Abstract
Objective:
To evaluate the diagnostic performance of contrast-enhanced ultrasound (CEUS) combined with conventional ultrasound of axillary lymph nodes (ALNs) in predicting metastatic ALNs in patients with breast cancer.
Methods:
This retrospective study included 259 patients with breast cancer who underwent conventional ultrasound and CEUS. The parameters and patterns evaluated on conventional ultrasound included short axis diameter (S), long axis/short axis (L/S) ratio, cortical thickness, resistive index (RI), lymph node (LN) morphology of greyscale ultrasound, hilum and vascular pattern. Meanwhile, enhancement pattern, wash-in time, time to peak (TP), maximum signal intensity, and duration of contrast enhancement were evaluated on CEUS. Univariate and multiple logistic regression analyses were performed to identify independent factors of ALN status. Three models (conventional ultrasound, CEUS, and combined parameters) were established. Receiver operating characteristic (ROC) curves were applied to evaluate the accuracy of the three predictive models.
Results:
On conventional axillary ultrasound, LN morphology and vascular pattern were independent factors in predicting metastatic ALNs. On CEUS, maximum signal intensity, duration of contrast enhancement, and TP were independent factors in predicting metastatic ALNs. When combining conventional ultrasound and CEUS features, five independent factors obtained from the conventional ultrasound and CEUS were associated with ALN status. ROC curve analysis showed that the use of CEUS markers combined with conventional ultrasound features (AUC = 0.965) was superior to the use of CEUS markers (AUC = 0.936) and conventional ultrasound features alone (AUC = 0.851).
Conclusion:
Combining conventional ultrasound and CEUS features can enable discrimination of ALN status better than the use of CEUS and conventional ultrasound features alone.
Advances in knowledge:
The axillary lymph node status in breast cancer patients impacts the treatment decision. Our ultrasonic data demonstrated that CEUS features of ALNs in breast cancer patients could be image markers for predicting ALN status. Combining conventional ultrasound and CEUS features of ALNs can improve specificity discrimination of ALN status better than the use of CEUS and the conventional ultrasound features alone, which will help the treatment planning optimization.
Introduction
Breast cancer is the most commonly diagnosed cancer among females worldwide and is the second leading cause of cancer-related deaths. 1,2 The detection of metastasis in axillary lymph nodes (ALNs) is essential for prognosis and treatment decisions in patients with breast cancer. 3 Axillary lymph node dissection (ALND) has traditionally been the standard method of lymph node staging in breast cancer. 4 In recent years, radiotherapy or surgery (AMAROS) trials among females with T stage 1–2 invasive primary breast cancer and one or two sentinel lymph nodes (SLNs) containing metastases showed that ALND might be omitted if followed by radiotherapy, as supported by Z0011 and mapping of the axilla. 5,6 Sentinel lymph node biopsy (SLNB) is significantly less invasive than ALND. SLNB has completely replaced ALND as a staging procedure for patients with negative clinically detected LNs due to serious associated complications of the latter. 7,8 Although the morbidity after SLNB is significantly lower than ALND, the incidence rates of lymphoedema, weakened shoulder, painful scars, shoulder/arm pain and numbness of the upper arm (3.5%, 3.5%, 3.7%, 8.1%, and 10.9%, respectively) cannot be ignored. 8 In addition, the results of SLNB are influenced by the existence of unusual lymphatic drainage, the quality of preoperative lymphoscintigraphy, and the experience of surgeons.
Conventional ultrasound (US) has long been a routine imaging modality for preoperatively characterising breast lesions and determining ALN status. 9 Ultrasound features are commonly used to evaluate ALNs based on dimensional, morphological, or colour Doppler 10 ; however, these methods are not entirely reliable due to low sensitivity (SENS) and specificity (SPEC) values of 27.4–92.0% and 55.6–98.1%, respectively. 11,12 A previous study showed that the diagnostic performance of greyscale ultrasound is unsatisfactory to determine ALN status and the area under the receiver operating characteristic curve (AUC) is 0.585–0.719. 13 However, the accuracy and reproducibility of the interpretations of ultrasound images often rely on the skill of ultrasonographer, resulting in interobserver variability.
In the past decade, the clinical use of CEUS has expanded research on its application. Previous studies reported that CEUS is better than conventional ultrasound in predicting benign and malignant tumours and evaluating molecular subtypes of tumours. 14–16 Rubaltelli et al. 15 concluded that CEUS has higher accuracy than any other current ultrasound techniques. Ouyang et al. 14 demonstrated that CEUS could predict tumour aggressiveness and play an important role in discriminating metastatic from non-metastatic LNs in patients with breast cancer; however, the specificity is 76%, which is insufficient for making a clinical decision. Therefore, we added the CEUS biomarkers to conventional axillary ultrasound and evaluated the diagnostic performance of CEUS combined with the conventional ultrasound in predicting metastatic ALNs among patients with breast cancer.
Methods
Patients
This retrospective study was approved by the institutional review board. Since CEUS is the standard of care in The First Affiliated Hospital of Nanjing Medical University institution, informed consent is not required for this study. A total of 259 consecutive patients who underwent breast and conventional axillary ultrasound and CEUS at our institution between November 2013 to May 2018 were enrolled. The inclusion criteria were as follows: (1) patients with recorded ultrasound primary tumour characteristics and ALN status; (2) patients with primary breast cancer confirmed after biopsy or resection; (3) patients who did not accept surgery, chemotherapy or radiotherapy for ipsilateral or contralateral malignant lesion or existing distant metastasis before ultrasound examination; and (4) final histopathological results obtained through surgical or ultrasound-guided biopsy. The exclusion criteria were as follows: (1) patients with benign or metastatic breast mass; (2) patients who had contraindications for the use of ultrasound contrast agents; (3) patients with previous history of axillary surgery or radiotherapy for ipsilateral or contralateral breast cancer and unplanned axillary surgery; and (4) patients with multifocal lesions or bilateral diseases. Finally, 259 patients (age range 23–80 years; mean age, 49.56 ± 10.66 years) were included in this study.
