Characterization of Indeterminate Breast Lesions based on Pressure Estimates by Noninvasive 3D Contrast-enhanced Ultrasound

Abstract

Objective:

To assess the ability of the pressure gradient between breast lesions and adjacent normal tissue estimated by 3D subharmonic-aided pressure estimation (SHAPE) to characterize indeterminate breast lesions.

Methods:

This prospective study enrolled patients scheduled for ultrasound-guided needle biopsies of a breast lesion. Before the biopsy, 3D SHAPE data were collected from the breast lesion during the infusion of an ultrasound contrast agent (Definity) as well as after clearance of the agent. Direct, invasive pressure measurements in the lesion and adjacent normal tissue were then obtained using an intra-compartmental pressure monitoring system (C2DX) before tissue sampling as part of the biopsy procedure. The mean SHAPE gradient and invasive measurement gradient between the lesion and adjacent normal tissue were compared to the biopsy results. The SHAPE gradients were also compared to the invasive pressure gradients.

Results:

There were 8 malignant and 13 benign lesions studied. The SHAPE gradients and invasive pressure gradients were significantly different between the benign and malignant lesions (2.86 ± 3.24 vs. −0.03 ± 1.72 a.u.; p=0.03 and 9.9 ± 8.5 vs. 20.9 ± 8.0 mmHg; p=0.008, respectively). The area under the curves, specificities, and sensitivities for detecting malignancy by SHAPE gradients and invasive pressure gradients were 0.79 and 0.88, 77% and 92%, and 88% and 50%, respectively. A weak negative correlation was found between the SHAPE and invasive pressure gradients (r = −0.2).

Conclusion:

The pressure gradient between a breast lesion and adjacent normal tissue estimated by 3D SHAPE shows potential for characterizing indeterminate breast lesions.

INTRODUCTION

Breast cancer is the most common new cancer diagnosis for women in the United States (1). Early detection through population-based screening programs, increased breast cancer awareness, and treatment advances are thought to have led to an overall 43% decline in breast cancer mortality since the late 1980s (2). Even with these advances, breast cancer remains the second leading cause of cancer death in women in the United States (1). Mechanisms to improve early and accurate diagnosis are critical to continue to improve outcomes for these patients. Screening mammography remains the backbone for screening patients for breast cancer, with diagnostic tools subsequently deployed to confirm if an indeterminate finding warrants biopsy, surveillance, or no additional imaging. The Breast Imaging Reporting and Data System (BI-RADS^®) manual is a classification system created by the American College of Radiology to standardize breast cancer risk assessment and reporting for mammography, ultrasound, and magnetic resonance imaging (MRI) (3). Since its initial publication in 1993, the BI-RADS manual has evolved to adapt to emerging imaging techniques to improve its performance. Despite these advances, current diagnostic tools have a positive predictive value for malignancy of 20-40% for lesions meeting criteria for biopsy in accordance with BI-RADS (PPV2) (3–5). Thus, integrating new diagnostic imaging markers or techniques in the diagnostic workup of breast lesions may reduce unnecessary biopsies and patient anxiety ultimately leading to better patient care.

As solid tumors grow, proliferating tumor cells and angiogenesis induce solid stress causing compression and leakiness of blood vessels and lymphatics (6, 7). As a result, interstitial fluid pressure (IFP) becomes significantly elevated within solid tumors and matches tumor microvascular pressure (MVP) (6, 8). This can cause numerous secondary effects, such as creating outward fluid flow towards the adjacent normal tissue causing difficulty in delivery of therapeutic agents as well as vascular occlusion with ultimate necrosis facilitating cancer cell invasion (6, 9, 10, 11). In most recent studies, IFP could independently induce tumor malignancy by creating gradients of autocrine factors such as C-C chemokine ligands 19 and 21 (CCL 19 and CCL21) that promote tumor migration in the fluid flow direction via the activation of C-C chemokine receptor type 7 (CCR7) (12–14). Other findings were that IFP could induce rheotactic migration of cancer cells towards nearby stromal vasculature creating a metastatic tumor environment (7, 12, 15). Based on these findings, the levels of IFP in benign and malignant tumors are inferred to be different considering their clear difference in invasiveness and metastatic characteristics. Indeed, Nathanson and Nelson reported IFPs in invasive breast tumors (mean value of 29 mmHg) were higher than those in benign tumors (not specifically presented but mean value of around 4 mmHg) (p=0.002) (16). Hence, the IFP level could be a new marker to characterize indeterminate breast lesions.

