en
×

分享给微信好友或者朋友圈

使用微信“扫一扫”功能。
通讯作者:

陈婷,E-mail:Chentingwzc@163.com

中图分类号:R737.33

文献标识码:A

文章编号:1007-4368(2024)01-052-08

DOI:10.7655/NYDXBNSN220980

参考文献 1
COHEN P A,JHINGRAN A,OAKNIN A,et al.Cervical cancer[J].Lancet,2019,393(10167):169-182
参考文献 2
KOH W J,ABU⁃RUSTUM N R,BEAN S,et al.Cervical cancer,version 3.2019,NCCN clinical practice guide⁃ lines in oncology[J].J Natl Compr Cancer Netw,2019,17(1):64-84
参考文献 3
ELIT L,KENNEDY E B,FYLES A,et al.Follow ⁃up for cervical cancer:a program in evidence⁃based care syste⁃ matic review and clinical practice guideline update[J].Curr Oncol Tor Ont,2016,23(2):109-118
参考文献 4
ZHONG M L,WANG Y N,LIANG M R,et al.Consolida⁃ tion chemotherapy in early ⁃stage cervical cancer patients with lymph node metastasis after radical hysterectomy [J].Int J Gynecol Cancer,2020,30(5):602-606
参考文献 5
MABUCHI S,ISOHASHI F,YOKOI T,et al.A phaseⅡ study of postoperative concurrent carboplatin and pacli⁃ taxel combined with intensity⁃modulated pelvic radiothe⁃ rapy followed by consolidation chemotherapy in surgically treated cervical cancer patients with positive pelvic lymph nodes[J].Gynecol Oncol,2016,141(2):240-246
参考文献 6
CASARIN J,BUDA A,BOGANI G,et al.Predictors of re⁃ currence following laparoscopic radical hysterectomy for early⁃stage cervical cancer:a multi⁃institutional study[J].Gynecol Oncol,2020,159(1):164-170
参考文献 7
COLTURATO L F,SIGNORINI FILHO R C,FER⁃ NANDES R C M,et al.Lymph node micrometastases in initial stage cervical cancer and tumoral recurrence[J].Int J Gynaecol Obstet,2016,133(1):69-75
参考文献 8
BENOIT L,KOUAL M,NGUYEN⁃XUAN H T,et al.Does a pre ⁃ operative conization improve disease ⁃free survival in early⁃stage cervical cancer?[J].Arch Gynecol Obstet,2021,303(1):231-239
参考文献 9
KIM S I,CHO J H,SEOL A,et al.Comparison of survival outcomes between minimally invasive surgery and conven⁃ tional open surgery for radical hysterectomy as primary treatment in patients with stageⅠB1⁃ⅡA2 cervical can⁃ cer[J].Gynecol Oncol,2019,153(1):3-12
参考文献 10
LIN G,YANG L Y,LIN Y C,et al.Prognostic model based on magnetic resonance imaging,whole ⁃tumour ap⁃ parent diffusion coefficient values and HPV genotyping for stageⅠB⁃ⅠV cervical cancer patients following chemo⁃ radiotherapy[J].Eur Radiol,2019,29(2):556-565
参考文献 11
DAPPA E,ELGER T,HASENBURG A,et al.The value of advanced MRI techniques in the assessment of cervical cancer:a review[J].Insights Imaging,2017,8(5):471-481
参考文献 12
SPOHN S K B,BETTERMANN A S,BAMBERG F,et al.Radiomics in prostate cancer imaging for a personalized treatment approach⁃current aspects of methodology and a systematic review on validated studies[J].Theranostics,2021,11(16):8027-8042
参考文献 13
LUCIA F,VISVIKIS D,DESSEROIT M C,et al.Predic⁃ tion of outcome using pretreatment 18F ⁃FDG PET/CT and MRI radiomics in locally advanced cervical cancer treated with chemoradiotherapy[J].Eur J Nucl Med Mol Imaging,2018,45(5):768-786
参考文献 14
FANG J,ZHANG B,WANG S,et al.Association of MRI⁃ derived radiomic biomarker with disease ⁃free survival in patients with early ⁃ stage cervical cancer[J].Theranos⁃ tics,2020,10(5):2284-2292
参考文献 15
ZHANG Z H,KATTAN M W.Drawing Nomograms with R:applications to categorical outcome and survival data [J].Ann Transl Med,2017,5(10):211
参考文献 16
ABU ⁃ RUSTUM N R,YASHAR C M,BEAN S,et al.NCCN guidelines insights:cervical cancer,version 1.2020 [J].J Natl Compr Cancer Netw,2020,18(6):660-666
参考文献 17
LI Z,LI H,WANG S,et al.MR ⁃ based radiomics nomo⁃ gram of cervical cancer in prediction of the lymph⁃vascu⁃ lar space invasion preoperatively[J].J Magn Reson Imag⁃ ing,2019,49(5):1420-1426
参考文献 18
WANG T,GAO T T,GUO H,et al.Preoperative predic⁃ tion of parametrial invasion in early⁃stage cervical cancer with MRI ⁃ based radiomics nomogram[J].Eur Radiol,2020,30(6):3585-3593
参考文献 19
HOU L,ZHOU W,REN J,et al.Radiomics analysis of multiparametric MRI for the preoperative prediction of lymph node metastasis in cervical cancer[J].Front On⁃ col,2020,10:1393
参考文献 20
TAKADA A,YOKOTA H,WATANABE NEMOTO M,et al.A multi⁃scanner study of MRI radiomics in uterine cer⁃ vical cancer:prediction of in⁃field tumor control after de⁃ finitive radiotherapy based on a machine learning method including peritumoral regions[J].Jpn J Radiol,2020,38(3):265-273
参考文献 21
ZHANG J,SUN J Q,HAN T,et al.Radiomic features of magnetic resonance images as novel preoperative predic⁃ tive factors of bone invasion in meningiomas[J].Eur J Radiol,2020,132:109287
参考文献 22
CHAI R,MA H,XU M,et al.Differentiating axillary lymph node metastasis in invasive breast cancer patients:a comparison of radiomic signatures from multiparametric breast MR sequences[J].J Magn Reson Imaging,2019,50(4):1125-1132
参考文献 23
RAVANELLI M,FARINA D,MORASSI M,et al.Texture analysis of advanced non⁃small cell lung cancer(NSCLC)on contrast⁃enhanced computed tomography:prediction of the response to the first⁃line chemotherapy[J].Eur Radiol,2013,23(12):3450-3455
参考文献 24
OKADOME M,NAGAYAMA R,SHIMOKAWA M,et al.Prognosis of bulky pTIIB cervical cancer treated by radi⁃ cal hysterectomy comparing adenocarcinoma with squa⁃ mous cell carcinoma using propensity score matching[J].Int J Gynecol Obstet,2021,153(1):56-63
参考文献 25
SHIMADA M,NISHIMURA R,NOGAWA T,et al.Com⁃ parison of the outcome between cervical adenocarcinoma and squamous cell carcinoma patients with adjuvant ra⁃diotherapy following radical surgery:SGSG/TGCU Inter⁃ group Surveillance[J].Mol Clin Oncol,2013,1(4):780-784
参考文献 26
TAKEUCHI S.Biology and treatment of cervical adeno⁃ carcinoma[J].Chin J Cancer Res,2016,28(2):254-262
参考文献 27
ZHAO Z T,ZHANG X Y,ZHAO X H,et al.SOX1 and PAX1 are hypermethylated in cervical adenocarcinoma and associated with better prognosis[J].Biomed Res Int,2020,2020:3981529
参考文献 28
HUANG B X,FANG F.Progress in the study of lymph node metastasis in early ⁃ stage cervical cancer[J].Curr Med Sci,2018,38(4):567-574
参考文献 29
PEDONE ANCHORA L,CARBONE V,GALLOTTA V,et al.Should the number of metastatic pelvic lymph nodes be integrated into the 2018 figo staging classification of early stage cervical cancer?[J].Cancers,2020,12(6):1552
参考文献 30
HUANG H,FENG Y L,WAN T,et al.Effectiveness of se⁃ quential chemoradiation vs concurrent chemoradiation or radiation alone in adjuvant treatment after hysterectomy for cervical cancer:the STARS phase 3 randomized clini⁃ cal trial[J].JAMA Oncol,2021,7(3):361-369
参考文献 31
LEE S I,ATRI M.2018 FIGO staging system for uterine cervical cancer:enter cross⁃sectional imaging[J].Radio⁃ logy,2019,292(1):15-24
参考文献 32
SHU T,ZHAO D,LI B,et al.Prognostic evaluation of postoperative adjuvant therapy for operable cervical can⁃ cer:10 years’experience of National Cancer Center in China[J].Chin J Cancer Res,2017,29(6):510-520
参考文献 33
ÁYEN Á,JIMÉNEZ MARTÍNEZ Y,BOULAIZ H.Targeted gene delivery therapies for cervical cancer[J].Cancers,2020,12(5):1301
目录contents

