If DNA doesn’t add up, array CGH
  steps in

Related articles:
"No cookbook approach"
DNA Samples (PDF document 1,431 K)

September 2005
Feature Story

In the laboratory sweepstakes, microarray-based comparative genomic hybridization (array CGH) has the appearance of a winner. With the speed and power of microarray technology and the focus and resolution of CGH, this newly foaled technique has an excellent pedigree. Its ability to track down constitutional chromosomal abnormalities puts it at the head of the pack. Although array CGH has made only a few appearances in the clinical laboratory, this new entry has already generated tremendous enthusiasm. Whether it will earn a major financial prize, however, is a question that will be settled outside the laboratory performance arena, in deliberations over molecular devices at the FDA, issuance of new CPT codes, and determination of Medicare reimbursement. There may also be disputes over just which syndicate owns this high-performing thoroughbred. All in all, at the present time no one can fix the odds on its success.

Array CGH is "totally revolutionizing chromosome diagnosis," says Arthur Beaudet, MD, chair of the Department of Molecular and Human Genetics at Baylor College of Medicine, one of only two places in the U.S. where this technique is being offered as a clinical service. "The old-fashioned form of karyotype analysis will disappear," Dr. Beaudet predicts. "You can diagnose things with array CGH that you can’t diagnose with karyotyping alone. And you would have to combine hundreds of FISH tests at the same time to equal its power." He has been in clinical genetics for 35 years, he says, and this has been "by far the biggest revolution" he has seen. "In terms of changes in lab methodology of direct clinical relevance, I don’t think it gets any more impressive than this," he says.

Bassem Bejjani, MD, FACMG, medical director of the other U.S. laboratory offering array CGH for clinical diagnosis of chromosomal abnormalities, says array CGH will create "a new place" in cytogenetics. "FISH will not be eliminated, but its role will be transformed," he says. Dr. Bejjani, who is research professor at the Health Research and Education Center at Washington State University, co-director of the molecular diagnostics laboratory at Sacred Heart Medical Center, and co-founder (with Lisa G. Shaffer, PhD, FACMG) and medical director of Signature Genomics Laboratory, predicts that array CGH will often be the first-choice test for suspected chromosomal abnormalities. "I already have two physician clients who send specimens and ask for array CGH first," he says. "If it is normal, then we do a standard chromosome study. In the future, I think this is exactly what will happen."

Scientists at Quest Diagnostics-Nichols Institute, San Juan Capistrano, Calif., are developing their own version of this technology. "As our understanding of the human genome increases, we’re finding that the frequency of variations to our genomic content is much higher than anticipated. Array CGH is an ideal tool to detect and characterize these variations," says Mansoor Mohammed, PhD, director of the Department of Advanced Technologies at Quest Diagnostics-Nichols Institute.

He notes that, in the clinical diagnostic setting, array CGH is useful in two primary situations. First, when there is an unclear phenotypic presentation, array CGH can be used as a first option to screen the entire genome for aberrations that would otherwise be difficult to predict. Second, when a child presents with a classic syndrome but traditional karyotyping or FISH analysis reaches a roadblock, array CGH can often detect the underlying cryptic abnormality. In addition, when a standard technique does suggest a chromosomal abnormality, "you can bring in array CGH to confirm and clarify the chromosomal abnormality at a significantly higher resolution," Dr. Mohammed says. He argues that array CGH will become a cost-saving method either when used independently or in conjunction with chromosome analysis, as opposed to having to rely on potentially several rounds of sequential FISH tests.

One application in which array CGH will be especially helpful will be in identifying abnormalities in regions adjacent to the telomeres (sometimes called subtelomeric abnormalities). Clones for these regions have been developed by David H. Ledbetter, PhD, director of the Division of Medical Genetics at Emory University School of Medicine, in collaboration with researchers in the United Kingdom. These clones have revealed abnormalities in three percent to five percent of cases of children with developmental delay or mental retardation who had normal banding analysis. "Telomere FISH moved rapidly into clinical practice," says Dr. Ledbetter, "but it is limited obviously by looking for cryptic or submicroscopic abnormalities only in one part of the genome at a time." It would be helpful to be able to look for very small deletions and duplications anywhere in the karyotype. "Current array technology probably represents the best chance to do that," Dr. Ledbetter says.

A commercial array for CGH is being developed at Spectral Genomics in Houston, a spinoff from the Baylor College of Medicine. "This type of technology, once approved for clinical diagnostic laboratories, will start replacing FISH analyses, effectively offering hundreds or thousands of FISH tests at the same time," says Robert Johnson, PhD, Spectral’s CEO and president. Dr. Johnson also expects to see some displacement of conventional karyotyping, but to a lesser degree than FISH.

