There is a "critical need" for early radiographic markers for metastatic melanoma treated with immune checkpoint inhibitors such as pembrolizumab, because of atypical responses - e.g., pseudoprogression - to these drugs, Dr. Laurent Dercle of Columbia University Medical Center in New York City told Reuters Health by email.
"The most striking result of our study was that the signature could be used to identify pseudoprogressors, since it estimated a favorable outcome and a favorable overall survival in 82% of patients whose tumors were categorized as pseudoprogression at month 3," he said. "We previously found that similar features could be used in patients with lung cancers and colorectal cancers."
In non-small cell lung cancer, for example, "We trained algorithms to predict tumor sensitivity to three systemic cancer therapies: the immunotherapeutic agent nivolumab, the chemotherapeutic agent docetaxel, and the targeted therapeutic gefitinib," he said.
As reported in JAMA Oncology, the prognostic study in the melanoma patients used radiomics and machine learning to retrospectively analyze CT images obtained at baseline and first follow-up in the KEYNOTE-002 (https://bit.ly/347Wnmg) and KEYNOTE-006 (https://bit.ly/3o07O6J) multicenter trials.
Participants included 575 patients with advanced melanoma randomly assigned to training and validation sets. KEYNOTE-002 involved trial groups testing IV pembrolizumab, 2 mg/kg or 10 mg/kg every two or three weeks, or investigator-choice chemotherapy. KEYNOTE-006 included trial groups testing IV ipilimumab, 3 mg/kg every three weeks and IV pembrolizumab, 10 mg/kg every two or three weeks.
A random forest model combined 25 imaging features extracted from tumors segmented on CT images to identify the combination (signature) that best estimated OS with pembrolizumab among all participants. The signature combined four imaging features - two related to tumor size and two reflecting changes in tumor imaging phenotype.
In the validation set (287 patients treated with pembrolizumab), the signature reached an area under the curve of 0.92 for estimation of OS. By contrast, the standard method of estimating OS (Response Evaluation Criteria in Solid Tumors 1.1), achieved an AUC of 0.80, and classified tumor outcomes as partial or complete response (32.4%), stable disease (31.3%), or progressive disease (36.2%).
Dr. Dercle said, "Our next research steps are to...go one step closer to using AI in clinical practice (by) prospectively validating our signatures and applying them to new clinical trials to understand how they can be used."
"Eventually, I see radiomics becoming part of a precision medicine approach to cancer treatment by helping physicians choose the most appropriate therapy for each individual patient and modify it more quickly when it's not working," he said. "The amazing thing is that it doesn't require more tests or more scans. We are simply applying computing power to analyze the medical images that patients are already getting."
Dr. Michael Farwell of the Perelman School of Medicine at the University of Pennsylvania in Philadelphia, coauthor of a related editorial, commented in an email to Reuters Health, "It will likely be a few years before automated segmentation software for volumetric lesion analysis, and radiomics software, become available for clinical use. But when they do, they will be a welcome addition to radiology clinical practice."
"Other approaches that assess immunotherapy response at even earlier time points are being explored, including CD8 PET/CT and granzyme B PET/CT for imaging CD8+ T cells and immune activation, respectively, and FDG PET/CT to image the metabolic flare in responding tumors as early as one week after starting treatment," Dr. Farwell concluded.
SOURCE: https://bit.ly/3fYD2H6 and https://bit.ly/3fYDkOc JAMA Oncology, online January 20, 2022.
By Marilynn Larkin
Posted on
site created by:
© 2024 Medicom Medical Publishers. All rights reserved. Terms and Conditions | Privacy Policy