A Study of Dose Painting Radiotherapy Guided by Three-Dimensional Arterial Spin labeling Perfusion Magnetic Resonance Imaging for Non-Enhancing Low-Grade Gliomas

Background: The purpose of this study was to investigate the feasibility and dosimetric characteristics of dose painting for non-enhancing low-grade gliomas (NE-LGGs) guided by three-dimensional arterial spin labeling (3D-ASL) perfusion magnetic resonance imaging. Methods: Eighteen patients with NE-LGGs were enrolled. 3D-ASL, T2 uid attenuated inversion recovery (T2 FLAIR) and contrast-enhanced T1-weight magnetic resonance images were obtained. The gross tumor volume (GTV) was delineated on the T2 FLAIR. The hyper-perfusion region of the GTV (GTV-ASL) was determined by 3D-ASL, and the GTV-SUB was obtained by subtracting the GTV-ASL from the GTV. The planning target volume (PTV), PTV-ASL and PTV-SUB were developed by expanding the external margins of the GTV, GTV-ASL and GTV-SUB, respectively. Three plans were established for each patient: in plan 1, the traditional homogeneous prescription dose to the PTV was 45-60 Gy; in plan 2 and plan 3, the dose to the PTV-ASL increased by 10-20% based on plan 1, without the maximum dose constraint to the PTV-ASL in plan 3. The dosimetric differences among the three plans were compared. Results: Compared with plan 1, the dose to 2% (D 2% ), 98% (D 98% ) and 50% (D mean ) of PTV-ASL volumes increased by 14.67%, 16.17% and 14.31% in plan 2 and 19.84%, 15.52% and 14.27% in plan 3, respectively (P < 0.05); the D 2% of the PTV, PTV-SUB increased by 11.89% and 8.34% in plan 2, 15.89% and 8.49% in plan 3, respectively (P< 0.05). The PTV coverages were comparable among the three plans (P > 0.05). In Plan 2 and plan 3, the conformity indexs decreased by 18.60% and 12.79%; while the homogeneity indexs increased by 1.43


Background
Low-grade gliomas (LGGs) are the most common brain tumors in children, accounting for approximately 15% of all gliomas [1,2]. The most common treatment method for LGGs is maximum surgical resection, followed by postoperative adjuvant chemotherapy and other comprehensive treatments. However, radiotherapy is usually the rst option for patients who are unable to undergo surgery. Both postoperative adjuvant radiotherapy and radical radiotherapy play an important role in the treatment of LGGs.
Nevertheless, there is much disagreement about how to determine tumor targets in radiotherapy for nonenhancing low-grade gliomas (NE-LGGs) with computed tomography (CT) or magnetic resonance imaging (MRI). The vascular wall components of NE-LGGs are similar to those of healthy brain tissue. NE- LGGs have an intact blood-brain barrier, and the macromolecular contrast agent cannot pass the blood-brain barrier. The lesion appears on CT images as hypodense and on T1-and T2-weighted (T1W and T2W) MRI as hypo-and hyperintense areas, respectively, with visible edemas [3,4]. In traditional methods, in order to ensure the target coverage of NE-LGGs radiotherapy, the abnormally hyperintense area (edema area) on T2 uid attenuated inversion recovery (T2 FLAIR) is usually used as the standard for the delineation of tumor target region, which often has too large range and increases the risk of radiation damage to surrounding healthy tissues. Tatekawa et al. [5] found that there was no signi cant correlation between the high signal volume on FLAIR and overall survival (OS). Therefore, accurate delineation of tumor targets for NE-LGGs is critical for improving the curative effect and reducing injury.
As a non-invasive blood perfusion imaging technology, three-dimensional arterial spin labeling (3D-ASL) generates an image by magnetically labeling water molecules as an endogenous tracer as they travel through blood. Imaging does not require the destruction of the blood-brain barrier and can re ect microscopic changes in tissue blood perfusion and micro-vessel density [6,7]. 3D-ASL perfusion imaging can effectively distinguish intracranial tumors from non-neoplastic lesions, and the tumor cerebral blood ow (CBF) value is signi cantly negatively correlated with survival [8]. Teng et al. showed that glioma puncture guided by 3D-ASL had a higher success rate than conventional MRI. In this study, 6 cases of low-enhancing or non-enhancing gliomas were successfully punctured, indicating that 3D-ASL has great potential in guiding the determination of tumor targets for NE-LGGs [9].
3D-ASL is widely used in the diagnosis, grading and e cacy evaluation of gliomas [8, 10,11], while its application in guiding NE-LGG radiotherapy has rarely been reported. Therefore, we studied the segmentation of tumor sub-volumes for NE-LGGs and dose painting of individualized radiotherapy based on 3D-ASL.

