4 | DISCUSSION
Identifying patients is a critical but limiting step in the disease gene discovery process for rare disorders. Diagnostic laboratories provide meaningful contributions to these efforts by screening large testing cohorts for relevant phenotypes based on genotype matches, contacting clinicians of appropriate cases, and issuing proactive reclassification reports. Over the last 5 years, our laboratory has entered 246 candidate entries into GeneMatcher of which 45.93% are now characterized. There was no significant difference for the likelihood to be characterized based on inheritance type, zygosity, or alteration type. However, entries which met our candidate gene reporting criteria were significantly more likely to be characterized, suggesting that whilede novo and biallelic alterations are easier to identify and represent a large percentage of our GeneMatcher entries (56.91% and 33.74%, respectively), our process of evaluating existing evidence for a gene-disease association is a strong predictor of disease association regardless of mechanism.
By integrating standardized methods for evaluating the evidence available for gene-disease associations into the clinical DES analysis framework, diagnostic laboratories can be valuable partners in screening genes for potential disease associations. All probands with informative trios and negative results after characterized gene review undergo additional candidate gene analysis at our laboratory. Analysts consider the published literature and in-house DES data from previous cases. In some instances, we amass sufficient evidence to characterize a gene-disease association based on internal data. For example, we identified six internal probands with de novo rare ZNF292alterations and similar neurodevelopmental phenotypes more than one year before the manuscript describing the association was published. Our laboratory submits data on genes with substantial evidence for a disease association into GeneMatcher. This creates a more meaningful contribution to data-sharing compared to automated ‘data dumping’ practices where every rare variant in an uncharacterized gene is entered. Broad data deposits without evaluating the potential clinical relevance can create analytical noise and superfluous communications (Seaby, 2021). Follow-up communications to screen matches are laborious, so contributors to GeneMatcher should have standardized methods to appropriately vet candidate genes before submitting.
One advantage of DES is the ability to analyze the expanding catalog of characterized genes over time. Gene characterization has a cascading effect with the potential for multiple patients to receive a reclassified diagnostic finding once a gene-disease relationship is established. Here, we found GeneMatcher entries ultimately resulted in 480 probands receiving a reportable finding. This accounts for a substantial portion (11.96%) of our total DES cohort with reported characterized findings. Importantly, nearly half of the probands with reported variants in these genes received a reclassification report. Laboratories should have processes in place to curate and score new evidence GDV in real-time, and retrospectively evaluate cases for proactive reclassification. This is an ongoing effort and several cases reported here are under review for clinical overlap and reporting [Figure 2]. Without a process to re-evaluate previous cases, nearly half of this cohort would not have received an updated report. Establishing an accurate molecular diagnosis has profound impacts on patients, especially for individuals with rare disorders. A diagnosis can precipitate tailored clinical management, clearer recurrence risk counseling, and in some cases, eligibility for clinical trials.
Further, because of the ability to reclassify previously reported DES cases, diagnostic laboratories can quickly identify additional probands and further power studies to clarify genotype-phenotype correlations and phenotypic spectrum expansions. Often after an initial cohort describes a new gene-disease relationship there is a follow-up publication of a larger cohort to further delineate the newly characterized disorder. For example, we co-authored Peng et al. (2017) which described the association between biallelic FDXR alterations and auditory neuropathy and optic atrophy. The following year, we collaborated with Slone et al. (2018) on a follow-up study to include additional patient reports and functional evidence. We participated in 24 follow-up cohort studies as well as 14 publications describing a single case report or other data in support of the association. Lastly, there were 5 publications we participated in that described a phenotypic expansion for a gene to include major features not previously reported [citations listed in supplement Table 2]. Diagnostic laboratories can serve as a recruitment hub for researchers where more than one proband of interest can be identified through a single source. We can screen the phenotype of individuals with rare variants in a candidate gene of interest, identify the probands with phenotypic overlap, and reach out to multiple clinicians. In some cases, a candidate gene may not have met reporting criteria. In these instances, we can provide reports documenting the presence of an alteration, but not commenting on the clinical relevance until sufficient information is amassed and reported that meets our characterization status. Clinicians find these reports useful to facilitate discussion with families about an alteration which was not originally included on a DES report and the option to enroll in a study further investigating a disease association. Laboratories can further assist the research process by sharing remaining banked DNA (if available and with the appropriate permissions) with researchers and providing cascade testing to other family members for purposes of risk prediction, recurrence risk estimates, and variant segregation within a family.
While diagnostic laboratories can be an efficient tool for finding matches given there is a warehouse of clinical data to screen, the level of clinical data available is dependent on the information submitted at the time of testing. Comprehensive clinic notes that summarize the salient points of a proband’s phenotype should be submitted with all DES testing (Seaby, 2020). A complete clinical description of a proband’s presentation is not only important for variant classification, but also for characterizing new disease genes. In reviewing previously reported cases for collaborators, those with more specific and complete clinical information are more evident as positive matches compared to vague descriptions. Further, robust clinical data can help identify links between a proband’s features and a gene with limited evidence-based on animal models and other functional data.
There is still much work to be done, and many suspected gene-disease associations with insufficient evidence for characterization exist. In total, we have 70 genes that have been reported as a candidate, but remained uncharacterized for more than 2 years, indicating that there was sufficient experimental evidence at the time to associate the gene with disease; however, robust patient data has not yet become available despite the use of GeneMatcher. More than one novel candidate report was issued in only nine of these genes, making these true “n-of-one” cases in need of additional patient reports to corroborate the finding. For example, in 2015, compound heterozygous predicted loss-of-function alterations in the NMNAT2 gene were identified in a fetus with multiple congenital brain anomalies, contractures, and absence of muscle, which matched perfectly with a mouse model published years earlier (Hicks, 2012). Follow-up in-house structural studies and functional studies through a collaborator support the pathogenicity of these alterations (Lukacs, 2019); however, this gene remains uncharacterized as no additional patient reports have been published. Another example of the importance of robust phenotypic data is theCHD3 gene, which was reported as a gene of interest in patients with neurodevelopmental disorders in several large cohort studies (O’Roark, 2012; Iossifov, 2014; DDD, 2016), but did not meet our criteria for characterization until a case series with detailed phenotypic data five years after the first published report (Snijders Blok, 2017). Robust case series like this are needed to garnish enough evidence to meet the strict criteria for gene-disease characterization. However, these manuscripts are more likely to take longer to publish, delaying the necessary evidence for characterization to be available. This delay is most profound for gene-disease associations in extremely rare cases that are less conducive to cohort studies and disorders with clinical heterogeneity which require large cohorts to identify the phenotypic commonalities. Collaborative efforts such as GeneMatcher are needed to power these studies.
By involving diagnostic laboratories in research collaborations there is also a direct clinical impact that results in diagnostic findings going to patients faster. We previously found that clinically relevant reclassification results were issued faster for publications on which we were collaborators compared to genes for which we were not included on the publication (internal data). Involvement of diagnostic laboratories in data-sharing platforms like the Matchmaker Exchange may shorten the time from publication to reclassification report issuance.