切换至 "中华医学电子期刊资源库"

中华介入放射学电子杂志 ›› 2025, Vol. 13 ›› Issue (01) : 54 -61. doi: 10.3877/cma.j.issn.2095-5782.2025.01.011

影像诊断

乳腺癌腋窝淋巴结转移负荷的超声组学研究
赵高芳1,(), 陈京2, 王春梅1, 吴娟3   
  1. 1. 621000 四川绵阳,四川绵阳四〇四医院医学超声科
    2. 401366 重庆,重庆市东南医院放射科南岸
    3. 610300 四川成都,成都市青白江区人民医院医学影像科
  • 收稿日期:2024-07-10 出版日期:2025-02-25
  • 通信作者: 赵高芳

Ultrasound omics study of axillary lymph node metastatic burden in breast cancer

Gaofang Zhao1,(), Jing Chen2, Chunmei Wang1, Juan Wu3   

  1. 1. Department of Medical Ultrasound,Mianyang 404 Hospital,Sichuan Mianyang 621000
    2. Department of Radiology,Chongqing Southeast Hospital,Chongqing 401366
    3. Department of Medical Imaging,Qingbaijiang District People's Hospital,Sichuan Chengdu 610300; China
  • Received:2024-07-10 Published:2025-02-25
  • Corresponding author: Gaofang Zhao
引用本文:

赵高芳, 陈京, 王春梅, 吴娟. 乳腺癌腋窝淋巴结转移负荷的超声组学研究[J/OL]. 中华介入放射学电子杂志, 2025, 13(01): 54-61.

Gaofang Zhao, Jing Chen, Chunmei Wang, Juan Wu. Ultrasound omics study of axillary lymph node metastatic burden in breast cancer[J/OL]. Chinese Journal of Interventional Radiology(Electronic Edition), 2025, 13(01): 54-61.

目的

研究乳腺癌腋窝淋巴结转移负荷的超声组学。

方法

选取2019 年11 月至2023 年1 月四川绵阳四〇四医院接受手术治疗的乳腺癌患者100 例。依据术后病理淋巴结转移情况分为高淋巴结转移负荷(HNB)组(n=43)和低淋巴结转移负荷(LNB)组(n=57)。分析2 组患者的一般临床资料、常规超声特征及影像组学特征。提取影像组学特征,使用组内相关系数(ICC)进行一致性分析。采用mRMR 和LASSO 回归算法,筛选与淋巴结转移负荷相关的影像组学特征。采用单因素Logistic 回归分析选取临床因素及常规超声特征。构建常规超声评分模型、影像组学评分模型和联合预测模型并对模型进行评价。

结果

从超声图像中最终筛选出10 个非0 的影像组学特征。影像组学评分(Rad-score)、常规超声评分及联合得分(Combine-score)在LNB 组和HNB 组间差异均有统计学意义(P<0.05)。常规超声模型分别与影像组学模型、联合预测模型的受试者工作特征曲线下面积(AUC)比较差异有统计学意义(P<0.05),而影像组学模型与联合预测模型的AUC 比较差异无统计学意义(P>0.05)。Hosmer-Lemeshow 检验表明各模型拟合均较好(P>0.05)。

结论

联合预测模型的鉴别能力优于常规超声模型和影像组学模型,提示影像组学特征联合常规超声特征用于预测乳腺癌腋窝高淋巴结转移负荷具有一定可行性。

Objective

To investigate the ultrasound radiomics of axillary lymph node metastasis burden in breast cancer.

Methods

A total of 100 patients with breast cancer who received surgical treatment in Mianyang 404 hospital, Sichuan Province from November 2019 to January 2023 were selected.According to the postoperative pathological status of lymph node metastasis, the patients were divided into high lymph node metastasis burden (HNB) group (n=43) and low lymph node metastasis burden (LNB) group (n=57).The general clinical data, conventional ultrasound features and image omics features of the 2 groups were analyzed.The image omics features were extracted and the intra-group correlation coefficient (ICC) was used for consistency analysis.The mRMR and LASSO regression algorithms were used to screen the image omics features related to lymph node metastasis burden.Univariate Logistic regression analysis was used to select clinical factors and routine ultrasound features.The conventional ultrasound scoring model, imaging omics scoring model and joint prediction model were constructed and evaluated.

