Comparison of normalized cerebral blood flow between different post-processing methods of dynamic susceptibility contrast perfusion-weighted imaging and arterial spin labeling in gliomas with different grading

Introduction

Brain glioma is the most common malignant tumor of the nervous system (1). According to World Health Organization 2021, it is classified into 4 grades, with grade 1–2 classed as low-grade glioma (LGG) and grade 3–4 as high-grade glioma (HGG) (2,3). A higher grade of glioma is associated with a higher recurrence rate and a worse prognosis. Currently, magnetic resonance imaging (MRI) is commonly used in clinical diagnosis and follow-up of neurological diseases (4). MRI perfusion imaging methods, such as dynamic susceptibility contrast perfusion-weighted imaging (DSC-PWI) and arterial spin labeling (ASL), have been shown to characterize and differentiate glioma (5-11). The perfusion sequences, especially DSC-PWI, provide information about tumor vascular and microvascular environments (12). However, DSC-PWI and ASL have completely different theoretical bases. Traditionally, the gadolinium contrast media is necessary for DSC-PWI (13), and the combination of time-intensity curve (TIC) of brain tissue and the arterial input function (AIF) can generate cerebral blood volume (CBV), cerebral blood flow (CBF), mean transit time (MTT), and time to peak (TTP) (14,15). Different from DSC-PWI, the labeled water is used for ASL, and the crucial scanning parameter is post-label delay (PLD) (16-18). Due to the different technical characteristics, the perfusion parameters derived from the two methods are also different, which has been confirmed in ischemic stroke (19,20).

In theory, the precondition for DSC-PWI to generate perfusion parameters is intact blood-brain barrier (BBB), which is based on the indicator dilution theory (21). However, due to the biological characteristics of glioma, the BBB is often invaded and destroyed, resulting in gadolinium leakage, which lead to failure to truly reflect tumor perfusion (10). The result of gadolinium leakage can be characterized by raw TIC, regardless of post-processing methods, and in order to ensure the efficiency of DSC-PWI to characterize glioma, it is necessary to rectify the raw TIC (10,22-24). Traditionally, reducing the effect of gadolinium leakage mainly focuses on the application of lower flip angle (FA), preload of gadolinium, and post-processing methods (14). Generally, DSC-PWI can provide two post-processing methods, namely, AIF and gamma-variate fitting (GVF), both of which can correct raw TIC to obtain perfusion parameters. However, the raw TIC of DSC-PWI can visually and quantitatively reflect gadolinium leakage using percentage of signal recovery (PSR) (10), and the PSR has been also demonstrated to differentiate intracranial tumors (10,25-27). The AIF can also derive two maps of T2* leakage indicator and T1 leakage indicator to show the two leakage effects. Unlike AIF, which uses the deconvolution signal to correct raw TIC, the GVF can correct raw TIC to reduce the influence of the curve instability caused by the contrast agent recirculation and leakage (28,29). However, the post-processing methods of DSC-PWI used for evaluating brain tumors have been inconsistent in previous reports, and the consistency of normalized CBF derived from different post-processing methods of DSC-PWI and ASL for gliomas with different grading currently remains unclear. In particular, the T2* leakage and T1 leakage caused by the disruption of BBB and other reasons (e.g., cellularity, vascular and microvascular architecture) further increases the uncertainty of perfusion differences (10,24,30). The aim of this retrospective study was to investigate the consistency of normalized CBF derived from different post-processing methods of DSC-PWI and ASL in gliomas, and to verify the value of T2* leakage and T1 leakage indicators of AIF in characterizing both leakage effects and differentiating gliomas. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1076/rc).

Methods

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethics Committee of Cangzhou Central Hospital (No. 2023-028-02[z]), and individual consent for this retrospective analysis was obtained.

Patients

The sample size of the study was determined by PASS (https://www.ncss.com/software/pass/) (power: 0.9, significance level: 0.05). We retrospectively included a total of 65 patients with pathologically confirmed glioma from 1 January 2020 to 15 December 2023. The inclusion criteria were as follows: (I) patients underwent conventional MRI (non-contrast MRI and contrast MRI), ASL, and DSC-PWI before resection or puncture biopsy of intracranial glioma; (II) no treatment for gliomas; (III) the grading of glioma was confirmed by pathology. The exclusion criteria were as follows: (I) MRI images with obvious artifacts; (II) both cerebral hemispheres with glioma, which resulted in an inability to draw mirror region of interest (ROI); (III) failure to label ASL because of carotid stent. Finally, 56 patients with 56 gliomas were enrolled (Figure 1). Non-contrast MRI was performed in a subset of patients within 1 week (median: 3 days, range: 1–7 days) prior to perfusion imaging and contrast MRI. Of the 56 patients, 24 with LGG (grade 1–2) and 32 with HGG (grade 3–4) gliomas (Table 1) were recorded as the LGG group and HGG group, respectively. The genetic profiles of the gliomas are presented in Table 1.

Figure 1 The flowchart of enrolled patient in the study. ASL, arterial spin labeling; LGG, low-grade glioma; HGG, high-grade glioma; MRI, magnetic resonance imaging.

Perfusion imaging and data processing

All MRI scanning sequences were performed on 3T MRI scanner (GE Discovery 750W, GE Healthcare, Waukesha, WI, USA). The 16-channel phased-array coil was used for MRI examinations. MRI sequences included non-contrast MRI [axial T2-weighted imaging (T2WI), axial T1‑weighted imaging (T1WI), axial T2 fluid‑attenuated inversion recovery (FLAIR), axial diffusion-weighted imaging (DWI), sagittal T1WI], contrast T1WI (axial, coronal, sagittal), ASL, and DSC-PWI. The gradient-recalled echo echo-planar imaging (GRE-EPI) sequence was used to perform DSC-PWI during the administration of gadolinium contrast media (gadoterate meglumine; Hengrui, Jiangsu, China) (0.1 mmol/kg) at a rate of 3.5 mL/s, followed by 15 mL saline at the same rate. A total of 50 phases were obtained for DSC-PWI. The preload of gadolinium was not employed in the protocol of DSC-PWI, and the FA used in DSC-PWI was 90°. The scanning parameters of MRI sequences are listed in Table 2. The data of DSC-PWI and ASL was processed on GE AW 4.7 workstation (Advantage for Windows; GE Healthcare). The post-processing of perfusion data was performed by two radiologists with 8 and 10 years of experience in neuroimaging diagnosis, respectively, who were blinded to all clinical information of patients.