"It’s going to be very interesting to see how this works out," says Jean Amos Wilson, PhD, scientific director of human genetics at Focus Diagnostics, Cypress, Calif. "It’s almost like cystic fibrosis was in 1992, when we knew we had to test for a lot of mutations but we didn’t have good platforms to develop homebrew assays and there were no ASRs." For a laboratory to develop array CGH in-house now requires an intensive development program. (Signature and Baylor both design and make their own arrays.) CGH has enormous potential significance," says Dr. Amos Wilson. She notes that Dr. Bejjani’s laboratory has used it to detect a case of Duchenne’s muscular dystrophy, which is often caused by deletions, in a child too young to be symptomatic.

Consideration of array CGH begins with advantages. As location is to real estate, resolution is to array CGH. "You miss a lot looking at chromosomes under a microscope," Dr. Bejjani says. With G banding, in the best hands resolution is three to five megabases (Mb), while resolution for any region with a bacterial artificial chromosome is 70 to 90 kilobases (Kb). "That is our experience, and that of others," Dr. Bejjani says. Resolution of FISH is similar to that of array CGH, but here other advantages of CGH come into play. "In FISH you use a single probe and must have a strong suspicion of which probe to use," Dr. Bejjani says. Array CGH is equivalent to using hundreds of FISH probes in parallel on a sample. In addition, array CGH uses not metaphase chromosomes but extracted DNA, so culture is not needed, allowing a shorter turnaround time. Dr. Bejjani says that in practice he can deliver a result in less than five days.

Before describing what array CGH is and how it works, Dr. Bejjani is careful to distinguish what it is not. "CGH arrays have nothing to do with gene expression microarrays," he says. CGH arrays are structural, not functional—they look for DNA that is missing or duplicated, not DNA that is expressed.

In a way, array CGH turns FISH upside down. In FISH, sample DNA is laid down on a substrate and dye-labeled probes are layered over this chromosome spread. Array CGH, on the other hand, starts with printing hundreds or thousands of probes on a microscope slide. These probes are bacterial artificial chromosomes, or BACs, which are sequences of cloned human DNA 120-180 Kb long that have been mapped to specific locations on specific chromosomes. (Most of the current library of BACs was prepared for the Human Genome Project.) To the array is added DNA from a patient (test DNA) and DNA from an apparently normal person (reference or control DNA), each labeled with a different dye. The essence of the experiment is that the two DNA samples compete for hybridization to the BAC probes on the slide—the "comparative" aspect of CGH.

When each DNA sample contains equal amounts of the DNA corresponding to a particular probe, equal amounts of each prep will hybridize to that probe. If the patient’s DNA has a deletion, the reference DNA will be twice as likely to hybridize to BACs covering that region. For a duplication, there will be excess hybridization of patient DNA to corresponding probes. After hybridization, the array is read by a laser scanner and a computer calculates color ratios and translates them into gain or loss of a chromosome region, which can identify a clinical syndrome.

"Array CGH works for any chromosomal rearrangement that shows a dosage abnormality," Dr. Bejjani says. This means anything that changes the copy number of a DNA segment—microduplications, microdeletions, and unbalanced translocations. It does not detect balanced translocations or inversions because all the DNA is present—it is just rearranged. "Whenever you have a change in gene dosage, you will have clinical consequences," Dr. Bejjani says. "Unbalanced chromosomal abnormalities make up the vast reservoir of our patients. Balanced abnormalities make up only a small percentage."

Several details of array CGH are important. Each affected region is represented by more than one BAC on the array that Signature developed. In the affected region of Potocki-Shaffer syndrome on 11p11.2, for instance, Signature’s array contains nine BACs. "In each patient," Dr. Bejjani says, "for any one locus that deviates from baseline, the array results should deviate for all BACs at that locus."

While Dr. Bejjani uses BACs as probes on his arrays, other groups use synthetic oligonucleotides, or "oligos," which can be as small as 45 base pairs in length, as compared with 100,000 base pairs for a small BAC. In theory, oligos offer better resolution, since the longer BACs may sometimes bridge the affected region. Using multiple overlapping BACs for each affected region minimizes this problem. Oligos, on the other hand, have a higher noise-to-signal ratio, Dr. Bejjani says, and thus are more difficult to interpret.

BACs are purchased from commercial sources and don’t always contain what is stated on the label. All BACs must be tested in-house to see whether the clone contains the DNA that it is supposed to contain—otherwise it will mismap—and to ensure that it contains unique DNA, not repetitive sequences, or it will hybridize to multiple chromosome sites.

