分享:
分享到微信朋友圈
X
临床研究
基于O-RADS MRI评分的临床与多参数MRI预测模型在子宫附件良恶性肿块鉴别中的应用价值
张晓琴 林斯宏 林泽林 林少帆 林黛英

Cite this article as: ZHANG X Q, LIN S H, LIN Z L, et al. The value of a clinical-multiparametric MRI prediction model based on O-RADS MRI scoring system in differentiating between benign and malignant adnexal masses of the uterus[J]. Chin J Magn Reson Imaging, 2025, 16(7): 39-46.本文引用格式:张晓琴, 林斯宏, 林泽林, 等. 基于O-RADS MRI评分的临床与多参数MRI预测模型在子宫附件良恶性肿块鉴别中的应用价值[J]. 磁共振成像, 2025, 16(7): 39-46. DOI:10.12015/issn.1674-8034.2025.07.006.


[摘要] 目的 构建基于附件影像报告和数据系统(ovarian-adnexal reporting and data system, O-RADS)MRI评分的临床与多参数MRI预测模型,探索其在附件良恶性病变鉴别诊断中的实用价值。材料与方法 回顾性分析2020年至2023年165例行盆腔MRI平扫+增强检查并具有病理诊断结果的附件肿物患者的影像及临床资料,收集患者术前临床指标和影像学特征,单因素分析比较良恶性肿块组间各指标的差异,多因素逻辑回归筛选出预测附件恶性肿块的独立风险因子,并构建logistic回归预测模型,以列线图展示。采用受试者工作特征(receiver operating characteristic, ROC)曲线、DeLong检验、综合判别改善指数(integrated discrimination improvement, IDI)、净重新分类指数(net reclassification index, NRI)比较基于O-RADS MRI评分的logistic回归模型与单纯O-RADS MRI评分在鉴别诊断性能上的差异,绘制校准曲线以评估logistic回归模型的校准能力,并通过决策曲线分析(decision curve analysis, DCA)评价二者的临床净收益。结果 共收集患者165例共170个肿块,其中良性肿块84个,恶性肿块86个,年龄范围11~87岁,良性组年龄为49.50(28.75,60.75)岁,恶性组年龄为50.50(38.75,62.00)岁。单因素分析结果显示良、恶性肿块组间在糖类抗原125(carbohydrate antigen 125, CA125)、人附睾蛋白4(human epididymis protein4, HE4)、血小板计数(platelet count, PLT)、病灶边界清晰度、O-RADS MRI评分、实性组织表观扩散系数平均值(mean apparent diffusion coefficient, ADCmean)和囊液ADC值的差异均具有统计学意义(P值均<0.05)。通过多因素logistic回归分析得出HE4水平升高[比值比(odds ratio, OR)=1.011,P=0.028]、O-RADS MRI评分增高(OR=3.085,P=0.001)、ADCmean值降低(OR=0.005,P<0.001)是附件肿块恶性病变的独立预测因子。联合O-RADS MRI评分、ADCmean、HE4建立logistic回归模型,用于区分附件良恶性肿块的曲线下面积(area under the curve, AUC)为0.944,敏感度为84.9%,特异度为90.5%,优于单纯O-RADS MRI评分(AUC为0.849,敏感度为89.5%,特异度为81.0%);DeLong检验显示两者的AUC差异有统计学意义(P<0.001);NRI、IDI显示logistic回归模型较O-RADS MRI评分对附件肿块的鉴别诊断性能更好(P<0.05);校准曲线显示logistic回归模型的校准度良好;DCA表明logistic回归模型的临床净收益大于O-RADS MRI评分。结论 联合O-RADS MRI评分、ADCmean和HE4构建的logistic回归模型在附件良恶性肿块的鉴别诊断中具有较高的效能,区分效能优于单纯的O-RADS MRI评分,可用于术前有效区分附件良恶性肿块。
[Abstract] Objective To develop and validate a clinical-multiparametric MRI predictive model incorporating the ovarian-adnexal reporting and data system (O-RADS) MRI score, and to evaluate its utility in distinguishing benign from malignant adnexal lesions.Materials and Methods A retrospective study was performed to analyze 165 cases of adnexal masses that underwent pelvic MRI plain scan + enhanced examination and were confirmed by pathological histology from 2020 to 2023. The preoperative clinical indicators and imaging characteristics of the patients were collected. The differences in various indicators between benign and malignant mass groups were compared by univariate analysis. Multivariate logistic regression was used to screen out independent risk factors for predicting adnexal malignant masses, and a logistic regression prediction model was constructed and displayed in a nomogram. Receiver operating characteristic (ROC) curve, DeLong test, integrated discrimination improvement index (IDI) and net reclassification index (NRI) were used to evaluate and compare the difference in differential diagnostic performance between the logistic regression model based on O-RADS MRI score and the simple O-RADS MRI score. Calibration curves were drawn to evaluate the calibration ability of the logistic regression model. Decision curve analysis (DCA) was used to evaluate the clinical net benefits of the two models.Results After screening, a total of 165 patients (with 170 masses) were collected. The age ranged from 11 to 87 years old. The median age of the benign group was 49.50 (28.75, 60.75) years old, and the median age of the malignant group was 50.50 (38.75, 62.00) years old. The results of univariate analysis showed that there were significant differences in carbohydrate antigen 125 (CA125), human epididymis protein 4 (HE4), platelet count (PLT), lesion boundary clarity, O-RADS MRI scores, mean apparent diffusion coefficient (ADCmean) of solid components, and ADC values of cystic fluid between benign and malignant masses (all P < 0.05) . Multifactorial logistic regression analysis showed that increased HE4 level (OR = 1.011, P = 0.028), increased O-RADS MRI score (OR = 3.085, P = 0.001), and decreased ADCmean value (OR = 0.005, P < 0.001) were independent predictors of malignant lesions of adnexal masses. The logistic regression model was established by combining O-RADS MRI score, ADCmean, and HE4. The area under the curve (AUC) for distinguishing benign and malignant adnexal masses was 0.944, the sensitivity was 84.9%, and the specificity was 90.5%, which were better than the simple O-RADS MRI score (AUC was 0.849, sensitivity was 89.5%, and specificity was 81.0%). DeLong test showed that the difference in AUC between the two models was statistically significant (P < 0.001). NRI and IDI showed that the logistic regression model had better differential diagnostic performance for adnexal masses than the O-RADS MRI score, and the difference were statistically significant (P < 0.05). The calibration curves showed that the calibration of the logistic regression model was good; DCA showed that the clinical net yield of the logistic regression model was greater than that of the O-RADS MRI score.Conclusions The logistic regression model constructed by combining O-RADS MRI score, ADCmean and HE4 has high efficacy in the differential diagnosis of benign and malignant adnexal masses, and its differentiation efficiency is better than that of the simple O-RADS MRI score, and can be used to effectively distinguish benign and malignant adnexal masses before surgery.
[关键词] 卵巢-附件影像报告和数据系统;附件肿块;鉴别诊断;预测模型;列线图;磁共振成像
[Keywords] ovarian-adnexal reporting and data system;adnexal masses;differential diagnosis;prediction model;nomogram;magnetic resonance imaging

