Diagnostic value of ultrasound peritumoral viscoelasticity parameters in breast cancer: enhancing BI-RADS classification performance

Objectives

Aims to access the diagnostic performance of different regions ultrasound viscoelasticity parameters especially margin of breast cancer and to determine whether the use of margin viscoelasticity can improve its accuracy in the Breast Imaging Reporting and Data System (BI-RADS).

Materials & methods

234 benign and 90 malignant lesions were subjected to standard breast ultrasound and viscoelasticity examinations. The doctors selected region of interest (ROI) to measure viscoelasticity. ROI-1, ROI-2, and ROI-3 represent the tumor, peritumoral, and peripheral areas, respectively. The viscosity modulus (VMean, VMin, VMax, VSD) and elasticity modulus (EMean, EMin, EMax, ESD) of 3 ROIs were analyzed. The diagnostic performance of viscoelasticity of three regions was assessed by receiver operating characteristic curves (ROC). Comparison of the effectiveness of B-Mode ultrasound and viscoelastic parameters in BI-RADS diagnosis of breast cancer based on true positive (TP) and false negative (FN).

Results

The optimal viscoelasticity-related parameters for differentiating breast lesions were determined to be 2-EMax and 2-VMax, with area under the curve (AUC) values of 0.84 (0.79–0.90) and 0.85 (0.80–0.90), respectively. Using ≥ 30.4 kPa and ≥ 3.3 Pa·s as cutoff values, the BI-RADS classification was then modified. The joint model improves diagnoses of benign lesions in category 4 (69/73). In the same way, 2-EMin + 2-VMin can improve the diagnosis rate of benign lesions (227/234). Viscoelastic parameters have better diagnostic performance than viscosity and elasticity alone.

Conclusion

Ultrasound quantitative viscoelasticity parameters of breast mass especially lesion margin can show more comprehensive information. Viscoelasticity improves diagnostic accuracy of BI-RADS categories and reduces unnecessary biopsies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12880-025-02144-8.

Relevance statement

Peritumoral ultrasound viscoelasticity should also be taken seriously, providing a new idea for early detection of breast cancer. It potentially improves diagnostic accuracy of BI-RADS assessment for patients with breast lesions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12880-025-02144-8.

Introduction

Breast cancer remains the most prevalent malignancy and leading cause of cancer-related mortality among women globally, with the World Health Organization (WHO) underscoring early detection as a critical determinant for improving prognostic outcomes and survival rates [1]. The histological examination through breast biopsy serves as the diagnostic gold standard. But the inherent risks and resource intensiveness of repeated biopsies render this approach impractical for longitudinal monitoring and population-level early detection initiatives. Mammography, MRI, and PET/CT have high sensitivity in the diagnosis of breast cancer [2–6]. However, the diagnostic efficacy of mammography is significantly affected by breast density, which is particularly challenging for Asian populations with higher breast tissue density [7].In contrast, ultrasound imaging has emerged as a cornerstone modality in breast disease assessment, particularly valued for its non-invasive characteristics, operational versatility, and favorable cost-benefit profile. Studies have found that ultrasound elastography can differentiate the mechanical properties from breast lesions, thereby enhancing the diagnostic accuracy of BI-RADS [8–11]. Some studies had reported some malignant breast lesions showed typical peritumoral stiffness [12, 13], attributed to the thickening and linearization of collagen fibers commonly observed in regions of active tissue invasion and tumor angiogenesis [14–16]. Incorporating shear wave elastography (SWE) features, particularly the stiff rim sign in the display settings of conventional ultrasound, has the potential to enhance the differentiation of breast lesions [17, 18].

