Name: Howard Slater
Title: Head, Cytogenetics Laboratory, Murdoch Childrens Research Institute, Parkville, Australia
Background: Since 1990, Slater has served as head of the Cytogenetics Laboratory in the Victorian Clinical Genetic Services unit at the Murdoch Childrens Research Institute in Parkville, Australia.
Slater holds a PhD in genetics from the University of Glasgow and had worked previously at the Duncan Guthrie Institute of Medical Genetics in that city, and the Inverness Cytogenetic Laboratory, also in Scotland.
Slater is an accreditation assessor with the National Association of Testing Authorities in Australia; an honorary fellow of the MCRI and the University of Melbourne; and a fellow of the Royal College of Pathologists.
The market for array-based cytogenetic testing has attracted platform vendors of all sizes in recent years. Companies like Affymetrix, Agilent Technologies, BlueGnome, Illumina, Oxford Gene Technology, Roche NimbleGen, and others offer catalog products for cytogeneticists. Meantime, shops like Signature Genomic Laboratories and CombiMatrix Molecular Diagnostics offer array-based cytogenetic testing services.
Recently, a team of researchers from the Murdoch Childrens Research Institute in Parkville, Australia, investigated the feasibility of replacing "time-consuming, locus-specific testing for specific microdeletion and microduplication syndromes with microarray analysis."
In their study, which appears in the current Journal of Medical Genetics, the authors performed genome-wide copy number analysis on 117 patients using Affymetrix 250K microarrays. This analysis yielded 434 CNVs, including 18 pathogenic CNVs and nine identified as ‘‘potentially pathogenic," leading the authors to conclude that array-based analysis "improved diagnostic success" in this group of patients.
[Bruno, et al. Detection of cryptic pathogenic copy number variations and constitutional loss of heterozygosity using high resolution SNP microarray analysis in 117 patients referred for cytogenetic analysis and impact on clinical practice. 2009 Feb;46(2):123-31.]
Additionally, the authors stated that the arrays provided the potential to make genetic diagnoses that were "not evident in the clinical presentation, with implications for pretest counseling and the consent process."
To learn more about this study and the advance of array-based cytogenetic testing in Australia, BioArray News spoke with co-author Howard Slater, head of the cytogenetics lab at MRCI, last week. Below is an edited transcript of that interview.
What was your experience with array technology prior to this study?
It has probably been seven or eight years since we started. We tried to make our own array using bacterial artificial chromosomes, and I quickly realized there were two real problems. One was quality assurance and the second was bioinformatics. I could see it was going to be a real challenge to do BAC arrays in house. So I started looking around for a commercial array. At that time there was little around, but Affymetrix was selling SNP arrays for genotyping. I wondered if there was any possibility if they would be quantitative and have the potential to be used in cytogenetics. We started using the very early 10K arrays. They were very crude, but they had been working on making denser arrays and they had 100K arrays in development. We set up a collaboration with them to try the arrays out. We published our first results in the American Journal of Human Genetics five years ago. It was good, but not as good as it is now. The bioinformatics is good now, too.
Why did you decide to work with Affymetrix SNP arrays?
We latched on to them we didn't have resources to look at everything, so we stuck with them. While the whole CGH thing was building up but, as I said, we weren't very interested in BAC arrays. We knew the resolution would be pretty crude. When the oligonucleotide arrays started to appear we didn't use them again because we didn't have the resources to look at another system. We have started recently using other array types as well, but we are still working with Affymetrix.
[ pagebreak ]
What other platforms are you using?
We are more or less looking at everything. When you get to the point of making a decision on what you platform to use in a diagnostic setting, you have to be very confident that it is robust, that the quality of the manufacturing is good, and that the bioinformatics is good. I worry about smaller biotech companies and how solid their finances might be because, if they go [out of business], it can be disastrous for your research. We are looking at [arrays from] Agilent, Illumina, BlueGnome, and NimbleGen. They are all very good, some have their particular advantages. They are all developing and getting better so I can't commit at this moment to any particular platform.
What are some of the advantages of using SNP arrays over BAC CGH, for example?
We didn't think they would be good enough to find small, submicroscopic abnormalities. Sometimes on these arrays there would be a single BAC showing an abnormality and you never knew whether it was an artifact or not. With a SNP array there are more features showing abnormalities, so the statistics of it being real are higher.
Can you provide an overview of the structure of your recent evaluation?
We used one of the 250K arrays. The real aim was to ask the question, 'Can we find abnormalities in patients that have been shown to have a normal karyotype using microarray approach?' We have a large clinical group here with 10-12 clinical geneticists. We let them select patients to test and we didn't want them to be too restrictive. We wanted them to select patients they thought might have a chromosomal abnormality. We didn't use a textbook, pre-selection procedure. We quickly found out that 15 percent of the samples that we looked at were clearly abnormal. What surprised us was that there were a few patients that had extremely mild phenotypes, and yet here they were showing pathogenic chromosomal abnormalities.