    摘要

    目的:对比最优磁共振组学参数及临床病理参数预测早期宫颈癌患者无病生存期的诊断效能。方法:回顾性分析2013年1月—2018年6月行根治性子宫切除及淋巴结清扫术的186例早期宫颈癌患者[术前国际妇产科联盟(Federation International of Gynecology and Obstetrics,FIGO)分期ⅠB~ⅡA],搜集患者多模态磁共振组学参数、临床特征、术后病理特征及患者术后无病生存期,用最小决策收缩和选择算子(LASSO)及比例风险回归模型来计算影像组学评分(Rad-score),构建临床病理特征模型(包括有意义的临床特征和病理特征)、影像组学模型(T1CE、DWI、T2WI、T1CE+DWI、T1CE+T2WI、DWI+T2WI 和T1CE+DWI+T2WI的影像组学评分)及联合模型对早期宫颈癌患者无病生存期的预测效能进行相互比较。结果:在预测早期宫颈癌无病生存期中,T1CE序列的组学模型(C指数:训练集0.798,验证集0.758)优于临床病理模型(C指数:训练集0.746,验证集0.641)。联合模型(病理类型、淋巴结转移及T1CE的影像组学评分)拥有最高的诊断效能(C指数:训练集0.848,验证集 0.784)。结论:基于T1CE的影像组学评分联合临床病理特征对早期宫颈癌无病生存期具有较高的预测效能。

    Abstract

    Objective:To determine the optimal preoperative magnetic resonance imaging(MRI)radiomics parameters and compare the values in predicting disease-free survival(DFS)of early-stage cervical cancer patients with clinicopathological factors. Methods:A total of 186 patients with early-stage cervical cancers(preoperative FIGO stage ⅠB-ⅡA),who underwent radical hysterectomy with lymphadenectomy during January 2013 and June 2018,were retrospectively reviewed. Multi-sequence MR data,clinicopathologic data, and DFS data were collected;least - absolute shrinkage and selection operator(LASSO)and the Cox proportional hazard regression model were applied to construct a Rad - score. Clinicopathological models(with significant clinicopathological features),radiomics models(Rad-scores for T1CE,DWI,T2WI,T1CE+DWI,T1CE+T2WI,DWI+T2WI,and T1CE+DWI+T2WI),and combined models (the Rad - score model combined with significant clinicopathological features)were compared using a multivariable Cox model. Results:The T1CE radiomics model was the most stable and optimal model(C-index for training,0.798;for validation,0.758)of all the radiomics models. The radiomics model of T1CE showed higher prognostic performance than the clinicopathological model(C-index for training,0.746;for validation,0.641). The combined model(histological type of adenocarcinoma,lymph-node metastasis together with Rad-score for T1CE)showed the highest prognostic performance in estimating DFS(C-index for training,0.848;for validation,0.784). Conclusion:An MRI-derived Rad-score of T1CE combined with a clinicopathological model is optimal in predicting DFS in early-stage cervical cancers.

