Hippocampal sparing approach in fractionated stereotactic brain VMAT radio therapy: A retrospective feasibility analysis

Abstract Volumetric Modulated Arc Therapy (VMAT) techniques for fractioned stereotactic brain radiotherapy (FSBRT) can achieve highly conformal dose distribution to intracranial lesions. However, they can potentially increase the dose to hippocampus (H) causing neurocognitive toxicity during the first four months after irradiation. The purpose of this study was to assess the feasibility of hippocampal‐sparing (HS) treatment plans in 22 patients with brain metastasis treated with VMAT technique. Firstly, we retrospectively analyzed hippocampal doses in all 22 VMAT original (not hippocampal‐sparing, NHS) plans. Plans with hippocampal dose exceeding constraints (9 out of 22) were re‐planned considering dose constraints on the hippocampus (H) and on hippocampal avoidance zone (HAZ) generated using 5 mm isotropic margin to the hippocampus. Conformity (CI) and homogeneity indexes (HI) on the target and MUs, were maintained as close as possible to the original plans. Mean CINHS and CIHS obtained were: 0.79 ± 0.11 and 0.81 ± 0.10, respectively (P = 0.75); mean HINHS and HIHS were 1.05 ± 0.02 and 1.04 ± 0.01 respectively (P = 0.72). In both sets of plans, the mean MU values were similar: 1033 ± 275 and 1022 ± 234 for NHS and HS respectively. In HS plans, the mean hippocampal dose was decreased by an average of 35%. After replanning, the Dmax (21.3 Gy) for HAZ and H was met by 45% (4/9) and 78% (7/9) of the NHS plans, respectively. The worst results were obtained for cases with target volumes extention closer than 12 mm to H, because of the difficulty to spare hippocampus without compromising target coverage. After replanning D40% constraint value (7.3 Gy) was met by all the 9 NHS plans. In conclusion, this study suggests that an hippocampal‐sparing approach to FSBRT is feasible resulting in a decrease in the dose to the hippocampus without any loss in conformity or increase in treatment time.


| INTRODUCTION
Hippocampal injuries play a fundamental role both in short and longterm memory loss and cognitive impairment. 1,2 Cranial irradiation can induce hippocampus damage, as suggested by some studies. [3][4][5][6][7][8][9] In particular, cognitive impairment caused by whole brain irradiation (WBRT) has been investigated. 10 These studies suggest that radiation-induced neurocognitive toxicity occurs after irradiating neural stem cells of the hippocampus, potentially compromising patients quality of life (QoL).
WBRT has long been considered the mainstay treatment for patients with multiple brain metastases; nowadays, due to innovative technologies, fractioned stereotactic brain radiotherapy (FSBRT) and radiosurgery (SRS) can represent valid alternative therapeutic options to WBRT 11,12 allowing a better sparing of organs at risk, an improved outcome, and an increase in life expectancy; as a consequence, late onset radiation induced neurological sequels on hippocampi could be revealed in the course of life. Moreover, being the hippocampus very often close to the target, it could receive very high doses in extreme hypofractionated FSBRT treatments. Despite the large number of patients treated with these techniques, hippocampus is not routinely considered among organs at risk and the few clinical data available are not able to establish the correlation between dose on the hippocampus and cognitive effects. Results of the phase II RTOG 0933 study, 13 show that some benefit in neuro-cognitive functioning is achieved by hippocampal-sparing in brain radiotherapy; however, phase III trial studies are necessary to validate the approach and confirm these findings. In spite of the paucity of clinical data, many authors focus on the feasibility of hippocampal-sparing (HS) treatment plans. They applied the HS approach to WBRT followed by a radio-surgical boost or WBRT and simultaneous integrated boost treatments using highly conformal techniques such as IMRT, helical tomotherapy or VMAT [14][15][16][17][18][19] demonstrating that HS plans were effectively able to spare the commonly delineated OARs including the hippocampus, while maintaining the same dose coverage and homogeneity of target volumes as the original ones. In these studies, the reduction in mean hippocampus dose was used as parameter to evaluate the appropriateness of the HS plans.
With regard to issues mentioned above, we conducted a retrospec-  CBCT was performed daily before each treatment session to evaluate set-up errors. CBCT acquisition volume (clip-box) was determined to include whole PTV and OARs. The 3D-3DCBCT-CT planning scan co-registration was performed using the Gray level algorithm.

2.C | Retrospective hippocampus study
The Hippocampus (H) was retrospectively delineated on the original plans by a radiation oncologist on gadolinium contrast-enhanced T1weighted MRI. Delineation was performed on axial images using the RTOG 0933 atlas 20 as reference. Afterwards, the contours in sagittal, coronal, and axial projections were revised by a neuroradiologist. A hippocampal avoidance zone (HAZ) was generated adding an isotropic 5 mm margin. The maximum dose (D max ) and dose to 100% of hippocampus volume (D 100% ) were documented for all VMAT plans, according to the RTOG 0933 dosimetric compliance criteria. 13 Moreover, the dose to 40% of hippocampus volume (D 40% ) was considered. 7 Since RTOG hippocampal constraints were defined for a prescribed dose of 30 Gy in ten fractions, they were converted to biologically equivalent doses in 2 Gy fractions (EQD2). As the D 40% constraint was more restrictive than D 100% , the latter was no longer considered in this study.
Due to the alternative fractionation FSBRT schema, the analysis of DVHs was performed, for all the OARs, reconverting all dose values to equivalent doses in 2-Gy fractions assuming an a/b ratio of 2 Gy.
In case original VMAT plans exceeded hippocampal constraints (non-hippocampal-sparing, NHS), HS plans were generated. The HS plans were elaborated following the same arc optimization systematic strategy as NHS regarding objectives, weights, and MU limit.
Hippocampal constraints have been applied on HAZ. In addition, the distance between HAZ and the adjacent lesion was measured.
The plan analysis included the evaluation of all the other OARs doses.
Dose delivery accuracy of all plans was assessed by measuring the 3D dose distributions with OCTAVIUS detector 729 device (PTW, Freiburg, Germany) and the agreement between measured and calculated dose profiles was checked using the gamma passing rate of 3% local dose (LD) and 3 mm distance to agreement (DTA) with a 10% threshold.