However, besides elasticity, biological tissues also possess viscosity. Current SWE relies on linear elastic models, treating tissues as purely elastic and thereby neglecting viscosity. While viscoelastic models have been more commonly applied in liver ultrasound [19–22], emerging evidence indicates that viscosity parameters such as shear viscosity can also help differentiate between malignant and benign breast masses [23]. But the research result is debatable because small population and uncertain identification of the mass boundary. Jia et al. emphasize that viscosity can enhance the accuracy of BI-RADS but give less attention to the viscosity of lesion margins [24]. This study aims to investigate the viscoelastic characteristics of different regions, particularly the margins, of benign and malignant breast lesions and to assess the diagnostic performance of viscoelasticity parameters.

Materials and methods

This retrospective study was approved by the institutional review board and local ethics committee (Ethics Approval No. K2024-075-01). All participants signed informed consent forms. All ultrasound examinations were performed using by Probe L11-3U of the same ultrasound system (Resona R9, Mindary) equipped with the sound touch visco imaging (STVi) technology.

Participants

Consecutive patients were recruited for this study from January 2023 and April 2024.

The enrolled patients’ breast lesions were visible on the conventional ultrasound. Then they underwent STVi technology to measure breast viscoelasticity. They have no needle biopsy or surgery before ultrasound examination and would accept the breast surgery or biopsy in our hospital in a week.

If the mass has a large diameter, the instrument’s measurement frame may not encompass the entire lesion, leading to inaccurate viscoelasticity measurements. Additionally, if the mass contains mixed echoes, assessing the elasticity of the regions becomes challenging, resulting in unreliable elasticity measurements for mixed echo lesions. A mixed echo lesion of the breast is defined as a complex abnormal area on ultrasound imaging characterized by the simultaneous presence of anechoic, iso- or hyperechoic components, which may also incorporate hyperechoic foci representing calcifications. The exclusion criteria were as follows: (a) no mass or no biopsy; (b) treated by neoadjuvant chemotherapy or systemic chemoradiotherapy; (c) mass with a large diameter; (d) refused to acknowledge the relevant investigation; (e) poor image quality. The flow chart of this study is shown in Fig. 1.

Fig. 1.

The flow chart of this study

Imaging protocol

The examinations were performed jointly by 2 doctors with over 5 years of experience in breast grayscale ultrasound and color Doppler flow imaging. Based on grayscale ultrasound images, the doctors selected the lesion with the highest BI-RADS category or, if multiple lesions had the same BI-RADS category, the one with the largest diameter. The final lesion selection was made by consensus between the 2 doctors.

The ultrasound viscosity imaging examination required participants to hold their breath. The probe was carefully positioned with minimal pressure to ensure high-quality images meeting quality control standards. When the Motion-Strain Time Box (M-STB) index in the upper left corner of the screen displays five stars, the image quality the image quality was considered high and more reliable.

The definition of all ROIs was centralized to a single senior doctor with over 5 years of experience, ensuring uniformity in the measurement protocol. The doctor first manually drawn the lesion on grayscale ultrasound images to derive the tumor’s viscoelasticity parameters, defining this area as ROI-1. Subsequently, the system’s “Shell” tool was used to automatically generate a 2-mm thick, concentric annular region surrounding ROI-1. Since tumor growth is often non-concentric, the shell itself does not fully represent the peritumoral area. Within this shell, ROI-2 is defined as a 1-mm-diameter circular sub-region. Its precise center is determined manually by the doctor using the following protocol: starting from the intersection point of the tumor’s longest axis with its boundary, a 1 mm extension is measured outward along the same axis into the shell. A 1-mm-diameter circle is then drawn with its center at this point, ensuring it is tangent to the tumor boundary. This location is selected as a site of peritumor. Furthermore, to minimize the confounding effect of tissue depth on viscoelastic measurements, ROI-3 and ROI-2 are positioned on the same horizontal plane. A 1-mm-diameter area is selected, located as far from the lesion as feasible within the valid measurement frame. Then measured the viscoelasticity of each ROIs three times, and calculated the average values to form a comprehensive database, including viscosity modulus (VMean, VMin, VMax, VSD) and elasticity modulus (EMean, EMin, EMax, ESD). To be specific, viscosity parameters (n-VMean, n-VMin, n-VMax and n-VSD) and elasticity parameters (n-EMean, n-EMin, n -EMax and n-ESD) were obtained, among which n represented ROIs including ROI-1, ROI-2 and ROI-3 (Figs. 2 and 3).