The post-processing of DSC-PWI needed motion correction, which cost almost 30 seconds. Subsequently, AIF and GVF were used to derive CBF, respectively. The auto mode was used in the AIF, and the semi-automatic mode was required when the blood vessel could not be automatically marked. In addition, the types of primary TIC of all gliomas in AIF and GVF (1 balanced, 2 descending, 3 ascending) had been recorded respectively (Figure 2). The CBF derived from ASL, AIF, and GVF were recorded as ASL-CBF, AIF-CBF, and GVF-CBF, respectively. ROI was outlined for the maximum section of enhanced tumors, which described in previous report (31), avoiding hemorrhage and cystic necrosis as much as possible (Figure 3A-3C). Subsequently, the mirror ROI was placed on the contralateral cerebral hemisphere. The ROI of lesion (ROI_lesion) and mirror ROI were cloned to all phases of current series to analyze the T2* leakage and T1 leakage indicators. Finally, the CBF derived from DSC-PWI and ASL was normalized by mirror ROI. The following calculation formula was used:

AIF-rCBF=AIF−CBFlesion/mirror ROI[1]

GVF-rCBF=GVF−CBFlesion/mirror ROI[2]

ASL-rCBF=ASL−CBFlesion/mirror ROI[3]

Figure 2 The TIC derived from AIF and GVF of the three representative cases were presented (the yellow dotted line represents raw TIC and the pink dotted line represents the corrected TIC. The blue or gray dotted lines represent TIC of mirror region of interest). The TIC of patient with pilocytic astrocytoma graded as LGG, diffusion astrocytoma graded as HGG, and glioblastoma graded as HGG present balanced (A), sightly descending (B), and markedly ascending (C) after the negative enhancement. TIC, time-intensity curve; AIF, arterial input function; GVF, gamma-variate fitting; LGG, low-grade glioma; HGG, high-grade glioma.

Figure 3 The fused images of T2* and T1 leakage indicators obtained from the 3 representative cases with LGG (A,D), HGG graded as 4 (B,E), and HGG graded as 3 (C,F), respectively, the leakage indicator maps were fused with T1WI (A-C) and GRE-EPI (D-F), respectively. The three gliomas were located in the left cerebral hemisphere, and the manually delineated ROI on the right cerebral hemisphere was the mirror ROI of the gliomas (blue lines). The point of both leakage indicators of three cases interpretated by two radiologists using 4-point scale is provided in the bottom of the drawing. The LGG (A,D) and HGG (B,E) exhibited positive difference of point between T2* leakage indicator and T1 leakage indicator, moreover, the other HGG (C,F) exhibited negative difference of point between both leakage indicators. LGG, low-grade glioma; HGG, high-grade glioma; T1WI, T1 weighted imaging; GRE-EPI, gradient recalled echo-echo planar imaging; ROI, region of interest; GR, gradient recalled.

The normalized CBF was the average of two radiologists.

Using “cloned to all phases” can satisfy the analysis of T2*&T1 leakage indicators and acquisition of normalized CBF at the same ROI and section. Two radiologists used a 4-point scale (0 none, 1 mild, 2 moderate, 3 severe) to interpreted the weight of T2* leakage and T1 leakage on the fused imaging integrated with GRE-EPI, which take the mirror ROI as the reference (Figure 3D-3F) to reduce the interference caused by background noise. Another radiologist with 12 years of experience in neuroimaging diagnosis arbitrated any disagreement between the above two radiologists. In order to ensure the accuracy of interpretation of T2* and T1 leakage effects, the three radiologists were blinded to the raw TIC. Then, the difference of point between T2* leakage indicator and T1 leakage indicator was calculated (point of T2* leakage indicator − point of T1 leakage indicator) to analyze the weight between the two leakage effects.

Subsequently, the PSR of raw TIC derived from AIF was obtained according to the previous reports (10,32), and PSR was taken as the average of the other two radiologists with 5 and 7 years of respective experience in neuroimaging diagnosis. In this study, the PSR, as a quantitative metric, was used to verify the hypothesis that subjective visual approach, namely, 4-point scale, of T2* leakage and T1 leakage indicators, could characterize leakage effects and evaluate glioma grading.

Statistical analysis

The software GraphPad Prism version 9.5.1 (GraphPad Software, La Jolla, CA, USA) and SPSS 22.0 (IBM Corp., Armonk, NY, USA) were employed to test the data. The Shapiro-Wilk test was used for normality testing. Normally distributed data were expressed as mean ± standard deviation. The correlation and consistency of normalized CBF between DSC-PWI and ASL was tested by Pearson’s correlation coefficient r (<0.25: low correlation, 0.25 to <0.5: moderate correlation, 0.5 to <0.75: strong correlation, ≥0.75: excellent correlation), linear regression analysis, and Bland-Altman plots in the LGG and HGG groups, respectively. The differences of normalized CBF derived from DSC-PWI and ASL between the LGG and HGG groups was compared by Student’s t-test. The correlation between (point of T2* leakage indicator − point of T1 leakage indicator) and PSR was tested by Spearman correlation analysis (<0.25: low correlation, 0.25–<0.5: moderate correlation, 0.5–0.75: strong correlation, >0.75: excellent correlation). The differences of PSR between the LGG and HGG groups were verified by t-test. The intra- and inter-group differences of point for T2* and T1 leakage indicators were compared by χ² test and/or Fisher’s exact test. Inter-observer consistency of normalized CBF was tested using the intra-group correlation coefficient (ICC) (<0.5: poor, 0.5–<0.75: moderate, 0.75–0.9: good, >0.9: excellent) (33). The Kappa test was used to test the consistency of judgment of T2* leakage and T1 leakage (34). Two-tail test was used for statistical analysis, and P<0.05 was considered a significant difference.

