May 01, 2013
by Julia Karow of genomeweb.com
Columbia University's Laboratory of Personalized Genomic Medicine has launched a whole-exome sequencing test for inherited genetic disorders, likely the first such test with conditional approval from the New York State Department of Health.
In addition to offering the exome test and an existing mitochondrial genome sequencing test, the lab is working on other NGS-based diagnostics – among them a 1,000-gene "superpanel," a whole-genome sequencing test, and a cancer panel – all of which it plans to release later this year.
The PGM laboratory, which is headed by Mahesh Mansukhani and is part of Columbia's department of pathology and cell biology, offers a variety of diagnostic tests in the areas of genetics, oncology, cytogenomics, and molecular microbiology. It was created last year from several formerly separate units and is accredited by CLIA, the College of American Pathologists, and the Clinical Laboratory Evaluation Program of the New York State Department of Health.
Two years ago, the lab launched its first NGS-based diagnostic – a mitochondrial genome sequencing test – after validating it on the 454 GS Junior and receiving preliminary approval from the health department.
However, the lab soon realized that it was unable to call some pathogenic mutations in homopolymer stretches unambiguously, even at high coverage, according to Peter Nagy, one of the lab's directors, who is a pathologist by training and in charge of developing NGS-based tests for the lab.
"There were a lot of blind spots," Nagy told CSN, so they revalidated the test on the Illumina MiSeq and launched it on that platform about a year ago. The new test has a turnaround time of about three weeks and is able to detect mutations down to 1 percent heteroplasmy, and sometimes lower. So far, Columbia has run the test on about 100 patients.
Last year, the lab started validation work for an exome test and a large gene panel test on a second-hand Illumina Genome Analyzer. After realizing that the cost of running an exome would be three times higher on that platform than on the HiSeq, the group leased a HiSeq 2500 from Illumina, which arrived last fall.
So far, the HiSeq has performed extremely well in their hands – there has not been a single bad or failed run yet – and the data is "much cheaper and much higher quality" than on the GA, Nagy said.
Nagy thinks Illumina's platforms will remain "the game in town" for clinical sequencing assays for the next few years. He currently has "limited interest" in Life Technologies' Ion Proton because he believes that platform has similar issues with homopolymers as the 454.
Developing its own exome test, rather than outsourcing exome sequencing to an external provider, made sense for Columbia because of its in-house expertise in genetics, Nagy said, and because it makes it easier to transition results into research projects. It also allows the researchers to retain control over and go back to the raw data if needed.
In February, the lab submitted its validation for the exome test – a 200-page document that includes assessments of the instrumentation, bioinformatics pipeline, and clinical evaluation of the data – to the State of New York and has received conditional approval to offer the test while the health department is reviewing the application, which can take more than a year.
Nagy said that, to his knowledge, theirs is the first whole-exome test for constitutional disorders that has been okayed by the state. Mount Sinai School of Medicine's Genetic Testing Laboratory is also working on an exome test (CSN 1/16/2013), which it plans to submit to the health department by the end of May, according to Lisa Edelmann, the lab's director. Until then, her lab is offering the test through the health department's non-permitted laboratory test system, she said, using those cases as part of its validation.
Columbia's exome test uses Agilent's SureSelect method – which performed better in their hands than Illumina's TruSeq capture – to enrich about 70 megabases of DNA.
Libraries are sequenced on the HiSeq in standard mode, using 100-base paired-end reads, which takes about 11 days, though a new run can be started each week because the two flow cells operate independently.
The lab runs fifteen samples and a control per flow cell, yielding about 20 to 30 gigabases of mapped data per sample, or an average coverage of 200x. It sequences 98 percent to 99 percent of the target with at least 15-fold coverage, its cut-off point for the data analysis. Regions that don’t achieve this coverage are consistent between runs and are listed on the lab's website.
The scientists use Softgenetics' NextGene software for calling mutations, which Nagy said is "a user-friendly and quite powerful software package."
To identify potentially disease-causing mutations, they have developed their own set of filters that classifies mutations into 12 categories, using various databases of previously described pathogenic mutations, allele frequencies, and effects of mutations on protein function.
Among them is a database of 28 exomes from healthy individuals ranging in age from 85 to 100 − a control group from an Alzheimer study – that were sequenced at Columbia.