The IFP of breast lesions can be directly measured invasively (17, 18). Since invasive methods are difficult to apply in patients, noninvasive imaging techniques such as dynamic contrast-enhanced (DCE) MRI, convection MRI, and diffusion-weighted MRI have been studied to estimate tumor IFP indirectly (18–20). However, MRI is not available to all patients due to its high cost and technique-specific limitations such as claustrophobia, metal implants, and compromised renal function (18).

Contrast-enhanced ultrasound (CEUS) is a non-nephrotoxic imaging technique using gas-filled microbubbles as a contrast agent (21, 22). CEUS is purely intravascular and can also be used for the quantification of microvascular perfusion, which is limited in Doppler-based ultrasound imaging (22, 23). Furthermore, CEUS can be used to estimate tumor IFP through subharmonic-aided pressure estimation (SHAPE) (24, 25). Microbubbles can act as pressure sensors producing multiple harmonics of the transmitted ultrasound frequency whose amplitudes vary with ambient pressure. Among the harmonics, subharmonic signals showed the highest sensitivity to pressure changes and SHAPE generally utilizes the negative linear relationship between the amplitude of subharmonic signals and hydrostatic pressure (26, 27). Thus, this study assessed the ability of the pressure gradient between breast lesions and normal adjacent tissue estimated by 3D SHAPE to characterize indeterminate breast lesions recommended for biopsy. Additionally, these 3D SHAPE pressure gradients were compared to direct, invasive pressure measurements.

MATERIALS AND METHODS

Subjects

This study was approved by our institutional review board and subjects signed informed consent. Between December 2020 and July 2023, 25 patients scheduled for breast biopsy of a breast mass at least 1 cm in the longest dimension and located within 3 cm of the skin surface were prospectively enrolled. The tumor depth was limited to 3 cm to ensure the ability to obtain direct intra-compartmental pressure measurement using a pressure monitoring system with a 6.3 cm needle (C2DX, Schoolcraft, MI, USA). Patients who were pregnant or nursing were excluded. Subjects had to be medically stable and not have known hypersensitivity or allergy to the ultrasound contrast agent used in this study (Definity; Lantheus Medical Imaging, North Billerica, MA, USA).

Ultrasound examination

Subjects underwent CEUS before the biopsy. The breast lesions were imaged using a Logiq E9 scanner with a 3D linear probe RSP6-16 (GE HealthCare, Waukesha, WI, USA) by an RDMS certified sonographer or a radiologist. This is off-label use of Definity, which is approved for echocardiograms by FDA. After confirming the lesion size and location using 2D B-mode with the same probe, subjects received a continuous IV infusion of 3 ml of Definity suspended in 50 ml saline via the antecubital vein, with infusion rates of 4-10 ml/min (titrated to effect). This dosage was selected based on previous studies (24, 28). First, optimal transmit pulse power was determined on an individual basis to achieve the highest sensitivity to pressure change for the SHAPE data collection (29). After determining the optimal transmit power, the infusion was stopped for about 5 -8 minutes for the clearance of bubbles. Then, three sets of 3D conventional CEUS (second harmonic), as well as three sets of 3D SHAPE (subharmonic) data, were collected. The conventional CEUS data were collected using the same imaging settings, such as transmit pulse power (29%) and receive gain (48 dB) for all subjects to directly compare the second harmonic signal of tumors among the subjects, while the SHAPE data were collected using individually optimized imaging settings. Then, the infusion was restarted, and another three sets of 3D conventional CEUS and 3D SHAPE data were collected. This paired data collection allowed us to compare the second and subharmonic signals from the lesion with and without contrast agent. The ultrasound exam procedure is summarized in Figure 1.

Figure 1.

Procedural workflow for the contrast-enhanced ultrasound (CEUS) exam. SHAPE: subharmonic-aided pressure estimation.