Also, a BAC must not be from a polymorphic region of the genome. When Dr. Bejjani’s laboratory tested a group of probes against the DNA of 50 apparently healthy individuals, many probes showed gain or loss. "We decided that for diagnostics we want to avoid these regions," Dr. Bejjani says. "Every time you find a polymorphism, you have to follow up on it, which is costly."

In sum, out of 906 clones that Dr. Bejjani tested, 61 (seven percent) mapped to the wrong region and 149 (17 percent) cross-hybridized to more than one region. Only 589 (65 percent) turned out to be suitable for the array. Once the array was designed and printed, it had to be validated by running it against many samples with known abnormalities and known normal controls (Bejjani BA, et al. Am J Med Genet A. 2005;134:259-267). Dr. Bejjani emphasizes that this is a targeted array—one in which every BAC corresponds to a genetic locus implicated in a congenital disorder. Arrays can be made that cover the entire genome, but with that type of array many results will not be interpretable in clinical terms.

At Signature, each array CGH analysis includes a male-female comparison—normal DNA of the opposite gender from the test sample—as an internal control. Between this pair of samples, there will always be a gene dosage difference for the X chromosome. If a run shows no difference between the test and reference samples, but does have the expected difference between the male and female samples, that shows that the lack of difference is not due to a technical deficiency in the run.

Each patient dataset is actually made up of two experiments on the Signature platform—one in which the test and reference DNA samples each have a different dye color and a second in which the dyes are switched. Both experiments are plotted on a single graph and should show the exact inverse of each other (Related article: DNA samples). For technical reasons, data are plotted as log 2, rather than arithmetically.

Signature’s first array contained 831 BACs covering 126 loci. Dr. Bejjani’s laboratory is developing a new array with greater coverage of the subtelomeric and pericentromeric regions. Signature first offered array CGH as a clinical service in March 2004. "In the first couple of months we had only a handful of cases," Dr. Bejjani says. "Now we are doing 65 to 70 cases a week. We started with three employees, now we have 20." To accommodate the increasing demand, Signature is moving toward automation. "This technology allows you to automate more easily than conventional cytogenetics," Dr. Bejjani says. "It is not as limited by human skills and training."

Dr. Beaudet at Baylor also uses BAC arrays, but says, "There is plenty of literature to indicate that oligo arrays have enormous attraction. Maybe they will be the final format for this technology." They can print BAC arrays now cost-efficiently, he says: "We can print the arrays and do a complete analysis for a little over $1,000." If they switched to oligo arrays, simply purchasing the arrays themselves would cost $1,000 or more, Dr. Beaudet says. Affymetrix, which holds the right to printing oligo arrays, "is unwilling to negotiate a license," according to Dr. Beaudet. Nonetheless, he believes that oligo technology will eventually predominate. "Signal-to-noise problems can be overcome," he says.

Just as Dr. Bejjani did, Dr. Beaudet had to validate his array in-house (Cheung SW, et al. Genet Med. 2005;7:422-432). Regarding polymorphisms, or benign variants, Dr. Beaudet says, "We have tried to weed out those clones. If we do use one, we check the [healthy] parents to see if the same gain or loss is present in their genome," which would indicate that it is probably not the cause of the child’s problem.

"We are just beginning to offer array CGH on a prenatal basis," Dr. Beaudet says. "It has tremendous attraction. It can diagnose a number of disorders that would not otherwise be diagnosed." Examples include diGeorge syndrome, Angelman syndrome, and Prader-Willi syndrome. "For someone having an amniocentesis, array CGH becomes a possibility," Dr. Beaudet says. "Should we be offering all women an invasive test such as amniocentesis?" he asks. "Some people are arguing that the answer is yes" (Caughey AB, et al. Obstet Gynecol. 2004;103:539-545).

Noninvasive alternatives include finding fetal cells in maternal circulation, isolating fetal cells with a cervical cell collection, and isolating fetal DNA from maternal plasma. It is unlikely that prolonged presence of cells or DNA from previous pregnancies would be a problem with the latter two methods, according to Dr. Beaudet. "In five to 10 years we may be able to do array CGH on noninvasive samples," he says. "That would be a huge revolution in genetic diagnosis." When they counsel parents-to-be now they say that there is "some merit" to doing amniocentesis in all pregnant women and, if parents elect an amniocentesis, they offer array CGH as an additional test.

