ASL-Perfusion for Intrinsic Brain Tumor Diagnosis. Analysis of 253 Patients.

Purpose The aim of the study was to evaluate the role of pseudo-continuous ASL-perfusion (pCASL-perfusion) in preoperative assessing of cerebral glioma grades. Methods The study group consisted of 253 patients aged 7 to 78 years with supratentorial gliomas (65 had low-grade gliomas (LGG), 188 – high-grade gliomas (HGG)). Maximal tumor blood ow (maxTBF) in small ROIs (20 mm 2 ± 10 mm 2 ) were evaluated by subsequently normalized tumor blood ow (nTBF) calculation which was compared with normal appearing white matter of center semiovale of the contralateral hemisphere. has demonstrated both high and specicity in differentiating LGG and HGG, grade and III gliomas, but low sensitivity and specicity in distinguishing grade III and grade IV gliomas. We did not observe a signicant difference in TBF in astrocytomas and oligodendrogliomas.


Introduction
Gliomas are the most common type of primary brain tumors and comprise about 80% of all malignant brain neoplasms. Preoperative predicting glioma grades is important for development of an optimal treatment strategy and making a prognosis [1,2].
It is well-known, that modern diagnostic modalities, like MR-perfusion, are more effective in detecting of brain tumors grades compared to the routine MRI [3,4]. T2*MR-perfusion (dynamic susceptibility contrast) is the "gold standard" for assessing tumor blood ow [5][6][7].
Arterial spin labeling (ASL) is a non-invasive method of obtaining CBF (cerebral blood ow) maps. Some authors reporting on the use of pulsed ASL (PASL) and pseudo-continuous ASL (pCASL) marked that CBF maps derived by ASL (ASL-CBF) were really helpful in detecting cerebral glioma grades [8- 14,15] and predicting prognosis of the disease [10,16,17]. Despite the fact that pCASL is an advanced ASL method compared to PASL and CASL [18][19][20][21], the recent data remain controversial. Some studies claim ineffectiveness of pCASL for glioma differentiation [4], while other papers, on the contrary, prove its informativity validate its virtue [22-26, 27, 28, 29-32]. pCASL sensitivity and speci city thresholds in differentiating LGG and HGG in the aforementioned studies vary considerably. Probably, these differences can be related to the methods of ROI detection and normalization technique of TBF.
The aim of the present study was to evaluate the potential of pseudo-continuous ASL-perfusion (pCASLperfusion) in preoperative assessment of cerebral glioma grades. We suggest that measuring of maximal TBF and normalized TBF values using small ROI might be more informative for distinguishing cerebral gliomas.

Materials And Methods
The study group consisted of 253 patients (118 men and 135 women) aged 7 to 78 years (mean age 45 ± 15 years) with cerebral supratentorial glial tumors which were diagnosed and then surgically treated (tumor removal or stereotactic biopsy, chemo-radiotherapy) at Burdenko Neurosurgery Center from 2011 to 2018 (Table 1). MR-studies were performed on 3 Tesla MRI scanner General Electric Signa HD (GE Healthcare) with 8channel head coil. Imaging included: T1 FSPGR BRАVO with isotropic voxel 1x1x1mm and no gap before and after contrast enhancement (or axial T1 weighted imaging with 5 mm slice thickness and 1 mm gap before contrast enhancement and postcontrast axial, sagittal and coronal T1 weighted images), axial T2 weighted images with 5mm slice thickness and 1 mm gap, T2-FLAIR with 5 mm slice thickness and 1mm gap, DWI ASSET with 5 mm layers and 1 mm gap, as well as 3D pCASL.
Postprocessing was performed with ReadyView (GE Healthcare). To assess tumor blood ow, ROI (region of interest) 20 mm 2 ± 10 mm 2 was chosen in the highest CBF zone (detected by color blood ow maps).
Mean tumor blood ow (maxTBF) was then evaluated within this ROI. Normalization with intact white matter of center semiovale of the contralateral hemisphere was performed to exclude individual blood ow patterns by placing ROI of the same size (20 mm 2 ± 10 mm 2 ) in the centre semiovale like ROI in the tumor. Normalization was obtained by dividing the maximal TBF by CBF in centre semiovale in the contralateral hemisphere: nTBF = maxTBF/ CBFcent.semiov.
CBF maps were fused with structural images (T2WI, T2-FLAIR, postcontrast T1WI) by means of NeuroRegistration software (GE Healthcare) in all cases.
Statistical analysis was performed with R-project program (https://www.r-project.org), pROC library was used for ROC-analysis. We used nonparametric methods in this study. Between-group comparisons were performed using the two-tailed Mann-Whitney rank-sum tests; continuous dependences were evaluated using Spearman rank correlations.

