Added Value of a New Strain Elastography Technique in Conventional Ultrasound for the Diagnosis of Breast Masses: A Prospective Multicenter Study

Objective This study aimed to explore the value of elasticity score (ES) and strain ratio (SR) combined with conventional ultrasound in distinguishing benign and malignant breast masses and reducing biopsy of BI-RADS (Breast Imaging Reporting and Data System) 4a lesions. Methods This prospective, multicenter study included 910 patients from nine different hospitals. The acquisition and analysis of conventional ultrasound and strain elastography (SE) were obtained by radiologists with more than 5 years of experience in breast ultrasound imaging. The diagnostic sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under curve (AUC) of conventional ultrasound alone and combined tests with ES and/or SR were calculated and compared. Results The optimal cutoff value of SR for differentiating benign from malignant masses was 2.27, with a sensitivity of 60.2% and a specificity of 84.8%. When combined with ES and SR, the AUC of the new BI-RADS classification increased from 0.733 to 0.824 (p < 0.001); the specificity increased from 48.1% to 68.5% (p < 0.001) without a decrease in the sensitivity (98.5% vs. 96.4%, p = 0.065); and the PPV increased from 52.2% to 63.7% (p < 0.001) without a loss in the NPV (98.2% vs. 97.1%, p = 0.327). All three combinations of conventional ultrasound, ES, and SR could reduce the biopsy rate of category 4a lesions without reducing the malignant rate of biopsy (from 100% to 68.3%, 34.9%, and 50.4%, respectively, all p < 0.001). Conclusions SE can be used as a useful and non-invasive additional method to improve the diagnostic performance of conventional ultrasound by increasing AUC and specificity and reducing the unnecessary biopsy of BI-RADS 4a lesions.


INTRODUCTION
The morbidity of breast cancer is the highest in the world, and the mortality ranks fifth among all cancers but first in female cancers (1). Early detection and timely diagnosis of breast cancer are closely related to the prognosis of patients. Ultrasound is widely used in the examination of patients with breast abnormalities. However, the lack of specificity of B-mode ultrasound in the diagnosis of breast masses leads to unnecessary biopsy (2), which leads to negative effects such as pain, anxiety, and complications (3).
Strain elastography (SE) is easily performed and provides elastic images with a high spatial resolution by evaluating tissue deformation (4). In general, malignant breast tissue is harder than normal breast tissue and produces less strain (5). Differentiating benign and malignant breast masses and upgrading or downgrading the Breast Imaging Reporting and Data System (BI-RADS) classification to avoid unnecessary biopsy are clinical indications for elastography according to the WFUMB guidelines and recommendations for clinical use of ultrasound elastography to breast (6). Ultrasound elastography technique may improve the specificity of B-mode ultrasound in the differential diagnosis of breast masses by measuring tissue stiffness (2,7), even for breast masses smaller than 1 cm in diameter (8). Elasticity assessment has been incorporated into the fifth edition of BI-RADS lexicon to further describe the characteristics of breast masses (9). The combination of conventional ultrasound and SE can reduce unnecessary biopsy of breast masses by down-staging the BI-RADS classification (10). SE was strongly recommended as a supplementary diagnostic tool for conventional ultrasound by the latest EFSUMB (European Federation of Societies for Ultrasound in Medicine and Biology) guidelines and recommendations for the clinical practice of elastography for non-hepatic applications released in 2018 (11). Three diagnostic methods of SE including elasticity score (ES), strain ratio (SR), and strain size ratio (EI/B ratio) were mainly used to classify breast lesions in clinic (6).
Most previous studies explored the value of SE in breast masses using Hitachi ultrasound equipment (5,12,13). However, the SE in different brands of ultrasound systems has different reference standards for clinical use. Recently, a new SE technique, with the function of measuring ES and SR, has been equipped in Samsung ultrasound systems. At present, only one single study has explored the diagnostic performance of SE of this system in differentiating benign and malignant breast masses (14). More studies are needed to explore the added value of ES and SR in the differential diagnosis of breast masses.
The prospective multicenter study aimed to determine the cutoff value of SR and to explore the value of ES and SR in combination with conventional ultrasound in distinguishing benign and malignant breast masses and reducing biopsy of BI-RADS 4a lesions.

MATERIALS AND METHODS
This prospective multicenter study enrolled patients from nine institutions in different regions of China between April 2019 and November 2020. It was approved by the ethics committee of Tongji Medical College of Huazhong University of Science and Technology and registered in ClinicalTrials.gov (NCT 03887598). The informed consent of all participants was obtained in this study.

