Combining genome sequencing and functional assays, the Duke Task Force for Neonatal Genomics aims to assist in the early diagnosis of newborns and young children with challenging genetic diseases.
Formed in 2011, the interdisciplinary group, a collaboration of researchers and clinicians from various disciplines at Duke, aims to "determine how and under what parameters genome and exome sequencing can be implemented as a first-state diagnostic tool to inform and guide clinical management and treatment for individuals with congenital defects," according to its website.
Members of the group have expertise in neonatal medicine, maternal fetal medicine, medical genetics, molecular biology, human disease genetics, statistical genetics, genetic counseling, clinical genetic testing, genetics policy, and human participants research.
"The motivation is to alter the way we practice genetic medicine," said Nicholas Katsanis, director of the Center for Human Disease Modeling at Duke, who heads the effort. So far, he said, genetic testing has mostly been used to confirm a suspected diagnosis, and the goal of the task force is to make it a first-line diagnostic test with high predictive value.
Besides sequencing many individuals, this will require the validation of results by functional assays. "We feel functional testing is absolutely necessary for interpretation," said Katsanis, who is also a professor of cell biology and pediatrics at Duke.
The study is designed to answer a number of questions: Whether trio exome sequencing – and eventually, whole genome sequencing – is valuable as a first-line diagnostic, how the data can be interpreted using computational tools alone, how much functional biological assays add to the interpretation, what can be learned about genetic architecture, and what useful information can be reported to patients.
So far, the group has identified about 110 families who have passed its inclusion criteria, has recruited 38 families, and has completed analyses on about 15. It plans to to publish the results of these 15 analyses in the near future. Funding currently comes mostly from institutional funds but the group has a pending application for a clinical exome sequencing grant from the National Institutes of Health.
The consortium focuses on those phenotypes that can be modeled with high sensitivity and specificity in an in vivo model. It chose zebrafish as its "workhorse model" because of its low cost, transience, and because many developmental processes between zebrafish and humans are highly conserved. In addition, the researchers use an array of functional in vitro assays, which need to be "physiologically relevant," Katsanis said.
In particular, the group looks for newborns or young children with complex anatomical defects that affect at least two organ systems – the nervous system, cardiovascular system, skeletal system, skeletal muscles, or visceral organs — or where several anomalies occur within the same organ.
Patients must have normal chromosomal results and no molecular diagnosis at the time of enrollment, and both parents must be willing to participate in the study. Cases where there are known environmental or maternal factors involved are excluded.
So far, the group has "piggybacked" on already scheduled clinical visits of patients, which may only occur every six or 12 months. However, "if there is potentially clinical benefit, we step on the gas," Katsanis said.
Following informed consent, the child and his or her parents provide a blood or saliva sample for exome sequencing. Sequencing and initial bioinformatic analysis are performed at the Baylor Human Genome Sequencing Center, Duke's partner in the project, while the functional interpretation is done at Duke. The task force is currently working with Baylor to switch the sequencing to Baylor's CLIA laboratory while still allowing the Duke team to use the results for its research study. According to its website, the task force is also collaborating with GeneDx for clinical sequencing.
For the sequence data analysis, the team uses only two filters: one that searches for variants with a minor allele frequency of less than 1 percent, and another that looks for recessive or de novo variants. "We think that the fewer hypotheses we have, the more likely it is that we will reach the truth," Katsanis said.
The filters typically leave them with six to 10 candidate genes for families with Northern European ancestry. Instead of predicting the gene function and effect of mutations computationally and guessing which gene is the causative one, the researchers design zebrafish models for each candidate gene, which costs between $3,000 and $20,000 per gene.
If suppression of the gene in a zebrafish embryo leads to a similar phenotype, they test whether the patient's allele is able to rescue that phenotype. "And if it fails to rescue, then the conclusion has to be that this allele is deleterious to protein function, and that this protein function is relevant [to the phenotype]," Katsanis said.
So far, he and his team have built functional assays for about 250 human disease genes, a number that is growing weekly, so in the future, they will not have to develop so many new assays.
For about a third of the children they have studied so far, they identified genes previously known to be associated with a genetic disease that is consistent with the child's phenotype, a result that is in line with other clinical exome sequencing studies.
However, after adding findings from the functional models to the genomic data, they identified a "very strong candidate" gene in 14 out of the 15 families studied so far. Katsanis cautioned that they cannot be sure that these are the causative genes – for that, one would need another patient with the same mutation and phenotype. But based on control studies where they tried to find candidate genes for a fictional phenotype, the specificity of the approach appears to be high. "Of course the numbers are still low and we will need to do a lot more work to get to a point where we have a high degree of confidence," he said.
In several cases, two genes appear to be contributing to the phenotype, which may or may not be coincidence. "Once we get to 100 or 200 trios, I think these statements will start to become a lot more robust," he said.
So far, the task force has reported results to several families where they found new mutations in genes known to be associated with the child's phenotype, and where they were able to confirm the effect of the allele in a functional assay. According to its website, findings that are "of immediate clinical relevance" are reported to the family's physician in the form of a written report after they have been confirmed in a CLIA lab.
"We are beginning to return things that are a little bit more uncertain, but we are returning them under the guise of a genomic hypothesis that predicts new clinical tests," Katsanis said. The threshold is lower for actionable results, he said, and his team always weighs the potential risk for the patient if the genomic hypothesis is wrong.
"So far, the community has been very positive, and most important of all, the patients' families have been really positive," he said.
In at least three cases, the results have had direct effects on clinical management. In one instance, the results prevented a potentially harmful clinical procedure, which was already scheduled and was postponed in order to conduct further tests.
In another case, the findings predicted that the clinical symptoms of the child would become weaker in the future, helping doctors adjust the child's medication.
In yet another patient, the results provided an entirely different molecular diagnosis than the suspected diagnosis, redirecting the patient "in a very different avenue of both clinical investigation and care," Katsanis said.
The group has also looked into incidental findings – including carrier status for Mendelian diseases and pharmacogenomic markers — and is working on a formula for returning these to patients, who can opt into obtaining such results. "We cannot CLIA confirm all of these things; it's just too expensive," Katsanis said.
As part of the study, they will also re-analyze the exomes of patients every 18 months.
In the meantime, the task force is expanding its team and scaling its operation, though functional assays remain labor-intensive. By the end of this year, it plans to have the capacity to analyze at least 100 trios per year. "We're not a mass production place; we are a boutique exome interpretation unit," Katsanis said.