Selecting the appropriate panel for a patient can be difficult. “Not every patient fits into a neat phenotypic box or meets diagnostic criteria for a particular genetic condition. WES removes the challenge of trying to find the single correct panel, or the need to create a custom panel, for your patient as all protein-coding genes are covered in one single test,” senior geneticist Jennifer Schleit explained.
Whole Exome Sequencing (WES) is often chosen for patients with complex phenotypes affecting many organs or body systems, when more than one disorder is suspected, when previous genetic testing has not yielded informative results, or when the suspected genetic disorder might not have a specific test available. “However, there are many important factors to consider when choosing a whole exome test for your patient,” Schleit said.
#1. There is a big difference between 99% coverage and 97% coverage
How much of the exome is actually being sequenced? WES technology is relatively recent, and it is not yet possible to capture and sequence 100% of the exome at high quality.
“On average, we capture and sequence >99.4% of the exome with a quality enabling reliable variant calls. Coverage also refers to how many times each nucleotide is being sequenced. This is sometimes referred to as sequencing depth, and it is ideal to have a minimum depth in the order of 20x”, Schleit says. The average sequencing depth does not indicate how much of the coding regions are covered, so it is important to have both good sequencing depth and good, uniform coverage.
Small differences in percentage coverage may not seem significant; however, they translate to big differences at the sequence level and the diagnostic impact can be substantial.
“The difference between 99% coverage and 97% coverage can mean hundreds of genes and thousands of exons are not covered optimally. These differences in coverage are especially important when it comes to difficult-to-sequence genes, like PKD1 associated with polycystic kidney disease and RPGR associated with X-linked retinitis pigmentosa,” Schleit explained.
“We have worked hard to get maximized coverage with customized sequencing solutions in difficult-to-sequence genes like PKD1, RPGR and SMN1/SMN2 and we will present a validation study for improved coverage of PKD1 at ACMG 2019 in Seattle and RPGR at ARVO 2019 in Vancouver.”
#2. Known disease-causing deep intronic variants are not included in every exome
It is important to remember that genes are made up of more than just exons; approximately 15% of disease-causing variants are within introns. Many of these occur at the exon-intron boundary. However, there is increasing information about the clinical importance of deep-intronic, noncoding variants. Schleit provided an example.
“We analyzed a patient suspected to have a hereditary immunodeficiency. Previous genetic testing had been negative. With our whole exome test, we identified a hemizygous, likely pathogenic variant in the promoter region. This variant would have been missed with an out-of-the box exon-only capture kit,” Schleit said.
“We are hoping to get more comprehensive statistics from our internal data on how significant an impact deep intronic variants have had on the diagnostic hit rate. At the moment, we believe that their importance is probably still underestimated in the field,” Schleit continued.
#3. High resolution CNV detection is a powerful diagnostic tool
The detection of CNVs (copy number variations) is a relatively recent capability in NGS platforms but is a crucial component of high-quality exome analysis. CNVs are an important disease mechanism that should be evaluated in all patients with a suspected inherited disorder.
“There was a really good study about this by Truty et al in June 2018. They looked at 143,000 individuals who underwent genetic testing. A pathogenic or likely pathogenic CNV was found in nearly 10% of individuals, a clinically significant result for sure,” Schleit underlined.
“Diagnoses also come from small deletions and duplications. Depending on the clinical specialty, CNVs can represent a significant portion of the disease-causing variants identified. Truty et al also found that in some clinical specialties, CNVs accounted for up to 35% of pathogenic variants identified.”
#4. A phenotype-first approach means that some genes, and therefore variants, are filtered out before analysis even begins
Having high-quality sequencing data is important, but how that data are analyzed is also important. A genotype-first approach ensures disease-causing variants are not filtered out of the analysis based only on our current understanding of genes and the phenotypes they cause.
The genotype-first approach allows identification of atypical presentations of known disorders, diagnoses in newly established genes or even candidate genes. It can be especially powerful in neonates and young children when the phenotype might not be complete.
“Again, genotype-first is valuable for those patients who do not fit into the box of a certain phenotype or diagnostic criteria for a particular condition. However, using a genotype-first approach does not remove the importance of having accurate and detailed clinical information to support high-quality interpretation of the data,” Schleit said.
#5. The inclusion of candidate genes can result in a diagnosis for your patient
In cases where known disease genes are negative for pathogenic and likely pathogenic variants or where only part of the phenotype is explained, it can be useful to also analyze and report candidate variants in genes not yet known to be associated with disease.
“Last year, I worked on the exome of a severely affected 3-month-old patient born preterm, with respiratory distress, a likely skeletal dysplasia and possible scoliosis, and with family history of short stature. We did identify a likely pathogenic variant in the SHOX gene, but it didn’t seem to explain the entire phenotype. So, we kept looking in the exome data, and identified a homozygous nonsense variant in SLC10A7. This variant caught our attention because it is predicted to completely abolish expression of SLC10A7 and individuals with homozygous loss of function variants are not present in control populations,” Schleit described.
According to Schleit, this variant had not been reported in population databases and, at the time, a PubMed search was negative. A pre-publication by Ashikov et al (2018) was identified using another search engine and actually reported a small series of patients with variants in this gene and the same phenotype.
“Having a genotype first approach allowed us to identify a highly suspicious variant for further investigation”, Schleit concludes, mentioning that soon after, further evidence to support the gene-disease association was reported (Dubail et al 2018).
#6. WES reduces the time from presentation to diagnosis, reduces the cost of a diagnosis and has an increased diagnostic yield
Recent publications have shown that WES can reduce the time to, and cost of, a diagnosis. A prospective study by Vissers et al (2017) revealed that WES identified significantly higher conclusive diagnoses (29.3%) than the standard care pathway (7.3%) without incurring higher costs.
Currently, WES also provides a better cost per diagnosis ratio than Whole Genome Sequencing (WGS). Recently Alfares et al (2018) studied 108 patients with negative initial WES and arrayCGH using WGS and found a new genetic diagnosis in 9.3% (10/108) of the patients. Three of the initially missed diagnoses were already present in the original WES data with decent coverage but remained unnoticed, thus being detectable by simple WES reanalysis. WGS was only able to achieve a 7% higher detection rate than their standard WES.
“There is a narrow gap between our WES and WGS since our customized WES assay includes >1500 clinically relevant non-coding regions and provides high-quality CNV detection. We analyzed all the variants missed by WES analysis in their study, and found out that our customized WES would have detected all of them including the PHOX2B alanine repeat expansion (25/20), the TPM3 deletion (5 exons), and the deep intronic variant in TSC2,” said Schleit.
This means that a customized WES may provide similar clinical sensitivity as WGS in a clinical cohort. The conclusion: reanalysis of old WES data is a reasonable first step.