  • 宫颈癌发病率较高,是全球第四大女性恶性肿瘤[1]。根治性全子宫切除术及盆腔淋巴结切除术是早期宫颈癌的主要治疗方法[2-3]。如果术后存在病理高危因素,如淋巴结转移等,需要进一步进行术后放化疗。然而,在当前的治疗方案下,20%~30% 的患者仍然会出现复发,18%~25%的患者会出现远处转移[4]。因此,即使临床病理特征类似的患者,肿瘤的异质性分析仍具有较大的临床意义。早期准确预测预后可以缩短随访时间和实施个性化的积极治疗策略(如巩固化疗周期),进一步提高生存率[5]。因此,在早期宫颈癌预后评估中,探索更准确的预后生物标志物尤为重要。

  • 既往临床研究认为肿瘤体积[6]、淋巴结转移[7]、术前锥切史[8] 和微创手术[9] 是预测早期宫颈癌术后无病生存期(disease⁃free survival,DFS)常用的临床病理高危因素。然而,这些临床病理危险因素大多是单独进行比较并且只有通过活检或手术才能得到预测结果。因此,进一步探寻非侵入性的预测标志物具有较大的临床应用价值。

  • 前期关于磁共振参数预测宫颈癌预后的研究主要集中于晚期宫颈癌患者,如表观扩散系数联合动态对比增强磁共振参数预测晚期宫颈癌患者放化疗后DFS[10-11]。近年来,放射组学引起了越来越多的关注,相较于传统的影像学或临床病理学参数表现出更高的预测能力[12-13]。前期只有一项基于磁共振影像组学分析预测早期宫颈癌术后 DFS 的研究,该研究发现,相对于临床病理特征,T2加权成像 (T2WI)联合T1对比增强成像(T1CE)的影像组学特征有更好的预测价值[14]。但是,上述研究仅采用不同序列间的比较且相对简单,一些重要的临床病理特征,如肿瘤大小、手术方式和宫颈间质浸润深度都没有纳入研究。

  • 本研究旨在比较不同磁共振序列以及不同序列的组合模型,构建早期宫颈癌患者术后DFS的最优影像组学预测模型;并进一步对影像组学模型与临床病理模型及联合模型的预测效能进行比较。

  • 1 对象和方法

  • 1.1 对象

  • 回顾性分析2013年1月—2018年6月在南京医科大学第一附属医院行根治性子宫切除术及淋巴结清扫术的早期宫颈癌患者[术前国际妇产科联盟 (Federation International of Gynecology and Obstetrics, FIGO)分期ⅠB~ⅡA]。进一步搜集多参数磁共振特征、临床病理数据和 DFS 数据。纳入标准如下: ①术前磁共振 FIGO 分期为ⅠB~ⅡA 期的宫颈癌; ②规范化的术前盆腔磁共振平扫和增强检查;③所有患者均行根治性子宫切除术和双侧盆腔淋巴结清扫术。排除标准如下:①术前行新辅助化疗或放疗;②罕见类型的宫颈癌(囊性黏液腺癌、神经内分泌肿瘤、癌肉瘤);③临床或磁共振数据不完整;④有其他恶性肿瘤或合并恶性肿瘤病史的患者。本研究经医院伦理审查委员会批准,所有患者均知情同意。

  • 最终纳入 186 例患者,平均年龄(45.88±8.36) 岁,并以7∶3的比例随机分配到训练集(n=131)和验证集(n=55)。根据病理结果的危险因素,166 例患者进行了根治手术后的补充放化疗。1~2年内每季度进行1次定期随访,3~5年内每年2次随访,此后每年1次随访。DFS是指从手术日到复发、转移、死亡或最后随访的日期,通过妇科检查、影像学检查、血液检查或病理学检查进行全面评估。

  • 1.2 方法

  • 1.2.1 磁共振图像获取与分割

  • 核磁共振检查采用 3.0T 核磁共振扫描仪 (MAGNETOM TrioTim,西门子公司,德国),配有16通道盆腔线圈。矢状位 T1CE:TR/TE 650/9.3 ms;视野279 mm×279 mm;层厚3.5 mm。轴位T2WI:TR/TE 4 430/129 ms;视野 250 mm×250 mm;层厚 3.5 mm。轴位 DWI:TR/TE 6 800/98 ms;视野,25 cm;层厚 3.0 mm。选择B值为1 000 s/mm2 的图像用于感兴趣区域的勾画。

  • 使用 ITK ⁃ SNAP(v.3.6.0;www.itksnap.org)对磁共振图像进行手动三维分割。肿瘤的感兴趣区域由一位 7 年盆腔磁共振诊断经验的放射科医生勾画,由具有20年盆腔磁共振诊断经验的高级放射科医生对分割区域进行确认。可靠性通过组间相关系数(intraclass correlation coefficient,ICC)计算, ICC高于0.8的特征被保留。

  • 1.2.2 组学特征提取及筛选

  • 通过 AK 软件(Analysis Kit,GE Healthcare,美国)提取磁共振图像组学特征,从 T1CE、T2WI 和 DWI 序列中分别提取 1 040 个特征,包括一阶统计学特征、基于尺寸和形状的特征、灰度级共生矩阵特征、灰度形成矩阵特征、灰度区域大小矩阵特征、灰度依赖矩阵特征和相邻灰阶差异矩阵特征。

  • 采用两步特征选择法保留 DFS 相关的组学特征。首先,用单因素风险回归法筛选训练集中组学特征,并计算预测的 P 值,P <0.05 的特征被保留。接着,多变量逻辑回归与最小决策收缩和选择算子(least absolute shrinkage and selection operator, LASSO)进行多变量特征的选择。LASSO涉及一个参数λ,以使C⁃指数最大并控制所选最优特征的数量。影像组学评分(Rad⁃score)由所选组学特征的指数组合来计算。在训练集中,多元比例风险模型也被用来预测影像组学评分。用 C⁃指数来评估影像组学评分的预测价值。C⁃指数表示模型预测的 DFS 和实际 DFS 间的一致性(<0.5 表示预测值较差,>0.7表示预测值较好)。