The test report that goes out to clinicians is one to two pages long and lists the findings as either positive, negative, or of uncertain significance. It also includes incidental findings for highly penetrant and actionable mutations, following guidelines released by the American College of Medical Genetics and Genomics in March (CSN 3/27/2013), but it does not include carrier mutations. All mutations listed in the report are confirmed by Sanger sequencing, and patients can opt in or out of receiving incidental findings.
Every test result is signed out by a group of five faculty members who each look at every case. The turnaround time for the test is about two months, although some cases are completed more quickly and others take longer.
The exome test has an out-of-pocket price of $6,000, which includes exome sequencing of up to three family members, who are chosen in collaboration with the ordering clinician. The cost increases by $1,000 for each additional family member.
The lab expects insurance to pay for the test, even for Medicare and Medicaid patients, Nagy said, and Columbia's hospital is willing if the test is clinically justified to pick up the tab for patients where Medicaid does not provide payment.
So far, Columbia has accepted about 30 samples for the exome test, returned results on four, and partially completed another four. In half the cases, the test was able to find a potentially disease-causing mutation.
One way to improve the test's success rate will be to use stricter criteria for selecting patients whose conditions have a clear genetic basis, Nagy said.
Another test his group has been working on is a "superpanel" of about 1,000 genes associated with constitutional disorders, called the Columbia Combined Genetic Panel.
Validation for that panel has been completed and the lab plans to submit it to New York state and start offering it to clinicians this month.
Though the test will sequence all the genes on the panel, physicians will order defined sub-panels for disease areas that match a patient's phenotype, or any set of genes they choose, and only those will be reported.
The panel, which covers about 7 megabases of sequence and uses Agilent's SureSelect capture, can run on the MiSeq, GA, or HiSeq. Unlike for the exome test, the lab guarantees full coverage of all genes, which are sequenced at 1,000-fold mean coverage, by filling in any gaps with Sanger sequencing.
Pricing for the test ranges from $2,000 to $4,000, depending on the size of the panel or number of genes reported. Physicians who later decide to "upgrade" to exome sequencing if the panel test was negative will receive partial credit for the panel test.
"If you have a well-defined phenotype that goes with a relatively manageable number of genes, it makes sense to order those first," Nagy said, whereas the exome test is geared at patients with a condition that is clearly genetic "but all bets are off [as to] what it could be."
Nagy said he expects demand for the panel to be higher than for the exome test because in many cases a patient's phenotype is well defined and because the panel is easier to analyze.
Besides the exome and panel test, the PGM lab is working on a whole-genome sequencing test, which it has almost validated and which it plans to submit to New York state by the end of the month. That test, which will cost about $10,000 per sample, will likely not be covered by insurance and will be reserved for difficult cases.
The lab is also working on a small panel for amyotrophic lateral sclerosis, which will include testing for a repeat expansion in one gene, that it will submit shortly.
Further, in the area of cancer, it is in the midst of validating a glioblastoma test that will combine exome sequencing from blood with exome and transcriptome sequencing from brain tissue. Combining exome and transcriptome results for that test, Nagy said, is a "more complicated analytical problem" than the other tests.
In collaboration with Illumina, the lab is also working on an updated version of Illumina's TruSeq Amplicon cancer panel, and it is working on a transcriptome sequencing test for hematological malignancies.
One of the greatest challenges in developing NGS-based clinical tests has been the lack of a "unified, reliable, rigorously curated" list of disease-causing mutations, Nagy said. While such lists already exist for certain disease types, no single comprehensive database exists.
Another challenge was putting together the consent for the exome test, which took several months. "There was a lot of anxiety about whether the patient wants incidental findings; whether we should report carrier status; or only focus on what we believe is truly the cause of the disease," he said. "We always find a mutation that sort of makes sense – but it has to be more than that. It has to be a very good clinical fit and be supported by good allele frequency data, and, sometimes, models."
Sharing results between clinical labs also will be important in the future. "We have to develop a framework where we can use each others' databases, even if it's for a fee," Nagy said.
Finally, educating physicians on NGS tests and the benefits they offer will be crucial. Clinicians were involved in selecting the genes for the panel test, Nagy said, and it will be important to have them participate in the sign-out of results as well.
"People have to stop considering pathology as just a lab," Nagy said. "We really have to be partners in this. This is bigger than clinical medicine, bigger than laboratory medicine; this is a new field, and it really requires very close interaction of all the partners involved.