SRS132 - Development and evaluation of a non-linear mathematical model of breast cancer growth

Abstract Objectives This study aimed to develop and validate two non-linear mathematical models, exponential and quadratic, to predict tumour growth in sedentary and exercised murine breast cancer models. The primary research question was whether these models could accurately replicate tumour growth dynamics. A secondary objective was to assess the impact of exercise on model performance. Design This was a computational modelling study using secondary data from a published experimental study. The data were analysed retrospectively with no new data collection or interventions. Setting Data were derived from pre-clinical experimental research conducted in a controlled laboratory environment. Participants The dataset included tumour size measurements from 25 sedentary and 25 exercised female mice, the latter having completed an eight-week endurance training programme. Inclusion required complete tumour size data from day 7 to day 30 (sedentary) and day 32 (exercised). All available data were included. Primary and Secondary Outcome Measures The primary outcome was the coefficient of determination (R²), quantifying model accuracy. Secondary outcomes included the models’ ability to replicate observed tumour growth and the influence of exercise on model performance. Results Both models demonstrated high accuracy (R² > 0.94) in both groups. The quadratic model consistently outperformed the exponential model (R² = 0.9771 versus 0.9495 for sedentary; 0.9937 versus 0.9864 for exercised mice), better capturing the peak and deceleration of tumour growth. Conclusions These findings support the use of non-linear models in preclinical cancer modelling and highlight the quadratic model’s potential for broader applicability. Future work should incorporate biological variables to enhance predictive power.