Value analysis of the prediction model based on multimodal MRI characteristics for the differential diagnosis of benign and malignant BI-RADS 4 types of breast tumors

doi:10.3760/cma.j.cn371439-20231109-00115

Abstract

Abstract:

Objective To explore the value of the prediction model based on multimodal MRI characteristics for the differential diagnosis of benign and malignant breast tumors of breast imaging reporting and date system （BI-RADS） 4 types. Methods A total of 204 patients with BI-RADS 4 types of breast tumors confirmed by contrast-enhanced MRI in 3201 Hospital from January 2018 to January 2023 were retrospectively included， and were divided into the malignant group （124 cases） and the benign group （80 cases） according to surgical histopathology. Clinical and MRI imaging characteristics of the two groups were compared. Multivariate logistic regression analysis was performed for the differential diagnosis indexes of benign and malignant breast tumors of BI-RADS 4 types. A prediction model for differential diagnosis of benign and malignant BI-RADS 4 types of breast tumors was constructed. Receiver operator characteristic （ROC） curve was used to compare the differential diagnostic value of each index for benign and malignant BI-RADS 4 types of breast tumors. Results There were statistically significant differences in age （t=7.78， P<0.001）， internal enhancement type （χ²=14.50， P=0.002）， apparent diffusion coefficient （t=-6.77， P<0.001） longitudinal relaxation time （T₁） value （t=-6.15， P<0.001）， and longitudinal relaxation rate （R₁） value （t=7.02， P<0.001） between the malignant and benign groups. Multivariate analysis showed that age （OR=1.16， 95%CI： 1.07-1.25， P<0.001）， internal reinforcement type （uneven： OR=8.08， 95%CI： 2.21-29.51， P=0.002）， apparent diffusion coefficient （OR=0.01， 95%CI： 0.00-0.05， P<0.001）， T₁ value （OR=0.99， 95%CI： 0.99-1.00， P<0.001）， and R₁ value （OR=1 043.50， 95%CI： 46.48-2 3426.36， P<0.001） were all independent factors influencing the differential diagnosis of benign and malignant BI-RADS 4 types of breast tumors. According to the results of multivariate analysis， a logistic regression model for differential diagnosis of benign and malignant BI-RADS 4 types of breast tumors was constructed. logit （P）=0.05+0.15×age+2.09×internal enhancement type-5.21×apparent diffusion coefficient-0.01×T₁ value+6.95×R₁ value. ROC curve analysis showed that age， internal reinforcement type， apparent diffusion coefficient， T₁ value， R₁ value， and logistic regression model P-value were used for differential diagnosis of benign and malignant breast tumors， the Jordan indexes were 40.60%， 39.68%， 49.44%， 38.23%， 43.27%， and 75.70%， respectively. The areas under the ROC curve were 0.757， 0.647， 0.718， 0.724， 0.757， and 0.924， respectively. Conclusion Multimodal magnetic resonance indexes， including internal reinforcement type， apparent diffusion coefficient， T₁ value and R₁ value， can be used for the differential diagnosis of benign and malignant BI-RADS 4 types of breast tumors. The differential diagnostic model based on the above indexes has good differential diagnostic efficacy for benign and malignant BI-RADS 4 types of breast tumors.

Key words: Breast neoplasms, Magnetic resonance imaging, Breast imaging reporting and data system, Diagnosis， differential

Zhu Bin, Wan Tao, Xu Hua, Jia Hao, Chen Shixin. Value analysis of the prediction model based on multimodal MRI characteristics for the differential diagnosis of benign and malignant BI-RADS 4 types of breast tumors[J]. Journal of International Oncology, 2024, 51(11): 678-683.

