Artificial Intelligence May Help Predict Response to Immunotherapy Among Patients With Metastatic Melanoma
Algorithms can be trained to analyze patient data and histology images to predict if immune checkpoint blockade treatment will be successful.
Response to immunotherapy among patients with advanced melanoma can be predicted with the help of a computer program trained to analyze histology images of their tumors and other clinical information, according to results published in Clinical Cancer Research, a journal of the American Association for Cancer Research (AACR).
Specifically, the program can predict if the therapy known as immune checkpoint blockade, which helps the body’s immune system to better attack cancer cells, is likely to be effective in a particular patient.
“While immune checkpoint inhibitors have profoundly changed the treatment landscape in melanoma, many tumors do not respond to treatment, and many patients experience treatment-related toxicity,” said Iman Osman, MD, an author of the study who is a medical oncologist in the Departments of Dermatology and Medicine (Oncology) at New York University (NYU) Grossman School of Medicine and director of the Interdisciplinary Melanoma Program at the NYU Langone’s Perlmutter Cancer Center.
“An unmet need is the ability to accurately predict which tumors will respond to which therapy,” Osman said. “This would enable personalized treatment strategies that maximize the potential for clinical benefit and minimize toxicity.”
The system developed by the researchers uses an algorithm that can use a process called machine learning to analyze histology images of the patients’ metastatic tumors of melanoma derived from a biopsy. The algorithms – called deep convolutional neural networks (DCCN) – are trained to analyze digital images and identify patterns associated with treatment response. This information was combined with specific clinical data, including patients’ Eastern Cooperative Oncology Group (ECOG) performance status and their immunotherapeutic treatment regimen, to develop a prediction model for response to immune checkpoint blockade.
Using this model, the researchers could stratify patients into high versus low risk for disease progression, with significantly different progression-free survival outcomes between the two groups.
“Our approach shows that responses can be predicted using standard-of-care clinical information such as pre-treatment histology images and other clinical variables,” said another author of the study, Aristotelis Tsirigos, PhD, a professor in the Institute for Computational Medicine at the NYU Grossman School of Medicine and member of the NYU Langone’s Perlmutter Cancer Center.
“There is potential for using computer algorithms to analyze histology images and predict treatment response, but more work needs to be done using larger training and testing datasets, along with additional validation parameters, in order to determine whether an algorithm can be developed that achieves clinical-grade performance and is broadly generalizable,” Tsirigos said.