NEW YORK (GenomeWeb) – Exome sequencing tests may not cover completely all medically relevant genes and may miss pathogenic variants in disease genes, researchers from the University of Texas Southwestern Medical Center and Thomas Jefferson University reported recently in a study in Clinical Chemistry.
A number of clinical laboratories now offer exome sequencing as a diagnostic test, with most giving patients and families the option of not only receiving diagnostic information but also of receiving secondary findings in 56 genes that the American College of Medical Genetics and Genomics have deemed medically relevant due to their ability to impact patient care.
Clinical exome tests have "what appear to be very good overall stats," Jason Park, medical director of the Advanced Diagnostics Laboratory at the Children's Medical Center Dallas and lead author of the study, told GenomeWeb, in terms of base coverage. "But, those good overall statistics fall apart when you look at small subsets of genes."
In the study, the researchers evaluated 57 exome datasets, 12 of which were performed in a clinical setting and 45 of which were performed in a research setting. Although all datasets had a high mean coverage — between 74x and 120x — the group found that depending on the capture technology and sequencing platform, between 83 percent and 87 percent of bases were covered at 20x or more. Looking at the subset of 56 ACMG genes, they found that more than 50 percent of HGMD variants in seven genes had "inadequate coverage."
The group evaluated three different exome capture kits with two next-gen sequencing platforms: Thermo Fisher's TargetSeq kit with the SOLiD 5500xl, Agilent's SureSelect with the 5500xl, and Illumina's TruSeq kit with the HiSeq 2000.
No pathogenic variants were called in the 56 ACMG genes in any of the samples. However, each method missed a substantial number of coding pathogenic variants in the genes. Looking at the 18,336 variants indicated in the HGMD as pathogenic, SureSelect, TargetSeq, and TruSeq missed 2,352, 1,779, and 719 of them, respectively. Additionally, the three methods poorly covered (defined as below 20x) 842, 1,295, and 1,269 variants, respectively.
Park said that there was a variety of reasons for poor or no coverage of genes, including high GC content and the presence of a pseudogene. Those are known issues, and "we knew there would be insufficient coverage" of those regions, "just not to what extent."
The study is adding to a growing body of evidence suggesting that comprehensive sequencing may not always be the best option for diagnosing disease. For "patients with well-defined test needs" it may make more sense to do a panel, Park said. For instance, when evaluating hereditary cancer syndromes. However, exome sequencing is still a good option for patients with rare or undiagnosed diseases, he added.
Researchers at Harvard's Ocular Genomics Institute also came to a similar conclusion when they compared panels and exomes for diagnosing inherited eye disorders: panels are often better at diagnosing well-defined diseases because exome tests can miss important variants.
Some groups are now offering enhanced exomes to get around this problem. Emory's Genetics Laboratory offers a medical exome with enhanced coverage in around 4,600 medically actionable genes.
Personalis offers its ACE Clinical Exome, which has enhanced coverage in over 8,000 genes. The company says the test has a diagnostic yield of 50 percent, well above the standard 25 percent to 30 percent rate of many clinical labs.
Park said that his group has not tested either of these options but noted that while the enhanced exomes are clearly an improvement, they still miss things. Even whole-genome sequencing leaves a significant number of genes insufficiently covered, he added. Park indicated a study published this March in the Journal of the American Medical Association by a group at Stanford University that found that whole-genome sequencing in 12 volunteer adults had incomplete coverage of some disease genes, among other problematic issues.
As the technology gets used clinically more frequently, "it will be important that when clinicians order this test, they know how well specific genes are covered," Park said.
He added that he thinks there should be more data on false negatives, particularly in secondary genes, such as the 56 recommended by the ACMG for testing.
Park said that his lab at the University of Texas Southwestern does not currently offer a clinical exome test for these very reasons. The lab offers a targeted cancer panel that they run on the Illumina MiSeq and recently also validated an epilepsy panel. The lab is also in the process of trying to figure out what single-gene tests physicians order most often and will then look to try to combine those into various panel tests, Park said.
"As the genomic exome field develops, we'll have to see what the role is for the various tests," Park said. He thinks that even as sequencing costs come down and technology progresses, there will still be a role for panel tests, as well as exome and whole-genome tests.
In the clinic, we "need the right test at the right time," he said, which is different depending on the specific patient and disease.