September 2002
Cover Story
Karen Titus
Given the chorus of conjectures that rise up from any discussion on brain tumors, it’s tempting to start with perhaps the simplest statement of all. Which is: Identifying loss of chromosomal arms 1p and 19q in anaplastic oligodendroglioma can help direct patients toward effective chemotherapy.
It’s a nice, neat picture. Even better—it’s true. That’s one reason laboratories are ramping up to offer testing for 1p/19q deletions, and why those not performing the tests are, more and more, sending samples to labs that do.
An even tidier picture would be to see this as a triumph of genetics over morphology, and to anticipate the day when molecular diagnostics reveals itself to be a unified theory of gliomas, if not of laboratory medicine. A sort of medical Big Bang, if you will.
But as physicists have known for decades, such grand-scheme theories evoke more questions with every answer. That’s why they keep scratching their heads and asking things like, Was there one Big Bang, or many? How do we reconcile the elegance of relativity with quantum chaos? Now that we’ve discovered dark energy, where does it fit into our theories?
In medicine, the march toward the molecular, though unquestionably rewarding, has been similarly vexing. In the case of oligodendroglioma, knowing 1p/19q status is both a breakthrough and a burden.
Loss of 1p/19q in oligodendroglioma—an admittedly rare, though not inconsequential, tumor—"is interesting in its own right. But more than that, it’s a glimpse into the future," says J. Gregory Cairncross, MD, professor and chairman of the Department of Oncology, University of Western Ontario, London, Canada.
"Certainly the oligodendroglial tumor field has crystallized around this particular test. It’s the first toehold," says Clayton Wiley, MD, PhD, director of neuropathology at University of Pittsburgh Medical Center and president of the American Association of Neuropathologists. "My guess is that we’ll slowly start adding on individual neoplasms, and that those neoplasms will each have their own unique profile." Within 10 years, he predicts, pathologists will be able to subdivide neoplasms that have a very wide spectrum of outcome.
That’s looking forward. Looking backward, to the emerging picture framed by the molecular arrivistes 1p/19q, is just as engaging.
The two chromosomes stepped into the spotlight four years ago, with an article published in the Journal of the National Cancer Institute (Cairncross JG, et al. 1998; 90: 1473--1479).
This came on the heels of earlier genetic research looking at clinicopathological types of glioblastoma, says David Louis, MD, director of the Division of Molecular Pathology and Research in the Department of Pathology, Massachusetts General Hospital, Boston, and one of the article’s authors. It had become apparent that TP53 mutations and EGFR amplification marked the two primary, mutually exclusive pathways by which glioblastoma emerges, and that pathological differences in tumors, such as predominance of either small cells or giant cells, correlated with a specific genotype. "But there wasn’t much clinical interest in this information, because glioblastomas are universally poor-prognosis tumors," says Dr. Louis.
Because Dr. Cairncross and colleagues had already shown that anaplastic oligodendrogliomas respond frequently—about 50 percent of the time—to a combination chemotherapy known as PCV (procarbazine, lomustine, and vincristine) and to PCV-like drugs, such as temozolomide, these tumors became the next target of interest. "So we teamed up with Dr. Cairncross to ask, Could we do what we had done for glioblastoma with the anaplastic oligos—could we use genetics to subdivide the tumors into groups?" Dr. Louis says. "And if we could subdivide them, was there a clinical correlation?"
The answer is "yes," as Drs. Cairncross and Louis and colleagues showed in their 1998 article and in a study published last year in Clinical Cancer Research (Ino Y, et al. 2001; 7: 839-845).
Dr. Louis explains: An anaplastic oligodendroglioma with allelic loss of chromosomal arm 1p—that is, the short arm of chromosome 1—has a 100 percent chance of responding to PCV chemotherapy. About 50 percent of those responses are complete and durable. Patients who have loss of 1p coupled with loss of 19q—the two almost always occur together in these tumors—have an average survival time of more than 12 years. "That’s a remarkable number when one considers that before the era of PCV chemotherapy, the average survival of these patients was under two years," Dr. Louis says.