Ultrasound examination
Conventional ultrasound imaging of the breast and axilla region was performed thoroughly with a ultrasound scanner (Mylab Twice, ESAOTE S.p.A. Italy) with a 5–13 MHz probe LA523. Colour Doppler Flow ultrasound was used to evaluate intralesional vessels and compare images of different planes. The most abundant vascularity plane was selected for CEUS. Ultrasound examinations were carried out with optimised colour Doppler parameters set at low-velocity scale and low wall filter. Colour gain was adjusted dynamically to maximise and eliminate colour noise.
CEUS was performed with the above device and a 4–9 MHz probe LA522. Low mechanical index (MI) values (MI: 0.05–0.1) were used to reduce microbubble destruction. The machine parameters were set as follows: gain, 100–120 dB; and frame rate, 12–15 frames per second. The contrast agent was SonoVue (Bracco, Milan, Italy). Sulphur hexafluoride powder was diluted with 5 ml of physiological saline, and a 2.4 ml of the contrast agent was administered in bolus manner via the cubital vein. Continuous imaging was performed for 4 min immediately after the injection of the contrast agent. Time–intensity curves and ultrasonographic perfusion maps were obtained by locating a region of interest on the nodes of the maximum–signal intensity zones by using Qontraxt software. No parameter was changed during the examination. Ultrasound video clips and images were recorded for subsequent research.
Image analysis
Conventional axillary ultrasound and CEUS were performed by two radiologists (LWD, HYG) with 5 and 10 years of experience in breast ultrasound. These two examinations on the same patient were performed by the same radiologist. Conventional axillary ultrasound images, enhancement patterns, and kinetic data were retrospectively reviewed by a single radiologist (CYL) with 20 years of experience in breast ultrasound, who was blinded to patients’ clinical data. Lymph node stage was determined according to the American Joint Committee on Cancer Staging Systems for Breast Cancer on the basis of the most significant metastases in either SLNs or non-SLNs: node negative (pN0), ITC [≤0.2 mm, pN0(i+)], micrometastases (>0.2–2 mm, pN1mi) or macrometastases (>2 mm, pN1). 17
On conventional ultrasound, the target breast lesion was measured on the largest diameter plane, and the ultrasound Breast Imaging-Reporting and Data System (BI-RADS) 18 was used for classification. After the breast ultrasound examination, the radiologists routinely performed conventional axillary ultrasound examination and recorded the ultrasound characteristics of ALNs. When the axillary lymph nodes of patients with breast cancer were at different levels, the one with the most malignant signs was selected. If there were 2–3 lymph nodes at a particular level, the largest one was selected. For each ALN, the parameters tested included long axis diameter (L), short axis diameter (S), L/S ratio, cortical thickness, and resistive index (RI). All data were measured on the long axis of ALNs. According to the cortical morphological features, each ALN was divided into one of the following four types: Type I, thin and uniform hypoechoic cortex; Type II, uniform hypoechoic cortex, cortex thickness is less than 3 mm; Type III, focal cortex thickening or uneven thickening, the thickness is greater than or equal to 3.0 mm; and Type IV, totally hypoechoic node with no hyperechoic hilum (Figure 1). 19 According to the modifications of the criteria by Na and Yang et al, 20,21 the vascular pattern of each ALN was assessed. The patterns were classified into one of the following categories based on colour flow: perihilar vascularity, central vascularity, peripheral vascularity, and mixed vascularity. 19 Peripheral vessels were defined as vessels coursing along the margin of a lymph node. Mixed vascularity was described as a node having more than one vascular pattern. Central vascularity, peripheral vascularity, and mixed vascularity were considered metastatic (Figure 2).
Figure 1.
Types I-–IV, according to cortical morphologic features on longitudinal Greyscale ultrasound images from four patients. (a) A 50-year-old female with invasive ductal carcinoma and benign node ; thin and uniform hypoechoic cortex (Type I). (b) A 24-year-old female with invasive ductal carcinoma and benign node, uniform hypoechoic cortex, cortex thickness is 2.2 mm (Type II). (c) A 46-year-old female with carcinoma muciparum and metastatic; focal cortex thickening or uneven thickening, thickness greater than 3.0 mm (Type III). (d) A 67-year-old female with invasive ductal carcinoma and metastatic; totally hypoechoic node with no hilum (Type IV).
Figure 2.
Vascularity patterns on longitudinal colour Doppler ultrasound images from four patients. (a) A 52-year-old female with invasive ductal carcinoma and benign node; Simple central hilar vessel within a normal axillary node. (b) A 44-year-old female with invasive ductal carcinoma and metastatic; Central centrifugal branches seen as blue and red linear structures within a node with thickening cortex and partially obliterated fatty hilum. (c) A 67-year-old female with invasive ductal carcinoma and metastatic; the vessels are coursing along the margin within a node with no hilum. (d) A 41-year-old female with invasive ductal carcinoma and metastatic; the node has more than one vascular pattern with no hilum.
