Shrinking-field concept spars the periventricular region in proton therapy of gliomas

Recent findings suggest an increased radiosensitivity of the cerebral periventricular region (PVR; Eulitz 2019, Harrabi 2019) in primary brain tumor patients. Shrinking-field concepts (SFC) are used in proton therapy, e.g. of the brain, to spare normal brain tissue. Since there is a correlation between treatment associated brain injury and dose / LET (Peeler 2019), the evaluation of the impact of SFC in PVR dose and LET sparing is necessary. We compared observed radiation-induced brain injuries after proton therapy for glioma patients treated either conventional (CC) or with SFC, and introduce an approach for PVR-adapted proton treatment planning.

All grade II and III glioma patients treated between 2014 and 2018 with (adjuvant) proton radio(chemo)therapy to a total dose (D) of 54-60 Gy(RBE) were considered for analysis. 33% of the patients received SFC (with sequential- or simultaneously-integrated proton boost (SIB)) with a prescribed dose reduction of 6-10 Gy(RBE) in the outer part of the target volume. Contrast enhancements (CE) on follow-up MRI (fuMRI) diagnosed as treatment-related brain injury lesions (symptomatic or clinically silent) were traced back to the fuMRI of first appearance, delineated and deformably co-registered to the planning CT. The distance between CE lesions to the cerebral ventricles was determined. The PVR was estimated as a 4 mm band around the segmented cerebral ventricles. The PVR volume VX% receiving more than X% of prescribed dose as well as D and LET within the CE lesions were calculated. Brain injury-free survival (in/outside PVR) was derived in a Kaplan-Meyer analysis. For a SIB patient with a CE lesion 10 months after proton therapy, PVR sparing treatment planning was performed.

For the SFC and UD patient cohort, the observed CE lesions clustered in direct proximity to the cerebral ventricles with median distances of 2.6 mm and 2.3 mm, respectively. Mean dose at the CE lesion was 54.4±3.5 Gy(RBE) and 56.4±4.3 Gy(RBE) and the corresponding LET value 2.7±0.4 keV/µm and 3.2±0.9 keV/µm, respectively. The SFC reduced V100% and V90% in the PVR by 11.3% and 35.3%, respectively. No significant difference was found in one-year symptomatic (p = 0.15) and asymptomatic (p = 0.75) injury free survival. An average CE lesion dose of 55 Gy(RBE) was derived within PVR tissue for all patients and used as PVR tolerance dose. Incorporating the PVR as OAR in treatment plan optimization reduced the V55Gy within the CE lesion and PVR contour by 19.1% and 2.0%, respectively, without compromising target coverage, plan robustness or clinical dose constrains (Fig. 1).

For both treatment concepts, late brain injury showed a remarkably similar proximity to the cerebral ventricles and dependence on dose and LET. The SFC spares parts of the PVR from high dose and has the potential to improve treatment outcome. However, significant reduction of brain toxicity may require a dedicated PVR dose sparing planning strategy minimizing V55Gy.