Fig. 2.

Schematic diagram of 3 regions of interest (ROIs). The schematic illustrates the region definitions for ultrasonic viscoelastic measurement. The gray box denotes the parameter sampling area. The blue area represents the entire tumor, defined as ROI-1, with its boundary manually delineated by the doctor (black line). Surrounding ROI-1, the system automatically generates a 2-mm-thick “Shell” region (orange). The red dashed line indicates the longest axis of the tumor. Along this axis, ROI-2 (purple dot) is defined within the annular region at a point 1 mm from the tumor boundary. ROI-3 (dark blue dot), serving as a reference region, is placed on the same horizontal line at a position distant from the lesion

Fig. 3.

Regions of interest (ROIs) of begin lesion. This is a patient whose pathological biopsy was fibroadenoma. The red box is the elasticity value and the blue box is the viscosity value. (a) is ROI-1. (b) No.1 is ROI-2 and No.2 is ROI-3

The reference standards

All participants underwent surgery or core needle biopsy for the target lesion after ultrasound imaging. Histological classification follows WHO breast tumor classification, and molecular classification follows Chinese Society of Clinical Oncology (CSCO) Breast Cancer guidelines.

Statistical analysis

Statistical analysis was performed using SPSS software (version 26.0). The Kolmogorov-Smirnov test was performed to assess the normality of the data. Data that followed a normal distribution were presented as mean ± standard deviation (SD), while data that did not follow a normal distribution were described as median (interquartile range). The Mann-Whitney U-test, independent t-test and Chi-square test were used for statistical comparison. A binary logistic regression model was constructed to combine multiple parameters. With histological results as reference standard, receiver operating characteristic (ROC) curve was used to calculate the area under the curve (AUC), sensitivity and specificity of all parameters. The DeLong test was used to compare the AUCs of different parameters. A level of P < 0.05 was considered statistically significant.

Result

Participant characteristics

A total of 324 women were eligible to accept breast ultrasound viscoelasticity examinations. 324 index breast lesions with 234 (72.2%) benign and 90 (27.8%) malignant lesions were enrolled in this study. The Tables S1, S2 and S3 are the patients’ basic information, pathological classification, and molecular classification.

Diagnostic performance of BI-RADS

B-mode BI-RADS categories are summarized in Table 1. For benign lesions, 161 were category 3, 73 were category 4. For malignant lesions, the number of lesions classified as category 3, 4a, 4b, 4c, and 5 was1, 11, 21, 47and 10, respectively. If we set BI-RADS category 4a as the optimal cut-off, the sensitivity, specificity, and accuracy for diagnosing malignant lesions were 98.8%, 68.8%, and 77.2%, respectively.

Table 1.

B-mode BI-RADS categories

BI-RADS category	N	Benign	Malignant
3	162	161	1
4a	74	63	11
4b	31	10	21
4c	47		47
5	10		10
N	324	234	90

Diagnostic performances of viscoelasticity parameters

The viscoelasticity parameters of lesions were significantly higher in the malignant group than in the benign group (P < 0.05) (Table S1). The ROC curves of the quantitative parameters are shown in Fig. 4. As shown in Fig. 4, the maximum and minimum values demonstrated superior diagnostic performance, evidenced by higher AUCs than other parameters, and were thus selected as the core indicators. And the sensitivity, specificity, and accuracy of the parameters with better AUC are shown in Table 2. According to ROC analysis, the optimal cut-off was 30.4 kPa for 2-EMax, indicating a sensitivity of 71.1% and specificity of 87.2%. For 2-VMax, the optimal cut-off was 3.3 Pa·s, with a sensitivity of 72.2% and specificity of 85.9%. For 2-EMin and 2-VMin, The sensitivity of 91.5% and 93.2%, respectively.