Results

In this study, 41% of gliomas were type 1, 48% of gliomas were type 3, and 11% of gliomas were type 2 (Table 3). Most (16/23=70%) of the type 1 gliomas showed the equal point between T2* leakage and T1 leakage indicators, however the type 2 and 3 gliomas exhibited positive and negative difference between T2* leakage and T1 leakage indicators, respectively (Table 3).

In the HGG group, the AIF-rCBF and GVF-rCBF showed excellent correlation with ASL-rCBF (r=0.871, P<0.0001; r=0.757, P<0.0001); meanwhile, AIF-rCBF and GVF-rCBF exhibited strong correlation with ASL-rCBF in the LGG group (r=0.744, P<0.001; r=0.763, P<0.001), and there was no significant difference in linear regressions in both groups (F=0.456, P=0.503; F=2.806, P=0.099) (Figure 4). In both groups, ASL overestimated normalized CBF compared with DSC-PWI (AIF and GVF) (Figure 5).

Figure 4 The correlation and linear regression analysis of normalized CBF between AIF, GVF and ASL in LGG group (r=0.744, P<0.001; r=0.763, P<0.001) (A) and HGG group (r=0.871, P<0.0001; r=0.757, P<0.0001) (B). Meanwhile, there was no significant difference in linear regressions in LGG (F=0.456, P=0.503) and HGG groups (F=2.806, P=0.099). The solid line represents the regression equations lines, and the dashed line represents the 95% confidence interval of the regression equations. CBF, cerebral blood flow; ASL, arterial spin labeling; AIF, arterial input function; GVF, gamma-variate fitting; LGG, low-grade glioma; HGG, high-grade glioma; rCBF, relative cerebral blood flow.

Figure 5 Bland-Altman plots analysis of agreement of normalized CBF between AIF, GVF and ASL in LGG group (A,B) and HGG group (C,D). Compared with AIF (A,C) and GVF (B,D), ASL overestimates the normalized CBF in LGG and HGG groups, and the difference of normalized CBF, which is presented as mean ± standard deviation, between ASL and AIF was significant between LGG and HGG groups (0.02±0.49 vs. 0.28±0.29, P=0.017); however, the difference of normalized CBF between ASL and GVF was not significant between two groups (0.06±0.39 vs. 0.24±0.38, P=0.085). CBF, cerebral blood flow; AIF, arterial input function; GVF, gamma-variate fitting; ASL, arterial spin labeling; LGG, low-grade glioma; HGG, high-grade glioma.

The difference of ASL-rCBF and AIF-rCBF (ASL-rCBF minus AIF-rCBF) was significant between the LGG and HGG groups (0.02±0.49 vs. 0.28±0.29, P=0.017); however, the difference of (ASL-rCBF minus GVF-rCBF) was not significant between two groups (0.06±0.39 vs. 0.24±0.38, P=0.085).

The Spearman correlation analysis demonstrated that the difference of point for T2* and T1 leakage indicators was strongly correlated with PSR (r=−0.739, P<0.0001) (Figure 6A). The difference of PSR between the HGG and LGG groups was significant (t=2.043, P=0.04) (Figure 6B).

Figure 6 The correlation between the difference of point for T2* and T1 leakage indicators and PSR is negative (r=−0.739, P<0.0001) (A), and the HGG with higher PSR compared with LGG (t=2.043, P=0.04) (B). *, P<0.05. The dashed blue line represents that the PSR is 1 (A). PSR, percentage of signal recovery; LGG, low-grade glioma; HGG, high-grade glioma.

The difference of point between T2* leakage indicator and T1 leakage indicator was only significant in the HGG group (χ²=12.45, P=0.006) (Figure 7). The inter-group difference of T2* leakage and T1 leakage demonstrated that the proportion of gliomas with T2* leakage and T1 leakage in the HGG group was larger than that without T2* leakage and T1 leakage in the LGG group, respectively (all P<0.001) (Figure 8A,8B); meanwhile, the difference of point between T2* leakage indicator and T1 leakage indicator was significant between the LGG and HGG groups (χ²=11.28, P=0.004) (Figure 8C). ICC [0.939; 95% confidence interval (CI): 0.932–0.945] and kappa-value (0.915) both demonstrated that the intra-observer consistency of normalized CBF and the judgement of T2* leakage and T1 leakage indicators were good. ICC (0.927; 95% CI: 0.912–0.937) also revealed that intra-observer consistency of PSR was good.

Figure 7 The comparison of grading between T2* leakage and T1 leakage in LGG (A) and HGG (B), respectively. The difference of point between T2* leakage and T1 leakage is significant in HGG group (χ²=12.45, P=0.006); however, the difference of point between T2* leakage and T1 leakage is not significant in LGG group (χ²=3.700, P=0.296). LGG, low-grade glioma; HGG, high-grade glioma.

Figure 8 The comparison of point for T2* leakage indicator and T1 leakage indictor between LGG (A) and HGG (B), and the inter-group comparison of the difference of point between T2* leakage indicator and T1 leakage indicator (C). The proportion of gliomas with T2* leakage and T1 leakage in HGG group was larger than that without T2* leakage and T1 leakage in LGG group, respectively (all P<0.001) (A,B). The difference of point between T2* leakage and T1 leakage is significant between the LGG and HGG groups (χ²=11.28, P=0.004) (C). LGG, low-grade glioma; HGG, high-grade glioma.

