IVIM、ASL及DWI对高级别胶质瘤1年无进展生存期的预测价值
发布时间：2018-09-14 12:06
【摘要】：目的:本研究旨在探讨高级别胶质瘤患者的MRI多功能成像(3D-PCASL、常规DWI及IVIM-DWI技术)和临床特征,探讨其对高级别胶质瘤1年无进展生存期的预测价值,找出独立预后因素。材料与方法:1.收集高级别胶质瘤患者31例(男性15例,女性16例),年龄范围:12-73岁,平均年龄53.1岁,包括WHO III级14例,WHO IV级17例。2.所有患者均在3.0T磁共振完成常规颅脑平扫序列(横断位T1WI、T2WI、DWI、矢状位T1WI、冠状位T2FLAIR);多b值扩散加权成像序列(IVIM-DWI)和三维伪连续式动脉自旋标记序列(3D-PCASL);增强扫描(横断位、矢状位及冠状位T1WI)。3.通过AW4.6工作站Function tool软件,IVIM-DWI经后处理获得慢扩散系数D(slow ADC)、快扩散系数D*(fast ADC)及快扩散所占的容积分数f(fraction of fast ADC)三种参数化图像;常规DWI经后处理获得扩散系数(ADC)图像;ASL经后处理获得脑血流量(CBF)图像。4.结合多方位T1WI增强扫描图像,在IVIM后处理中选取肿瘤D图最低区域手动绘制感兴趣区,得到参数D值、D*值和f值。在ASL后处理中选取高灌注区绘制感兴趣区,得到CBF值,同时对侧正常脑实质放置相应感兴趣区,计算出rCBF值(rCBF值=病灶CBF值/镜像侧CBF值)。在DWI后处理中选取ADC图最低区域手动绘制感兴趣区,得到参数ADC值。均绘制3个感兴趣区,尽量避开肿瘤囊变、坏死区域及血管走形区域,取均值,ROI面积约50mm2。5.所有患者均行手术完全切除或部分切除肿瘤,并术后进行替莫唑胺的辅助化疗和同步放疗。所有患者术后每3个月做一次MRI检查随访。本研究的起点是肿瘤手术时间,研究终点是肿瘤影像学进展时间或死亡日期,或最后一次随访观察日期(即2017年3月1日)。统计出患者的无进展生存时间。6.统计学分析:(1)所有患者的肿瘤平均rCBF、ADC、D、D*及f值均进行ROC曲线分析,得到其阈值、敏感度及特异度,并与病理学分级进行比较。(2)采用Kaplan-Meier方法统计患者生存率,并采用Log rank法对患者生存情况进行检验。(3)采用Cox比例风险回归模型对所有影像学参数和临床特征分别进行单因素和多因素分析。所有统计分析均以P0.05表示差异有统计学意义。结果:1.临床特征:(1)高级别胶质瘤患者(共31例):男性/女性:15/16例;年龄12-73岁,平均年龄53.1岁;手术完全/部分切除肿瘤:20/11例;PFS为63-732天;1年内无进展/进展人数:16/15例。(2)III级(共14例):男性/女性:6/8例;年龄30-67岁;手术完全/部分切除肿瘤:10/4例;PFS为90-473天;1年内无进展/进展人数:8/6例。(3)IV级(共17例):男性/女性:9/8例;年龄12-73岁;手术完全/部分切除肿瘤:10/7例;PFS为63-732天;1年内无进展/进展人数:8/9例。2.ROC曲线分析(各参数值对于高级别胶质瘤1年PFS预测价值):(1)IVIM:(1)D*值的Az=0.602,阈值为3.28,敏感度和特异度分别为86.7%和50.0%。(2)D值的Az=0.725,阈值为0.522,敏感度和特异度分别为66.7%和87.5%。(3)f值的Az=0.708,阈值为0.408,敏感度和特异度分别为86.7%和62.5%。(2)ASL:rCBF值的Az=0.825,阈值为6.408,敏感度和特异度分别为66.7%和87.5%。(3)DWI:ADC值的Az=0.771,阈值为0.741,敏感度和特异度分别为80.0%和81.2%。3.Kaplan-Meier曲线分析(各参数值预测生存期的价值有无统计学差异):(1)IVIM:(1)D*3.28×10-3mm2/s时,PFS无统计学差异(P=0.067)。(2)D≤0.522×10-3mm2/s时,PFS明显下降,有统计学差异(P=0.00004)。(3)f≤0.408时,PFS明显下降,有统计学差异(P=0.007)。(2)ASL:rCBF6.408时,PFS明显下降,有统计学差异(P=0.003)。(3)DWI:ADC≤0.741×10-3mm2/s时,PFS明显下降,有统计学差异(P=0.00005)。4.Cox回归模型单因素分析(高级别胶质瘤1年PFS的影响因素):(1)临床特征:(1)性别、年龄的P值分别为0.086、0.072,HR分别为2.570、1.046,95%CI分别为0.876-7.542、0.996-1.099。(2)完全切除肿瘤、病理分级的P值分别为0.0002、0.420,HR分别为8.167、1.532,95%CI分别为2.688-24.80、0.544-4.313。(2)IVIM:(1)D值、D*值的P值分别为0.0003、0.088,HR分别为0.134、3.676,95%CI分别为0.044-0.404、0.826-16.36。(2)f值的P值为0.018,HR为0.164,95%CI为0.037-0.731。(3)ASL:rCBF值的P值为0.006,HR为4.587,95%CI为1.552-13.563。(4)DWI:ADC值的P值为0.001,HR为0.107,95%CI为0.029-0.388。5.Cox回归模型多因素分析(高级别胶质瘤1年PFS的独立预后因素):(1)临床特征:完全切除肿瘤的P值为0.023,HR为4.582,95%CI为1.233-17.025。(2)IVIM:(1)D值的P值为0.620,HR为0.617,95%CI为0.091-4.166。(2)f值的P值为0.006,HR为0.080,95%CI为0.013-0.490。(3)ASL:rCBF值的P值为0.029,HR为5.162,95%CI为1.180-22.577。(4)DWI:ADC值的P值为0.081,HR为0.132,95%CI为0.014-1.285。结论:ASL成像中的rCBF值和IVIM成像中的f值均是预测高级别胶质瘤1年PFS较好的独立预后因素,其中rCBF值的预测效能最高(AUC=0.825,阈值为6.408),IVIM-f值的预测效能中等(AUC=0.708),其它的影像学参数均不是预测PFS的独立预后因素。而IVIM扫描时间过长,为ASL的1.5倍,图像后处理较繁琐。DWI作为常规成像,从信号强度判断良恶性有一定价值,ADC值的预测效能不如rCBF值。因此,三种MRI功能成像中,ASL的rCBF值对于高级别胶质瘤1年无进展生存期的预测价值是最好的。
[Abstract]:Objective: To investigate the MRI multifunctional imaging (3D-PCASL, conventional DWI and IVIM-DWI) and clinical features in patients with high-grade gliomas, and to explore its predictive value for 1-year progression-free survival of high-grade gliomas, and to identify independent prognostic factors. Materials and methods: 1. 31 patients with high-grade gliomas (15 males and 16 females) were collected. Age range: 12-73 years, mean age 53.1 years, including WHO III grade 14 cases, WHO IV grade 17 cases.2. All patients were performed routine brain plain scan sequence (transverse T1WI, T2WI, DWI, sagittal T1WI, coronal T2FLAIR); Multi-b value diffusion weighted imaging sequence (IVIM-DWI) and three-dimensional pseudo-continuous arterial spin labeling sequence (3D-PCASL); Strong scan (transverse, sagittal and coronal T1WI). 3. Through AW4.6 workstation Function tool software, IVIM-DWI after post-processing to obtain slow diffusion coefficient D (slow ADC), fast diffusion coefficient D * (fast ADC) and fast diffusion volume fraction of fast ADC (fraction of fast ADC) three parametric images; conventional DWI after post-processing to obtain diffusion coefficient (ADC) images; Cerebral blood flow (CBF) images were obtained by SL post-processing. 4. Combined with multi-directional T1WI enhanced images, the lowest region of tumor D was selected to draw the region of interest manually in IVIM post-processing, and the parameters D, D * and F were obtained. After DWI, the lowest region of ADC was selected to draw the region of interest manually, and the parameters of ADC were obtained. All three regions of interest were drawn to avoid cystic degeneration, necrotic area and vascular transverse area. The mean ROI area was about 50mm2.5. All patients underwent complete resection. All patients received adjuvant chemotherapy and concurrent radiotherapy with temozolomide every three months. The starting point of this study was the time of tumor surgery. The end point of the study was the time of tumor imaging progression or death, or the last follow-up observation date (March 1, 2017). Statistical analysis: (1) ROC curves were used to analyze the tumor mean values of rCBF, ADC, D, D * and f, and the thresholds, sensitivity and specificity were obtained, and compared with pathological grades. (2) Kaplan-Meier method was used to calculate the survival rate of patients, and Log rank method was used to test the survival of patients. (3) The Cox proportional hazard regression model was used to analyze all imaging parameters and clinical features. All statistical analyses were statistically significant with P 0.05. Results: 1. Clinical features: (1) High-grade glioma patients (31 cases): male / female: 15/16 cases; age 12-73 years old, average age 53.1 years old; operation finished; Total/partial tumor resection: 20/11 cases; PFS: 63-732 days; no progress / progression in 1 year: 16/15 cases. (2) Grade III (14 cases): male/female: 6/8 cases; age 30-67 years; total/partial tumor resection: 10/4 cases; PFS: 90-473 days; no progress / progression in 1 year: 8/6 cases. (3) Grade IV (17 cases): male/female: 9/8 cases; age 12-73 years; surgery; Complete/partial resection of the tumor: 10/7; PFS 63-732 days; no progress/progression within one year: 8/9.2. ROC curve analysis (predictive value of each parameter value for PFS in high-grade gliomas for one year): (1) IVIM: (1) D * Az = 0.602, threshold value 3.28, sensitivity and specificity 86.7% and 50.0% respectively. (2) D Az = 0.725, threshold value 0.522, sensitivity and specificity (3) Az = 0.708, threshold value is 0.408, sensitivity and specificity are 86.7% and 62.5% respectively. (2) Az = 0.825, threshold value is 6.408, sensitivity and specificity are 66.7% and 87.5%. (3) Az = 0.771, sensitivity and specificity are 80.0% and 81.2% respectively. (2) When D < 0.522 *10-3mm2/s, PFS decreased significantly (P = 0.00004). (3) When f < 0.408, PFS decreased significantly (P = 0.007). (2) When ASL: rCBF6.408, PFS decreased significantly (P = 0.007). Statistical difference (P = 0.003). (3) When DWI: ADC < 0.741 *10-3 mm2/s, PFS decreased significantly, with statistical difference (P = 0.00005). 4. Cox regression model univariate analysis (influencing factors of PFS in high grade gliomas for one year): (1) Clinical features: (1) P values of sex, age were 0.086, 0.072, HR were 2.570, 1.046, 95% CI were 0.876-7.542, 0.996-1.099, respectively. (2) For complete resection of tumor, the P values of pathological grade were 0.0002, 0.420, 0.0002, 0.420, HR 8.167, 1.532, 95% CI were 2.688-24.80, 0.544-4.313. (2) IVIM: (1) D value, P value of D * value 0.0003, 0.088, P value 0.088, HR 0.134, 3.676, 95% CI were 0.134, 3.134, 3.676, 95% CI were 0.044-0.044-0.404, 0.40.826-16.36. (2) F value was 0.018, HR 0.018, HR 0.164, HR 0.164, 95% CI 0.164 731. (3) ASL: rC The P value of BF was 0.006, HR was 4.587, 95% CI was 1.552-13.563. (4) The P value of DWI: ADC was 0.001, HR was 0.107, 95% CI was 0.029-0.388.5. Multivariate analysis of Cox regression model (independent prognostic factors of 1-year PFS in high grade gliomas): (1) Clinical features: P value of completely resected tumors was 0.023, HR was 4.582, 95% CI was 1.233-17.025. (2) IVIM:(1) P value (2) P value of F value was 0.006, HR was 0.080, 95% CI was 0.013-0.490. (3) P value of ASL: rCBF value was 0.029, HR was 5.162, 95% CI was 1.180-22.577. (4) P value of DWI: ADC value was 0.081, HR was 0.132, 95% CI was 0.014-1.285. Among the independent prognostic factors, rCBF had the highest predictive power (AUC = 0.825, threshold 6.408), IVIM-f had the moderate predictive power (AUC = 0.708), and other imaging parameters were not independent prognostic factors for PFS. The predictive power of ADC value is inferior to that of rCBF value. Therefore, ASL rCBF value is the best predictive value for 1-year progression-free survival of high-grade gliomas.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R445.2;R739.41
本文编号：2242662
[Abstract]:Objective: To investigate the MRI multifunctional imaging (3D-PCASL, conventional DWI and IVIM-DWI) and clinical features in patients with high-grade gliomas, and to explore its predictive value for 1-year progression-free survival of high-grade gliomas, and to identify independent prognostic factors. Materials and methods: 1. 31 patients with high-grade gliomas (15 males and 16 females) were collected. Age range: 12-73 years, mean age 53.1 years, including WHO III grade 14 cases, WHO IV grade 17 cases.2. All patients were performed routine brain plain scan sequence (transverse T1WI, T2WI, DWI, sagittal T1WI, coronal T2FLAIR); Multi-b value diffusion weighted imaging sequence (IVIM-DWI) and three-dimensional pseudo-continuous arterial spin labeling sequence (3D-PCASL); Strong scan (transverse, sagittal and coronal T1WI). 3. Through AW4.6 workstation Function tool software, IVIM-DWI after post-processing to obtain slow diffusion coefficient D (slow ADC), fast diffusion coefficient D * (fast ADC) and fast diffusion volume fraction of fast ADC (fraction of fast ADC) three parametric images; conventional DWI after post-processing to obtain diffusion coefficient (ADC) images; Cerebral blood flow (CBF) images were obtained by SL post-processing. 4. Combined with multi-directional T1WI enhanced images, the lowest region of tumor D was selected to draw the region of interest manually in IVIM post-processing, and the parameters D, D * and F were obtained. After DWI, the lowest region of ADC was selected to draw the region of interest manually, and the parameters of ADC were obtained. All three regions of interest were drawn to avoid cystic degeneration, necrotic area and vascular transverse area. The mean ROI area was about 50mm2.5. All patients underwent complete resection. All patients received adjuvant chemotherapy and concurrent radiotherapy with temozolomide every three months. The starting point of this study was the time of tumor surgery. The end point of the study was the time of tumor imaging progression or death, or the last follow-up observation date (March 1, 2017). Statistical analysis: (1) ROC curves were used to analyze the tumor mean values of rCBF, ADC, D, D * and f, and the thresholds, sensitivity and specificity were obtained, and compared with pathological grades. (2) Kaplan-Meier method was used to calculate the survival rate of patients, and Log rank method was used to test the survival of patients. (3) The Cox proportional hazard regression model was used to analyze all imaging parameters and clinical features. All statistical analyses were statistically significant with P 0.05. Results: 1. Clinical features: (1) High-grade glioma patients (31 cases): male / female: 15/16 cases; age 12-73 years old, average age 53.1 years old; operation finished; Total/partial tumor resection: 20/11 cases; PFS: 63-732 days; no progress / progression in 1 year: 16/15 cases. (2) Grade III (14 cases): male/female: 6/8 cases; age 30-67 years; total/partial tumor resection: 10/4 cases; PFS: 90-473 days; no progress / progression in 1 year: 8/6 cases. (3) Grade IV (17 cases): male/female: 