张晓琴    林斯宏    林泽林    林少帆    林黛英 *  

汕头市中心医院磁共振室,汕头 515041

通信作者:林黛英,E-mail: lindaiying917@163.com

作者贡献声明:林黛英设计本研究方案,对稿件的重要内容进行了修改;张晓琴获取、分析、解释本研究的数据,起草和撰写稿件;林斯宏、林泽林、林少帆协助获取、分析、解释本研究的数据,对稿件的重要内容进行了修改;林斯宏获得了汕头市医疗卫生科技计划项目基金资助;全体作者一致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。


基金项目: 汕头市医疗卫生科技计划项目 211117116490993
收稿日期:2025-03-01
接受日期:2025-06-05
中图分类号:R445.2  R737.3 
文献标识码:A
DOI: 10.12015/issn.1674-8034.2025.07.006
本文引用格式:张晓琴, 林斯宏, 林泽林, 等. 基于O-RADS MRI评分的临床与多参数MRI预测模型在子宫附件良恶性肿块鉴别中的应用价值[J]. 磁共振成像, 2025, 16(7): 39-46. DOI:10.12015/issn.1674-8034.2025.07.006.

0 引言

       卵巢癌是女性生殖系统第二大常见癌症,也是妇科癌症导致死亡的主要原因,其早期表现隐匿,晚期患者的五年生存率低于50%,但早期患者若接受标准治疗,五年生存率可达90%以上[1, 2, 3]。因此,早期诊断和准确分期至关重要。超声成像作为一种无创、便捷的检查手段,通常作为附件肿块的首选筛选方法,但仍有多达三分之一的附件肿块难以定性,需进一步影像评估[4]。多参数磁共振成像能对附件病变准确定性,还能优化所需手术干预的时机和范围,避免良性/交界性肿瘤不必要或过度手术干预,从而保护年轻女性的生育能力并提高其生存率[5]。卵巢附件报告和数据系统(ovarian-adnexal reporting and data system, O-RADS)MRI评分是一种风险分层系统,2022年由美国放射学会(American College of Radiology, ACR)正式发布[6],用于评估超声不确定的附件肿块,其诊断敏感性和特异性已得到验证[7]。然而,在实际应用中,单纯依赖O-RADS MRI评分仍存在一定局限。RIZZO等[8]的一项荟萃分析发现,O-RADS MRI 3~5分的恶性概率跨度较大(3分2%~13%,4分25%~84%,5分87%~100%),易导致临床决策困难。因此,自O-RADS MRI评分提出以来,如何优化O-RADS MRI评分以提高其诊断能力一直是学术界关注的重要问题。目前相关研究都侧重于优化O-RADS MRI 4分肿块风险分层的准确性[9, 10],或探索联合O-RADS MRI评分与单一影像参数的应用价值[11, 12];同时,多数研究在纳入新参数时,缺乏对各诊断参数在诊断策略中权重的统计学量化分析,致使新参数的权重分配在实际应用中更多依赖于主观经验,而非基于客观数据的支持,使其在临床应用中仍存在一定的局限性。

       本研究旨在联合O-RADS MRI评分、其他MR征象及临床因素,构建logistic回归模型,用于区分附件良恶性肿块,以期进一步优化O-RADS MRI评分2~5分附件肿块良恶性鉴别诊断的准确度。

1 材料与方法

1.1 研究对象

       本研究为回顾性研究,经汕头市中心医院伦理委员会批准,免除受试者知情同意,批准文号:(2025)科研015号。收集2020年3月至2023年11月期间在本院接受手术治疗并具有病理诊断结果的附件肿物患者的影像及临床资料。入组标准:(1)患者术前未曾接受过任何附件区或卵巢的手术及放化疗;(2)患者术前均接受了完整的磁共振检查,所获影像信息可用于O-RADS MRI评分系统。排除标准:(1)有附件恶性肿瘤病史;(2)O-RADS MRI 0分和1分的病例;(3)临床、病理资料不完整。最终本研究纳入165例患者共170个病灶,见图1

图1  研究对象入组流程图。O-RADS:卵巢-附件影像报告和数据系统。
Fig. 1  Process flowchart for enrollment of study population. O-RADS: ovarian-adnexal reporting and data system.