Fig. 4.

The Receiver Operating Characteristic (ROCs) of Viscoelasticity Parameters. (a) is the ROCs of 3 regions elasticity parameter. (b) is the ROCs of 3 regions viscosity parameter

Table 2.

The sensitivity, specificity and accuracy of viscoelasticity parameter

	Cut off	Sensitivity	Specificity	AUC	P
Elasticity(Kpa)
2-EMean	26.8	0.656	0.889	0.82(0.76–0.88)	<0.001
2-EMax	30.4	0.711	0.872	0.84(0.79–0.90)	<0.001
2-EMin	24.7	0.6	0.915	0.78(0.71–0.84)	<0.001
Viscosity(Pa.s)
2-VMean	2.3	0.7	0.833	0.84(0.78–0.89)	<0.001
2-VMax	3.3	0.722	0.859	0.85(0.80–0.90)	<0.001
2-VMin	2.3	0.567	0.932	0.75(0.68–0.82)	<0.001

AUC: area under the curve

Diagnostic performances of ROI-2 viscoelastic parameter joint model

Joint models incorporating ROI-2 viscosity and elasticity parameters were established using binary logistic regression. The ROC curves are presented in Fig. 5. The 2-EMax + 2-VMax joint model yielded an AUC of 0.86, with a sensitivity of 72.2% and a specificity of 88.9%. The 2-EMin + 2-VMin model yielded an AUC of 0.82, with a sensitivity of 65.6% and a specificity of 91.5%. We compared the diagnostic performance of the two joint models using the Delong test. The result indicated significant difference in AUC between the 2-EMax + 2-VMax model and the 2-EMin + 2-VMin model (Z = 2.592, P = 0.01).

Fig. 5.

The Receiver Operating Characteristic (ROCs) of Viscoelasticity Parameters Joint Model. Comparison by Delong test: 2-EMax + 2-VMax model vs. 2-EMin + 2-VMin model, P = 0.01

Diagnostic performances of BI-RADS with ROI-2 viscoelasticity parameters

In differentiating benign and malignant lesions among BI-RADS 4a and 4b categories, the combined parameter 2-EMax + 2-VMax demonstrated high sensitivity. It correctly identified 64 benign and 21 malignant lesions, including one case misclassified as BI-RADS 3. While standard BI-RADS management for all 105 patients yields a malignancy detection rate of 30.5% (32/105), the 2-EMax + 2-VMax model would have recommended biopsy for only 30 patients, achieving a superior detection rate of 70% (21/30). Furthermore, the 2-EMin + 2-VMin model showed high specificity by correctly identifying 227 benign lesions. Its application would have reduced necessary biopsies to only 16 patients, allowing 84.8% of the cohort to avoid invasive procedures (Table 3).

Table 3.

Diagnostic performances of BI-RADS with ROI-2 viscoelasticity parameters

BI-RADS category	2-EMax		2-EMin		2-VMax		2-VMin			2-EMax + 2-VMax		2-EMin + 2-VMin
	Benign	Malignant	Benign	Malignant	Benign	Malignant	Benign	Malignant		Benign	Malignant	Benign	Malignant
3	145/161	0/1	150/161	0/1	141/161	0/1	154/161		0/1	148/161	1/1	158/161	0/1
4a	50/63	8/11	55/63	8/11	51/63	9/11	55/63		7/11	55/63	8/11	60/63	6/11
4b	9/10	13/21	9/10	8/21	9/10	14/21	9/10		10/21	9/10	13/21	9/10	6/21
N	204/234	21/33	214/234	16/33	201/234	23/33	218/234		17/33	212/234	22/33	227/234	12/33

Discussion

As demonstrated in Table 1, subcategories 4a and 4b of the fifth edition of the BI-RADS ultrasound lexicon demonstrate reduced diagnostic performance compared to subcategory 4c. If we set BI-RADS category 4a as the optimal cut-off, the sensitivity, specificity, and accuracy for diagnosing malignant lesions were 98.8%, 68.8%, and 77.2%, respectively. BI-RADS category 3 is considered likely benign, with short-term follow-up recommended, while category 4 indicates suspicion for malignancy, with a biopsy recommended. Some benign lesions were misdiagnosed as malignant using B-mode BI-RADS, resulting in a low accuracy rate. These misdiagnosed cases lead to an increase in unnecessary biopsies.

