Workflow visual abstract. Credit: Nature Medicine (2025). https://doi.org/10.1038/s41591-025-03532-x
Researchers with City of Hope and Memorial Sloan Kettering (MSK) Cancer Center have created a tool that uses machine learning to assess a non-Hodgkin lymphoma (NHL) patient’s likely response to chimeric antigen receptor (CAR) T cell therapy before starting the treatment, according to study results published in Nature Medicine.
CAR T cell therapy is one of the most promising recent advances made in the fight against blood cancers. But more than half of NHL patients who do not respond to standard lines of treatment also relapse or progress within six months of CAR T therapy.
Known as InflaMix (Inflammation Mixture Model), the new tool was developed to assess inflammation, a potential cause of CAR T failure, by testing for a variety of blood biomarkers in 149 patients with NHL.
With the help of machine learning, a type of artificial intelligence that uses algorithms to learn from sets of information and draw conclusions from patterns found in that data, the model was able to find an inflammatory biomarker from a series of unique blood tests not usually employed in standard clinical practice.
By analyzing the inflammatory signature that InflaMix identified, the researchers found it was associated with a high risk of CAR T treatment failing, including increased risk of death or disease relapse. InflaMix is an unsupervised model, meaning that it was trained without any knowledge of clinical outcomes.
“These studies demonstrate that by using machine learning and blood tests, we could develop a highly reliable tool that can help predict who will respond well to CAR T cell therapy,” said Marcel van den Brink, M.D., Ph.D. president of City of Hope Los Angeles and City of Hope National Medical Center, the Deana and Steve Campbell Chief Physician Executive Distinguished Chair in Honor of Alexandra Levine, M.D., and a senior author of the paper.
“With a rigorous statistical approach, we demonstrated that this is one of the most thoroughly validated tests we have for predicting CAR T outcomes in lymphoma patients and could be used by oncologists everywhere to assess the risk of CAR T in an individual patient.”
According to the team, the machine learning model is very flexible and worked well even when they used only six available blood tests—all of which are typically evaluated for patients with lymphoma—to assess InflaMix’s capabilities with less data. Researchers said this is important because it means this test can be made available for most, if not all, patients with lymphoma.
“Prior studies had hinted that inflammation might be a risk factor for poor CAR T cell efficacy,” said medical oncologist Sandeep Raj, M.D., who specializes in bone marrow transplants at MSK and is lead author of the Nature Medicine paper.
“Our goal was to refine this concept and build a robust and reliable clinical tool that characterizes inflammation in the blood and predicts CAR T outcomes.”
Studies of three independent cohorts comprising 688 patients with NHL who had a wide range of clinical characteristics and disease subtypes and used different CAR T products were also used to validate the team’s initial findings.
Next, City of Hope and MSK researchers plan to investigate whether blood inflammation defined by InflaMix directly influences CAR T cell function and learn more about the source of this inflammation.
“InflaMix could be used to reliably identify patients who are about to be treated with CAR T and are at high risk for the treatment not working,” said Dr. Van den Brink. “By identifying these patients, doctors may be able to design new clinical trials that can boost the effectiveness of CAR T with additional treatment strategies.”
The work was primarily done at MSK where Dr. Van den Brink worked for more than two decades before coming to City of Hope in 2024.
More information:
Sandeep S. Raj et al, An inflammatory biomarker signature of response to CAR-T cell therapy in non-Hodgkin lymphoma, Nature Medicine (2025). DOI: 10.1038/s41591-025-03532-x
Provided by
City of Hope National Medical Center
Citation:
Machine learning helps predict immunotherapy response in lymphoma patients (2025, April 1)
retrieved 2 April 2025
from https://medicalxpress.com/news/2025-04-machine-immunotherapy-response-lymphoma-patients.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