Figures/Tables 5

References 20

[1]	段帅, 郭晨明, 李慧芳, 等. 1990—2019年中国女性乳腺癌发病死亡趋势及年龄-时期-队列模型分析[J]. 国际肿瘤学杂志, 2022, 49(10): 586-591. DOI: 10.3760/cma.j.cn371439-20220726-00116.
[2]	宋德领, 张玉姣, 朱月香, 等. 动态对比增强MRI影像组学特征鉴别乳腺良恶性病变的价值[J]. 实用放射学杂志, 2021, 37(11): 1810-1813, 1846. DOI: 10.3969/j.issn.1002-1671.2021.11.015.
[3]	Zhou J, Zhang Y, Chang KT, et al. Diagnosis of benign and malignant breast lesions on DCE-MRI by using radiomics and deep learning with consideration of peritumor tissue[J]. J Magn Reson Imaging, 2020, 51(3): 798-809. DOI: 10.1002/jmri.26981. pmid: 31675151
[4]	Granzier RWY, Verbakel NMH, Ibrahim A, et al. MRI-based radiomics in breast cancer: feature robustness with respect to inter-observer segmentation variability[J]. Sci Rep, 2020, 10(1): 14163. DOI: 10.1038/s41598-020-70940-z. pmid: 32843663
[5]	徐兵河, 江泽飞, 胡夕春. 中国晚期乳腺癌临床诊疗专家共识2016[J]. 中华医学杂志, 2016, 96(22): 1719-1727. DOI: 10.3760/cma.j.issn.0376-2491.2016.22.002
[6]	中国抗癌协会乳腺癌专业委员会. 中国抗癌协会乳腺癌诊治指南与规范(2021年版)[J]. 中国癌症杂志, 2021, 31(10): 954-1040. DOI: 10.19401/j.cnki.1007-3639.2021.10.013.
[7]	Zhao YF, Chen Z, Zhang Y, et al. Diagnosis of breast cancer using radiomics models built based on dynamic contrast enhanced MRI combined with mammography[J]. Front Oncol, 2021, 11: 774248. DOI: 10.3389/fonc.2021.774248.
[8]	张书海, 王小雷, 朱芸, 等. 多模态影像组学在鉴别乳腺良恶性病变中的应用研究[J]. 临床放射学杂志, 2021, 40(11): 2098-2104.
[9]	Huang Y, Wei L, Hu Y, et al. Multi-parametric MRI-based radiomics models for predicting molecular subtype and androgen receptor expression in breast cancer[J]. Front Oncol, 2021, 11: 706733. DOI: 10.3389/fonc.2021.706733.
[10]	Daimiel Naranjo I, Gibbs P, Reiner JS, et al. Radiomics and machine learning with multiparametric breast MRI for improved diagnostic accuracy in breast cancer diagnosis[J]. Diagnostics (Basel), 2021, 11(6): 919. DOI: 10.3390/diagnostics11060919.
[11]	Lin F, Wang Z, Zhang K, et al. Contrast-enhanced spectral mammography-based radiomics nomogram for identifying benign and malignant breast lesions of Sub-1 cm[J]. Front Oncol, 2020, 10: 573630. DOI: 10.3389/fonc.2020.573630.
[12]	Chen HL, Zhou JQ, Chen Q, et al. Comparison of the sensitivity of mammography, ultrasound, magnetic resonance imaging and combinations of these imaging modalities for the detection of small (≤2 cm) breast cancer[J]. Medicine (Baltimore), 2021, 100(26): e26531. DOI: 10.1097/MD.0000000000026531.
[13]	Niu S, Jiang W, Zhao N, et al. Intra- and peritumoral radiomics on assessment of breast cancer molecular subtypes based on mammography and MRI[J]. J Cancer Res Clin Oncol, 2022, 148(1): 97-106. DOI: 10.1007/s00432-021-03822-0.
[14]	Gillies RJ, Schabath MB. Radiomics improves cancer screening and early detection[J]. Cancer Epidemiol Biomarkers Prev, 2020, 29(12): 2556-2567. DOI: 10.1158/1055-9965.EPI-20-0075.
[15]	Jiang M, Li CL, Luo XM, et al. An MRI-based radiomics approach to improve breast cancer histological grading[J]. Acad Radiol, 2023, 30(9): 1794-1804. DOI: 10.1016/j.acra.2022.12.014. pmid: 36609032
[16]	Li X, Jiang N, Zhang C, et al. Value of conventional magnetic resonance imaging texture analysis in the differential diagnosis of benign and borderline/malignant phyllodes tumors of the breast[J]. Cancer Imaging, 2021, 21(1): 29. DOI: 10.1186/s40644-021-00398-3. pmid: 33712070
[17]	Conti A, Duggento A, Indovina I, et al. Radiomics in breast cancer classification and prediction[J]. Semin Cancer Biol, 2021, 72: 238-250. DOI: 10.1016/j.semcancer.2020.04.002. pmid: 32371013
[18]	Langman EL, Kuzmiak CM, Brader R, et al. Breast cancer in young women: imaging and clinical course[J]. Breast J, 2021, 27(8): 657-663. DOI: 10.1111/tbj.14261.
[19]	尤超, 彭卫军, 顾雅佳, 等. 乳腺X线摄影及MRI结合临床特征对乳腺高危病变诊断价值[J]. 中华放射学杂志, 2020, 54(3): 203-208. DOI: 10.3760/cma.j.issn.1005-1201.2020.03.006.
[20]	Tagliafico AS, Piana M, Schenone D, et al. Overview of radiomics in breast cancer diagnosis and prognostication[J]. Breast, 2020, 49: 74-80. DOI: 10.1016/j.breast.2019.10.018. pmid: 31739125

