By Monica Heger
This story has been updated from a previous version posted Nov. 12 to include outside comment.
Using whole-genome sequencing in a single patient, researchers at Washington University's Genome Center identified a mutated gene in an acute myeloid leukemia relapse tumor — that they confirmed with targeted sequencing in a larger cohort — that is correlated with a lower survival rate and a more aggressive form of the disease — a finding that they believe could help guide treatment.
About 60 percent of AML patients have a normal cytogenetic profile, but vary widely in terms of how they respond to treatment. For some, chemotherapy is successful at achieving remission, but for others, the treatment seems to have no impact. The study, published this week in the New England Journal of Medicine, points to a new way to molecularly determine whether a patient with a normal cytogenetic profile has a good chance of responding to treatment or not, Rick Wilson, co-director of the Genome Center at Wash U and a senior author of the paper, told In Sequence.
"It allows us for the first time to have a good marker for whether or not those patients should be further classified as high risk or low risk," Wilson said.
The team used the Illumina Genome Analyzer to sequence the relapse tumor from a patient for whom they had already sequenced the primary tumor and matched normal sample (IS 11/11/2008).
Elaine Mardis, co-director of the Genome Center, presented preliminary results of the sequencing at Cold Spring Harbor's Biology of Genomes meeting last spring (IS 5/18/2010). At the meeting she reported that the team discovered three tier 1 mutations in the relapse tumor that had been missed in the primary tumor.
Improvements to the sequencing technology enabled the discovery of the additional mutations, the authors reported. The primary tumor was sequenced with shorter read lengths of only about 36 base pairs, to lower coverage, and with single-end reads, while the relapse tumor was sequenced with a paired-end sequencing strategy to about 30-fold coverage with read lengths of 75 to 100 base pairs.
"The improved sequence coverage, with paired-end reads and improved mutation calling algorithms, allowed us to identify several nonsynonymous tier 1 mutations that we did not detect in our initial sequencing effort," the authors wrote in the NEJM study.
One of the missed mutations was a single-base deletion in the DNMT3A gene. The researchers then used Sanger sequencing to sequence the exons of the DNMT3A and DNMT3L genes in 281 other AML samples. Of those patients, 62 had mutations in the DNMT3A gene that were predicted to affect translation, including 18 different missense mutations, six frameshift mutations, six nonsense mutations, three splice-site mutations, and a 1.5 megabase pair deletion encompassing DNMT3A.
The Wash U researchers have currently sequenced about 50 AML tumors and matched normal samples (IS 5/18/2010). At the time of the study, they had sequenced and analyzed 38 of those genomes, 11 of which had mutations in DNMT3A.
Strikingly, none of the mutations in DNMT3A were found in the 79 patients with a favorable risk, while patients with DNMT3 mutations had "significantly worse survival" — even among patients with normal or intermediate-risk cytogenetic profiles. These findings were independent of age and the presence of other mutations, and occurred "regardless of the type of mutation or genetic location," the authors wrote.
In the study, patients with DMNT3A mutations survived for a median of just over a year, compared to the median survival of around 3.5 years among those without the mutation.
These findings mean that the gene could be a good candidate for determining treatment. A mutated DNMT3A gene is a "strong marker for poor outcome," said Wilson, making patients with those mutations "good candidates for bone marrow transplantation early on in treatment," since they are less likely to respond to chemotherapy, he said.
Currently, more than half of AML patients are classified as having an intermediate risk and are then treated with standard chemotherapy. The study's results suggest, however, that patients with DNMT3A mutations would benefit from more aggressive treatment.
Nikolas Papadopoulos, director of translational genetics at the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University who has used sequencing to identify a frequently mutated gene in ovarian cancer (IS 9/14/2010), said that the finding could have both immediate and long term consequences.
"The first application, the low hanging fruit, is to stratify sets of patients that will benefit from certain types of clinical management," he said. Long term, he said the results could be used to find new drug targets, once scientists piece together the mechanism involved.
The study builds on sequencing efforts at the center that found that mutations in the IDH1 and IDH2 genes were predictive of outcomes in AML patients, and suggests that eventually treatment could be tailored to a patient's molecular subtype.
While the finding will have to be confirmed in other samples, Wilson said that the gene appears to be a promising biomarker. "The key thing is that right away this is a new marker that one could use to determine how aggressive treatment might need to be," he said.
Have topics you'd like to see covered in In Sequence? Contact the editor at mheger [at] genomeweb [dot] com.