Objectives: The aim of this study was to compare patient-specific radiobiological parameters to population averages in predicting clinical outcome after radiotherapy using a tumor control probability (TCP) model based on BED.
Methods: A previously published material of forty-six head and neck carcinomas with individually identified radiobiological parameters; SF2 and Tpot, and known tumor size was investigated. These patients had all been treated with external beam radiotherapy and the majority had also received brachytherapy. TCP for each individual based on BED using patient-specific radiobiological parameters was compared to TCP based on BED using average radiobiological parameters (α = 0.3 Gy-1 and Tpot = 3 days).
Results: Forty-three patients remained in the final analysis. There was only a weak trend for increasing local tumor control with increasing BED in both groups. However, when TCP was calculated the use of patient-specific parameters was better to identify local control correctly. Sensitivity and specificity for tumor-specific parameters were 63% and 80%, respectively. The corresponding values for population-based averages were 0% and 91%, respectively. Positive predictive value was 92% when tumor-specific parameters were used compared to 0 % for population-based. A receiver operating characteristic (ROC) curve confirmed the superiority of patient-specific parameters over population averages in predicting local control.
Conclusions: Individual radiobiological parameters are better than population derived averages when used in a mathematical model to predict tumor control probability after curative radiotherapy in head and neck carcinomas.
Advances in knowledge: TCP based on individual radiobiological parameters are better than TCP based on population based averages.
2013. Vol. 86, no 1025, 20130015