指标	恶性组（n=124）	良性组（n=80）	t/χ²值	P值
年龄（岁）	51.54±7.50	43.81±5.93	7.78	<0.001
病变最大径（mm）	17.19±3.37	17.02±3.60	0.34	0.732
肿块可触及	103（83.06）	61（76.25）	1.43	0.231
病灶形态不规则	48（38.71）	25（31.25）	1.18	0.278
病灶边缘
光整	8（6.45）	10（12.50）
不规则	102（82.26）	64（80.00）	2.76	0.252
星芒状	14（11.29）	6（7.50）
BI-RADS类型
4A	22（17.74）	17（21.25）	0.39	0.534
4B~C	102（82.26）	63（78.75）	0.39	0.534
内部强化类型
均匀	7（5.65）	14（17.50）	14.50	0.002
不均匀	85（68.55）	35（43.75）
环形强化	20（16.13）	18（22.50）
低信号分隔	12（9.67）	13（16.25）
时间-强度曲线类型
流入	22（17.74）	16（20.00）	2.54	0.281
平台	48（38.71）	38（47.50）
流出	54（43.55）	26（32.50）
表观弥散系数（×10^-3 mm²/s）	0.83±0.15	1.05±0.31	-6.77	<0.001
T₁值（ms）	1 250.81±137.13	1 372.87±140.14	-6.15	<0.001
T₂值（ms）	92.23±8.05	94.49±10.72	-1.72	0.088
PD值（pu）	69.79±10.47	72.82±13.95	-1.77	0.079
R₁ 值（s^-1）	0.88±0.15	0.74±0.12	7.02	<0.001
R₂值（s^-1）	13.04±3.26	12.82±2.54	0.51	0.610

影响因素	β值	SE值	Wald χ²值	OR值	95%CI	P值
年龄	0.15	0.04	14.15	1.16	1.07~1.25	<0.001
内部强化类型
均匀				Ref
不均匀	2.09	0.66	10.01	8.08	2.21~29.51	0.002
环形强化	1.54	1.00	2.38	4.66	0.66~32.99	0.122
低信号分隔	1.40	0.85	2.70	4.05	0.76~21.44	0.100
表观弥散系数	-5.21	1.08	23.42	0.01	0.00~0.05	<0.001
T₁值	-0.01	0.01	16.09	0.99	0.99~1.00	<0.001
R₁值	6.95	1.59	19.17	1 043.50	46.48~23 426.36	<0.001
常量	0.05	3.01	0.01	1.05		0.986

项目	最佳截断值	ROC曲线下面积	敏感性（%）	特异性（%）	约登指数（%）
年龄	51.85	0.757	44.35	96.25	40.60
内部强化类型	1.50	0.647	84.68	55.00	39.68
表观弥散系数	1.03	0.718	91.94	57.50	49.44
T₁值	1 265.36	0.724	53.23	85.00	38.23
R₁值	0.83	0.757	64.52	78.75	43.27
logistic回归模型P值	0.63	0.924	87.90	87.50	75.70