NEW YORK (GenomeWeb) − Highlighting the potential clinical utility of whole-genome sequencing, researchers in the Netherlands have found that it provided a diagnosis for 42 percent of patients with intellectual disability where other tests, including exome sequencing and genomic microarrays, failed to yield an answer. Based on their results, they estimated that whole-genome sequencing has a potential diagnostic yield of 62 percent for patients with the disorder.
In an article published in Nature today, the researchers, from the Radboud University Medical Center in Nijmegen, presented results from the study for which they sequenced the genomes of 50 patients with severe intellectual disability and their unaffected parents. The patients had undergone extensive genetic testing in the past, including targeted gene sequencing, exome sequencing, and diagnostic microarrays, but not received a diagnosis. Based on an analysis of de novo single nucleotide variants and copy number variations they found in the coding regions of the genome, they were able to provide a firm diagnosis for 21 of those patients.
The study is a "very nice illustration of the power of genome sequencing," said Joris Veltman, a professor of translational genomics at Radboud UMC and the senior author of the paper. "For genetically heterogeneous diseases like intellectual disability, this is definitely the best test. It's better than a microarray and better than an exome, and it combines [the capabilities of] those two."
For their study, BGI's Complete Genomics sequenced the genomes of 50 patient-parent trios to 80-fold average coverage. The families were part of a cohort of 100 patients who had previously undergone exome sequencing, which had yielded a diagnosis in 27 cases. All patients were part of an even larger cohort of about 1,500 who had previously tested negative with genomic microarrays, which had a diagnostic yield of about 12 percent.
The researchers' analysis focused on de novo SNVs and CNVs, which are known to play an important role in intellectual disability. They found 84 de novo SNVs in protein-coding regions and eight de novo CNVs, which were enriched in known genes for intellectual disability as well as candidate genes.
For diagnostic purposes, they further evaluated the pathogenicity of these mutations using various criteria, including a comparison of the patient's phenotype with the phenotype described in the literature. Based on that analysis, they reached conclusive diagnoses for 14 patients with mutations in known disease genes and six patients with mutations in candidate genes. In another patient, they found a compound heterozygous mutation in a known disease gene, where one deletion was inherited from the father and the other from the mother.
In addition, they found 43 de novo mutations in promoters, introns, or untranslated regions of known intellectual disability genes, which they are currently following up on with functional studies in order to determine their possible role in disease.
Many of the new diagnoses were based on point mutations in coding regions, begging the question why exome sequencing did not pick them up before. According to Veltman, most were in coding regions that were not covered well by the exome test. He also pointed out that the exomes were generated in 2010, and that capture technology has since improved.
Past studies by others such as Mike Snyder's group at Stanford found that exome sequencing can detect important variants that whole-genome sequencing misses, mostly because it generates better coverage.
The Dutch researchers looked if that was also the case for their samples and found that genome sequencing picked up 16 out of 17 de novo variants that had been previously identified by exome sequencing. "There is very little that we missed," Veltman said, noting that one important result from their study is that "a genome gives you a very good exome."
Another recent study by Stanford researchers found that whole-genome sequencing of 12 individuals generated insufficient coverage for between 10 percent and 19 percent of inherited disease genes and did not detect variants with important clinical effects reproducibly.
"We looked at our data and completely disagree with that," Veltman said. One reason for that discrepancy might be advances in technology, he suggested – while the data for the Stanford study was generated in 2011 and 2012, their own data was produced last year.
The majority of new CNVs that genome sequencing discovered in their study were too small to be picked up by the types of diagnostic microarrays that are used by most groups. "Genome sequencing is much more powerful for structural variation detection than even the highest resolution microarrays," Veltman said.
Also, genome sequencing yields additional information about the breakpoints of structural variants that microarrays cannot detect but that might be important for diagnoses. For example, in one patient, genome sequencing identified a duplication that moved from chromosome 4 to the X chromosome, where it disrupted a known intellectual disability gene.
Not everyone agrees with the results of the study. "I really think that their idea that they are picking up such a larger percentage by doing whole genome rather than whole exome is inflated significantly," said Sherri Bale, managing director of genetic diagnostic firm GeneDx.
For one thing, she said, the patients in the study received diagnostic genomic microarrays between 2003 and 2013, and the diagnostic yield from modern arrays is likely higher than that from arrays 10 years ago. Also, exome sequencing should have picked up more of the mutations in coding regions that genome sequencing did, which might be explained by the fact that the researchers used older exome technology.
In addition, she does not agree with the researchers' definition of a candidate gene – a gene that was mutated in one to four patients with intellectual disability – because such genes could also be mutated in healthy individuals.
Bale said she believes whole-genome sequencing will allow doctors to make additional diagnoses in the future, once we know how to interpret mutations in non-coding regions, but it will take time to get there. For now, "we are not going to be able to interpret a lot of what we see," she said. "We have enough trouble interpreting what we see in exomes."
Veltman agreed that more research is needed to interpret mutations in non-coding regions. "If we were to implement genome sequencing in diagnostics, we would do the diagnostics on the coding regions [first], and have the best assay for both point mutations and structural variations, and then, in addition, we would learn, by collecting data from many patients, about variation in the non-coding part," he said.
Several barriers exist for implementing whole-genome sequencing as a diagnostic test for intellectual disability, he said, including cost, sequencing capacity, and infrastructure issues related to data volume.
Several genetics groups in the Netherlands are currently collaborating to jointly invest in a sequencing system, either from Illumina or BGI's Complete Genomics, in order to be able to offer clinical whole-genome sequencing, he said, but they need to secure additional funding first.
While exome sequencing might still be the most efficient way to test as many patients as possible, for an individual patient with a genetically heterogeneous or extremely rare disease, starting with genome sequencing might be the quickest way to reach a diagnosis. "I'm sure the genome will be completely standard within, hopefully, five years," he said.