Discussion
In gliomas, particularly malignant ones, neovascularity was significantly increased with extremely irregular morphologies and composed of endothelial cells and a basement membrane with incomplete structures. In this neovascularity, vascular resistance significantly increased and intravascular pressure rose, which usually resulted in an increase in vascular permeability and the high likelihood of ruptures and bleeding. To some extent, DCE-MRI and/or SWI reflected that the above pathophysiological changes in tumor neovascularity were somehow associated with glioma malignancy [
4,
5,
8-
10]. In low grade gliomas (LGGs), especially astrocytomas, there were no or sporadic dot-like ITSS within tumors, which displayed low K
trans values. Thus, there was consistency between ITSS and K
trans values, indicating the unchanged permeability and low density in tumor neovascularity, while the vascular characteristics of oligodendrogliomas and oligoastrocytomas were quite different from those of astrocytomas since their grades of ITSS were higher and K
trans values were relatively lower. Additionally, a phenomenon of non-correspondence between the ITSS tufts and the areas of maximal K
trans values often occurred in oligodendrogliomas and oligoastrocytomas. This phenomenon indicated that vascular structures in the significantly increased tumor neovascularity were nearly complete. Thus, the vascular permeability was not significantly changed, whereas in HGGs, conglomerated dot-like and fine linear ITSS were frequently observed with significantly increased K
trans values. Notably, areas of the highest value of K
trans did not always accurately correspond to the exact region with the densest ITSS. Considering these radiographic inconsistencies between DCE-MRI and SWI in HGGs, the main reason was hypothesized to be that the detection of ITSS in HGGs not only reflected tumor vascularity distribution but also indicated considerable susceptibility associated with micro-hemorrhage and necrosis within tumors, while the areas of the highest value of K
trans represented vascularization with a high proportion of immature, hyperpermeable microvessels.
In this study, K
trans values of HGGs were significantly higher than those of LGGs, which meant that HGGs had higher microvascular permeability. So, contrast agents were transferred from plasma to the EES more easily, and plenty of malignant, immature, and hyperpermeable microvessels existed in HGGs. The ROC curve analysis results showed that the cut-off values of K
trans provided good diagnostic efficacy (their diagnostic sensitivity and specificity were either near or above 90%) for distinguishing between LGGs and HGGs and between grade II and grade IV, which was consistent with the previous report [
14]. Therefore, grading gliomas via the assessment of tumor vascular permeability is highly feasible. Spearman’s correlation analysis showed that the K
trans value was strongly correlated with tumor grade (
r = 0.782,
P < 0.01), so K
trans could be a better biomarker to assess the grading of gliomas. However, there was no statistical difference in K
trans between grade III and grade IV. This result was concordant with some reports [
8]. This might be attributed to their similar pathological microvascular patterns and because there was abundant microvascular hyperplasia within those two malignant progressive gliomas [
15]. The pathological data showed that the mean values of both MVD and VD of LGGs were significantly lower than those of HGGs. This was largely because HGGs were prone to stimulating the secretion of VEGF, which might be the vascular morphogen that formed the abnormally large vessels [
16,
17]. One of the most important factors affecting the K
trans value was blood flow. The increase of MVD and VD values caused the increased leakage of the contrast agent in unit time, which occurred more frequently in HGGs. Spearman’s correlation analysis showed K
trans values were moderately correlated with MVD values (
r = 0.474,
P < 0.01) and strongly correlated with VD values (
r = 0.692,
P < 0.01), indicating that K
trans values were more vulnerable to the impacts of VD sizes.
V
e is defined as the volume fraction of contrast agent transfer from the vessel into the EES, and many findings on the relationship between V
e and glioma grade have been mentioned. The present results, being concordant with previous studies [
8,
18], showed that the mean V
e value in LGGs was significantly lower than that in HGGs, indicating that the leakage volume of contrast agent into EES was greater in HGGs than in LGGs. The ROC curve analysis showed that the cut-off value of V
e (0.296) also provided high sensitivity (92.9%) and specificity (91.7%), which helped differentiate LGGs from HGGs. The cut-off value of V
e (0.345) also provided the best combination of sensitivity (88.9%) and specificity (93.3%), which helped differentiate grade II from grade IV gliomas. Both AUC were greater than 80%. Spearman’s correlation analysis showed that V
e values were strongly correlated with K
trans values (
r = 0.823,
P < 0.01). Moreover, V
e could be expressed mathematically as the ratio of the contrast agent quantity that leaked into the EES to the contrast agent quantity that returned to the plasma space [
19], indicating a close relationship between K
trans and V
e values. V
e values were weakly correlated with MVD values (
r = 0.379,
P < 0.05) but moderately correlated with VD values (
r = 0.586,
P < 0.01). This suggested that VD values within gliomas played an important role in the influence of V
e values. Thus, the present study also demonstrated that V
e could be a biological marker for glioma grading. EES was easily influenced by some factors such as cell density, necrosis, cystic lesions, and extracellular stroma. A previous study showed that necrotic or cystic regions increased the volume of EES [
7], while the area with higher cellularity decreased it. M. Aref et al. [
20] demonstrated that extracellular spaces and V
e measured by both DCE-MRI and microscopic analysis were statistically similar. Therefore, if the volume of the EES was changed, it resulted in a corresponding change in V
e. With the rapid growth and metabolism requirements of HGGs, the tumors more easily produced regional cellular hypoxia and necrosis or cystic degeneration, which consequently increased EES volume. An animal experiment also confirmed that both the progression of tumor vascularization and the increase of the EES were closely related with tumor growth [
21]. The present results showed that the V
e value was strongly correlated with tumor grade (
r = 0.717,
P < 0.01), and there were significant differences between LGGs and HGGs and between grade II and grade IV gliomas. This phenomenon could be explained by the larger volume of EES in HGGs due to great physiological and metabolic changes.