Participants
The inclusion criteria of participants were as follows: (i) patients had definite pathological results after ultrasound examination, and (ii) patients were at least 18 years old. The exclusion criteria were as follows: (i) patients who received radiotherapy or chemotherapy before the examination; (ii) patients who did not have reliable SE images or SR analyses; and (iii) patients who were lactating or pregnant.

Image Acquisition: B-Mode Ultrasound and SE
Conventional ultrasound imaging and SE technique were performed with the Samsung RS80A ultrasound system (Samsung Madison Co., Ltd., Seoul, South Korea) in all patients. The acquisition of ultrasound images and the analysis of SE images were performed by nine radiologists with more than 5 years of experience in breast ultrasound imaging. The standard data acquisition process was established, and all operators received rigorous training before the enrollment of patients. The study was conducted only after five qualified test cases were uploaded from every single center and checked by the principle investigator.
Breast B-mode ultrasound was performed in the supine position in all patients using a 3-12 MHz linear transducer. Bmode videos of the lesions were documented in both the long axis and short axis. SE imaging was performed using the same 3-12 MHz linear transducer based on WFUMB guidelines (6). The SE images were obtained by manually applying slight vibration with the probe perpendicular to the skin under the guidance of the quality indicator. After the elastic image was stabilized, SR and ES were acquired on a representative static image by the same operator. Strain A was obtained by placing the ROI in the target mass, and strain B was obtained by placing the ROI in the subcutaneous fibroglandular tissue at the same depth as the mass (Figure 1). The SR (the ratio of strain B to strain A) calculated by the system was recorded. SE videos of the lesions were documented in both the long axis and short axis.

Image Analysis
Conventional ultrasound features of breast masses were analyzed by two experienced radiologists (more than 5 years of experience in breast ultrasound imaging) who were blinded to the pathological results according to BI-RADS classification (12) and finally evaluated as category 2, 3, 4a, 4b, 4c, and 5. The final assessments would be given after a discussion of two radiologists when there was a disagreement. Category 4a was considered as the cutoff value: benign, category 2 or 3; malignant, category 4a, 4b, 4c, or 5.
The elastic scoring criteria of breast masses are shown in Table 1 (8). Scores 1 to 3 were considered benign, while scores 4 and 5 were considered malignant. The optimum cutoff value of SR was determined by receiver operating characteristic (ROC) analysis. A breast mass was considered malignant when the SR value was higher than the cutoff value. Otherwise, it was considered benign.

Combination Criteria of B-Mode Ultrasound and ES and/or SR
The combined analysis of B-mode ultrasound, ES, and SR of all images was based on the long-axis section of the breast mass. The BI-RADS classification of the breast mass was reassessed when combined with the ES and SR. Only BI-RADS categories 3 and 4a were upgraded or downgraded in this study. When conventional ultrasound was combined with ES or SR, BI-RADS category 3 was upgraded to category 4a if the result was malignant; BI-RADS category 4a would be downgraded to category 3 if benign was recommended. When conventional ultrasound was combined with ES and SR, category 3 was upgraded to category 4a if both ES and SR results are malignant; category 4a would be downgraded to category 3 if both were recommended as benign; otherwise, the BI-RADS classification of the mass would be unchanged.

Statistical Analysis
The histopathological results were considered the reference standard for this study. The diagnostic sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), area under the curve (AUC) value, and positive and negative diagnostic likelihood ratios (LR+, LR-) of ES and SR on two different sections (long axis and short axis) were calculated and compared. The diagnostic value of the combination of conventional ultrasound and ES and/or SR in differentiating benign and malignant breast masses and reducing biopsy of BI-RADS 4a lesions were analyzed and compared: conventional ultrasound and ES, conventional ultrasound and SR, and conventional ultrasound and ES and SR.
Quantitative data such as patient age and tumor size were expressed as means and standard deviations, and compared using t test or Mann-Whitney U test. The chi-square test and Fisher's test were used to compare categorical variables. The comparison between AUC values was performed by the DeLong method (15). The SPSS (version 22, IBM Corp.) and MedCalc software (V.19.0.7, MedCalc Software, Ostend, Belgium) were used for all statistical analyses. p-values less than 0.05 were assumed to be statistically significant.