Array CGH will definitely remain a sendout for a while, says Dr. Amos Wilson. Laboratories desiring array CGH analysis will be sending samples to either the Signature or Baylor laboratories. They would have a choice of how to proceed, Dr. Amos Wilson explains. "If the clinician suspects an unbalanced chromosome rearrangement, the choices are not doing routine cytogenetics at all, or maybe doing routine cytogenetics to look for large alterations, then sending a sample to one of these laboratories." For now, array CGH will be secondary to routine cytogenetics but will allow medical geneticists to skip FISH, in Dr. Amos Wilson’s view. "CGH could be considered superFISH," she says. "You do FISH because you suspect a specific chromosome alteration. With array CGH you can look at the whole genome when you don’t have a suspected target." Referring laboratories will also have to make a financial decision, she says. "Laboratories make no money if they send out."

A case studied by Quest’s Dr. Mohammed and Shelley Gunn, MD, PhD, illustrates the power of array CGH. A child presented with a classic phenotype consistent with what would be expected as a consequence of a large deletion of the distal portion of 18q—facial abnormalities, malformation of limb development, and abnormalities of white brain matter. Yet chromosome analysis appeared normal. "Large regions of 18q can be deleted and still be consistent with life," Dr. Mohammed says, "so this anomaly should have been easily seen on chromosome analysis." After years of chromosome and FISH analyses, the case reached Dr. Mohammed’s laboratory. FISH probes confirmed a large deletion of 18q. Array CGH uncovered the basis of the abnormality: the large (approximately 7 Mb) deletion on 18q was balanced by a concomitant duplication of material from chromosome 4q in the segment of 18q from which DNA had been deleted. This case highlights the need for "innovative molecular cytogenetic techniques," the authors wrote (Gunn SR, et al. Am J Med Genet A. 2003;120:127-135). "Within the year," Dr. Mohammed says, "a colleague in France, Dr. Anne Moncla, who had a patient with a classic presentation of 18q deletion syndrome that could not be confirmed by chromosome studies, used our array CGH approach and found that her patient in fact had the exact breakpoints and unbalanced translocation as our child" (Am J Med Genet A. 2004;131[3]:131). The odds of uncovering this finding by any other approach would have been "incredibly small," he adds.

Neither the Baylor laboratory nor Signature Genomics sells their arrays. However, at least two companies are developing commercial CGH arrays. At Abbott Molecular Inc. (formerly Abbott/Vysis), a product for research use only is available, says medical director Timothy T. Stenzel, MD, PhD, FACMG. are targeting the research market right now," Dr. Stenzel says. "But we do have a plan to offer array-based CGH for diagnostic purposes in the future and would like to get FDA regulatory clearance."

Spectral Genomics has two products for research use, Dr. Johnson says. One is a chip with more than 2,600 BAC clones spaced roughly 1 Mb apart that gives a whole-genome view of changes in copy number. The second is a chip with a much smaller number of BACs that is targeted to constitutional syndromes. In its current research version, it is designed to pick up 42 congenital disorders as well as 41 subtelomeres. Spectral initially approached the FDA in October 2003 with the 1-Mb chip, but the agency expressed a desire for a more focused device, Dr. Johnson says. "So we honed in on constitutional syndromes." Dr. Johnson notes that there is a body of literature describing which chromosomes and which areas of those chromosomes show changes corresponding to specific constitutional syndromes. As with Dr. Bejjani’s and Dr. Beaudet’s diagnostic arrays, a change in a specific spot correlates with a specific syndrome. Spectral has submitted a pre-IDE that is under review. It intends to go for a 510(k) designation, Dr. Johnson says.

In Dr. Beaudet’s view, the "FDA is currently a potential obstacle to this technology." The FDA appears inclined to think about a relatively limited number of clones, he says, whereas people are already using arrays with 800 to 1,000 BACs and are likely to go higher.

Future additions to clinical diagnostic arrays for CGH will include more BACs covering subtelomeric regions, which Dr. Ledbetter prefers to call "unique telomere clones." He began working more than 10 years ago—before the Human Genome Project’s physical map was as complete as it is now—to identify clones for DNA regions that are unique to each chromosome end and as close to the end as possible. Abnormalities near the ends of chromosomes underlie many developmental anomalies, particularly mental retardation syndromes. However, Dr. Ledbetter says, "The very end of each human chromosome arm contains repeat sequences common to many chromosomes." To get diagnostically useful clones that are specific or unique required a lot of physical mapping. Once such probes were obtained and validated, they were incorporated into FISH kits and are now standard of care in unexplained cases of developmental delay or mental retardation with normal banding analysis. Dr. Ledbetter says he was "surprised" when the unique telomere clones were found to explain three percent to five percent of such cases. "When we were first developing the probe set and thinking about the frequency with which you might find telomere imbalances below the resolution of karyotyping," he says, "I thought one percent would be useful and clinically important." Finding the much higher value actually makes these probes "clinically extremely valuable," he says.