In contrast, patients with anaplastic oligodendrogliomas without 1p/19q loss, but with any number of other genetic changes, such as CDKN2A deletions, 10q loss of heterozygosity, or EGFR amplification, very rarely respond to chemotherapy. Those who do have incomplete, nondurable responses. "These patients have short survivals, more on the order of a year and a half," Dr. Louis says.
How useful is this information? Because so few successful treatments exist for oligodendroglioma, some might contend that all oligodendrogliomas, including those that look like oligodendrogliomas but are actually astrocytomas or so-called mixed oligoastrocytomas, should be treated with PCV, just on the chance they’ll respond.
"The argument is, ’Why not? Suppose he overdiagnoses a few oligodendrogliomas—so what? At least we’re getting them all,’" says Peter Burger, MD, professor of pathology at Johns Hopkins University, Baltimore. Though Dr. Burger does not necessarily advocate this approach, he’s certainly familiar with it, having had no shortage of requests from patients to review their slides in the hopes that this second opinion will render a diagnosis of high-grade oligodendroglioma-thus opening the door to PCV chemotherapy. "Some patients will almost beg you to find an astro/oligo component. And if you say, ’Maybe this is an oligodendroglioma,’ then PCV could be useful. It’s not as if a patient with a high-grade astrocytoma and treated with PCV will be missing out on a better option. There’s no such thing as a very specific, very effective astrocytoma treatment," Dr. Burger says.
Why not, indeed?
If only matters were that simple. They’re not, of course. This is a tripartite problem, involving not just therapy but diagnostic and prognostic matters as well.
To be sure, there are practical clinical sequelae to knowing a tumor’s 1p/19q status.
"In many cases it has a profound influence on our recommended management plan," says Gene Barnett, MD, chairman of the Brain Tumor Institute, Cleveland Clinic.
"It already influences management decisions for individual patients. That’s the beauty of it," says Dr. Cairncross. "It allows me to make individual recommendations, rather than recommendations that on average are OK—but for the individual are not necessarily OK." And, predicts Dr. Cairncross, 1p/19q information is likely to affect treatment decisions "more and more in the months and years to come."
Working through the various scenarios, Dr. Cairncross and others explain how the information helps them:
Patients with a newly diagnosed malignant oligodendroglioma with 1p/19q loss stand a good chance of responding to treatment. Chemotherapy is part of the initial management. This is usually—but not always—followed by radiation. "There are some circumstances in which radiation won’t always be absolutely necessary to a good outcome. In fact, I think there are situations in which delaying radiation has advantages, because you’re able to postpone certain side effects," Dr. Cairncross says.
Then there are the newly diagnosed anaplastic oligodendrogliomas lacking the telltale genetic signature. "They behave very badly," says Dr. Cairncross. "Today’s chemotherapy has virtually no role in treatment." Radiation, on the other hand, does, and should be given straightaway, he says, although prognosis remains poor even with radiation.
In these cases, knowing that a high-grade oligodendroglioma is 1p/19q-intact may help clinicians steer their patients toward investigational treatments at the outset, and away from PCV. As Dr. Barnett points out, chemotherapy is not necessarily benign. "To be giving chemotherapy when it’s minimally effective subjects patients to unnecessary risk. And it’s costly," he says.
How common is 1p/19q loss? Says Arie Perry, MD, assistant professor of pathology, Washington University, St. Louis: "Generally, in the pure oligodendrogliomas, people find 1p and 19q deletions—it’s been reported between 50 and 80 percent. In our lab it’s been about 70 percent."
The emphasis to date has been on grade 3 oligodendrogliomas, because of the strong evidence that late-stage tumors with 1p/19q loss are chemosensitive. But the incidence is the same in grade 2 oligodendrogliomas.
"The literature and our own experience indicate there are a lot of reasons to justify performing 1p/19q testing routinely on all neoplasms with an oligodendroglioma phenotype," says Susan M. Staugaitis, MD, PhD, staff pathologist, Cleveland Clinic.
