Translational tumor control probability modeling for NSCLC: A two-dimensional maximum-likelihood framework
Ryoichi Hinoto1,2, Takeji Sakae2,3, Kenta Takada4
1Department of Radiation Oncology, Saitama Red Cross Hospital, Saitama, Saitama, Japan.
Medical Physics
|July 31, 2026
Summary
This study developed a new 2D framework for tumor control probability (TCP) modeling in early-stage non-small cell lung cancer (NSCLC). The model accurately accounts for dose uncertainties and reveals inter-patient radiosensitivity as the main driver of TCP steepness.
Area of Science:
- Radiation Oncology
- Medical Physics
- Cancer Modeling
Background:
- Traditional 1D tumor control probability (TCP) models for early-stage non-small cell lung cancer (NSCLC) often neglect dose uncertainties, leading to errors-in-variables (EIV) bias.
- This bias can skew estimates of the dose required for 50% TCP (TCD50) and affect the understanding of TCP curve steepness.
- The mechanistic origins of TCP steepness and its link to radiobiology require further elucidation.
Purpose of the Study:
- To create a mechanistic TCP framework that addresses EIV bias by modeling uncertainties in both dose and local control (LC).
- To differentiate the contributions of clonogenic heterogeneity and inter-patient radiosensitivity to TCP steepness.
- To validate the translational consistency of the developed framework.
Main Methods:
- Fitted cell-survival data from A549 NSCLC cells using saturable repair and repairability (SRR), universal survival curve (USC), and linear-quadratic (LQ) models.
- Employed a 2D maximum-likelihood framework for clinical TCP modeling using data from 34 early-stage NSCLC studies.
- Quantified TCP steepness by measuring inter-patient variability and decomposed it into clonogenic heterogeneity and radiosensitivity components.
Main Results:
- The SRR and USC models accurately fit cell-survival data; the LQ model required inflated parameters.
- The 2D framework demonstrated that effective dose-axis uncertainty is greater than LC-axis uncertainty.
- Inter-patient radiosensitivity was the dominant factor (85%-97%) in TCP steepness, with clonogenic heterogeneity contributing minimally (3%-15%).
Conclusions:
- The proposed 2D maximum-likelihood framework effectively mitigates EIV bias and connects clinical TCP to mechanistic cell-survival models.
- In early-stage NSCLC, inter-patient radiosensitivity, not clonogenic cell number variability, primarily determines TCP steepness.
- This framework offers a robust foundation for future TCP modeling and clinical dose-prescription optimization.