Another clinical value of the unique telomere probes is that parents of children with telomere imbalances have been found to have a 50 percent probability of being carriers of balanced telomere rearrangements. It is known from studying couples with multiple miscarriages that a parent with a balanced translocation can contribute a cytogenetically visible unbalanced translocation to an offspring. "We have now discovered that cryptic unbalanced rearrangements are also inherited from a balanced and normal parent carrier," Dr. Ledbetter says. "That’s important for counseling."

While array CGH sails smoothly along, ahead lies a potentially dangerous iceberg—intellectual property conflicts. "We have a very strong patent position in this area," says Dr. Stenzel of Abbott Molecular. They have exclusively licensed the University of California’s fundamental CGH patents, which Dr. Stenzel says are related both to CGH and array CGH (Kallioniemi A, et al. Science. 1992;258:818-821), as well as UC’s array CGH patents and Deutsches Krebs forschungszentrum’s matrix CGH patents out of the University of Heidelberg. Laboratories and companies offering, or intending to offer, array CGH have responded to this situation in various ways.

Dr. Beaudet says, "Affymetrix believes that they have patents of relevance to CGH, and we are seeking a license from them." The Baylor laboratory is in "the final stages" of negotiating a license from Affymetrix for BAC arrays. "Eventually there will be a fight or negotiation between Abbott Molecular and Affymetrix," Dr. Beaudet predicts. "One possibility is that we will need a license from both of them."

Dr. Bejjani is more cryptic about Signature’s approach. "We are in discussion with a number of companies regarding licensing agreements," he says. Confidentiality agreements prevent him from being more specific.

Spectral’s work is covered by exclusive licenses from Baylor for crosslinking DNA and attaching it to any surface and for the preparation and manufacture of arrays, and by a license from Affymetrix that covers a large number of patents on the ability to print arrays, Dr. Johnson says. Asked about the UC’s fundamental CGH patent, he says, "Based on what we know about the patent, that is not something we are concerned about."

Quest is taking a different approach and has developed a proprietary method for array CGH. "The fundamental premise of current array CGH approaches hinges upon using two colors to differentiate binding derived from the test and reference samples respectively," Dr. Mohammed explains. Quest’s innovation is to use only one color. The initial steps of standard CGH are actually performed with no color, but each DNA sample is tagged with a short unique DNA sequence. After hybridization, when the comparative genomic step has been done, a fluorochrome is added to which is attached the complement to the tag on the reference sample. Scanning at this point gives the amount of reference sample hybridized. Then the same fluorochrome is added, but this time it has attached to it the complement to the tag on the test sample. Scanning this reaction gives the total signal. Subtracting gives the test signal. Since these steps are fast, little time is added to the procedure, Dr. Mohammed says.

Reimbursement is another bureaucratic hurdle. "It will be very important to know that a laboratory can be reimbursed for this," says Dr. Amos Wilson. "As with all molecular testing, as technology changes we don’t have CPT codes that are specifically applicable."

Says Dr. Bejjani: "Right now we do institutional billing, insurance billing, patient billing—the same as with other tests, such as FISH. We are starting to make headway with reimbursement with some carriers." He agrees that CPT codes are a problem. He uses CPT codes from other technologies for individual steps, such as DNA preparation, hybridization, and interpretation and bundles them, "as in the early days of FISH."

Dr. Beaudet says that at present the Baylor laboratory requires guaranteed payment from hospital sendout laboratories. "I am very optimistic about reimbursement within the next two years," he says.

Dr. Bejjani brings up one final consideration: Who should interpret array CGH?

"With this new technique, the conceptual line has been blurred between cytogenetic conditions that are diagnosed with G band analysis and molecular conditions that have been traditionally diagnosed with techniques such as PCR and Southern blots," he says. Array CGH is both a cytogenetic and a molecular technique. "I feel cytogeneticists need to learn this technique and do the interpretation," Dr. Bejjani says. "For chromosomal diagnosis we need someone who knows meiosis and who knows which family members need to be tested. Cytogeneticists have been doing this forever and are more attuned to the mechanisms of chromosomal abnormalities."

At the same time, he acknowledges, "All of this separation is artificial. I think we are moving toward a unified world, in which cytogeneticists are trained to know molecular and molecular biologists are trained to know cytogenetics. Those who will navigate this field will be those who are cross-trained."