Knowing the 1p/19q status of low-grade oligodendrogliomas is indeed helpful, according to Dr. Cairncross. "I think if it has a 1p/19q loss, it’s going to have a very favorable natural history—it will be very slow-growing. And chemotherapy almost certainly will be useful when the time comes to use it. Radiation might also be helpful, but there’s no rush to do it." For these patients, surgical management would be the principal treatment in the first five to 10 years.
In such cases, Dr. Cairncross says, "I’m generally trying to postpone the chemotherapy and radiation of today, figuring that better things are in the offing. And if a patient can live with the tumor, I don’t want to treat them prematurely with somewhat helpful, somewhat harmful treatments."
Low-grade oligodendrogliomas without 1p/19q loss are more likely to cause trouble sooner, Dr. Cairncross says, and less likely to respond to current chemotherapies. In these cases, radiation is probably the more dependable therapy when the time comes to act.
There’s also clinical evidence to suggest that 1p deletion in mixed tumors, and even in astrocytomas—though only rarely in the latter—may be associated with high chemosensitivity. "For the astrocytoma story, we still don’t have enough information," says Dr. Staugaitis. 1p and 19q loss in astrocytomas is very rare—it occurs in only about 10 percent of these tumors. In mixed gliomas, the number appears to be slightly higher—perhaps 15 to 20 percent. Mounting evidence suggests that patients with astrocytomas with combined 1p/19q loss may also have a better prognosis. "But we still haven’t been able to systematically examine a sufficient number to say that with certainty," she says.
Likewise, it’s possible that the small number of astrocytic tumors that respond to chemotherapy may actually be misclassified oligodendrogliomas. "But I don’t think that’s the case most of the time," says Dr. Cairncross. "I think there are genetic subtypes of astrocytic tumors that are also sensitive to today’s treatments—we just haven’t learned what the marker is."
The confidence clinicians gain by knowing a tumor’s 1p/19q status is "the tip of the iceberg," Dr. Cairncross says. "Undoubtedly our ability to classify oligodendroglioma and to predict response to treatment will get better and better. We’ll refine it further. But we’ve got our foot in the door, anyway, with this tumor type."
Unfortunately, there’s no sine qua non regarding the diagnosis of oligodendroglioma. "There’s a lot of controversy over exactly what criteria should be used to diagnose them," says Dr. Perry.
Gliomas are vagabonds, migrating across and through the brain; oligodendrogliomas commonly are contaminated with normal glial cells, explains Robert Jenkins, MD, PhD, co-director of the cytogenetics laboratory and professor of laboratory medicine and pathology at Mayo Clinic. In addition to this cellular commingling, some tumors comprise both astrocytic and oligodendroglial components. "It’s often difficult to say whether a particular tumor is one type or the other," Dr. Jenkins says. "It’s basically a spectrum."
"Certainly, I’ve seen cases that have circulated among the best neuropathologists in the country, and you get diagnoses that run the gamut," says Richard Prayson, MD, staff pathologist, Cleveland Clinic. "There’s a significant number of cases where it’s sort of in the eye of the beholder."
Others are less charitable. Says one observer, who demanded anonymity: "I’m often astounded at some of the cases that get called oligodendroglioma by well-known neuropathologists. To me, there’s just nothing remotely oligodendroglioma about it."
Harsh, yes—but hardly a lone voice in the wilderness.
"Once you start looking for an oligodendroglioma quality, it’s hard to stop," cautions Dr. Burger, whose article "What is an oligodendroglioma?" recently appeared in Brain Pathology (2002;12:257-259). "You look harder, and you spend more time, and you accept that the cell is round, that there’s a little halo around it. And you say, ’Aha! Mixed glioma.’ It’s a hard habit to shake. And pretty soon you rarely find yourself diagnosing a pure astrocytoma. It all ends up oligodendroglioma or mixed. And that’s what the patient wants to hear, and what the clinician wants to hear. If we call it oligodendroglioma, everybody’s happy."
