Predicting response to pembrolizumab in metastatic melanoma by a new personalization algorithm

Area of Science:

• Oncology
• Immunotherapy
• Mathematical Modeling

Background:

• Advanced melanoma treatment relies on immune checkpoint inhibitors like pembrolizumab, offering durable responses but with limited efficacy in up to 50% of patients.
• High treatment costs and poor prognosis for non-responders necessitate predictive tools to identify potential responders before initiating therapy.
• Personalized medicine approaches are crucial for optimizing immunotherapy outcomes in advanced melanoma.

Purpose of the Study:

• To develop and validate a personalization algorithm for predicting time to disease progression in advanced melanoma patients treated with pembrolizumab.
• To create a mathematical model integrating tumor-immune interactions and drug effects for personalized response prediction.
• To identify key clinical parameters that can be used to personalize the predictive algorithm.

Main Methods:

• Developed a mathematical model simulating tumor-immune system-pembrolizumab interactions.
• Created a personalization algorithm using clinical pretreatment data from 54 advanced melanoma patients.
• Correlated pretreatment measurements with mathematical model parameters to predict individual patient time to progression.
• Validated the algorithm's predictive capacity using Leave-One-Out cross-validation.

Main Results:

• Baseline tumor load, Breslow thickness, and nodular melanoma status correlated significantly with CD8+ T cell activation and tumor growth rates.
• The personalized mathematical model predicted time to progression with moderate accuracy (Cohen's κ = 0.489).
• Predicted versus clinical time to progression showed moderate accuracy (R² = 0.505) in progressing patients.

Conclusions:

• A simple mathematical model, personalized by pre-treatment clinical data, can predict individual patient response to pembrolizumab.
• The algorithm demonstrates moderate accuracy in predicting time to progression, with potential for improvement through larger datasets and independent validation.
• This approach offers a promising tool for personalizing immunotherapy in advanced melanoma treatment.