Results

Finally, 10 non-zero image omics features were screened out from ultrasonic images.There were significant differences in Radscore, conventional ultrasound score and combination-score between the LNB group and HNB group (P<0.05).There were statistically significant differences in the area under the receiver working characteristic curve(AUC) between conventional ultrasound model and imaging omics model and combined prediction model respectively (P<0.05), while there was no statistically significant difference in the AUC between imaging omics model and combined prediction model (P>0.05).The Hosmer-Lemeshow test showed that all models fit well (P>0.05).

Conclusion

The differential ability of the combined prediction model is better than that of conventional ultrasound model and imaging omics mode, suggesting that it is feasible for the combined imaging omics features and conventional ultrasound features to predict high axillary lymph node metastatic burden of breast cancer.

图1 乳腺癌病灶勾画示意图 1A:原始图像;1B:分割图像。
表1 2 组患者临床病理特征比较
临床病理特征 LNB组(57例) HNB组(43例) t/χ2 P
年龄(岁,xˉ±s) 45.34±6.21 42.93±5.37 2.034 0.045
婚姻情况[例(%)] 0.107 0.743
已婚 52(91.23) 40(93.02)
未婚 5(8.77) 3(6.98)
生育情况[例(%)] 0.353 0.553
已育 48(84.21) 38(88.37)
未育 9(15.79) 5(11.63)
月经状况[例(%)] 0.103 0.748
未绝经 44(77.19) 32(74.42)
已绝经 13(22.81) 11(25.58)
肿瘤家族史[例(%)] 0.273 0.602
6(10.53) 6(13.95)
51(89.47) 37(86.05)
病理类型[例(%)] 1.427 0.490
导管癌 11(19.30) 12(27.91)
小叶癌 15(26.32) 8(18.60)
其他 31(54.39) 23(53.49)
组织学分级[例(%)] 20.067 <0.001
34(59.65) 8(18.60)
16(28.07) 16(37.21)
7(12.28) 19(44.19)
乳腺手术方式[例(%) (%)] 0.235 0.628
保乳手术 10(17.54) 6(13.95)
乳房切除术 47(82.46) 37(86.05)
ER[例(%)] 0.086 0.770
阴性 17(29.82) 14(32.56)
阳性 40(70.18) 29(67.44)
PR[例(%)] 0.043 0.835
阴性 24(42.11) 19(44.19)
阳性 33(57.89) 24(55.81)
HER2[例(%)] 45.175 <0.001
阴性 51(89.47) 10(23.26)
阳性 6(10.53) 33(76.74)
Ki-67表达[例(%)] 0.065 0.799
<14% 9(15.79) 6(13.95)
≥14% 48(84.21) 37(86.05)
分子分型[例(%)]
Luminal A 5(8.77) 4(9.30) 0.355 0.641
Luminal B(HER2-) 27(47.37) 18(41.86)
Luminal B(HER2+) 8(14.04) 8(18.60)
HER2过表达 9(15.79) 8(18.60)
三阴性 8(14.04) 5(11.63)
表2 2 组的常规超声特征比较[例(%)]
常规超声特征 LNB组(57例) HNB组(43例) χ 2 P
肿瘤大小(mm) 4.399 0.036
<20.0 20(35.09) 7(16.28)
≥20.0 37(64.91) 36(83.72)
淋巴结长径(mm) 6.229 0.013
<20.0 44(77.19) 23(53.49)
