Time to make rare disease diagnosis accessible to all

Abstract

Studies have demonstrated the utility of genomic analysis for rare disease diagnosis, yet accessibility is still in its infancy; global data sharing will be needed to further advance our knowledge of all causes of rare disease.

Substantial advances have been made in recent decades towards understanding the genetic basis for many human diseases and several key lessons have been learned. First, monogenic forms of disease are individually extremely rare, but collectively common. Second, there is a broad spectrum of phenotypic expression controlled by both genetic and non-genetic factors. Third, application of sequencing technologies is critical for diagnosis and fourth, advancing our understanding of the basis of genetic diseases will require global data sharing. In a recent issue of the New England Journal of Medicine, investigators from the UK’s 100,000 Genomes Project reported on a pilot for implementing rare disease diagnosis in their national healthcare system.¹ This study is an exemplar of all four of these key lessons and also represents a major step forward in implementing what is increasingly viewed as the key path forward for comprehensively diagnosing and treating genetic disease.

Human diseases with a genetic component range from common to rare, both ends of the spectrum gaining increasing focus over the last decade as the genetic underpinnings of these diseases are identified. Historically, common and rare diseases have been dichotomized but there is increasing evidence of a broad continuum between the two, with pathogenic variation causing variable expression of disease, in terms of the severity and number of phenotypic features. Both genetic and non-genetic modifiers of disease have been implicated in this variation but much work remains to discern a more comprehensive understanding of the variable expression and incomplete penetrance of many genetic diseases.

In the UK study, participants had been identified as having a rare disease with a likely genetic cause, but had not received a genetic diagnosis. Although genomic sequencing was used as the diagnostic tool of choice, the pilot began with the use of virtual panels, whereby disease-focused sets of genes are first used to constrain the analysis of causal variation. However, negative cases were then subjected to expanded analysis, encompassing the full set of genes and non-coding regions in the human genome, revealing that 26% of solved cases have causal variation residing outside of pre-defined, disease-associated panels. This underscores the inadequacy of the panel-based approach to genetic diagnosis and the necessity of ‘genotype first’ approaches to overcome the limitations of phenotype-driven analysis. Likely our field will settle on a combined approach, as used in most laboratories today, whereby suspicious variation (e.g. known pathogenic variants in the ClinVar public database, de novo variation, predicted loss of function variation) across the entire genome is scrutinized along with more comprehensive analysis of variation in genes previously linked to the patient’s symptoms (Austin-Tse et al 2021, in press). Together, these approaches bring a powerful solution to rare disease diagnosis.

Unfortunately, despite major advances, we are far from being able to identify all genes for which pathogenic variation is causal for one or more forms of rare disease. The UK study yielded an overall diagnosis rate of 25% across all probands, a percentage similar to other research studies.^2,3,4 While this yield is higher than most diagnostic tools applied in the practice of medicine, it points to a continued need to identify the causes of the remaining forms of genetic disease. Most causal genetic variants occur at extremely low frequency, so may not be identified as causal in a single individual; therefore, establishing causality requires a globally-shared evidence-base. The first step involves sharing classified variants through the ClinVar database⁵ which has now attracted submissions from over 2000 clinical laboratories and researchers from over 80 countries and has become an essential tool for rare disease diagnosis. ClinVar is a knowledgebase of classified variants with aggregated evidence for their association to disease. But missing from most entries is a direct link to the phenotypic presentation of each individual with the variant, as well as other genetic and functional data that may have been observed or generated to interpret the variant. To further advance variant interpretation, as well as identify novel gene-disease associations, we must share the individual-level data such as demonstrated in the DECIPHER database⁶ as well as the results of functional analyses.⁷ Platforms such as the Matchmaker Exchange have enabled researchers and clinicians to aggregate individual cases with candidate genes in common,⁸ but we still lack robust mechanisms for sharing cases for which no candidate genes or variants are identified. The standards and approaches being developed by the Global Alliance for Genomics and Health will be critical for launching innovative approaches to share this data, yet still respecting the privacy and security of each individual.⁹

Advances in sequencing technology have allowed an individual’s entire genomic sequence to be deciphered at a cost that is feasible within routine medical practice, yet still, most individuals with rare disease have not had access to whole genome sequencing. There are several barriers that have been in place and are only now finally being broken down. For example, the added yield of a genome sequence over an exome sequence has been marginal due to technical issues – this added yield should come from several contributions including detection of structural and more complex variation within genomes, but there is a lack of robust and accessible computational tools to comprehensively characterize these. To overcome this, we must define and enhance access to the best-in class tools for genomic analysis and make them freely available to all laboratories. Another barrier comes from the fact that interpretation of non-coding variation remains in its infancy. Addressing non-coding variation will require larger datasets and more cost-effective and scalable methodologies to enhance signal out of background noise and determine the functional impact of most non-coding variation. Finally, there is concern over the added burden for patients when they receive results of genomic tests that reveal variants of uncertain significance (VUS). Interestingly, this may well be unfounded when one compares clinically available disease-focused panels, for which standard practice is to report all variants classified as ‘uncertain’, ‘likely pathogenic’ or ‘pathogenic’, compared to genomic approaches where only variants with high suspicion for a causal role are reported.¹⁰ An analysis of the rates of VUS reporting in panels vs exomes and genomes will answer this question.

In summary, hundreds of studies have demonstrated the personal, clinical and economic utility of applying sequencing approaches to rare disease, and countries are beginning to implement these services at a national scale. But more must be done to ensure that access to these services is accessible to all individuals, regardless of their background or country of origin. Furthermore, we must work as a global community to share a common evidence base to ensure that each individual can achieve the most informed understanding of the variation in their genomes, and which variation may be causal or put them at risk for genetic disease.

Figure. The cycle of rare disease testing, discovery and treatment.

Progress in the diagnosis and treatment of rare disease depends on a tight relationship between clinical testing, research and data sharing.