Study Results
During our study there were obtained mean maximal tumor blood ow levels and mean maximal normalized tumor blood ow levels in gliomas of different grades. We found a signi cant difference (p < 0.001) in tumor blood ow for LGG (grade I + II) and HGG (grade III + IV), with higher TBF values being marked in high grade gliomas.
Maximal TBF levels and normalized TBF levels are summarized in Table 2. ROC-analysis revealed sensitivity and speci city values of ASL-perfusion in the differential diagnosis of LGG and HGG. Data are summarized in Table 3 and Fig. 1.
ASL-perfusion exhibited both high sensitivity and speci city in distinguishing HGG and LGG. Threshold was determined as 64ml/100g/min for maximal TBF and 3.6 for normalized TBF. AUC > 0.95 for maximal TBF and normalized TBF.
Our study revealed distinct differences in TBF for low-and high-grade gliomas (p < 0.001 However, ROC-analysis revealed relatively low diagnostic value of ASL-perfusion when distinguishing gliomas grade III and grade IV (Table 3, Fig. 2).
We also analyzed TBF in high-grade gliomas excluding anaplastic oligodendrogliomas. maxTBF for anaplastic astrocytomas was 123.37 ± 89.57 ml/100g/min and nTBF was 6.70 ± 4.59 and both parameters were signi cantly lower compared to those for glioblastomas (p = 0.0004 for maxTBF, p = 0.001 for nTBF). Despite of that, exclusion of anaplastic oligodendrogliomas did not result in higher sensitivity and/or speci city when distinguishing grade III and grade IV gliomas by ASL-perfusion. (Table 3).
Moreover, comparison of TBF and nTBF in diffuse astrocytomas and oligodendrogliomas, anaplastic astrocytomas and anaplastic oligodendrogliomas did not reveal any signi cant difference (all p > 0.05).
Grade III gliomas demonstrated signi cantly higher maxTBF and nTBF values than grade II gliomas (p < 0.001). According to ROC-analysis ASL-perfusion was proved to be highly informative in these tumors ( Table 3). Exclusion of oligodendrogliomas and anaplastic oligodendrogliomas affected neither sensitivity nor speci city of the method.

Discussion
Recent studies have demonstrated a high potential of PASL and CASL in differentiating high-and lowgrade gliomas before surgery [8- 14,31]. Although ASL-perfusion is a relatively new method, it has already proved to be effective in diagnosis of cerebral gliomas. There is a number of studies establishing a high correlation between tumor blood ow derived from ASL-perfusion and DSC-perfusion which is known as the "gold standard" in perfusion studies [33,34].
Several recent studies were devoted to ASL-perfusion in differentiating cerebral gliomas. They showed contradictory results regarding sensitivity and speci city of this method in distinguishing LGG and HGG and TBF threshold values. An our opinion, these differences are the result of different approaches used for selecting ROI/VOI in TBF assessing as well as different methods of TBF normalization.  [22]. ROI size was not speci ed in the paper. The authors did not use TBF normalization for the differential diagnosis in this paper. Ma et al. (2017) used color maps to select the highest TBF and ROI was set to 50-60 mm 2 [23]. In contrast, we used smaller ROI size of 20 ± 10 mm 2 and performed normalization to the ROI within tumor in the mirror-like area of the contralateral hemisphere. The approach used by Xiao et al. (2015) was mostly close to the one we used: the researchers placed several 28-32 mm 2 ROI scattered within the whole tumor volume. Then ROIs with maximal TBF were picked for analysis [24]. Normalization was performed to CBF in the cerebellar white matter.
All the above mentioned papers were based upon pCASL technique.
Studies using small-size ROI and VOI demonstrated higher sensitivity and speci city in distinguishing LGG and HGG. According to meta-analysis performed by Alsaedi et al. (2019) maxTBF was proved to be more informative rather than meanTBF in differentiating cerebral glioma grades.
Our results coincide with the aforementioned studies, but demonstrate higher sensitivity and speci city in distinguishing LGG and HGG and much higher AUC. maxTBF for low-grade gliomas in our study group was much lower compared to other studies, and much higher for high-grade gliomas. The observed difference could be explained by different ROI selection for TBF measuring.
Our study is also different by nTBF: the difference was mostly de ned by the site of normalization -we used center semiovale of the contralateral hemisphere. In our study we used both absolute (TBF) and normalized (nTBF) maximal tumor blood ow and demonstrated higher sensitivity and speci city in distinguishing LGG and HGG. We found a signi cant difference in TBF for grade III and grade IV gliomas, although low sensitivity and speci city did not let us using ASL-perfusion for differentiating gliomas of grades III and IV. Importantly, excluding anaplastic oligodendrogliomas affected neither sensitivity nor speci city. We also failed to detect any statistically signi cant difference in TBF for diffuse astrocytomas and oligodendrogliomas as well as anaplastic astrocytomas and anaplastic oligodendrogliomas. On the contrary, in the study conducted by Zeng et al. It is well-known, that oligodendrogliomas are characterized by even higher microvascular density within the whole tumor volume [35]. Inclusion of the whole tumor volume on the slice in the measurement area results in the increased TBF on perfusion map in oligodendrogliomas. Our results suggest that application of small ROI showed enable measuring maxTBF which is the same for astrocytomas and oligodendrogliomas.

Conclusion
pCASL with a small ROI used for measuring maxTBF and nTBF has demonstrated both high sensitivity and speci city in distinguishing high-and low-grade gliomas, as well as grade II and grade III gliomas.
The study was supported by the Russian Foundation for Basic Research (grant № 18-29-01018).

Declarations Funding
The study was supported by the Russian Foundation for Basic Research (grant № 18-29-01018). No other funding was received for this study.

Con ict of interest
The authors declare that they have no con ict of interest.

Availability of data and material
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics approval
This retrospective and prospective chart review study involving human participants was in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Approval was granted by the

Consent to participate
Informed consent was obtained from all individual participants included in the study. Informed consent was also obtained from the parent and / or legal guardian of the minors who participated in the study.

Consent for publication
Patients signed informed consent regarding publishing their data and photographs.