Hippocampal sparing radiotherapy for glioblastoma patients: a planning study using volumetric modulated arc therapy

Glioblastoma multiforme (GBM) constitutes one of the most frequent and most lethal primary brain tumours, with an annual incidence rate of approximately 3 in 100,000 people in the US and only about 36 % of patients surviving the first year after initial diagnosis [1]. The standard treatment approach for GBM is trimodal, with maximal safe resection, if feasible, followed by radiochemotherapy and temozolomide-based maintenance chemotherapy. However, radiotherapy of the brain may have side-effects, which include radiation necrosis, cognitive impairment, and others [2‐5]. The dose to the hippocampus is associated with radiation induced memory impairment [6‐9]. Since modern radiotherapy techniques such as intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) allow for the delivery of highly conformal dose distributions, the idea arose to selectively spare the hippocampus during brain radiotherapy [10]. In this planning study, the potential to spare the contralateral hippocampus in radiotherapy for glioblastoma using the VMAT technique is investigated for 27 patients. Resulting dose volume histogram (DVH) statistics are compared to clinical standard 3D-conformal radiotherapy (3D-CRT) plans.

Datasets of 27 patients who had received 3D-CRT for glioblastoma at the department of radiation oncology at the Ludwig-Maximilians University of Munich (LMU) between 2010 and 2012 were included in this planning study. For all patients, the prescription to the planning target volume (PTV) was 60Gy in 30 fractions. The patient characteristics are shown in Table 1. Patients were positioned in supine position and immobilized with a thermoplastic mask. Normal structures like the eyes, the lenses, the optic nerves, the chiasm, the brain stem were contoured and designated as organs-at-risk (OARs). Expanded contours were created with safety margins of 3 mm around the brain stem and the chiasm and 5 mm around the optic nerve. The hippocampus was contoured on the T1 MRI sequence as described by Chera et al. [11]. The VMAT plans were optimised in the research treatment planning system (TPS) Hyperion V2.44 (equivalent to Elekta Monaco 5.1) which relies on the XVMC algorithm (X-ray voxel Monte Carlo [12]) for dose calculation. All VMAT plans were generated for a 6MV Elekta Axesse linear accelerator (LINAC) equipped with an Agility multileaf-collimator (Elekta, Stockholm, Sweden). The original clinical 3D-CRT plans were generated in the TPS Oncentra Masterplan (Elekta, Stockholm, Sweden) using a pencil beam dose algorithm. 3D-CRT plans were created for Siemens LINACs (Oncor (19 patients), Primus (4) and Mevatron M (4)) equipped with an 80 or 58 leaves MLC. Since the aim of this study was a comparison of clinical planning strategies rather than a technical comparison between VMAT and 3D-CRT, the use of two different TPS and dose calculation algorithms is not considered to bias the results significantly. In both systems, planning was performed independently. We supposed that slight differences between the beam models only have small influence on the optimal plan a dosimetrist can create using these models. The dose grid was 3×3×3mm³ for both VMAT and 3D-CRT plans.

In Fig. 2, a comparison of the isodose lines in the same transversal slice between the VMAT and the 3D-CRT plans is shown for one exemplary patient case. The reduction of the hippocampal dose exposure can be seen in the 30 % (=18Gy) isodose line.