Patient characteristics
A total of 18 patients that had been pathologically diagnosed with LGGs at Shandong Cancer Hospital from December 2018 to May 2021 were retrospectively analysed. These patients showed no enhancement on contrast-enhanced T1-weighted MR images. The ages of the patients ranged from 5 to 74 years, with an average age of 42 ± 21 years, and there were 12 (67%) males and 6 (33%) females. Patient characteristics are presented in Table 1 (included at the end of the document text le). Our retrospective study followed medical ethics guidelines, and the Institutional Review Board of Shandong Cancer Hospital approved this study. All patients signed informed consent forms. We protected patient privacy and excluded patient identi cation information from our analysis.

Simulation
CT simulation: For simulations, a CT scanner (Brilliance Big Bore, Philips, Netherlands) with a slice thickness of 3 mm and an interslice gap width of 3 mm was used, scanning from the foramen magnum to the end of the skull in a caudocranial direction. scan parameters for each sequence are described in Table 2.

Target volume de nition
All MRI images were imported into MIM Maestro software (6.8.8, USA). Two senior radiologists delineated the tumor target volume and organs at risks (OARs), and another senior radiologist con rmed the target region and modi ed it as needed. Axial T2 FLAIR sequences were used to contour the hyperintense signal abnormality as the gross tumor volume (GTV). The GTV-ASL was de ned as a high perfusion region under the guidance of 3D-ASL sequence that was free of cysts, necrosis, calci cations, haemorrhage, and large vessels. The GTV-SUB was obtained by subtracting the GTV-ASL from the GTV.
The planning target volume (PTV) and PTV-ASL were generated by adding a 0.5 and 0.3 cm margin to GTV and GTV-ASL, respectively. The PTV-SUB was obtained by subtracting the PTV-ASL from the PTV. OARs included the bilateral eyeballs, bilateral lenses, bilateral optic nerves, optic chiasma and brain stem. A schematic diagram is shown in Fig. 1.

Treatment planning
All treatment plans were designed in the Eclipse (Version 15.6, Varian, USA) treatment planning system using intensity-modulated radiotherapy (IMRT). Three plans were designed for each patient, including plan 1 for conventional treatment and plan 2 and plan 3 for dose painting. In plan 1, patients were prescribed a traditional homogeneous dose of 45-60 Gy to the PTV, with a maximum dose constraint of 110% of the prescription dose, normalized to the PTV average dose. In plan 2, the dose of PTV-ASL increased by 10-20% based on plan 1, with a maximum dose constraint of 110% of the dose to PTV-ASL. The plan 3 was obtained from plan 2 without the maximum dose constraint. The dose normalizations were set to the PTV-ASL average dose in plans 2 and plan 3.
The eld angles were kept constant for all three treatment plans. Dose calculation was performed in anisotropic analytical algorithm optimization mode (AAA, version 15.512) using 6 MV X-rays. The calculated grid size was 2.5 mm × 2.5 mm, and the prescription covered 95% of the target volume.

Plan evaluation
To compare the three treatment plans, the dose distribution and dose-volume histograms (DVHs) of the PTV, PTV-SUB, PTV-ASL and OARs were calculated. The D 2% and D 98% (where Dx indicates the dose to x% of the target volume) as well as D mean were compared for the PTV, PTV-SUB and PTV-ASL. Furthermore, the target coverage, conformity index (CI) and homogeneity index (HI) were calculated. A CI close to 1 indicates that the shape of the region receiving the reference dose closely matches the shape of the target region, and an HI close to 0 indicates good uniformity.  (2) Compared with plan 2, the D 2% of the PTV, PTV-SUB and PTV-ASL increased by 3.58%, 0.51% and 4.52% in plan 3, respectively, with statistically signi cant differences for the PTV and PTV-ASL (P < 0.05); D 98% and D mean decreased by 0.41%, 0.43%, and 0.56% and 0.50%, 0.86%, and 0.03%, respectively, with no statistically signi cant differences (P > 0.05).
(3) In the PTV-ASL, the D 2% , D 98% and D mean increased by at least 10% in plan 2 and plan 3 than plan 1, with the most signi cant increase of D 2% in plan 3 at 19.84%, as shown in Table 3 (provided as an additional le) and Fig. 5.