1.2 资料采集

       通过电子病历系统获取临床资料,包括年龄、绝经状态、血清学指标人附睾蛋白4(human epididymis protein4, HE4)、血小板计数(platelet count, PLT)、糖类抗原125(carbohydrate antigen 125, CA125);影像学指标包括:病灶最大直径及边界、表观扩散系数(apparent diffusion coefficient, ADC)值、O-RADS MRI评分。

1.3 MRI检查方法

       MRI扫描使用德国西门子3.0 T磁共振设备(Magnetom Verio; Siemens Medical Solutions, Erlangen, Germany),采用8通道盆腔相控阵线圈接收信号。扫描体位为仰卧位,头先进,中心定位线位于耻骨联合处上方2 cm。

       扫描序列及参数:横轴位T1WI,TR 550 ms,TE 13 ms,层厚5 mm,FOV 380 mm×380 mm;高分辨率横轴位T2WI,TR 3800 ms,TE 85 ms,层厚3.0 mm,FOV 200 mm×200 mm;矢状位T2WI,TR 5000 ms,TE 85 ms,层厚3.5 mm,FOV 260 mm×260 mm;扩散加权成像(diffusion weighted imaging, DWI),运用单次激发快速自旋回波序列,TR 4500 ms,TE 93 ms,层厚5 mm,FOV 260 mm×260 mm,b值设定为50、400、800 s/mm2;横轴位动态对比增强(dynamic contrast-enhanced, DCE),采用T1WI脂肪抑制容积内插屏气检查(fat suppression-volumetric interpolated breath-hold examination, FS-VIBE),TR 3.92 ms,TE 1.39 ms,层厚3 mm,FOV 380 mm×380 mm,连续采集6期动态增强图像,时间分辨率为12.6 s,先获取1期T1WI-FS序列蒙片图像,从第2个时相开始,按照0.2 mmol/kg、2 mL/s静脉团注对比剂钆喷酸葡胺(Gd-DPTA;马根维显),以相同流率加20 mL生理盐水冲管。

1.4 图像分析

       由2名具有盆腔MRI诊断经验的放射科医生(分别为具有15年影像诊断经验的副主任医师和具有5年影像诊断经验的主治医师)在未获知病理结果的条件下,进行了图像评估与数据测量。O-RADS MRI评分依据ACR发布的标准[6],当两位医师的评分出现不一致时,由该两位医师通过回顾MRI影像并进行讨论,以确定最终的O-RADS MRI评分。利用工作站软件进行DWI及动态增强后处理,生成ADC图像及时间-信号强度曲线(time-signal intensity curve, TIC),在病变明显强化处勾画感兴趣区(region of interest, ROI)测量TIC。在ADC图上测量病灶的ADC值,在实性成分及囊液区域连续3个层面分别勾画ROI并测量ADC值,获取原始数据,除去最大值及最小值,计算相应平均值得到最终的实性组织ADCmean和囊液ADC值。勾画病变的实性部分时ROI应尽可能避开囊变、脂肪和出血等区域,为准确定位病灶并避免部分容积效应,使用T2WI、DWI及DCE作为参考图像。

1.5 观察标准

       根据术后获得的病理诊断结果将肿块分为良性及恶性两组,若术后病理报告显示为交界性或低度恶性肿瘤,将该类病变纳入恶性肿瘤的统计分析中[7, 13]

1.6 统计学分析

       数据整理与分析通过SPSS 26.0统计软件完成。计数数据以频数(百分比)的形式表述,组间对比则使用卡方检验进行分析。计量资料用Kolmogorov-Smirnov检验进行正态性检验,符合正态分布时采用(x¯±s)表示,采用两独立样本t检验比较组间差异;不符合正态分布的计量数据,用中位数(P25,P75)形式来呈现,采用Mann-Whitney U检验比较组间差异。运用加权Kappa检验评估观察者间一致性。构建二元逻辑回归预测模型,并将其结果以列线图形式展现。采用受试者工作特征(receiver operating characteristic, ROC)曲线评估预测模型与O-RADS MRI评分的区分效能,并通过DeLong检验、净重新分类指数(net reclassification index, NRI)和综合判别改善指数(integrated discrimination improvement, IDI)对预测模型与O-RADS MRI评分的鉴别诊断效能进行比较分析。利用Hosmer-Lemeshow检验和Bootstrap重复抽样(1000次)进行内部验证,同时绘制校准曲线以评估模型的拟合度。在R4.2.0软件中,借助rmda包绘制决策曲线分析(decision curve analysis, DCA)曲线,测试模型的临床应用价值。在DCA中,临床净收益是指权衡真阳性带来的益处和假阳性带来的损失后的净效果,通过计算不同阈值概率下的净效益,以评估模型的临床效用。当双侧检验P<0.05时,认为差异具有统计学意义。

