TAMPA, Fla. — Talks at this year's annual Clinical Genetics Meeting of the American College of Medical Genetics, held here last week, underscored the expanding role of cytogenetic microarrays in the clinical setting, but also touched upon a number of issues that clinicians face as they look to bring the arrays into their practice.
There was evidence at the conference that SNP and CGH arrays have made inroads into the cytogenetics market over the past few years. For example, nearly 25 percent of the 329 posters at the conference mentioned microarrays, compared with around 10 percent several years ago, several attendees noted.
Within cytogenetics, nearly two-thirds of the cytogenetics posters — 31 out of 50 — involved arrays, and all but one of the eight talks in the cytogenetics platform session discussed microarrays, prompting one speaker to apologize that her talk dealt only with karyotyping.
Indeed, there were also a few signs that microarrays could eventually replace karyotyping in certain applications in the future. Christine Eng, medical director of the DNA Diagnostic Laboratory at Baylor College of Medicine, mentioned that her group is currently involved in a study comparing arrays to karyotyping in prenatal screening.
And David Ledbetter, director of the division of medical genetics at Emory University, spoke about a workshop hosted last year by the International Standard Cytogenetic Array Consortium where participants reached a "consensus" recommendation that cytogenetic arrays be used as the "first-line diagnostic" for unexplained mental retardation or developmental delay in children.
He said that David Miller, assistant director of the Genetics Diagnostic Laboratory at Children's Hospital Boston, is currently preparing a manuscript outlining the recommendation.
Ledbetter told BioArray News after the session that despite the promise of cytogenetic arrays, it will likely take some time for most labs to transition from karyotyping because array platforms remain relatively new. He said that even some labs that run both platforms still tend to run karyotyping first, and then turn to arrays only after the karyotype is normal or inconclusive.
However, he added that he believes this will change because arrays offer a number of advantages over karyotyping, which is "subjective and of variable quality."
As an example, Ledbetter said that conventional wisdom has held that karyotyping should be able to detect all chromosomal variations of five megabases or larger, but the advent of arrays has shown that karyotyping often misses variations as large as 10 megabases.
"The accuracy of karyotyping is a lot lower than people thought," he said.
Another factor that might drive adoption of arrays in the clinic is insurance billing, he said. Most payors will only allow clinics to bill for one type of assay, leaving clinicians to decide whether to request karyotyping or arrays. Since the yield and resolution of arrays are so much higher, more clinicians will likely start requesting arrays, he said.
Some vendors at the conference said they expect cytogenetic arrays to be a key growth area for their businesses over the next few years. For example, Perry John Aiello, director of business development at bioinformatics firm InfoQuant, told BioArray News that the company was founded around three years ago with the intention to focus only on data analysis for CGH and SNP arrays.
While the research market for these arrays has probably plateaued, he said, the clinical market is still at the beginning of its adoption curve and the company expects strong growth in the years ahead.
"I would not be surprised if we see a 100 percent growth rate over the next few years," Aiello said.
As further evidence that arrays are becoming established as a standard clinical cytogenetics tool, the latest edition of the International System for Cytogenetic Nomenclature, ISCN 2009, includes a new chapter devoted to nomenclature for reporting data from arrays alongside karyotyping and FISH results. In a talk outlining the new array nomenclature, Kathy Rao, a cytogeneticist at the University of North Carolina Chapel Hill who serves on the ISCN committee, explained that the array nomenclature is structured the same way as that of the other cytogentics platforms, and can be combined with those results in a single report.
Work in Progress
Growing adoption of cytogenetic arrays in the clinical setting is not without its challenges, however. In particular, many ACMG speakers cited the interpretation of results as their biggest challenge in this area. Not only is this a new field for many clinicians, with a steep learning curve, but there is still a lot of information missing from public databases that would help improve the interpretation of array findings, speakers said.
[ pagebreak ]
In his talk about the ISCA, Emory's Ledbetter noted that the "biggest problem" with cytogenetic array results is "how to determine what is pathogenic and what is benign," largely because public databases do not include sufficient data to support clinical decision making.
