NEW YORK (GenomeWeb) –Weill Cornell Medicine is gearing up to expand whole-exome sequencing to several thousand advanced cancer patients per year based on results from an ongoing clinical study.
This week, a team led by Mark Rubin and Olivier Elemento at the Caryl and Israel Englander Institute for Precision Medicine at New York Presbyterian Hospital-Weill Cornell Medicine published a clinical validation study of the test, called EXaCT-1 (Exome Cancer Test), in NPJ Genomic Medicine. The test, which can detect point mutations, insertions and deletions, and copy number variations, received approval from the New York State Department of Health last year.
In addition to detecting genomic alterations that may point to targeted therapies, the cancer exome test could also be useful for selecting patients likely to respond to immunotherapy. "It's a major justification for doing whole-exome sequencing in the clinic," said Elemento, an associate professor and the associate director of the Institute for Computational Biomedicine at Weill Cornell Medicine.
In parallel to expanding access to the test, the researchers are working on an updated version, called EXaCT-2, that will combine exome sequencing and deep sequencing of hotspots and is expected to be available by the end of 2017.
EXaCT-1 uses Agilent's HaloPlex target enrichment system, followed by Illumina sequencing on a HiSeq system in rapid mode in a CLIA laboratory at Weill Cornell that handles the entire testing process from DNA isolation to clinical reporting. Initially, the test ran on the HiSeq 2500, which has now been replaced by a HiSeq 4000.
Elemento said his team looked at four target selection platforms before settling on the HaloPlex, which at the time was the only platform that could handle low amounts of DNA. Also, it was compatible with both fresh-frozen and formalin-fixed paraffin-embedded samples, which was important because patient samples can be either type, and because the lab was seeking NY state approval for analyzing both sample types.
The test involves sequencing both tumor and matched constitutional DNA, which Elemento said is critical for distinguishing somatic mutations in the tumor from background germline mutations.
In their paper, the researchers reported an average turnaround time of three weeks, from sample acquisition to clinical reporting, and a reagent cost of approximately $1,400 per tumor-normal pair. Turnaround time is now about two to three weeks, Elemento said, and can be as fast as one week "when everything lines up nicely." The test costs several thousand dollars, he said, which is currently not covered by health insurance. Weill Cornell, the hospital, and philanthropic funding are covering the costs, so patients don't have to pay.
The clinical report, a multi-page PDF file that along with structured data is also pushed automatically into a patient's Epic electronic health record, lists somatic mutations in three categories or tiers: clinically actionable mutations in more than 50 genes, which are often linked to a US Food and Drug Administration-approved drug; other genomic alterations in almost 600 cancer genes that likely drive tumor growth; and variants of unknown significance in other genes, many of which are passenger mutations.
Clinically actionable mutations are linked to Weill Cornell's in-house Precision Medicine Knowledge Base (PMKB), which contains information about mutations in certain tumor types and their clinical response to specific drugs. There are also efforts to integrate information from databases from different institutions, or at least to be able to query multiple databases, Elemento said.
These passenger mutations reported in the third tier may be critical for immunotherapy, an aspect that Elemento's team is exploring in a follow-up paper it is about to submit. "Very often, those passenger mutations don't drive the tumor, but they are potentially exposed on the surface of tumor and immune cells," he explained. "They are the mutations that the immune system is able to recognize — mutations that can potentially trigger an immune response against the tumor." Such mutations appear to be biomarkers of response to immunotherapy, he added, along with other types of genomic information, such as tumor RNA expression and T cell receptor gene expression. "Looking at both DNA and RNA within the tumor can get you a set of biomarkers that together can predict, with reasonable accuracy, which patients are most likely to respond," he said.
Another reason for analyzing tumors by whole-exome sequencing, rather than more targeted testing, is that it may uncover important mutations involved in therapy resistance. Oftentimes resistant tumor cells activate pathways that are potentially druggable, "but typically not through the typical driver genes," Elemento said. "We want to be able to detect those mutations, which could potentially occur anywhere in the genome."
The report does not include links to clinical trials that enroll based on specific mutations, in part to avoid providing inaccurate information. "We are very conservative with regard to information that goes into the reports," Elemento said. "We don't want to say that there is a clinical trial that's open for a particular patient if a patient is not eligible to enter the trial."
Instead, clinical trials are often discussed by tumor boards, where experts gather to talk about specific patient cases and the results of their genomic tests. "Those tumor boards are very well attended by physicians, and physicians have been very positive about the ability of whole-exome sequencing to help them better treat patients," Elemento said.
The test currently does not report germline variants, but the scientists have studied them for research purposes, looking at about 100 genes, and the plan is to include germline variants in future reports. This will require some additional steps, for example genetic counseling to explain the results to patients. Based on their initial analysis, 10 to 15 percent of patients appear to have reportable germline variants, Elemento said.
Validation of the test for New York state approval took about a year and a half, he said, and involved sequencing hundreds of samples that were mutation-positive in certain genes.
For their paper, the researchers initially sequenced two HapMap samples as well as 57 matched tumor-normal pairs, representing a large spectrum of tumor and sample types that had a known mutation in one of five cancer genes — KRAS, BRAF, JAK2, EGFR, or HER2/neu.
The overall sensitivity of the assay across the exome was 95.7 percent, its specificity was 99.9 percent, and the positive predictive value 99.2 percent, with a 10 percent limit of detection for validated mutations at a minimum coverage of 30x.
A small number of exons was not covered at all, the researchers noted, and about 1 percent of the HaloPlex targets were covered poorly. Overall, about 20 percent of clinically relevant genes did not have sufficient coverage to definitely rule out mutations, meaning low-level mutations, particularly in tumor suppressor genes, might be missed.
As part of a clinical study, the researchers then went on to analyze 337 prospective primary and metastatic tumor samples from advanced cancer patients with a variety of diagnoses — including breast, lung, colon, bladder, prostate, and hematological cancers — and found mutations in 168 out of 558 cancer genes they analyzed. Of these mutations, 72 were classified as clinically relevant tier 1 mutations, and 475 as tier 2 mutations, and such mutations were found in 13 percent and 69 percent of cases, respectively.
Elemento said the clinical study is still open and has so far sequenced the cancer exomes of about 750 patients. He and his colleagues are working with the hospital to expand the protocol, with the goal of providing the EXaCT-1 test to 2,000 to 3,000 patients per year over the next few years.
Test results have already been helpful, both for patient treatment and on a research level, he said. In several cases, the test identified mutations that are typically seen in other types of cancer and might have been missed by other tests. For example, it found a HER2 amplification, which is usually seen in breast cancer, in a bladder cancer patient. As a result, she received herceptin, which targets the mutated gene product, and responded well to that drug.
Overall, the percentage of cases where doctors have acted on the test results is still small — on the order of 10 percent, according to Elemento's estimate — and if they do, it is usually when all standard treatment options have been exhausted.
In the meantime, the researchers are already working on EXaCT-2, which will combine cancer exome sequencing with targeted deep sequencing of about 50 genes, currently a separate test that is used in parallel with ExaCT-1.
EXaCT-2, which the team hopes to get approved by the end of 2017, will have additional probes to augment coverage in the hotspot areas and will likely use a different target enrichment platform than HaloPlex, Elemento said.