分享:
分享到微信朋友圈
X
临床研究
胰腺神经内分泌肿瘤的CT和MRI特征对预测其病理分级的价值
王夕江 郭炜 刘剑羽

Cite this article as: WANG X J, GUO W, LIU J Y. The value of CT and MRI features of pancreatic neuroendocrine neoplasm in predicting the pathological grade[J]. Chin J Magn Reson Imaging, 2025, 16(1): 127-134.本文引用格式:王夕江, 郭炜, 刘剑羽. 胰腺神经内分泌肿瘤的CT和MRI特征对预测其病理分级的价值[J]. 磁共振成像, 2025, 16(1): 127-134. DOI:10.12015/issn.1674-8034.2025.01.019.


[摘要] 目的 探究胰腺神经内分泌肿瘤(pancreatic neuroendocrine neoplasm, panNEN)的CT和MRI特征对预测其病理分级的价值。材料与方法 回顾性分析北京大学第三医院106例panNEN患者的临床及影像资料,本研究遵循世界卫生组织(World Health Organization, WHO)2019年第五版的分类和分级标准,将panNEN中的G1、G2、G3级神经内分泌肿瘤(neuroendocrine neoplasm, NEN)和神经内分泌癌(neuroendocrine carcinoma, NEC)分别划分为低级别组(G1级NEN)和中高级别组(包括G2、G3级NEN和NEC)。对患者性别、年龄和病灶的形态、位置、体积、囊实性质、CT特征(平扫、增强动脉期和静脉期相CT值、动脉期和静脉期CT图像的增强模式)、MRI特征[T1、T2加权MRI图像上的信号强度、扩散加权成像(diffusion-weighted imaging, DWI)序列b值=1000 s/mm2图像的信号强度及表观扩散系数(apparent diffusion coefficient, ADC)图像的信号强度],以及血管侵犯和肝转移进行统计学分析。运用t检验、Mann-Whitney U检验、卡方检验及Wilcoxon秩和检验比较panNEN不同病理分级和病灶相关参数的差异,并采用二元logistic回归构建预测模型,使用受试者工作特征曲线下面积(area under the curve, AUC)评估模型预测效能,采用DeLong检验比较模型间的AUC值的差异。校准曲线评估模型的拟合度,决策曲线分析评估模型的临床价值。结果 低级别组与中高级别组在肿瘤体积、肝转移和血管侵犯方面的差异具有统计学意义(P<0.05),而在性别、年龄、囊实性质和发生部位方面的差异无统计学意义(P>0.05)。CT和MRI特征中,仅DWI和ADC图信号特征差异具有统计学意义。多因素logistic回归分析显示,肿瘤体积、肝转移和血管侵犯是panNEN病理分级的独立预测因素,联合后构建的模型预测panNEN中高级别组的AUC达0.861(95% CI:0.798~0.923),敏感度为78.1%,特异度为83.3%。结论 基于肿瘤体积、肝转移和血管侵犯的联合模型在术前能有效预测panNEN病理分级。
[Abstract] Objective To investigate the value of CT and MRI features of pancreatic neuroendocrine neoplasm (panNEN) in predicting its pathological grade.Materials and Methods The clinical and imaging data of 106 patients with panNEN in the Third Hospital of Peking University were analyzed retrospectively. According to the World Health Organization (WHO) classification and classification standard of 2019, the patients were divided into low-grade group [neuroendocrine neoplasm (NEN) of G1 grade] and middle-high-grade group [NEN of G2, G3 grade and neuroendocrine carcinoma (NEC)]. Sex, age, tumor shape, tumor location, tumor volume, cystic and solid nature, CT and MRI signal characteristics, vascular invasion and hepatic metastasis were analyzed. t test, Mann-whitney U test, chi-square test and Wilcoxon rank-sum test were used to analyze the data, and binary logistic regression was used to construct the prediction model.Results There were significant differences in tumor volume, hepatic metastasis and vascular invasion between low-grade group and middle-high-grade group, but no significant differences in sex, age, cystic nature and location. On CT and MRI, only diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) images showed significant differences in signal characteristics. Multivariate logistic regression analysis showed that tumor volume, hepatic metastasis and vascular invasion were independent predictors of panNEN pathological grade, the combined model predicted the AUC of the high-grade group in panNEN to be 0.861 (95% CI: 0.798 to 0.923), with a sensitivity of 78.1% and a specificity of 83.3%.Conclusions The combined model based on tumor volume, hepatic metastasis and vascular invasion can effectively predict panNEN pathological grade before operation and is helpful for clinical treatment decision.
[关键词] 胰腺;神经内分泌肿瘤;磁共振成像;体层摄影,X-线计算机;病理分级
[Keywords] pancreas;neuroendocrine neoplasm;magnetic resonance imaging;tomography, X-ray computed;pathological grade

王夕江 1   郭炜 2   刘剑羽 2*  

1 晋中市第二人民医院影像科,晋中 030800

2 北京大学第三医院放射科,北京 100191

通信作者:刘剑羽,E-mail:jyliubysy@163.com

作者贡献声明:刘剑羽设计本研究的方案,参与论文重要内容的修改;王夕江起草和撰写稿件,参与论文重要内容的修改,获取、分析和解释本研究的数据;郭炜获取、分析或解释本研究的数据,对论文重要内容进行了修改;全体作者都同意发表最后的修改稿,同意对本研究的所有方面负责,确保本研究的准确性和诚信。


收稿日期:2024-06-21
接受日期:2025-01-10
中图分类号:R445.2  R735.9 
文献标识码:A
DOI: 10.12015/issn.1674-8034.2025.01.019
本文引用格式:王夕江, 郭炜, 刘剑羽. 胰腺神经内分泌肿瘤的CT和MRI特征对预测其病理分级的价值[J]. 磁共振成像, 2025, 16(1): 127-134. DOI:10.12015/issn.1674-8034.2025.01.019.

0 引言

       胰腺神经内分泌肿瘤(pancreatic neuroendocrine neoplasm, panNEN)是一类罕见且具有高度异质性的肿瘤[1, 2, 3],在胃肠道胰腺(gastroenteropancreatic, GEP)系统中是第三常见的神经内分泌肿瘤(neuroendocrine neoplasm, NEN)亚型[4]。其特征在于具有神经内分泌分化并表达神经内分泌标志物,约占所有胰腺肿瘤的2%~5%[5]。世界卫生组织(World Health Organization, WHO)于2019年发布了第五版分类和分级标准,该标准依据panNEN的组织分化程度及细胞增殖活性进行分类和分级,其中细胞增殖活性通过有丝分裂计数和Ki-67指数进行评估[5]。因此,panNEN被分为G1、G2、G3级NEN及神经内分泌癌(neuroendocrine carcinoma, NEC)、混合型神经内分泌-非神经内分泌肿瘤(mixed neuroendocrinenonneuroendocrine neoplasms, MiNEN)[1]。近年来,panNEN的发病率显著增加[6, 7, 8]。目前,对于panNEN患者而言,尽管在靶向治疗和免疫治疗等方面取得了一定进展[9, 10],最佳的长期预后治疗方法仍主要依赖于手术[11]。即使是直径小于2 cm的panNEN患者,仍建议进行手术治疗[12, 13]。制订手术方案时,需要综合考虑肿瘤的功能特性、体积、位置、可切除性、分期、病理学类别及分级等多个因素,以及手术的风险与收益[14, 15]。术前准确评估和预测panNEN的病理分级对指导临床治疗决策及预后判断至关重要[16]。不同分级的panNEN表现出不同的生物学行为、治疗策略和预后,患者的整体生存期与其病理分级直接相关[17, 18]。先前的研究已证实,panNEN的CT、MRI和超声影像特征可提示肿瘤的病理分级[19, 20, 21, 22, 23, 24, 25]。然而,关于联合CT和MRI检查结果与panNEN病理分级之间关联性的国内报道仍存在空白[20]。因此,本研究旨在依据2019版WHO消化系统肿瘤分类标准,通过比较不同病理分级的panNEN的CT及MR影像特征,探讨多种影像特征指标联合构建的模型对panNEN术前病理分级的预测价值。

1 材料与方法

       本研究遵守《赫尔辛基宣言》,经北京大学第三医院伦理委员会批准,免除受试者知情同意,批准文号:M2024969。本研究回顾性分析了自2016年9月至2023年9月期间北京大学第三医院134例经手术和病理证实为panNEN患者的影像资料。纳入标准:(1)经手术病理确诊为panNEN;(2)均在术前30 d内进行过腹部CT平扫及增强和上腹部MRI平扫两种影像技术检查;(3)病灶的CT及MRI图像清晰。排除标准:(1)panNEN太小导致CT或MRI未能检出;(2)临床、影像和病理资料不完整。经纳入和排除标准筛选后106例病例参与本研究,其中包括132个肿瘤病灶。这些病例按照2019年WHO消化系统肿瘤分类标准被分为G1(42例)、G2(52例)、G3(11例)级别的NET以及NEC(1例)。

1.1 研究分组

       遵循2019版WHO第5版分类和分级标准,本研究将panNEN中的G1、G2、G3级NEN和NEC分别划分为低级别组(G1级)和中高级别组(包括G2、G3级及NEC)。尽管MiNEN的分级因其NEN和非NEN成分的多样性而复杂,但在大多数MiNEN病例中,这两种成分通常表现为低分化。NEN成分的增殖指数与其他神经内分泌癌相似[5],因此在本研究中将其暂时归入中高级别组。我们将106例panNEN患者划分为低级别组(G1级)42例和中高级别组(G2、G3级NEN和NEC)64例。

1.2 影像学检查

       所有患者术前30天内进行了腹部CT平扫及增强扫描和上腹部MRI平扫检查。使用的设备包括GE Revolution 256排512层超高端螺旋CT机(Revolution CT VICTOR,GE Medical Systems,LLC公司,美国)、Siemens双源CT机(Somaton Definition Flash,SIEMENS公司,德国)和3.0 T Discovery MR750超导MRI扫描仪(GE Healthcare公司,美国),Siemens 3.0 T Prisma MRI扫描仪(西门子医疗系统有限股份公司,德国)和体线圈。

       CT扫描参数:管电压120 kV,管电流强度200~600 mA,重建层厚5 mm,层间距5 mm。肘静脉穿刺建立注药通道,注入对比剂碘普罗胺(碘浓度300 mg/mL,拜耳医药公司,德国)80~100 mL,流率2~3 mL/s,分别于注射对比剂碘普罗胺后35 s、50~60 s扫描动脉期、静脉期。

       MRI扫描序列及参数:T1WI容积内插屏气检查(volume interpolated body examination, VIBE)脂肪抑制(fat saturation, FS)序列,TR 4.0 ms,TE 1.8 ms,层厚3 mm,层间距1 mm,FOV 360 mm×360 mm;T2WI-FS序列,TR 15 000 ms,TE 82 ms,层厚5 mm,层间距1 mm,FOV 320 mm×320 mm;T2WI半傅立叶采集单次激发快速自旋回波(half-fourier acquisition single-shot turbo spin-echo, HASTE)序列,TR 2000 ms,TE 9 ms,层厚5 mm,层间距1 mm,FOV 230 mm×906 mm;扩散加权成像(diffusion-weighted imaging, DWI)序列行自旋-平面回波成像(spin echo echo planar imaging, SE-EPI),TR 5455 ms,TE 62 ms,层厚5 mm,层间距1 mm,FOV 350 mm×350mm,b值为0和1000 s/mm2

1.3 图像分析

       在本研究中,所有影像均由两位经验丰富的放射科医师(1位12年放射诊断经验的主治医师和1位17年经验的副主任医师)独立审阅。评估过程中,两位医师在不知晓患者手术病理分级结果的情况下,共同分析了病变的多个特征,包括其位置、形态、体积、边界、密度、信号强度、强化特征、胰腺周围血管侵犯情况、淋巴结转移及肝转移等信息。

       在动脉期或静脉期影像中,阅片者会选择病灶实性部分强化最为显著的区域,手动设定感兴趣区(region of interest, ROI),并在PACS系统中利用CT值测量工具进行相关操作。为了确保CT平扫与增强各期影像中病灶位置的一致性,以及ROI面积的可比性,需遵循以下步骤:首先,使用多平面重建CT图像对病灶的最大直径和形态进行评估,确保两位放射科医生在测量结果上达成一致;其次,在CT增强动脉期影像中,通过与正常胰腺实质进行对比,目视确认实体部分的强化模式,并在静脉期影像中检查是否存在渐进式强化现象;最后,ROI应精确放置于病灶的实性部分,两位放射科医生需协商一致以确定ROI的位置和大小,绘制时尽量覆盖实性部分的最大区域。ROI的平均面积约为(502±901)mm2,范围为15~4662 mm2

1.4 统计学处理

       采用SPSS 29.0.1.0软件进行统计分析。采用Kolmogorov-Smirnov检验计量资料是否符合正态分布,正态分布的计量资料用(x¯±s)表示;不符合正态分布的计量资料采用中位数(四分位数间距)表示。符合正态分布检验和方差齐性检验的,采用独立样本t检验;不符合正态分布的采用非参数独立样本Mann-Whitney U检验或Wilcoxon符号秩和检验进行组间的差异性分析。运用多因素二元logistic回归分析(前进法)筛选panNEN中高级别组的独立预测因素,并构建预测模型。最后,利用受试者工作特征(receiver operating characteristic, ROC)曲线评估预测模型对于panNEN中高级别组的诊断效能,计算AUC、阈值、95% CI、敏感度和特异度。P<0.05为差异具有统计学意义。

2 结果

2.1 panNEN低级别组与中高级别组在临床和影像一般特征的组间比较

       在本研究中,panNEN低级别组包含42例患者,男25例,女17例,年龄范围33~67岁,平均年龄54岁,标准偏差13岁。该组中有38例无肝转移,2例发生可疑肝转移,2例明显肝转移;3例表现为肠系膜上静脉或脾静脉边缘欠清,诊断血管可疑侵犯。中高级别组共有64例患者,男37例,女27例,年龄范围39~63岁,平均年龄51岁,标准偏差12岁。在这一组中,发现7例患者可疑肝转移,24例明显肝转移;13例出现了肠系膜上静脉或脾静脉边缘模糊,9例出现血管管腔内癌栓形成。结果显示,两组在性别、年龄、囊实性以及发病部位的差异无统计学意义(P>0.05),但低级别组与中高级别组在病灶形态、体积、肝转移及血管侵犯程度的判断方面的差异具有统计学意义(P<0.05)。具体而言,中高级别组病灶的体积较大,形态更不规则,肝转移和血管侵犯的发生率均较低级别组更高(表1表2)。

表1  胰腺神经内分泌肿瘤低级别组与中高级别组患者一般临床情况比较
Tab. 1  Comparison of general clinical conditions between low-grade and high-to-intermediate grade groups of patients with pancreatic neuroendocrine neoplasm
表2  胰腺神经内分泌肿瘤低级别组与中高级别组病灶一般影像特征比较
Tab. 2  Comparison of general imaging features of pancreatic neuroendocrine neoplasm between low-grade and high-to-intermediate grade groups

2.2 panNEN低级别组与中高级别组在MRI多参数特征的组间比较

       在MRI图像上,低级别组肿瘤在T1WI序列上通常表现出较低的信号强度,而在T2WI序列上则显示出较高的信号强度(图1E~1G),并且在DWI的表观扩散系数(apparent diffusion coefficient, ADC)图上表现出较低信号(图1I);相比之下,中高级别组的情况更加复杂且多样化(图2B~2C、2F及3E~3F)。然而,在T1WI和T2WI信号特征上,两组间的比较差异无统计学意义(P>0.05);相反,在DWI和ADC图信号特征上,两组间比较的差异具有统计学意义(P<0.05),其中两组均表现出DWI高信号和ADC图低信号的特点(图1H~1I2D~2E)(表3)。

图1  男,27岁,panNEN G1级患者。胰头部实性结节,病灶约21 mm×20 mm,CT平扫呈等密度(1A,箭);增强CT动脉期“血池样”明显均匀强化(1B、1D,箭);门静脉期均匀强化,强化程度减低(1C,箭);T1WI-water胰头部等信号结节(1E,箭);脂肪抑制呈稍高信号T2WI(1F,箭);T2WI呈等信号(1G,箭);DWI呈稍高信号(1H,箭),ADC图低信号(1I,箭)。panNEN:胰腺神经内分泌肿瘤;DWI:扩散加权成像;ADC:表观扩散系数。
Fig. 1  Male, 27-year-old, patient with panNEN G1 grade. A solid nodule in the head of the pancreas measuring about 21 mm × 20 mm, on non-enhanced CT scan (1A, arrow), the lesion appeared as iso-dense; during the arterial phase of contrast-enhanced CT (1B, 1D, arrow), it exhibit significant and homogeneously "blood pool-like" enhancement; and in the portal venous phase (1C, arrow), it shows even enhancement with a decreased intensity; on T1WI-water imaging (1E, arrow), the signal intensity is iso-significant; on fat saturation T2WI (1F, arrow), there is a slightly high signal; on standard T2WI (1G, arrow), the signal is also iso-intense; on DWI (1H, arrow), there is a slightly high signal intensity; and on the ADC map (1I, arrow), the signal is low. panNEN: pancreatic neuroendocrine neoplasm; DWI: diffusion-weighted imaging; ADC: apparent diffusion coefficient.
表3  胰腺神经内分泌肿瘤低级别组与中高级别组病灶在MRI多参数特征的比较
Tab. 3  Comparison of multiparametric MRI characteristics between low-grade and high-to-intermediate grade groups of pancreatic neuroendocrine neoplasm

2.3 panNEN低级别组与中高级别组在CT平扫期、动脉期及静脉期的CT值比较

       对panNEN低级别组与中高级别组在CT平扫期、动脉期及静脉期的CT值进行比较,结果显示两组间各期CT值比较差异无统计学意义,P值均大于0.05(表4)。

图2  女,59岁,panNEN G2级患者。胰尾部占位,病灶约52 mm×43 mm,T2-HASTE呈稍高信号(2A、2F,箭);T1WI呈低信号(2B,箭);脂肪抑T2WI呈等稍高混杂信号(2C,箭);DWI呈混杂高信号(2D,箭),ADC图呈低信号(2E,箭);CT平扫为等密度(2G,箭);增强CT动脉期(2H,箭)明显不均匀强化;静脉期(2I,箭)渐进性强化,强化程度减低。panNEN:胰腺神经内分泌肿瘤;DWI:扩散加权成像;ADC:表观扩散系数。
Fig. 2  Female, 59-year-old, patient with panNEN G2 grade. A mass in the tail of the pancreas, measuring approximately 52 mm × 43 mm, on T2-HASTE sequences (2A, 2F, arrow), the lesion exhibits a slightly high signal intensity; on T1WI (2B, arrow), it displays a slightly low signal; on fat saturation T2WI (2C, arrow), the signal is isointense to slightly mixed high; on DWI (2D, arrow), a mixed high signal intensity is observed, with a low signal on the ADC map (2E, arrow); on non-enhanced CT scanning (2G, arrow), the mass is of iso-density; during the arterial phase of contrast-enhanced CT (2H, arrow), it shows marked heterogeneous enhancement; and in the venous phase (2I, arrow), it exhibits progressive enhancement with a decrease in enhancement degree over time. panNEN: pancreatic neuroendocrine neoplasm; DWI: diffusion-weighted imaging; ADC: apparent diffusion coefficient.
表4  胰腺神经内分泌肿瘤两组患者间CT平扫期、增强动脉期及静脉期的CT值比较
Tab. 4  Comparison of CT values in plain scan phase, arterial phase, and venous phase between the two groups of pancreatic neuroendocrine neoplasm patients

2.4 panNEN低级别与中高级别组在增强CT动脉期和静脉期的强化类型特征比较

       在对比分析panNEN低级别组与中高级别组在增强CT动脉期和静脉期的强化特征时发现,在动脉期相和门静脉期相,低级别组的panNEN病灶表现为均匀强化特性(图1B1C),而中高级别组则普遍呈现为不均匀的明显强化现象(图2H、3B3D)。相对地,中高级别组的肿瘤更常出现不均匀强化现象以及侵袭性血管特征(图4A~4C)。尽管如此,两组在动脉期及门静脉期强化类型的特征比较差异无统计学意义(P>0.05)(表5)。

图3  男,47岁,panNEN G3级患者。胰体部占位,病灶约57 mm×56 mm,CT平扫病灶为等密度(3A,箭);增强CT动脉期明显不均匀强化,且见肝脏转移(3B、3D,箭);门静脉期病灶呈渐进性均匀强化,强化程度增高(3C,箭);T1WI-water为不均匀稍低信号(3E,箭);脂肪抑制T2WI为不均匀稍高信号(3F,箭)。
图4  男,54岁,panNEN G3级患者。胰腺颈部不规则形占位,病灶约50 mm×41 mm,增强CT见胰十二指肠动脉受侵犯(4A~4C,红箭)。panNEN:胰腺神经内分泌肿瘤;DWI:扩散加权成像;ADC:表观扩散系数。
Fig. 3  Male, 47-year-old, patient with panNEN G3 grade. Neuroendocrine neoplasm locate in the body of the pancreas, the lesion measures about 57 mm × 56 mm, on non-contrast CT scan (3A, arrow), the lesion appears as an iso-density mass; during the arterial phase of contrast-enhanced CT (3B, 3D, arrow), it shows significant heterogenous enhancement, along with evidence of hepatic metastasis; in the portal venous phase (3C, arrow), the lesion demonstrates progressive and uniform enhancement with an increased enhancement degree; on T1WI-water imaging (3E), there is unevenly slightly low signal intensity; and on fat saturation T2WI (3F, arrow), an uneven slightly high signal intensity is seen.
Fig. 4  Male, 54-year-old, patient with panNEN G3 grade. An irregular-shaped mass in the neck of the pancreas, measuring around 50 mm × 41 mm, enhanced CT scans (4A to 4C, red arrow) show invasion of the pancreaticoduodenal artery, rough vessel wall, narrow and occlude lumen in this case. DWI: diffusion-weighted imaging; ADC: apparent diffusion coefficient.
表5  胰腺神经内分泌肿瘤低级别组与中高级别组间增强CT动脉期和静脉期强化类型的比较
Tab. 5  Comparison of enhancement patterns during the arterial phase and the venous phase on enhanced CT between low-grade and high-grade groups of pancreatic neuroendocrine neoplasm

2.5 panNEN多因素二元logistic回归分析及ROC曲线预测模型的效能

       针对panNEN,我们纳入了差异具有统计学意义的变量进行分析,包括病灶体积、病灶形态、肝转移和血管侵犯程度的判断情况,以及DWI和ADC图像信号特征。我们将panNEN分为低级别和中高级别两组作为因变量,并进行了多因素二元logistic回归分析(前进式条件法)。结果表明,病灶体积、肝转移和血管侵犯程度是panNEN中高级别的重要独立预测因子。基于这三项指标构建的联合预测模型在panNEN病理分级预测中的效能分析结果显示,联合模型在预测panNEN中高级别组的AUC值达到0.861(95% CI:0.798~0.923),敏感度和特异度为78.1%和83.3%(表6图5)。表明该联合预测模型在区分panNEN中高级别风险状态方面具有优越性能。

图5  病灶体积、肝转移和脉管侵犯及其联合预测panNEN病理分级的ROC曲线。panNEN:胰腺神经内分泌肿瘤;ROC:受试者工作特征;AUC:曲线下面积。
Fig. 5  ROC curves for lesion volume, hepatic metastasis, vascular invasion, and their combined predictive ability for the pathological grade of panNENs. ROC: receiver operating characteristic; panNEN: pancreatic neuroendocrine neoplasm; AUC: area under the curve.
表6  不同指标及其联合指标预测胰腺神经内分泌肿瘤中高级别组的效能
Tab. 6  Efficiency of different indicators and their combined metrics in predicting high-grade pancreatic neuroendocrine neoplasm

3 讨论

       本研究通过比较不同病理分级的panNEN的CT和MRI的影像特征,以及定量参数,包括术后患者的临床资料,运用二元logistic回归分析法构建一个综合模型,以探讨这些联合指标在术前对panNEN病理分级的预测价值。研究结果表明,该综合模型在评估panNEN中高级别组的风险状态方面展现出显著的优越性。这一研究填补了国内在联合影像检查结果预测panNEN病理分级方面的研究空白。此外,术前对panNEN病理分级的准确评估和预测,对于指导临床治疗决策及预后判断具有重要的临床意义。

3.1 panNEN临床和影像特征与病理分级的研究

       本研究发现panNEN病理分级与患者的性别结构、年龄分布相比差异没有统计学意义,与韩滨泽等[26]、刘黎明等[27]、PROCACCI等[28]前期研究一致。同时,病灶囊实性及发病部位差异也无法显著预测panNEN病理分级。然而,结果显示panNEN病灶多为实性且集中于胰头、体尾部。高级别的panNEN更多见囊性坏死等特征[28],这一结果与我们的观察并不完全一致。本研究发现,肿瘤体积、形态、肝脏转移和血管侵犯是panNEN病理分级的关键预测因素,与以往研究[29, 30, 31]报道G1组和G2/G3组在最大直径和临床分期上存在显著差异,肿瘤体积与恶性程度之间存在密切联系这一结果一致。高级别panNEN更容易表现出病灶体积大、囊性坏死、轮廓不规则、边界不清、胰周组织或血管侵犯、淋巴结肿大以及远处转移等特点。这是因为panNEN的病理分级标准基于有丝分裂计数和Ki-67指数,这两者皆反映出肿瘤细胞的增殖活性和侵袭性[32]。肿瘤分级越高,通常表示细胞增殖活跃、侵袭性强。

3.2 panNEN的CT和MRI影像特征对病理分级预测的价值

3.2.1 CT定量参数特征

       研究显示,对于panNEN患者,在CT平扫时,多数病变区域的密度与正常胰腺组织相近,进而使得小病灶易被遗漏。然而,通过增强CT的动脉期扫描,显著提升了病灶检测的准确性。动脉期的强化程度揭示了其与肿瘤分级之间的关系,这一点与刘黎明等[27]先前的研究结果保持一致。研究进一步阐明,低级别或高度分化的panNEN相较于高级别或低分化的肿瘤,在动脉供血上表现出更强的优势,这也是为何大部分panNEN在动脉期图像上表现出高衰减的现象,即它们具有丰富的毛细血管网络。

       不过,我们所进行的研究并未显示出CT量化参数与panNEN病理分级之间存在显著的相关性。相对地,LUO等的研究指出,在门静脉期,低级别panNEN的静脉回流较轻微,并且肿瘤的病理级别与肿瘤内部的微血管密度成反比[33],即级别越低,实质部分的血供越充足,增强后表现出的强化程度也就越明显,尤其是在动脉期的表现强于门静脉期。

       此外,多期增强CT扫描在panNEN的诊断和分期中发挥着重要作用,其诊断敏感度和特异度平均可达82%和96%[34]。除了诊断之外,多期增强CT扫描在预测肿瘤病理分级和疗效评估方面也发挥着关键作用。研究表明,级别越高的panNEN更常表现为肿瘤体积增大、囊性坏死、形态不规则、边界模糊以及伴随主胰管扩张等特征[35]

       尽管我们的研究未明确证实CT各时期参数与病理分级之间的显著联系,但呈现出这一潜在的趋势。有趣的是,本研究结果表明,在区分panNEN的不同病理分级上,CT增强扫描的动脉期与门静脉期定量参数差异并无统计学意义,这与HORIGUCHI等[19]、LIANG等[29]及LUO等[30]的研究结论相悖。我们推测,这种差异可能源于测量精度上的误差。尽管如此,近年来影像组学作为一门有效评估肿瘤异质性的定量工具,在panNEN的鉴别诊断、生物学行为分析、病理分级以及预后预测等方面已得到广泛应用[35, 36, 37],因此在未来的研究中,我们应该充分利用这项技术以获取更为精确的评估结果。

3.2.2 MRI参数特征

       panNEN低级别与高级别组在T1WI和T2WI信号强度差异无统计学意义,但本研究结果显示,不论病理级别如何,panNEN病灶以T1WI略低、T2WI略高偏多,信号类似正常胰腺实质,提示小型病灶在MRI平扫中也易漏诊,这一发现与前述CT平扫研究结果基本吻合,并得到了刘黎明等[27]的研究支持。相反,DWI和ADC图的信号强度在不同病理级别间差异有统计学意义。DWI图像显示高信号强度,ADC图示低信号强度占比较高。已知DWI有助于提高panNEN检出率,甚至优于PET/CT[34],且ADC与Ki-67标记指数相关,可预测内分泌肿瘤生长,认为panNEN的ADC值的变化可能受肿瘤细胞质、细胞质比例和细胞外纤维化联合作用的影响[34]。在本研究中,DWI和ADC值在panNEN不同病理分级间差异具有统计学意义,表明二者可用于病理分级预测。

3.3 panNEN不同病理分级预测模型及分析

       本研究发现,经多变量分析和二元有序logistic回归模型评估,肿瘤体积、肝转移及血管侵犯程度这三个因素与panNEN的病理分级存在显著相关性。在高级别panNEN中,这些现象更为突出,可能归因于肿瘤细胞的高侵袭性。虽然病灶形态、DWI和ADC图像在非参数检验(Wilcoxon秩和检验)中对病理分级显示出统计学意义,但在ROC曲线评价预测模型时却未纳入这些因素,推测可能源于这些特性与肿瘤体积、肝脏转移和血管侵犯之间的交互效应。据此判断,在panNEN病理分级评估过程中,肿瘤体积、肝脏转移和血管侵犯的重要性可能超过其他特征;尤其是肝转移作为独立预测因子,其OR值高达18.83。本研究将肿瘤体积、肝转移和血管侵犯整合至联合预测模型,结果显示该模型在预测panNEN中高级别组方面表现出高预测效能,AUC值达到0.861,敏感度为78.1%,特异度为83.3%,这说明多元指标联合应用能够提升模型的稳定性和准确性。

3.4 研究的局限与展望

       本研究存在如下局限性:第一,由于是一项回顾性研究,采用的CT和MR扫描设备及扫描参数不尽统一,可能对研究结果产生一定影响;第二,纳入样本量偏小,可能导致统计结果产生偏倚,需要扩大样本量以进一步确认;第三,在选用CT值等定量参数时,尤其是针对囊实性或不均匀强化的实性肿瘤部分,可能存在感兴趣区域选择的偏差,影响测量精度;第四,在MRI检查中,T1WI、T2WI、DWI等非定量参数易受图像视觉分析偏差和观察者一致性问题影响。至于ADC作为定量参数,受限于本研究的回顾性性质,很多病例未能完成ADC值测量,降低了ADC值的特异性,故未来研究中应优先采用定量值进行分析。

4 结论

       综上所述,本研究构建的基于多种指标的联合预测模型在预测panNEN病理分级方面展现出较高效能,具有一定的临床价值,有助于医师在术前对患者的风险评估,为制订个性化的手术及治疗方案以及评估术后预后提供了有力支持。

[1]
KHANNA L, PRASAD S R, SUNNAPWAR A, et al. Pancreatic neuroendocrine neoplasms: 2020 update on pathologic and imaging findings and classification[J]. Radiographics, 2020, 40(5): 1240-1262. DOI: 10.1148/rg.2020200025.
[2]
HOFLAND J, FALCONI M, CHRIST E, et al. European Neuroendocrine Tumor Society 2023 guidance paper for functioning pancreatic neuroendocrine tumour syndromes[J/OL]. J Neuroendocrinol, 2023, 35(8): e13318 [2024-12-20]. https://pubmed.ncbi.nlm.nih.gov/37578384. DOI: 10.1111/jne.13318.
[3]
MATROOD S, MELMS L E, BARTSCH D K, et al. The expression of autophagy-associated genes represents a valid footprint for aggressive pancreatic neuroendocrine neoplasms[J/OL]. Int J Mol Sci, 2023, 24(4): 3636 [2024-12-20]. https://pubmed.ncbi.nlm.nih.gov/36835048/. DOI: 10.3390/ijms24043636.
[4]
FORSYTHE S D, PU T, ANDREWS S G, et al. Models in pancreatic neuroendocrine neoplasms: current perspectives and future directions[J/OL]. Cancers, 2023, 15(15): 3756 [2024-12-20]. https://pubmed.ncbi.nlm.nih.gov/36835048/. DOI: 10.3390/cancers15153756.
[5]
NAGTEGAAL I D, ODZE R D, KLIMSTRA D, et al. The 2019 WHO classification of tumours of the digestive system[J]. Histopathology, 2020, 76(2): 182-188. DOI: 10.1111/his.13975.
[6]
FAN J H, ZHANG Y Q, SHI S S, et al. A nation-wide retrospective epidemiological study of gastroenteropancreatic neuroendocrine neoplasms in China[J]. Oncotarget, 2017, 8(42): 71699-71708. DOI: 10.18632/oncotarget.17599.
[7]
HALLET J, LAW C H L, CUKIER M, et al. Exploring the rising incidence of neuroendocrine tumors: a population-based analysis of epidemiology, metastatic presentation, and outcomes[J]. Cancer, 2015, 121(4): 589-597. DOI: 10.1002/cncr.29099.
[8]
DAS S, DASARI A. Epidemiology, incidence, and prevalence of neuroendocrine neoplasms: are there global differences?[J/OL]. Curr Oncol Rep, 2021, 23(4): 43 [2024-12-20]. https://pubmed.ncbi.nlm.nih.gov/33719003/. DOI: 10.1007/s11912-021-01029-7.
[9]
LU F Y, YE M J, HU C H, et al. FABP5 regulates lipid metabolism to facilitate pancreatic neuroendocrine neoplasms progression via FASN mediated Wnt/β-catenin pathway[J]. Cancer Sci, 2023, 114(9): 3553-3567. DOI: 10.1111/cas.15883.
[10]
PTASNUKA M, TRUSKOVS A, OZOLINS A, et al. Sporadic pancreatic neuroendocrine neoplasms: a retrospective clinicopathological and outcome analysis from a Latvian study group[J/OL]. Front Surg, 2023, 10: 1131333 [2024-12-20]. https://pubmed.ncbi.nlm.nih.gov/37021091/. DOI: 10.3389/fsurg.2023.1131333.
[11]
WIESE D, HUMBURG F G, KANN P H, et al. Changes in diagnosis and operative treatment of insulinoma over two decades[J/OL]. Langenbecks Arch Surg, 2023, 408(1): 255 [2024-12-20]. https://pubmed.ncbi.nlm.nih.gov/37386194/. DOI: 10.1007/s00423-023-02974-6.
[12]
TOSHIMA F, INOUE D, KOMORI T, et al. Is the combination of MR and CT findings useful in determining the tumor grade of pancreatic neuroendocrine tumors?[J]. Jpn J Radiol, 2017, 35(5): 242-253. DOI: 10.1007/s11604-017-0627-x.
[13]
ZIOGAS I A, SCHMITZ R, MORIS D, et al. The role of surgery for pancreatic neuroendocrine tumors[J]. Anticancer Res, 2022, 42(2): 629-639. DOI: 10.21873/anticanres.15520.
[14]
SCOTT A T, HOWE J R. Evaluation and management of neuroendocrine tumors of the pancreas[J]. Surg Clin North Am, 2019, 99(4): 793-814. DOI: 10.1016/j.suc.2019.04.014.
[15]
李炳琦, 原春辉. 胰腺神经内分泌肿瘤肝转移的外科治疗争议与展望[J]. 中华外科杂志, 2023, 61(10): 839-844. DOI: 10.3760/cma.j.cn112139-20230319-00114.
LI B Q, YUAN C H. Controversies and prospects for surgical treatment of pancreatic neuroendocrine neoplasms with liver metastases[J]. Chin J Surg, 2023, 61(10): 839-844. DOI: 10.3760/cma.j.cn112139-20230319-00114.
[16]
赵利辉, 张岱, 穆洁, 等. 基于超声影像组学对胰腺神经内分泌肿瘤肝转移的预测价值[J]. 中华超声影像学杂志, 2023, 32(8): 685-691. DOI: 10.3760/cma.j.cn131148-20230206-00060.
ZHAO L H, ZHANG D, MU J, et al. Value of predictive liver metastasis in pancreatic neuroendocrine neoplasms based on ultrasonographic radiomics[J]. Chin J Ultrason, 2023, 32(8): 685-691. DOI: 10.3760/cma.j.cn131148-20230206-00060.
[17]
SINGHI A D, KLIMSTRA D S. Well-differentiated pancreatic neuroendocrine tumours (PanNETs) and poorly differentiated pancreatic neuroendocrine carcinomas (PanNECs): concepts, issues and a practical diagnostic approach to high-grade (G3) cases[J]. Histopathology, 2018, 72(1): 168-177. DOI: 10.1111/his.13408.
[18]
ITO T, HIJIOKA S, MASUI T, et al. Advances in the diagnosis and treatment of pancreatic neuroendocrine neoplasms in Japan[J]. J Gastroenterol, 2017, 52(1): 9-18. DOI: 10.1007/s00535-016-1250-9.
[19]
HORIGUCHI S, KATO H, SHIRAHA H, et al. Dynamic computed tomography is useful for prediction of pathological grade in pancreatic neuroendocrine neoplasm[J]. J Gastroenterol Hepatol, 2017, 32(4): 925-931. DOI: 10.1111/jgh.13594.
