NEW YORK – Researchers at Children's Hospital of Philadelphia and Al Jalila Children's Specialty Hospital in Dubai have found that RNA sequencing in a cell line derived from blood can identify pathogenic variants in patients with neurodevelopmental disorders that were missed by DNA testing.
The approach, which the researchers plan to develop into a clinical test, could complement diagnostic exome or genome sequencing.
In a proof-of-concept study published in Genetics in Medicine on Wednesday, the researchers, led by Ahmad Abou Tayoun at Al Jalila and Mahdi Sarmady at CHOP, applied their method to patients with Cornelia de Lange syndrome, a neurodevelopmental disorder with a wide array of symptoms. In nine out of 10 patient samples, they were able to validate known pathogenic variants, and in three out of five patients with no known mutations, they established a genetic diagnosis.
The use of RNA sequencing to bolster molecular diagnoses of patients with genetic conditions is not entirely novel — Ambry Genetics, for example, last year added RNA-seq to its hereditary cancer risk tests to help identify and interpret splice site variants. Others have used RNA-seq in blood to help diagnose various rare genetic diseases, RNA-seq in muscle biopsies to diagnose patients with rare muscle disorders, and RNA-seq in cultured fibroblast to provide a diagnosis to patients with mitochondrial disorders.
What sets the new work apart is its focus on neurodevelopmental disorders and that it establishes RNA-seq in B-lymphoblastoid cell lines (LCLs), which are derived from B cells in the blood, as a way to diagnose them, according to Abou Tayoun, director of the genomic center at Al Jalila.
Abou Tayoun, who is also a professor at Mohammed Bin Rashid University College of Medicine, explained that only about 40 to 50 percent of patients with neurodevelopmental disorders obtain a genetic diagnosis from clinical exome sequencing, even if the data is reanalyzed later on. "We were thinking that the next step is RNA sequencing because it tells you the consequences of a deep intronic variant," for example, that has no established function, he said. "Our goal was to do this for a [specific] disease."
Since CHOP, where he worked when the study began, had cell lines for a number of patients with Cornelia de Lange syndrome available who were consented for research, he and his colleagues decided to focus on that rare disorder. About 60 percent of patients with CdL syndrome have a pathogenic variant in the NIPBL gene; about 10 percent have a mutation in SMC1A, SMC3, HDAC8, or RAD21; and the remaining 30 percent have an unknown genetic cause.
Initially, they compared gene expression profiles in LCLs and in whole blood and found that LCLs expressed nearly twice as many genes as blood and more Mendelian neurodevelopmental disease genes. In addition, the expression of transcript isoforms overlapped 63 percent between LCLs and brain tissue from GTEx (the Genotype-Tissue Expression sequencing project).
"We are able to detect [these isoforms] in LCLs, so if there are any intronic or splice events in the brain, we can catch them in LCLs," Abou Tayoun said. There were some neuro-specific genes that were not well expressed in LCLs, such as ion channels, pointing to a limitation of the approach, but overall, the data showed that LCLs "present a good source for RNA testing for neurodevelopmental disorders," covering about 1,000 relevant genes, he said.
“We are increasingly finding that cell types that seem unrelated to a rare disease can be informative for identifying dysregulated gene function," said Stephen Montgomery, an associate professor at Stanford University School of Medicine who was not involved in the study, in an email. Last year, he and his team published a study that used RNA-seq in blood to identify rare disease genes. "This paper provides one example where an easy-to-acquire cell type is informative for neurodevelopmental diagnoses," he said.
Jim Dowling, a researcher at the Hospital for Sick Children in Toronto, who was also not involved in the project, said in an email that neurodevelopmental disorders “present a unique challenge for RNA sequencing because the ideal source material (cortical neurons) is rarely available for testing.” In their study, the researchers overcame this barrier by using LCLs. “Somewhat surprisingly, LCLs have a transcriptome that greatly overlaps with that of the brain, both in terms of meaningful gene expression and isoform profile,” he said, very different from whole blood, which represents the CNS transcriptome poorly.
Next, the researchers selected 10 LCLs from CdL syndrome patients with known pathogenic mutations in four genes and ran them through their transcriptome sequencing pipeline, restricting their analysis to 14 disease-causing genes. Overall, they were able to call nine of the 10 pathogenic variants, missing one nonsense variant, probably due to nonsense-mediated decay of the transcript. For one of the patients, they uncovered a mutation in a second gene, leading to an additional diagnosis with a syndrome that has overlapping clinical features with CdL syndrome.
They then went on to test their approach on five unsolved patient cases who had negative genetic testing and were suspected to have CdL syndrome or a related condition. For three of these, they obtained a genetic diagnosis – two abnormal splice events in NIPBL and one missense variant in BRD4. The BRD4 variant could have been detected by an DNA-based test, they noted, if the gene had been known at the time of the test.
One potential drawback of the LCL RNA-seq approach is the time and cost required to establish the cell line, the authors wrote, which takes about a month. However, Abou Tayoun said he does not believe the test would be cost prohibitive and that LCLs are relatively easy to make.
Also, though the researchers only tested their approach on CdL syndrome, it is likely applicable to a long list of other neurodevelopmental disorders including glyoscylation disorders and mitochondrial disorders.
Abou Tayoun said he plans to start offering the test for clinical diagnosis at Al Jalila for patients with a high suspicion of a neurodevelopmental disorder that involves one of the genes the test covers well. One option would be to provide it as a reflexive test after negative DNA testing or DNA testing that identified variants of unknown significance. "Having [a test like this] out there for Mendelian pediatric disease I think is quite novel," he said. "The focus now is to push it to diagnostics for those genes that we think are amenable to RNA analysis."
Getting reimbursement for the test, though, will likely require additional studies. "We need more papers like this across different disease areas to show the utility of RNA-seq," he said.
In addition, according to Dowling, further work is needed to identify suitable tissues or cell lines for the 20 percent or so of neurodevelopmental disorders that are not well represented by the LCL transcriptome. “That stated, this study represents a major advance in terms of the applicablility of RNA-seq for clinical genetic diagnostics,” he said.