Subpopulation Analysis of Melanoma

Advanced melanoma is a serious form of skin cancer that has spread to other parts of the body. Each year, tens of thousands of people worldwide are diagnosed with advanced melanoma. While new treatments have improved outcomes for some patients, many still do not benefit. One major challenge is that patients respond very differently to the same treatments, and doctors do not yet fully understand why.

Immunotherapy is a type of cancer treatment that helps the body’s own immune system recognize and attack cancer cells. One promising approach combines two immunotherapy drugs: PLX3397, which blocks a protein called CSF1R (colony-stimulating factor 1 receptor) that affects how certain immune cells behave, and pembrolizumab, a PD-1 (programmed death-1) antibody that removes a “brake” on immune cells so they can better attack cancer. Although this drug combination has shown encouraging results in some patients with advanced melanoma, others see little benefit or experience side effects. Understanding which patients benefit most, and why, remains an important unmet need.

We will address this problem by re-examining data from an existing Phase 1/2a clinical trial called the Study of Double-Immune Suppression Blockade. Instead of relying only on traditional statistical methods, which can struggle with small and complex trial datasets, we will use an explainable artificial intelligence approach. Artificial intelligence is a computer-based method that can find patterns in complex data. “Explainable” means that the results are clear and understandable, rather than coming from a hidden or unclear process.

We will use a platform called NetraAI to identify groups of patients within the trial who share similar characteristics and treatment outcomes. These characteristics may include immune markers (measurable signs of immune activity), basic clinical features such as age or disease stage, or treatment-related side effects. We will use this approach to better understand which patients respond well to the drug combination and what distinguishes them from patients who do not.

This research is necessary because it can help explain why responses to immunotherapy vary so widely in advanced melanoma. By learning more from data that already exist, we aim to improve the design of future clinical trials and move closer to more personalized treatment, where patients are more likely to receive therapies that will help them.

We will not recruit new participants or collect new data. All analyses will be conducted using de-identified information from the completed clinical trial, accessed securely through Vivli’s research environment. Patient privacy will be fully protected at all times.

Our ultimate goal is to show how explainable artificial intelligence can speed up learning from existing clinical trials, reduce development costs, and improve outcomes for people with advanced melanoma by supporting better and more precise clinical trial design.