On CEUS, the parameters evaluated included wash-in time, time to peak (TP), maximum signal intensity, and duration of contrast enhancement. The area of interest was selectively placed in the fastest and strongest areas of cortical enhancement. The quantitative parameters were classified as follows. 14 Wash-in time was defined as the time when the first microbubble was seen to enter the lesion. Maximum signal intensity was defined as the peak intensity of the time–intensity curve. TP was defined as the time required to reach the peak intensity from the first microbubble entering the lesion. The classification of the enhancement patterns of ALNs was based on literature. 22–24 Internal enhancement was classified as a homogeneous or heterogeneous enhancement. Homogeneous enhancement pattern was marked as diffuse and overall enhancement of the lesion (Figure 3). Heterogeneous enhancement pattern was defined as heterogeneous enhancement of the lesion and included high partial enhancement, low enhancement and perfusion defects (Figure 3). 25,26
Figure 3.
Enhancement patterns on longitudinal CEUS images from four patients. (a, b) A 46-year-old female with invasive ductal carcinoma and benign node; the node was homogeneous enhancement with the uniform hypoechoic cortex. (c, d) A 46-year-old female with mucinous carcinoma and metastasis; the node was heterogeneous enhancement with a 4.8 × 6.8 mm perfusion defects. (e, f) A 55-year-old female with invasive ductal carcinoma and metastatic; the node was heterogeneous enhancement with an 8.1 × 6.8 mm high enhanced area. (g, h) A 44-year-old female with invasive ductal carcinoma and metastatic; the node was heterogeneous enhancement with a 6.6 × 8.9 mm low enhanced area. CEUS, contrast-enhanced ultrasound.
Statistical methods
Non-parametric Mann–Whitney U test or Student’s t-test was performed for continuous data. Fisher’s exact or Pearson χ2 test was used to compare differences in categorical variables. Wilcoxon rank-sum test was performed to compare proportions and medians among groups. Univariate analysis was used to assess the relationship between ultrasound features and metastatic ALNs. Given the large number of variables used in this study, Bonferroni correction was applied (p < .0042). The statistically significant ultrasound features were used as input variables for multivariate logical regression analysis to identify independent predictors. A multivariate analysis model was established using backward stepwise elimination method. Comparisons among three AUC operating characteristic curves (ROC) were conducted with the non-parametric approach of DeLong, DeLong, and Clarke-Pearson. 27 Other parameters such as SENS, SPEC, negative predictive value (NPV), and positive predictive value (PPV) were also used to evaluate the model performance. Backward elimination method was used to select the final model. All data were collected and statistically analysed by MedCalc Statistical Software v. 15.6.1 (MedCalc Software bvba, Ostend, Belgium).
Results
Correlation of ALN status with clinical features
A total of 259 patients (age range 23–80 years; mean age 49.56 ± 10.66 years) were included in this study. All patients with breast cancer had undergone either SLNB or ALND. The pathological results revealed that 138 (52.28%) were N0, 0 (0%) were N0 (i+), 2 (0.78%) were N1mi and 119 (45.95%) were N1. The mean size of the ALNs was 2.5 cm (range 0.7–7.5 cm). Table 1 shows data on patient age, ultrasound size, human epidermal growth factor receptor 2 (HER-2), Ki-67 proliferation index, oestrogen receptor (ER) status, progesterone receptor (PR) status, BI-RADS category, and tumour type. No significant difference in age, HER-2, ER, and PR between metastatic and non-metastatic ALNs was found. Metastatic ALNs had a significantly higher incidence rate in larger malignant breast masses than in smaller masses (p < 0.001). The incidence rate of ALNs metastases in higher BI-RADS category of breast cancer was higher than that in lower category ones (p < 0.001). A significant correlation was found between ALN status and tumour type (p < 0.001).
Table 1.
Demographic and histological information of two groups
| Characteristics | All patients | Non-metastasis | Metastasis | P | Univariate OR |
|---|---|---|---|---|---|
| Number | 259 | 138 (53.28%) | 121 (46.72%) | ||
| Age, mean ± SD, years | 49.56 ± 10.66 | 49.44 ± 11.18 | 49.69 ± 10.12 | 0.852 | 1.002 (0.980, 1.025) |
| Ultrasound size of breast tumour, mean ± SD, mm | 26.48 ± 15.71 | 22.55 ± 14.55 | 30.69 ± 15.87 | <0.001 | 1.041 (1.020, 1.062) |
| HER2 | |||||
| Positive | 177 | 98 (61.01%) | 79 (65.29%) | 0.323 | Reference |
| Negative | 82 | 40 (28.99%) | 42 (34.71%) | 0.768 (0.454, 1.297) | |
| Ki-67 | |||||
| Positive | 149 | 77 (55.80%) | 72 (59.50%) | 0.547 | 1.164 (0.710, 1.909) |
| Negative | 110 | 61 (44.20%) | 49 (40.50%) | Reference | |
| ER | |||||
| Positive | 184 | 97 (70.29%) | 87 (71.90%) | 0.776 | 1.082 (0.631, 1.854) |
| Negative | 75 | 41 (29.71) | 34 (28.10%) | Reference | |
| PR | |||||
| Positive | 164 | 84 (60.87%) | 80 (66.12%) | 0.382 | 1.254 (0.754, 2.086) |
| Negative | 95 | 54 (39.13%) | 41 (33.88%) | Reference | |
| BI-RADS category | <0.001 | ||||
| 4A | 29 | 25 (18.11%) | 4 (3.31%) | <0.001 | 0.110 (0.036, 0.330) |
| 4B | 28 | 20 (14.49%) | 8 (6.61%) | 0.004 | 0.274 (0.114, 0.661) |
| 4C | 47 | 30 (21.74%) | 17 (14.05%) | 0.006 | 0.388 (0.197, 0.763) |
| 5 | 155 | 63 (45.65%) | 92 (76.03%) | Reference | |
| Tumour type | <0.002 | ||||
| Invasive ductal carcinoma | 226 | 112 (81.16%) | 114 (94.21%) | 0.001 | 6.362 (2.145, 18.868) |
| Invasive lobular carcinoma | 4 | 1 (0.72%) | 3 (2.48%) | 0.021 | 18.750 (1.543, 2.781) |
| Other tumour types | 29 | 25 (18.12%) | 4 (3.31%) | Reference |
SD, standard deviation.