Plan evaluation
The PTV target coverages for plan 1, plan 2 and plan 3 were comparable (P > 0.05). Compared with plan 1, the CIs decreased by 18.60% in plan 2 and 12.79% in plan 3, respectively, with statistically signi cant differences (P < 0.05), but the CIs were comparable between plan 2 and plan 3(P > 0.05). Compared with plan 1, the HIs increased by 1.43 in plan 2 and 2 times in plan 3, respectively; and in increased by 23.53% in plan 3 than in plan 2 (P < 0.05).

Dosimetric Comparison of OARs
The Dmax of the bilateral eyeballs, bilateral lens, and bilateral optic nerves were comparable among plan 1, plan 2, and plan 3 (P > 0.05). Compared to plan 1, the D0.1cc of brain stem and the Dmax of optic chiasma were slightly increased by 3.98% and 4.52% in plan 2 and 4.33% and 6.27% in plan 3, respectively, with statistically signi cant differences (P < 0.05), but all absolute doses were safe. And they were comparable between plan 2 and plan 3 (P>0.05), as shown in Table 4 (provided as an additional le).

Discussion
The main innovation of this study is the demonstration that MR 3D-ASL images can be used to quantify and analyse tumor perfusion differences in NE-LGGs, and that sub-volume segmentation of the GTV can be performed based on this perfusion difference, allowing for safe dose escalation in hyper-perfused regions by the boost technique.
LGGs grow slowly and have a long disease duration, and most LGGs inevitably develop into high-grade malignant gliomas, with an incidence of malignant transformation of up to 17% and a poor prognosis, with a median overall survival time of only 2.4 years after malignant transformation [12][13][14]. Radiotherapy is an effective treatment for LGGs, allowing for the control of primary tumor lesions and improving patient survival. However, failure of local control remains the predominant cause of LGG treatment failure. Most LGG recurrences after radiotherapy predominantly occur in the radiation eld in conventional radiotherapy, which is signi cantly associated with insu cient doses to the tumor target region [15,16]. Prospective clinical trials, such as EORTC 22844 and NCCTG 86-72-51, showed that highdose radiotherapy did not improve the survival of patients with LGGs, but rather increased the risk of radiation necrosis and reduced the quality of life [17][18][19]. However, Liu et al. [20] demonstrated the survival bene ts of high-dose radiotherapy in patients with IDH wild-type (IDHwt) LGGs. In this study, 52 patients with IDHwt LGGs treated with radiotherapy were divided into two groups: a high-dose group with doses >54 Gy and a low-dose group with doses ≤54 Gy. The results showed that high-dose radiotherapy was associated with increased survival rates. Liu et al. [21] showed that high-dose radiotherapy (> 54 Gy) was an independent prognostic factor of LGGs and was associated with an increase in the survival rate of patients with O6-methylguanine-DNA methyltransferase promoter nonmethylation (MGMT-non-pM). Therefore, increasing the radiation dose can improve the survival rates of patients with special molecular subtypes of LGGs. And in this study, more than 70% the participants were patients with IDHwt LGGs, so we conducted the study of escalation.
The EORTC 22844 and NCCTG 86-72-51 trials were mainly based on conventional anatomical MR images for homogeneous dose boosting, with a simultaneous dose increase for healthy tissues and an inevitable increase in radiation damage. In contrast, dose painting radiotherapy guided by functional imaging, such as functional MR and PET, is expected to achieve a safe dose boost. Kim et al. [22] investigated the region of hyperperfused/hypercellular (TV HCV /TV CBV ) >1 cm 3 in high b-value diffusion-weighted MRI and dynamic enhanced perfusion MRI (DCE) for glioblastoma with dose boosting (75 Gy/30 f) and found that it resulted in good survival rates (12-month OS rate of 92% (P=0.03) and median OS of 20 months) with fair neurocognitive function and quality of life.