Direct pressure measurement

Several invasive methods have been used to measure IFP in tissue and tumor directly [17, 18]. In this study, the C2DX STIC intra-compartmental pressure monitoring system was used mainly due to its availability and clinically proven safety in patients (cleared by the FDA and commonly used for the assessment of intra-compartmental pressure). The device was set up using the manufacturer’s standard protocol. The patient was positioned supine, and the area of interest was localized using real-time ultrasound guidance. Using clean technique and after antiseptic preparation of the skin, the skin puncture and biopsy sites were infiltrated with 1% Lidocaine. While using ultrasound guidance, three separate pressure measurements were obtained within the central solid-appearing portion of the mass and adjacent breast tissue along the path of the needle.

Data analysis

All ultrasound data were analyzed offline using Matlab software (Mathworks, Natick, MA, USA). The volumes of conventional CEUS and SHAPE data were composed of 115~197 and 46~182 2D slices depending on the field of view (determined by lesion size and depth) and selected image quality level (three with mid-level and the rest with the highest level), respectively. The data analysis was performed only for the 2D slices including the lesion. To estimate the vascularity in the lesion, the mean second harmonic amplitude in the lesion volume was obtained from conventional CEUS data by subtracting the second harmonic signal in the lesion without Definity from that with Definity and averaging it over the lesion volume. The final mean second harmonic amplitude in the lesion was determined by the mean value of three volumes (from three separate CEUS data sets). Similarly, to estimate the pressure gradient between the lesion and the adjacent normal tissue, the gradient of mean subharmonic amplitude was obtained by subtracting the mean subharmonic amplitude of the adjacent normal tissue from that of the lesion. The region of interest (ROI) size of adjacent normal tissue was similar to the lesion size except for the cases where the lesion was too large within the field of view (e.g., Figure 2). The final value of mean subharmonic amplitude gradient was obtained by averaging the results from three SHAPE data sets. The direct pressure measurements of the breast lesion and adjacent normal tissue were also averaged over each of the three measurements.

Figure 2.

Example of the region of interest selection in the lesion (red color) and adjacent normal tissue (yellow color) on 2D plane of 3D data for subharmonic-aided pressure estimation.

Additional parameters

The lesion size was calculated as a volume assuming an ellipsoid using 3-dimensional ultrasound measurements. As a proliferation marker, Ki-67 levels were obtained from the histology results if available.

Statistical methods

Three parameters were evaluated to characterize indeterminate breast lesions: 1) the mean second harmonic amplitude in the lesion, 2) the mean subharmonic amplitude gradient between the lesion and adjacent normal tissue, and 3) the direct measurement gradient between the lesion and adjacent normal tissue. These three parameters were compared to the biopsy results using t-test and receiver operating characteristic (ROC) curve analyses. The correlation among the parameters were assessed using the Pearson correlation coefficient. Additionally, the breast lesion sizes and Ki-67 levels of the malignant lesions (as proliferation markers) were compared with the mean subharmonic amplitude gradients and direct pressure gradients using the Pearson correlation coefficient.

RESULTS

Of the 25 subjects initially enrolled, 21 subjects were ultimately included in this study. The flow chart for the study is presented in Figure 3. The subjects were all women with an age of 21~69 years (mean ± standard deviation (std): 43 ± 15 yo). There were 7 White or Caucasian participants (33%), 6 Asians (29%), 6 Blacks or African Americans (29%), and 2 Hispanics (9%). Their biopsy results showed 8 malignant (38%; six BI-RADS 4 and two BI-RADS 5) and 13 benign (62%; all BI-RADS 4) lesions. Six of the malignant lesions were grade 3 and two were grade 2. The tumor size of malignant and benign lesions were 4.3 ± 3.6 cm³ (mean ± std) and 1.9 ± 2.2 cm³, respectively (p=0.06). Three-dimensional SHAPE data were available for 21 subjects, with 3D conventional CEUS data available in 10 of them (4 malignant and 6 benign lesions) since 3D conventional CEUS data collection was added to the study protocol later.

Figure 3.

