify.
Chinnaiyan said that around 25 percent of the patients that have gone through the Mi-OncoSeq protocol fall into this category, where potentially druggable mutations are found but either no compound currently exists, the patient is not eligible for the clinical trial, or the clinical trial is not in a location accessible to the patient.
"I think drug companies are certainly aware of this issue," said Chinnaiyan, adding that he thinks clinical trials will eventually be modified so that they consider "the mutations and pathways, rather than where the cancer originated."
Informed Consent
Like other groups looking to implement clinical sequencing, the UMich team has developed its own protocol to guide the return of results. Chinnaiyan said that there is a flexible consent model, to give patients a choice about the type of information they receive.
Patients can opt to receive all relevant genomic information, whether about their particular cancer or another disease. Patients can also opt to receive information that pertains only to their particular cancer and can even opt out of receiving information that could have implications for family members.
Chinnaiyan said that so far, the team has not faced any ethical challenges with the patients that have enrolled. However, the protocol was developed with the anticipation that germline mutations predictive of cancer risk that could impact other family members would be found.
Results are then returned through a genetic counselor and the patient's oncologist, who also sits on the sequencing tumor board that makes treatment recommendations.
Chinnaiyan said the team is also working to develop a model to create a report for the patient's medical record. "The challenge with generating so much information is, 'What do you put in the report?'" he said. Right now, the team presents to the board only filtered sequence data with potentially relevant results. He added, however, that the sequence data will be stored in a database so that as patients progress the data can be revisited.
Mi-OncoSeq is currently only available for patients with advanced cancer or metastatic disease. Offering it to these patients first poses fewer ethical challenges than offering it to all cancer patients, said Chinnaiyan, but also makes it less likely that the patient will see lasting benefits.
"We're biasing it against ourselves because we're focused on patients with such advanced disease," he said. "Even if we see mutations for therapies, the chance that they work are less likely because the patients have picked up so many different types of mutations and resistance mechanisms," he said.
Ultimately, the goal is that patients would be enrolled in Mi-OncoSeq at the time of diagnosis, and that the protocol would become the standard of care. That's where "we'll increase the chances of having better therapeutic outcomes," said Chinnaiyan.
Have topics you'd like to see covered by Clinical Sequencing News? Contact the editor at mheger [at] genomeweb [.] com.