Discussion

In the investigation, both AIF-rCBF and GVF-rCBF had good consistency with ASL-rCBF. However, the difference of normalized CBF between AIF and ASL was significant between the LGG and HGG groups; conversely, the difference between GVF-rCBF and ASL-rCBF was not significant among two groups. In addition, combining the analysis of raw TIC and PSR, as well as the point of the T2* and T1 leakage indicators, showed that HGG is more prone to T2* leakage and T1 leakage than LGG, and the weight of T1 leakage was larger than that of T2* leakage. The subjective visual interpretation of T2* leakage and T1 leakage indicators correlated with PSR, which is also effective for characterizing leakage effects and differentiating the gliomas grading.

The T2* GRE-EPI sequence, which is sensitive to susceptibility signal, is utilized for DSC-PWI. If the BBB is damaged, leakage of gadolinium will inevitably pollute the raw TIC of DSC-PWI, resulting in instability of semi-quantitative parameters after tissue response function deconvolution (22,29,35,36). Previous studies have shown that using preload leakage-correction, the spin echo (SE) sequence and different echo time (TE), as well as lower FA, can improve the accuracy of DSC-PWI (14,23,29,37,38). However, it is mainly focused on reducing the T1 relaxation effect. The post-processing of DSC-PWI can compensate for the distortion of raw TIC. Unlike AIF, GVF employs gamma-fits to minimize the influence of recirculation and gadolinium leakage after the negative enhancement (29). Additionally, ASL is not affected by the status of BBB (39).

In theory, the TIC of craniocerebral lesions obtained from DSC-PWI is the basis for the acquisition of perfusion parameters, which contains the perfusion information. Previous studies have also shown that the TIC and PSR derived from DSC-PWI can be used to differentiate brain gliomas from other brain tumors (e.g., intracranial solitary metastatic and primary central nervous system lymphoma) and characterize the different types of isocitrate dehydrogenase (IDH) mutant gliomas (10,25-27,40). In the present study, the majority of TIC was type 3, type 2 was the minority (Table 3). The main leakage weighted for the type 2 curve and type 3 curve were the T2* effect and T1 effect, respectively. However, the premise for the T2* effect is complex, due to variations in the magnetic susceptibility between different chambers (e.g., intravascular to extravascular/extracellular or intracellular to extracellular) (10,30,41,42). Another condition of T2* effect is the concentration accumulation (29). At the dose of 0.1 mmol/kg of gadolinium used in the study, the likelihood of causing T2* leakage is significantly reduced compared with the previous report (29), which can interpret the least number of type 2. In addition, the majority of gliomas with type 3 curve are HGG, and the number of LGG and HGG is similar in type 1 and type 2 curves. Those findings could indicate that HGG may be more susceptible to T1 leakage compared with T2* leakage at the dose of gadolinium (0.1 mmol/kg), which was also reported in previous research (35). From the technical point of view, the other reason for the large weight of T1 leakage is the non-usage of small FA and preload in this study. The type 1 curve included 12 gliomas without T2* and T1 leakage (12/23), most of which were LGG (11/12), and 11 gliomas with T2* and T1 leakage included 9 HGG (9/11) and 2 LGG (2/11). The composition of the gliomas with and without T2 * and T1 leakage is also the reason for the larger number of gliomas with type 1 curve. It is not difficult to understand that the gliomas without T2* and T1 leakage can perform type 1 curve. However, for gliomas with T2* and T1 leakage (11/23), as interpreted by 4-point scale, the influence of the two types of leakage neutralizes each other due to the similar point of both leakages (±1), so that the curve still shows a balanced performance after negative enhancement (Figure 2A). The analysis of the type 1 curve can be also revealed by the correlation between the difference of point for two leakage indicators and PSR. The PSR with a small degree of dispersion (1.15±0.24), which is close to 1, corresponds to the zero point (point of T2* leakage − point of T1 leakage). Moreover, because of the relatively larger weight of T2* leakage compared with T1 leakage, the TIC exhibits a slight decrease (Figure 2B). The influence of leakage has also been reported in previous studies of gadolinium extravasation (10,24,29,43). The analysis of three curve types indicates that HGG may have a greater probability of T2* leakage and T1 leakage than LGG, and the weight of T1 leakage was larger than T2* leakage. Those findings are also consistent with the biological behavior of HGG with higher invasion ability compared to LGG.