2 结果

2.1 附件肿块病理类型及O-RADS MRI评分

       170个肿块中良性肿块84个,恶性肿块86个。O-RADS MRI分类评分:5分肿块46个,其中恶性肿块35个,良性肿块11个;4分肿块47例,其中恶性肿块41个,良性肿块6个;3分肿块61个,其中恶性10个,良性肿块51个;2分肿块16个,均为良性,见表1图2~3。2名诊断医师对病灶的O-RADS MRI评分结果一致性高,Kappa值为0.818,P<0.001。

图2  女,39岁,O-RADS MRI评分3分,HE4水平:50.64 pmol/L,术后病理:右侧卵巢卵泡膜细胞瘤。2A:T2WI显示肿块实性成分呈高信号;2B:DWI显示实性成分弥散呈高信号;2C:肿块实性成分ADC值为1.354×10-3 mm2/s;2D:增强扫描显示实性成分强化程度低于子宫肌层;2E:低风险曲线。
图3  女,38岁,O-RADS MRI评分5分,HE4水平:44.12 pmol/L,术后病理:右侧卵巢甲状腺肿。3A:T2WI显示肿块实性成分呈低信号;3B:DWI显示实性成分呈低信号;3C:肿块实性成分ADC值为1.404×10-3 mm2/s;3D:增强扫描示实性成分强化高于子宫肌层;3E:高风险曲线。O-RADS:卵巢-附件影像报告和数据系统;HE4:人附睾蛋白4;DWI:扩散加权成像;ADC:表观扩散系数。
Fig. 2  Female, 39-year-old, with O-RADS MRI 3, HE4 level: 50.64 pmol/L, postoperative pathology: thecoma of the right ovarian. 2A: T2WI shows high signal intensity in the solid component of the mass; 2B: DWI shows high signal intensity in the diffusion of the solid component; 2C: The ADC value of the solid tissue is 1.354 × 10-3 mm2/s; 2D: On contrast-enhanced images, the solid component shows enhancement less than that of the myometrium; 2E: Low-risk TIC curve.
Fig. 3  Female, 38-year-old, with O-RADS MRI 5, HE4 level: 44.12 pmol/L, postoperative pathology: struma ovarian of right ovarian. 3A: T2WI shows low signal intensity in the solid component of the mass; 3B: DWI shows low signal intensity in the diffusion of the solid component; 3C: The ADC value of the solid component is 1.404 × 10-3 mm2/s; 3D: On contrast-enhanced images, the solid component shows enhancement more than that of the myometrium; 3E: High-risk TIC curve. O-RADS: ovarian-adnexal reporting and data system; HE4: human epididymis protein 4; DWI: diffusion-weighted imaging; ADC: apparent diffusion coefficient; TIC: time-intensity curve.
表1  O-RADS MRI评分与病理的对应关系
Tab. 1  Correlation between O-RADS MRI scores and pathological diagnosis

2.2 筛选良恶性肿块的相关因素及logistic回归模型的建立

       单因素分析显示,两组患者年龄、绝经状态、病灶最大直径比较差异均无统计学意义(P>0.05)。两组间CA125、HE4、PLT、边界“不清晰”、O-RADS MRI评分、囊液ADC值、ADCmean差异均有统计学意义(P<0.05),见表2。logistic回归分析显示HE4、O-RADS MRI评分、ADCmean是预测病理恶性的独立风险因素,见表3。根据二元逻辑回归分析结果,联合上述独立风险因素得出模型的回归方程为logit(P)=1.498+0.011×HE4+1.127×O-RADS MRI评分-5.372×ADCmean,模型以列线图的形式呈现(图4)。

图4  logistic回归模型列线图。HE4:人附睾蛋白4;O-RADS:卵巢-附件影像报告和数据系统;ADC:表观扩散系数。
Fig. 4  Nomogram of the logistic regression model. HE4: human epididymis protein4; O-RADS: ovarian-adnexal reporting and data system; ADC: apparent diffusion coefficient.
表2  影响附件肿块良恶性诊断的单因素分析
Tab. 2  Univariate analysis of factors affecting the diagnosis of benign and malignant adnexal masses
表3  附件良恶性肿块鉴别诊断的多因素分析
Tab. 3  Multifactorial analysis for differentiating benign and malignant adnexal masses

2.3 logistic回归模型与O-RADS MRI评分的诊断效能比较

       logistic回归模型用于区分附件良恶性肿块的曲线下面积(area under the curve, AUC)为0.944(95% CI:0.913~0.976)(P<0.001),特异度为90.5%,敏感度为84.9%,约登指数为0.754。O-RADS MRI评分的AUC为0.849(95% CI:0.786~0.912)(P<0.001),特异度为81.0%,敏感度为89.5%,约登指数为0.705,最佳截断值为3.5。DeLong检验结果表明,在预测附件肿块良恶性风险方面,logistic回归模型AUC优于单独使用O-RADS MRI评分,差异有统计学意义(Z=3.840,P<0.001)。NRI为0.408,差异有统计学意义(Z=5.403,P<0.001);IDI为0.218,差异有统计学意义(Z=3.087,P=0.002)。详见表4图5