2.D | Dosimetric evaluation of VMAT plans
Dosimetric evaluation of both NHS and HS plans was carried out by calculating conformity and homogeneity indexes. Healthy brain mean dose as a surrogate of the integral dose for said tissue 21 was also evaluated.
The conformity index CI 22 was defined as: where V T,pi was the volume of the target covered by the prescription isodose, V T was the target volume, and V pi was the volume of the prescription isodose. The homogeneity index (HI), 23  To evaluate healthy brain mean dose (NTMD), an additional structure [called non-tumor (NT)] consisting of brain minus PTV was created. From DVHs, the mean doses were then extracted both for NHS (NTMD NHS ) and HS (NTMD HS ) plans.

| RESULTS
Twenty-two patients with 38 brain metastases were evaluated.
Mean volumes for PTV and H were 5.7 cm 3 (range 1-23.6) and 4.5 cm 3 (range 2-6) respectively. The closest distance between the lesion and the hippocampal surface was found to be 2 mm (mean distance = 11.7 AE 6 mm

| DISCUSSION
The role of the hippocampus in memory function is well-established in literature 1,2 ; although some studies affirm that cranial irradiation may damage neurogenic stem cells located in the subgranular layer of the hippocampal dentate gyrus inducing neurocognitive toxicity, [3][4][5]24,25 dose constraints for hippocampus have not yet been elucidated, making it difficult to establish the potential benefits of hippocampal-sparing approach. Some constraints were derived from the initial results of an analysis of patients affected by brain metastasis undergoing WBRT (D max < 16 Gy and D 100% < 9 Gy at 3 Gy per fraction for a total dose of 30 Gy) and low/high grade gliomas receiving fractionated stereotactic radiation therapy (D 40% < 7.3 Gy on bilateral hippocampus at 2 Gy per fraction). 7,10,13 D max and D 100% constraints were also adopted by Pokhrel et al. 26 who retrospectively investigated the plan quality and accuracy of using hippocampal-sparing intensity modulated arc therapy in WBRT treatments.
In another works, the mean dose was considered as dose reference to optimize plans in a hippocampal-sparing approach when highly conformal techniques such as IMRT, helical tomotherapy or VMAT are used. 14 suggest the "non-applicability" of constraints to HAZ observing that sparing HAZ (hippocampus + 5 mm margin) poses a theoretical risk of disease progression in the margin region. 6 Taking into account these issues, we evaluated the feasibility of using VMAT to deliver FSBRT with hippocampal avoidance. Our results show that plan quality after hippocampal sparing is still well within the published standards of conformity and homogeneity.
Mean NTMD NHS value was found comparable with mean NTMD HS (P = 0.73). This means that changing treatment parameters passing from NHS to HS plans maintaining the same target coverage does not increase the total energy deposited to the healthy brain.
In our retrospective analysis, we found that 9 of 22 cases had D max above the limits; 45% of them were recovered for HAZ (4/9 plans); one more plan was recovered for H (they passed from 6/9 to In summary, our data suggest that for FSBRT performed with VMAT technique (a) targets adjacent to hippocampus could only partially benefit from an hippocampal-sparing approach, in case good conformity and target coverage is needed considering D max as a rigid constraint in the optimization tool; and (b) the doses undergoing a more important decrease (mean dose and D 40% ), may benefit much more from HS approach. On the other hand, our HS plans were optimized with respect to D max and D 40% . In a future work, we intend to investigate the impact of D 40% and the mean dose as constraints in the optimization tool in HS approach and use the constraint on D max only in selected cases. In any case, the total number of MU did not change significantly among NHS and HS plans, so we think that an hippocampal-sparing approach should always be attempted.

| CONCLUSION
This study suggests that hippocampal-sparing approach in fractionated stereotactic brain radiotherapy VMAT treatments is feasible, resulting in an overall decreased dose to the hippocampus. HS plans maintain the same target conformity and homogeneity, the same mean dose to surrounding healthy tissues and the same treatment time of the original plans. In case of hippocampal distance from the target larger than 12 mm, all the considered dose constraints are respected. Anyhow, a reduction in 35% has been obtained for the mean dose and D 40% . Although safe threshold doses for the hippocampus have not been defined yet, it is strongly advisable to delineate the hippocampus and put in practice all the necessary strategies to reduce doses especially in patients with a reasonable life expectancy.

CONFLI CT OF INTEREST
The authors declare that they have no conflict of interest.