The quantitative analysis demonstrated that the normalized CBF derived from DSC-PWI and ASL were both greatly correlated in both groups, which was consistent with a previous study (44). However, compared with GVF, the difference of normalized CBF derived from ASL and AIF was significant between LGG and HGG groups (0.02±0.49 vs. 0.28±0.29, P=0.017). This finding demonstrated that AIF-rCBF may have a larger bias compared to ASL-rCBF between LGG and HGG groups. The difference may also be interpreted by the more aggressive biological behavior of HGG and the greater probability and degree of BBB destruction compared with LGG, which could increase the bias of normalized CBF between AIF and ASL. Contrary to the underestimation of normalized CBF derived from DSC-PWI compared with ASL in this study, Ma et al. (31) reported that the normalized CBF of DSC-PWI was greater than that of the ASL. The reason may be the difference in the normalization of CBF. As mentioned above, there are differences in the normalization of perfusion quantitative parameters in the investigation of craniocerebral diseases using perfusion technique, and in order to verify whether the T2* leakage indicator and T1 leakage indicator can reflect the grading of gliomas, our study employed mirror ROI to normalize CBF. Compared with the normalized by white matter or gray matter, mirror ROI may outline part of large blood vessels and include them in quantitative analysis. Compared with white matter or gray matter with stable perfusion (25,45), mirror ROI increased the perfusion difference compared with previous studies. This inference can be verified on the TIC of the target glioma and its mirror ROI (Figure 2B and Figure 3B). It can be found that the TIC of mirror ROI (right cerebral hemisphere ROI) is higher than that of HGG (left cerebral hemisphere ROI) (Figure 2B and Figure 3B). This discrepancy can be interpreted by the locations of glioma itself and the mirror ROI, namely, the larger blood vessels have been brought into the perfusion analysis. The mirror ROI of gliomas in Figure 3B is located in the area of the proximal branch of the middle cerebral artery, and the method of the normalization of CBF used in the study may also be the potential reason for the difference in quantitative comparison between DSC-PWI and ASL. In addition, another nonnegligible reason for the discrepancy with previous studies is the heterogeneity of gliomas, which greatly increases the probability of differences in normalized CBF. Compared with 3D voxel of interest, the utilization of 2D ROI could be another reason for the increased probability of overlooking gliomas heterogeneity in the investigation (10,25). Previous studies of correcting perfusion parameters of DSC-PWI mainly focus on CBV; meanwhile, the T1 effect caused by gadolinium leakage after correction is still present (29,46). From the perspective of acquisition sequence, the acquisition of original perfusion signals by dual echo is conducive to reducing the influence of T1 effect on CBV (47), and the usage of lower FA and preload of gadolinium are also effective methods to alleviate T1 effect. However, dual echo sequence, lower FA, and preload gadolinium were not used in the study. Contrary to the result of T1 effect which caused the underestimation of perfusion parameter, T2* effect caused it to be overestimated. However, the number of gliomas with T2* leakage is smaller compared with those with T1 leakage. As reported in previous studies (29,46), the technically unavoidable T1 effect may be responsible for the results of normalized CBF derived from DSC-PWI still be underestimated compared with ASL in this study. Meanwhile, the predominance of HGG (19/27) in gliomas with type 3 curves (Table 3) also explains the finding that the differences of normalized CBF between DSC-PWI (AIF and GVF) and ASL were greater in the HGG group than they were in the LGG group (Figure 5).

Although it has been previously reported that gadolinium leakage occurring at the relatively later stage of the TIC has little effect on CBF (48), the introduction of AIF increases the uncertainty of CBF from the analysis of the deconvolution operation method for obtaining CBF. In addition, the above views on the small effect of gadolinium leakage on CBF are based on the model assumption that the contrast agent is still in the blood vessels during the initial capillary transport stage (36,49). In summary, the inevitable T1 effect and the uncertainty of AIF may be responsible for the differences between post-processing methods of DSC-PWI and ASL in gliomas with different grading.

The strong correlation between the difference of point for two leakage indicators and PSR (Figure 6A) demonstrates that interpretation of both leakage indictors derived from AIF can also effectively characterize the balance between T2* and T1 leakage effects. In addition, the quantitative PSR can also be used to distinguish grading of gliomas, which is consistent with previous studies (10,30). Furthermore, HGG with higher PSR (1.22±0.23) indicates (Figure 6B) the larger T1 effect weight, and this finding is also supported by the correlation analysis between PSR and the point of two leakage indicators.

The analysis of T2* and T1 leakage indicators demonstrated that the proportion of HGG with T2* leakage and T1 leakage is higher than that in LGG. The grading weight of T1 leakage in the HGG group was greater than that in the LGG group. These findings indicate that HGG is more prone to T2* and T1 leakage comparing with LGG, and HGG exhibits high weight of T1 leakage compared with T2* leakage. Those results are also verified by the proportion of HGG (19/27) in gliomas with the type 3 curve and the proportion of HGG (9/11) with T2* and T1 leakage in gliomas with the type 1 curve. However, inconsistent with the previous study (50), the three cases of pilocytic astrocytoma attributed to the LGG group did not show dominant T1 effect, and PSR (0.97±0.25) was not higher compared other gliomas in this study. The smaller sample size (3 vs. 11) and the gliomas’ heterogeneity may be the explanation for this difference. It is necessary to comprehensively analyze the effects of T2* leakage and T1 leakage on DSC-PWI (30,35,51). A previous study demonstrates that transverse relativity at tracer equilibrium, which reflects the balance of T2* and T1 leakage effects, is also correlated with PSR (r=−0.87) (30). Those results echo our finding of the correlation between the visual interpretation of both leakage indicators and PSR (r=−0.739). In this investigation, the proportion of gliomas with predominantly T2* leakage was almost 18% (10/56) slightly lower than reported by Bjornerud et al. (35). This discrepancy may be due to differences in the concentration of gadolinium contrast agent used in DSC-PWI (0.1 vs. 0.2 mmol/kg). T2* leakage may indicate higher vascularization, namely, higher perfusion parameters (30). Notably, the normalized CBF of the six gliomas with type 2 curve was not higher than that of the gliomas with type 1 curve, which may be the reason for the sample size and without absolute perfusion parameters. In terms of the biological behavior of gliomas, IDH wild-type may exhibit a greater tendency of T2* leakage effect compared to IDH mutants (30). However, the enrolled majority of adult gliomas in this study were IDH mutants (27 vs. 17), which could also clarify the difference of two leakage effects. Furthermore, the inter-group and intra-group differences of T2* and T1 leakage indicators also echo the underestimation of normalized CBF by DSC-PWI compared with ASL, and the difference of (ASL-CBF minus AIF-rCBF) between LGG and HGG groups. Additionally, the intra-observer consistency of both leakage indicators is good. Combined with the above analysis, it is reasonable to conclude that the subjective interpretation of both leakage indicators derived from AIF can be used as a surrogate of PSR for characterizing the leakage effects and distinguishing glioma grading.

In this study, the patients enrolled were preoperative glioma patients. As mentioned above, disruption of BBB may interfere with the ability of DSC-PWI to obtain stable perfusion parameters (43). In addition, a previous study suggested that the application of ASL in postoperative glioma assessment may be more appropriate (12), and this difference of normalized CBF between DSC-PWI and ASL was also verified in preoperative gliomas. However, the normalized CBF derived from different post-processing methods of DSC-PWI and ASL had good consistency in HGG and LGG groups, and two leakage indicators can aid in characterizing leakage effect and distinguishing glioma classification.