图5  logistic回归模型与O-RADS MRI评分的ROC曲线。O-RADS:卵巢-附件影像报告和数据系统;ROC:受试者工作特征;AUC:曲线下面积。
图6  logistic回归模型校准曲线。
图7  logistic回归模型与O-RADS MRI评分的决策曲线分析图。O-RADS:卵巢-附件影像报告和数据系统。
Fig. 5  ROC curves of the logistic regression model and O-RADS MRI score. ROC: receiver operating characteristic; AUC: area under the curve; O-RADS: ovarian-adnexal reporting and data system.
Fig. 6  Calibration curve of the logistic regression model.
Fig. 7  Decision curve analysis curves of the logistic regression model compared with O-RADS MRI score. O-RADS: ovarian-adnexal reporting and data system.
表4  构建模型并测试模型的区分效能
Tab. 4  Model development and evaluation of discriminative performance

2.4 logistic回归模型的校准度

       经Hosmer-Lemeshow检验发现,实际病理恶性发生概率和预测概率比较,差异无统计学意义(χ2=5.806,df=8,P=0.669)。经过Bootstrap法重复抽样1000次进行模型的内部验证和校准图显示logistic回归模型在预测附件病灶恶性概率方面,与临床实际观察结果呈现出较高的一致性(图6)。

2.5 logistic回归模型与O-RADS MRI评分的临床实用性及其比较

       DCA表明在较大的阈值概率范围内,采用logistic回归模型评估附件良恶性肿块的临床净收益高于O-RADS MRI评分,展现出较优的临床应用价值(图7)。

3 讨论

       本研究最终联合O-RADS MRI评分、ADCmean、HE4构建了logistic回归模型,用于区分附件良恶性肿块,并通过列线图可视化。相比于单纯O-RADS MRI评分,预测模型有效提高了对附件良恶性肿块的诊断效能及临床净收益,并表现出良好的校准特性。

3.1 临床因素对附件肿块的诊断效能

       本研究发现恶性组患者血清HE4水平高于良性组(P<0.001),与文献研究结果一致[14, 15]。既往针对O-RADS MRI评分在附件良恶性肿块鉴别诊断中的不足,虽已有研究建议联合临床因素进行优化,但改良方案多聚焦在CA125指标[10, 16]。与传统的CA125相比,HE4作为卵巢癌的新型肿瘤标志物,不仅与肿瘤细胞增殖、浸润及转移密切相关,在卵巢癌的早期诊断中也显示出更高的特异性和敏感性[15],并且HE4在卵巢癌患者中的表达水平显著高于良性卵巢疾病患者和健康对照组[14]。此外,我们的研究还观察到恶性组患者的CA125、PLT水平较良性组高,与文献报道基本一致[17, 18]。而与既往研究结果不同[19],这次研究年龄在良恶性组间未显示统计学差异(P>0.05),分析可能原因是将交界性肿瘤纳入恶性肿瘤类别进行分析,而交界性肿瘤通常比浸润性肿瘤至少早10年确诊,约有三分之一在40岁之前确诊[20, 21]。我们通过多因素逻辑回归分析证实临床因素中只有HE4是预测附件恶性肿瘤的独立风险因素(OR=1.011,P=0.028),这一结果提示HE4作为临床高危因素,在附件恶性肿瘤的早期诊断、风险评估中具有重要的临床价值,值得进一步深入研究。然而,HE4水平可能受年龄、绝经状态等因素影响(绝经后水平显著升高)[22],因此,单独依赖HE4水平进行评估可能存在不足。本研究创新性地将HE4与O-RADS MRI评分相结合,旨在通过多参数联合分析提高诊断效能。

3.2 影像征象对附件肿块的诊断效能

       在本研究中,ADCmean与附件良恶性病变存在显著相关性,可作为独立预测因素,肿块ADCmean越低,恶性可能越高,这与肿瘤生物学特征密切相关。随着恶性程度增高,肿瘤细胞增殖活跃、数量增加、排列紧密,导致细胞外间隙变窄,从而使水分子扩散受限,在DWI上表现为高信号而ADC值降低[23]。KIM等[24]纳入了21项研究进行荟萃分析,旨在探讨ADC值在区分恶性与良性附件肿块中的诊断性能,其中5项研究利用ADC阈值区分良性和恶性附件肿块,范围为(1.15~1.25)×10-3 mm2/s。孙碧霞等[25]发现全肿瘤ADC图纹理分析的应用有助于提高对良性、交界性和恶性卵巢上皮肿瘤的鉴别。也有研究报告显示,在软组织病变中,基于最大截面ROI的ADC测量显示出与全病灶ADC直方图相当的诊断效能[26]。我们的研究突出了ADCmean在附件良恶性病变鉴别诊断中的价值,与既往文献报道基本相符[27, 28, 29]

       本研究组单独运用O-RADS MRI评分系统鉴别附件良恶性肿块的AUC为0.849(95% CI:0.786~0.912),特异度为81.0%,敏感度为89.5%。该评分系统除了需要测量TIC曲线来区分O-RADS MRI 3~5分外,无需测量其他复杂的数据,即可系统评估附件肿块恶性风险,方案简便高效[6]。然而,文献报道的O-RADS MRI诊断效能波动范围较大(AUC为0.838~0.926)[10, 11],分析其原因主要有两方面:首先,0-RADS MRI评分会受到诊断医师主观因素影响,诊断效能相对不稳定;其次,O-RADS MRI 3~5分的界定仅依赖“实性组织强化程度”这一单一特征,但在实践中,不典型或罕见病例可能由于缺乏特征性影像表现而导致诊断困难,这些肿块的增强曲线与其他常见的良恶性病变存在重叠,因此样本中不同病理类型分布可能导致诊断效能存在差异。这些发现提示,虽然O-RADS MRI系统具有临床应用简便的优势,但对于不典型病例仍需结合其他指标进行综合判断。

