NEW YORK (GenomeWeb News) – In a paper appearing online today in the New England Journal of Medicine, researchers from Washington University have published results from the second sequenced acute myeloid leukemia genome.
Using massively parallel, paired-end sequencing, the team sequenced the genomes of AML tumor and matched skin samples from a 38-year old male patient. They subsequently detected and validated a dozen somatic mutations in the protein-coding regions and 52 point mutations in conserved and/or regulatory regions — as well as many more mutations in poorly characterized parts of the genome that are yet to be validated.
Last fall, the same group became the first to sequence a full cancer genome, sequencing matched tumor and normal skin samples from a woman in her 50s who had the M1 type of AML. And earlier this year, lead author Elaine Mardis, co-director of Washington University's Genome Sequencing Center, reported on the team's progress in sequencing a second AML genome at the American Association for Cancer research annual meeting in Denver.
Now, the researchers have published those results, are in the midst of sequencing dozens more AML genomes, and have set their sites on other cancer types.
The researchers used an Illumina Genome Analyzer II to sequence AML tumor and matched samples from a 38-year-old man with AML. Although the man had the same AML subtype and similar pathology to the woman tested previously, the patients' outcomes were very different. Whereas the woman from the first paper died, the man tested for the current paper has been in remission for about three years.
The overall approach used in both papers was similar, Mardis told GenomeWeb Daily News. For instance, in both studies the researchers used Affymetrix Genome-Wide Human SNP 6.0 arrays to gauge the level of genome coverage for the tumor and normal skin samples.
Even so, Mardis noted that advances in sequencing and other technology made it possible to sequence the second AML more completely than the first AML genome using fewer sequencing runs.
When they examined the second AML genome, the team found and verified 12 mutations in gene coding regions. These changes, dubbed "tier 1 mutations" by the authors, are what Mardis calls the "low hanging fruit" of the tumor genome. Because these somatic changes affect known genes, understanding their potential effects is relatively straightforward.
Next, the team looked at mutations in conserved and/or regulatory regions of the genome, identifying and validating 52 somatic point mutations in conserved or regulatory genome sequences. Projects such as ENCODE have helped to elucidate the possible role of some of these "tier 2" mutations, Mardis explained, and researchers expect to gain an even better understanding of these parts of the genome in the relatively near future.
Meanwhile, the team has not yet validated any of the hundreds of "tier 3" mutations detected in the rest of the genome — mainly because much of the genome is poorly characterized, Mardis explained.
"Obviously there's still a lot about the genome that we don't understand," she said. But by getting as much information about each genome as possible now, she added, it should be possible to gain even more insights about the genomes down the road.
Although AML studies so far have come up with few recurrent mutations, there were exceptions in the current paper.
When they screened through 188 AML tumors, the researchers found that four of the 64 validated somatic mutations turned up in at least one additional tumor. Among them were mutations in NRAS and NPM1, which have been implicated in AML in the past, and IDH1, which has not.
Past studies suggest IDH1 mutations may be linked to survival times in brain cancer. But IDH1 mutations have not been found in AML in the past. Mardis said finding the IDH1 mutation "was a complete surprise. It kind of rocked us back on our heels," she said. And besides turning up in the AML2 genome, the researchers also found IDH1 mutations in another 15 AML samples.
Although IDH1 status did not appear to have prognostic value based on the samples assessed in this paper, Mardis said that with enough statistical analysis of IDH1 in the context of AML clinical factors of the disease, it's possible that the gene could present itself as a useful marker to help predict patient outcomes and direct treatment. Still, she cautioned, its much too early to make that kind of suggestion based on data available so far.
In a perspectives paper appearing in the same issue of NEJM, James Downing, a pathology researcher at St. Jude's Children's Hospital in Memphis noted that the study "opens a clear window into the rapid advancements that are being made in cancer-genome sequencing."
Downing also praised the technical improvements that the team has made in less than a year.
"Although the DNA sequence of the first cancer genome was reported by this group less than 10 months ago, the present study achieved a deeper (and thus more accurate) coverage of the genome, despite a reduction in the number of runs by a factor of more than six," he noted. "Moreover, improved computational algorithms have increased the accuracy of the identification of potential mutations, resulting in a decrease in the number of putative mutations that must be validated."
Mardis and her co-workers have another 40 or more AML genomes that are either completed or in their sequencing pipeline. In the future, they also plan to look at other cancer types, including breast and lung cancer. In the context of The Cancer Genome Atlas, Mardis added, the Washington University group and collaborators at the Broad Institute and the Baylor College of Medicine have gotten permission to sequence glioblastoma samples which they characterized last year using other methods.