NE-
LGGs, as a group of tumors with signi cant heterogeneity, require a more individualized radiotherapy dose to be applied to the tumor target region. NE-LGGs have an intact blood-brain barrier and do not show enhancement on CT or MRI. There is considerable blindness when the GTV is delineated with the edema area as the main criterion. Jakola et al. [23] showed that the blood-brain barriers of NE-LGGs were disrupted after malignant transformation, and the region that appeared to be enhanced was much smaller than the edema area. Although coverage of the radiotherapy dose to the tumor can be guaranteed if the edema area is directly de ned as the GTV range of the NE-LGG, it is di cult to increase the radiotherapy dose due to the large irradiation range. Although it is di cult to de ne tumor margins in imaging without enhanced LGGs, there are still signi cant biological differences between the blood ow and metabolism of the tumor and the brain tissue in the edema area [24]. Quantifying this difference with functional imaging can clarify the distribution of tumor cell-enriched areas, which will provide a reliable reference for the heterogeneous distribution of radiotherapy doses.
3D-ASL is used for the brain tumors diagnosis and radiotherapy e cacy/toxicity predictions by labeling water molecules for perfusion imaging of tissues, effectively avoiding the risks of exogenous contrast agents [6]. Related studies have shown that the CBF determined by 3D-ASL is signi cantly positively correlated with micro-vessel density and vascular endothelial growth factor expression (r=0.604) but signi cantly negatively correlated with survival (r=-0.714) and can be an independent risk factor for OS (HR=1.028) [25,26]. This suggests that dose boosting in regions of high CBF in LGG radiotherapy is necessary.
We found that NE-LGGs showed no abnormalities on MR-enhanced T1 but showed local high perfusion on 3D-ASL. And all the local hyper-perfusion regions were within edema areas through the fusion images of T2 FLAIR and 3D-ASL. We also found that the volume of the hyper-perfusion region (GTV-ASL) was signi cantly smaller than the volume of the edema area (GTV) and by volume analysis of each target region. Therefore, boosting the local hyper-perfusion sub-volume region of 3D-ASL, the target volume of dose boosting was reduced, so as to the risk of radiation damage to surrounding healthy brain tissue.
In this study, plan 1 is the standard plan and plan 2 and plan 3 are dose painting plans. Compared with plan 1, the D 2% , D 98% and D mean of the PTV-ASL increased by more than 14% in plan 2 and plan 3; however, these values increased by less than 9% in the PTV-SUB, with D 98% increased by less than 1%.
This suggests that under the guidance of 3D-ASL, the dose painting plan can achieve a signi cant dose boost in hyper-perfused sub-volume region for NE-LGGs while having less impact on the peripheral dose and avoiding dose boost blindness. This is similar to the study of Thureau et al., which was based on 18 FDG PET-CT image-guided dose boost for locally advanced non-small-cell lung cancer (NSCLC) [27]. On the basis of not increasing the dose to OARs, the dose to high recurrence risk areas increased by 12.12%, whereas the doses to other areas of the GTV only increased by 2.88%. We also found that the D 2% of the PTV, PTV-SUB and PTV-ASL gradually increased in plan 1, plan 2 and plan 3, whereas the D 98% and D mean of each target region were comparable between plan 2 and plan 3. Therefore, the elimination of maximum dose constraints is favourable to boost the maximum dose but has no signi cant effect on the minimum or average dose.
The target coverages of the PTV were comparable among plan 1, plan 2 and plan 3, with the prescribed isodose line wrapping around approximately 95% of the PTV, indicating that all treatment plans for plan 2 and plan 3 met the minimum dose requirements of the target region. Because only the local hyperperfused regions on 3D-ASL were escalated in plan 2 and plan 3, whereas the edema areas surrounding the hyper-perfused regions received standard doses, the HIs were larger in plan 2 and plan 3 than in plan 1. In this study, although the conformality and homogeneity of plan 2 and plan 3 decreased, the dose to the tumor target region was signi cantly increased. Therefore, reducing the homogeneity of the radiation dose is bene cial for dose escalation and protecting OARs.
During radiotherapy for patients with NE-LGGs, radiation damage to surrounding healthy tissues should not be ignored. The dose boost to the tumor target region must be balanced with protecting OARs. The results of this study showed that, compared with plan 1, the D0.1cc of brain stem and the Dmax optic chiasma increased by less than 5% in plan 2 and plan 3, but they were within a safe range. The other OARs were comparable among the three treatment plans, which were consistent with the ndings of Kosztyla et al. [28].
The main limitations of our study were as follows: (1) The small number of patients is due to the relatively low incidence rate of NE-LGGs, as well as the fact that the vast majority of patients underwent surgery and were excluded from analysis after surgery due to the disruption of the blood-brain barrier, which often shows enhancement on conventional enhanced MRI. (2) Because this is a preclinical technical study, it is necessary to increase the sample size and include the PFS and OS of patients to evaluate the value of this technique for further clinical applications.

Conclusion
It is feasible to segment sub-volume in NE-LGG radiotherapy based on the perfusion differences under the guidance of 3D-ASL. The radiation dose to highly perfused regions can be safely increased by 10-20% by dose painting without increasing the radiation dose to OARs, which is conducive for improving the survival rate of patients with NE-LGGs and provides a basis for future individualized and precise radiotherapy for NE-LGGs.