Correlation of conventional ultrasound and CEUS features with ALN status
Associations between conventional ultrasound and CEUS features and ALN pathological status are illustrated in Table 2. On conventional ultrasound, significant differences in S, cortical thickness, L/S ratio, LN morphology of greyscale ultrasound, hilum, vascular pattern, and RI were found between the two groups (p < .001). ALNs with longer S, cortical thickness ≥3.0, smaller L/S ratio, absence of a hilum, non-perihilar vessel, and RI ≥0.7 were more likely to have metastasis than non-metastasis. The types of malignant ALNs tended to be types II (26.45%), III (37.19%), and IV (23.97%), while the types of benign ALNs tended to be type I (68.84%) (p < 0.001). On CEUS, significant differences in TP, maximum signal intensity, duration of contrast enhancement, and enhancement pattern were found between the two groups (p < 0.001). A significantly higher degree of enhancement (higher maximum signal intensity and longer duration of enhancement) was observed in malignant ALNs than in benign ALNs (p < 0.001). Of the 121 metastatic ALNs, 62 (51.24%) showed signs of high partial enhancement, low enhancement, and perfusion defects. A total of 120 (86.96%) ALNs showed homogeneous enhancement, and 18 (13.04%) showed heterogeneous enhancement in the benign ALN group. No significant correlation was observed between wash-in time and ALN status (p = 0.640).
Table 2.
Univariate analysis for the conventional ultrasound and CEUS features of ALNs
| Non-metastasis (n = 138) |
Metastasis (n = 121) |
P | |
|---|---|---|---|
| Conventional ultrasound | |||
| S | 6.55 ± 2.40 | 8.95 ± 3.70 | <0.001 |
| L/ S ratio | 2.80 ± 1.03 | 2.28 ± 1.02 | <0.001 |
| Cortical thickness | <0.001 | ||
| ≥3 mm | 28 (20.29%) | 90 (74.38%) | |
| ＜3 mm | 110 (79.71%) | 31 (25.62%) | |
| LN morphology of Greyscale ultrasound | <0.001 | ||
| I | 95 (68.84%) | 15 (12.40%) | |
| II | 20 (14.49%) | 32 (26.45%) | |
| III | 20 (14.49%) | 45 (37.19%) | |
| IV | 3 (2.17%) | 29 (23.97%) | |
| Hilum | <0.001 | ||
| Existence | 135 (97.82%) | 92 (76.03%) | |
| Absence | 3 (2.17%) | 29 (23.97%) | |
| Vascular pattern | <0.001 | ||
| Perhilar vessel | 130 (94.20%) | 57 (47.11%) | |
| Central vessel | 1 (0.77%) | 6 (4.96%) | |
| Peripheral vessel | 1 (0.77%) | 7 (5.85%) | |
| Mixed vascularity | 6 (4.35%) | 51 (42.15%) | |
| RI | <0.001 | ||
| <0.7 | 135 (97.83%) | 93 (76.86%) | |
| ≥0.7 | 3 (2.17%) | 28 (23.14%) | |
| CEUS | |||
| Wash-in time | 12.86 ± 3.28 | 12.66 ± 3.05 | 0.6399 |
| Time to peak | 29.68 ± 6.92 | 21.69 ± 8.07 | <0.001 |
| Maximum signal intensity | 66.34 ± 7.22 | 78.49 ± 9.10 | <0.001 |
| Duration of contrast enhancement | 173.30 ± 47.64 | 270.66 ± 62.00 | <0.001 |
| Enhancement pattern | <0.001 | ||
| Homogeneous | 120 (86.96%) | 59 (48.76%) | |
| Heterogeneous | 18 (13.04%) | 62 (51.24%) | |
CEUS, contrast enhanced ultrasound; RI, resistive index.
Multivariable analyses of conventional ultrasound and CEUS features and ROC
Curve analysis
The results of the logistic regression analysis are summarised in Table 3. For the model with conventional ultrasound features alone, the LN morphology of greyscale and vascular pattern was proven to be independent prognostic factors of benign and malignant LNs. Model A was built up by the LN morphology of greyscale ultrasound and vascular pattern (AUC = .851, SENS = 89.26%, SPEC = 68.84%). For the model with CEUS features alone, the TP, maximum signal intensity, and duration of contrast enhancement were identified to be independent prognostic factors of benign and malignant LNs. Model B was built up by TP, maximum signal intensity and duration of contrast enhancement (AUC = 0.936, SENS = 92.56%, SPEC = 87.68%). Finally, for the model with both conventional ultrasound and CEUS features of ALNs, regardless of statistical significance, the five ultrasound features obtained from the conventional ultrasound and CEUS were kept in the model. Model A + B was built by the five ultrasound features (AUC = 0.965, SENS = 92.56%, SPEC = 91.30%) (Table 4). The specificity of the combined model significantly increased from 68.84 to 91.30% when CEUS features were added (p < 0.05). Statistical differences were found in the AUC areas among the three models (p < 0.05) (Figure 4).
