Early adopters of clinical sequencing are all taking slightly different approaches — from whole-genome sequencing to exome sequencing to targeted panels — but one of the key hurdles they all face is interpretation and analysis, according to speakers at Cambridge Healthtech Institute's Clinical Genome Conference held earlier this month in San Francisco.
Additionally, the complex intellectual property landscape surrounding gene patents can pose challenges for returning clinically relevant results from sequence data.
The "clinical genome is much more than just good data," Tina Hambuch, a senior scientist at Illumina, said during a presentation at the conference. Not only do sequencing calls have to be made accurately, but all that information must then be evaluated according to the American College of Medical Genetics and Genomics' standards for "what level of evidence is available to characterize this variant as pathogenic or likely pathogenic."
During her talk, Hambuch discussed a study Illumina performed in order to evaluate methods for interpreting whole-genome sequence data from a healthy individual.
For one of the first healthy genomes the company sequenced, it used the Human Gene Mutation Database to look for potentially pathogenic variants, Hambuch said. On the first pass, it considered all variants reported in HGMD to be associated with Mendelian disease without looking at the strength of the evidence for those variants, she said.
From that one healthy genome, the researchers identified 16 homozygous mutations that should cause disease and 48 heterozygous mutations, "but this healthy adult did not have any of those diseases," Hambuch said.
After going back through the literature surrounding all of those variants, none of the homozygous mutations met ACMG standards to call as pathogenic.
"The point was not to say that the database was inaccurate," Hambuch said. "We're asking it to do something that it was not designed to do" — clinical interpretation.
Following that initial healthy genome, company researchers then looked at 434 genes that represented 158 highly penetrant Mendelian conditions across five healthy genomes. They only looked for variants that had been reported in HGMD and other databases. Across the five genomes, 75 such variants were found, 60 of which were only found once.
Further investigation revealed that 17 variants could be classified as benign, 29 as likely benign, 22 were of unknown significance, 13 were probably pathogenic, and four were pathogenic.
"This is a doctor's nightmare," Hambuch said, because there are so many variants that are not conclusively benign or pathogenic that figuring out what to do with that information will be difficult. She noted, however, that the more genomes that are sequenced, the more those variants will segregate into either the benign or pathogenic classifications.
Whole-genome interpretation has to take into account the clinical context of the patient, and it "falls into the active practice of medicine and requires someone with appropriate medical training," Hambuch said.
Elizabeth Chao, director of translational medicine at Ambry Genetics, spoke of the challenges Ambry has faced in interpreting results of its Clinical Exome test. Ambry launched the test last fall and reported this spring that it had used it to successfully diagnose three cases (CSN 3/7/2012). It offers its test for between $5,000 and $7,999, depending on whether it is sequencing just the one affected individual or a trio.
Chao said that so far, the company has found diagnoses in between half to two-thirds of all cases, though she did not disclose how many cases the company has tested.
One of the main challenges, she said, is when the test reveals variants in genes that have not been previously implicated in disease. In that scenario, "the line becomes blurred between clinical testing and research," she said.
For example, the company tested a newborn that suffered from hypotonia and severe respiratory distress. At five months, the family withdrew respiratory support and the infant died, but the parents still wanted to know what caused the disorder because they were interested in having more children.
Exome sequencing uncovered only one notable finding — a de novo hemizygous mutation to the gene LAS1L, which is involved in ribosome biogenesis, but had never been implicated in human disease.
However, there are a number of ribosomopathies that have been well documented and included some of the clinical neurodevelopment features seen in the patient.
Chao said that while the case was not clear-cut, Ambry reported the finding back to the family's physician, and is now working with a group in Texas to try and validate the finding.
"When we find a novel finding, we work to get people together to find out if that might be disease causing," she said. Such work includes trying to get more biological data on the gene and functional consequence of the mutation, as well as looking for additional families with similar phenotypes.
Rare diseases can be particularly challenging because there are often no other individuals with the same disorder, so variants cannot always be confirmed.
During a panel discussion at the conference, Gholson Lyon, an assistant professor at Cold Spring Harbor Laboratory, questioned whether variants could be deemed pathogenic if they had not been observed in other individuals. The ACMG has different sets of guidelines for calling variants as pathogenic, one of which is that the mutation should be present in three other unrelated individuals with similar phenotypes.
However, Chao said that in the realm of rare diseases, this does not always happen. To report variants as pathogenic in cases where other individuals have not been identified, the ACMG says that the variant should be a deleterious mutation in a gene that is predicted to cause the phenotype or a nonsense mutation in known disease genes.
She said that novel mutations are frequently uncovered in the CF gene, which causes cystic fibrosis, as well as cancer predisposition genes.
Another challenge with interpretation, Chao said, is that the information changes constantly as new discoveries are made. So in some cases where an exome sequencing test has not yielded a diagnosis, future publications may have the potential to shed more light on variants of unknown significance and could offer evidence that would lead to a diagnosis. Chao said the company is now working to "implement [an infrastructure] to allow us to revisit cases and update when necessary."
In response to a question from the audience on gene patent issues, Chao said that there are fewer than 100 genes for which Ambry does not report results due to IP restrictions. In cases where the company finds variants of significance in a patented gene, the company contacts the physician so that he or she can order the appropriate single-gene test as a follow-up, she said.
Exome-based sequencing tests are more likely to face intellectual property issues than whole-genome tests because gene patents typically cover methods that would specifically target or enrich for the gene, including exome enrichment.
In response to another question, Chao added that whether a patent covers findings from RNA-seq studies would depend on the language of the patent. Typically patents cover methods for "the isolation of DNA," she said, "but some include RNA as well."