For a little over a year, doctors at Columbia University have been testing clinical exome sequencing to help diagnose patients with suspected genetic disorders of unknown origin. The exome test enabled them to make firm diagnoses in about one-third of cases, but it will be important to demonstrate the test's clinical utility and cost-effectiveness going forward.
Columbia clinicians recently completed exome tests for the first 150 cases, trying out and comparing different laboratories in the process. "I think it's an incredibly powerful tool," Wendy Chung, an associate professor of pediatrics and medicine and director of the clinical genetics program at Columbia University, told Clinical Sequencing News. "I can easily see this becoming a first-line test."
However, she said, turnaround times have been increasing, in one case to a full year, which might hamper the wider adoption of exome testing. "The biggest problem that I'm seeing is one of capacity," she said. "Everyone is starting to realize how valuable this in fact is."
Chung and her colleagues decided to send out patient samples to four different laboratories that offer clinical exome sequencing – Ambry Genetics, Baylor College of Medicine's Medical Genetics Laboratories, GeneDx, and Columbia's own Personalized Genomic Medicine Laboratory – in order to compare their services, though they did not send the same sample to more than one lab.
Services differ in a several areas, Chung said. For example, some labs allow doctors to opt out of incidental or secondary findings, an option that Chung said many of her Orthodox Jewish patients prefer.
Also, some labs offer to reanalyze the data after a period of time, during which new research might have linked additional genes to a disease. "We found that helpful," Chung said, noting that re-analysis enabled them to solve two cases where they did not get an answer initially – once within a week, when a new scientific paper came out, the other time after a year, during which a new gene had been found.
Another important difference between labs is whether parents or other family members, such as affected siblings, are included in the testing. This, she said, increases the sensitivity for de novo and rare inherited mutations, adding that "we found it more helpful to work with laboratories that sequence the [exomes of the] parents upfront."
From a practical standpoint, some laboratories were easier to deal with in terms of their billing policies and how they handle insurance, she said.
Across laboratories, turnaround times have increased over the last year, Chung noted, as demand for clinical exomes has increased and the labs have reached capacity. Ambry, for example, took a year to return a result, which she said might have had other reasons than capacity but caused problems nevertheless. "A year is a long time to wait for a diagnosis."
Overall, the Columbia team has had "better luck" with GeneDx's service, which offers re-analysis after a year, includes family members in the analysis, and has handled insurance issues well, Chung said.
The first 150 Columbia patients undergoing exome testing included some who Chung had been following for years without getting an answer, but where she had "a strong clinical sense that there was an underlying genetic problem."
A disproportionately high number of the first cases suffered from neurological conditions, but also included heart conditions, syndromic birth defects, and hearing loss.
For about a third of the 150 patients, she and her colleagues were able to come up with a definite and unambiguous diagnosis. Part of the reason for this relatively high success rate is likely related to who they send for testing, based on their clinical judgment. "If I sent everyone who walked through my door, I would have a much lower positive rate," Chung said.
Another likely reason is the inclusion of good clinical data and of other family members, in addition to the patient. "We armed the lab with the best possible package, both in terms of clinical data to correctly interpret the [genomic] data as well as the right family members to be able to filter the data," she said.
In some cases, they were only able to get a diagnosis because of some prior knowledge and time-consuming manual data analysis. "In one case, I knew for a particular reason that I was limited to a certain genomic interval, I knew the mutation had to be there, and it was a matter of going through the unaligned sequences that were in the garbage can, or going through [areas] with low read depth," Chung said. "I know there are things that we routinely miss that if you knew to look at a certain place, you could find them, but it's not easy to do."
Many of the successful diagnoses were for well-known genetic disorders, and the mutated gene had previously been associated with the disease, but "if you had gone by the textbook in terms of what the clinical description was, our patient did not fit that," she said. The new findings will help to redefine and broaden the disorder.
In about a quarter of cases, the clinicians identified a new gene as the probable cause of a new type of disease, supported by how the mutation segregated in the family. For some of these patients, they were able to find a second family with the same disorder, sometimes carrying the exact same mutation in the same gene, mostly through literature searches and networking with doctors internationally.
A centralized database would facilitate finding other families that match these "N of 1" cases, Chung said. "It would be tremendously helpful if you could, in a de-identified way, go in, query these things, and have what I call an honest broker to be able to connect people to each other."
While the NIH's ClinVar database might serve this purpose for clinical diagnostic laboratories, it does not cater so much to research laboratories, which are "sitting on a lot more genomic data than the clinical laboratories," she said. Depositing data into ClinVar is not trivial, for example, and research labs might not have the bandwidth to do so, especially since unlike clinical laboratories, they might not directly benefit from this.
For about half the patients, the Columbia team has been unable to identify a disease-causing mutation, indicating that "this technology clearly is not perfect," Chung said. The exome might be incomplete, calling mutations in insertions or deletions is still difficult, and mutations might reside in non-coding regions of the genome.
In cases where the exome failed – sometimes because parts of it were not captured or covered adequately – Chung and her colleagues are now following up with whole-genome sequencing. But even that may not yield an answer because there are not enough reference genomes available yet to interpret mutations in regions outside the exome.
Finally, the disease might have other causes. "Not everything is genetic," she said.
Going forward, Chung expects exome testing to be used earlier in the diagnostic process than before. "At the beginning, we used this mostly for diagnostic odysseys, patients that had already gone through panel testing, microarrays and metabolic workups, and this was the fifth test," she said. "Increasingly, as we realize how useful it can be, it's becoming more of a first-line test."
Indications for the test will also likely expand beyond mostly neurological problems, she said. "Now that patients and other providers are hearing about our success, many patients with a wide variety of familial conditions that are thought to be genetic, not just neurological, will come asking for help."
It will also be important to convince health insurance companies that the test, which most labs offer at several thousand dollars, is worth paying for. "I can easily see from some back-of-the-envelope calculations we have done … how it changes medical care, how we avoid additional diagnostic testing, other procedures, [and prevent] additional affected children," Chung said, but more evidence is needed.
"That's clearly going to be important to the equation of whether or not laboratories will continue offering this service, and whether or not the fee structure will continue to be as it is," she said. So far, none of her patients has been faced with a large bill from one of the testing labs, but it is unclear how much those labs are getting in reimbursement.
To prove clinical utility and cost-effectiveness, Chung is collaborating with health economists at the Columbia Mailman School of Public Health and others, collecting data on how patients use the exome results and how the results affected them, what types of further tests were avoided, and how the test changed patient care. "Families and patients do make particular decisions about being more or less aggressive about their medical care based on having better information about prognosis," she said, something that is not always easy to quantify.
She and her colleagues will also continue to try out new exome sequencing services. The New York Genome Center, for example, is working on opening a CLIA-approved and state-licensed laboratory in order to be able to offer clinical testing. "I'm excited about the possibility, they have got really good people there in terms of informatics," Chung said. "We'll test drive them just like we did everyone else in terms of how well they perform and how user-friendly they are, and I hope they are another good option."