Discussion
In this study we report the clinical application of WES in the diagnosis of 162 cases referred to a single academic genetics center. The overall molecular diagnostic yield reached 52.5%, slightly higher of that reported in many previous studies (Jalkh et al., 2019; Sawyer et al., 2016; Seaby et al., 2016; Valencia et al., 2015). This rate was observed to vary depending on the disease category (Figure 3), but is also likely influenced by the strategy applied. In our department, the strategy included: (1) Analytical clinical description of the patient’s phenotype along with comprehensive family and medical history to include as many data; (2) Stringent phenotype-driven strategies adopted for variant filtration and interpretation (Figure 2); (3) Close collaborations between clinicians and laboratory geneticists for variant evaluation and phenotype-genotype correlation; (4) Family segregation strategies with Sanger sequencing to support evaluation of variant pathogenicity when necessary. Besides potentially optimizing the diagnostic yield, the strategy supported a reduction of the number of variants in need of classification and an acceptable turn-around time and reduced the overall cost to the patient, paramount in the absence of NGS-reimbursement in Greece. Every attempt was made to use new and/or updated annotation and pathogenicity prediction tools and variant databases, of great value in the context of variant evaluation. Recognizing the importance of data sharing, 48 novel pathogenic or likely pathogenic variants were identified in this cohort and uploaded to public databases.
A few cases in this cohort demonstrated that complex phenotypes may be caused by complex genotypes, potentially complicating both clinical evaluation and data interpretation. For example, in cases 8039 and 9151 (Table 2), the findings in TGFB3 and MYH7 genes respectively, were classified as VUS due to the lack of cardiological data at the time of WES analysis and the report included a recommendation for subsequent re-evaluation.
Rare genetic phenomena including variable expressivity, incomplete penetrance and genetic heterogeneity, should also be considered when assessing WES results. At least 2 cases had secondary variants potentially contributing to the clinical expression of the primary condition. In case 9081 (Table 2), referred with polycystic kidney disease at the age of 31, the detection of the co-inheritance of the pathogenic PKD2: c.1837C>T variant associated with autosomal dominant PKD2, and variant c.6992T>A in thePKHD1 gene, (associated with autosomal recessive PKD4) could explain the relative early onset of the disease, althoughPKHD1 :c.6992T>A classification is VUS (Bergmann et al., 2011).
Case 9115 (Table 1) presented with spastic paraplegia with lower limb spasticity and weakness. WES identified a known pathogenic variant c.1245+5G>A (Polymeris et al., 2016) of paternal origin inSPAST gene, along with a known modifier polymorphism c.131C>T of maternal origin in the same gene (Parodi et al., 2018). Usually variants in SPAST are associated with spastic paraplegia type 4, where the age of onset ranges between 20 to 60 years; the infantile onset in this case might be attributed to the presence of the modifying polymorphism in trans .
Case 9045 (Table 1), highlights the importance of adjunct genetic analysis, such as classic karyotyping, since ZFPM2 pathogenic variant are disease causative only in 46,XY individuals. A 6 years-old female with a 46,XY karyotype, was referred for Disorders of Sex Development (DSD). A novel, likely pathogenic variant c.192T>G of maternal origin was detected in theZFPM2 (FOG2 ), related to 46,XY sex reversal 9 (SRXY9). Functional studies have suggested that failure of testis development is most likely caused by impaired or no interactions of the mutant ZFPM2 (proteins) with GATA4 gene, which regulates early testis development (Bashamboo et al., 2014).
Pleiotropy was another interesting observation in this cohort.GJB2 gene is associated with autosomal recessive deafness or autosomal dominant Vohwinkel syndrome (De Zwart-Storm et al., 2011). Two unrelated patients with deafness (Table 1) were homozygous for pathogenic variants c.71G>A (case 8016) and c.35delG (case 8039) in GJB2 . Interestingly, a likely pathogenic variant c.524C>G was revealed in the same gene in case 9088 (Table 1) with ichthyosis and keratitis but no hearing impairment; heterozygosity of GJB2: c.524C>G likely explains the dermatological findings. Unlike recessive GJB2 variants, in which the spectrum and phenotype-genotype correlations have been analyzed clearly, only a few studies of GJB2 dominant variants have been reported. However approximately two-thirds of dominant GJB2mutations are shown to cause syndromic hearing loss associated with diverse skin lesions(H. Wang et al., 2017).
In cases where one gene is linked to more than one disease or disease variants, like PTPN11, distinction may be based on differential clinical features or previous reports of the specific variant.PTPN11 is usually associated with Noonan syndrome 1, as additionally supported by findings concerning 5 unrelated patients in this study (8112, 8133, 9080, 9138, 9148) (Table 1). In case 9165 (Table 1), a PTPN11 variant c.1381G>T was associated with Leopard syndrome 1 (Carcavilla et al., 2013), but the patients’ diagnosis was based on the phenotypic findings rather than the genotype (Osawa et al. 2009). Diligent clinical examination, biochemical tests and other pre-WES genetic tests, such as conventional and molecular karyotype, all have important roles in supporting the interpretation of WES results.
In conclusion, classical genetic diagnosis strategies involve targeted single gene/exon analysis often with negative or inconclusive results. Within the referred period that our academic diagnostic laboratory uses WES, 85 cases were resolved upon the detection of 94 pathogenic variants 48 of which were novel contributing valuable information to the genotype spectrum of the disease. Despite the technical and diagnostic challenges associated with NGS, (phenotyping, clinical and genetic heterogeneity and variable expressivity), WES provides a unique opportunity to resolve the genetic etiology of disorders, end diagnostic odysseys and support the provision of appropriate medical management and genetic counseling. When used within the context of effective multidisciplinary collaboration between clinicians and laboratories, the clinical WES diagnostic tool becomes efficient and cost-effective and thus appropriate for first-tier genomic analysis.
However, it is important to recognize the limitations of WES, including low coverage of coding regions, inefficient capture of GC-rich regions, misalignment of reads to reference genome, presence of pseudogenes, homologous or repetitive regions, large deletions/duplications, complex rearrangements, triplet repeat and imprinting variants (Bertier et al. 2016; Taylor et al. 2015; White et al. 2017). Polygenic and/or multifactorial diseases and limited data on VUS in coding genes not yet associated with human diseases hinder interpretation and call for alternative approaches, such as functional studies. Finally, Whole Genome Sequencing potentially addresses many of the limitations of WES and can additionally detect non-coding (deep intronic or non-coding RNAs) and structural variants (Smith et al., 2019).