New Methodologies for Predicting Individualized Response to Immune Checkpoint and Protein Kinase Inhibitors in Metastatic Melanoma

• Applicant: Professorin Dr. Ulrike Leiter-Stöppke, since 11/2025
• Subject Area: Dermatology ; Bioinformatics and Theoretical Biology ; Medical Informatics and Medical Bioinformatics
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 560972842

Background. Not all patients with melanoma benefit from Immune Checkpoint Inhibitors (ICI) due to intrinsic or acquired resistance. Individuals with melanoma who carry BRAF mutations are treated with targeted protein kinase inhibitors (PKI). Still, the response to PKIs is limited and less durable than the response to ICIs. As ICIs are costly and associated with immune-related toxicity and response to PKIs is limited, it is essential to predict which drugs should be used in individual patients based on their genetic profile and immune characteristics. This is particularly true when managing symptomatic brain metastases since corticosteroids, given to treat associated symptoms, decrease ICI efficacy. Hence, symptomatic brain metastases are treated with PKIs in patients with the BRAFV600 mutation, even though some patients may derive more benefits from ICIs. We intend to develop personal mathematical disease models based on individual features to predict the risk of disease progression under different therapies. A major hindrance is the need for a large sample, including lesion sizes, which are not currently documented as standard of care. AI-based algorithms that automatically evaluate lesion sizes can be of help. In this setting, we developed an algorithm that predicts the individual response to pembrolizumab. The current project will improve the algorithm's mathematical model by increasing its complexity, accounting for inter-patient heterogeneity, and extending the algorithm to all first-line drugs. Methods. Pseudo-anonymized retrospective data will be collected. A radiologist will retrospectively evaluate lesion sizes based on CT/MRI images (300 patients; ~3000 CT and MRI images). We will also assess lesion sizes with a new AI-based algorithm (SimU-Net). We will develop a general mathematical model integrating the genetic, molecular, and immune effects on the sizes of tumor lesions and retrospectively validate it. Further, we will create an algorithm using the patient's clinical data to individualize the general mathematical model, simulate the personal model under various first-line treatments, and estimate the time to disease progression. Expected results, significance, and future work. The new mathematical model will enable further research into this complex topic. Following external retrospective and observatory prospective validation, the algorithm will support decision-making in selecting optimal patient treatments, increasing the chances of durable benefits and sparing the patient from potentially toxic therapies. The validation of the AI-based algorithm's accuracy in evaluating the sizes of brain, liver, and lung metastases will boost the present and further biomedical research, which is currently limited by the shortage and cost of expert radiologists.

• DFG Programme: Research Grants
• International Connection: Israel
• Partner Organisation: The Israel Science Foundation
• Cooperation Partners: Professorin Zvia Agur; Professor Dr. Leo Joskowicz
• Ehemalige Antragstellerin: Dr. Teresa Amaral, Ph.D., until 11/2025