Purpose: Estimations of complications from stereotactic radiosurgery usually rely simply on dose-volume or dose-diameter isoeffect curves. Due to the sparse clinical data available, these curves have typically not considered the target location in the brain, target histology, or treatment plan conformality as parameters in the calculation. In this study, a predictive model was generated to estimate the probability of cranial neuropathies as a result of acoustic schwannoma radiosurgery.
Methods and materials: The dose-volume histogram reduction scheme was used to calculate the normal tissue complication probability (NTCP) from brainstem dose-volume histograms. The model's fitting parameters were optimized to provide the best fit to the observed complication data for acoustic neuroma patients treated with stereotactic radiosurgery at the University of Florida. The calculation was then applied to the remainder of the patients in the database.
Results: The best fit to our clinical data was obtained using n = 0.04, m = 0.15, and alpha/beta = 2.1 Gy(-1). Although the fitting parameter m is relatively consistent with ranges found in the literature, both the volume parameter, n, and alpha/beta are much smaller than the values quoted in the literature. The fit to our clinical data indicates that brainstem, or possibly a specific portion of the brainstem, is more radiosensitive than the parameters in the literature indicate, and that there is very little volume effect; in other words, irradiation of a small fraction of the brainstem yields NTCPs that are nearly as high as those calculated for entire volume irradiation. These new fitting parameters are specific to acoustic neuroma radiosurgery, and the small volume effect that we observe may be an artifact of the fixed relationship of acoustic tumors to specific regions of the brainstem. Applying the model to our patient database, we calculate an average NTCP of 7.2% for patients who had no cranial nerve complications, and the average NTCP for was 66% for patients who sustained a cranial neuropathy. For the entire patient population, the actual percentage of patients suffering either facial or trigeminal neuropathy was 14.7%, whereas the calculated average NTCP was 14.8%.
Discussion: NTCP calculations using brainstem dose-volume histograms can be used to estimate the rate of cranial neuropathies from acoustic neuroma radiosurgery. More clinical data and further study will lead to refinement of the model with time.