Proton RBE for in vivo model systems: LET and dose dependencies for early and late biological endpoints

Proton therapy applies a constant relative biological effectiveness (RBE) of 1.1, despite increasing evidence that RBE varies with physical and biological factors such as linear energy transfer (LET), dose, and tissue type. Variable RBE models from in vitro data may not reflect in vivo radiation responses. This study aimed to compile and analyze a comprehensive database of published in vivo RBE data from proton irradiations.

In vivo proton RBE studies were identified through a literature review. RBE values, doses, biological endpoints were extracted from 25 studies. Dose-averaged LET (LET_d) was estimated by reconstructing experimental setups in a treatment planning system. Data were stratified into three endpoint groups: early jejunal crypt regeneration, other early endpoints, and late endpoints. Regression fits were obtained for RBE as a function of LET and fraction dose (D_p).

174 RBE data points were compiled. Across endpoints, RBE increased with LET, with statistically significant slopes in most groups and a more pronounced dependency for late endpoints. A negative correlation between RBE and D_p was observed for crypt-, other early-, and pooled data, and was also retained when excluding experiments using kV X-rays as reference.

The present in vivo analysis supports proton RBE dependencies on LET, dose, and endpoint previously observed in vitro and highlights the need for standardized methodologies and more extensive in vivo data to enable reliable endpoint-specific analyses. The compiled database, with consistently derived LET estimates across studies, provides a solid foundation for continued in vivo RBE research.