Charting a Course to Genome-Guided Cancer Medicine
"It is not difficult to foresee a time when a person's individual genome can be sequenced for as little as $100, putting genetic studies in the realm of a routine laboratory test."
—Drs. Vincent J. DeVita and Steven A. Rosenberg
"Two Hundred Years of Cancer Research," New England Journal of Medicine
If the cost of sequencing DNA continues to decline, patients with cancer could have their tumors profiled as part of routine care. But how this might actually work is far from clear.
To begin developing some answers, NCI's Cancer Diagnosis Program (CDP) held a workshop last month in Bethesda, MD. More than 200 participants shared their ideas about the challenges involved in trying to bring new DNA sequencing technologies into the clinic.
These tools, known as next-generation DNA sequencers, can decode a relatively large amount of DNA in a short period of time (days or weeks). Although the accuracy of these tools is still being assessed, the instruments are increasingly affordable for clinical labs.
"We know that next-generation sequencing will be used in the clinic," said co-organizer Dr. Barbara Conley, associate director of CDP. "So this workshop focused on the barriers to making this happen and how we could address them."
The meeting included experts on all aspects of cancer care—pathologists, oncologists, laboratory scientists, statisticians, bioinformaticians, and drug and device manufacturers. A representative from the Food and Drug Administration and patient advocates also gave presentations.
Clinically Meaningful Mutations
Next-generation sequencing is being used at some medical centers, and some private companies are exploring its potential as well. Genetic information can help diagnose a patient's cancer or match patients with drugs that target specific molecular changes within their tumors.
A theme of the meeting was the need for a database that doctors could use when making decisions about a patient's treatment. It would include "clinically actionable information," such as genetic markers that identify candidates for a particular treatment.
At the moment, however, no such database exists. This may be the most significant challenge for managing and reporting genetic information, noted Dr. Jane Gibson of the University of Central Florida College of Medicine.
Discussions are under way at NCI about how best to address the need for a repository, an NCI official noted at the meeting.
Every Genome Is Different
Even with the costs of DNA sequencing dropping, clinical labs have limited resources. Rather than sequencing genomes or exomes (the protein-coding regions of genes), clinical labs could begin by sequencing panels of genes associated with cancer, several participants said.
Tumor DNA would need to be sequenced about 500 times to achieve the sensitivity needed for clinical decision-making, noted Dr. Maureen Cronin, an industry consultant. In comparison, many published genomes have been sequenced about 30 times, but this is "barely sufficient for tumor DNA," she added.
"Every patient's genome is different," said Dr. Elaine Mardis, co-director of the Genome Institute at Washington University. "And, ultimately, one wants to know exactly what these differences are."
There was no clear agreement about what a complete molecular workup of a tumor would look like. But whatever form the report to clinicians takes, it would be important not to overwhelm them with sequencing data, a few participants said.
"You want the Reader's Digest version of the results so that a clinician can make sense of them," said Stephen Lincoln of Complete Genomics. "But you also want the research behind those results."
The role of the DNA sequencing and analysis team, Dr. Mardis noted, is to offer evidence to the oncologist. "We're not in a position to tell the oncologist what to do or to advocate one therapy over another," she said. "We're just trying to help out."
Choosing a Technology
Participants at the meeting reported using various next-generation sequencing tools, or platforms. These technologies are updated frequently, which has made it impractical to establish technological standards for clinical DNA sequencing.
Another challenge participants mentioned was the fact that not all cells within a tumor contain the same mutations. Furthermore, different amounts of tumor may be required for analysis depending on the type of tumor and the experiment.
"The elephant in the room here is which platform has the right mix of simplicity, cost efficiency, and high-quality data," said Dr. John Iafrate of Massachusetts General Hospital.
Perhaps even more important than the platform, some participants noted, is the ability to make sense of all the data generated. "The centerpiece of any clinical next-generation sequencing program would be bioinformatics," Dr. Gibson stressed.
What the field needs are "tools that can allow us to digest and present the data to clinicians so that they can ask meaningful and intelligent questions," said Dr. John Quackenbush of the Dana-Farber Cancer Institute.
Dr. Mardis, who has written about the cost of data analysis, added: "I fear that too much is being made across the board about the cost of sequencing when, really, the impactful and expensive activity in all [of] this is the analysis of the data."
When DNA sequencing becomes routine has less to do with the cost of the technology than with the perceived value of the results to the community, some participants said.
"It's not the cost that's going to drive this, it's the value [to patients and clinicians]," said Dr. Richard Press of the Knight Cancer Institute at Oregon Health and Science University. And for clinical labs, coverage for services by insurers will be critical, he added.
"When our data are considered to be clinically useful, there won't be any more complaints about the cost," added Dr. Iafrate. He cautioned, however, that the current costs of sequencing can be tens of thousands of dollars, and insurance might not cover the expense.
"You're trying to help a patient," Dr. Iafrate went on. "But if you don't find anything actionable in the results, and the patient gets a bill for $5,000, that is a problem."
Sharing Is Important
A recurring question at the workshop was: What should the level of evidence be for reporting a mutation to a clinician? One person suggested that the field eventually will need to have information from tens of thousands of patients in a well-annotated database, as well as definitions of clinically actionable variants.
Everyone seemed to agree that, whatever happens, it will be important to share data. Data sharing will be necessary in cases in which different doctors treat the same patient years apart, as well as to create the most informative, useful database.
"Everyone believes we have a powerful opportunity to bring these tools into the clinic," said co-organizer Dr. Mickey Williams, of the Molecular Characterization and Clinical Assay Development Laboratory at the NCI Frederick National Laboratory for Cancer Research. "But we want to make sure that it happens in a thoughtful way."