This study had several limitations. Firstly, the relatively small sample size limited further subgroup analysis. Secondly, different MRI scanning platforms, dose of gadolinium, and FA (14,38) were not employed in this study, and prospectively comparative investigation may be necessary. Thirdly, since the glioma was the subject enrolled in the study, the 4-point scale of T2* and T1 leakage indicators cannot be used to identify the other intracranial solid tumors.

Conclusions

The normalized CBF derived from DSC-PWI and ASL have good consistency in gliomas with different grading, regardless of post-processing methods of DSC-PWI. Moreover, the GVF can provide the less bias of normalized CBF between LGG and HGG groups compared with AIF; however, T2* and T1 leakage indicators derived from AIF, which can be used for a surrogate of PSR, can aid to characterize leakage effects and identify glioma classification.

Acknowledgments

Funding: This work was supported by Science and Technology Planning Project of Cangzhou City (No. 222106148).

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1076/rc

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1076/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethics Committee of Cang zhou Central Hospital (No. 2023-028-02[z]), and individual consent for this retrospective analysis was obtained.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Ostrom QT, Gittleman H, Liao P, Vecchione-Koval T, Wolinsky Y, Kruchko C, Barnholtz-Sloan JS. CBTRUS Statistical Report: Primary brain and other central nervous system tumors diagnosed in the United States in 2010-2014. Neuro Oncol 2017;19:v1-88. [Crossref] [PubMed]
2. Louis DN, Perry A, Wesseling P, Brat DJ, Cree IA, Figarella-Branger D, Hawkins C, Ng HK, Pfister SM, Reifenberger G, Soffietti R, von Deimling A, Ellison DW. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol 2021;23:1231-51. [Crossref] [PubMed]
3. Pons-Escoda A, Majos C, Smits M, Oleaga L. Presurgical diagnosis of diffuse gliomas in adults: Post-WHO 2021 practical perspectives from radiologists in neuro-oncology units. Radiologia (Engl Ed) 2024;66:260-77. [Crossref] [PubMed]
4. Stockham AL, Tievsky AL, Koyfman SA, Reddy CA, Suh JH, Vogelbaum MA, Barnett GH, Chao ST. Conventional MRI does not reliably distinguish radiation necrosis from tumor recurrence after stereotactic radiosurgery. J Neurooncol 2012;109:149-58. [Crossref] [PubMed]
5. Zhang J, Wang Y, Wang Y, Xiao H, Chen X, Lei Y, Feng Z, Ma X, Ma L. Perfusion magnetic resonance imaging in the differentiation between glioma recurrence and pseudoprogression: a systematic review, meta-analysis and meta-regression. Quant Imaging Med Surg 2022;12:4805-22. [Crossref] [PubMed]
6. Purohit B, Kamli AA, Kollias SS. Imaging of adult brainstem gliomas. Eur J Radiol 2015;84:709-20. [Crossref] [PubMed]
7. Manias KA, Peet A. What is MR spectroscopy? Arch Dis Child Educ Pract Ed 2018;103:213-6. [Crossref] [PubMed]
8. Galanaud D, Chinot O, Metellus P, Cozzone P. Magnetic resonance spectroscopy in gliomas. Bull Cancer 2005;92:327-31.
9. Hu X, Xue M, Sun S, Zou Y, Li J, Wang X, Liu X, Ma H. Combined application of MRS and DWI can effectively predict cell proliferation and assess the grade of glioma: A prospective study. J Clin Neurosci 2021;83:56-63. [Crossref] [PubMed]
10. Pons-Escoda A, Garcia-Ruiz A, Naval-Baudin P, Martinez-Zalacain I, Castell J, Camins A, Vidal N, Bruna J, Cos M, Perez-Lopez R, Oleaga L, Warnert E, Smits M, Majos C. Differentiating IDH-mutant astrocytomas and 1p19q-codeleted oligodendrogliomas using DSC-PWI: high performance through cerebral blood volume and percentage of signal recovery percentiles. Eur Radiol 2024;34:5320-30. [Crossref] [PubMed]
11. Pons-Escoda A, Naval-Baudin P, Viveros M, Flores-Casaperalta S, Martinez-Zalacaín I, Plans G, Vidal N, Cos M, Majos C. DSC-PWI presurgical differentiation of grade 4 astrocytoma and glioblastoma in young adults: rCBV percentile analysis across enhancing and non-enhancing regions. Neuroradiology 2024;66:1267-77. [Crossref] [PubMed]
12. Chatha G, Dhaliwal T, Castle-Kirszbaum MD, Amukotuwa S, Lai L, Kwan E. The utility of arterial spin labelled perfusion-weighted magnetic resonance imaging in measuring the vascularity of high grade gliomas - A prospective study. Heliyon 2023;9:e17615. [Crossref] [PubMed]
13. Belliveau JW, Kennedy DN Jr, McKinstry RC, Buchbinder BR, Weisskoff RM, Cohen MS, Vevea JM, Brady TJ, Rosen BR. Functional mapping of the human visual cortex by magnetic resonance imaging. Science 1991;254:716-9. [Crossref] [PubMed]
14. Boxerman JL, Quarles CC, Hu LS, Erickson BJ, Gerstner ER, Smits M, et al. Consensus recommendations for a dynamic susceptibility contrast MRI protocol for use in high-grade gliomas. Neuro Oncol 2020;22:1262-75. [Crossref] [PubMed]
