College of American Pathologists
Printable Version

  With AML genetic profiling, it takes all kinds


CAP Today




Genetic mutations—primarily FLT3, CEBPA, and NPM1—have been part of the acute myeloid leukemia picture since 2008, says Dr. Gail Vance, whose lab at Indiana University tests for the mutations in cases of AML with normal chromosomes. A recent ECOG trial found those three to have some company.
Genetic mutations—primarily FLT3, CEBPA, and NPM1—have been part of the acute myeloid leukemia picture since 2008, says Dr. Gail Vance, whose lab at Indiana University tests for the mutations in cases of AML with normal chromosomes. A recent ECOG trial found those three to have some company.

June 2012
Feature Story

Karen Titus

If only genetic profiling for acute myeloid leukemia were like a horse race. No matter how close, no matter how many competitors, there would be only one winner, even in a photo finish—one champion mutation that would provide all prognostic and treatment answers.

If only. In recent years it’s been more like a game of Go Fish. Researchers find a marker, describe it in a publication, and attempt a multivariate analysis to compare its significance either to other genetic markers or to more traditional markers, such as cytogenetics or white counts.

The game has recently become more sophisticated, however, with the publication of an Eastern Cooperative Oncology Group, or ECOG, study, published in the New England Journal of Medicine (Patel JP, et al. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. 2012;366:1079–1089). The phase three clinical trial, E1900, provides the most nuanced view of AML genetic profiles to date, with researchers performing a mutational analysis of 18 genes in 398 AML patients under age 60.

The title says it all, says Mark Litzow, MD, a study coauthor and cochair of the ECOG leukemia committee. “‘Integrated genetic profiling’—trying to integrate the results of multiple markers and trying to relate them to one another. And helping us as clinicians understand, OK, we’ve got these results—what do we do with them?” says Dr. Litzow, professor of medicine, Mayo Clinic, and chair of Mayo’s myeloid disease group.

“I’m really happy to see this paper,” says Rita Braziel, MD, who was not part of the ECOG study. “Everyone I’ve talked to is happy to see this paper.” While gene expression profiling in AML isn’t new, “We haven’t had this type of extensive genetic profiling by mass array methods. I was excited to see it,” says Dr. Braziel, professor, Department of Pathology, and director of hematopathology, Oregon Health & Sciences University, Portland, Ore.

This panoptic view has produced a few surprises for the study’s authors, among them:

  • A whopping 97.3 percent of AML patients in the study had at least one somatic mutation.

  • The well-known markers FLT3, NPM1, and CEBPA, are no longer the winning trifecta.

  • In certain patients, high-dose daunorubicin improves patient survival.

In short, the AML picture just got clearer and more complicated. It also raises the question, How far will laboratories go in identifying AML mutations? Like Moses gazing at the Promised Land, everyone can see the ultimate destination—but there’s no guarantee of crossing over.

The NEJM paper stands at the crossroads of AML, providing a crystalline look at where the field has been, current standard of care, and where things likely are headed. Writ large, it’s also the story of how knowledge of cancer heterogeneity is sinking in on a clinical level: So this is what we’re really dealing with.

AML is genetically diverse, as are all tumors, notes Rhett Ketterling, MD, a study coauthor. While physicians group them together based on their morphology, it’s well known that there are any number of different genetic pathways that can be disrupted or activated to generate the same malignancy, says Dr. Ketterling, pathologist and codirector, Cytogenetics Laboratory, Mayo Clinic.

By now even a bright 5th grader—and pathologists are far smarter than a 5th grader—might know this. The study, in that sense, isn’t revolutionary. But Dr. Ketterling calls it “further evidence proving what most of us knew: This is complicated.”

Until recently, cytogenetic abnormalities were the leading prognostic indicators for AML. Early studies suggested it was possible to divide patients into three basic groups: those with favorable cytogenetics, those with unfavorable cytogenetics, and those with intermediate cytogenetics. Those in the latter group were “somewhere in the middle,” says Dr. Ketterling, most with normal cytogenetics. About one third of AML patients fell into this group, making it a good place to start looking for molecular markers. “It’s not that they didn’t have AML—we just didn’t know what the genetic abnormalities were,” Dr. Ketterling says of this bit of medical predestination. “There was nothing wrong with their karyotype; there was nothing from FISH or other genetic evaluation that we could identify.”

