Room: AAPM ePoster Library
Variations of the relative biological effectiveness (RBE) of proton therapy with linear energy transfer (LET(d)) are predicted from pre-clinical experiments. However, RBE values may substantially vary for different clinical endpoints. In this study, we assess the feasibility of integrating LET(d) into normal tissue complication probability (NTCP) models for patient-rated head and neck cancer toxicity.
We included 32 consecutive head and neck cancer patients treated with proton therapy. Physical dose (D) and product of D and LET(d) (D·LET(d)) were calculated on weekly verification CTs using the Raystation v6R Monte Carlo algorithm and accumulated on the planning CT.
The means and covariance matrix of the accumulated D and D·LET(d) of all relevant organs-at-risk (OARs) were used to simulate 2.500 data sets of different sizes. For each dataset, an attempt was made to make a multivariable NTCP model including D and D·LET(d) parameters of the associated OAR for xerostomia, tube feeding and dysphagia. The probability of creating an NTCP model with statistically significant parameters (i.e. power) is calculated as a function of the simulated sample size for various RBE models.
D and D·LET(d) were highly correlated (figure 1). This is likely explained by a lack of variation in LET(d) due to the standardized beam setup. The sample size required to have a power of at least 80% to show an independent effect of D·LET(d) on toxicity is over 5.000 patients (figure 2) for all toxicities.
It is not feasible to model NTCP with both Dphys and D·LET(d) for current clinical practice. LET(d) treatment plan optimization is likely to reduce the correlation which may make relating LET(d) to toxicity possible for some patients. Alternatively, functional imaging may provide spatial information on biological damage which can be directly related to high LET(d) regions within OARs.