Patients are hardly martinets in these matters. Reports Dr. Burger: "Some people are so anxious to get the diagnosis, I’ll say to them, ’If you want, I can send these slides to someone who has looser criteria, and we can easily get a diagnosis of oligo or mixed.’ And they’ll do it, because it makes them feel better, and they can get the treatment. And, maybe they’re right."
In reality, says Dr. Burger, oligodendrogliomas are relatively uncommon lesions. Ditto mixed gliomas. "In my opinion, we’re seeing an overdiagnosis of these lesions. The criteria are loosened. It’s like an epidemic of these things, which I think are purely artifact. Most of these things are not oligodendrogliomas, in my opinion, and they’re probably not mixed gliomas, but no one knows exactly how to define them."
Mixed gliomas offer their own maddening quirks. Ideally, says Dr. Burger, these tumors would exhibit distinct areas of classic astrocytoma and oligodendroglioma characteristics. "The problem is, very few tumors look like that." Instead, tumors defined as mixed usually are a sloppy stew of both types of features.
Further complicating matters, some genetically defined oligodendrogliomas acquire astrocytic features—more cytoplasm, for example—as they become more malignant. "Then you end up in a very difficult situation, because these tumors really do look like both types of tumors," Dr. Burger says. "And genetic studies of these tumors have been more confusing than anything."
Mixed gliomas are problematic, no doubt about it. "They’re overcalled—I do it knowingly myself," Dr. Burger says. "You’ll find something that you really just can’t call either oligodendroglioma or astrocytoma. So you call it mixed. It’s an out. A lot of people don’t admit it, but that’s what it is."
This is no mere back-office squabble among pathologists. Clinicians are taking note as well. "There certainly has been a tendency in the last five years or so for pathologists to overcall the oligodendroglioma phenotype," says Dr. Barnett, who quickly adds that their intentions are good: "They don’t want to deny a patient who might be sensitive to chemotherapy."
Dr. Cairncross recalls that when he first became interested in anaplastic oligodendroglioma, some 15 years ago, very few tumors were classified as such; those that were always responded to treatment. "Now, when we go back and look at them genetically, they always have 1p and 19q loss."
Today, he notes wryly, anaplastic oligodendrogliomas—or tumors classified as such—respond only half the time to chemotherapy. "There’s been some liberalization of the histopathologic diagnosis criteria, and about half of them have 1p/19q loss."
Dr. Cairncross’ response to the more inclusive approach: "You can call it what you like, but you better dump it through the gene analyzer."
At an informal meeting he attended this past summer, he was treated to an eye-opening demonstration using tissue samples that had been banked 20 years earlier. The basic breakdown of the original diagnoses was one-third astrocytoma, one-third oligodendroglioma, and one-third mixed glioma. The samples were then reviewed by a contemporary pathologist, whose diagnoses shifted considerably: Most of the tumors originally called astrocytoma were now called mixed, and the group called oligodendroglioma expanded. When gene expression profiling was done and the results compared to the two eras of classification, the better match was with the 20-year-old diagnoses.
"I don’t know what the truth is, but I think we can bring greater clinical relevance by genetic diagnosis," Dr. Cairncross says.
There’s also the possibility—OK, certainty—that genetic testing will cause added confusion, at least early on. But there’s no going back. "The molecular techniques have become very powerful and are rapidly evolving to the point where they’re going to become standard," says Dr. Burger.
"I personally believe that in the not-too-distant future, we really won’t be paying very much attention at all to the histology, in terms of whether it’s an oligo or a mixed or an astrocytoma," says Dr. Barnett. "Rather, we’ll be looking at the molecular signature of the tumor."
But few are prepared to cashier morphology just yet, if ever. Sydney Finkelstein, MD, associate professor of pathology at the University of Pittsburgh Medical Center, suggests that oligodendroglioma will be the first area that will convert to nearly a pure molecular analysis of tumors. "It’s actually going that route at this time," he says. "But morphology will orient each case, to help select the areas of the tumor that are important."