Table 3.
Multivariable logistic regression analysis of the ALNs conventional ultrasound and CEUS features
| Variable | B | SE | Relative risk (95% CI) | P |
|---|---|---|---|---|
| Conventional ultrasound | ||||
| S | 0.007 | 0.076 | 1.007 (0.868–1.168) | 0.926 |
| L/ S ratio | −0.106 | 0.196 | 0.900 (0.613–1.321) | 0.900 |
| LN morphology of Greyscale ultrasound | 0.008 | |||
| I | −2.362 | 1.391 | 0.094 (0.006–1.442) | 0.090 |
| II | −0.556 | 1.377 | 0.573 (0.039–8.528) | 0.686 |
| III | −0.605 | 1.372 | 0.546 (0.037–8,030) | 0.695 |
| Cortical thickness | −0.479 | 0.488 | 0.619 (0.238–1.612) | 0.326 |
| Absence of hilum | 0.732 | 1.392 | 2.061 (0.135–31.570) | 0.604 |
| Vascular pattern | 0.038 | |||
| Perhilar | −2.416 | 0.842 | 0.089 (0.017–0.456) | 0.004 |
| Central | −1.343 | 1.272 | 0.261 (0.022–3.157) | 0.291 |
| Peripheral | −1.334 | 1.514 | 0.263 (0.014–5.125) | 0.378 |
| RI ≥0.7 | −1.273 | 1.023 | 0.452 (0.061–3.357) | 0.438 |
| CEUS | ||||
| Time to peak | −0.123 | 0.025 | 0.884 (0.841–0.929) | <0.001 |
| Maximum signal intensity | 0.108 | 0.028 | 1.114 (1.054–1.177) | <0.001 |
| Duration of contrast enhancement | 0.024 | 0.004 | 1.023 (1.013–1.033) | <0.001 |
| Enhancement pattern | 0.799 | 0.581 | 2.224 (0.712–6.946) | 0.196 |
ALN, axillary lymph node; CEUS, contrast enhanced ultrasound; RI, resistive index.
Table 4.
The prediction of different models for ALN status
| Model | SENS (%) | SPEC (%) | PPV (%) | NPV (%) | AUC |
|---|---|---|---|---|---|
| Model A | 89.26 (82.3–94.2) |
68.84 (60.4–76.4) |
71.5 (63.6–78.6) |
88.0 (80.3–93.4) |
0.851 (80.2–89.2) |
| Model B | 92.56 (86.3–96.5) |
87.68 (81.0–92.7) |
86.8 (79.7–92.1) |
93.1 (87.3–96.8) |
0.936 (89.9–96.3) |
| Model A + B | 92.56 (86.3–96.5) |
91.30 (85.3–95.4) |
90.3 83.7–94.9) |
93.3 87.7–96.9) |
0.945 (93.5–98.4) |
ALN, axillary lymph node;AUC, area under the receiver operating characteristic curve; NPV, negative predict value; PPV, positive predictive value; SENS, sensitivity; SPEC, specificity.
95% confidence intervals are included in brackets. Source data are provided as a Source Data file. Model A conventional ultrasound, Model B contrast-enhanced ultrasound, Model A + B CEUS plus conventional ultrasound.
Figure 4.
Receiver operating characteristic (ROC) curves comparison between different models for predicting metastatic ALNs in breast cancer patients. ALN, axillary lymph node; ROC, receiver operating characteristic.
Discussion
Accurate identification of ALN involvement in patients with breast cancer is important for prognosis and therapy decisions. SLNB is a standard procedure for regional lymph node staging. 28 However, SLNB still brings out a risk of acute and long-term complications, 29 including lymphedema, wound infection, seroma formation, and pain. 30 Hence, there is an increasing need for predicting metastatic extent of ALN accurately in a noninvasive way.
In our study, we validated and developed CEUS features combined with conventional axillary ultrasound features for preoperative prediction of ALN status. We demonstrated that LN morphology, vascular pattern, maximum signal intensity, duration of contrast enhancement, and TP were five independent factors in predicting metastatic ALNs on the combined model. Moreover, the combined model showed better diagnostic performance than any single method in differentiating patients with metastatic and non-metastatic ALNs.
With regard to LN morphology, 68.84% of ALNs (95/138) showed Type I in no-metastatic ALNs, whereas 87.60% (106/121) ALNs showed types II–IV in metastatic ALNs. We found that the cortical morphological characteristics of ALNs can be used as predictors of breast cancer metastasis, consistent with the findings of Bedi et al. 31 This finding may be explained by the fact that metastatic cells in lymph fluid pass through one or more afferent lymphatic channels to reach the marginal sinus on the periphery of the lymph node. Lymphatic fluid then passes from the marginal sinus through the cortex and the paracortex, filtering towards the hilum, which contains lymphocytes and phagocytic cells. Metastatic deposits are blocked by these cells around the nodules, causing the cortex to enlarge, which can be local or eccentric. 32 Therefore, this cortical uplift may usually precede the extensive expansion of the cortex and eventually the replacement of the entire node. Therefore, ALNs can be classified according to cortical morphological features to predict the occurrence of metastasis.