3.3 临床与多参数MRI预测模型的临床价值分析

       本研究首次联合O-RADS MRI、ADCmean和HE4建立logistic回归预测模型,与单独使用O-RADS MRI评分相比,预测模型特异度有所增加(90.5% vs. 81.0%),但同时敏感度稍降低(84.9% vs. 89.5%),整体诊断性能显著提升,AUC从0.849提升至0.944;NRI为0.408、IDI为0.218,差异均有统计学意义,提示预测模型对良恶性病变的区分能力较O-RADS MRI有显著提高。因此,本研究构建的预测模型可为临床决策提供重要参考:对于预测为良性的病变,临床医师可优先考虑腹腔镜等微创手术方式或短期影像学随访,这对保留育龄期患者的生育功能有重要意义;而对于预测为恶性的病例,建议及时转诊至妇科肿瘤专科,以便尽早制定规范化治疗方案[30]。此外,该预测模型还具有多方面的临床应用优势:首先,MR能够以无创且可重复的方式获得附件肿块的影像特征,ADC值和HE4均为易于获取的数据,无需额外的后处理工具,保证了临床实施的可行性;其次,这种方法在提高O-RADS MRI评分诊断准确度的同时,避免了额外检查成本的增加,展现出良好的临床推广潜力和广泛适用性。然而,值得注意的是,本研究利用预测模型导致的误诊病例以交界性黏液性肿瘤、较少见的性索-间质肿瘤为主,推测误诊原因可能是:第一,交界性黏液性肿瘤的生物学和影像学特征与良性重叠,肿瘤内黏液的黏稠性可能降低扩散受限程度,导致ADC值偏高[31];第二,性索-间质肿瘤的罕见性导致研究样本量不足、异质性导致影像表现复杂[32],模型可能未充分学习其特征;第三,HE4主要与上皮性癌相关[14],而性索-间质肿瘤的特异性标志物未被纳入模型,这可能导致该模型在此类肿瘤的诊断中存在局限性。因此,在实际应用中仍需结合临床病史和其他检验结果、多模态影像以甄别。

       目前,国内外学者做了相关研究以优化O-RADS MRI评分的应用。例如,ELSHETRY等[11]联合O-RADS MRI与ADCmean评估对良恶性病变的分类效能,通过将新引入的诊断参数和O-RADS MRI评分依据阈值设置为2个层次节点,构建出树状结构,这种操作简单且容易实施,但其权重分配较为主观。而本次研究通过多因素逻辑回归引入新参数,使诊断策略中各参数的权重分配更为精确且更具客观性。近年来集成MRI(synthetic MRI, SyMRI)等新技术在相关研究中显示出一定潜力,也有学者探讨了SyMRI对O-RADS MRI 3~5分附件肿块良恶性病变鉴别的效能[33],将SyMRI中的T1值、T2*与ADC值结合起来,结果显示SyMRI+ADC模型的AUC为0.879。但SyMRI的实现需要额外的采集序列和分析软件,不是大多数医院盆腔检查的常规序列,在国内应用尚不普遍。目前,运用SyMRI评估卵巢病变的研究案例鲜见报道,故该技术在卵巢病变诊断中的价值尚待进一步的临床试验加以验证。而本研究建立的预测模型基于广泛应用的常规检查参数,具有更好的普适性和推广价值。

3.4 局限性

       本研究为单中心回顾性研究,样本量较小,且未对不同的病理学亚型进行详细评估;此外,未应用影像组学等更先进、更敏感的技术进行提取特征,无法获取更加全面及深层的影像特征。因此,未来的研究还需进一步扩大样本量,采用前瞻性、多中心的方法并运用影像组学等先进技术进一步提高模型的效能和稳健性。

4 结论

       本研究首次联合O-RADS MRI评分、ADCmean及HE4构建logistic回归模型,用于区分附件良恶性肿块。预测模型对附件良恶性肿块的区分效能及临床净收益均优于经典O-RADS MRI评分,可作为术前预测附件良恶性肿块的有力工具,更好地指导临床决策。