9/8 cases; age 12-73 years; surgery; Complete/partial resection of the tumor: 10/7; PFS 63-732 days; no progress/progression within one year: 8/9.2. ROC curve analysis (predictive value of each parameter value for PFS in high-grade gliomas for one year): (1) IVIM: (1) D * Az = 0.602, threshold value 3.28, sensitivity and specificity 86.7% and 50.0% respectively. (2) D Az = 0.725, threshold value 0.522, sensitivity and specificity (3) Az = 0.708, threshold value is 0.408, sensitivity and specificity are 86.7% and 62.5% respectively. (2) Az = 0.825, threshold value is 6.408, sensitivity and specificity are 66.7% and 87.5%. (3) Az = 0.771, sensitivity and specificity are 80.0% and 81.2% respectively. (2) When D < 0.522 *10-3mm2/s, PFS decreased significantly (P = 0.00004). (3) When f < 0.408, PFS decreased significantly (P = 0.007). (2) When ASL: rCBF6.408, PFS decreased significantly (P = 0.007). Statistical difference (P = 0.003). (3) When DWI: ADC < 0.741 *10-3 mm2/s, PFS decreased significantly, with statistical difference (P = 0.00005). 4. Cox regression model univariate analysis (influencing factors of PFS in high grade gliomas for one year): (1) Clinical features: (1) P values of sex, age were 0.086, 0.072, HR were 2.570, 1.046, 95% CI were 0.876-7.542, 0.996-1.099, respectively. (2) For complete resection of tumor, the P values of pathological grade were 0.0002, 0.420, 0.0002, 0.420, HR 8.167, 1.532, 95% CI were 2.688-24.80, 0.544-4.313. (2) IVIM: (1) D value, P value of D * value 0.0003, 0.088, P value 0.088, HR 0.134, 3.676, 95% CI were 0.134, 3.134, 3.676, 95% CI were 0.044-0.044-0.404, 0.40.826-16.36. (2) F value was 0.018, HR 0.018, HR 0.164, HR 0.164, 95% CI 0.164 731. (3) ASL: rC The P value of BF was 0.006, HR was 4.587, 95% CI was 1.552-13.563. (4) The P value of DWI: ADC was 0.001, HR was 0.107, 95% CI was 0.029-0.388.5. Multivariate analysis of Cox regression model (independent prognostic factors of 1-year PFS in high grade gliomas): (1) Clinical features: P value of completely resected tumors was 0.023, HR was 4.582, 95% CI was 1.233-17.025. (2) IVIM:(1) P value (2) P value of F value was 0.006, HR was 0.080, 95% CI was 0.013-0.490. (3) P value of ASL: rCBF value was 0.029, HR was 5.162, 95% CI was 1.180-22.577. (4) P value of DWI: ADC value was 0.081, HR was 0.132, 95% CI was 0.014-1.285. Among the independent prognostic factors, rCBF had the highest predictive power (AUC = 0.825, threshold 6.408), IVIM-f had the moderate predictive power (AUC = 0.708), and other imaging parameters were not independent prognostic factors for PFS. The predictive power of ADC value is inferior to that of rCBF value. Therefore, ASL rCBF value is the best predictive value for 1-year progression-free survival of high-grade gliomas.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R445.2;R739.41
【参考文献】
相关期刊论文 前1条
1 王健;钱银锋;杨奇芳;李敏;;多b值DWI鉴别高级别胶质瘤及脑转移瘤[J];临床放射学杂志;2015年10期
，本文编号：2242662
本文链接：https://www.wllwen.com/yixuelunwen/fangshe/2242662.html