  • 1.2.3 影像组学、临床病理模型及联合模型的建立

  • 分别计算T1CE、DWI、T2WI、T1CE+DWI、T1CE+ T2WI、DWI+T2WI 和 T1CE+DWI+T2WI 的影像组学评分,并进一步比较各序列及组合的 C⁃指数。不同序列中影像组学评分的效能在验证集中进一步验证。

  • 临床病理特征包括年龄、术后化疗、肿瘤大小、组织学类型、锥切史、手术方式(微创与否)、组织学分级、淋巴结转移、淋巴血管间隙侵犯、宫颈间质浸润深度和宫旁浸润。这些参数首先在训练集和验证集之间进行比较,然后利用单变量比例风险模型在训练集中筛选有意义的预后因素(P <0.05)。将有意义的变量纳入多变量风险预测模型中,建立预测 DFS 的临床病理模型。进一步在验证集中计算模型的效能。

  • 将筛选出的有意义的临床病理特征分别联合不同序列的影像组学评分,构建联合的多变量风险预测模型,通过C⁃指数比较预测效能。进一步在验证集中计算联合模型的效能。

  • 1.2.4 临床应用

  • 列线图是一种易于使用的临床诊断工具,将有意义的参数通过风险量表的形式呈现[15]。本文绘制了具有最大诊断效能模型的列线图。

  • 1.3 统计学方法

  • 使用R4.2.1(https://www.r⁃project.org)、MedCalc (15.0版)和SPSS 18.0进行统计分析。用独立样本t 检验比较连续变量(年龄、肿瘤大小和影像组学评分),用卡方检验比较训练集和验证集间分类变量 (放化疗、组织学类型、锥切史、手术方式、组织学等级、淋巴结转移、淋巴血管间隙侵犯、宫颈间质浸润深度和宫旁浸润)。使用“生存”软件包建立比例风险预测模型。使用“Surminer”软件包构建生存曲线并计算。P <0.05为差异有统计学意义。

  • 2 结果

  • 2.1 患者特征

  • 共纳入186例患者,平均年龄(45.88±8.36)岁,中位随访时间 57.5 个月。临床病理特征在训练集和验证集间无统计学差异(年龄,P=0.75;肿瘤长径、宽径、纵径,P=0.88、0.65、0.53;化放疗,P=0.75;组织学类型,P=0.33;锥切,P=0.78;手术方法,P=0.10;组织学等级,P=0.85;淋巴结转移,P=0.25;淋巴血管间隙侵犯,P=1.00;宫颈间质浸润深度,P=0.06;宫旁浸润,P=0.79)。

  • 2.2 组学模型预测DFS的效能

  • 对于T1CE,共筛选了8个组学特征。在训练集中,放射学评分的危险比(hazard ratio,HR)为 3.71 (95%CI:2.31~5.97),训练集中预测效能的C⁃指数为 0.798,验证集为0.758,在训练集和验证集中具有最优的预测效能。

  • 对于 DWI,共筛选了 2 个组学特征。在训练集中,放射学评分的 HR 为 4 793.88(95% CI: 32.11~7 156.09),训练集中预测效能的 C⁃指数为 0.740,验证集为0.525。

  • 对于T2WI,共筛选了6个组学特征。在训练集中,放射学评分的 HR 为 5.20(95%CI:2.23~12.12)。训练集中预测效能的C⁃指数为0.782;验证集为0.493。

  • 对于 T1CE+DWI,共筛选了 5 个组学特征。在训练集中,放射学评分的 HR 为 6.40(95%CI:3.13~13.10)。训练集中预测效能的C⁃指数为0.759;验证集为0.699。

  • 对于 T1CE+T2WI,共筛选了 10 个组学特征。在训练集中,放射学评分的 HR 为 4.37(95% CI: 2.59~7.37)。训练集中预测效能的C⁃指数为0.822; 验证集为0.667。

  • 对于DWI+T2WI,共筛选了13个组学特征。在训练集中,放射学评分的HR为15.41(95%CI:6.47~36.69)。训练集中预测效能的C⁃指数为0.903;验证集为0.578。

  • 对于 T1CE+DWI+T2WI,共筛选了 4 个组学特征。在训练集中,放射学评分的HR为6.47(95%CI: 3.16~13.24)。训练集中预测效能的 C ⁃ 指数为 0.762;验证集为0.695。

  • 2.3 临床病理特征模型预测DFS的效能

  • 通过单因素和多因素分析筛选出3个有意义的临床病理特征(组织学类型、淋巴结转移和宫旁浸润)并构建预测模型。在训练集中,肿瘤的病理组织学类型的 HR 为 3.43(95%CI:1.74~6.76),淋巴结转移的 HR 为 3.53(95%CI:1.40~8.94),而宫旁浸润的HR为2.50(95%CI:0.79~7.86)(表1)。训练集中预测效能的C⁃指数为0.746;验证集为0.641。

  • 表1 训练集中临床病理因素预测DFS的单变量和多变量Cox回归分析

  • Table1 Univariate and multivariate Cox regression analysis of clinicopathological factors for predicting DFS in training cohort

  • 患者1进行根治性全子宫切除及盆腔淋巴结清扫术。术后病理宫颈鳞癌,淋巴结阳性,脉管内癌栓阴性,术后矫正 FIGO 分期ⅢCp;术后补充化疗,随访 19 个月出现远处转移;列线图预测 3 年 DFS 概率 63%,鳞癌 0 分,淋巴结 15 分,影像组学评分 72 分。患者 2 进行根治性全子宫切除及盆腔淋巴结清扫术。术后病理宫颈鳞癌,淋巴结阳性,脉管内癌栓阴性,术后矫正 FIGO 分期ⅢCp; 术后补充化疗,随访 79 个月正常;列线图预测 3 年 DFS 概率 >90%,鳞癌0分,淋巴结15分,影像组学评分25分。