BI-RADS subdivides category 4 lesions into categories 4a, 4b, and 4c for ultrasound findings, with the positive predictive value (PPV) of 2% -10%, 10% -50%, 50% -95%, respectively. The accuracy of subcategory 4c in diagnosing malignancy is 100% and the accuracy of subcategory 4a and 4b is 14.9% and 67.7% (Table 1). This result was in accordance with the research of Zou et al. [35]. Most benign lesions are misdiagnosed as malignant, leading to unnecessary biopsies. In the present study, we use ROI-2 viscoelasticity parameters to diagnose breast cancer and compare the results with B-mode BI-RADS. When 2-EMin and 2-VMin were used individually for diagnosis, the accuracy of benign diagnosis showed minimal differences (214/234, 218/234). The 2-EMin + 2-VMin diagnosis rate of benign lesions increased compared to each individual parameter (227/234). The viscoelastic combination model exhibits better diagnostic performance than models using only viscosity or elasticity alone, as shown in Figs. 4 and 5. But it showed a low malignant diagnosis rate (12/30) in category 4, which increased the probability of missed malignant diagnoses. For lesions diagnosed as category 4a, whether benign or malignant can only be determined through biopsy. Most category 4a nodules are benign, and performing a biopsy in such cases may lead to unnecessary patient injury and anxiety. Therefore, ROI-2 Min value is more suitable for preliminary screening of lesions in categories 4a that are vague and have a low possibility of malignancy. 2-EMin + 2-VMin can improve the accuracy of benign diagnosis. Of the 63 benign lesions in category 4a, 60 were diagnosed as benign by 2-EMin + 2-VMin. For doctors, this improves the confidence in diagnosing benign and reduces unnecessary punctures.

However, ROI-2 Max value is more competent than Min in diagnosing benign and malignant tumors. When 2-EMax + 2-VMax was used for diagnosis, although the diagnostic rate of malignant lesions remained unchanged compared with each individual parameter. In addition, the number of diagnoses of benign lesions increased in category 4 (69/73). If biopsies had been performed based on viscoelastic diagnosis for BI-RADS 4a and 4b lesions, only 28.6% (30/105) of the lesions would have required biopsy, with a resulting malignancy detection rate of 70% (21/30). This strategy effectively minimizes unwarranted biopsies. However, there are 11 false-negative lesions. Small diameter (< 1 cm) and moderate to high differentiation may contribute to false negatives. Some malignancies may have benign characteristics when they are small.

Elastography has become a valuable tool in breast imaging and tumor assessment, now serving as an integrated part of the breast examination. Real biological tissue exhibits both viscous and elastic properties. The viscoelasticity of malignant lesions is higher than that of benign lesions (Table S1). According to ROC analysis, all viscoelastic parameters derived from ROI-2, regardless of being elasticity or viscosity metrics, yielded consistently superior diagnostic performance over those from ROI-1 and ROI-3. This finding can be attributed to the unique pathological changes in the peritumoral extracellular matrix (ECM). ECM in cancer undergoes alterations in biochemical properties, biomechanical characteristics, structural features, and topological morphology [25]. In breast cancer, increased deposition of type I, III, and V fibrous collagen promotes collagen cross-linking. The function of lysyl hydroxylase 2 (LH2) is to remodel the cross-linking pattern, shifting from the lysyl oxidase (LOX) -derived collagen cross-links (LCCs) to hydroxylysine aldehyde-derived cross-links (HLCCs), which are more abundant in load-bearing tissues. This transition promotes structural stabilization and stiffening of the collagen matrix, thereby conferring greater resistance to external forces and heightened elasticity [26, 27]. Through its collaboration with proteoglycans (PGs), hyaluronic acid (HA) draws in water, creating an osmotic pressure that elevates fluid viscosity and provides mechanical damping [28]. During cancer progression, HA levels rise significantly, thereby enhancing the viscous properties of tumors [29].