Clinically, distinguishing grade II from grade III gliomas is very important because the prognoses for patients with grade II gliomas are significantly better than for patients with grade III gliomas [
22]. The present study showed that both K
trans and V
e values of grade II gliomas were significantly lower than those of grade III gliomas (
P < 0.01). The cut-off values of K
trans = 0.045 min
−1 and V
e = 0.296 were adopted for differentiation between grade II and grade III gliomas, and high sensitivity and specificity (greater than 85%) were achieved. Therefore, these results provided important clinical information for judging the development or progression of grade II to grade III gliomas.
Several studies demonstrated SWI was a promising noninvasive method for differentiating between LGGs and HGGs according to the different frequencies and appearances of ITSS [
23-
25]. ITSS could simultaneously reflect the intratumoral venous structures and micro-bleeding. The present results indicated that the highest degrees of ITSS were observed in almost all HGGs (except for one case of glioblastoma), suggesting that ITSS can be a potentially helpful sign for the correct diagnosis of HGGs. Also, there was a significant difference in ITSS degree between LGGs and HGGs (P < 0.01) and between grade II and grade III or IV gliomas (
P < 0.01,
P < 0.05, respectively). This finding was similar to that described by Park et al. [
10]. Thus, the degree of ITSS could be used for grading gliomas. However, the present study also found that the degree of ITSS showed a moderate correlation with glioma grade (
r = 0.515,
P < 0.01), which was very different from previous studies. The main reason was assumed to be that the grade II oligodendrogliomas and oligoastrocytomas were also enrolled in this study, while former studies only focused on the differences between high grade and lower grade astrocytomas [
9,
10]. Significantly increased angiogenesis and highly dense vascularity or mild bleeding were observed in either oligodendrogliomas or oligoastrocytomas, so their grades of ITSS were generally rather high. The present study showed that the degree of ITSS was highly correlated with VD (
r = 0.629) and moderately correlated with MVD, indicating that, with the exception of angiogenesis and micro-bleeds, the VD size may have hugely impacted magnetic susceptibility to some extent.
At the early stage of gliomas, microvessels are similar to normal brain capillaries. However, in the intermediate stage, they become tortuous, disorganized, and dilated, and in the advanced stage, they change into anarchic and aberrant structures with topographies such as multilayered “glomeruloid tufts,” “garland vessels,” and huge dilated vessels [
26]. Remarkably, one glioblastoma showed no evidence of ITSS on SWI, indicating that no micro-hemorrhaging, necrosis, or calcification existed within the tumor. However, this tumor prominently showed high K
trans values on DCE-MRI, indicating the high permeability of microvasculature. It was unknown why these microvasculatures were not detected by SWI, though it was speculated that the fine microvasculature within the tumors did not have a great enough susceptibility effect, which was depicted using ITSS. Further studies with larger populations are needed to test this deduction.
Tumor enhancement is the relaxation enhancement caused by pericerebral collections of contrast leakage from the blood-brain barrier (BBB) due to BBB destruction or disintegration. BBB destruction in gliomas may result from the vascular damages caused by tumor formation or the immature endothelium of the tumor neovascularity, as these two factors can both contribute to the contrast leakage and further lead to the enhancement [
27,
28]. K
trans values mainly reflected the permeability of the neovascularity. Therefore, intratumoral distributions of high K
trans value areas and contrast collection areas were significantly different, with the former located medially to the tumor enhancement areas and the latter residing inside the tumor. ITSS also reflected the angiogenesis degrees. In low grade astrocytomas, changes and distributions of ITSS were consistent with K
trans values. However, in oligodendrogliomas, the degree of ITSS was dense while K
trans values were not too high, and the areas of densely prominent ITSS did not completely correspond with the areas of maximal K
trans values. Therefore, when analyzing the degrees of tumor enhancement and tumor vascular characteristics to obtain a correct grading assessment, special attention should be paid to the changes in intratumoral vascular characteristics reflected by the various MRI parameters.
This study had some limitations. First, the patient population was small, especially for grade IV tumors, and it was difficult to draw meaningful statistical conclusions from these studies. Second, it was a challenge to assess the voxel-to-voxel correlation between abnormal signals of DCE-MRI and pathological specimens. Third, ITSS on SWI were associated with tumor micro-hemorrhage and necrosis, which could potentially degrade the gradient echo images used for DCE-MRI so that the Ktrans measurements might be unreliable in the areas of considerable ITSS. Therefore, a larger sample size and more appropriate method should be adopted in future studies to test these results.
Authors’ contributions
XGL performed the study and drafted the manuscript. YSZ and HYK performed the statistical analyses and image post-processing. YLZ performed the scanning sequence. HPL and SMW participated in the design of the study and performed the analysis with constructive discussions. WGZ conceived the study idea, participated in its design and coordination, and helped draft the manuscript. All authors read and approved of the final manuscript.