15. Pons-Escoda A, Smits M. Dynamic-susceptibility-contrast perfusion-weighted-imaging (DSC-PWI) in brain tumors: a brief up-to-date overview for clinical neuroradiologists. Eur Radiol 2023;33:8026-30. [Crossref] [PubMed]
16. Golay X, Ho ML. Multidelay ASL of the pediatric brain. Br J Radiol 2022;95:20220034. [Crossref] [PubMed]
17. Yu H, Ouyang Y, Feng Y, Sun S, Zhang L, Liu Z, Tian H, Xie S. Comparison of single-postlabeling delay and seven-delay three-dimensional pseudo-continuous arterial spin labeling in the assessment of intracranial atherosclerotic disease. Quant Imaging Med Surg 2023;13:2514-25. [Crossref] [PubMed]
18. Ma X, Wang Y, Wang M, Zhang M, Meng N, Zhang L, Zhang J, Dou S, Wang M. Evaluation of infarct core and ischemic penumbra by absolute quantitative cerebral dynamic susceptibility contrast perfusion magnetic resonance imaging using self-calibrated echo planar imaging sequencing in patients with acute ischemic stroke. Quant Imaging Med Surg 2022;12:4286-95. [Crossref] [PubMed]
19. Huang YC, Liu HL, Lee JD, Yang JT, Weng HH, Lee M, Yeh MY, Tsai YH. Comparison of arterial spin labeling and dynamic susceptibility contrast perfusion MRI in patients with acute stroke. PLoS One 2013;8:e69085. [Crossref] [PubMed]
20. Zhang SX, Yao YH, Zhang S, Zhu WJ, Tang XY, Qin YY, Zhao LY, Liu CX, Zhu WZ. Comparative study of DSC-PWI and 3D-ASL in ischemic stroke patients. J Huazhong Univ Sci Technolog Med Sci 2015;35:923-7. [Crossref] [PubMed]
21. Rosen BR, Belliveau JW, Vevea JM, Brady TJ. Perfusion imaging with NMR contrast agents. Magn Reson Med 1990;14:249-65. [Crossref] [PubMed]
22. Boxerman JL, Schmainda KM, Weisskoff RM. Relative cerebral blood volume maps corrected for contrast agent extravasation significantly correlate with glioma tumor grade, whereas uncorrected maps do not. AJNR Am J Neuroradiol 2006;27:859-67.
23. Hu LS, Baxter LC, Pinnaduwage DS, Paine TL, Karis JP, Feuerstein BG, Schmainda KM, Dueck AC, Debbins J, Smith KA, Nakaji P, Eschbacher JM, Coons SW, Heiserman JE. Optimized preload leakage-correction methods to improve the diagnostic accuracy of dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging in posttreatment gliomas. AJNR Am J Neuroradiol 2010;31:40-8. [Crossref] [PubMed]
24. Quarles CC, Ward BD, Schmainda KM. Improving the reliability of obtaining tumor hemodynamic parameters in the presence of contrast agent extravasation. Magn Reson Med 2005;53:1307-16. [Crossref] [PubMed]
25. Pons-Escoda A, Garcia-Ruiz A, Naval-Baudin P, Cos M, Vidal N, Plans G, Bruna J, Perez-Lopez R, Majos C. Presurgical Identification of Primary Central Nervous System Lymphoma with Normalized Time-Intensity Curve: A Pilot Study of a New Method to Analyze DSC-PWI. AJNR Am J Neuroradiol 2020;41:1816-24. [Crossref] [PubMed]
26. Pons-Escoda A, García-Ruíz A, Naval-Baudin P, Grussu F, Viveros M, Vidal N, Bruna J, Plans G, Cos M, Perez-Lopez R, Majós C. Diffuse Large B-Cell Epstein-Barr Virus-Positive Primary CNS Lymphoma in Non-AIDS Patients: High Diagnostic Accuracy of DSC Perfusion Metrics. AJNR Am J Neuroradiol 2022;43:1567-74. [Crossref] [PubMed]
27. Garcia-Ruiz A, Pons-Escoda A, Grussu F, Naval-Baudin P, Monreal-Aguero C, Hermann G, Karunamuni R, Ligero M, Lopez-Rueda A, Oleaga L, Berbís MÁ, Cabrera-Zubizarreta A, Martin-Noguerol T, Luna A, Seibert TM, Majos C, Perez-Lopez R. An accessible deep learning tool for voxel-wise classification of brain malignancies from perfusion MRI. Cell Rep Med 2024;5:101464. [Crossref] [PubMed]
28. Benner T, Heiland S, Erb G, Forsting M, Sartor K. Accuracy of gamma-variate fits to concentration-time curves from dynamic susceptibility-contrast enhanced MRI: influence of time resolution, maximal signal drop and signal-to-noise. Magn Reson Imaging 1997;15:307-17. [Crossref] [PubMed]
29. Paulson ES, Schmainda KM. Comparison of dynamic susceptibility-weighted contrast-enhanced MR methods: recommendations for measuring relative cerebral blood volume in brain tumors. Radiology 2008;249:601-13. [Crossref] [PubMed]
30. Sanvito F, Raymond C, Cho NS, Yao J, Hagiwara A, Orpilla J, Liau LM, Everson RG, Nghiemphu PL, Lai A, Prins R, Salamon N, Cloughesy TF, Ellingson BM. Simultaneous quantification of perfusion, permeability, and leakage effects in brain gliomas using dynamic spin-and-gradient-echo echoplanar imaging MRI. Eur Radiol 2024;34:3087-101. [Crossref] [PubMed]
31. Ma H, Wang Z, Xu K, Shao Z, Yang C, Xu P, Liu X, Hu C, Lu X, Rong Y. Three-dimensional arterial spin labeling imaging and dynamic susceptibility contrast perfusion-weighted imaging value in diagnosing glioma grade prior to surgery. Exp Ther Med 2017;13:2691-8. [Crossref] [PubMed]
32. Cha S, Lupo JM, Chen MH, Lamborn KR, McDermott MW, Berger MS, Nelson SJ, Dillon WP. Differentiation of glioblastoma multiforme and single brain metastasis by peak height and percentage of signal intensity recovery derived from dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging. AJNR Am J Neuroradiol 2007;28:1078-84. [Crossref] [PubMed]
