AML genome answers no longer MIA

BETHESDA, Md.—The Cancer Genome Atlas (TCGA) ResearchNetwork, a massive effort to increase molecular knowledge of cancer throughidentification of the genetic mutations that lead to its various subtypes, hasannounced the release of a detailed classification of the genomic alterationsand mutations that lead to acute myeloid leukemia (AML). TCGA is jointlysupported and managed by the National Human Genome Research Institute (NHGRI)and the National Cancer Institute, both of which are part of the U.S. NationalInstitutes of Health.

AML is the most common acute form of adult leukemia, andoccurs when immature white blood cells do not mature, but instead collect inthe bone marrow, reducing the production of healthy blood cells and leading toanemia, abnormal bleeding, infections and, if untreated, death.

The researchers' work, which appeared online May 1 in the New England Journal of Medicine, resultsfrom their study of the tumor genomes of specimens from 200 adult cases ofspontaneously occurring, newly diagnosed AML, representing all known subtypesof the cancer (in roughly the same proportion as is seen in the generalpopulation). Each patient's AML tumor genome was compared to a normal genomederived from their skin sample. Fifty of the 200 samples were analyzed withwhole-genome sequencing, while the protein-coding regions of the genome wereanalyzed in the other samples. In each case, the researchers also usedsequencing to examine RNA as well.

"Rather than just random snapshots about individualpatients, this study provides a more detailed look at the aberrant genomes ofAML than we have ever had before," NHGRI Director Dr. Eric D. Greensaid in a press release. "It has the potential to open up new directions in AMLresearch, and perhaps, in the design of new therapeutics, its impact could befelt in the near future."

The results of the study, in addition to being informative,were also surprising. The size of the sample allowed researchers to predictthat they had identified nearly all mutations that occur in at least 5 percentof AML patients—and that in general, there are comparatively few mutations,making AML tumors some of the least mutated among adult cancers. Solid tumorsseen in breast, lung or pancreatic cancer generally present with hundreds ofgenetic mutations, but the average number for each AML genome in the TCGA'swork was 13.

The team found more than 1,600 genes that were mutated atleast once in the sampled tumors, and classified the recurrently mutated genesinto nine categories based on the involved pathways and the genes' functions.Of the observed mutated genes, epigenetic modifiers represented the mostfrequently mutated class of genes.

Though relatively few recurrently mutated genes were found,the discovery of frequent alterations in genes that control gene expressionstands to help highlight new potential drug targets and disease markers.Abnormal chromosome arrangement and gene fusions are often used to providediagnoses and prognosis for AML patients, and the study not only revealed thatnearly half of all the examined samples displayed gene fusions, it also drewattention to many fusions that had not previously been identified.

"Although gene fusions resulting from chromosomalrearrangements have been studied in AML for many years, this study uncoveredmany previously unknown, some which could be new targets," says BradOzenberger, Ph.D., TCGA program director. "An important message from this studyand other TCGA reports is that each tumor is different. It is now quite clearthat drug regimens for cancer will increasingly be customized for theparticular patient, based on the specific spectrum of mutations in his/hercancer."

Previously, it had been known that signaling gene mutationswere very common in AML and assumed that all tumor samples would display atleast one such mutation, but the study showed that these genes are mutated inonly 60 percent of cases. Mutations of the FLT3 gene, which plays a role inblood cell development, are seen in roughly a third of cases. FLT3, NPM1 andDNMT3A are commonly mutated genes in AML, and the researchers found that manyAML patients present with concurrent mutations of these genes. Recurringmutations were also found in cohesin genes, which play an important role incell division.

"This data set helps to integrate what was previouslyfragmented information," said Dr. Timothy J. Ley, associate director forcancer genomics at The Genome Institute at Washington University School ofMedicine in St. Louis and co-leader of the study. "We didn't realize how fewrecurrent mutations there were, and no one was thinking even five years agothat AML was associated with a high frequency of mutations in genes that encodeepigenetic modifiers."

Ozenberger says an important result of these findings "is amuch better stratification of patients based on their specific mutationalpatterns," noting that the results "begin to describe which genetic markersshould direct a physician to treat a patient more aggressively."

He says that he "absolutely" sees epigenetics, likegenomics, playing a larger role in cancer study, pointing out that byperforming multiple genomic analyses—such as "mutation, methylation, geneexpression, rearrangements"—on the same samples, TCGA can then integrate thedata "to reveal the correlations between gene sequence disruption and effectson gene expression." Methylation patterns also represent potential biomarkers,Ozenberger adds, as this work identified "particular methylation signaturesassociated with increased risk for recurrence."

AML is not the first type of cancer the TCGA ResearchNetwork has published such an analysis on; the group has also released reportson glioblastoma multiforme, ovarian serious adenocarcinoma, colorectaladenocarcinoma, lung squamous cell carcinoma and invasive breast cancer. Andother analyses are on the way: Ozenberger says the group "has a paper in presson kidney clear cell carcinoma," and will be releasing reports soon on lungadenocarcinoma, melanoma and thyroid carcinoma. He adds that "with data nowgenerated in TCGA on 8,000 cancer specimens in 26 different diseases, theNetwork has begun working hard on pan-cancer analyses," and later this yearTCGA plans to publish "comparisons of the genomes across 1000s of tumors,revealing unanticipated similarities between tumor types previously not knownto have the same genomic drivers."

"Elucidation of the genetic cause of disease cannot happeneffectively without the basic atlas of possible changes that occur in that disease,"says Ozenberger. "Each TCGA paper is historic, similar to the completion of thefirst representation of the human genome only 10 years ago. Each TCGA tumorreport provides the essential parts list that then allows for the informedapplication of research dollars, private and public, to translate thatinformation to therapies. That process cannot be started efficiently withoutthe comprehensive views that TCGA is providing."

"We'venever had such a complete picture of AML, and this data set will be mined byresearchers for years," Dr. Richard Wilson, director of WashingtonUniversity's Genome Institute and a co-leader of the study, commented in astatement. "These findings have probably identified every pathway in which amodification—and perhaps new drugs—might be beneficial. They also furtherrefine our understanding of the importance of individual mutations for diseaseclassification and prognostication, and will help us build better diseasemodels."