NEW YORK (GenomeWeb) – Gene panels have significant practical and cost advantages over clinical exome sequencing for diagnosing Mendelian diseases, according to researchers from Saudi Arabia.
In a study published online in Genome Biology last month, the team found that 13 gene panels they developed covering about 3,000 Mendelian disease genes yielded a diagnosis in 43 percent of patients, a rate comparable to whole-exome sequencing but at a fraction of the cost and in a shorter timeframe.
"We wanted something that's very practical, that any clinician can order," said Fowzan Alkuraya, a principal clinical scientist at King Faisal Specialist Hospital and Research Center in Riyadh and the study's author. "Exome sequencing is popular, but it's too expensive, and it takes time."
According to Joris Veltman, a researcher at the Radboud University Medical Center in Nijmegen in the Netherlands, whose team has championed clinical exome sequencing, this is an "interesting study showing that targeted panels will continue to play a role in genetic diagnostics, especially when price is the major driver."
Alkuraya and his colleagues have done large-scale exome sequencing on a research basis as part of the Saudi Human Genome Program, but diagnostic exome sequencing is currently not available in Saudi Arabia and is outsourced to international laboratories, including in the US and in Germany, which typically charge more than $4,000 per patient, and more for parent-patient trios. "We thought gene panels would be a good alternative," he said.
Two years ago, he and his colleagues set out to design 13 gene sequencing panels encompassing more than 3,000 Mendelian disease genes, each panel designed around a broad clinical theme, such as cardiovascular disease, deafness, dermatology, or vision.
The idea was that selecting a panel test would require "very little clinical input," from physicians, Alkuraya explained, making it easy to order the most appropriate test. For example, if a patient showed any type of facial abnormality, clinicians would select the dysmorphology-dysplasia panel.
The turnaround time for a panel is about 10 days compared to typically months for clinical exome tests, and the cost per panel is between $75 and $150, according to the study.
Alkuraya and his colleagues tested their panels in more than 2,300 patients who presented with a wide range of suspected genetic diseases. For each patient they chose just one of the 13 panels, based on their most prominent clinical feature.
They amplified the panel genes using AmpliSeq technology and sequenced them on Thermo Fisher Scientific's Ion Proton machine, followed by variant calling and data analysis to identify pathogenic or likely pathogenic variants. Potentially causal variants were validated by Sanger sequencing. A major reason for using the Ion Torrent platform was its compatibility with AmpliSeq, Alkuraya said, allowing for high multiplexing and, thus, low assay costs.
Overall, panel sequencing detected likely causal variants in 43 percent of cases, a diagnostic yield much higher than the 25 percent or so reported in several large clinical exome sequencing studies. One reason for this might be the interpretative challenge of exome sequencing.
"With the exome, you have way more variants to sift through compared to the panel," Alkuraya said. "Even though they automate a lot of the variant interpretation in these large exome sequencing labs, there is still the human factor — you go through the final list of variants and try to make sense of them," and the right answer may be missed.
In addition, he said, because consanguinity is higher in their Saudi patient population, they had more recessive mutations, which are easier to spot. Also, older clinical exome studies might have missed variants in genes that were not known to be involved in disease at the time.
Because large causative structural variants would be missed by the panels, the researchers also performed microarray-based karyotyping on 213 patients that tested negative in the panel assay and found likely pathogenic de novo copy number variants in 35 of them.
For the remaining 178 patients, they conducted exome sequencing on the Ion Proton, which uncovered mutations in known disease gene for 20 of them. Of those, 14 had mutations in genes that were not included in the panels, and six had mutations in genes that were actually present in the panels but were missed because of a "limitation of the analytical sensitivity" of the Ion Proton platform.
Conversely, two patients received diagnoses from the panel tests after previously testing negative in an exome test — their mutations were present in the exome data but were missed during the interpretation. The panels uncovered them, "likely because of the smaller number of variants" to be considered, according to the study.
The fact that most of the variants the panels missed but the exomes uncovered were in known genes is good news, Alkuraya said, because those genes can easily be added to the panels through an additional PCR reaction, which costs less than $10. His team is in the process of updating all the panels with Mendelian disease genes that were discovered since 2013 — fewer than 200 in total.
One sticking point is the low sensitivity of the Ion Proton platform to insertions or deletions. In a validation study of the panels, the researchers found the sensitivity for single nucleotide variants to be 93 percent, but for indels, it was only 72 percent, largely due to errors in homopolymer regions. "That's something we really need to fix," Alkuraya said.
For known mutations, bioinformatics can provide a temporary fix by lowering the threshold for calling variants in certain areas. "But it's not going to work for the unknown mutations," he said. "For that, we really need to improve the overall sensitivity and specificity of the platform."
A possible solution is the Hi-Q sequencing chemistry that Thermo Fisher recently released for the Ion Proton, which is supposed to improve accuracy for both SNVs and indels, and Alkuraya's team is currently testing the new chemistry, expecting results within the next few weeks. "I really hope the Hi-Q solution is going to live up to its promise and will bring about the 80 percent improvement," he said. "If not, we should definitely consider other possibilities as well."
According to Veltman, clinical exome sequencing does have advantages over gene panels. For example, exomes pick up variation in all genes across all disorders, which he said "is essential for diagnostic interpretation."
Also, for patients suffering from multiple disorders, it may be difficult to pick the right panel. "With whole-exome sequencing, this is no problem. You can simply perform the analysis for the deafness genes and next look at the blindness genes, for example," he said.
The diagnostic yield of exome sequencing is also likely to increase as new disease genes are being discovered, and old exome data sets can easily be reanalyzed, whereas panel tests need to be repeated with updated panels.
Finally, in Veltman's experience, the cost differential between panel and exome sequencing is less pronounced than what the authors cited in their study.
In the meantime, the panels have been well-received by clinicians in Saudi Arabia, Alkuraya said, allowing them to screen large numbers of genes for a much lower price than more narrow gene panels offered by other diagnostic laboratories.
At his hospital, panel testing has become the first tier test for patients with suspected Mendelian disease, he said, and is followed by exome sequencing as a second-tier test. In patients with no family history of disease, a copy number array is run in parallel with the panel to discover structural variants.
Whole-genome sequencing remains reserved "for very few selected cases," because they are so much more difficult to interpret, he said.
Even for research work, the panel approach has proved helpful, "because we can easily eliminate cases that are caused by known genes and focus our energy on the novel genes," he said. "Instead of spreading our attention over 100 cases, now we can focus on the 20 cases that are likely to be novel."