Colour Doppler flow ultrasound provides information about morphology and flow. The use of high-frequency transducers improves the detectability of low-velocity signals from superficial structures. Na et al 21 evaluated the hilar, central, and peripheral vascularity of ALN by using colour Doppler flow ultrasound; the characteristics of colour Doppler flow ultrasound can be applied to identify disease in patients with breast cancer. Benign nodes showed normal patterns of nodal vascularity, such as radial symmetric central vascularity, central vascularity, and no peripheral vascularity. Yang et al. 20 demonstrated that colour Doppler flow was performed equally well in benign and malignant ALNs among different groups. For all ALNs, the peripheral flow of malignant ALNs was significantly higher than benign ones (p < 0.001), whereas central perihilar flow and central flow of malignant ALNs were significantly lower (p < 0.002 and <0.001, respectively) than benign ones. In our study, LN morphology and vascular pattern were independent factors when discriminating benign and malignant LNs on conventional ultrasound. Our model A was built up by the LN morphology and vascular pattern. Previous studies demonstrated that the overall diagnostic performance of preoperative axillary ultrasound results was not satisfactory, with AUC of 0.585–0.719. 33 In the present study, the diagnostic performance of Model A results was average, with an AUC of 0.851, which is slightly higher than previous studies. This finding may be explained by the retrospective analysis of conventional ultrasound. In addition, CEUS features were retrospectively reviewed by a single professional doctor, which can reduce the difference among different observers.
On CEUS, multivariate analyses showed the maximum signal intensity, duration, and TP of CEUS features were the discriminating criteria for ALN metastasis (p < 0.001). Maximum signal intensity was a significant factor when screening lymph nodes. Our results revealed a significantly higher degree of enhancement (higher maximum signal intensity and longer duration of enhancement) in metastatic ALNs than in negative ones (p < 0.001). This result is relevant to Steppan I’s findings. Steppan I et al 34 found that the duration and maximum signal intensity of enhancement were significantly higher in metastatic nodes. Angiogenesis plays a vital role in tumour growth and metastasis. 35,36 During metastasis, blood vessels may increase due to angiogenic factors in tumour cells. Blood flow may be reduced due to the encapsulation of blood vessels, compression of tumour tissues, high endothelial fluid exchange, or poor lymphatic circulation function. We found that metastatic ALNs had shorter TP than benign ones (p < 0.001). Ouyang et al 14 found the TP in reactive LNs was slightly longer than that in metastatic LNs, but no statistical difference was recorded (p = 0.300). The TP reflected the perfusion rate of microbubbles, which is different between benign and malignant ALNs. These conflicting results from different studies may be caused by their differences in the study grouping methods and designs.
Most of the prediction models for ALNs were based on several clinicopathological parameters, such as age, tumour location, tumour size, menopausal status, ER, PR, HER2, and histologic grade. However, some data could not be available preoperatively. Our model A + B was built up by CEUS and conventional ultrasound features. Compared with previous studies, our study achieved better diagnostic performance by combining conventional ultrasound and CEUS features, which can provide more information. Limiting the features extracted from the image makes the model more robust. Therefore, model A + B shows satisfactory prediction results in terms of SENS, SPEC, and AUC. In addition, for patients suspected of having breast lesions, breast and axillary ultrasound is a common practice used to characterise breast lesions and axillary lymph node conditions. Compared with other imaging methods, the proposed model has the advantages of cost-effectiveness and lack of ionising radiation.
We acknowledge that this study possesses several limitations. Firstly, prospective multicentre studies with large data sets are needed to further verify the repeatability and ruggedness of our prediction model. Secondly, this study did not assess intra- and interobserver variability in image interpretation. Thirdly, N0 patients did not receive ALND to validate the accuracy of their axillary lymph node status. Finally, abnormal ALNs displayed by the ultrasound cannot be completely matched with those in postoperative pathology. In the future, sentinel LN CEUS and titanium clips will be used as markers to locate sentinel LN more precisely.
Conclusion
CEUS features of ALNs with breast cancer can be used as image markers for ALN status. Combining conventional ultrasound and CEUS features of ALNs can enable discrimination of ALN status better than the use of CEUS and conventional ultrasound features alone.
Footnotes
The authors Li-Wen Du and Hong-Li Liu contributed equally to the work.
Contributor Information
Li-Wen Du, Email: duliwen092@sina.com.
Hong-Li Liu, Email: lhl_njmu@163.com.
Hai-Yan Gong, Email: ghy_njmu@yeah.net.
Li-Jun Ling, Email: llj_njmu@163.com.
Shui Wang, Email: wangshui_njmu@163.com.
Cui-ying Li, Email: lcy_njmu@163.com.
Min Zong, Email: mzong@njmu.edu.cn.
