Uncovering Rare Pediatric Diseases Through Atypical Gene Associations

A conversation with Carol Saunders, PhD, FACMG, FABMGG and Isabelle Thiffault, PhD, FABMGG

Carol Saunders, Ph.D., FACMG, FABMGG, is the division director of the Clinical Genetics and Genomics Laboratory (CGGL) at Children’s Mercy Kansas City and clinical director of the Genomic Medical Center at Children’s Mercy Research Institute (CMRI), where she is currently focused on clinical testing and research specifically on discovering rare disease gene associations. Isabelle Thiffault, Ph.D., FACMG, FABMGG, is assistant director of molecular genetics in the CGGL and director of translational genetics at CMRI. She works closely with Saunders on genome sequencing testing. Together, Saunders’ and Thiffault’s team averages at least 10 to 15 publications a year focused on gene associations, with hopes of finding similar gene associations across global publications.

How long have you worked in the field of genomics and what brought you to the genome center at Children’s Mercy?

Saunders: I came to Children’s Mercy in 2002 right out of fellowship training in clinical molecular genetics, to direct the Molecular Genetics Laboratory. As was typical at the time, we specialized in single gene testing, meaning one or a few genes at a time. In 2011, Children’s Mercy made the decision to invest in genomic medicine and brought in three researchers to launch a big multi-gene panel using next-generation sequencing (NGS.) These researchers had brilliant technical knowledge but needed a clinical laboratory director to guide the design of both the laboratory space and testing process, so they sort of “inherited” me. Together we built the first pediatric genome center — what is now the Genomic Medicine Center at Children’s Mercy Research Institute. That multi-gene panel was offered as a broad symptom-driven test for rare pediatric diseases and was followed by clinical genome and exome sequencing. Since we were such early adopters of this technology, we have a vast amount of data, which has translated into a lot of interesting observations about phenotypic expansion, genotype-phenotype correlations, and new gene-disease associations. This has fueled my research.

Thiffault: I am from Montreal, Canada. I studied and trained mostly at McGill University. I have a master’s degree in cancer genetics, a Ph.D. in molecular genetics, and a post-doc in translational medicine. After postdoc, I decided to apply to the Molecular Genetics Fellowship Program or CCMG (Canadian College of Medical Genetics) program, which is the equivalent of the ABMGG (American College of Medical Genetics Genomics). Both programs are comparable and recognized — CCMG/ABMGG trainees can take Canadian and/or American board exams without additional training.

During my fellowship, I read an academic paper by Saunders and the Children’s Mercy team and it grabbed my attention. I emailed Dr. Carol Saunders, who at the time was leading this research, to see if I could do an “elective CCMG rotation” at Children’s Mercy to learn about “50-hour differential diagnosis by whole-genome sequencing.” She kindly replied that they were too busy, but they needed to hire – and I might be a good fit. I started the position in March 2014; the first months were “credited” as a an elective CCMG rotation and I took the ABMGG board exam in 2015. It was a win-win!

Discuss the type of gene sequencing you are currently conducting. What are the benefits of this type of sequencing? Why is it important to identify atypical gene associations, particularly in the grand scheme of rare disease research?

Saunders: We use both exome and genome sequencing to help diagnose rare pediatric disease. This broad testing has resulted in a critical change in how providers go about their clinical workup. In the past, they would order sequencing on a single gene or a small list of genes based on how the patient presented, since that was what was feasible at the time. As a result, there is a huge ascertainment bias in the literature as far as how patients with a given genetic disorder should look. Today it’s very typical for a provider to do exome sequencing right off the bat, which results in the diagnosis of patients based on their genotype.

So, patients who don’t necessarily have the typical features expected for a certain genetic disease are broadening the phenotypic spectrum – we have observed many patients who present differently than expected for a given disorder, even very well-characterized diseases. It’s important to report such findings so these features become recognized and accepted as part of the phenotypic spectrum of the disease. Otherwise, clinicians are left wondering whether the diagnosis fully explains their patient. New gene-disease associations are being published at an astonishing rate thanks to large-scale sequencing. Until a gene has been associated with a disease, patients with pathogenic variants in that gene will go undiagnosed.

There are certain types of variants that get our attention because of a predicted functional effect, or the fact that they are “de novo,” i.e., absent in the parents and in control data sets, which is a rare occurrence in the genome – typically, a child will have only one or two de novo variants in their exome. New gene-disease associations need more than one patient with consistent features and inheritance pattern to be considered causative vs. occurring by chance. To find other patients, we use matchmaking websites, which allow us to find clinicians and researchers from all over the world with similar patients. This effort has resulted in hundreds of collaborations to describe new gene-disease associations to the medical community.

Can you share any clinical exome sequencing examples where you discovered an atypical gene association? Once you discover an atypical gene associated with a rare disease, why is it important to find another patient (or multiple) with the same gene variant and how will these discoveries be used to improve clinical trials for clinicians in the future?

