AML Gene Panel Moves Targeted Sequencing Closer to Clinical Utility
Memorial Sloan-Kettering’s Ross Levine is using a gene panel developed with RainDance to better understand leukemia and to stratify patient samples by predicted outcome.
Ross Levine, an associate member of the Human Oncology and Pathogenesis Program at Memorial Sloan-Kettering Cancer Center in New York City, is well into a pilot project aimed at establishing a better understanding of the genetic basis of leukemia progression. The research program uses the RainDance Technologies targeted sequencing platform to study a panel of genes associated with acute myeloid leukemia in a cohort of 300 patients who have received various treatments and therapies in the last few years.
Levine is one of a relatively new breed of physician-scientists, or lab scientists who also practice medicine — providing a built-in feedback loop that encourages research areas with clear implications for medicine. Being part of both worlds “grounds you in asking the right questions in the lab that are clinically relevant,” Levine says.
With a medical background as a hematologist and oncologist, Levine’s fascination with the research world stems from a fellowship at the Dana-Farber Cancer Institute. Working with Gary Gilliland, he used genome sequencing technology to interrogate the tyrosine kinome, identifying JAK2 mutations in patients with myeloproliferative neoplasms. Just last year, the first drug acting on JAK2 mutations was approved by the US Food and Drug Administration. For a physician-scientist, it was a great display of how meaningful research advances could be for patient care: “That really encouraged us to believe that sequencing efforts could lead to clinically relevant findings,” Levine says. “We were asking questions in the lab that were not just biologically important, but had the potential to inform the development of novel therapies in the clinic.”
Since joining Memorial Sloan-Kettering in 2007, Levine “became interested in trying to carefully look at the frequency of known mutations in a large cohort of AML patients to determine which mutations occur most commonly and which predict for good or bad outcome in leukemia,” he says. His team used a Sanger-based pipeline to study a large AML cohort in a paper published this year in the New England Journal of Medicine entitled “Prognostic Relevance of Integrated Genetic Profiling in Acute Myeloid Leukemia.” In it, the authors describe a study of 18 genes in 502 AML patients which allowed them to define mutations that improved risk stratification for patient outcome.
“That work allowed us to develop prognostic classification schema that requires targeted sequencing as a way to figure out which patients are likely to relapse and which patients are most likely to be cured,” Levine says. “The question we started to ask ourselves was, how were we going to efficiently translate that into a clinical context?”
The RainDance Technologies platform, which is currently approved for research use only, offered “a very elegant approach” in upgrading from the Sanger-based pipeline. Levine had used the technology for other experiments and had been impressed “with the coverage and particularly with its ability to get high-quality data in regions that are difficult to sequence, like high GC content,” he says.
“We have to be thinking about what aspects of these data are going to be most important when developing a test — and it’s not just our ability to get inexpensive resequencing data,” Levine says. “It’s our ability to get very high-quality genomic information such that we can give patients the information that they’re positive or negative for sets of genes with prognostic relevance.” Other platforms that miss important regions will not be clinically viable in the long term, so Levine and his team prefer to focus their research on technologies that could one day deliver patient information “without losing the data that would be clinically important.”
In planning for a possible clinical pipeline, Levine has to develop and optimize his panel in the research realm. To that end, he has worked with RainDance to design a panel of 32 genes known to be associated with AML. “The panel has been developed and has been optimized,” Levine says. “We are testing it on patient samples collected during the last couple of years, trying to get a sense for how well it performs with large data sets.” That pilot project of 300 patients is well underway, evaluating whether mutations in any of these genes can predict outcome in various courses of treatment. “Our approach is that you annotate all the genes, and then you use the ones that you know are important for clinical care,” Levine says. “And if prognostic schema change, you will have the data on the relevant genes at hand.”
If all goes smoothly, the next step would be to work with molecular pathologists to figure out how such a pipeline could be used in the clinic. “The platform is working well. We’re starting to get lots of data back, and we’re very encouraged,” Levine adds. “Our hope is to collaborate with experts in diagnostics to transition this platform into the clinical setting.”
— Meredith Salisbury