Skip to main content
Premium Trial:

Request an Annual Quote

The Clinical Case for Array CGH

Premium

Clinicians have known for years that chromosomal structural changes can be related to disease, and thanks to high-resolution microarrays for probing SNPs and CNVs across the genome, scientists have been able to catalog more and more of these. In the past five years, the technology has gained a foothold in the clinical realm, and while the platforms are still evolving, arrays are being increasingly used in cytogenetics labs to diagnose disease.

"Overall, what I think is that people who are using this became more aware of the power of this
approach, and are using this now more widely and more frequently," says the Medical College of Wisconsin's Ulrich Broeckel, who has added diagnostic testing using arrays to his lab in the past six months.

Ankita Patel, co-director of the cytogenetics lab at Baylor College of Medicine, says that more than 50 labs in the US are now employing array CGH — or SNP/CNV chips — to look genome-wide at chromosomes for duplications and deletions that might cause disease.

While larger cytogenetics labs like those of Baylor, CombiMatrix Molecular Diagnostics, and Signature Genomics — the leading commercial services lab for cyto testing assays — have a corner on the market, experts say that smaller labs will be investing in the technology in the future. "More and more, when I talk to colleagues, there is a lot of interest from cytogenetics labs, smaller pathology labs, to start using this technology," Broeckel says. "If you can run it inside your lab and you don't have to send the samples outside, it's probably beneficial."

In fact, Patel says that her lab's sample volume has decreased from 600 samples per month to 400 simply because a lot of cytogenetics labs are bringing the technology in house.

"The whole field is moving very rapidly toward microarrays," says Joris Veltman, assistant professor at the Nijmegen Centre for Molecular Life Sciences in the Netherlands. "What you see in Europe is that all the academic labs are setting this up."

The consensus is that not only have arrays gained traction in the clinical lab, but also that they're bound to phase out more traditional approaches, like karyotype analysis. In Veltman's lab, which developed a microarray for clinical genetics services in 2002, scientists have been doing karyotype analysis and microarrays together for many years; now that microarray technology has been established and turnaround time has gone down, they plan to eventually replace all karyotyping with arrays.

David Ledbetter, a leader in clinical genetics at Emory University, predicts a similar future across the board. "At this point, the great majority, if not 100 percent of labs, want to transition to this new technology," Ledbetter says. "I'm not aware of any individual lab or lab director that I've talked to in the last year who doesn't think that everybody should move to this technology platform. I'd say virtually everyone is making plans to make this transition and is in agreement that this will become the standard cytogenetic technology platform and will largely replace — and be much more powerful than — the conventional karyotype."

Making the case

The main benefit to using arrays in the clinic is that they're more powerful, since the technology is high-resolution and can detect smaller duplications and deletions, or microduplication and microdeletion events, which were previously missed by karyotyping or other traditional, small-scale techniques. "The yield is about twice the yield of a karyotype," Ledbetter says. "The number and percentage of clinically significant deletions and duplications — or losses and gains — that are determined to be pathogenic is about double with this technology compared to conventional karyotype."

Using conventional karyotyping, says Baylor's Patel, the resolution is low — 3 megabases and above. "You miss a lot of the microdeletion or the smaller gene disorders," she says. "With microarrays we can go down to the resolution of an exon." The lab at Baylor was one of the first academic labs to offer clinical array testing beginning in early 2004, and they now custom-design their arrays through Agilent. Their most recent array is an exon chip, so genes that are known to cause a disorder are targeted at the level of an exon. "This would be hard by conventional karyotyping," Patel says. While they haven't gotten that much data yet with the newer chip, Patel says she thinks that for certain disorders they could identify newer exonic deletions that were not identified before. The Ledbetter lab has developed an array that allows detection of deletions or duplications anywhere in the genome greater than 500 kb in size, and more than 50 kb in selected critical regions.