Because of the abundance of genetic diversity in gliomas, each area of a tumor may have a different profile. This is frequently the case in patients who have survived for a period of time and undergone re-biopsy or re-resection, says Dr. Wiley. "You’ll see that different clones pop up." Chances are, then, that there will be a fair amount of noise in the genetic analyses—noise that is assuaged by genetic testing.
"You wouldn’t think morphology would add anything" to a sophisticated genetic fingerprint, says Dr. Wiley. "A small number of pathologists have already made that leap and said the heck with morphology. Unfortunately, it’s a little early to do that."
Ultimately, the sharp boundaries of morphology and genetics may simply erode. "Maybe our job is to say, ’This is an infiltrating lesion. It could be either oligodendroglioma or astrocytoma,’ and then we triage those for genetic testing," says Dr. Burger.
Most are leery of relying solely on 1p/19q status for diagnoses. One of the biggest problems, says Dr. Perry, is no specific markers exist to identify oligodendrogliomas histochemically. "Some would argue that the 1p/19q genetic test is the closest thing to supporting a tumor as being oligodendroglioma. And in that case, it would be a very useful diagnostic tool.
"There’s still some question about that, though, because anywhere from 20 to 50 percent of what people think are oligodendroglioma don’t have this deletion," he continues. "So while it’s probably very specific, I’m not sure that the sensitivity is high enough to use it by itself as a diagnostic test."
"I would be cautious," agrees Dr. Prayson, though he suggests 1p/19q may have diagnostic utility in differential diagnoses involving an oligodendroglioma look-alike—for example, clear cell ependymoma, neurocytoma, or dysembryoplastic neuroepithelial tumor. In reviewing DNET cases at their institution, pathologists at the Cleveland Clinic found that in the group they looked at, all were 1p-intact. "So in cases like that, if it’s 1p-deleted, it would speak more in favor of oligodendroglioma. But if it’s 1p-intact, it’s obviously of no help," Dr. Prayson says, since a subset of oligodendrogliomas are 1p-intact; likewise, a small percentage of astrocytomas and mixed gliomas demonstrate 1p deletion.
At UPMC, pathologists do a battery of loss of heterozygosity analyses, including 1p/19q, for yet another diagnostic purpose: to help determine whether minute, stereotactic brain biopsies represent a gliosis or merely a reactive condition. "This hasn’t been well-proven yet, but the assumption has been, at least in the conditions that we’ve looked at, that gliosis does not ever lead to a loss of heterozygosity for the genes that we look at," says Dr. Wiley. "And therefore, if we determine from our panel that a gene has been lost, in a sort of ambiguous biopsy, it helps guide us toward diagnosing a neoplastic disorder."
Clearly 1p/19q testing has its limitations from a diagnostic perspective. Dr. Jenkins sounds another note of caution, this one regarding its prognostic value. Warning that the distinction is a subtle one, he says, "Patients with oligodendrogliomas with deletions may have a better prognosis independent of therapy; they may also respond better to chemotherapy. Those are two different things. Predicting a response to chemotherapy is different than predicting whether a patient will live longer."
Though the data from Dr. Cairncross and his colleagues are very convincing, "the prognostic and predictive values have only been shown retrospectively," Dr. Jenkins says. At Mayo, he and his colleagues have also shown, retrospectively, that patients with deletions have a better prognosis. "But we haven’t been able to show that there’s any association with any kind of therapy—because all the patients were treated."
Two ongoing trials, one by the RTOG (Radiation Therapy Oncology Group) and one by the EORTC (European Organisation for Research and Treatment of Cancer), randomized patients with high-grade oligodendrogliomas to PCV and radiation therapy or to radiation therapy alone; as an adjunct to both trials, researchers are also looking at 1p and 19q status. Dr. Jenkins, who is doing this testing for the RTOG study, says that taken together, the studies may have enough power to determine the prognostic and predictive worth of 1p/19q. "We’re going to need three or four years, though, before the followup has accrued long enough to know what the response to therapy was, and what the overall survival is."
