2100 - Fitting Modified Microdosimetric Kinetic Model Parameters for Tumor Control Probability Estimation in Prostate Cancer Patients Receiving Photon or Carbon Ion Radiotherapy

02:30pm - 04:00pm PT

Hall F

Screen: 18

POSTER

Presenter(s)

Yu-Wen Hu, MD, PhD - Veterans General Hospital, Taipei, Taiwan

Y. W. Hu¹, W. C. Yang², and C. C. Chiou¹; ¹Department of Heavy Particles and Radiation Oncology, Taipei Veterans General Hospital, Taipei, Taiwan, ²VGHTPE, Taipei, Taiwan

Purpose/Objective(s): The modified microdosimetric kinetic model (mMKM) underpins clinical dose calculations in carbon ion radiotherapy (CIRT). It expresses survival similarly to the linear-quadratic (LQ) model but with an alpha term dependent on microdosimetric lineal energy distribution, accounting for saturation or overkill in high-LET radiation. However, its parameters are mainly derived from in vitro cell survival data. To predict clinical tumor and tissue responses, challenges remain in addressing inter-cellular and inter-tissue variability and bridging the in vitro–in vivo gap. Additionally, debate continues over applying photon radiotherapy expertise to CIRT. This study proposes a framework based on mMKM and LQ models to calculate and compare biologically effective doses (BED) across photon and particle therapies.

Materials/Methods: From the literature, data on 10-year biochemical failure-free probability as the endpoint for tumor control probability (TCP) were collected for low-risk prostate cancer patients treated with photon radiotherapy or CIRT under various dose-fractionation schemes, without androgen deprivation therapy. Additional data included patient numbers, fractions, and dose per fraction. BEDs for photon radiotherapy were estimated using the LQ model with the photon alpha /beta ratio. For CIRT, assuming y* is insensitive to other parameters, the carbon alpha /beta ratio was derived from the photon alpha /beta ratio and domain radius (rd) using mMKM, enabling BED calculation. TCP parameters (BED50, gamma50) and BED parameters (photon alpha /beta, rd) were then fitted. The main formulas are listed below:

BED for photon = N * d_photon * (1 + (d_photon / (alpha_photon / beta)))
BED for particle = N * ((alpha_photon / beta + (y_particle* - y_photon*) / (rho * pi * r_d^2)) + d_particle) * (d_particle / (alpha_photon / beta))
TCP = 1 / (1 + exp(4 * gamma_50 * (1 - (BED / BED_50))))

Results: Seven photon and three carbon dose-fractionation schemes were included for fitting, ranging from 35 Gy/5 Fx to 81 Gy/45 Fx for photons and from 57.6 Gy(RBE)/16 Fx to 66 Gy(RBE)/20 Fx for CIRT, with TCP values ranging from 50% to 95%, with a total of 1,010 patients. The fitted parameters and their 95% confidence intervals were as follows: gamma50 = 1.185 [0.011, 2.359], BED50 = 134.882 Gy [88.151, 181.613], alpha /beta ratio = 1.381 Gy [0.295, 2.467], and rd = 0.373 µm [0.294, 0.452]. For goodness-of-fit, the R-square is 0.451. After excluding one outlier cohort (n=31), the R-square is 0.844, indicating a good fit of the TCP curve for both photon radiotherapy and CIRT clinical data.

Conclusion: This study presents a promising workflow for identifying tumor- and tissue-specific parameters, not only improving clinical dose estimation in CIRT but also establishing a unified framework for calculating BED across photon, proton, and carbon ion therapies. This approach enables comparisons across different modalities and supports the development of multimodal treatment strategies.