The Need for Clinical Diagnostic Proficiency in the Genomic Era

Genomic testing is increasingly being integrated into clinical care.1,2 Genetic testing has also become more complex, transitioning from hypothesis-driven targeted testing to broad-based testing that includes over 20 000 genes.3 Prior to the 2010s, genetic testing largely included karyotypes, fluorescence in situ hybridization, chromosomal microarrays, and sequencing of select single genes, which respectively allowed for detection of chromosomal deletions, duplications, and structural abnormalities and variants in select target genes. Completion of the human genome sequence and declining technological costs allowed for development of gene panel testing, which permits detection of sequence and copy number changes in one or multiple genes associated with a given disease or phenotype. These targeted testing methodologies are being replaced with exome and genome sequencing, which simultaneously interrogate all genes and can detect single nucleotide variants, copy number variants, and trinucleotide repeat expansions in an unbiased manner.4 Although the advancement of genomic testing represents a triumph in clinical diagnostics, these tests have limitations and there are important considerations for ensuring accessibility to genetic testing and professionals trained in test selection, interpretation, and follow-up.Selecting the appropriate genetic test is critical. Though most monogenic syndromes can be detected through broad-based testing such as exome or genome sequencing, there are important diagnoses that are missed using these technologies, highlighting that there is no universal, one-test-fits-all test for all genomic conditions. For example, balanced and unbalanced translocations are better resolved by karyotype or fluorescence in situ hybridization depending on size. For aneuploidy syndromes in which mosaicism plays a role in management, such as Turner syndrome (monosomy X) in which the presence of Y-chromosome material signals increased risk of gonadoblastoma and prompts gonadectomy, a combination of karyotype and chromosomal microarray capable of detecting low-level mosaicism is warranted. For syndromes associated with epigenetic changes such as Beckwith Wiedemann syndrome and Prader Willi syndromes, DNA methylation studies are required. Trinucleotide repeat expansion disorders, such as Fragile X syndrome and myotonic dystrophy, can be detected by whole genome sequencing (WGS); however, these diagnoses can be missed, or the reported repeat length can be inaccurate, and WGS may not include all possible conditions caused by trinucleotide repeats.5The results of genetic testing may require specialized interpretation. Not all genetic differences are deleterious; natural variation exists across populations. Though efforts are being made to catalog natural variation in people of diverse ancestries through public databases and repositories, these databases are not exhaustive, and individuals of non-European descent are often poorly represented.6,7 Therefore, genetic testing reports often include “variants of uncertain significance,” genetic changes that are possibly but not definitively disease causing. Variants of uncertain significance are found most in individuals of non-European ancestry that were historically underrepresented in genomic research.8 The significance of these variants can often be clarified through careful physical examinations and nongenetic laboratory tests aimed at detecting suggestive clinical features of disease, and by cascade familial screening, DNA methylation testing for diseases with characterized episignatures, and functional studies in cellular or animal models.9 A clinical geneticist can help navigate these testing modalities.Another challenge of genetic testing interpretation is the possibility of a single gene being associated with multiple syndromes. This occurs when distinct variants in one gene give rise to different diseases due to variable disease mechanisms such as gain of function (increased activity of the gene product causing pathology) vs loss of function (reduced production of a gene product causing disease). For example, loss-of-function variants in MYCN cause Feingold syndrome (OMIM #164280), characterized by microcephaly, esophageal and duodenal atresia, and mild to moderate neurocognitive phenotypes, whereas gain-of-function MYCN variants cause megalencephaly-polydactyly syndrome (OMIM #620748), characterized by megalencephaly, postaxial polydactyly, intellectual disability, and increased neuroblastoma risk. The parent of origin of the disease-causing variant can also influence phenotype, as observed in Prader Willi syndrome (OMIM #176270) vs Angelman syndrome (OMIM #105830). Phenotypic variability between individuals with the same variant can also complicate variant interpretation, especially if the proposed causal variant is inherited from a seemingly unaffected parent. Clinical genetic testing reports often do not elaborate on these mechanistic nuances, leading to confusion regarding the correct diagnosis and the appropriate evaluation.Some individuals have dual molecular diagnoses and consequent blended phenotypes, which also complicates variant interpretation. Furthermore, a genetic testing report may list as a dual diagnosis conditions that reflect a single pathological event. This is particularly common in translocations, for which a laboratory may report a chromosomal deletion and duplication in a single individual as isolated diagnoses. Recognizing that complex aneuploidy could be the result of an unbalanced chromosomal translocation is an important consideration for recurrence risk counseling, given the possibility of a parent being a balanced translocation carrier.Clinical testing can be nondiagnostic, sometimes because the syndrome has not yet been described, but also because of current technological limitations. Broad-based genetic testing modalities such as exome and genome sequencing can miss structural, mosaic, intronic, and noncoding variants, and these factors often require the input of a clinical geneticist to resolve. Here we present a case series in which first-pass genetic testing was nondiagnostic, but careful clinical evaluation suggested a diagnosis, which was later confirmed on molecular testing after communication with the genetic testing laboratory (Table 1). Case 1 is a 14-year-old boy with a suspected diagnosis of the TSC2/PKD1 contiguous gene deletion syndrome with previous nondiagnostic testing. He ultimately received a molecular diagnosis through manual review of the clinically suspected genomic region, which identified a mosaic deletion. Case 2 is a 21-month-old girl with clinically suspected branchio-oculo-facial syndrome also missed on targeted clinical testing due to low-level mosaicism, which was ultimately resolved on research-based genomic testing with manual review of the causal TFAP2A gene. Case 3 is a 15-year-old girl with clinically suspected Hardikar syndrome, similarly missed on genome sequencing due to low-level mosaicism, ultimately diagnosed through manual review of the relevant MED12 locus. Cases 4 and 5 are individuals who presented with congenital ichthyosis (case 4) and hepatosplenomegaly with developmental regression (case 5) with high clinical suspicion for harlequin ichthyosis and Niemann Pick Disease Type C, respectively. First-pass clinical testing was nondiagnostic due to atypical variant types including splice and intronic variants. Both probands were ultimately molecularly diagnosed upon manual review of the relevant clinically suspected genomic loci (Appendix 1, Appendix Figures 1–5, Table 1). Molecular diagnosis guided comorbidity screening, treatment initiation, and recurrence risk counseling. Successful diagnosis through clinical insights after nondiagnostic broad-based genetic testing has been reported previously, highlighting the continued importance of clinical genetics evaluation in diagnosis.10–12Important lessons come from the etiologies of the missed diagnoses. Cases 1 to 3 were not detected by clinical testing, as the causal variants were present at very low-level mosaicism in the sampled tissue, meaning that most DNA sequencing reads did not include the variant allele. When a variant is present at low levels, bioinformatic pipelines often discard the variant as noise, and manual review is required for detection, which is not feasible at the genome level. The causal variants for cases 4 and 5 were not initially detected, as they are noncoding, localizing to intronic regions and splice sites. These variant types are often missed by genomic testing, which focuses mostly on protein-coding regions of the genome. Furthermore, the significance of deep intronic variants remains difficult to establish, as their effect on RNA splicing and protein production is poorly understood. Taken together, these cases illustrate that clinical acumen, the ability to phenotype a patient, recognize the limitations of available molecular testing, and communicate phenotype and suspected molecular diagnosis to a genomics laboratory, continues to drive accurate genetic diagnoses. The ability to clinically diagnose patients is also important, as genetic testing can take days to weeks until results, and pretest diagnoses based on physical exam findings and nongenetic laboratory and imaging results can improve care during this turnaround time.Beyond genetic test selection and interpretation, a geneticist can help guide clinical care based on genetic testing results. A molecular diagnosis can inform comorbidity screening and anticipatory guidance in organ systems outside the ordering physician’s realm of expertise. For example, a nephrologist may send a gene panel to identify the cause of seemingly isolated kidney dysfunction and learn that the true diagnosis is nephronophthisis, which can have significant liver, ophthalmological, and audiological disease that requires evaluation. Medical geneticists can help identify other specialists that need to be involved and serve as a medical home for affected families. Furthermore, a molecular diagnosis often prompts cascade familial screening to facilitate early identification of presymptomatic, at-risk individuals so appropriate screening or prophylactic measures can be taken or to identify carriers to inform reproductive decision making.With the explosion in precision-based therapies, clinical genetics input is also essential for selecting appropriate interventions. A prime example is Noonan syndrome, caused by both gain- and loss-of-function variants in genes encoding components of the RAS/MAPK signaling pathway. MEK and mTOR inhibitors have demonstrated efficacy in treating the cardiac and lymphatic complications of Noonan syndrome; however, they are contraindicated in individuals whose causal variants are loss of function.13Importantly, the genetics work force remains limited, with most geneticists clustered in large urban academic centers with limited access in rural areas.14 Even in areas with large genetics centers, patient referrals have increased whereas the workforce size has not, leading to longer wait times and delayed care.15 To address this unmet need, great strides are being made to increase access to genetic testing through integration of genetic counselors into subspecialty clinics, increasing genetic counselor clinical autonomy, and educational efforts to support nongenetics physicians.16–19 Approaches to increase the efficiency of the genetics evaluation to increase clinic throughput are also being investigated, including artificial intelligence and algorithm-based phenotyping approaches.20–23In conclusion, although genomic testing is increasingly used to facilitate appropriate anticipatory guidance, comorbidity screening, and recurrence risk counseling, the skills of careful history-taking, pedigree analysis, and dysmorphology are not yet to be forgotten. In many cases, careful clinical evaluation and partnership between the medical and laboratory genetics teams are critical to establish the correct molecular diagnosis. The cases presented here highlight the importance of physical exam skills, familiarity with clinical criteria for genetic syndromes, and knowledge of disease mechanisms and phenotypes to facilitate appropriate comorbidity screening and recurrence risk counseling. These cases also demonstrate the importance of intronic, splice, and mosaic variants in human disease, highlighting the importance of bioinformatic pipelines that can detect these types of variants. We further emphasize the central role of a geneticist in facilitating reanalysis of existing genomic data in nondiagnostic cases and facilitating open communication between the clinical and laboratory teams to inform careful reexamination of relevant loci.The authors would like to thank the families for their participation in this study.