Arrays won't totally make older techniques irrelevant, though. FISH, or fluorescent in situ hybridization, will still be used in complement to array CGH to find exactly where those duplications and deletions appear on the chromosome. And in the end, karyotype analysis still must be used to find balanced translocations, which arrays can't pick up yet. Ledbetter's lab tests children with developmental disabilities, undiagnosed or unexplained developmental delay, mental retardation, and autism; where array CGH comes in is when there is no specific genetic disease diagnosis, which would instead be confirmed with a targeted genetic test. While arrays can turn up small, usually de novo, deletions and duplications on specific genes, most of the changes Ledbetter sees involve 10 or 20 or more genes. "We often don't know which one is the specific one causing the [disease] in the child," he says. "There's a fair amount of misconception that the purpose of the test is to find a specific, individual genetic disease as opposed to find a cytogenetic abnormality, which usually involves multiple genes."

The significant improvement in the rate of diagnosis alone merits a good look at clinical array CGH. Most labs interviewed related a double-digit percentage increase in clinically significant numbers of detection. "It's been a significant improvement," says Ledbetter, who notes that using karyotype analysis, they could find 5 percent to 10 percent of clinically significant abnormalities; targeted BAC arrays added 5 percent, and whole genome oligo arrays added yet another 5 percent, totaling the current 20 percent of clinical abnormalities detected.

Veltman says that in patients with mental retardation, karyotyping and FISH would yield up to 10 percent diagnosed, but microarrays have increased that to 25 percent. "We hope with the move to higher-resolution platforms we can increase this percentage," Veltman says, adding that both arrays and increased knowledge of CNVs will help.

The big challenge

Even though they can identify more and smaller deletions, the biggest challenge to the field is determining whether these are pathogenic or not. And with more and more data on both normal and pathogenic CNVs becoming available to cytogeneticists, parsing out the good from the bad is taking more know-how.

Variants fall into three categories: benign, pathogenic, and unknown. In general, when trying to name an unknown disorder, the goal is to find de novo CNVs, "a very important proof that a CNV is linked to disease," Veltman says, after excluding normal variants and then testing parental samples to see whether they harbor the same variant. It's also important to consider whether the CNV has been implicated in a disease before.

All this takes consulting ever-increasing catalogs of normal variation, like the Database of Genomic Variants at The Centre for Applied Genomics in Toronto or other online databases for interpreting the results. And the basic research continues to add up. Steve Scherer, director of TCAG, noted in a story earlier this year in Genome Technology that 2008 had been a "watershed year" for schizophrenia, and that more and more studies were revealing CNVs to be involved in diseases as diverse as autism, cancer, macular degeneration, obesity, and others. Several large-scale studies have linked a microdeletion or duplication on chromosome 16p11.2 as the second most frequent chromosomal disorder associated with autism. A study out of Scherer's lab this September increased the range of possible variation linked to autism spectrum disorders, finding that both de novo and inherited deletions and duplications at 16p11.2 are associated with autism, mental retardation/developmental delay, and possibly other primary psychiatric disorders.

Likewise, a couple of papers published last year linked de novo CNVs to schizophrenia. The International Schizophrenia Consortium used SNP arrays to find that large, rare deletions on 15q13.3 and 1q21.1 were associated with developing schizophrenia, while a second from Decode Genetics found rare deletions at both of those loci as well as at 15q11.2.

With low-res techniques, data analysis is pretty straightforward, says Graham Snudden, VP of engineering at BlueGnome, which sells custom-printed Agilent arrays along with analysis software. However, higher-res chips have made quick analysis and interpretation a key challenge. "Things have changed, primarily the resolution of the platforms," Snudden says, adding that his company began with 2,000 probes and now offers chips with 180,000 to multiple million probes. Compared to two or three years ago, the large amounts of both functional genetic and CNV data have become daunting for labs, especially those that don't have bioinformatics people to help out with the data crunching.

To that end, the company recently launched a software upgrade that will be able to pool disparate sources of information into a curated database linked to the software, making CNV analysis much easier — and faster. Called DecisionTrack, the new software platform will allow labs to mine their own data and pool data from other labs, and will make curated annotation information available from online databases, along with detailed information from previous array CGH cases. Snudden sees this as the future, certainly when it comes to identifying smaller, and rarer, CNVs. "What you really want is a national-level consortium or something on that scale where you can very quickly identify relatively small numbers of patients who share the same imbalances," he says, noting that several countries are already launching projects of this nature. "I think we'll see more consortiums, [and] I think we'll see more standardization."

