As next-generation sequencing is increasingly being adopted in the clinic to help diagnose rare disorders in children, two early adopters — the Medical College of Wisconsin and Baylor College of Medicine — recently provided updates on their experiences with the approach, highlighting the advances that each institution has made as well as the remaining challenges.
At last month's Advances in Genome Biology and Technology conference in Marco Island, Fla., Christine Eng, director of Baylor's DNA Diagnostic Laboratory, presented on the first 450 cases to go through Baylor's clinical exome sequencing pipeline, while MCW's Liz Worthey highlighted a few challenging cases that have gone through MCW's whole-genome sequencing pipeline. Additionally, Worthey said that the center is now looking to implement a whole-genome sequencing pipeline within its Neonatal Intensive Care Unit on the Illumina HiSeq 2500 for rapid diagnosis of babies born with rare disorders.
MCW's Pipeline
MCW's experience with clinical sequencing began in 2009, when it used exome sequencing to provide a diagnosis for a 6-year-old boy with a debilitating case of inflammatory bowel disease, which led to a successful bone marrow transplant.
Afterwards, MCW, in conjunction with the Children's Hospital of Wisconsin, instituted a clinical sequencing pipeline through which physicians can nominate for sequencing pediatric cases that have eluded a diagnosis by other means. A board hears those cases and decides which ones would be most likely to benefit from whole-genome sequencing.
The center has a CLIA-certified, CAP-accredited laboratory, which contains an Illumina HiSeq 2000 that will be upgraded to the 2500, and it also makes use of Illumina's CLIA lab for sequencing when it needs increased capacity.
During her presentation, Worthey said that the team is evaluating how it might incorporate the clinical whole-genome sequencing pipeline into the neonatal intensive care unit. The NICU "might be the ideal place for whole-genome sequencing," she said, because in those cases, there is no time to do a complete family history, the babies might not be displaying symptoms that would lead to a diagnosis, and the babies are usually in such critical condition that there is no time to go through the diagnostic odyssey that many children with rare genetic disorders go through.
In the NICU, "treatment often takes place before diagnosis, within the first 48 hours," she said. So for whole-genome sequencing to truly be effective, it would need to be done within that time frame to avoid potentially harmful or ineffective treatments.
In order to implement the sequencing protocol in the NICU, Worthey said one main goal for the center is to reduce the turnaround time.
As of the end of 2012, the entire protocol took around 334.5 hours, or about two weeks, said Worthey during her presentation. By the end of 2013, the goal is to reduce that to 39 hours.
Reducing turnaround time will involve converting the HiSeq 2000 to the 2500, which will reduce sequencing time from around 10 days to around 27 hours. Additionally, the team is working on making improvements to its secondary analysis, interpretation, and reporting, and making improvements to its LIMS system.
The need to reduce turnaround time is important not just for the NICU, but for other cases as well, she said. For instance, insurance companies sometimes deny reimbursement for the test because the wait time is too long, and in two cases from the initial pilot program, the team uncovered a diagnosis that would have led to a change in management, but the answer came too late.
For such urgent cases, Worthey said the team has started to utilize Illumina's STAT-seq service — a protocol originally published by Stephen Kingsmore of Children's Mercy Hospital last year in Science Translational Medicine, and which Illumina offers as a service on its HiSeq 2500 through its CLIA lab (CSN 10/3/2012).
Aside from reducing the turnaround time, other pressing concerns are analyzing regions of the genome that are currently missed by sequencing technology, and the need for more comprehensive and accurate gene/disease databases, Worthey said.
For instance, more than 90 diseases and 52 genes are poorly covered with Illumina's short-read technology, even at 40x to 50x coverage, she said. She described one case that phenotypically appeared to be familial neutropenia. The affected female had a brother with the same phenotype who died at 1 year and also a cousin who died shortly after birth. Standard testing yielded no diagnosis, and trio exome sequencing yielded no clinically reportable mutations. But, she said, there were gaps in five genes known to be causative or related to familial neutropenia, as well as a non-protein-coding gene that is associated with the disorder.
Improved databases are also necessary, she said, because a lack of clinical evidence has hindered the return of putative diagnoses in certain cases. For example, Worthey described one case of an infant with a large spinal hemorrhage. Whole-genome sequencing uncovered a candidate mutation that was confirmed by Sanger sequencing and identified as de novo in the patient. The mutation was in a gene that had previously been linked to hereditary hemorrhagic telangiectasia, an autosomal dominant disorder that causes nosebleeds, digestive tract bleeding, and other symptoms that matched the patient's phenotype.
Further literature searching found that the particular variant had been classified as a benign polymorphism, but the classification "came from a conference proceeding without a published paper," Worthey said. "Based on that, all subsequent papers labeled the variant as a polymorphism," despite the fact that in some of those papers, the frequency of the variant in a sample hereditary hemorrhagic telangiectasia population was found to be higher than in the normal population, she said.
No other candidate variant in the patient was identified, but "because of the lack of evidence to confirm it not being labeled a polymorphism, we couldn't report it," she said.
While the team plans to reanalyze the case every six months, Worthey said this is an example of why correctly annotated databases are so important. "The biggest limitation is the availability of correct genotype/phenotype associations," she said.
Baylor's First 450 Exomes
Baylor's Eng also agreed that having correct genotype/phenotype associations is critical for making a diagnosis. Baylor implemented a clinical exome test in 2011, and has since sequenced over 1,000 patients, and of the first 450 cases, around 25 percent obtained a diagnosis.
Of those without a diagnosis, Eng said that there are many cases "where we have deleterious changes in genes with no association with disease currently." That information is kept on file so as new disease-gene associations are made, those can be reported. In other cases, "we find variants that might be responsible for some of the disorder, but the overlap is not as complete as we'd like," she said. Those are reported as variants of unknown significance.
Some of the more difficult cases have involved patients with more than one disease phenotype. For example, she said that the team has identified four patients for which two pathogenic alleles were identified. The patients "have overlapping clinical features," but they "can't be combined into one diagnosis."
In one case, a 9.5-year-old male had a history of weakness, apnea, and respiratory secretions and had been diagnosed with cardiomyopathy when he was eight. He also had a sister with similar symptoms who passed away at 20 months, and the rest of his siblings were normal.
Before having exome sequencing done, the family had already spent more than $25,000 on genetic tests, muscle biopsies, and other physical examinations and workups.
Exome sequencing uncovered a compound heterozygous mutation in a gene associated with congenital myasthenic syndrome, which fit the patient's clinical presentation. Additionally, one of the variants was confirmed in the mother and the other in the father. A treatment is available for that disorder, and has so far had "some efficacy," Eng said. However, she said, there is also a possible second diagnosis — a mutation in the gene ABC9, which is associated with cardiomyopathy dilated type 10, for which the mother was heterozygous.
As the center gains more experience with clinical exomes, Eng said that it will likely expand the types of variants it reports. Aside from variants that are determined to be causative for the patient's disease, the center also reports out variants that impact pharmacogenomics, secondary findings that are medically actionable, variants associated with preventable diseases like BRCA1/2 variants, and carrier status for a number of diseases.
But, she said, "we report only genes that are recommended by medical societies for population testing," such as the genes related to sickle cell disease, or disease genes that are specific to communities such as the Ashkenazi population. However, "we will probably expand as physicians become more comfortable," she said.
Additionally, she expects that as more exomes are sequenced, the diagnostic success rate will grow. Currently, one of the main challenges is reporting the non-phenotype findings, she said. But as more evidence builds, Eng said she expects that the phenotypic spectrum for many disorders will be expanded.