Verified by 
Artificial Intelligence
AI Predicts Immunotherapy Risks for Cancer Patients
AI advances precision medicine by helping to identify immunotherapy candidates.
Posted March 6, 2024 Reviewed by Kaja Perina
Not every cancer patient is a good fit for immunotherapy. Having a timely way for clinicians and health care providers to assess risk and patient outcomes is critical. A new study published in the Journal of Clinical Oncology Clinical Cancer Informatics unveils an artificial intelligence (AI) machine learning (ML) model that can predict cancer patient responses to immune checkpoint inhibitors (ICIs) using electronic health record (EHR) data.
“To our knowledge, this is the first ML solution that assesses individual ICI risk-benefit profiles based predominantly on routine structured EHR data,” wrote corresponding author Levente Lippenszky, MS, at GE HealthCare along with researchers affiliated with Vanderbilt University Medical Center (VUMC) in Nashville, Tennessee.
Globally by 2050, the number of cancer cases globally is projected to increase by 77% to more than 35 million compared to 20 million in 2022 according to 2024 estimates by the World’s Health Organization (WHO) International Agency for Research on Cancer (IARC).
There are many ways to treat cancer. Options include surgery, chemotherapy, radiation therapy, hormone therapy, hyperthermia, photodynamic therapy, biomarker testing, stem cell transplant, targeted therapy, and immunotherapy according to the National Cancer Institute at the National Institute of Health (NIH).
Immunotherapy is a biological therapy where the body’s own immune system is used to battle cancer cells. There are various forms of immunotherapy, such as therapeutic vaccines, CAR T-cell therapy, immune system modulators (immunomodulators), and monoclonal antibodies (MABs) or immune checkpoint inhibitors (ICIs).
Ever wonder how the body’s immune system knows which cells are normal versus harmful? Checkpoint proteins on immune cells function like on/off switches according to the American Cancer Society. When checkpoint proteins bind to partner proteins, the switch goes off and this keeps a type of white blood cell called a T cell (also known as Thymus-derived lymphocyte or T lymphocytes) from attacking the harmful foreign cells.
Immune checkpoint inhibitors are immunotherapy medications that block checkpoint proteins from binding to their partner protein which allows T cells to fight pathogens and cancer per the National Cancer Institute.
The 2018 Nobel Prize in Physiology or Medicine was awarded to Jim Allison, PhD, currently a professor at the University of Texas MD Anderson Cancer Center in Houston and Tasuku Honjo, MD, PhD, a professor at Kyoto University for their development of a novel cancer treatment through the inhibition of negative immune regulation.
In the mid 1990’s Allison, at that time the Director of the UC Berkeley Cancer Research Laboratory, discovered that by blocking the immune checkpoint protein called CTLA-4, the immune system can continue to attack cancer unfettered and inhibit or stop tumor growth in lab mice. This discovery led to the development of the first FDA approved immune checkpoint inhibitor that blocks CTLA-4 called ipilimumab (Yervoy) in 2011.
A well-known immunotherapy success story is that of former U.S. President Jimmy Carter, who announced that he was free of melanoma after a combination treatment of radiation with the immunotherapy drug pembrolizumab (Keytruda) in 2015. At the age of 90, he survived metastatic melanoma that had spread to his brain. The immune checkpoint inhibitor pembrolizumab is a monoclonal antibody, a protein that can spot and block the protein PD-1 receptor by binding that is on a T cell so that the immune system can strike and slay cancer cells, per the National Cancer Institute. The 99-year-old Nobel Peace Prize recipient and 39th U.S. president will be turning 100 this October.
Side effects of immune checkpoint inhibitors treatment are called immune-related adverse events (irAEs). According to the scientists, these unwanted side effects can happen to over half of cancer patients on a combination of immune checkpoint inhibitor medications and 20%-30% of those on a single immune checkpoint inhibitor.
“Although immune checkpoint inhibitors (ICIs) have improved outcomes in certain patients with cancer, they can also cause life-threatening immunotoxicities,” the researchers wrote.
How to find out which cancer patients are at risk? The scientists say that the current method for predicting toxicity-effectiveness requires measurement of biomarkers such as polygenic signatures and CD4+ lymphocytes for immune-related adverse events that are often not part of larger-size studies of participants and not routinely tested. Moreover, other studies that did use more practical data were less comprehensive and either had fewer participants, lacked verification by humans, did not evaluate both efficacy and toxicity, or had other limitations according to the GE Healthcare and VUMC team.
For this study, the team aimed to have a comprehensive approach and focused the AI model on predicting not only the one-year overall survival after initiation of ICI treatment, but also the risk of developing inflammation of the liver (hepatitis), lung (pneumonitis), or colon (colitis) for the period of one year after initiation of immune checkpoint inhibitor treatment.
What also sets this research apart is the large amount of real-world electronic health record data used. The AI machine learning model was trained using data from thousands of cancer patients who were treated with immune checkpoint inhibitors.
“In contrast to past studies, we developed and validated our algorithms using a large cohort of more than 2,200 ICI-treated individuals from a National Cancer Institute–designated Comprehensive Cancer Center with human-verified outcomes data,” shared the researchers.
The AI models had an accuracy up to 76% the researchers reported. With this proof-of-concept, as next steps the team plans to expand the scope to longitudinal predictions and are currently validating the AI models with additional datasets internationally, and from other health care institutions.
At the intersection of cutting-edge AI machine learning, electronic health record databases, precision oncology, medicine, and biotechnology are innovating cancer treatment for better outcomes in the future.
Copyright © 2024 Cami Rosso All rights reserved.