Saunders: We find patients with atypical phenotypes through clinical testing almost daily. When such features haven’t been reported in association with the gene in question, it is sometimes questionable whether we have the correct diagnosis — does the variant fully explain the patient’s features, or is more testing needed? Such testing is expensive for the family, so we want to make sure it’s necessary. An example is a patient we tested for developmental delay and growth failure, whom we found to have a pathogenic variant in a gene called SF3B4, associated with Nager syndrome, which is defined by specific facial features and limb anomalies. Since our patient didn’t have either of these findings, it was difficult to call it Nager syndrome. In addition, although growth failure and developmental delay had been reported in Nager syndrome, these features were hardly discussed in the literature. Since this was the reason for referral, this diagnosis was not felt initially to be a good fit for this patient and a reanalysis of the exome data was ordered a year later, yielding the same interpretation but perhaps more accurately called SF3B4-related disease rather than Nager syndrome. We published this case in the Journal of Medical Genetics so other providers will realize pathogenic variants in this gene may result in a phenotype inconsistent with what they might expect for Nager syndrome.

Thiffault: Sometimes, as more patients are identified with specific variants in a gene, genotype-phenotype correlations become possible. This can be helpful for prognostication and family planning. Such is the case with a recurrent pathogenic variant we identified in a gene called GNB1, which is associated with severe neurodevelopmental disease and only rarely cleft palate. At first, we wondered whether the clefting was unrelated to this variant, but in reading further we observed that while overall clefting was rare for this gene, it was relatively common in patients with the specific variant observed in our patient. Such findings from our research, “Genotype-phenotype correlation in GNB1-related neurodevelopmental disorder: association of p.Leu95Pro with cleft palate,” published in the American Journal of Medical Genetics, are useful for clinicians to understand the phenotypic spectrum for genetic disease. When there is insufficient genetic evidence for a gene-disease association, further cases are needed before a variant in that gene may be called pathogenic.

For many neurodevelopmental diseases, the phenotype is nonspecific and variable; others may have more striking features that make the similarities between them more obvious. Such was the case with a patient in whom we identified a variant in GNB2: At the time there was only one patient reported; however, our patient’s facial features bore a striking resemblance and the variant itself affected the very same amino acid in GNB2. Around the same time, we published this observation, and another group published a series of 12 additional patients with GNB2 variants, which is now fully recognized as a new neurodevelopmental syndrome.

Children’s Mercy Research Institute is building a repository of pediatric rare disease genomes with the goal of creating a database of 100,000 genomes through their Genomic Answers For Kids Program (GA4K). How does your team work in collaboration with GA4k to identify new rare genetic variants in children?

Thiffault: GA4K is one of the many research programs at work within CMRI. The partnership between the Clinical Sequencing Laboratory and GA4K is unique in the sharing of technology and patients. The goal is for new technology in the research lab to be translated into state-of-the-art clinical testing, such as long read sequencing. In turn, there is a constant stream of patients tested clinically who have uncertain or negative result; we recommend they be consented for GA4K to be further studied. GA4K is unique because it has a large patient pool to pull genomic associations from both national and international sources through data sharing.

The diagnostic journey for a family with rare disease is often long due to lack of access to genetic testing or inconclusive results at the time of testing. By sharing data and using specialized technologies, we hope to accelerate the diagnosis and understanding of rare pediatric diseases. It is always surprising to discover that we have more than one patient from unrelated families within GA4K with the same rare disease. Such findings are an opportunity to delineate the clinical characteristics and to examine the effect of variants on the gene function and disease mechanism. Moreover, it also allows families to connect and share what they are going through. Parents of a child with a rare disease diagnosis often feel alone in a sea of unknowns in terms of prognosis, treatments, and available medical support.

To summarize the greater impact of being able to connect directly with clinicians at Children’s Mercy treating rare disease patients, we are fortunate to have such direct access to the patients’ medical records as well as our clinical colleagues; both of which facilitate analysis by providing information to help us find answers. For example, clinical correlation with pictures of the patient is extremely helpful in determining whether a gene in question is a good candidate. We have had cases where a variant seems potentially interesting, but patients with the corresponding disease usually have a feature that wasn’t mentioned in the medical record, either because the family failed to reveal this or the physician didn’t find it noteworthy.

Clarifying conversations are immensely helpful. For patients, we often personalize the genetic reports with the hope to share information and recommendations in a digestible manner, to help the clinical team and patients’ families to understand the diagnosis or candidates to pursue. After reporting, we discuss the genomic findings with the clinicians to determine the next appropriate steps — whether it’s pursuing publication, functional studies, holding data for reanalysis, enrolling more family members, etc. In many cases, a gene has only a handful of patients reported with the associated disease, or the patient’s clinical phenotype is different than previously reported, and/or the molecular mechanism still needs to be explored. We also work with patient advocacy groups, disease foundations, and associations interested in ultra-rare disease treatments. For the clinical team and the families, a genetic diagnosis is not the end of their journey — it represents the first step, often raising more questions and highlighting the knowledge gap in effective disease management strategies or treatments.

How has the advancement of gene-sequencing technology over the past 10 to 20 years impacted your research and ability to check DNA and identify disease-causing mutations?

Thiffault: Clinical exome or genome sequencing is done with NGS. At Children’s Mercy, patients with genetic disease, especially those with negative or inconclusive testing, are offered enrollment in the GA4K study. Their exome may be reanalyzed; if this remains negative, we use a new technique called long-read sequencing which offers advantages over short-read NGS, including detection of structural variants, copy number variants, repeat expansions, and mapping certainty across problematic regions. In addition, it allows variant phasing, de novo assembly, transcript isoform identification, and methylation (epigenomic modifications). The availability of RNA sequence and epigenetic profiles is useful for interpreting structural variants detected with long-read sequences. Children’s Mercy Research Institute has been fortunate to recruit researchers with diverse pediatric research interests to complement GA4K, with the ultimate goal of improving diagnostic yields.