Prediction of diffuse glioma grade and tumor cell proliferative activity by synthetic MRI combined with three dimensional arterial spin labeling imaging

葛鑫,孙胜玉,刘文潇,赵建国,申颖,吕瑞瑞,黄雪莹,王晓东
DOI: https://doi.org/10.3760/cma.j.cn112149-20210617-00569
2022-01-01
Abstract:Objective:To evaluate the value of synthetic MRI combined with three dimensional-arterial spin labeling (3D-ASL) imaging in the grading of diffuse glioma and its correlation with tumor cell proliferative activity (Ki-67).Methods:This study was prospective. The clinical and imaging manifestations of 66 patients with diffuse glioma who underwent synthetic MRI combined with 3D-ASL imaging from August 2020 to June 2021 in General Hospital of Ningxia Medical University were analyzed. Among 66 patients, there were 36 males and 30 females, aged 4-76 years, and divided into low grade glioma (LGG) group ( n=25) (WHO Ⅱ) and high grade glioma (HGG) group ( n=41) (WHO Ⅲ and vⅣ). T 1, T 2, proton density (PD) and cerebral blood flow (CBF) of tumor parenchyma were measured by GE ADW4.7 postprocessing software. The Ki-67 label index (Ki-67 LI) in postoperative pathological sections was detected by immunohistochemistry. Independent sample t test or Mann-Whitney U test was used to compare the differences of quantitative parameters between HGG group and LGG group. Receiver operating characteristic (ROC) curve was used to analyze the diagnostic efficacy of T 1, PD, CBF and the combination. Spearman test was used to analyze the correlation between the parameters and Ki-67 label index (LI). Results:T 1[(1 573±173)ms], PD[(86.2±2.4)pu] and CBF[(129±48)ml·100 g -1·min -1] in HGG group were significantly higher than those in LGG group [(1 376±134)ms, (83.0±2.5)pu and (77±49)ml·100g -1·min -1 respectively], and difference had statistical significance ( t=-4.86, -5.08, -4.24, P<0.01). ROC confirmed that the area under curve (AUC) of T 1, PD and CBF in differentiating HGG from LGG were 0.847, 0.843 and 0.777, respectively. In multi-parameter analysis, the combination of three parameters had the best diagnostic efficiency (AUC=0.973) and the sensitivity and specificity were 87.8% and 100%, respectively. In LGG and HGG groups, there was no correlation between T 1, T 2, PD, CBF and Ki-67 LI. In the overall cohort, T 1, PD and CBF had slight positive correlation with Ki-67 LI ( r=0.394, 0.411 and 0.406, respectively, all P<0.01). There was no correlation between T 2 and Ki-67 LI ( r=-0.100, P=0.423). Conclusion:Synthetic MRI and 3D-ASL can noninvasively evaluate the pathological grade of glioma and predict the expression of Ki-67, among which T 1 and PD are novel imaging marks.
What problem does this paper attempt to address?