The purpose of this planning study is to investigate the feasibility of sparing the contralateral hippocampus in radiotherapy for glioblastoma using VMAT. The hippocampal dose and the whole brain V _30Gy could be significantly improved, while maintaining the same dose constraints for other OARs. The other investigated parameters (target coverage, homogeneity index and conformity index) could be kept stable or even improved (dose homogeneity and conformity, whole brain dose). These results are in agreement with the results found in similar studies by Canyilmaz et al., who compared standard IMRT to hippocampal sparing IMRT and VMAT plans for 20 patients [17], and Marsh et al. who compared standard IMRT plans to IMRT plans with sparing of the contralateral neural stem cell compartment, hippocampus and limbic circuit for 5 patients [18]. The planning strategy in these studies differed from the concept used in our planning study. Both studies used a dose prescription of 46 Gy in 23 fractions followed by a sequential boost of 14Gy in 7 fractions (RTOG consensus) and concluded that a substantial reduction of the dose to the contralateral hippocampus is feasible with the used techniques without compromising other treatment parameters. In contrast to those studies, we also attempted to identify influencing factors for the hippocampal dose exposure. Studies suggest that it is safe to reduce the applied margins in IMRT for glioblastoma [19]. A study by Ali et al. evaluated the effect of reduced PTV margins on hippocampal dose [20]. They compared the standard margin of 2 cm (gross tumour volume (GTV) to clinical target volume (CTV)) with a reduced margin of 8 mm and reported a significant reduction of the bilateral hippocampal dose when applying the reduced margin concept. Concordantly, a significant correlation between the size of the PTV and the dose to the contralateral hippocampus was observed for VMAT in our planning study. This correlation was not significant for the original 3D-CRT plans, which might be explained by the fact that in the standard clinical protocol hippocampal sparing was not a treatment goal, even though it was attempted to create a dose distribution as conformal as possible, but the hippocampus was not designated as an OAR for which the dose exposure should be selectively reduced. This is also a limitation of this study. It is not suited to investigate the question to what extent VMAT is better in sparing the hippocampus compared to 3D-CRT from a purely technological perspective, since the planning objectives were not the same. However, it allows us to draw conclusions about how much reduction of the dose exposure can be achieved clinically, when the standard 3D-CRT planning strategy is replaced by a VMAT approach and hippocampal sparing is incorporated as a planning objective. An unexpected result of our study is that a parietal tumour localisation correlated with a higher hippocampal dose for VMAT, while a temporal localisation correlated with a lower hippocampal dose. A possible explanation may be the used planning strategy - for some angles of the non-coplanar arcs, the contralateral hippocampus is in the field when the target volume is located on the parietal lobe. Given that the subgroup of patients with parietal involvement was small (5 patients), which limits the statistical power of the test and that the p-value for the comparison between patients with and without temporal involvement was just under the significance threshold (p = 0.04), these particular results should be interpreted with caution. While there have been few studies investigating the feasibility of sparing the hippocampus in radiotherapy for glioblastoma, more research has been carried out concerning hippocampal sparing during whole-brain radiotherapy (WBRT) for patients with brain metastases. Several studies have investigated the feasibility of sparing both hippocampi during WBRT [21‐24]. Clinical data suggests that the approach is safe and favourable in terms of neurological outcome [25‐27]. For hippocampal sparing WBRT, also the influence of patient positioning on the hippocampal dose exposure has been investigated, and there is data suggesting that an inclined head angle might be beneficial [28]. In our study, it was not investigated whether an inclined head angle is able to further improve hippocampal dose in radiotherapy for glioblastoma, as all patients were positioned in the standard way. In the setting of WBRT for patients with brain metastases, sparing the hippocampus is not expected to affect the therapeutic ratio, since brain metastases rarely occur inside or close to the hippocampus [29, 30]. By contrast, sparing the hippocampus in radiotherapy for GBM is more controversial because of potential involvement of the subventricular zone (SVZ) in tumour genesis [31]. Higher doses to the ipsilateral SVZ have been shown to correlate with improved progression-free survival [32], and no SVZ contact of the tumour volume has been shown to be a predictor for long-term survival [33]. For this reason, only the contralateral hippocampus was spared in this planning study. In the cases with bilateral tumour, no involvement of the SVZ was presumed due to the small tumour volume on the spared side. Whether hippocampal sparing is possible without increasing the risk of relapse is a matter of ongoing debate [34, 35]. Possible benefits for patients with regard to neurological toxicity of radiotherapy for glioblastoma still need to be evaluated in future clinical trials. A retrospective analysis by Bodensohn et al. concluded that sparing might make sense for about 50 % of the patients receiving radiotherapy for glioblastoma [36].

Using VMAT instead of the standard 3D-CRT planning procedure, a significant reduction of the dose to the contralateral hippocampus (median reduction: 36 %) is feasible. Other treatment parameters can be improved as well or at least be maintained. Whether this approach is able to improve the outcome of patients with glioblastoma needs to be investigated in the setting of prospective clinical trials. A larger size of the PTV is a predictor for a higher hippocampal dose in the VMAT plans.