33. Koo TK, Li MY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J Chiropr Med 2016;15:155-63. [Crossref] [PubMed]
34. Gisev N, Bell JS, Chen TF. Interrater agreement and interrater reliability: key concepts, approaches, and applications. Res Social Adm Pharm 2013;9:330-8. [Crossref] [PubMed]
35. Bjornerud A, Sorensen AG, Mouridsen K, Emblem KE. T1- and T2*-dominant extravasation correction in DSC-MRI: part I--theoretical considerations and implications for assessment of tumor hemodynamic properties. J Cereb Blood Flow Metab 2011;31:2041-53. [Crossref] [PubMed]
36. Emblem KE, Bjornerud A, Mouridsen K, Borra RJ, Batchelor TT, Jain RK, Sorensen AG T. (1)- and T(2)(*)-dominant extravasation correction in DSC-MRI: part II-predicting patient outcome after a single dose of cediranib in recurrent glioblastoma patients. J Cereb Blood Flow Metab 2011;31:2054-64. [Crossref] [PubMed]
37. Kelm ZS, Korfiatis PD, Lingineni RK, Daniels JR, Buckner JC, Lachance DH, Parney IF, Carter RE, Erickson BJ. Variability and accuracy of different software packages for dynamic susceptibility contrast magnetic resonance imaging for distinguishing glioblastoma progression from pseudoprogression. J Med Imaging (Bellingham) 2015;2:026001. [Crossref] [PubMed]
38. Schmainda KM, Prah MA, Hu LS, Quarles CC, Semmineh N, Rand SD, Connelly JM, Anderies B, Zhou Y, Liu Y, Logan B, Stokes A, Baird G, Boxerman JL. Moving Toward a Consensus DSC-MRI Protocol: Validation of a Low-Flip Angle Single-Dose Option as a Reference Standard for Brain Tumors. AJNR Am J Neuroradiol 2019;40:626-33. [Crossref] [PubMed]
39. Clement P, Mutsaerts HJ, Václavů L, Ghariq E, Pizzini FB, Smits M, Acou M, Jovicich J, Vanninen R, Kononen M, Wiest R, Rostrup E, Bastos-Leite AJ, Larsson EM, Achten E. Variability of physiological brain perfusion in healthy subjects - A systematic review of modifiers. Considerations for multi-center ASL studies. J Cereb Blood Flow Metab 2018;38:1418-37. [Crossref] [PubMed]
40. Pons-Escoda A, Garcia-Ruiz A, Naval-Baudin P, Grussu F, Fernandez JJS, Simo AC, Sarro NV, Fernandez-Coello A, Bruna J, Cos M, Perez-Lopez R, Majos C. Voxel-level analysis of normalized DSC-PWI time-intensity curves: a potential generalizable approach and its proof of concept in discriminating glioblastoma and metastasis. Eur Radiol 2022;32:3705-15. [Crossref] [PubMed]
41. Quarles CC, Gochberg DF, Gore JC, Yankeelov TE. A theoretical framework to model DSC-MRI data acquired in the presence of contrast agent extravasation. Phys Med Biol 2009;54:5749-66. [Crossref] [PubMed]
42. Semmineh NB, Xu J, Skinner JT, Xie J, Li H, Ayers G, Quarles CC. Assessing tumor cytoarchitecture using multiecho DSC-MRI derived measures of the transverse relaxivity at tracer equilibrium (TRATE). Magn Reson Med 2015;74:772-84. [Crossref] [PubMed]
43. Sorensen AG. Perfusion MR imaging: moving forward. Radiology 2008;249:416-7. [Crossref] [PubMed]
44. Teunissen WHT, Lavrova A, van den Bent M, van der Hoorn A, Warnert EAH, Smits M. Arterial spin labelling MRI for brain tumour surveillance: do we really need cerebral blood flow maps? Eur Radiol 2023;33:8005-13. [Crossref] [PubMed]
45. Cho NS, Hagiwara A, Sanvito F, Ellingson BM. A multi-reader comparison of normal-appearing white matter normalization techniques for perfusion and diffusion MRI in brain tumors. Neuroradiology 2023;65:559-68. [Crossref] [PubMed]
46. Jackson A. Analysis of dynamic contrast enhanced MRI. Br J Radiol 2004;77:S154-66. [Crossref] [PubMed]
47. Vonken EP, van Osch MJ, Bakker CJ, Viergever MA. Simultaneous quantitative cerebral perfusion and Gd-DTPA extravasation measurement with dual-echo dynamic susceptibility contrast MRI. Magn Reson Med 2000;43:820-7. [Crossref] [PubMed]
48. Bammer R, Amukotuwa SA. Dynamic Susceptibility Contrast Perfusion, Part 2: Deployment With and Without Contrast Leakage Present. Magn Reson Imaging Clin N Am 2024;32:25-45. [Crossref] [PubMed]
49. Schmiedeskamp H, Andre JB, Straka M, Christen T, Nagpal S, Recht L, Thomas RP, Zaharchuk G, Bammer R. Simultaneous perfusion and permeability measurements using combined spin- and gradient-echo MRI. J Cereb Blood Flow Metab 2013;33:732-43. [Crossref] [PubMed]
50. Pons-Escoda A, Garcia-Ruiz A, Garcia-Hidalgo C, Gil-Solsona R, Naval-Baudin P, Martin-Noguerol T, Fernandez-Coello A, Flores-Casaperalta S, Fernandez-Viñas M, Gago-Ferrero P, Oleaga L, Perez-Lopez R, Majos C. MR dynamic-susceptibility-contrast perfusion metrics in the presurgical discrimination of adult solitary intra-axial cerebellar tumors. Eur Radiol 2023;33:9120-9. [Crossref] [PubMed]
51. Stokes AM, Semmineh N, Quarles CC. Validation of a T1 and T2* leakage correction method based on multiecho dynamic susceptibility contrast MRI using MION as a reference standard. Magn Reson Med 2016;76:613-25. [Crossref] [PubMed]