By 2008, genetic abnormalities became part of the picture, primarily FLT3, CEBPA, and NPM1, says Gail H. Vance, MD, a study coauthor and professor of medical and molecular genetics, Indiana University School of Medicine.

But those three now face additional competition, as the NEJM study shows. Dr. Vance, who was involved in E1900’s extensive cytogenetics central review, says she was surprised by the frequency of the DNMT3A, or methylation, gene. Its overall frequency was 23 percent. (See Fig.1, (PDF, 606 KB), Mutational Complexity of AML). FLT3 mutations had an overall frequency of 37 percent (ITD, 30 percent; TKD, seven percent), and NPM1, 29 percent. CEBPA had a relatively low frequency of nine percent.

DNMT3A is a newcomer,” Dr. Vance explains. “I knew about FLT3, I knew about NPM1, I knew about CEBPA. Mutations of these three genes were fairly well established, even in this new age of molecular mutation evaluation.” In patients with normal cytogenetics, for example, an FLT3 mutation dictated an overall poorer prognosis. An NPM1 mutation without FLT3 meant a good prognosis. And CEBPA, be it one mutation or homologous mutations, was a good prognosis.

“And then this gene, DNMT3A, arrives on the scene,” says Dr. Vance. It’s not the only newcomer. Other mutations with relatively high frequency (and studied as part of E1900) include TET2, WT1, and IDH2 (each with eight percent overall frequency), and IDH1 (seven percent).

“For lack of an eloquent way of framing this, there’s now a bucket of mutations,” says Dr. Vance. Physicians had their pet mutations, she says, but until the E1900 study, there had been no clear way to relate these mutations to one another, either by frequency or by outcome.

The study is both welcome and unsettling. “We had just gotten comfortable with CEBPA and NPM1 and FLT3,” Dr. Vance says. “Whether they be wild type or mutational, we were all convinced we understood gene test results would frame up the possible outcome, the overall survival for this patient.”

“One of the things that I found so interesting,” she continues, “was that we had thought AML normal chromosomes and NPM1 mutation had a better outcome. But then the data shows that mutational status of NPM1 alone didn’t define that favorable subset—there actually had to be different alleles, or additional mutations, that stratified it further.”

Dr. Vance likes another aspect of the study: the finding that molecular and cytogenetic mutations “held” even if a patient relapsed. So an overall favorable prognosis in a patient meant he or she tended to do better if they relapsed; an overall unfavorable prognosis meant they tended to do poorly, regardless of whether they were treated with high- or low-dose daunorubicin.

Dr. Ketterling professes to being startled by the high percentage of patients with identifiable mutations. That 97.3 percent figure shows the power of the genes selected for the study. “Very impressive,” he says. “That’s a surprisingly high number.”

He was also intrigued by findings regarding MLL, a relatively common translocation gene in AML. Historically this has been associated with a relatively unfavorable prognosis, he says, but the study suggests it actually confers an improved rate of survival when patients were given high-dose induction chemotherapy.

“In the end,” says Dr. Ketterling, “it’s getting at what trumps what. There are certain genetic subgroups that are mutually exclusive, as you would anticipate, but it also points out how genetically diverse this group is.”

The beauty of the study can be summed up in Fig.1 (PDF, 606 KB) . While the sheer number of markers would seem to demand a spreadsheet, the researchers used a more vivid tool to bring the results to life—a Circos diagram. As it colorfully demonstrates—Dr. Ketterling wryly notes, “I know it’s spelled differently, but it looks like a big circus tent”—not every patient has every marker, nor are markers distributed randomly. He says it beautifully shows the complexity and interrelationships of the various genes. “You could learn a lot just by studying this.”

Seeing the combinations is crucial, Dr. Litzow agrees. Patients who are NPM1 positive but FLT3 negative, for example, tend to fare well, and don’t necessarily need a bone marrow transplant. But those with a poorer prognosis, who are likely to relapse even if they go into remission at first, might be good candidates to receive a transplant when they’re in remission. This NPM1/FLT3 combo was an important observation of an earlier study (Schlenk RF, et al. N Engl J Med. 2008;358:1909–1918); this current study confirms it.

