Program Information
Computational Model of Observed Heterogeneity in Bone Metastases
M Turk1*, U Simoncic1,2 , A Roth3 , D Valentinuzzi1,2 , R Jeraj1,2,3 , (1) Faculty of Mathematics and Physics, Ljubljana, Slovenia, (2) Jozef Stefan Institute, Ljubljana, Slovenia, (3) University of Wisconsin, Madison, WI
Presentations
SU-I-GPD-T-655 (Sunday, July 30, 2017) 3:00 PM - 6:00 PM Room: Exhibit Hall
Purpose: Intrinsic (pre-existent) and acquired (drug induced) resistance lead to heterogeneity in observed treatment response. This work aims to describe observed heterogeneity in metastatic prostate cancer patients with computation model.
Methods: A population model was used to simulate cellular dynamics in individual metastases based on data from eighteen patients with 686 lesions (min. 13/patient). We assumed that metastases initially consist only of drug-sensitive cells. Different biological parameters (BP) included rate of proliferation, mutation of drug-sensitive into drug-resistant cells, and treatment efficiency (TE, describing killing of drug-sensitive cells). We assumed that number of cells in metastases is proportional to bone lesion burden (SUVtotal), extracted from [F-18]-NaF PET baseline and mid-treatment (acquired before the therapy and 12 weeks after the start of therapy, respectively) scans. First, a uniform set of BP for lesions within a patient, but different across patients was assumed. Subsequently, variation of TE between lesions was allowed. Simulated SUVtotal change for individual lesions and total change of tumor burden for individual patients were compared with measured data.
Results: With uniform BP within patients good agreement was found between model prediction and measured total change of tumor burden for individual patients, with 6% (range 0.003%-41%) mean deviation. However, model poorly predicted SUVtotal change for individual lesions, with 70% mean deviation from measured value. By allowing TE to vary between lesions, mean total change of tumor burden differed between the model and measured data by 3.5% (range 0-29%). Importantly, model predictions of SUVtotal change for individual lesions improved, with 94% of lesions having SUVtotal deviations of less than 5% at mid-treatment.
Conclusion: In both cases (uniform BP or varying TE) model successfully describes heterogeneity of total change of tumor burden between patients. However, by varying TE between lesions, the model captured not just heterogeneity between patients, but also within patients.
Contact Email: