Despite its proven utility for providing molecular diagnoses in patients with genetic disorders, exome sequencing is rarely the first-line diagnostic test ordered by most physicians. There are numerous valid medical and practical reasons for this. In some ways, exome sequencing is a victim of its own success: by enabling rapid discovery of many disease-causing genes, it has facilitated the development of numerous commercial gene panels that can be ordered (usually with lower cost and quick turnaround) for patients with suspected diagnoses.
It seems likely that this evolution makes healthcare systems more efficient, and that’s certainly a good thing. However, a necessary consequence of serial testing is that easy-to-diagnose cases rarely proceed to exome sequencing. In the extreme sense, unbiased exome sequencing becomes the “last resort” of clinical diagnostic testing. That undoubtedly skews the conclusions we might draw about large, longitudinal WES cohorts.
I came across a recent study of clinical exome sequencing in a highly consanguineous population led by researchers in Saudi Arabia that was published in the American Journal of Human Genetics. Although a number of exome cohort studies have been published over the past eight years or so, this one is unique for a couple of reasons. First and most obvious is the Saudi population, which has a high fertility rate and a shockingly high consanguinity rate of >50%. Second, the country’s Ministry of Health began covering clinical whole exome sequencing (WES) as a first-line diagnostic test. This financial commitment is unusual, and enabled WES for thousands of patients by a single clinical laboratory.
Because of this support, any clinician could order WES, and it was also their decision whether additional family members would be sequenced as well. Nearly half of the ordering physicians, interestingly, were not clinical geneticists. About 75% elected to send the proband only for WES. Normally, that’s a disadvantage, as full trios (child and both parents) tend to have a higher diagnostic yield. However, in this population it might not be as detrimental since the expectation is a homozygous causal allele.
Increased Diagnostic Yield and Associated Patterns
The well-documented diagnostic yield (success rate) of clinical exome/genome sequencing for unselected sequential patients with rare disorders is around 30%, with a community-accepted range of perhaps 25-35%. This figure has remained remarkably consistent across several years and numerous different clinical diagnostic laboratories, including our own. Although some groups have achieved and reported higher rates, these typically reflect subsets of phenotypes or some type of case selection. Generally speaking, 30% is the expected diagnostic yield.
In the Saudi study, the authors reached a conclusive diagnosis for 961 out of 2,219 families, for an overall diagnostic yield of 43.3%. That’s a very good diagnostic rate, and supports the notion that employing WES as a frontline test in an at-risk population tends to yield more success.
Among the 961 definitively solved cases:
- 77.2% were autosomal recessive (98.4% were homozygous-recessive)
- 20% were autosomal dominant (25% proven de novo; the rest did not have parental samples available)
- 3.9% were X-linked
The proportion of recessive diagnoses is higher than even I’d have expected given the population. For a point of comparison, the DDD study reported thousands of molecular diagnoses for neurodevelopmental disorders, finding that just 3.6% were due to autosomal recessive inheritance.
WES Result versus Clinical Diagnosis
Another key finding of this study was that an overwhelming majority of individuals with solved diagnoses (78%) did not carry the correct clinical diagnosis when the WES was ordered. That’s a compelling figure. It suggests that, particularly in this study population, an early comprehensive laboratory test provided new information. Perhaps even information that changed management or care, though that’s not investigated here.
Molecular Autopsy by Proxy
Another powerful application of clinical WES was showcased by this study: the ability to perform a molecular autopsy when material from a deceased child is not available. There were 80 couples who had lost at least one child and were referred for prenatal counseling. In 45 of these (56.3%), WES of the parental samples provided a likely molecular cause. All of these were recessive disease genes where both parents were heterozygous for a pathogenic variant. The majority of those (28, or 62%) were loss-of-function variants, which undoubtedly was a factor in the previous losses.
This naturally highlights the potential positive impact of prenatal testing in populations (like this one) where there’s an increased risk of recessive disorders. The authors plunge into an exploration of population-specific and private alleles that inform this topic, but I won’t cover them here.
Confirmed and Candidate Disease Genes
One very appealing aspect of this study is that a single laboratory conducted all of the WES, enabling them to catalogue genomic findings across thousands of families. After all, the 56.7% of cases that did not achieve a definite molecular diagnosis are fertile grounds for gene discovery. The authors thus included in their study a list of 236 proposed candidate genes for genetic disorders, some of which were observed in multiple families in their cohort or had supporting evidence in animal model systems.
Their efforts also “confirmed” the disease association for 60 genes which had been reported to cause disease on the basis of a single family and/or were not otherwise recorded in OMIM. This work thus represents a valuable resource to the research community, since rare disease genes tend to be, you know, rare. The authors also helped delineate and expand the phenotypic spectrum of many disorders, which is always a valuable addition to the public record.
One intriguing class of findings are the 10 cases (8 genes) that appear to have recessive forms of genes previously believed to be autosomal dominant. Some of these presented with a phenotype that matched the AD disorder, while others showed earlier onset or a more severe presentation. For at least one disease gene, the recessive disorder apparent in the family was quite distinct from the known AD disorder.
Rare But Fascinating Diagnoses
A few other known but uncommon genetic testing phenomena were observed in this cohort.
- Parental gonadal mosaicism was noted in at least 3 families, highlighting the importance of looking for this in apparent AR pedigrees.
- A de novo mutation “second hit” in a recessive disease gene was noted in 1 family. This is something a number of analysis pipelines might not detect, so it’s reassuring that it’s somewhat rare.
- 3.1% of cases had more than one genetic diagnosis (called multi-locus diagnoses in this study), either multiple AR disorders (1.57%), multiple AD disorders (0.4%), or a combination (1.35%).
The latter phenomenon is something that we’ve noticed in our own laboratory, most notably in the form of de novo alterations in consanguineous cases where we expect to find a recessive cause of disease. It happens enough that we’ve been discussing the possibility that as-yet-unidentified biallelic loss of some key DNA repair/integrity mechanisms might be contributing to the high incidence of de novos. Food for thought.
In summary, this is a very useful study that emphasizes the power of WES as a first-line diagnostic tool and a second-line discovery vehicle, particularly in this unique population. I hope to see more studies like these in years to come.
Jessica Chong says
Re: “The proportion of recessive diagnoses is higher than even I’d have expected given the population. For a point of comparison, the DDD study reported thousands of molecular diagnoses for neurodevelopmental disorders, finding that just 3.6% were due to autosomal recessive inheritance.”
This actually isn’t surprising. The DDD study was specific to NDDs, which are highly enriched for conditions caused primarily by de novo variants. Our analysis (in AJHG “Mendelian Gene Discovery: Fast and Furious with No End in Sight”) shows that ~90% of all Mendelian conditions are not due to de novos and ~55% are AR. Of course this is by count of distinct conditions not by cases due to each condition, but still, recessives are a lot more common than people think.