Temporal lobe reactions after radiotherapy with carbon ions: incidence and estimation of the relative biological effectiveness by the local effect model.

Purpose: To identify predictors for the development of temporal lobe reactions (TLR) after carbon ion radiation therapy (RT) for radiation-resistant tumors in the central nervous system and to evaluate the predictions of the local effect model (LEM) used for calculation of the biologically effective dose.

Methods And Materials: This retrospective study reports the TLR rates in patients with skull base chordomas and chondrosarcomas irradiated with carbon ions at GSI, Darmstadt, Germany, in the years 2002 and 2003. Calculation of the relative biological effectiveness and dose optimization of treatment plans were performed on the basis of the LEM. Clinical examinations and magnetic resonance imaging (MRI) were performed at 3, 6, and 12 months after RT and annually thereafter. Local contrast medium enhancement in temporal lobes, as detected on MRI, was regarded as radiation-induced TLR. Dose-volume histograms of 118 temporal lobes in 59 patients were analyzed, and 16 therapy-associated and 2 patient-associated factors were statistically evaluated for their predictive value for the occurrence of TLR.

Results: Median follow-up was 2.5 years (range, 0.3-6.6 years). Age and maximum dose applied to at least 1 cm(3) of the temporal lobe (D(max,V - 1 cm)3, maximum dose in the remaining temporal lobe volume, excluding the volume 1 cm(3) with the highest dose) were found to be the most important predictors for TLR. Dose response curves of D(max,V - 1 cm)3 were calculated. The biologically equivalent tolerance doses for the 5% and 50% probabilities to develop TLR were 68.8 ± 3.3 Gy equivalents (GyE) and 87.3 ± 2.8 GyE, respectively.

Conclusions: D(max,V - 1 cm)3 is predictive for radiation-induced TLR. The tolerance doses obtained seem to be consistent with published data for highly conformal photon and proton irradiations. We could not detect any clinically relevant deviations between clinical findings and expectations based on predictions of the LEM.