While there are a number of databases that include information on copy number variants, such as the Database of Structural Variants, hosted by the Centre for Applied Genomics at Toronto's Hospital for Sick Children, and the Decipher database, hosted by the Wellcome Trust Sanger Institute, Ledbetter noted that there is a dearth of phenotypic data linked to particular variations that would help clinicians determine whether a particular variation is pathogenic or not.
For example, Decipher, which was developed as a repository for structural variation data linked to phenotypes, only has a few hundred entries. "We'd like to see that in the thousands," he said.
If clinical labs were to share their own phenotype and variation data, it would go a long way toward improving interpretation, he said. "We know there are 50,000 to 100,000 clinical array cases in labs that are not being shared right now."
Ledbetter noted that ISCA is currently working with the National Center for Biotechnology Information to include cytogenetic array studies as part of the Database of Genotype and Phenotype, which NCBI launched several years ago to house data from genome-wide association studies.
Ledbetter said that several ISCA members, including Emory, are currently working with NCBI on a "trials submission" process for depositing into dbGaP de-identified patient data from cytogenetic array studies. Once NCBI is certain that the mechanism works smoothly, it will begin accepting submissions from other groups, Ledbetter said.
Darrel Waggoner, a clinical geneticist at the University of Chicago, outlined some of the specific difficulties that clinicians face when interpreting cytogenetic array data. As an example, he used the case of a large deletion that includes 62 genes — far more information than most labs would be willing to annotate on their own, especially if there are no details of the patient's clinical phenotype on the test requisition.
In this case, to help them interpret the results, it would be up to the clinician to consult public databases such as the University of California-Santa Cruz Genome Browser, Ensembl, and the Database of Structural Variants, Waggoner said.
"I think as clinicians, many of us are not comfortable using these databases," he said, noting that it would be a good idea for training programs for genetic counselors and residents to ensure that students are familiar with these resources.
Waggoner demonstrated several examples of how a clinician might use public genomic databases to help interpret array results, and explained that many of these resources have a great deal of clinically useful information, though it may not always be easy to find.
"There is a perception that the UCSC Browser is only for researchers, but just by clicking on the right buttons, it's also a great tool for clinicians," he said.
'Collateral Damage'
Other speakers touched upon the ethical questions that some clinicians face when they use information from cytogenetic arrays. Kelly Ormond, program director in the department of genetics at Stanford University School of Medicine, noted that much of the available research information on copy number variation has not been validated in multiple studies, and that it is widely known that there is a great deal of variation in the results from different array platforms. As a result, an array may provide information about deletions or duplications that may not necessarily account for the clinical phenotype.
"This raises the issue of how much information to disclose" because some or all of that information may not be medically actionable, Ormond said. "If we don't know what the CNVs do, do we need to tell the patient or the parent?"
Mildred Cho, an associate professor at the Stanford Center for Biomedical Ethics, echoed Ormond's point, noting that CNVs are only now "crossing the blurry boundary between research and clinical use," which means that the ethical obligations for doctors and genetic counselors are "just as blurred."
Beth Tarini, a pediatrician with the University of Michigan, put these questions in the context of the general practitioner's office with an anecdote that outlined the impact that incomplete genetic information could have on a family. She described a hypothetical case of a 4-year-old autistic boy found to have a 1.8-megabase CNV. The boy's baby sister had the same CNV but has not exhibited any symptoms of autism.
Tarini said that even though the sister was developing normally, the mother was put in the position of "waiting for the other shoe to drop." This, in turn, led the mother to interpret minor characteristics that might otherwise have been overlooked — such as not looking the mother in the eye all the time, or sometimes not wanting to be cuddled — as the onset of autism.
Tarini stressed that there is a difference between "information" and "knowledge," and that there is a risk in providing patients with incomplete information that may not be helpful, or that may actually cause harm.
While not advocating against testing for CNVs, Tarini urged that clinicians take care to avoid the "collateral damage" that is possible from the increase in CNV array studies.