Changes in the physical properties of the ECM, such as viscoelasticity, influence the prognosis of breast cancer. Increased ECM stiffness as a biomechanical cue that activates ERK and PI3K signaling pathways in tumor cells, thereby driving cancer progression and malignant transformation [26]. Matrix stiffness can trigger epithelial-mesenchymal transition (EMT) in breast cancer through mechanotransduction. This process leads to loss of cellular polarity and enhanced invasiveness, ultimately compromising tissue integrity [27, 30]. ECM remodeling directly alters pore size, thereby physically impedes cell migration. In response, cells secrete proteases to degrade ECM components and apply mechanical forces to disrupt fibers, thereby creating migration paths. This entire process subsequently modifies the viscoelastic properties of the matrix [31, 32].

A key regulator of these physical properties is matrix metalloproteinase 13 (MMP-13). It’s a calcium-dependent endopeptidase present in the ECM. MMPs cleave type I and III collagen in the ECM. Inna et al. reported that intact, non-degraded ECM exhibits higher viscosity, whereas MMP-13–mediated degradation reduces ECM viscosity [33]. Although MMP-13 initially softens the ECM, the process releases growth factors that upregulate LOXs, promoting the deposition and cross-linking of type I and III collagen and ultimately transforming the ECM into a stiffened, fibrotic tumor matrix. Beyond its direct biomechanical impact on the ECM, MMP-13 also regulates EMT through mechanotransduction and reduces intercellular adhesion, thereby modifies the physical properties of tumors. Notably, the role of ECM viscoelasticity in promoting tumor progression is not limited to breast cancer. Studies in hepatocellular carcinoma have shown that advanced glycation end products (AGEs)–mediated structural changes enhance ECM viscoelasticity, which in turn promotes tumor progression in vivo, independent of stiffness [34]. This mechanism is also relevant to breast cancer: the process of ECM remodeling—including increased collagen cross-linking and altered matrix stiffness—confers distinct viscoelastic characteristics to tumor tissue. These alterations provide a critical biological foundation and diagnostic rationale for ultrasound viscoelastic imaging. These biomechanical changes are often manifested as specific imaging features. Previous studies have identified the “stiff rim sign”, increased peritumoral stiffness on elastography, as a valuable diagnostic indicator [17, 18]. The presence of the “stiff rim sign” is closely linked to specific changes in the tumor ECM. However, this feature is qualitative and subject to interpreter variability on strain elastography. While SWE can quantify this sign, it typically relies on concentric peritumoral measurements. Moreover, the key biological events that determine invasiveness occur in a distinct micro-region within the peritumoral area - the “invasive front”. As shown by Koorman et al., the collective invasion of ductal carcinoma is enhanced by increased collagen bundling and alignment, which are locally induced by lysyl oxidase-like 3 (LOXL3) at the invasion front, rather than driven by uniform stromal sclerosis. Conversely, systemic sclerosis or a homogeneous increase in matrix stiffness inhibits invasion [35]. Mechanistic studies by Schwager et al. revealed that tumor cell-derived microvesicles (MVs) effectively activate cancer-associated fibroblasts (CAFs) only on substrates mimicking the high stiffness of the peritumoral stroma [36]. This indicates that the invasion front constitutes a distinct and biomechanically heterogeneous region. This concept is further supported by the synthetic ECM model of Hiraki et al., which demonstrated that the physical heterogeneity of the ECM - defined by stiffness and fiber density - directly regulates tumor cell invasion patterns [37]. Therefore, traditional holistic measurements (such as ROI-1 or the “stiff rim” sign) possess inherent limitations. By averaging signals, they may obscure the critical signature of the invasion front and potentially conflate inhibitory overall stiffening with promotive local stiffening. In contrast, our point-measurement strategy (ROI-2) overcomes this limitation. It enables targeted sampling of the most invasion-prone “hotspot”, thereby directly capturing the specific viscoelastic signature reflecting the aforementioned active, local ECM remodeling at the invasion front. This explains why the diagnostic performance of ROI-2 significantly surpasses that of ROI-1: we are not measuring a homogeneous “ring”, but rather the mechanical “engine” itself that drives malignant progression.

