How AI-Driven Continuous Monitoring Can Improve Management of Immunotherapy-Related Cytokine Release Syndrome in Oncology Patients

The field of immunotherapy has significantly improved survival rates for patients with cancer. Immunotherapies that redirect T cells to target tumor-specific antigens, such as chimeric antigen receptor (CAR)-modified T cells and T cell-engaging antibodies, have demonstrated success, particularly in hematologic or blood cancers, such as lymphomas, some forms of leukemia, and multiple myeloma.

Currently, seven CAR T-cell therapies¹ and six T-cell engager therapies have been approved by the FDA.² Despite these advancements and their clear benefits, immunotherapy-based treatments carry risks of severe, life-threatening toxicities, including cytokine release syndrome (CRS) and neurotoxicity.³

Cytokine release syndrome

CRS is a potentially serious adverse effect (AE) associated with immunotherapy. It occurs when activated immune cells release significant quantities of cytokines into the bloodstream.

This is a primary concern with potent immunotherapies, such as CAR T-cell therapy, with incidence rates ranging from 42%-100%, and severe CRS (grade 3 or higher) might occur in up to 46% of patients.⁴ CRS can present with symptoms of fever, nausea, hypotension, and, in the most severe cases, organ failure and death.

Prompt management of CRS is critical to prevent further deterioration and improve patient outcomes.⁵ For this reason, patients undergoing immunotherapy treatments are closely monitored in clinical settings, often requiring several days of hospitalization post-infusion to ensure safety.

This requirement not only amounts to a significant barrier to patient access but can also complicate patient recruitment in clinical trials due to higher participant burdens, care costs, and limited access to qualified treatment centers. Addressing these challenges requires a deeper understanding of how immunotherapy-related AEs develop post-infusion and how they can be more effectively managed in an outpatient setting.

Multivariate digital data and AI-powered CRS monitoring

Leveraging multivariate vital signs collected during standard-of-care bispecific T-cell engager immunotherapy administration, AI algorithms have been developed that have shown promising potential for detecting risks for developing CRS.

Figure 1 presents two real-world case studies. On the left, a patient-maintained pre-infusion temperature and blood pressure levels for 16 hours post-treatment, indicating no CRS development.

In contrast, the second patient exhibited signs of fever and hypotension, progressing to CRS Grade ≥2 post-infusion. Continuous monitoring could allow for meaningful deviations in vital sign signals to be picked up earlier or more nuanced patterns predictive of CRS onset to be learned.

If CRS is monitored continuously and changes in a patient’s physiology are identified quickly, most symptoms are reversible, with a reported mortality rate of less than 1% across 84 studies.⁶ Digital health technology presents an opportunity to provide remote monitoring for CRS in both healthcare settings and clinical trials.

Recently, the Digital Medicine Society shared a set of core resources via the DATAcc, aimed at leveraging digital innovations to support the development of risk-mitigating products aimed at monitoring immunotherapy patients at risk for CRS.⁷

Improving the treatment and monitoring pathway

A new approach that employs digital technology would see immunotherapy patients enrolled on a digital health platform in which baseline physiological data is collected via a wearable device before infusion. Following treatment, patients are monitored for AEs.

Continuous vital sign data and AI-powered algorithms assist healthcare professionals (HCPs) in determining when patients can be safely discharged to an outpatient setting. Even after discharge, with wearables, HCPs maintain continuous oversight, with the algorithms providing alerts if AEs emerge that require inpatient care. This means that patients can be promptly readmitted to the hospital.

Currently, the standard of care for CRS involves episodic monitoring of patient vitals and treatment decisions mainly based on physician discretion, which can introduce subjectivity. In contrast, continuously collecting objective physiological data gives HCPs a complete view of the immune response and when it may become dangerous.

Taking a technology-based approach will bring a better understanding of how to develop effective immunotherapy treatments safely.

A digitized future for CRS monitoring

As an emerging field, there is a lack of standardization of safety monitoring post immunotherapy. Digital health and AI technology offer a promising solution to reduce the risk of serious AEs and alleviate the burden of prolonged hospital stays and associated costs.

Remote monitoring solution for CRS paves the way for more consistent patient care, a deeper understanding of immune responses, ultimately widening patient access to life-saving immunotherapy and reducing recruitment barriers in clinical trials.

References

1. https://www.cancer.gov/about-cancer/treatment/research/car-t-cells
2. https://www.aacr.org/blog/2023/11/14/bites-help-immune-system-destroy-cancer-cells/
3. https://pmc.ncbi.nlm.nih.gov/articles/PMC8227172/
4. https://pmc.ncbi.nlm.nih.gov/articles/PMC8600921/
5. Abramson et al., 2021.
6. https://pubmed.ncbi.nlm.nih.gov/34359816/
7. https://datacc.dimesociety.org/de-risking-cytokine-release-syndrome/#practical-guide

About the Author

Christine Guo, PhD, Chief Scientific Officer, Ametris (formerly ActiGraph), leads the clinical and data science team at Ametris and is responsible for the scientific strategy and services supporting Ametris leadership in digital medicine. Christine has over 15 years of experience in clinical research and a vision for leveraging technology in clinical trials and practice. Prior to Ametris, Christine was Head of Scientific Innovation at Biogen Healthcare Solutions, leading the clinical development and validation of Biogen’s digital medicine products (Software as Medical device) in multiple sclerosis, neuromuscular, and neurodegenerative diseases. Christine brings unique scientific insights by bridging clinical and technical disciplines and is passionate about leveraging data and technology to improve people’s health. Christine holds a B.A. in biological sciences from Peking University and Ph.D. in neuroscience from Stanford University.