[20]
RIMBAŞ M, RIZZATTI G, TOSONI A, et al. Small nonfunctional pancreatic neuroendocrine neoplasms: Time for a step-up treatment approach?[J]. Endosc Ultrasound, 2023, 12(1): 1-7. DOI: 10.4103/EUS-D-22-00028.
[21]
TAKUMI K, FUKUKURA Y, HIGASHI M, et al. Pancreatic neuroendocrine tumors: Correlation between the contrast-enhanced computed tomography features and the pathological tumor grade[J]. Eur J Radiol, 2015, 84(8): 1436-1443. DOI: 10.1016/j.ejrad.2015.05.005.
[22]
MA J, WANG X Y, TANG M S, et al. Preoperative prediction of pancreatic neuroendocrine tumor grade based on 68Ga-DOTATATE PET/CT[J]. Endocrine, 2024, 83(2): 502-510. DOI: 10.1007/s12020-023-03515-3.
[23]
TAKADA S, KATO H, SARAGAI Y, et al. Contrast-enhanced harmonic endoscopic ultrasound using time-intensity curve analysis predicts pathological grade of pancreatic neuroendocrine neoplasm[J]. J Med Ultrason, 2019, 46(4): 449-458. DOI: 10.1007/s10396-019-00967-x.
[24]
WANG Y, CHEN Z E, YAGHMAI V, et al. Diffusion-weighted MR imaging in pancreatic endocrine tumors correlated with histopathologic characteristics[J]. J Magn Reson Imaging, 2011, 33(5): 1071-1079. DOI: 10.1002/jmri.22541.
[25]
BIAN Y, LI J, CAO K, et al. Magnetic resonance imaging radiomic analysis can preoperatively predict G1 and G2/3 grades in patients with NF-pNETs[J]. Abdom Radiol, 2021, 46(2): 667-680. DOI: 10.1007/s00261-020-02706-0.
[26]
韩滨泽, 纪元, 曾蒙苏, 等. 胰腺神经内分泌肿瘤计算机断层扫描增强特征对术前病理分级的预测价值[J]. 中华消化杂志, 2021, 41(9): 613-618. DOI: 10.3760/cma.j.cn311367-20210320-00162.
HAN B Z, JI Y, ZENG M S, et al. Predictive value of enhanced computed tomography in preoperative pathological grading of pancreatic neuroendocrine neoplasm[J]. Chin J Dig, 2021, 41(9): 613-618. DOI: 10.3760/cma.j.cn311367-20210320-00162.
[27]
刘黎明, 唐艳华, 王海屹, 等. 多层CT对胰腺神经内分泌肿瘤病理分级的可行性[J]. 中华放射学杂志, 2016, 50(2): 105-109. DOI: 10.3760/cma.j.issn.1005-1201.2016.02.006.
LIU L M, TANG Y H, WANG H Y, et al. CT imaging features of different histological grades of pancreatic neuroendocrine tumors[J]. Chin J Radiol, 2016, 50(2): 105-109. DOI: 10.3760/cma.j.issn.1005-1201.2016.02.006.
[28]
PROCACCI C, CARBOGNIN G, ACCORDINI S, et al. Nonfunctioning endocrine tumors of the pancreas: possibilities of spiral CT characterization[J]. Eur Radiol, 2001, 11(7): 1175-1183. DOI: 10.1007/s003300000714.
[29]
LIANG W J, YANG P F, HUANG R, et al. A combined nomogram model to preoperatively predict histologic grade in pancreatic neuroendocrine tumors[J]. Clin Cancer Res, 2019, 25(2): 584-594. DOI: 10.1158/1078-0432.CCR-18-1305.
[30]
LUO Y J, CHEN X, CHEN J, et al. Preoperative prediction of pancreatic neuroendocrine neoplasms grading based on enhanced computed tomography imaging: validation of deep learning with a convolutional neural network[J]. Neuroendocrinology, 2020, 110(5): 338-350. DOI: 10.1159/000503291.
[31]
UTSUMI M, UMEDA Y, TAKAGI K, et al. Correlation of computed tomography imaging features and pathological features of 41 patients with pancreatic neuroendocrine tumors[J]. Hepatogastroenterology, 2015, 62(138): 441-446.
[32]
BUETOW P C, PARRINO T V, BUCK J L, et al. Islet cell tumors of the pancreas: pathologic-imaging correlation among size, necrosis and cysts, calcification, malignant behavior, and functional status[J]. AJR Am J Roentgenol, 1995, 165(5): 1175-1179. DOI: 10.2214/ajr.165.5.7572498.
[33]
BETTINI R, PARTELLI S, BONINSEGNA L, et al. Tumor size correlates with malignancy in nonfunctioning pancreatic endocrine tumor[J]. Surgery, 2011, 150(1): 75-82. DOI: 10.1016/j.surg.2011.02.022.
[34]
SUNDIN A, ARNOLD R, BAUDIN E, et al. ENETS consensus guidelines for the standards of care in neuroendocrine tumors: radiological, nuclear medicine & hybrid imaging[J]. Neuroendocrinology, 2017, 105(3): 212-244. DOI: 10.1159/000471879.
[35]
LUO Y J, DONG Z, CHEN J, et al. Pancreatic neuroendocrine tumours: correlation between MSCT features and pathological classification[J]. Eur Radiol, 2014, 24(11): 2945-2952. DOI: 10.1007/s00330-014-3317-4.
[36]
GU D S, HU Y B, DING H, et al. CT radiomics may predict the grade of pancreatic neuroendocrine tumors: a multicenter study[J]. Eur Radiol, 2019, 29(12): 6880-6890. DOI: 10.1007/s00330-019-06176-x.
[37]
BIAN Y, JIANG H, MA C, et al. CT-based radiomics score for distinguishing between grade 1 and grade 2 nonfunctioning pancreatic neuroendocrine tumors[J]. AJR Am J Roentgenol, 2020, 215(4): 852-863. DOI: 10.2214/AJR.19.22123.

上一篇 构建并外部验证XGBoost模型鉴别乳腺非肿块病变良恶性
下一篇 增强T2*加权血管成像在T2WI低信号肾脏病变良恶性鉴别诊断中的价值
  
诚聘英才 | 广告合作 | 免责声明 | 版权声明
联系电话:010-67113815
京ICP备19028836号-2