There were some limitations in this study. First, while targeting a single point (ROI-2) at the invasive front along the longest axis provides a focused and reproducible measurement, we acknowledge that the invasive front is not a singular point. Biologically, circumferential heterogeneity may exist. A more comprehensive approach in future studies could involve sampling multiple points around the tumor periphery (e.g., along both the long and short axes) to capture a fuller picture of the peritumoral biomechanical landscape. This represents a valuable direction for our subsequent research. Secondly, mass with larger diameters were excluded from this study. This part also needs to be considered in further research. Finally, the lesion boundary is marked by the doctor, which is highly subjective, which could have affected our results.

In conclusion, quantitative viscoelastic parameters of breast masses, particularly at the lesion margins, can enhance the accuracy of BI-RADS diagnosis, especially for lesions with benign lesions in category 4. This can improve diagnostic confidence and reduce unnecessary biopsies. Future studies will further verify this conclusion and explore the correlation between viscoelastic properties and the biological behavior or prognosis of malignant breast lesions.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

WHO

The World Health Organization

BI-RADS

The Breast Imaging Reporting and Data System

SWE

Shear wave elastography

STVi

The sound touch visco imaging

ROI

Region of interest

SD

Standard deviation

ROC

Receiver operating characteristic

AUC

Area under the curve

M-STB

Motion-Strain Time Box

CSCO

Chinese Society of Clinical Oncology

ECM

Extracellular matrix

LH2

Lysyl hydroxylase 2

LCCs

Lysald-derived collagen cross-links

HLCCs

Hylald-derived cross-links

HA

Hyaluronic acid

PGs

Proteoglycans

EMT

Epithelial- mesenchymal transition

MMPs

Matrix metalloproteinase 13

LOXs

Lysyl oxidases

AGEs

Advanced glycation end products

LOXL3

Lysyl oxidase-like 3

MVs

Microvesicles

CAFs

Fibroblasts

PPV

Positive predictive value

Author contributions

JX: Data curation, Formal Analysis, Methodology, Visualization, Writing – original draft. JS: Conceptualization, Data curation, Formal Analysis, Writing – original draft. YH: Conceptualization, Methodology, Supervision, Validation, Writing-review & editing. JC: Conceptualization, Supervision, Writing- review & editing. GH: Data curation. WZ: Data curation. MC: Conceptualization, Formal Analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Writing - review & editing.

Funding

This study was jointly supported by Medical-Engineering Interdisciplinary Program of Donghua University & Shanghai Tongren Hospital (Grant No. 2023DHYGJC-YBA02) and Shanghai Tongren Hospital Strategic Discipline for Nurturing (Grant No. tx2023xk18).

Data availability

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Declarations

Ethics approval and consent to participate

This study strictly adheres to the ethical principles of the Declaration of Helsinki (2013 revision) and has been approved by the Ethics Committee of Shanghai Tongren Hospital (Ethics Approval No. K2024-075-01, Approval Date 2024-12-09).

Consent for publication

Not applicable.

Informed consent

Written informed consent was obtained from all subjects (patients) in this study.

Competing interests

The authors declare no competing interests.