ORLANDO – By integrating gene mutation and expression profiles of more than 1,300 patients with myeloid malignancies, researchers have been able to identify distinct subtypes for acute myeloid leukemia and myelodysplastic syndrome, which may improve understanding of patients' disease trajectories.
Myelodysplastic syndrome is a premalignant disease that affects myeloid cell. It is a precursor to acute myeloid leukemia, an aggressive blood cancer caused by the accumulation of immature blood cells. The increased use of sequencing in the past decade has improved the field's understanding of the genetic mutations that cause MDS and AML, but for the most part, those data have not been integrated with expression data.
In the research setting, while some groups have used transcriptomics in a limited way to study myeloid malignancies, gene expression analysis is rarely factored in to guide patient care. This has limited researchers' ability to fully characterize the spectrum of biomarkers associated with the diagnosis and prognosis of AML and MDS.
At the American Society of Hematology's annual meeting here, researchers led by Ilaria Iacobucci from St. Jude Children's Research Hospital presented data from a study in which they aimed to explore in an unbiased way the role of genetic mutations and gene expression in the diagnosis and prognosis of these diseases. "We need integrated gene expression [and] sequencing mutational analysis for better classifying myeloid malignancies," said Iacobucci, adding that this can also improve understanding of patients' prognosis.
For the study presented on Tuesday, researchers conducted whole-genome sequencing and transcriptomic analysis via RNA-seq of cancer samples from more than 1,300 adult patients — nearly 600 with AML and around 700 with MDS — and looked at how detected genetic variants tracked with gene expression patterns, patients' clinical disease features, and outcomes.
Iacobucci and colleagues were able to confirm the diagnosis of 11 percent of patients where AML was due to recurrent genetic abnormalities according to the World Health Organizations' classifications. The researchers also identified more than 7,000 variants (including somatic and germline mutations, chimeric fusions, and structural variants) in 839 genes, around a third of which were potential driver genes. Patients harbored between one and 18 mutations, and averaged five mutations.
Some genetic mutations overlapped between the two diseases but were more frequent in one setting than the other. For example, NPM1 mutations occurred in 27 percent of AML and around 1 percent of MDS cases.
The researchers also showed that NPM1 mutations can co-occur with mutations in other genes that impact patient outcomes. For example, when NPM1 mutations occurred with FLT3 mutations, patients had poor outcomes, while patients with mutations in NPM1 and cohesin genes had good outcomes.
RUNX1 alterations showed up in 12.5 percent and 9.5 percent of AML and MDS cases, respectively, and appeared to be associated with poor outcome.
TP53 mutations occurred in 12 percent of AML and 10 percent of MDS cases and were associated with complex karyotypes in both. They also tended to show up in older patients. Moreover, when TP53 mutations occured with complex karyotypes, they conferred poor outcomes, the researchers reported.
"This study, for the first time, provides a very detailed description of how different mutations cooperate together, and [shows how] this can be used to stratify patients [by] cataloging different mutations and correlating them with outcome," Iacobucci said in an interview.
Charles Mullighan, also from St. Jude and a coauthor on the study, added in an interview that the use of RNA-seq in the study allowed researchers to more precisely identify subgroups with distinct gene expression profiles and specific mutational patterns that weren't evident before.
Researchers identified three gene expression groups that accounted for 9 percent of cases, which had mutually exclusive mutations in RUNX1, TP53, and CEBPA and co-occurred with mutations in DNA methylation, splicing, and signaling genes. These groups were also associated with differential outcomes. For example, patients with RUNX1 mutations who had other co-occurring mutations and high MN1 expression had poor prognosis.
Iacobucci and colleagues showed that while AML cases had gene expression profiles that clustered with specific mutational patterns, expression profiles of MDS patients were not as variable even though they also had a complex landscape of mutations. Around 27 percent of MDS cases had mutations in SF3B1, which did not show up in 14 percent of patients with SFRS2 mutations and 6 percent of cases with U2AF1 mutations. Additionally, around 14 percent of MDS cases had TP53 mutations and 11 percent had RUNX1 mutations, which occurred with mutations in epigenetic regulators and were associated with patient outcomes.
"ASH is deeply invested in providing dictionaries to inform research and clinical interpretation in hematology and datasets like [this] … have real implications for the advancement of precision medicine efforts," Robert Brodsky, a professor of medicine at Johns Hopkins University who specializes in hematologic malignancies and is also ASH secretary, said at the meeting.
The AML and MDS cases used for the analysis are part of a larger study being conducted by Munich Leukemia Laboratory (MLL) in Germany, which has analyzed the genomes and transcriptomes of 5,000 cases across 30 hematologic malignancies. Mullighan noted that their colleagues at MLL have been able to show that they can perform whole-genome and transcriptomic analyses in seven days, which he said was a clinically useful time frame.
However, Mullighan also pointed out that this is not a clinical study but a very large proof-of-principle study. "We've done a lot of work describing the landscape, the constellations, and the more conventional associations with outcome," he said. "But that very question as to how you build a predictive model, that's a more complex question. That [work] is still ongoing on our end for this dataset."
Iacobucci noted that in addition to studying the genomics and transcriptomics of AML and MDS, researchers are also establishing a repository of patient-derived xenografts as a platform for testing out different molecularly informed treatment strategies before giving them to patients. "In this way, in addition to the genomic information, you have biological material to test [out] different therapies," she said. "We need to know the genome in order to guide treatment and prevent progression or relapse."