An artificial intelligence (AI) algorithm developed at Vanderbilt University Medical Center (VUMC) is showing promise in optimizing precision immunotherapy for cancer patients. The algorithm aims to predict both the likelihood of toxicity and the potential benefit from immunotherapy, addressing a critical challenge in cancer care.
Leveraging AI for Precision Immunotherapy
Travis Osterman, DO, MS, FAMIA, FASCO, principal investigator for the Digital Precision Oncology Project, presented data on this initiative, highlighting its goal of helping oncologists make more informed decisions about treatment options. The project, a collaboration between VUMC and GE Healthcare, focuses on leveraging routinely collected patient data to predict immunotherapy outcomes across a pan-tumor sample.
Unlike many studies that focus on specific cancers like lung cancer or melanoma, this project included data from 2200 patients treated with immunotherapy across various cancer types. This large, diverse cohort is a key differentiator, according to Osterman. Another key aspect is the focus on clinical utility, providing information that is directly useful to clinicians in making treatment decisions.
Improving Clinical Trial Outcomes
The AI algorithm has demonstrated the potential to significantly improve clinical trial outcomes. Published in the Journal of Clinical Oncology: Clinical Cancer Informatics, the study showed that the algorithm could be used to select patients most likely to benefit from clinical trials. By selecting a hypothetical group of 100 patients out of 200 based on the algorithm's output, overall survival in the clinical trial could be improved by approximately 20%. This improvement can lead to more efficient clinical trials with fewer patients and shorter timelines, accelerating the drug development process.
Reducing Immunotherapy Toxicity
In addition to predicting efficacy, the AI algorithm can also identify patients at lower risk of developing pneumonitis, a common and potentially severe side effect of immunotherapy. The study found that by using the algorithm to select patients least likely to experience this toxicity, the incidence of pneumonitis could be reduced from 9% to 4%. This is particularly important for investigational agents that may have significant promise but also carry a higher risk of toxicity. By identifying the right patients for these drugs, the algorithm can help balance the risk and benefit of treatment.
The Future of Cancer Care
The development of this AI algorithm represents a significant step forward in bringing the future of cancer care to the present. By leveraging existing patient data and focusing on clinical utility, this approach has the potential to transform how immunotherapy is used, improving outcomes for patients and accelerating the development of new cancer treatments. The hope is that algorithms like these can act as synthetic biomarkers, helping companies and patients find the treatments that are most likely to benefit them and least likely to cause harm.
