New Tool Aims for Accurate Risk Assessment in AML, Even With Gaps in Data

June 6, 2020
Jared Kaltwasser
Jared Kaltwasser

Accurate risk assessment is a highly technical process for patients with acute myeloid leukemia (AML). A new tool could help make such assessments available even when patients can’t get to a specialty care center.

A newly developed tool could help clinicians perform highly accurate risk assessments for patients with acute myeloid leukemia (AML), even when certain cytogenetic and molecular data is missing.

When a patient is diagnosed with AML, physicians use a variety of metrics to determine the severity of the patient’s disease and to gauge the appropriate aggressiveness of their treatment. Genetic risk classifications can be an important way to distinguish between patients with high- and low-risk cancer, but breaking those larger categories down into more specific risk groups can be a challenging task.

One widely used tool is the European LeukemiaNet 2017 (ELN2017), which leverages new molecular markers to create a more complex genetic risk assessment. While the tool is accurate, there’s a problem—the kinds of full cytogenetic and molecular information required to make the model work is not always available, particularly in countries with fewer health care resources.

Writing in the journal Blood Advances, Douglas Rafaele A. Silveira, MD, of the AC Camargo Cancer Center, in Brazil, and colleagues explain a method of getting around the problem. Their novel tool, called Adapted Genetic Risk (AGR), allows for accurate prognostic classification even if some cytogenetic or molecular information is missing.

The team began by assessing data from 167 patients using the ELN2017 method. The team then used that data to build the new AGR model, which leveraged additional clinical variables to account for gaps in cytogenetic-molecular data.

“Although AGR was built over the current biological concepts for AML classification (eg, NPM1 mutation confers good prognosis), we use some probabilistic simplifications to approach the missing data issues, and this is the novelty of AGR,” Silveira told The American Journal of Managed Care®.

Next, they evaluated 2 independent testing cohorts to validate the AGR for calculating overall survival and disease-free survival. The investigators then integrated AGR into a tool called the survival AML score (SAMLS), which uses widely available tests like white-blood-cell counts and serum albumin to further refine risk classifications, he said.

“To our knowledge, AGR is the first tool in oncology trying to cope with the missing data issue,” he added. Silveira said he expects the tool to fill a significant need. In Brazil, 80% of patients with AML are treated in resource-constrained public health facilities, and only a minority of AML treatment centers are able to perform full cytogenetic and molecular testing. Thus, for patients to receive testing, many would need to travel long distances to get to a specialty cancer center.

Overall, Silveira and colleagues found that about 30% of patients could not be classified using ELN2017, for a variety of reasons.

“The way in which such classification systems are devised often do not take into account variations in clinical practice and resource availability that would affect applicability and therefore usefulness of the classification tools, which have been primarily designed for use in a ‘state-of-the-art’ healthcare setting,” he said.

When treating his own patients, Silveira said he would choose to use a validated prognostic information that is appropriate for the patient’s context, rather than attempting to use a “state-of-the-art” scoring system that is not designed for missing diagnostic data.

Reference

Silveira DRA, Quek L, Santos IS, et al. Integrating clinical features with genetic factors enhances survival prediction for adults with acute myeloid leukemia. Blood Adv. 2020;4(10):2339‐2350. doi:10.1182/bloodadvances.2019001419