In another example, the thick reddish band stretching from the upper right to the bottom left of the diagram shows the reach of DNMT3A; patients with this mutation seemed to do better with the higher dose of daunorubicin, Dr. Litzow says, which he found a bit surprising. And, like Dr. Ketterling, he was a bit taken aback by the 97.3 percent figure. “It’s remarkable they found an alteration in almost all the patients,” he says. “In some ways, it’s actually exciting. We’ve always known they had abnormalities. It’s just that cytogenetics was too crude to pick that up.”

Knowing that nearly every patient has an abnormality, and being able to define it, opens the door to profiling every patient, Dr. Litzow continues, and brings physicians another step closer to personalized medicine.

Being a clinician, he brings a clinician’s concerns to the findings. More markers will help categorize patients with AML, yes, but just as importantly they might spur development of additional drugs to act against them. “We desperately need new agents,” he says, noting that AML treatment has barely budged in 30 to 40 years.

An accompanying editorial (Godley LA. N Engl J Med. 366;12:1152–1153) suggests that turnaround time for AML genetic profiles would have to be quick—“within the first few days of presentation,” the author writes.

Yes and no, Dr. Litzow says. It depends on available treatments. Obviously, if there’s a drug that’s specific to a certain subtype, physicians will need to know about the subtype ASAP. But that’s not the state of AML treatments right now. Moreover, he says, transplant decisions are not made at diagnosis.

What about high-dose daunorubicin? Because it hasn’t been shown to be detrimental, most physicians already give high doses to all their AML patients, Dr. Litzow says.

But once more drugs are available, physicians will indeed need that information swiftly. It’s a technological hurdle, but a low one, Dr. Litzow suggests.

Some clinicians will approach these findings with cautious enthusiasm, while others will want to launch gene panels sooner rather than later. In the meantime, laboratories will be scrambling to answer the seemingly endless stream of questions any good study raises.

“How do we interpret the results?” asks Dr. Vance. “What if we have two unfavorable markers and one favorable marker? What if you have competing mutations? Do you split them in half? What do you do?”

What is her institution considering? “We’re having multiple conversations,” she says, as she and her colleagues try to decide what genes to evaluate—some? all? just the predominant ones?—and whether to do it in a research setting or clinically.

Another challenge is sample availability. The time between diagnosis and treatment can be short as a Puritan’s haircut. In the interim come the finer points of lab analysis. A karyotype takes several days to generate and evaluate, Dr. Ketterling says, by which time a patient has usually undergone a first chemotherapy treatment. “These patients are very ill,” he says. But with the sample unavailable, labs need to have a means for saving specimen and extracting DNA, RNA, or both, for molecular evaluation. Such sample acquisition and availability are not classically part of the algorithms in genetics laboratories or hematopathology areas, he says, although Mayo has addressed this. “One of the first things we do now is extract the DNA and have that available for the genetic test in case we need it,” says Dr. Ketterling.

Not that the methods described in E1900 are speeding into clinical practice.

“It’s very exciting to be involved in these studies,” says Dr. Ketterling. “But it’s also frustrating to know that on a practical basis, this is not going to happen in my clinical laboratory anytime soon. On a day-to-day basis, I still sign out karyotypes. Because that’s where we are, in reality.” He says it’s a leap to go from standard cytogenetics to the methodologies employed in the NEJM study. “They did a beautiful job, but they were comparing [their work] to what we do in a standard cytogenetics laboratory—they had to compare it to the archaic gold standard,” he says.

The informatics puzzle remains just that—a puzzle. Physicians’ understanding of what is normal in the genome, and what is abnormal, is still evolving. That’s one of the most common refrains in medicine, in fact. Like a Phillip Glass piece, the information churns relentlessly, with no end in sight. “It’s just too much,” says Dr. Ketterling. AML is one hematologic malignancy, and a relatively uncommon one at that—yet AML alone can seem overwhelming, he says. “A lot of what I do is educate. And I educate smart people—the clinicians at Mayo are bright. But people can’t keep up.”