Flow chart of study participant enrollment. The SHAPE data of two subjects were mistakenly not collected with the optimal acoustic power determined during the study, specified as “wrong acoustic power”. CEUS: contrast-enhanced ultrasound, SHAPE: subharmonic-aided pressure estimation.

The summarized study results are presented in Table 1. The mean second harmonic amplitudes were not significantly different between the malignant and benign lesions (p=0.66) even though the malignant lesions showed a slightly higher average value (13.6 vs. 10.8 arbitrary unit (a.u.)). The area under the curve (AUC) in detecting malignancy was 0.54, with 100% specificity and 25% sensitivity. Examples of 3D conventional CEUS for malignant and benign lesions are shown in Figure 4. The mean subharmonic amplitude gradients between the lesion and adjacent normal tissue by 3D SHAPE were higher in the benign lesions (indicating lower pressure in the lesion given the inverse relationship of SHAPE method) than in the malignant lesions (2.86 ± 3.24 vs. −0.03 ± 1.72 a.u.; p=0.03). Examples of 3D SHAPE images for malignant and benign lesions are presented in Figure 5. The AUC for detecting malignancy with SHAPE was 0.79, with 77% specificity and 88% sensitivity. The direct intra-compartmental pressure gradients were significantly lower in the benign lesions than the malignant lesions (9.9 ± 8.5 vs. 20.9 ± 8.0 mmHg; p=0.008). The AUC for detecting malignancy was 0.85, with 92% specificity and 50% sensitivity.

Table 1.

Summary of study results (a.u.: arbitrary unit; N/A: not available).

Subject number	Direct pressure measurement (mmHg)		Difference in mean subharmonic amplitude (a.u.)		Difference in mean second harmonic amplitude (a.u.)	Biopsy result	Ki-67 level (%)
	Tumor	Tissue	Tumor	Tissue	Tumor
1	22.0	10.0	2.1	5.7	N/A	Malignant	36
2	13.0	3.0	−1.6	−11.3	N/A	Benign	N/A
4	38.0	8.7	1.2	0.4	N/A	Malignant	87
5	29.0	11.7	1.1	−4.2	N/A	Benign	N/A
6	26.3	8.3	0.2	−1.1	N/A	Malignant	28
8	12.7	11.7	−3.0	−0.9	N/A	Benign	N/A
9	5.3	13.3	4.8	−0.7	N/A	Benign	N/A
10	8.0	2.7	1.9	0.1	N/A	Benign	N/A
11	45.0	12.0	1.7	2.5	N/A	Malignant	72
12	35.3	11.0	0.0	−1.9	N/A	Benign	N/A
13	40.3	38.0	0.5	1.9	N/A	Benign	N/A
15	22.7	12.7	2.7	0.0	5.2	Benign	N/A
16	17.0	3.0	5.1	−1.2	16.7	Benign	N/A
17	35.7	16.0	1.4	1.3	14.9	Benign	N/A
19	25.0	14.7	1.0	−1.0	−0.2	Benign	N/A
20	15.3	5.3	9.9	7.4	18.6	Benign	N/A
21	26.0	11.0	4.7	2.6	30.8	Malignant	47
22	39.3	15.0	0.8	0.2	5.4	Malignant	54
23	34.0	10.0	0.0	−0.2	11.2	Malignant	30
24	24.0	12.0	2.1	−0.4	9.7	Benign	N/A
25	13.7	2.0	0.7	1.6	6.8	Malignant	81

Figure 4.

Three-dimensional conventional contrast-enhanced ultrasound images from (a) malignant lesion (white arrows) and (b) benign lesion (white arrows) without contrast agent (upper half) and with contrast agent (lower half). The transmit power and receiver gain used for both lesions were 29% and 27 dB, respectively.

Figure 5.

Three-dimensional images for subharmonic-aided pressure estimation from (a) malignant lesion (white arrows) and (b) benign lesion (white arrows) without contrast agent (upper half) and with contrast agent (lower half). The transmit power and receiver gain used were 35% and 44 dB for (a), respectively and were 35% and 48 dB for (b), respectively. The images were from the same lesions presented in Figure 4.