Though he finds the retrospective data compelling, Dr. Barnett agrees the entwined notions of prognosis and chemoreceptiveness are difficult to unravel, particularly since virtually everyone with an oligodendroglial tumor receives treatment. "What one really looks at, when one retrospectively looks at the 1p data, is how their prognosis was after treatment. I don’t think there’s much information on 1p in the absence of treatment."
A final word of caution comes from his colleague Dr. Staugaitis, who calls 1p/19q status "an aid in treatment planning," rather than a therapeutic marker. Treatment of brain tumors is highly individualized, based not only on morphologic and radiologic findings, but on the patient’s age and health and the location of the tumor, among other criteria. "You can’t imagine all the decisions neuro-oncologists and neurosurgeons are faced with," she says. Thus, while additional, genotypic information is helpful, it alone can’t point to the best therapy.
In light of the preceding dissonance, a messy discourse over the best method for testing for 1p/19q loss seems inevitable.
Not to disappoint, but it’s not going to happen. At least not anytime soon. Raging debates about FISH versus competing methods will remain, for now, with others, such as the HER2 crowd. Those testing for 1p/19q are quite amiable in their discussions, agreeing there is no regnant methodology and that all seem to be highly concordant. "Most people’s approaches have been based on what they’re most comfortable with in their own laboratories," says Dr. Louis.
In many of the 1p/19q studies, researchers used loss of heterozygosity, a method employed by Dr. Louis in his research as well as clinically, though his laboratory is now shifting to FISH. "We have done a lot of LOH analysis in the past, and the clinical laboratory at our hospital has done a lot of gel-based approaches," he says. "So we naturally went with an LOH-based approach."
LOH requires extraction of tissue sections from DNA and a sample—usually blood leukocytes, sometimes nontumor tissue—for comparing polymorphisms. It also requires a gel or capillary electrophoresis system. "It’s a lot of upfront work," Dr. Louis concedes. "Once you get the results, though, they’re very, very easy to read."
Oddly enough, obtaining blood samples from patients has been one of the stumbling blocks for labs. "It adds a layer of complexity, just for practical reasons," Dr. Louis says.
Another drawback with LOH is that the presence of normal, background brain cells interferes with reading the chromosomal losses. Because glial brain tumors are notoriously infiltrative, such contamination is common and can impede LOH or other quantitative methodologies.
FISH, which preserves the architecture of the tissue, sidesteps that problem, thought not completely. "You have to take into account, when you’re doing your spot counts, the number of signal counts for FISH, how many cells are normal and how many aren’t. That can be quite, quite tricky," Dr. Louis says.
Dr. Jenkins uses FISH, in part, he says, because Mayo has a highly experienced FISH lab. "We’re technically competent."
With several collaborations, his lab has shown good correlation between LOH and FISH, as well as comparative genomic hybridization and quantitative microsatellite analysis, in studies published in Oncogene (Smith JS, et al. 1999; 18: 4144-4152) and the American Journal of Pathology (Nigro JM, et al. 2000:158:1253-1262). (Drs. Perry, Burger, and Ken Aldape were also authors on these papers.)
In high-grade oligodendrogliomas, 1p/19q deletions usually involve the entire arm; only rarely do mitotic recombinations occur. In those few cases, LOH might pick up one or two additional cases missed by FISH. "But LOH might miss 10 cases out of 100 because of normal cell contamination," Dr. Jenkins says. Interestingly, Dr. Staugaitis and her colleagues at the Cleveland Clinic have found that partial chromosomal loss is more common in high-grade astrocytomas—anaplastic astrocytoma and glioblastoma multiforme. "If we looked at five different markers on the short arm of chromosome 1, for example, we would see maybe one would be lost, either at the telomere or at an interstitial location," she says.
At Cleveland Clinic, the laboratory has the capacity to perform FISH and PCR-based techniques. "We decided FISH is the preferred technique for several reasons," says Dr. Staugaitis. As Dr. Louis noted, sample coordination is a logistical challenge. Moreover, extracting the DNA required for LOH can be difficult on small samples. "Whereas, on the FISH, you’re looking at several high-power fields in a microscope, so the total amount of tissue you need is much, much smaller," says Dr. Staugaitis. Like Dr. Jenkins, she and her colleagues also found, during their initial validation, that FISH and LOH gave concordant results in the vast majority of cases they performed using both techniques. The costs of both tests were also comparable.