Clinical Characteristics
A total of 910 patients (mean age, 45.3 ± 10.9 years) were finally included in this study after the exclusion criteria were performed ( Figure 2). Invasive ductal carcinoma was the most common of 332 (36.5%) malignant breast masses, accounting for 83.4% (277/ 332). Among 578 (63.5%) cases of benign breast masses, proliferative disease (61.2%, 354/578) and fibroadenoma  Homogeneous green within the mass Score 2 Most of the area is light green, with some blue around and/or in the center of the mass Score 3 Half of the area is blue and half is green in the mass Score 4 Homogeneously blue with or without a little green within the mass Score 5 Homogeneously blue with or without a little green throughout the entire mass and its surrounding area (31.0%, 179/578) were the main ones. The characteristics of patients and masses are summarized in Table 2. Patients with benign breast masses are significantly younger than those with malignant masses (42.3 ± 9.9 vs. 50.4 ± 10.5, p < 0.001). The diameter (22.5 ± 10.4 vs. 14.6 ± 7.3, p < 0.001) and SR (2.8 ± 1.6 vs. 1.6 ± 0.8, p < 0.001) of the breast mass with histopathological findings of malignancy were significantly higher than those of the benign mass. In addition, there were significant statistical differences in the distribution of malignant and benign breast masses in the ES and BI-RADS classification (all p < 0.001).

SR in BI-RADS Classification
SR values of different BI-RADS categories are shown in Table 3. For masses classified as BI-RADS category 5, the SR value was significantly higher than that of category 4 (median value, 2.520 vs. 1.770, p < 0.001), and the SR value for category 4 masses was significantly higher than that of category 3 (median value, 1.770 vs. 1.330, p < 0.001, Table 3 and Figure 3A).  Table 3 and Figure 3B).

Diagnostic Performance of ES and SR
The optimal cutoff values of SR in the long-axis and short-axis sections were determined by the Youden index. In the long-axis section of the breast mass, 2.27 was the optimal cutoff value of SR, with a sensitivity of 60.2% and a specificity of 84.8%. In the short-axis section, 2.12 was the optimal cutoff value of SR, with a sensitivity of 63.6% and a specificity of 82.5% (  Table 4). Figure 4.

Diagnostic Value of BI-RADS Combined ES and SR
When combined with ES and SR, the diagnostic performance of the re-assessed BI-RADS classification was as follows: the AUC increased from 0.733 to 0.824 (p < 0.001); the specificity increased from 48.1% to 68.5% (p < 0.001) without a decrease in the sensitivity (98.5% vs. 96.4%, p = 0.065), and the PPV increased from 52.2% to 63.7% (p < 0.001) without a loss in the NPV (98.2% vs. 97.1%, p = 0.327, Table 5).
The AUC, specificity, and PPV were higher than those of BI-RADS classification alone by the addition of ES or SR to the BI-RADS classification (all p < 0.05). However, the sensitivity decreased from 98.5% to 96.1% (BI-RADS combined with ES, p = 0.039) and 94.9% (BI-RADS combined with SR, p = 0.004), respectively. There was no statistically significant difference in NPV, with all p > 0.05 ( Table 5). The comparison of the AUC of the three combination methods is shown in Figure 5.

The Value of ES and SR in Reducing Unnecessary Biopsy
The value of ES and SR in reducing biopsy of BI-RADS 4a lesions is summarized in Table 6. When BI-RADS was combined with ES, biopsy was avoided in 80 masses, and the biopsy rate decreased from 100% to 68.3% (p < 0.001). When combined with SR, 164 masses avoided biopsy, with a lower biopsy rate   (34.9% vs. 100%, p < 0.001). When combined with both ES and SR, biopsy of 125 masses was avoided and the biopsy rate decreased to 50.4% (p < 0.001). There was no statistical difference in the malignant rate of biopsy regardless of the combination of BI-RADS and SE (all p > 0.05).