Dr. Vance (standing), director of the IU Division of Diagnostic Genomics, with Shaochun Bai, PhD (left), director of the molecular diagnostic laboratory, and Lisa Lowe Wright, lab manager of the cytogenetics labs, which submit samples to ECOG for review.
Dr. Vance (standing), director of the IU Division of Diagnostic Genomics, with Shaochun Bai, PhD (left), director of the molecular diagnostic laboratory, and Lisa Lowe Wright, lab manager of the cytogenetics labs, which submit samples to ECOG for review.

But they keep trying. For proof, one need look no further than Dr. Braziel’s group at OHSU. A member of the Lymphoma/Leukemia Molecular Profiling Project and the Southwest Oncology Group, Dr. Braziel and colleagues at OHSU have incorporated a high-throughput screening method for AML mutational analysis into the routine diagnostic workup of all new AML patients at OHSU. They’re using the Sequenom MassArray system on a 31-gene panel and have tested about 150 patients over the past three years. Results for each patient are reported in the clinical chart.

Their work (Human Pathology, in press) is similar to that done by the ECOG group in terms of thought process, she says, with a slightly different technology and gene mix than those presented in the NEJM paper. “We thought we were going to be first, but then this paper came out, which is like our paper on steroids,” she says. However, it does offer a look at how AML integrated genetic profiling can make its way into the clinical setting, step by precarious step.

The challenges of doing broad AML workups are not insignificant, but most of the hurdles will eventually fade as technology improves, she predicts. Her biggest concerns lie outside the lab.

“Guess how many cases we’ve been reimbursed for?” she asks.

That would be zero.

“We can handle the technology,” she says. “The problem is, the insurers don’t understand it, and the FDA and CLIA don’t yet know how to QC these things.”

Gene profiling will not be a black box situation, Dr. Braziel says. While the technology may become more stable over time, the genetic features of interest will change constantly. Those who work in the field see it happen often enough: One gene drops off, another gene comes onboard. Labs need to be nimble—and they are. The FDA has been another matter. “They love black box kits. But it costs a fortune to build an FDA-approved kit, and it will likely be out of date by the time it’s approved,” Dr. Braziel says.

The equipment and testing are expensive, she continues, while the experience needed to interpret and validate results lies well beyond the ken of most community hospitals. That may ultimately change, she says, adding, “I do not anticipate that that will be the case in my professional lifetime. There’s no way a community hospital is going to dump hundreds or thousands or millions of dollars into something that’s not even being reimbursed.” A lab would need to see a minimum of 50, or maybe even 100, new AML cases a year to make it a worthwhile investment, she says.

She likes the partnership she sees between academic hematopathologists and those at community hospitals in her region. They meet regularly, she says, and the latter send their AML cases to OHSU for gene mutation analysis. “Our competitors are not one another,” she says. “Our competitors are the national labs, which have far greater resources. So we have to work together to offer patients the best testing at the best price.

“However, for us right now, that’s translating into zero dollars,” she adds with a laugh. OHSU’s work has been largely supported by research funds. “I don’t know how long we’ll be able to keep doing it.”

The other kill-joy could be patents. For now, the only patent issue concerns FLT3: “Invivoscribe Technologies [says the company’s Web site] holds the exclusive license to develop, manufacture, use, and sell testing reagents for FLT3/ITD mutation detection.”

In other words, “You can’t do FLT3 anymore—we have to send it out to a central lab in California,” says Dr. Vance, adding a sentence sure to stop most labs in their tracks: “The lawyers have called us.”

No one—except likely the folks at Invivoscribe—seems happy about this, and there’s concern that more genes could fall prey to patents in the future.

Nonetheless, an air of inevitability hovers over this work, above the practical considerations and frustrations and questions, and, yes, anger. Integrated genetic profiling for AML, as well as other leukemias, is bound to be more than a bright idea.

“We’re very excited about this data,” Dr. Litzow says. “I see this paper as futuristic. It’s not something that people are going to implement tomorrow. It complicates things, but it’s the wave of the future.”

Karen Titus is CAP TODAY contributing editor and co-managing editor.