Occasionally LOH is used as a confirmatory test. "The best example is if the specimen is very cellular, which makes it difficult to get an accurate count of the FISH signals by looking in the microscope," Dr. Staugaitis says. LOH results, on the other hand, are hardly ever ambiguous.
LOH is also used as a backup test from time to time. The FISH technique doesn’t work well on tissue processed with nonformalin fixatives—"Or at least we haven’t been able to successfully get it to work," says Dr. Prayson. "In these situations we might use the PCR technique."
Another FISH limitation: It’s a targeted assay, notes UPMC’s Dr. Wiley. "And you have to use the entire section for one particular analysis. You could build all your controls into that one slide, but you only get to ask one question." Most flow cytometry-type lasers, and even the confocal microscopy laser, can look at only four or five types of markers, he estimates. With PCR-based analysis, on the other hand, amplification of all the DNA allows users to look at a wider variety of potential genes.
In tumors that are very aneuploid or polyploid, FISH interpretation becomes complicated, requiring users to assess marker ratios, says Dr. Perry, whose own lab at Washington University uses FISH. Aneuploidy does not confound LOH, however, because the method focuses on allelic polymorphisms. Most oligodendro-gli-omas are diploid, he concedes. "But we have identified over time a small percentage that become aneuploid or polyploid, and it’s even worse in the astrocytomas." His lab uses reference probes to skirt this problem.
FISH more readily addresses a different complication: the histologic heterogeneity of brain tumor specimens. "We’ve found that correlating the area that we analyze with areas that are tumor, histologically, is critical," Dr. Staugaitis says. "So every test we do, we look at the H&E-stained slide, we circle an area for analysis—tumor that is representative—and we do the analysis on that." In PCR-based methods, microdissection can capture different areas of tissue from a slide. But that may limit the amount of tissue from which to extract the needed DNA, she says.
Unrolling the heterogeneity carpet even further, Dr. Staugaitis notes it’s becoming clear that neoplasms are evolving clones, with genetic heterogeneity occurring within different sections of the neoplasm. "I think this is the next area people will pursue with genetic analysis," she predicts.
Another method, quantitative microsatellite analysis, or QUMA, snags advantages from both LOH and FISH. Like LOH, it allows users to look at many markers simultaneously. Like FISH, it doesn’t require normal tissue or blood for comparison.
Ken Aldape, MD, associate professor of pathology at MD Anderson Cancer Center, Houston, uses QUMA, which is based on real-time PCR. The method is a straightforward way to detect change in copy number by measuring the rate of PCR product accumulation during the course of reaction. "It’s a quick assay—about two hours," Dr. Aldape says. "You can detect multiple loci all at once using a 96-well plate. And we optimized it for paraffin."
This method, he notes, is not necessarily better than LOH or FISH. "It’s different."
In certain cases, QUMA is at a disadvantage. For example, tiny amounts of tissue, such as from a needle biopsy, may not yield enough DNA to perform the assay. Second, the DNA must be from pure tumor—and the so-called find-and-grind method used in the QUMA technique obliterates the distinction between normal and tumor cells. Dr. Aldape considers this to be a minor limitation, however. "You look at the tissue before you grind it up, so you just have to make sure it’s pure tumor."
Microdissection is at the heart of genetic testing of gliomas at UPMC, enabling neuropathologists to select three to five of the best areas of the tumor for molecular analysis. Each section is microdissected out, then undergoes a detailed molecular analysis to assess glioma growth, development, and progression.
"We try to characterize each site for as many mutational changes as we can," says Dr. Finkelstein. He and his colleagues test for 15 different mutational markers; their goal is to boost that number to 30. "Even with 15, we get a very good feel for exactly how much acquired mutational change has occurred at these sites," he says. In addition to testing for 1p/19q loss, they test for 9p21, the location for the p16 gene, which, when mutated, is associated with aggressive biological behavior; 10q23, where PTEN is located; and 17p13, where the TP53 gene is located. They use approximately three markers per site. Measuring allelic loss is done by high-throughput PCR analysis.