[1]
SIEGEL R L, GIAQUINTO A N, JEMAL A. Cancer statistics, 2024[J]. CA A Cancer J Clin, 2024, 74(1): 12-49. DOI: 10.3322/caac.21820.
[2]
WEBB P M, JORDAN S J. Global epidemiology of epithelial ovarian cancer[J]. Nat Rev Clin Oncol, 2024, 21(5): 389-400. DOI: 10.1038/s41571-024-00881-3.
[3]
BRAY F, LAVERSANNE M, SUNG H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[J]. CA Cancer J Clin, 2024, 74(3): 229-263. DOI: 10.3322/caac.21834.
[4]
FROYMAN W, LANDOLFO C, DE COCK B, et al. Risk of complications in patients with conservatively managed ovarian tumours (IOTA5): a 2-year interim analysis of a multicentre, prospective, cohort study[J]. Lancet Oncol, 2019, 20(3): 448-458. DOI: 10.1016/S1470-2045(18)30837-4.
[5]
THOMASSIN-NAGGARA I, AUBERT E, ROCKALL A, et al. Adnexal masses: development and preliminary validation of an MR imaging scoring system[J]. Radiology, 2013, 267(2): 432-443. DOI: 10.1148/radiol.13121161.
[6]
SADOWSKI E A, THOMASSIN-NAGGARA I, ROCKALL A, et al. O-RADS MRI risk stratification system: guide for assessing adnexal lesions from the ACR O-RADS committee[J]. Radiology, 2022, 303(1): 35-47. DOI: 10.1148/radiol.204371.
[7]
THOMASSIN-NAGGARA I, PONCELET E, JALAGUIER-COUDRAY A, et al. Ovarian-adnexal reporting data system magnetic resonance imaging (O-RADS MRI) score for risk stratification of sonographically indeterminate adnexal masses[J/OL]. JAMA Netw Open, 2020, 3(1): e1919896 [2024-06-08]. https://pubmed.ncbi.nlm.nih.gov/31977064/. DOI: 10.1001/jamanetworkopen.2019.19896.
[8]
RIZZO S, COZZI A, DOLCIAMI M, et al. O-RADS MRI: a systematic review and meta-analysis of diagnostic performance and category-wise malignancy rates[J/OL]. Radiology, 2023, 307(1): e220795 [2024-12-04]. https://pubmed.ncbi.nlm.nih.gov/36413127/. DOI: 10.1148/radiol.220795.
[9]
BOUROUROU R, NOUGARET S, ROCKALL A, et al. Apparent diffusion coefficient analysis of solid tissue helps distinguish borderline from invasive malignant adnexal masses rated O-RADS MRI 4[J]. Diagn Interv Imag, 2024, 105(10): 386-394. DOI: 10.1016/j.diii.2024.05.004.
[10]
李彩红, 李易, 刘柳, 等. 磁共振结合临床指标对优化O-RADS MRI 4分肿块风险分层的价值[J]. 磁共振成像, 2024, 15(11): 103-109.
LI C H, LI Y, LIU L, et al. Value of MRI combining with clinical indicators in optimizing the risk stratification of O-RADS MRI Score 4[J]. Chin J Magn Reson Imag, 2024, 15(11): 103-109.
[11]
ELSHETRY A S F, HAMED E M, ABDEL FATTAH FRERE R, et al. Impact of adding mean apparent diffusion coefficient (ADCmean) measurements to o-rads MRI scoring for adnexal lesions characterization: a combined o-rads MRI/ADCmean approach[J]. Acad Radiol, 2023, 30(2): 300-311. DOI: 10.1016/j.acra.2022.07.019.
[12]
MANGANARO L, CIULLA S, CELLI V, et al. Impact of DWI and ADC values in ovarian-adnexal reporting and data system (O-RADS) MRI score[J]. Radiol Med, 2023, 128(5): 565-577. DOI: 10.1007/s11547-023-01628-3.
[13]
ASSOULINE V, DABI Y, JALAGUIER-COUDRAY A, et al. How to improve O-RADS MRI score for rating adnexal masses with cystic component [J]. Eur Radiol, 2022, 32(9): 5943-5953. DOI: 10.1007/s00330-022-08644-3.
[14]
陈琪. 血清HE4联合CA125检测在卵巢癌早期诊断及预后评估中的临床价值[J]. 临床输血与检验, 2016, 18(5): 478-481. DOI: 10.3969/j.issn.1671-2587.2016.05.025.
CHEN Q. Clinical value of serum HE4 combined with CA125 detection in early diagnosis and prognosis of ovarian cancer[J]. J Clin Transfus Lab Med, 2016, 18(5): 478-481. DOI: 10.3969/j.issn.1671-2587.2016.05.025.
[15]
薛玲玲. 血清HE4与CA125及ROMA在鉴别诊断卵巢良恶性肿瘤中的临床应用价值[J]. 实用癌症杂志, 2019, 34(8): 1369-1371, 1385. DOI: 10.3969/j.issn.1001-5930.2019.08.042.
XUE L L. Clinical application value of serum HE4, CA125 and ROMA in differentiating benign and malignant ovarian tumors[J]. Pract J Cancer, 2019, 34(8): 1369-1371, 1385. DOI: 10.3969/j.issn.1001-5930.2019.08.042.
[16]
WONG B Z Y, CAUSA ANDRIEU P I, SONODA Y, et al. Improving risk stratification of indeterminate adnexal masses on MRI: What imaging features help predict malignancy in O-RADS MRI 4 lesions [J/OL]. Eur J Radiol, 2023, 168: 111122 [2024-11-23]. https://pubmed.ncbi.nlm.nih.gov/37806193/. DOI: 10.1016/j.ejrad.2023.111122.
[17]
MUKUDA N, FUJII S, INOUE C, et al. Bilateral ovarian tumors on MRI: how should we differentiate the lesions [J]. Yonago Acta Med, 2018, 61(2): 110-116. DOI: 10.33160/yam.2018.06.003.
[18]
