A combination of targeted sequencing and mass spectrometry-based genotyping has revealed a group of somatic mutations that could help diagnose and better stratify patients with myelodysplastic syndromes, according to a new study.
The research, led by investigators at Brigham and Women's Hospital in Boston, used 454 sequencing, Sanger sequencing, and mass-spectrometric genotyping to analyze a set of cancer-associated genes in bone-marrow aspirate samples from 439 patients with MDS. The team found somatic mutations in 18 genes, including two that had not been previously reported in the syndromes. Half of all patients had at least one point mutation, including approximately 50 percent of those with otherwise normal cytogenetics.
Their study, published this month in the New England Journal of Medicine, also successfully associated clinical variables, including overall survival, with specific mutation patterns. Mutations in five genes — TP53, EZH2, ETV6, RUNX1, and ASXL1 — predicted poor overall survival in patients with MDS "independently of established risk factors," the authors reported. Including mutations in any of these five genes as a variable could improve the accuracy of current prognostic prediction, they wrote.
Rafael Behjar, one of the study authors and an instructor in medicine at Brigham and Women's Hospital, told Clinical Sequencing News that the varied approach reflected the group's interest in doing a broad survey of mutation in MDS.
"We were interested in using a next-gen sequencing approach in order to have a higher sensitivity to pick up mutations — things we might have missed by Sanger sequencing," he said. But our overriding goal was to be as inclusive as possible — to include as many genes that might be mutated in the disease as we could."
"No one had ever done a study this large that looked at this large a number of patients and this many genes at the same time [in MDS], so we had both breadth to identify a lot of mutations, and we also had depth in that we could also say how these mutations we found actually related to clinical features," said Bejar.
While the study results need to be confirmed, associating the discovered mutations with clinical features and outcome may allow for both improved diagnosis of MDS and the potential to more accurately stratify patients into either low or high-risk groups, he said.
Myelodysplastic syndromes are a clinically heterogeneous set of hematological conditions that confer a risk of progression to acute myeloid leukemia. Currently, patients with MDS fall on either side of what Bejar called a "big dividing line."
"Low risk patients … could just be followed. Or they could be given growth factors or transfusion support – things that are fairly side effect free," he said.
On the other side of the line, physicians tend to push for more aggressive therapy for higher-risk patients, "and the only curative treatment in [these syndromes] is a bone marrow transplant, [which is] a pretty toxic procedure so we really reserve that for the patients at the greatest risk," Bejar said.
Current prognostic scoring systems consider chromosomal abnormalities and a variety of clinical features. "Those things we know add to risk, and what we found is that these mutations tell you something more even if you already know all that stuff about a patient."
The authors reported in their paper that for all patients except those with the highest scores in the International Prognostic Scoring System, the presence of a mutation in TP53, EZH2, ETV6, RUNX1, or ASXL1 was associated with an overall survival more akin to patients in the next-highest risk group. An approach to improving the IPSS would therefore be to include the presence or absence of mutation in any of these genes as an additional prognostic variable, they wrote.
"I can't guarantee that if we treated these patients with mutations aggressively that they would necessarily do better," Bejar said. "But we know that treating them like other low-risk patients certainly doesn't help them."
In their study, the group first examined 191 bone marrow aspirates for abnormalities in known oncogenes. Specifically, they performed mass spectrometry-based genotyping of 953 mutations representing 111 genes, a technique chosen for its "high throughput detection of recurrent oncogene mutations" in well-characterized locations.
The researchers confirmed somatic mutations in five genes known to be mutated in myelodysplastic syndromes and found recurrent mutations in one, GNAS, "that have not been previously reported in hematologic cancers." Genome-wide analysis of copy-number changes in a 75-sample subgroup using Affymetrix 6.0 SNP arrays also revealed a single case with a focal deletion of ETV6, a gene not previously known to be mutated in MDS.
The team then evaluated all 439 bone marrow samples for the genes thus far identified, along with another 13 previously reported to be mutated in hematologic cancers, to examine the clinical effects of mutations.
They used the Roche 454 platform to sequence PCR-amplified exons of a handful of target genes, and used Sanger sequencing on an additional group. Behar said this was because the team wanted to include genes that were being discovered during and after the next-gen sequencing part of the project "quickly and at a reasonable cost."
To isolate their analysis to somatic mutations, the researchers excluded any mutations that were either present in a buccal sample or previously reported to be germline in the National Center for Biotechnology Information's dbSNP database. Matched buccal DNA was available for half of the samples analyzed in the study.
Overall, the team identified somatic mutations in 18 genes, and found that about half of the patients had at least one mutation. "One thing people didn't necessarily expect was the high rate of mutations – that more than half of our patients had a mutation we could identify," Bejar said. MDS can be hard to diagnose, he explained, so there is potential to use mutation status as an additional factor in diagnosis.
Chromosomal abnormalities are one of the main markers used to diagnose the syndromes, but "half of patients have totally normal chromosomes," Bejar said. We found that even in that subset, you could find mutations in half of those – so it's essentially another marker you could use to make the diagnosis as well."
The researchers compared mutation patterns with both survival and with clinical variables. Mutations in RUNX1, TP53, and NRAS were each associated with severe thrombocytopenia and increased bone marrow blasts. Mutations in TET2 were found to be most strongly associated with normal cytogenetic features, and TP53 mutations were represented strongly in samples with a complex karyotype. Because TP53 mutation was also linked with a lack of other mutations, the authors posit that TP53 mutated patients might be considered a distinct molecular subclass of MDS.
To discover the independent contribution of mutations to survival, the researchers also conducted a multivariable analysis from which mutations in TP53, EZH2, ETV6, RUNX1, and ASKL1 emerged as independent predictors of survival.
"Essentially, we found mutations in a subset [of genes] we looked at that told you something about the patients' prognosis that you wouldn't have known … if you looked at what we consider to be our standard risk stratification system," Bejar said.
The group is now doing a similar project looking at therapy response. While the researchers chose the Roche 454 platform for their initial study for "technical" and cost reasons, Behar said this new investigation will use the Illumina sequencing platform.
The group doesn't have plans to study other diseases, according to Bejar, but he said similar methods are "exactly the approach I think people will be taking with other tumors and cancers, especially as the cost of sequencing goes down."
Going forward, he said, he thinks researchers will more likely start with very broad surveys that include "essentially the entire genome. And from that they'll get a list of candidate areas they want to look at in a larger number of samples with a more targeted strategy."
Indeed, many other groups are already taking this tack, led by efforts such as the Cancer Genome Atlas and the International Cancer Genome Consortium, which are taking a genome-wide approach to identify mutations of interest in a wide range of tumor samples.
According to Bejar, the group does not plan to develop a commercial test based on the genes they identified. However, he said, the university has filed an IP claim on the gene set that could be licensed for commercial development.
A test based on this research would likely have to be sequencing-based, Bejar said, given "the nature of the genes that are in that signature. … Most of the genes we found don't have a single hotspot you could look at using a genotyping tool. You really have to sequence … to find what you are looking for."
Have topics you'd like to see covered in Clinical Sequencing News? Contact the editor at mashford [at] genomeweb [.] com.