≥20.0 13(22.81) 20(46.51)
淋巴结短径(mm) 5.200 0.023
<8.6 33(57.89) 15(34.88)
≥8.6 24(42.11) 28(65.12)
长径/短径比值 0.240 0.624
<1.8 28(49.12) 19(44.19)
≥1.8 29(50.88) 24(55.81)
腋窝可疑淋巴结(枚) 65.764 <0.001
1~2 48(84.21) 1(2.33)
≥3 9(15.79) 42(97.67)
淋巴门情况 0.326 0.850
可见 36(63.16) 26(60.47)
部分可见 7(12.28) 7(16.28)
完全消失 14(24.56) 10(23.26)
所在象限 0.381 0.439
内上 8(14.04) 9(20.93)
内下 3(5.26) 2(4.65)
外上 29(50.88) 22(51.16)
外下 11(19.30) 6(13.95)
乳晕区 6(10.53) 4(9.30)
内部回声 17.156 <0.001
均匀 41(71.93) 13(30.23)
不均匀 16(28.07) 30(69.77)
后方回声 2.522 0.112
未衰减 33(57.89) 18(41.86)
衰减或消失 24(42.11) 25(58.14)
形态 2.353 0.125
规则 34(59.65) 19(44.19)
不规则 23(40.35) 24(55.81)
边界 8.855 0.003
清晰 37(64.91) 15(34.88)
不清晰 20(35.09) 28(64.12)
边缘 0.608 0.435
光整 26(45.61) 23(53.49)
不光整 31(54.39) 20(46.51)
钙化 2.756 0.097
36(63.16) 20(46.51)
21(36.84) 23(53.49)
Alder血流分级 38.048 <0.001
0~Ⅰ级 46(80.70) 8(18.60)
Ⅱ~Ⅲ级 11(19.30) 35(84.40)
BI-RADS分类 42.821 <0.001
4a类 37(64.91) 4(9.30)
4b类 11(19.30) 6(13.95)
4c类 6(10.53) 11(25.58)
5类 3(5.26) 22(51.16)
表3 常规超声评分模型多因素Logistic 回归分析
图2 影像组学特征筛选 2A:使用LASSO 通过10 次交叉验证选择惩罚系数(λ),绘制λ 与二项偏差值的关系图;2B:影像组学特征的LASSO 系数分布。
图3 经mRMR 及LASSO 回归算法选择的特征
表4 联合预测广义线性回归模型
表5 各模型得分的组间比较[MP25P75)]
表6 各模型的诊断能力
图4 模型的校准曲线及决策曲线分析 4A:校准曲线。横坐标代表结局事件(高淋巴结转移负荷)的预测概率,纵坐标代表结局事件发生的实际概率,对角线代表最佳模型预测结果,各预测模型的曲线与对角线越接近则拟合效果越好;4B:决策曲线。纵坐标表示净获益,横坐标表示阈值概率。
[1]
Tsang JYS, Tse GM.Molecular classification of breast cancer[J].Adv Anat Pathol, 2020, 27(1):27-35.
[2]
Wang X, Chen L, Sun Y, et al.Evaluation of axillary lymph node metastasis burden by preoperative ultrasound in early-stage breast cancer with needle biopsy-proven metastasis[J].Clin Transl Oncol, 2020, 22(4):468-473.
[3]
Britt KL, Cuzick J, Phillips KA.Key steps for effective breast cancer prevention[J].Nat Rev Cancer, 2020, 20(8):417-436.
[4]
Jiang M, Li CL, Luo XM, et al.Radiomics model based on shearwave elastography in the assessment of axillary lymph node status in early-stage breast cancer[J].Eur Radiol, 2022, 32(4):2313-2325.
[5]
Jena S, Bhattacharya S, Gupta A, et al.Axillary reverse mapping in patients undergoing axillary lymph node dissection: a single institution experience from India[J].Cureus, 2021, 13(7):e16462.
[6]
Patel A, Hoda RS, Hoda SA.Papillary breastsssss tumors:continuing controversies and commentary on WHO's 2019 criteria and classification[J].Int J Surg Pathol, 2022, 30(2):124-137.
