Estimation of Tumor Control Probability of Carbon Ion Radiotherapy Using Cancer Type-specific Sensitivity Data

Background/Aim: This study aimed to evaluate whether the predictive performance of tumor control probability (TCP) by carbon ion radiotherapy (CIRT) could be improved by incorporating cancer type-specific carbon ion sensitivity data obtained from non-small cell lung cancer (NSCLC). Materials and Methods: A TCP model based on the linear-quadratic (LQ) formalism and fractionated irradiation was employed, with σ defined as an index of inter-tumor heterogeneity on radiosensitivity. The LQ model parameters α and β for NSCLC were obtained from 13 cell lines subjected to carbon ion irradiation under clinically relevant conditions, followed by clonogenic assays. The human salivary gland (HSG) cell line was used as a control. The α and β values were incorporated into the TCP model, and agreement with the clinical data (i.e., the local control rates for 48 NSCLCs treated with CIRT, as previously reported) was evaluated using the coefficient of determination (R2) derived from least-squares fitting. Results: When σ was set to 0.20, the NSCLC-derived TCP curve showed better agreement with the clinical data than the HSG cell-derived TCP curve (R2: 0.72 vs. 0.53, respectively). Similarly, when σ was set to 0.15, the NSCLC-derived curve showed better agreement with the clinical data (R2: 0.56 vs. 0.38, respectively). Furthermore, when weighting factors of 1, 3, and 10 were applied to the two data points with TCP=1.0, the NSCLC-derived TCP curve showed consistently better agreement with the clinical data than the HSG cell-derived TCP curve (R2: 0.74 vs. 0.55, 0.78 vs. 0.63, and 0.79 vs. 0.65, respectively). Conclusion: Incorporation of cancer type-specific carbon ion sensitivity data can improve the predictive performance of TCP modeling compared with the conventional HSG cell-based approach.