  • 2.4 联合模型预测DFS的效能

  • 通过多因素分析筛选联合模型特征,联合组织学类型腺癌、淋巴结转移及 T1CE 的影像组学评分在预测 DFS 中有最优的效能(训练集 C ⁃指数为 0.848;验证集C⁃指数为0.784,表2)。组织学类型腺癌的 HR 在训练集中为 2.04(95%CI:0.95~4.39),在验证集中为 2.37(95%CI:0.49~11.43),淋巴结转移的 HR 在训练集中为 2.74(95%CI:1.13~6.63),验证集中为2.17(95%CI:0.46~10.25),训练集中T1CE的影像组学评分的 HR 为 3.35(95%CI:1.95~5.77),验证集中为4.23(95%CI:1.13~15.77)(表3)。

  • 3 种模型的生存曲线分析,组学特征模型、临床病理特征模型及联合模型均可有效预测 DFS (P <0.05,图1)。

  • 表2 磁共振组学模型与联合临床模型预测DFS的效能对比

  • Table2 Comparison of performance between MR radiomi⁃ cs signature model and radiomics combining clini⁃ cal model in predicting DFS

  • 2.5 临床应用

  • 联合模型(组织学类型腺癌、淋巴结转移及 T1CE的影像组学评分)的风险量表被转换为列线图的形式呈现(图2)。通过列线图对早期宫颈癌DFS 率进行预测比较,2例ⅠB期宫颈癌患者肿瘤大小一致,均为鳞状细胞癌,淋巴结转移均阳性,DFS却完全不同。

  • 表3 T1CE组学特征联合临床病理特征预测DFS的多变量 Cox回归分析

  • Table3 Multivariate Cox regression analysis of T1CE radiomics signature combining clinical factors for predicting DFS

  • 3 讨论

  • 本研究比较了不同序列的放射组学评分与临床病理特征在预测早期宫颈癌 DFS 中的效能。 T1CE的放射组学评分与临床病理高危特征(腺癌和淋巴结转移)联合模型是预测DFS的最优模型。

  • NCCN指南里针对宫颈癌的临床预后评估和治疗策略没有纳入磁共振参数[16]。然而既往研究和本研究结果都表明,相对于临床病理危险因素,磁共振组学特征预测早期宫颈癌 DFS 具有更高的效能。磁共振组学特征联合临床病理危险因素,可以为早期宫颈癌患者提供更准确的个性化治疗策略。

  • 宫颈癌的磁共振影像组学分析广泛应用于预测淋巴血管间隙浸润[17]、宫旁侵犯[18]、淋巴结转移[19] 和化放疗监测[20]。前期有一项研究关注磁共振组学在预测早期宫颈癌术后DFS中的价值[14];作者发现 T1CE 联合 T2WI 的放射组学评分在预测 DFS 中最优(C⁃指数,0.753),优于临床病理特征(C⁃指数, 0.632)或两者组合(C⁃指数,0.714)。然而,该研究并没有比较其他不同组合的磁共振序列,他们的结果与本研究部分结果相似(对于临床病理特征,C⁃指数为 0.641;对于 T1CE+T2WI 的放射组学评分联合临床病理特征,C⁃指数为0.739)。本研究中将磁共振单一或组合序列的放射组学评分效能进行比较时, T1CE的放射组学评分预测DFS最优(训练集中C⁃指数为0.798;验证集中C⁃指数为0.758)。

  • 图1 在训练集及验证集中3种模型的生存曲线分析

  • Figure1 Kaplan⁃Meier analysis of the three models in the training and validation cohorts

  • 图2 联合模型的列线图

  • Figure2 Nomogram of the combined model

  • 前期相关研究也认为基于T1CE的组学预测效能较高,对脑膜瘤的骨侵犯预测 T1CE 组学效能优于T2WI[21];对侵袭性乳腺癌前哨淋巴结转移的预测也优于T2WI、DWI及二者联合[22]。T1CE可以通过增强模式反映肿瘤的血管生成异质性,本研究中有 8 个 T1CE 组学参数与 DFS 密切相关,多于 DWI 或 T2WI,这可能反映了 T1CE 参数与 DFS 的相关性更高。在这 8 个参数中,灰度共生矩阵值的差异是一个衡量异质性的指标,可以区分具有高度异质性增强成分的肿瘤。两个均一性参数也可以反映图像的均匀性并预示肿瘤新生血管。前期研究验证了均一性参数在预测晚期肺癌化疗反应中的重要价值[23]。灰度共生矩阵的集群阴影是衡量偏斜度和均一性的指标,意味着更大的不对称性。一阶最小值可以反映肿瘤最具异质性的部分。这些有意义的组学参数主要反映肿瘤的新生血管和强化方式的异质性,这可能与肿瘤的远处转移和复发有关。

  • 在联合模型中,腺癌及淋巴结转移是多变量回归分析计算后最终纳入的有意义的临床病理特征。Okadome 等[24] 研究同样发现腺癌相对于鳞癌具有更低的 5 年 DFS。部分研究者认为,术后放化疗作为宫颈腺癌患者的辅助治疗可能不一定有效[25],腺癌的转移率较高,对放化疗的敏感性相对较低[26-27]。淋巴结转移被认为是不同肿瘤DFS的一个独立危险因素[28-30]。2018年FIGO明确规定,有淋巴结转移的宫颈癌应归为ⅢC期[31]。Shu等[32] 认为,使用多变量分析预测宫颈癌 DFS 时,淋巴结转移是独立危险因素(HR=2.91;95%CI:1.94~4.36),这与本研究结果类似(HR=2.74;95%CI:1.13~6.63)。其他危险参数如淋巴血管间隙侵犯和宫旁浸润最终没有纳入联合模型,其预测能力可能受到其他风险因素的干扰。