In the US, Emory's Ledbetter has spent the past two years setting up a consortium promoting data sharing and standardization of CNV classification. A series of workshops in 2008 led to the development of the International Standard Cytogenomic Array Consortium (ISCA), whose goals are to "encourage experts to work together to develop standards for the new technology, both in terms of the appropriate clinical application and the proper design strategy, resolution, and, most importantly, the interpretation of what's clinically significant versus what's normal variation," Ledbetter says. Awaiting a likely NIH stimulus grant, ISCA will use the money to launch a public database hosted at NCBI that will house both research and clinical array CGH data from 70 labs in the US, Europe, Canada, and Australia and will consist of de novo pathogenic variants. Sharing of data and increasing the number of known pathogenic variants will ultimately advance the utility of arrays in the clinic.

That third type of variant, the unclassified category, Ledbetter says, "at the moment is large, but it's shrinking; and the rate of shrinking is going to accelerate as more people practice the technology on normal populations and on patient populations, [and] as long as they're willing to contribute this data to public databases so that we can all benefit from that information."

Choosing a platform

Not all labs agree on what's the best chip to use. For whole-genome analysis, everyone interviewed offers Agilent custom-designed oligo arrays, and various vendors sell and support these tools. When it comes to more targeted analysis, and scouting out balanced changes like uniparental disomy, Illumina and Affymetrix's SNP/CNV chips are preferred.

ISCA is working on a custom design, made with input from five major genetics labs that have previously been creating their own chips. They've merged the best features of all five designs into the 'ISCA design,' says Ledbetter, and have made that publicly available to everybody who would like to use it and to any vendor who would like to print it. The chip is currently available from Agilent, BlueGnome, Oxford Gene Technology, and Affymetrix. "We wouldn't recommend to anybody to use an off-the-shelf array designed by a vendor," he says.

Ledbetter says that the advantage for clinical labs to using Illumina and Affy chips, which offer both SNP and CNV content, is that they can identify uniparental disomy, which "can be clinically significant," he says. On the other hand, from his experience, copy number platforms "like Agilent, OGT, Blue-Gnome, which are all Agilent-printed, and NimbleGen give a little bit better-quality copy number data," he says. "Those companies are all investigating how to add SNP information to their high-quality copy number data, and the SNP platforms — Illumina and Affy — are trying to add better-quality copy number probes to their arrays." He adds, "Everybody's trying to move towards an ideal array that gives high-quality, high-resolution copy number and SNP information, but people have different assessments about which platform today comes the closest to that ideal."

Ulrich Broeckel solely uses Affy's 6.0 array, mostly because his lab has been using it for research for a long time and is very familiar with the platform. "I think the key feature in terms of what platform you use depends on the resolution, so what's the smallest segment size you can detect," he says. "And how well does it cover regions which are known to affect disease and how well does it cover known genes?"

Joris Veltman switched to Affy in 2005 after comparing platforms to see which had the best performance. He currently uses the 250K SNP array, as well as the 6.0 and new 2.7M array, which is the basis of Affy's Cytogenetics Research Solution. The 2.7M offers coverage of 2.7 million copy number markers and 400,000 SNPs for detecting loss of heterozygosity, uniparental disomies, and regions of the genome that are identical by descent. He also uses a NimbleGen array for target analysis, "when we want to specifically focus on a detailed region of the genome." According to Veltman, "The 2.7M array really seems like a platform that can now be used in diagnostics," so the lab is working to make the switch.

Deciding which array to use depends on the application, Veltman says. "For diagnostic applications, we like the SNP content because it gives us a lot of additional quality control." In Europe he sees a lot of variation, with labs using the Affy, Illumina, as well as CGH-based Agilent platforms. "There is still no decision yet," he says. "I think what is very important for routine cytogenetics labs [is] that interpretation is straightforward so that people can interpret the results in a routine diagnostics setting."

The Scan

Shape of Them All

According to BBC News, researchers have developed a protein structure database that includes much of the human proteome.

For Flu and More

The Wall Street Journal reports that several vaccine developers are working on mRNA-based vaccines for influenza.

To Boost Women

China's Ministry of Science and Technology aims to boost the number of female researchers through a new policy, reports the South China Morning Post.

Science Papers Describe Approach to Predict Chemotherapeutic Response, Role of Transcriptional Noise

In Science this week: neural network to predict chemotherapeutic response in cancer patients, and more.