What is a 'pathogenic CNV'?
Copy number variation is ubiquitous in everyone. Normal, healthy individuals have hundreds of CNVs in their DNA. The vast majority must be harmless. They must be benign copy number variants. We know that some patients have copy number changes which are very suspicious and could be pathogenic. The way we identify them is that they are often de novo. They happen in that one individual; they are not present in a parent. And if you can find another patient with the same copy number variant with the same phenotype, it suggests very strongly that the CNV has clinical significance. To prove a CNV is pathogenic is difficult. It involves clinical testing and who has the time for that?
Upon what resources do you rely to assess CNV significance?
If we find a CNV and we don't know if it means anything or not, we can look at a few databases. One is called DECIPHER, which is hosted by the Wellcome Trust Sanger Institute. It is a catalog of potential pathogenic CNVs. If you can go into that database and find a similar case you can get in touch with that other lab and pursue whether its pathogenic or not. This is very influential in making a decision that it is pathogenic. Another is the Database of Genomic Variants hosted by the Centre for Applied Genomics in Toronto. That is a catalog of CNVs that are thought to benign. If you find your CNV in that list, you can more or less discredit it as a pathogenic CNV. You have to be careful though as these databases are new. There are also a lot of networks of labs that work together on these problems. We are drafting a paper describing a new syndrome due to a specific CNV with people in Sweden, Holland, the US, Belgium, Australia, and Mexico. The web has enabled all of this very quick communication.
[ pagebreak ]
You say that this study has led to improved diagnostic success in your lab. How so?
All the patients that we are testing are patients that have had a whole battery of testing done, including karyotyping, and have shown a normal karyotype. A lot of these patients have gone through a diagnostic odyssey of expensive testing, including magnetic resonance imaging, and yet no diagnosis been made. But, in 15 percent of our patients, we were able to get a firm diagnosis because they have shown a particular CNV. We are still testing and using more dense arrays and we are still getting a 10 – 15 percent pickup in children with idiopathic mental retardation. That's an enormous success rate compared with karyotyping, which may have a 2 – 3 percent success rate.
Also with SNP arrays you get genomic information as well. We are surprised by how much consanguinity we are finding. Consanguinity is everywhere and often harmless. So we are helping doctor find out that patients are consanguinous. In those cases the patients have a recessive disorder. We had one family with three different phenotypes, so it is quite important clinically. It is wasting the doctor's time if he doesn't know that, but with a SNP array it is very easy to tell. It is also important to warn these parents that there is an enormous risk of recurrence if they have more children. This all comes out of the SNP arrays.
To what extent will you adopt array technology for cytogenetic testing purposes? Is there room for the older technologies?
We are right in the middle of switching over to arrays. We are going to switch all of our testing of mental retardation patients onto SNP arrays and we will stop karyotyping. I think it is accepted that arrays offer the most appropriate way to investigate these patients because the diagnostic pick up is so much higher, but we will keep karyotyping in situations where we don't think an array will pick up those abnormalities. Anything that has a balanced translocation won't be detected by any type of array. We are not using any prenatal diagnosis yet; there is a whole range of issues there, both ethical and technological.
What are you using to confirm a diagnosis?
We are using multiplex ligation-dependent probe amplification to confirm our results. We can do a lot at one time; we can check 20 at one time. In some cases, you don't need to check a diagnosis made with an array because there are so many data points that the statistics alone tell you it's real. We only check using MLPA if the abnormality is small or if we want to check it in a parent. So, rather than running an array we just run an MLPA; it's cheaper.
In your experience, how does the status of Australian cytogenetic testing compare to what is being done in the North America or Europe? Is it ahead of the curve or lagging behind?
I think we have done well here. There are two or three labs that are at the same stage as ours; who have reasonable experience and are now getting into using arrays in diagnostic practice. There are a few problems, though; the way pathology is funded here is very good for government to control escalating costs, but it's bad for innovation of any type. You have to go through the bureaucracy to get them to provide any funding. In the US or UK it is easier to get funding, but Australia is at disadvantage, without a doubt.
Research funding is more difficult to come by for CNV research specifically. I think that copy number variation is the biggest thing that has happened in genetics in 10 years, and yet the national research funding organizations have decided not to invest their money in CNV research while the Europeans and Americans have gone into this enormously. There is an enormous amount of EU funding going in to this. So we have to be imaginative as to where we look for funding.
There has always been instantaneous translation from basic research to patient management. It is disappointing that the importance of this research area hasn't been recognized by research funding bodies in Australia.