REFERENCES
- 1. Ginsburg O, Bray F, Coleman MP, Vanderpuye V, Eniu A, Kotha SR, et al. The global burden of women's cancers: a grand challenge in global health. Lancet 2017; 389: 847–60 10.1016/S0140-6736(16)31392-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018; 68: 394–424 10.3322/caac.21492 [DOI] [PubMed] [Google Scholar]
- 3. Ahmed M, Purushotham AD, Douek M. Novel techniques for sentinel lymph node biopsy in breast cancer: a systematic review. Lancet Oncol 2014; 15: e351–62 10.1016/S1470-2045(13)70590-4 [DOI] [PubMed] [Google Scholar]
- 4. Bevilacqua JLB, Kattan MW, Fey JV, Cody HS, Borgen PI, Van Zee KJ, Doctor VZKJ. Doctor, what are my chances of having a positive sentinel node? A validated nomogram for risk estimation. J Clin Oncol 2007; 25: 3670–9. doi: 10.1200/JCO.2006.08.8013 [DOI] [PubMed] [Google Scholar]
- 5. Giuliano AE, Ballman KV, McCall L, Beitsch PD, Brennan MB, Kelemen PR, et al. Effect of axillary dissection vs no axillary dissection on 10-year overall survival among women with invasive breast cancer and sentinel node metastasis: the ACOSOG Z0011 (Alliance) randomized clinical trial. JAMA 2017; 318: 918–26. doi: 10.1001/jama.2017.11470 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Donker M, Gv T, Straver ME, Meijnen P, Rutgers EJT. Radiotherapy or surgery of the axilla after a positive sentinel node in breast cancer (EORTC 10981- 22023. The Lancet Oncology 2014; 15: 1303–10. doi: 10.1016/S1470-2045(14)70460-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Lyman GH, Somerfield MR, Bosserman LD, Perkins CL, Weaver DL, Giuliano AE. Sentinel lymph node biopsy for patients with early-stage breast cancer: American Society of clinical oncology clinical practice guideline update. JCO 2017; 35: 561–4. doi: 10.1200/JCO.2016.71.0947 [DOI] [PubMed] [Google Scholar]
- 8. Langer I, Guller U, Berclaz G, Koechli OR, Schaer G, Fehr MK, et al. Morbidity of sentinel lymph node biopsy (SLN) alone versus SLN and completion axillary lymph node dissection after breast cancer surgery: a prospective Swiss multicenter study on 659 patients. Ann Surg 2007; 245: 452–61. doi: 10.1097/01.sla.0000245472.47748.ec [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Lee B, Lim AK, Krell J, Satchithananda K, Coombes RC, Lewis JS, et al. The efficacy of axillary ultrasound in the detection of nodal metastasis in breast cancer. AJR Am J Roentgenol 2013; 200: W314–20. doi: 10.2214/AJR.12.9032 [DOI] [PubMed] [Google Scholar]
- 10. Ecanow JS, Abe H, Newstead GM, Ecanow DB, Jeske JM. Axillary staging of breast cancer: what the radiologist should know. Radiographics 2013; 33: 1589–612. doi: 10.1148/rg.336125060 [DOI] [PubMed] [Google Scholar]
- 11. Alvarez S, Añorbe E, Alcorta P, López F, Alonso I, Cortés J. Role of sonography in the diagnosis of axillary lymph node metastases in breast cancer: a systematic review. AJR Am J Roentgenol 2006; 186: 1342–8. doi: 10.2214/AJR.05.0936 [DOI] [PubMed] [Google Scholar]
- 12. Mills P, Sever A, Weeks J, Fish D, Jones S, Jones P. Axillary ultrasound assessment in primary breast cancer: an audit of 653 cases. Breast J 2010; 16: 460–3. doi: 10.1111/j.1524-4741.2010.00952.x [DOI] [PubMed] [Google Scholar]
- 13. Youk JH, Son EJ, Kim J-A, Gweon HM. Pre-Operative evaluation of axillary lymph node status in patients with suspected breast cancer using shear wave elastography. Ultrasound Med Biol 2017; 43: 1581–6. doi: 10.1016/j.ultrasmedbio.2017.03.016 [DOI] [PubMed] [Google Scholar]
- 14. Ouyang Q, Chen L, Zhao H, Xu R, Lin Q. Detecting metastasis of lymph nodes and predicting aggressiveness in patients with breast carcinomas. J Ultrasound Med 2010; 29: 343–52. doi: 10.7863/jum.2010.29.3.343 [DOI] [PubMed] [Google Scholar]
- 15. Rubaltelli L, Beltrame V, Scagliori E, Bezzon E, Frigo AC, Rastrelli M, et al. Potential use of contrast-enhanced ultrasound (CEUS) in the detection of metastatic superficial lymph nodes in melanoma patients. Ultraschall Med 2014; 35: 67–71. doi: 10.1055/s-0033-1335857 [DOI] [PubMed] [Google Scholar]
- 16. Dudau C, Hameed S, Gibson D, Muthu S, Sandison A, Eckersley RJ, et al. Can contrast-enhanced ultrasound distinguish malignant from reactive lymph nodes in patients with head and neck cancers? Ultrasound Med Biol 2014; 40: 747–54. doi: 10.1016/j.ultrasmedbio.2013.10.015 [DOI] [PubMed] [Google Scholar]
- 17. Andersson Y, Bergkvist L, Frisell J, de Boniface J, Yvette A LB. Long-Term breast cancer survival in relation to the metastatic tumor burden in axillary lymph nodes. Breast Cancer Res Treat 2018; 171: 359–69. doi: 10.1007/s10549-018-4820-0 [DOI] [PubMed] [Google Scholar]
- 18. Berg WA, Cosgrove DO, Doré CJ, Schäfer FKW, Svensson WE, Hooley RJ, et al. Shear-Wave elastography improves the specificity of breast us: the BE1 multinational study of 939 masses. Radiology 2012; 262: 435–49. doi: 10.1148/radiol.11110640 [DOI] [PubMed] [Google Scholar]
- 19. Choi YJ, Ko EY, Han B-K, Shin JH, Kang SS, Hahn SY. High-Resolution ultrasonographic features of axillary lymph node metastasis in patients with breast cancer. Breast 2009; 18: 119–22. doi: 10.1016/j.breast.2009.02.004 [DOI] [PubMed] [Google Scholar]
