Public Release: 

Panel predicts whether rare leukemia will respond to treatment

Successful biomarker could spare more than half of patients from ineffective therapy

University of Michigan Health System


IMAGE: This is Maria E. Figueroa, M.D., University of Michigan Health System. view more

Credit: University of Michigan Health System

ANN ARBOR, Mich. -- Patients with chronic myelomonocytic leukemia have limited treatment options, and those that exist are effective only in fewer than half of patients. Now, a new study identifies a panel of genetic markers that predicted which tumor samples would likely respond to treatment.

The finding could lead to a test to help doctors know which patients could be spared from a lengthy course of treatment that's unlikely to work.

Finding an effective biomarker is particularly crucial for this disease because the current treatment is slow-acting, which means patients must undergo as much as six months of treatment before there are signs of response.

"The slow kinetics is what gets us. It's not just one week or one dose to see signs of response. A good biomarker test could potentially prevent patients who are unlikely to respond from receiving prolonged, unwarranted treatments," says study author Maria E. Figueroa, M.D., assistant professor of pathology at the University of Michigan Medical School.

Chronic myelomonocytic leukemia, or CMML, is a cancer that begins in the blood-forming cells of bone marrow. It primarily affects older adults. Previous research has tried to identify genetic differences but with little success.

In the current study, published in the Journal of Clinical Investigation, researchers from the University of Michigan Comprehensive Cancer Center and the University of Florence in Italy used next generation gene sequencing techniques to dive deep into the DNA of CMML tissue samples treated with the drug Decitabine.

By harnessing this advanced technology to look beyond the usual regions of a gene, the researchers were able to find a number of differences between the samples that responded to treatment and those that did not.

The team combined a series of markers into a panel, which they then tested against 28 additional CMML samples from a research team at the Gustave Roussy Cancer Center in France. They found the panel was 87 percent accurate in predicting response to Decitabine.

In addition, the researchers looked at the non-responders to try to understand why they were resistant to Decitabine. They found two proteins, CXCL4 and CXCL7, were overexpressed. When cells exposed to high levels of these chemokines were treated with Decitabine, the drug's effect was blocked.

"We are pursuing this to understand why these proteins block the effect of the drug and whether we can develop a new compound that could be used along with Decitabine to turn non-responders into responders," Figueroa says.

In addition, the team is working to refine and translate its panel to a viable biomarker test that could be used in the clinic. More research is needed before a test would be available.


Additional authors: Kristen Meldi, Tingting Qin, Francesca Buchi, Nathalie Droin, Jason Sotzen, Jean-Baptiste Micol, Dorothee Selimoglu-Buet, Erico Masala, Bernardino Allione, Daniela Gioia, Antonella Poloni, Monia Lunghi, Eric Solary, Omar Abdel-Wahab, Valeria Santini

Funding: Leukemia and Lymphoma Society, Sass Foundation, National Cancer Institute grants T32 CA967622 and K08 CA16064701, U.S. Department of Defense grant W81XWH-12-1-0041, the Josie Robertson Investigator Program, Damon Runyon Clinical Investigator Award, the Evans Foundation, the French National Cancer Institute, the Ligue Nationale Contre le Cancer, MIUR/University of Florence, FISM Postdoctoral Fellowship

Disclosure: None

Reference: Journal of Clinical Investigation, published March 30, 2015


U-M Cancer AnswerLine, 800-865-1125
U-M Comprehensive Cancer Center,
Clinical trials at U-M,
mCancerTalk blog,

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.