  • 本研究存在以下局限性:首先,临床血液学参数如鳞状细胞癌抗原等没有纳入,因为这些参数在前期研究中不是独立危险因素;其次,组学参数与预后相关的生物学标志物如DNA修复基因、P53和 E6/7 蛋白的相关性也没有计算[33],因此,组学联合治疗相关的生物标志物有待进一步研究;再次,本研究中感兴趣区域为手动分割,使用机器学习来自动分割和提取肿瘤特征有待在以后的研究中进一步探索;最后,虽然本研究中的患者数量多于之前的研究,但仍需要在多中心研究中进一步验证。

  • 综上所述,本研究认为 T1CE 的放射组学评分联合临床病理危险因素(腺癌、淋巴结转移)是预测早期宫颈癌 DFS 的最优模型。该模型的应用可有助于预测早期宫颈癌预后和优化治疗决策。

  • 参考文献

    • [1] COHEN P A,JHINGRAN A,OAKNIN A,et al.Cervical cancer[J].Lancet,2019,393(10167):169-182

    • [2] KOH W J,ABU⁃RUSTUM N R,BEAN S,et al.Cervical cancer,version 3.2019,NCCN clinical practice guide⁃ lines in oncology[J].J Natl Compr Cancer Netw,2019,17(1):64-84

    • [3] ELIT L,KENNEDY E B,FYLES A,et al.Follow ⁃up for cervical cancer:a program in evidence⁃based care syste⁃ matic review and clinical practice guideline update[J].Curr Oncol Tor Ont,2016,23(2):109-118

    • [4] ZHONG M L,WANG Y N,LIANG M R,et al.Consolida⁃ tion chemotherapy in early ⁃stage cervical cancer patients with lymph node metastasis after radical hysterectomy [J].Int J Gynecol Cancer,2020,30(5):602-606

    • [5] MABUCHI S,ISOHASHI F,YOKOI T,et al.A phaseⅡ study of postoperative concurrent carboplatin and pacli⁃ taxel combined with intensity⁃modulated pelvic radiothe⁃ rapy followed by consolidation chemotherapy in surgically treated cervical cancer patients with positive pelvic lymph nodes[J].Gynecol Oncol,2016,141(2):240-246

    • [6] CASARIN J,BUDA A,BOGANI G,et al.Predictors of re⁃ currence following laparoscopic radical hysterectomy for early⁃stage cervical cancer:a multi⁃institutional study[J].Gynecol Oncol,2020,159(1):164-170

    • [7] COLTURATO L F,SIGNORINI FILHO R C,FER⁃ NANDES R C M,et al.Lymph node micrometastases in initial stage cervical cancer and tumoral recurrence[J].Int J Gynaecol Obstet,2016,133(1):69-75

    • [8] BENOIT L,KOUAL M,NGUYEN⁃XUAN H T,et al.Does a pre ⁃ operative conization improve disease ⁃free survival in early⁃stage cervical cancer?[J].Arch Gynecol Obstet,2021,303(1):231-239

    • [9] KIM S I,CHO J H,SEOL A,et al.Comparison of survival outcomes between minimally invasive surgery and conven⁃ tional open surgery for radical hysterectomy as primary treatment in patients with stageⅠB1⁃ⅡA2 cervical can⁃ cer[J].Gynecol Oncol,2019,153(1):3-12

    • [10] LIN G,YANG L Y,LIN Y C,et al.Prognostic model based on magnetic resonance imaging,whole ⁃tumour ap⁃ parent diffusion coefficient values and HPV genotyping for stageⅠB⁃ⅠV cervical cancer patients following chemo⁃ radiotherapy[J].Eur Radiol,2019,29(2):556-565

    • [11] DAPPA E,ELGER T,HASENBURG A,et al.The value of advanced MRI techniques in the assessment of cervical cancer:a review[J].Insights Imaging,2017,8(5):471-481

    • [12] SPOHN S K B,BETTERMANN A S,BAMBERG F,et al.Radiomics in prostate cancer imaging for a personalized treatment approach⁃current aspects of methodology and a systematic review on validated studies[J].Theranostics,2021,11(16):8027-8042

    • [13] LUCIA F,VISVIKIS D,DESSEROIT M C,et al.Predic⁃ tion of outcome using pretreatment 18F ⁃FDG PET/CT and MRI radiomics in locally advanced cervical cancer treated with chemoradiotherapy[J].Eur J Nucl Med Mol Imaging,2018,45(5):768-786

    • [14] FANG J,ZHANG B,WANG S,et al.Association of MRI⁃ derived radiomic biomarker with disease ⁃free survival in patients with early ⁃ stage cervical cancer[J].Theranos⁃ tics,2020,10(5):2284-2292

    • [15] ZHANG Z H,KATTAN M W.Drawing Nomograms with R:applications to categorical outcome and survival data [J].Ann Transl Med,2017,5(10):211

    • [16] ABU ⁃ RUSTUM N R,YASHAR C M,BEAN S,et al.NCCN guidelines insights:cervical cancer,version 1.2020 [J].J Natl Compr Cancer Netw,2020,18(6):660-666

    • [17] LI Z,LI H,WANG S,et al.MR ⁃ based radiomics nomo⁃ gram of cervical cancer in prediction of the lymph⁃vascu⁃ lar space invasion preoperatively[J].J Magn Reson Imag⁃ ing,2019,49(5):1420-1426

    • [18] WANG T,GAO T T,GUO H,et al.Preoperative predic⁃ tion of parametrial invasion in early⁃stage cervical cancer with MRI ⁃ based radiomics nomogram[J].Eur Radiol,2020,30(6):3585-3593

    • [19] HOU L,ZHOU W,REN J,et al.Radiomics analysis of multiparametric MRI for the preoperative prediction of lymph node metastasis in cervical cancer[J].Front On⁃ col,2020,10:1393

    • [20] TAKADA A,YOKOTA H,WATANABE NEMOTO M,et al.A multi⁃scanner study of MRI radiomics in uterine cer⁃ vical cancer:prediction of in⁃field tumor control after de⁃ finitive radiotherapy based on a machine learning method including peritumoral regions[J].Jpn J Radiol,2020,38(3):265-273