- 20. Yang WT, Chang J, Metreweli C. Patients with breast cancer: differences in color Doppler flow and gray-scale us features of benign and malignant axillary lymph nodes. Radiology 2000; 215: 568–73. doi: 10.1148/radiology.215.2.r00ap20568 [DOI] [PubMed] [Google Scholar]
- 21. Na DG, Lim HK, Byun HS, Kim HD, Ko YH, Baek JH. Differential diagnosis of cervical lymphadenopathy: usefulness of color Doppler sonography. AJR Am J Roentgenol 1997; 168: 1311–6. doi: 10.2214/ajr.168.5.9129432 [DOI] [PubMed] [Google Scholar]
- 22. Wan CF, Du J, Fang H, Li FH, Zhu JS, Liu Q, Cai FW, Jing D, Hua F, Feng HL, Qiang L. Enhancement patterns and parameters of breast cancers at contrast-enhanced us: correlation with prognostic factors. Radiology 2012; 262: 450–9. doi: 10.1148/radiol.11110789 [DOI] [PubMed] [Google Scholar]
- 23. Wang Y, Fan W, Zhao S, Zhang K, Zhang L, Zhang P, et al. Qualitative, quantitative and combination score systems in differential diagnosis of breast lesions by contrast-enhanced ultrasound. Eur J Radiol 2016; 85: 48–54. doi: 10.1016/j.ejrad.2015.10.017 [DOI] [PubMed] [Google Scholar]
- 24. Tuncbilek N, Unlu E, Karakas HM, Cakir B, Ozyilmaz F. Evaluation of tumor angiogenesis with contrast-enhanced dynamic magnetic resonance mammography. Breast J 2003; 9: 403–8. doi: 10.1046/j.1524-4741.2003.09508.x [DOI] [PubMed] [Google Scholar]
- 25. Metz S, Daldrup-Unk HE, Richter T, Räth C, Ebert W, Settles M, et al. Detection and quantification of breast tumor necrosis with MR imaging: value of the necrosis-avid contrast agent Gadophrin-3. Acad Radiol 2003; 10: 484–90. doi: 10.1016/s1076-6332(03)80056-9 [DOI] [PubMed] [Google Scholar]
- 26. Pickles MD, Lowry M, Gibbs P. Pretreatment prognostic value of dynamic contrast-enhanced magnetic resonance imaging vascular, texture, shape, and size parameters compared with traditional survival indicators obtained from locally advanced breast cancer patients. Invest Radiol 2016; 51: 177–85. doi: 10.1097/RLI.0000000000000222 [DOI] [PubMed] [Google Scholar]
- 27. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837–45. doi: 10.2307/2531595 [DOI] [PubMed] [Google Scholar]
- 28. Garcia-Etienne CA, Ferrari A, Della Valle A, Lucioni M, Ferraris E, Di Giulio G, et al. Management of the axilla in patients with breast cancer and positive sentinel lymph node biopsy: an evidence-based update in a European breast center. Eur J Surg Oncol 2020; 46: 15–23. doi: 10.1016/j.ejso.2019.08.013 [DOI] [PubMed] [Google Scholar]
- 29. Lyman GH, Somerfield MR, Bosserman LD, Perkins CL, Weaver DL, Giuliano AE. Sentinel lymph node biopsy for patients with early-stage breast cancer: American Society of clinical oncology clinical practice guideline update. J Clin Oncol 2017; 35: 561–4. doi: 10.1200/JCO.2016.71.0947 [DOI] [PubMed] [Google Scholar]
- 30. Manca G, Rubello D, Tardelli E, Giammarile F, Mazzarri S, Boni G, et al. Sentinel lymph node biopsy in breast cancer: indications, contraindications, and controversies. Clin Nucl Med 2016; 41: 126–33. doi: 10.1097/RLU.0000000000000985 [DOI] [PubMed] [Google Scholar]
- 31. Bedi DG, Krishnamurthy R, Krishnamurthy S, Edeiken BS, Le-Petross H, Fornage BD, et al. Cortical morphologic features of axillary lymph nodes as a predictor of metastasis in breast cancer: in vitro sonographic study. AJR Am J Roentgenol 2008; 191: 646–52. doi: 10.2214/AJR.07.2460 [DOI] [PubMed] [Google Scholar]
- 32. Boughey JC, Ballman KV, Hunt KK, McCall LM, Mittendorf EA, Ahrendt GM, et al. Axillary ultrasound after neoadjuvant chemotherapy and its impact on sentinel lymph node surgery: results from the American College of surgeons Oncology Group Z1071 trial (alliance. J Clin Oncol 2015; 33: 3386–93. doi: 10.1200/JCO.2014.57.8401 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Zheng X, Yao Z, Huang Y, Yu Y, Wang Y, Liu Y, Zhou J, et al. Deep learning radiomics can predict axillary lymph node status in early-stage breast cancer. Nat Commun 2020; 11: 1236 11: 1236. doi. doi: 10.1038/s41467-020-15027-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Steppan I, Reimer D, Müller-Holzner E, Marth C, Aigner F, Frauscher F, et al. Breast cancer in women: evaluation of benign and malignant axillary lymph nodes with contrast-enhanced ultrasound. Ultraschall Med 2010; 31: 63–7. doi: 10.1055/s-0028-1109847 [DOI] [PubMed] [Google Scholar]
- 35. Shrivastav S, Bal A, Singh G, Joshi K. Tumor angiogenesis in breast cancer: pericytes and maturation does not correlate with lymph node metastasis and molecular subtypes. Clin Breast Cancer 2016; 16: 131–8. doi: 10.1016/j.clbc.2015.09.002 [DOI] [PubMed] [Google Scholar]
- 36. Su M-Y, Cheung Y-C, Fruehauf JP, Yu H, Nalcioglu O, Mechetner E, et al. Correlation of dynamic contrast enhancement MRI parameters with microvessel density and VEGF for assessment of angiogenesis in breast cancer. J Magn Reson Imaging 2003; 18: 467–77. doi: 10.1002/jmri.10380 [DOI] [PubMed] [Google Scholar]