Skip to main content
Premium Trial:

Request an Annual Quote

Inflammatory Gene Expression Score IDs High-Risk AML Patients


NEW YORK – A unique genetic signature of inflammation can stratify adults and children with acute myeloid leukemia by their risk for poor survival outcomes.

The signature, called iScore for inflammation risk score, consists of 38 core inflammation genes in adults and 11 genes in children that associate with molecular features of AML, such as bone marrow microenvironment remodeling and the fraction of abnormal immune cells.

The research, published recently in Nature Cancer, aimed to address knowledge gaps in the effects that inflammation has on the composition of the bone marrow microenvironment and on clinical outcomes in AML, and in the reasons underlying the so far modest results from targeting immune cell function in AML.

"Unlike solid tumors where immunotherapy is really something that has changed the way we treat patients, [in] AML, patients really don't have any treatment that specifically targets the immune system," said Audrey Lasry, a researcher at the New York University School of Medicine and the study's co-first author.

"When we started this project," she added, "our thought was that maybe we just don't understand enough about what the immune microenvironment looks like in AML patients. We should really have a very unbiased look at it, so we performed single-cell RNA sequencing for both adult and pediatric AML patients."

Specifically, Lasry and her colleagues examined the different microenvironment compositions of bone marrow samples from 10 healthy donors, 20 adults with AML, and 22 pediatric AML patients.

Results showed that in general, the bone marrow microenvironment undergoes notable alterations in AML, with discrete differences between adult and pediatric patients. In particular, specific hematopoietic stem and progenitor cell subsets and certain myeloid cell types increased, while pre- and pro-B cells, along with CD4+ and CD8+ T-cell fractions grew depleted in pediatric patients, and whereas mucosal-associated invariant T-cell and natural killer cell fractions declined in both adult and pediatric patients.

As a validation of these findings, the NYU team noted that their results compare favorably to a previously published AML scRNA-seq dataset.

The researchers observed in the tumor cells of AML patients an inflammatory gene expression profile that appeared to impair the immune response through an association with more inhibitory T and B cells, including an expanded population of atypical disease-associated B cells rarely seen in healthy individuals.

These results led to the development of the iScore, which segregated patients into high- and low-inflammation groups. Notably, both adult and pediatric patients with high iScore values showed much poorer overall survival outcomes than those with low iScore values, and the risk associated with those values largely tracked with risk as determined by more traditional factors such as specific mutations, age, and chromosomal abnormalities. The finding carries potential clinical utility, as risk stratification impacts treatment decisions.

"Patients that have a more favorable risk will often just get chemotherapy whereas patients that are more at risk will also get a bone marrow transplant," Lasry said.

If clinically validated, the iScore may be able to speed up such decisions by quantifying patient risk through a simple blood draw and rapid analysis.

"We think that identifying [high-risk] patients early on might help assign them to a more aggressive treatment, which might help their survival," Lasry said.

Jiyang Yu, a computational biologist at St. Jude Children's Research Hospital who recently published research on a somewhat similar risk score for breast cancer, praised the study as "convincing," while cautioning that the current results need to be corroborated in a larger dataset.

"The sample size is small, given the heterogeneity [seen in] AML," he said, adding that the score's current "lack of connection with specific AML molecular subtypes and genetic alterations could prove a drawback in clinical implementation."

Nonetheless, he commented that he sees good potential in the iScore and can envision uses for it in his own research.

Bettina Nadorp, co-first author of the iScore study, agreed that some important questions remain to be resolved. 

Despite broad similarities in bone marrow microenvironment composition between adult and pediatric patients, for instance, the inflammatory genes driving that composition appeared to differ markedly between patient populations.

"In the overall signature that we started with," she said, "about 30 percent of the genes were overlapping, but the ones that had the largest effect on overall survival in those patients really didn't overlap between adult and pediatric [patients], which was kind of interesting."

Another interesting observation in the study was that high-inflammation patients showed expansions in GZMK+CD8+ T cells, a population known to give rise to terminally exhausted CD8+ T cells that traffic to inflammation sites and that some research has shown to respond to immune checkpoint blockade therapy.

"We think it might suggest that the high inflammation patients will do well with immunotherapy, but this is purely hypothetical and we've never tested that," Lasry said.

NYU has patented the iScore and is currently looking for potential licensing partners who could advance its clinical development.

In the future, Lasry, Nadorp, and their colleagues are interested in applying the iScore methodology to other leukemias, and to seeing if they can target one of the inflammation-related B or T cell populations identified in this study.

"Another question that we have is whether targeting the inflammation itself using anti-inflammatory drugs would be beneficial," Lasry said. "Those are some aspects that we are considering."