DISCUSSION
This prospective multicenter study explored the auxiliary value of two SE diagnostic methods, ES and SR, in the assessment of Bmode ultrasound breast lesions. Our results indicate that ES has a higher AUC, sensitivity, and NPV, but lower specificity and PPV in differentiating benign from malignant breast masses compared with SR. The AUC, specificity, and PPV of BI-RADS combined with ES and SR were higher than those of BI-RADS alone, without the loss of sensitivity and NPV. In addition, BI-RADS combined with ES or/and SR can significantly reduce the biopsy rate of BI-RADS 4a lesions without affecting the malignant rate of biopsy.
Recently, a new SE technique equipped in Samsung ultrasound system has been more and more widely used, and a multicenter study to explore how to make better use of its strain technique is necessary. Studies have shown that the ES and SR of SE techniques show good diagnostic accuracy in distinguishing benign from malignant breast masses (16,17). These are consistent with our result that both ES and SR have statistical differences in the differentiation between benign and malignant breast masses. However, 3.5 (10), 4.2 (16), 2.3 (17), and 4.5 (18) were used as cutoff values of SR to distinguish benign and malignant breast masses. The difference in the cutoff value of SR may be caused by the difference in strain calculation methods of various equipment vendors (17). In addition, the measurement of SR is greatly affected by the initial shear modulus and elastic nonlinearity of the lesion, as well as the pre-compression during image acquisition (19). In the previous single-center study, the cutoff value of SR was 1.765, and the sensitivity and specificity were 76% and 75%, respectively. In our multicenter study, 2.27 was the optimal cutoff value of SR, with a sensitivity of 60.2% and a specificity of 84.8%. The data for this study come from nine different hospitals, and the cutoff value of SR may be more objective.
The influence of different planes of breast masses on elastography imaging was discussed in this study. The results showed that there were no statistical differences in the diagnostic performance of both ES and SR in the long-axis and short-axis sections, which reflected the stability of SE. With respect to the diagnostic performance of ES and SR in the diagnosis of breast lesions, our study showed that ES was superior to SR, with a higher AUC (0.829 vs. 0.787, p < 0.001). A previous study showed that ES was the most useful in the identification of benign and malignant breast masses among the four diagnostic methods (ES, SR, distance ratio, and area ratio) of SE (20), which was consistent with our result. However, some studies showed that there was no significant statistical difference in the diagnostic value of ES and SR in distinguishing benign and malignant breast masses (10,16). The operator dependence of elastography may be one of the reasons.
Elastography was considered to be helpful to improve the specificity of conventional ultrasound (11). In this study, the combination of BI-RADS and SE was in the following three forms: BI-RADS combined with ES, BI-RADS combined with SR, and BI-RADS combined with both ES and SR. Our results indicated that the combination of ES and/or SR could significantly improve the AUC and specificity of BI-RADS, which was consistent with the guidelines. Therefore, SE can be used as a useful additional method for a conventional ultrasound. Among the three combinations, the AUC of BI-RADS classification combined with SR was the highest, followed by that of BI-RADS classification combined with both ES and SR, and that of BI-RADS classification combined with SE was the lowest (0.846 vs. 0.824, p = 0.001; 0.846 vs. 0.783, p < 0.001; 0.824 vs. 0.783, p < 0.001). However, the combination of BI-RADS classification and ES or SR decreased the sensitivity compared to the BI-RADS evaluation alone (98.5% vs. 96.1%, p = 0.039; 98.5% vs. 94.9%, p = 0.004). When combined with both ES and SR, the sensitivity of the BI-RADS classification decreased from 98.5% to 96.4% with no statistical difference (p = 0.065). Therefore, BI-RADS classification combined with both ES and SR performed best to improve specificity without the loss of sensitivity.
The study showed that SR can be used as a valuable method for the evaluation of breast lesions in categories 3 and 4a, but not in categories 4b and 4c (21). Similarly, our results showed that the SR value for category 4a lesions was significantly higher than that for category 3 lesions, but there was no significant difference between category 4b and 4c lesions. For breast lesions that were highly suspected for malignancy by conventional ultrasound, the stiffness of the tissue had little effect on the patient's clinical decision-making (21). Elastography may play a role in improving the selection of biopsy for patients with low suspicion lesions (20). BI-RADS category 3 or 4a lesions were upgraded or downgraded based on the results of ES and SR. This multicenter study showed that all three combined methods could reduce the biopsy rate of category 4a lesions without reducing the malignant rate of biopsy, and BI-RADS classification combined with SR was found to be the most useful. Therefore, elastography imaging can be used as a noninvasive auxiliary method to reduce unnecessary biopsy of BI-RADS 4a lesions, thus avoiding negative emotions of patients and complications of tissue biopsy.
The main limitation of our study was the uneven distribution of the patient population and histopathological results. In addition, the repeatability of the breast SE technique with the Samsung ultrasound system was not explored in this study. The repeatability of elastography is mainly manifested in the variability of data acquisition and interpretation (22). Our prospective studies followed very strict procedural protocols to minimize differences among radiologists in data acquisition and interpretation. Lastly, some patients were unable to be included in this study because they did not get a reliable SE assessment. On the one hand, breast masses do not meet the requirements of elastic quality indicators. On the other hand, the breast mass is too large and occupies the whole elastic frame, so it is impossible to evaluate the elasticity.
In summary, the optimal cutoff value of SR for differentiating benign from malignant masses was 2.27, with a sensitivity of 60.2% and a specificity of 84.8%. In addition, SE can be used as a useful and non-invasive additional method to improve the diagnostic performance of conventional ultrasound by increasing AUC and specificity and reducing unnecessary biopsy of BI-RADS 4a lesions.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available from the corresponding author upon reasonable request.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Institutional Review Board of Tongji Medical College of Huazhong University of Science & Technology. The patients/participants provided their written informed consent to participate in this study.