[7]
Farshid G, Bilous M, Morey A, et al.ASCO/CAP 2018 breast cancer HER2 testing guidelines: summary of pertinent recommendations for practice in Australia[J].Pathology, 2019,51(4):345-348.
[8]
Niinikoski L, Hukkinen K, Leidenius MHK, et al.Axillary nodal metastatic burden in patients with breast cancer with clinically positive axillary nodes[J].Br J Surg, 2020, 107(12):1615-1624.
[9]
Goldhirsch A, Winer EP, Coates AS, et al.Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013[J].Ann Oncol, 2013, 24(9):2206-2223.
[10]
Wang XF, Zhang GC, Zuo ZC et al.A novel nomogram for the preoperative prediction of sentinel lymph node metastasis in breast cancer[J].Cancer Med, 2023, 12(6):7039-7050.
[11]
Liu H, Chen Y, Zhang Y, et al.A deep learning model integrating mammography and clinical factors facilitates the malignancy prediction of BI-RADS 4 microcalcifications in breast cancer screening[J].Eur Radiol, 2021, 31(8):5902-5912.
[12]
Yi CB, Ding ZY, Deng J, et al.Combining the ultrasound features of primary tumor and axillary lymph nodes can reduce falsenegative rate during the prediction of high axillary node burden in BI-RADS category 4 or 5 breast cancer lesions[J].Ultrasound Med Biol, 2020, 46(8):1941-1948.
[13]
Ha SM, Chang JM, Kim SY, et al.Prediction of axillary nodal burden in patients with invasive lobular carcinoma using MRI[J].Breast Cancer Res Treat, 2021, 186(2):463-473.
[14]
Yoo TK, Kang BJ, Kim SH, et al.Axillary lymph node dissection is not obligatory in breast cancer patients with biopsy-proven axillary lymph node metastasis[J].Breast Cancer Res Treat, 2020,181(2):403-409.
[15]
Fang S, Zhu J, Wang Y, et al.The value of whole-lesion histogram analysis based on field-of-view optimized and constrained undistorted single shot (FOCUS) DWI for predicting axillary lymph node status in early-stage breast cancer[J].BMC Med Imaging, 2022, 22(1):e163.
[16]
Qian X, Pei J, Zheng H, et al.Prospective assessment of breast cancer risk from multimodal multiview ultrasound images via clinically applicable deep learning[J].Nat Biomed Eng, 2021,5(6):522-532.
[17]
Liu Z, Wu K, Wu B, et al.Imaging genomics for accurate diagnosis and treatment of tumors: a cutting edge overview[J].Biomed Pharmacother, 2021, 135(6):e111173.
