Rapid Epigenetic Test Could Speed Diagnostics for Acute Leukemias

Acute leukemias are life-threatening blood conditions that need urgent treatment. Approved targeted therapies such as BCL2, IDH, and menin inhibitors have expanded treatment options for certain patients, but determining if a given patient might benefit from one of these treatments can take time. Doctors need to order multiple complex diagnostic workups that can take days or even weeks to complete.

“As someone who treats patients with leukemia, I frequently face the agony of waiting for test results,” says Evan Chen, MD, a hematologic oncologist in the Adult Leukemia Program.  

Now, a new tool developed by Dana-Farber Cancer Institute investigators could deliver precise results within hours. The tool, MARLIN™, relies on epigenetic rather than genetic markers to make a precise diagnosis. The Dana-Farber team that developed the tool – computational biologist Volker Hovestadt, PhD, pathologist Gabriel Griffin, MD, and Chen – is running a research study now to determine how best to use the test in clinical practice. The tool could provide physicians like Chen with early insights into treatment possibilities and help them more consistently identify patients who are candidates for precision therapies like menin inhibitors.  

“It’s an exciting new lens on leukemia,” says Chen. “We’ve viewed it through a lens of genetics for a decade now and made ground-breaking progress. Adding epigenetic tests could help us learn even more.” 

Building a map for machine learning 

The idea for the project grew out of previous research by Hovestadt, who had used epigenetic markers to develop diagnostic tools for brain cancer. Recently, these tools were adapted to utilize a new technology called nanopore sequencing. Hovestadt saw an opportunity to do something similar in acute leukemia. 

“Leukemia diagnostics are already very advanced, but no one had tried to use this new technology in the field,” he says. 

Nanopore sequencing creates a readout that includes a cell’s entire DNA sequence plus a readout of epigenetic tags on DNA, called DNA methylation, that flip genes on and off. The technology unravels a double strand of DNA, unzips it, and threads a single strand through a synthetic pore, similar to a pore on a cell membrane. As the single strand of DNA moves through the pore, electrical signals are generated that precisely identify the next DNA molecule in the sequence – A, T, C, or G – and the presence of any methylation tags. 

“This technology has really revolutionized DNA methylation profiling and is streamlined for a clinical laboratory environment,” says Griffin. 

To use the technology in leukemia, the team needed a way to map a DNA methylation signature to a familiar diagnosis. To build this map, Hovestadt’s lab assembled a comprehensive database of AML, acute lymphocytic leukemia (ALL), and other acute leukemia cases from public databases. These databases included DNA methylation profiling data, DNA sequencing data, and other important clinical data. 

Griffin, Chen, and Hovestadt analyzed the resulting dataset to align DNA methylation signatures with known molecular subtypes of acute leukemias. The project took about a year, and the result was a map created exclusively using DNA methylation signals, with each unique signal annotated with traditional diagnostic classifications. 

“It required everyone’s expertise plus the previous research of many other labs around the world to figure out the different classifications,” says Hovestadt. “We were the first to build a comprehensive reference map across age groups and different disease types.” 

Cross-checking and validating 

The team used that map to train a neural network to take in results from nanopore sequencing and make a diagnosis. The result, described in Nature Genetics, is MARLIN™.  

To validate MARLIN™, Chen and Griffin worked together to classify patients at Dana-Farber who agreed to participate in their research study. The moment Chen suspected acute leukemia, he would send a biopsy to Griffin’s lab for rapid nanopore sequencing and MARLIN™ classification.  

Griffin was able to deliver results to Chen within three hours on average. Days or weeks later, when standard diagnostic results came in, the team confirmed that the official diagnosis matched that of the classifier for the majority of cases. 

MARLIN™ is not yet ready to guide treatment decisions, but Griffin is working to make the tool a CLIA certified laboratory test, which would enable it to be used more routinely at Dana-Farber and its results to be included in a patient’s medical record. 

To learn more about how the classifier might be used in clinical practice, Chen has recently expanded their research study to validate MARLIN™ in a larger population of patients. The team doesn’t expect this diagnostic to replace existing diagnostics for acute leukemia, but they do think it will add value alongside them.  

“One of our hopes with DNA methylation is that it will provide valuable early information about a patient’s condition. It also may act as a catch-all that can help us make sure we don’t miss patients who could potentially benefit from a particular targeted drug,” says Chen. “In the long run, we hope to use MARLIN™ to expand the number of patients who can benefit new and effective therapies like menin inhibitors.” 

Written by: Beth Dougherty
Medically Reviewed By: Evan Chen, MD