Comparison of Photon Isoeffective Dose Models Based on In Vitro and In Vivo Radiobiological Experiments for Head and Neck Cancer Treated with BNCT

Abstract

Boron neutron capture therapy (BNCT) is a treatment modality for cancer that involves radiations of different qualities. A formalism that proved suitable to compute doses in photon-equivalent units is the photon isoeffective dose model. This study addresses the question whether considering in vitro or in vivo radiobiological studies to determine the parameters involved in photon isoeffective dose calculations affects the consistency of the model predictions. The analysis is focused on head and neck squamous cell carcinomas (HNSCC), a main target that proved to respond to BNCT. The photon isoeffective dose model for HNSCC with parameters from in vitro studies using the primary human cell line UT-SCC-16A was introduced and compared to the one previously reported with parameters from an in vivo oral cancer model in rodents. Both models were first compared in a simple scenario by means of tumor dose and control probability calculations. Then, the clinical impact of the different dose models was assessed from the analysis of a group of squamous cell carcinomas (SCC) patients treated with BNCT. Traditional dose calculations using the relative biological effectiveness factors derived from the SCC cell line were also analyzed. Predictions of tumor control from the evaluated models were compared to the patients' outcome. The quantification of the biological effectiveness of the different radiations revealed that relative biological effectiveness/compound biological effectiveness (RBE/CBE) factors for the SCC cell line are up to 20% higher than those assumed in clinical BNCT, highlighting the importance of using experimental data intimately linked to the tumor type to derive the model's parameters. The comparison of the different models showed that photon isoeffective doses based on in vitro data are generally greater than those from in vivo data (∼8-16% for total tumor absorbed doses of 10-15 Gy). However, the predictive power of the two models was not affected by these differences: both models fulfilled conditions to guarantee a good predictive performance and gave predictions statistically compatible with the clinical outcome. On the other hand, doses computed with the traditional model were substantially larger than those obtained with both photon isoeffective models. Moreover, the traditional model is statistically rejected, which reinforces the assertion that its inconsistencies are intrinsic and not due to the use of RBE/CBE factors obtained for a tumor type different from HN cancer. The results suggest that the nature of the radiobiological data would not affect the consistency of the photon isoeffective dose model in the studied cases of SCC head and neck cancer treated with BPA-based BNCT.