SAHA B, MATHUR T, TRONOLONE J J, et al. Human tumor microenvironment chip evaluates the consequences of platelet extravasation and combinatorial antitumor-antiplatelet therapy in ovarian cancer[J/OL]. Sci Adv, 2021, 7(30): eabg5283 [2024-12-04]. https://pubmed.ncbi.nlm.nih.gov/34290095/. DOI: 10.1126/sciadv.abg5283.
[19]
ALI A T, AL-ANI O, AL-ANI F. Epidemiology and risk factors for ovarian cancer[J]. Prz Menopauzalny, 2023, 22(2): 93-104. DOI: 10.5114/pm.2023.128661.
[20]
GOTLIEB W H, CHETRIT A, MENCZER J, et al. Demographic and genetic characteristics of patients with borderline ovarian tumors as compared to early stage invasive ovarian cancer[J]. Gynecol Oncol, 2005, 97(3): 780-783. DOI: 10.1016/j.ygyno.2005.02.022.
[21]
TROPÉ C G, KRISTENSEN G, MAKAR A. Surgery for borderline tumor of the ovary[J]. Semin Surg Oncol, 2000, 19(1): 69-75. DOI: 10.1002/1098-2388(200007/08)19:1<69::aid-ssu11>3.0.co;2-e.
[22]
TIAN Y P, WANG C X, CHENG L M, et al. Determination of reference intervals of serum levels of human epididymis protein 4 (HE4) in Chinese women[J/OL]. J Ovarian Res, 2015, 8: 72 [2024-12-04].https://pubmed.ncbi.nlm.nih.gov/26552478/. DOI: 10.1186/s13048-015-0201-z.
[23]
MARKO J, MARKO K I, PACHIGOLLA S L, et al. Mucinous neoplasms of the ovary: radiologic-pathologic correlation[J]. Radiographics, 2019, 39(4): 982-997. DOI: 10.1148/rg.2019180221.
[24]
KIM H J, LEE S Y, SHIN Y R, et al. The value of diffusion-weighted imaging in the differential diagnosis of ovarian lesions: A meta-analysis[J/OL]. PLoS One, 2016, 11(2): e0149465 [2024-09-12].https://pubmed.ncbi.nlm.nih.gov/26907919/. DOI: 10.1371/journal.pone.0149465.
[25]
孙碧霞, 朱大林, 张旭霞, 等. ADC图纹理分析对卵巢上皮肿瘤的鉴别诊断价值[J]. 磁共振成像, 2023, 14(2): 83-86, 108. DOI: 10.12015/issn.1674-8034.2023.02.014.
SUN B X, ZHU D L, ZHANG X X, et al. The value of ADC texture analysis in differential diagnosis of ovarian epithelial tumors[J]. Chin J Magn Reson Imag, 2023, 14(2): 83-86, 108. DOI: 10.12015/issn.1674-8034.2023.02.014.
[26]
OZTURK M, POLAT A V, SELCUK M B. Whole-lesion ADC histogram analysis versus single-slice ADC measurement for the differentiation of benign and malignant soft tissue tumors[J/OL]. Eur J Radiol, 2021, 143: 109934 [2024-09-12].https://pubmed.ncbi.nlm.nih.gov/34500411/. DOI: 10.1016/j.ejrad.2021.109934.
[27]
MALEK M, POURASHRAF M, MOUSAVI A S, et al. Differentiation of benign from malignant adnexal masses by functional 3 tesla MRI techniques: diffusion-weighted imaging and time-intensity curves of dynamic contrast-enhanced MRI[J]. Asian Pac J Cancer Prev, 2015, 16(8): 3407-3412. DOI: 10.7314/apjcp.2015.16.8.3407.
[28]
HOTTAT N A, BADR D A, VAN PACHTERBEKE C, et al. Added value of quantitative analysis of diffusion-weighted imaging in ovarian-adnexal reporting and data system magnetic resonance imaging[J]. J Magn Reson Imaging, 2022, 56(1): 158-170. DOI: 10.1002/jmri.28003.
[29]
LIN Y, HSIEH C Y, HUANG Y L, et al. Magnetic resonance spectroscopy for risk stratification of sonographically indeterminate ovarian neoplasms: preliminary study[J/OL]. Diagnostics (Basel), 2021, 11(10): 1847 [2024-06-04]. https://pubmed.ncbi.nlm.nih.gov/34679545/. DOI: 10.3390/diagnostics11101847.
[30]
DABI Y, ROCKALL A, RAZAKAMANANTSOA L, et al. O-RADS MRI scoring system has the potential to reduce the frequency of avoidable adnexal surgery[J]. Eur J Obstet Gynecol Reprod Biol, 2024, 294: 135-142. DOI: 10.1016/j.ejogrb.2024.01.016.
[31]
YANG X Y, LI X, MA F H, et al. MRI characteristics for differentiating mucinous borderline ovarian tumours from mucinous ovarian cancers[J]. Clin Radiol, 2022, 77(2): 142-147. DOI: 10.1016/j.crad.2021.10.022.
[32]
MITCHELL J R, SIEGELMAN E S, SUNDARAM K M. MR imaging of germ cell and sex cord stromal tumors[J]. Magn Reson Imaging Clin N Am, 2023, 31(1): 65-78. DOI: 10.1016/j.mric.2022.07.003.
[33]
李海蛟, 曹崑, 李晓婷, 等. 合成MRI在O-RADS MRI 3~5分卵巢附件肿物良恶性鉴别中的价值[J]. 磁共振成像, 2024, 15(5): 148-153, 161. DOI: 10.12015/issn.1674-8034.2024.05.023.
LI H J, CAO K, LI X T, et al. Value of synthetic MRI in differential diagnosis of benign and malignant ovarian adnexal lesions with O-RADS MRI score 3-5[J]. Chin J Magn Reson Imag, 2024, 15(5): 148-153, 161. DOI: 10.12015/issn.1674-8034.2024.05.023.

上一篇 MRI功能性肝脏成像评分和自发性门体分流在慢性乙肝患者肝功能评估及首次失代偿事件预测中的研究
下一篇 电影体积渲染技术在腰骶丛神经鞘瘤中的应用价值:一项与最大密度投影的对比研究
  
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