    • [21] ZHANG J,SUN J Q,HAN T,et al.Radiomic features of magnetic resonance images as novel preoperative predic⁃ tive factors of bone invasion in meningiomas[J].Eur J Radiol,2020,132:109287

    • [22] CHAI R,MA H,XU M,et al.Differentiating axillary lymph node metastasis in invasive breast cancer patients:a comparison of radiomic signatures from multiparametric breast MR sequences[J].J Magn Reson Imaging,2019,50(4):1125-1132

    • [23] RAVANELLI M,FARINA D,MORASSI M,et al.Texture analysis of advanced non⁃small cell lung cancer(NSCLC)on contrast⁃enhanced computed tomography:prediction of the response to the first⁃line chemotherapy[J].Eur Radiol,2013,23(12):3450-3455

    • [24] OKADOME M,NAGAYAMA R,SHIMOKAWA M,et al.Prognosis of bulky pTIIB cervical cancer treated by radi⁃ cal hysterectomy comparing adenocarcinoma with squa⁃ mous cell carcinoma using propensity score matching[J].Int J Gynecol Obstet,2021,153(1):56-63

    • [25] SHIMADA M,NISHIMURA R,NOGAWA T,et al.Com⁃ parison of the outcome between cervical adenocarcinoma and squamous cell carcinoma patients with adjuvant ra⁃diotherapy following radical surgery:SGSG/TGCU Inter⁃ group Surveillance[J].Mol Clin Oncol,2013,1(4):780-784

    • [26] TAKEUCHI S.Biology and treatment of cervical adeno⁃ carcinoma[J].Chin J Cancer Res,2016,28(2):254-262

    • [27] ZHAO Z T,ZHANG X Y,ZHAO X H,et al.SOX1 and PAX1 are hypermethylated in cervical adenocarcinoma and associated with better prognosis[J].Biomed Res Int,2020,2020:3981529

    • [28] HUANG B X,FANG F.Progress in the study of lymph node metastasis in early ⁃ stage cervical cancer[J].Curr Med Sci,2018,38(4):567-574

    • [29] PEDONE ANCHORA L,CARBONE V,GALLOTTA V,et al.Should the number of metastatic pelvic lymph nodes be integrated into the 2018 figo staging classification of early stage cervical cancer?[J].Cancers,2020,12(6):1552

    • [30] HUANG H,FENG Y L,WAN T,et al.Effectiveness of se⁃ quential chemoradiation vs concurrent chemoradiation or radiation alone in adjuvant treatment after hysterectomy for cervical cancer:the STARS phase 3 randomized clini⁃ cal trial[J].JAMA Oncol,2021,7(3):361-369

    • [31] LEE S I,ATRI M.2018 FIGO staging system for uterine cervical cancer:enter cross⁃sectional imaging[J].Radio⁃ logy,2019,292(1):15-24

    • [32] SHU T,ZHAO D,LI B,et al.Prognostic evaluation of postoperative adjuvant therapy for operable cervical can⁃ cer:10 years’experience of National Cancer Center in China[J].Chin J Cancer Res,2017,29(6):510-520

    • [33] ÁYEN Á,JIMÉNEZ MARTÍNEZ Y,BOULAIZ H.Targeted gene delivery therapies for cervical cancer[J].Cancers,2020,12(5):1301

  • 参考文献

    • [1] COHEN P A,JHINGRAN A,OAKNIN A,et al.Cervical cancer[J].Lancet,2019,393(10167):169-182

    • [2] KOH W J,ABU⁃RUSTUM N R,BEAN S,et al.Cervical cancer,version 3.2019,NCCN clinical practice guide⁃ lines in oncology[J].J Natl Compr Cancer Netw,2019,17(1):64-84

    • [3] ELIT L,KENNEDY E B,FYLES A,et al.Follow ⁃up for cervical cancer:a program in evidence⁃based care syste⁃ matic review and clinical practice guideline update[J].Curr Oncol Tor Ont,2016,23(2):109-118

    • [4] ZHONG M L,WANG Y N,LIANG M R,et al.Consolida⁃ tion chemotherapy in early ⁃stage cervical cancer patients with lymph node metastasis after radical hysterectomy [J].Int J Gynecol Cancer,2020,30(5):602-606

    • [5] MABUCHI S,ISOHASHI F,YOKOI T,et al.A phaseⅡ study of postoperative concurrent carboplatin and pacli⁃ taxel combined with intensity⁃modulated pelvic radiothe⁃ rapy followed by consolidation chemotherapy in surgically treated cervical cancer patients with positive pelvic lymph nodes[J].Gynecol Oncol,2016,141(2):240-246

    • [6] CASARIN J,BUDA A,BOGANI G,et al.Predictors of re⁃ currence following laparoscopic radical hysterectomy for early⁃stage cervical cancer:a multi⁃institutional study[J].Gynecol Oncol,2020,159(1):164-170

    • [7] COLTURATO L F,SIGNORINI FILHO R C,FER⁃ NANDES R C M,et al.Lymph node micrometastases in initial stage cervical cancer and tumoral recurrence[J].Int J Gynaecol Obstet,2016,133(1):69-75

    • [8] BENOIT L,KOUAL M,NGUYEN⁃XUAN H T,et al.Does a pre ⁃ operative conization improve disease ⁃free survival in early⁃stage cervical cancer?[J].Arch Gynecol Obstet,2021,303(1):231-239

    • [9] KIM S I,CHO J H,SEOL A,et al.Comparison of survival outcomes between minimally invasive surgery and conven⁃ tional open surgery for radical hysterectomy as primary treatment in patients with stageⅠB1⁃ⅡA2 cervical can⁃ cer[J].Gynecol Oncol,2019,153(1):3-12

    • [10] LIN G,YANG L Y,LIN Y C,et al.Prognostic model based on magnetic resonance imaging,whole ⁃tumour ap⁃ parent diffusion coefficient values and HPV genotyping for stageⅠB⁃ⅠV cervical cancer patients following chemo⁃ radiotherapy[J].Eur Radiol,2019,29(2):556-565

