Mathematical and data-driven analysis of atypical cancer antigen 15-3 patterns in male metastatic breast cancer

Area of Science:

• Oncology
• Biomarker Analysis
• Mathematical Modeling

Background:

• Cancer Antigen 15-3 (CA 15-3) is a key glycoprotein biomarker for monitoring breast cancer progression and treatment efficacy.
• Understanding the dynamic behavior of CA 15-3 is crucial for personalized cancer management.

Purpose of the Study:

• To analyze five years of CA 15-3 measurements in a male metastatic breast cancer patient.
• To characterize the temporal dynamics of CA 15-3 using integrated mathematical and data-driven approaches.
• To investigate atypical oscillatory patterns in biomarker levels.

Main Methods:

• Fitted a phenomenological model with oscillatory and exponential components to longitudinal CA 15-3 data.
• Employed Variational Mode Decomposition (VMD) and Hilbert Spectrum (HS) analysis for data-driven insights.
• Estimated key model parameters including baseline, clearance/growth rates, and oscillation characteristics.

Main Results:

• Identified a persistent, atypical sinusoidal pattern in CA 15-3 with a 170-day period over three cycles.
• Observed the emergence of oscillations coinciding with the start of metastatic-phase therapy, linked to Palbociclib.
• The phenomenological model accurately reproduced the observed periodic fluctuations and overall trends.
• VMD/HS analysis revealed instantaneous frequency components associated with metastatic onset and therapy intervals.

Conclusions:

• The study highlights an unusual, predictable oscillatory behavior in CA 15-3, potentially linked to specific therapies like Palbociclib.
• Combined mathematical and data-driven modeling enhances the interpretation of dynamic biomarker behavior.
• This integrated approach supports more personalized strategies for cancer monitoring and management.