As with the other methods, Dr. Finkelstein’s approach has its limitations as well as a certain edge over other methods. Normal cells can get mixed in with tumor cells if the pathologist isn’t careful, leading to spurious results. On the other hand, "We haven’t even begun to push the technology we use," he says—which could gain added importance as the list of potential markers grows.
The discussion could continue, and no doubt will—at least among laboratorians. So we’ll let Dr. Cairncross end it for now. "As a clinician, I would just say, ’Do it any way you like. Just tell me whether 1p/19q is there or not, and that’s good enough for me."
Not surprisingly, the 1p/19q testing motif has endless themes and variations.
1p deletions are seen primarily in oligodendroglial tumors. "Some people argue that since you almost always have 19q deletion with the 1p, that you can test just for 1p and not worry about 19q," says Dr. Perry. "But more studies need to be done on that. We’ve had some preliminary evidence that patients with 19q deletions alone are still doing relatively well. So it could be that either one predicts the better outcome. But the stronger data right now is on the 1p deletion or combined 1p /19q deletion."
At Mayo, Dr. Jenkins and colleagues are looking into whether the 19q deletion alone is significant beyond oligodendrogliomas—it occurs in all glioma subtypes: astrocytoma, ependymoma, and mixed oligoastrocytoma as well as oligodendroglioma. "That is one of my basic research goals: to identify the genes on 19 and see whether they could also be targets for therapeutic approaches," he says. Though 1p and 19q are the short arms of their respective chromosomes, each contains hundreds of genes—any number of which might be relevant.
In the coming years, an argosy of helpful genetic markers will likely arrive. "I think we’re just starting to learn what most of the common alterations are, and we’ll probably fine-tune the panel of genes that we want to look at in the future," says Dr. Perry.
One of the most common alterations seen in gliomas is amplification of the EGFR gene—it’s rearranged in 40 percent of primary glioblastomas. "It’s not ready for prime time, but there are several tyrosine kinase inhibitors designed for EGFR in particular, and there are several ongoing trials to determine whether patients with gliomas are responders," Dr. Jenkins says. "If that really pans out, that’s going to be another test we’ll be doing."
Progression-associated markers may include PTEN and p16 gene deletions as well as EGFR gene amplification. At Dr. Perry’s lab, all are fair game, though not in every case. "We’ve used them in cases where our differential diagnosis includes other forms of high-grade glioma, including small-cell glioblastoma, which looks a lot like oligodendroglioma." Small-cell glioblastomas tend not to have 1p or 19q deletions, but about 70 percent have EGFR amplification.
The folks at Cleveland Clinic test for EGFR routinely. They’ve also begun to look at array-based CGH in the hopes of identifying new genes that are amplified or deleted, which might serve as new therapeutic or prognostic markers. "We’re starting with anaplastic oligos, because we have a benchmark in the 1p deletion," says Raymond Tubbs, DO, chairman of the Department of Clinical Pathology. "Then we can use that benchmark to extrapolate the comparative genomic hybridization to FISH assays, which will allow us to study many more tumors and identify many more markers."
How big might the picture get? "It’s a very important question," says Dr. Staugaitis. "I think more markers will be examined and discarded as not being useful, than the number of markers that will be retained as tests. It’s also possible a combination of a few markers may be predictive."
No matter how this evolves, it’s clear that neurosurgeons and neuro-oncologists, who’ve pushed this testing into the everyday realm, will keep up the pressure.
"This is truly a breakthrough in classification, and it translates directly to how we manage brain tumors," says Dr. Barnett. "Hopefully it’s the beginning of a change that will affect everything."
Oligodendrogliomas are just a warm-up, agrees Dr. Cairncross. "We’re getting ready for the real game—glioblastoma. This is just a dress rehearsal."
Karen Titus is CAP TODAY contributing editor and co-managing editor.