    • [11] DAPPA E,ELGER T,HASENBURG A,et al.The value of advanced MRI techniques in the assessment of cervical cancer:a review[J].Insights Imaging,2017,8(5):471-481

    • [12] SPOHN S K B,BETTERMANN A S,BAMBERG F,et al.Radiomics in prostate cancer imaging for a personalized treatment approach⁃current aspects of methodology and a systematic review on validated studies[J].Theranostics,2021,11(16):8027-8042

    • [13] LUCIA F,VISVIKIS D,DESSEROIT M C,et al.Predic⁃ tion of outcome using pretreatment 18F ⁃FDG PET/CT and MRI radiomics in locally advanced cervical cancer treated with chemoradiotherapy[J].Eur J Nucl Med Mol Imaging,2018,45(5):768-786

    • [14] FANG J,ZHANG B,WANG S,et al.Association of MRI⁃ derived radiomic biomarker with disease ⁃free survival in patients with early ⁃ stage cervical cancer[J].Theranos⁃ tics,2020,10(5):2284-2292

    • [15] ZHANG Z H,KATTAN M W.Drawing Nomograms with R:applications to categorical outcome and survival data [J].Ann Transl Med,2017,5(10):211

    • [16] ABU ⁃ RUSTUM N R,YASHAR C M,BEAN S,et al.NCCN guidelines insights:cervical cancer,version 1.2020 [J].J Natl Compr Cancer Netw,2020,18(6):660-666

    • [17] LI Z,LI H,WANG S,et al.MR ⁃ based radiomics nomo⁃ gram of cervical cancer in prediction of the lymph⁃vascu⁃ lar space invasion preoperatively[J].J Magn Reson Imag⁃ ing,2019,49(5):1420-1426

    • [18] WANG T,GAO T T,GUO H,et al.Preoperative predic⁃ tion of parametrial invasion in early⁃stage cervical cancer with MRI ⁃ based radiomics nomogram[J].Eur Radiol,2020,30(6):3585-3593

    • [19] HOU L,ZHOU W,REN J,et al.Radiomics analysis of multiparametric MRI for the preoperative prediction of lymph node metastasis in cervical cancer[J].Front On⁃ col,2020,10:1393

    • [20] TAKADA A,YOKOTA H,WATANABE NEMOTO M,et al.A multi⁃scanner study of MRI radiomics in uterine cer⁃ vical cancer:prediction of in⁃field tumor control after de⁃ finitive radiotherapy based on a machine learning method including peritumoral regions[J].Jpn J Radiol,2020,38(3):265-273

    • [21] ZHANG J,SUN J Q,HAN T,et al.Radiomic features of magnetic resonance images as novel preoperative predic⁃ tive factors of bone invasion in meningiomas[J].Eur J Radiol,2020,132:109287

    • [22] CHAI R,MA H,XU M,et al.Differentiating axillary lymph node metastasis in invasive breast cancer patients:a comparison of radiomic signatures from multiparametric breast MR sequences[J].J Magn Reson Imaging,2019,50(4):1125-1132

    • [23] RAVANELLI M,FARINA D,MORASSI M,et al.Texture analysis of advanced non⁃small cell lung cancer(NSCLC)on contrast⁃enhanced computed tomography:prediction of the response to the first⁃line chemotherapy[J].Eur Radiol,2013,23(12):3450-3455

    • [24] OKADOME M,NAGAYAMA R,SHIMOKAWA M,et al.Prognosis of bulky pTIIB cervical cancer treated by radi⁃ cal hysterectomy comparing adenocarcinoma with squa⁃ mous cell carcinoma using propensity score matching[J].Int J Gynecol Obstet,2021,153(1):56-63

    • [25] SHIMADA M,NISHIMURA R,NOGAWA T,et al.Com⁃ parison of the outcome between cervical adenocarcinoma and squamous cell carcinoma patients with adjuvant ra⁃diotherapy following radical surgery:SGSG/TGCU Inter⁃ group Surveillance[J].Mol Clin Oncol,2013,1(4):780-784

    • [26] TAKEUCHI S.Biology and treatment of cervical adeno⁃ carcinoma[J].Chin J Cancer Res,2016,28(2):254-262

    • [27] ZHAO Z T,ZHANG X Y,ZHAO X H,et al.SOX1 and PAX1 are hypermethylated in cervical adenocarcinoma and associated with better prognosis[J].Biomed Res Int,2020,2020:3981529

    • [28] HUANG B X,FANG F.Progress in the study of lymph node metastasis in early ⁃ stage cervical cancer[J].Curr Med Sci,2018,38(4):567-574

    • [29] PEDONE ANCHORA L,CARBONE V,GALLOTTA V,et al.Should the number of metastatic pelvic lymph nodes be integrated into the 2018 figo staging classification of early stage cervical cancer?[J].Cancers,2020,12(6):1552

    • [30] HUANG H,FENG Y L,WAN T,et al.Effectiveness of se⁃ quential chemoradiation vs concurrent chemoradiation or radiation alone in adjuvant treatment after hysterectomy for cervical cancer:the STARS phase 3 randomized clini⁃ cal trial[J].JAMA Oncol,2021,7(3):361-369

    • [31] LEE S I,ATRI M.2018 FIGO staging system for uterine cervical cancer:enter cross⁃sectional imaging[J].Radio⁃ logy,2019,292(1):15-24

    • [32] SHU T,ZHAO D,LI B,et al.Prognostic evaluation of postoperative adjuvant therapy for operable cervical can⁃ cer:10 years’experience of National Cancer Center in China[J].Chin J Cancer Res,2017,29(6):510-520

    • [33] ÁYEN Á,JIMÉNEZ MARTÍNEZ Y,BOULAIZ H.Targeted gene delivery therapies for cervical cancer[J].Cancers,2020,12(5):1301