[18]
Zheng X, Yao Z, Huang Y, et al.Deep learning radiomics can predict axillary lymph node status in early-stage breast cancer[J].Nat Commun, 2020, 11(1):e1236.
[19]
Pesapane F, Rotili A, Agazzi GM, et al.Recent radiomics advancements in breast cancer: lessons and pitfalls for the next future[J].Curr Oncol, 2021, 28(4):2351-2372.
[20]
Dean PB.Radiomics and breast cancer management[J].Acad Radiol, 2022, 29(12):1783-1785.
[21]
Calabrese A, Santucci D, Landi R, et al.Radiomics MRI for lymph node status prediction in breast cancer patients: the state of art[J].J Cancer Res Clin Oncol, 2021, 147(6):1587-1597.
[1] 孙舒涵, 陈雅静, 宗晴晴, 栗翠英, 缪殊妹, 杨斌, 俞飞虹. 基于超声的深度学习列线图预测乳腺癌新辅助化疗后腋窝淋巴结状态的研究[J/OL]. 中华医学超声杂志(电子版), 2025, 22(02): 97-105.
[2] 王琪, 杨春凤, 宋宏萍, 许磊. 2021 至2023 年陕西省渭北农村地区基于超声的乳腺癌筛查结果及影响因素分析[J/OL]. 中华医学超声杂志(电子版), 2025, 22(02): 106-113.
[3] 聂全禹, 李明星, 刘莹, 王慧, 陈媛媛, 周鸿, 周洋. 基于超声的瘤内联合瘤周影像组学模型对乳腺癌HER2 表达状态的预测价值[J/OL]. 中华医学超声杂志(电子版), 2024, 21(12): 1103-1110.
[4] 洪玮, 叶细容, 刘枝红, 杨银凤, 吕志红. 超声影像组学联合临床病理特征预测乳腺癌新辅助化疗完全病理缓解的价值[J/OL]. 中华医学超声杂志(电子版), 2024, 21(06): 571-579.
[5] 刘伟, 牛云峰, 安杰. LINC01232 通过miR-516a-5p/BCL9 轴促进三阴性乳腺癌的恶性进展[J/OL]. 中华乳腺病杂志(电子版), 2024, 18(06): 330-338.
[6] 邱琳, 刘锦辉, 组木热提·吐尔洪, 马悦心, 冷晓玲. 超声影像组学对致密型乳腺背景中非肿块型乳腺癌的诊断价值[J/OL]. 中华乳腺病杂志(电子版), 2024, 18(06): 353-360.
[7] 何友新, 韩林荟, 杨贺庆. 新辅助化疗分别联合保乳术和改良根治术治疗Ⅱ、Ⅲ期乳腺癌的应用效果[J/OL]. 中华普外科手术学杂志(电子版), 2025, 19(02): 184-187.
[8] 尹泽新, 杨继林, 李有尧, 吴美龙, 刘利平. 肝癌微血管侵犯的术前预测研究进展[J/OL]. 中华肝脏外科手术学电子杂志, 2025, 14(01): 128-134.
[9] 陆镜明, 韩大为, 任耀星, 黄天笑, 向俊西, 张谞丰, 吕毅, 王傅民. 基于术前影像组学的肝内胆管细胞癌淋巴结转移预测的系统性分析[J/OL]. 中华肝脏外科手术学电子杂志, 2024, 13(06): 852-858.
[10] 杨艳, 邢长洋, 杨瑞静, 苏小佩, 周永刚, 金京兰. 经皮超声造影引导下导丝定位在早期乳腺癌前哨淋巴结的临床应用[J/OL]. 中华临床医师杂志(电子版), 2024, 18(11): 993-999.
[11] 梁博宇, 宋宏萍, 宋张骏, 巨艳, 王虎霞, 薛文欣. 人工智能辅助乳腺影像学检查造成的自动偏差[J/OL]. 中华临床医师杂志(电子版), 2024, 18(11): 1061-1065.
[12] 刘琦, 王守凯, 王帅, 苏雨晴, 马壮, 陈海军, 司丕蕾. 乳腺癌肿瘤内微生物组的研究进展[J/OL]. 中华临床医师杂志(电子版), 2024, 18(09): 841-845.
[13] 王誉英, 刘世伟, 王睿, 曾娅玲, 涂禧慧, 张蒲蓉. 老年乳腺癌新辅助治疗病理完全缓解的预测因素分析[J/OL]. 中华临床医师杂志(电子版), 2024, 18(07): 641-646.
[14] 王帅, 张志远, 苏雨晴, 李雯雯, 王守凯, 刘琦, 李文涛. 孟德尔随机化及其在乳腺癌研究中的应用进展[J/OL]. 中华临床医师杂志(电子版), 2024, 18(07): 671-676.
[15] 崔军威, 蔡华丽, 胡艺冰, 胡慧. 亚甲蓝联合金属定位夹及定位钩针标记在乳腺癌辅助化疗后评估腋窝转移淋巴结的临床应用价值探究[J/OL]. 中华临床医师杂志(电子版), 2024, 18(07): 625-632.
阅读次数
全文


摘要