A weak negative correlation existed between the subharmonic amplitude gradients and directly measured pressure gradients (r = −0.2), while a weak positive correlation was found between the subharmonic amplitude gradients and second harmonic amplitudes in the lesions (r = 0.2). There was no correlation between the direct pressure gradients and second harmonic amplitudes in the lesions (r = 0.05). The lesion volumes had weak negative (r = −0.2) or positive (r = 0.3) correlation with the subharmonic amplitude gradients or the gradient by direct measurement, respectively. Finally, the Ki-67 levels of the malignant lesions showed no (r = 0) or weak correlation (r = 0.3) with the subharmonic amplitude gradients or the gradients by direct measurements, respectively.

DISCUSSION

Previously, IFP was mainly studied to predict prognosis and response to cancer therapy (18). This is because elevated IFP was thought to promote disease progression primarily by acting as a barrier for drug delivery until recent studies found its independent role in promoting malignancy (12). In fact, a low IFP before the treatment or reduction of IFP induced by therapy was correlated with better cancer outcomes in most clinical studies (24, 30–33). Since most studies including mathematical and pre-clinical models focused on malignant lesions, there is limited knowledge of the IFP of benign lesions. Thus, this study aimed to evaluate IFP in the characterization of indeterminate breast lesions.

Conventionally, malignant lesions were known to evolve from a normal to malignant phenotype through a multi-step transition and progress through a favorable tumor microenvironment for proliferation and metastasis (34). While it is unknown if benign lesions follow a similar initiation and progression process, a pre-clinical study found a difference in tumor microenvironment between benign and malignant breast tumors (35). The benign breast tumor stroma had a lower number of cancer-associated fibroblasts and a higher number of endothelial cells with genes only identified in the benign tumor. This finding supports that malignant lesions have favorable microenvironments for proliferation and invasiveness. Cancer-associated fibroblasts promote tumor invasiveness, and a lack of vascular supply relating to strong proliferation causes large hypoxic or necrotic areas (35). Notably, these signaling-mediated, multidirectional interactions are facilitated by interstitial fluid, whose genesis, turnover, and drainage depend on many factors such as tumor type, grade, stage, and composition of the tumor microenvironment (36). The interstitial fluid itself was suggested as a serologic biomarker to identify breast cancer subtype (36) and IFP has emerged as a biomechanical driver affecting therapeutic outcomes as well as promoting tumor invasion (7, 14).

Elevated tumor IFP affects the migration of tumor cells, which can increase tumor invasiveness (37). Interestingly, the migration directions of tumor cells were in two opposite directions: upstream and downstream of fluid (38, 39). The downstream migration is created by the IFP gradient from the tumor core to the periphery, and its velocities are about 0.001-0.004 mm/s (38). Studies with in vitro microfluidic models have shown that the relative fraction of this migration was affected by cell density, CCR 7 activity, and interstitial flow velocity (related to IFP) (13, 14, 40). The upstream or rheotactic migration was also created by elevated IFP competing with the mechanism for downstream migration (14). This upstream migration was induced by asymmetric fluidic tension on the cells and facilitated by high cell density (12, 15). The upstream migration of cancer cells can promote metastasis by draining into the nearby stromal vasculature (12).

Another motivation to evaluate the IFP within breast lesions was its correlation and interdependence with tissue stiffness. Previously, the stiffness of a breast lesion has shown potential to be a diagnostic marker and BI-RADS for ultrasound includes the lexicon for elasticity assessment. Elevated IFP causes tissue deformation, changes in the temporal and spatial distribution of strains, and increased stiffness (18, 41, 42). Tissue stiffness represented by extracellular matrix stiffness is affected by the change in collagen content and organization, which also influences hydraulic conductivity and interstitial transport in tumors (17, 43).

In this study, the pressure gradient between the lesion and adjacent normal tissue was utilized instead of the pressure in the lesion itself to normalize the values across the subjects. The IFP and MVP in adjacent normal tissue were used as a reference for the direct measurement and SHAPE of IFP in the lesion, respectively. Our direct pressure gradient measurements showed that the IFP in the lesion was always higher than that in the normal tissue regardless of lesion phenotype except for one benign lesion. Also, the IFP gradients between the malignant lesion and normal tissue were about 10 mmHg higher on average than those between the benign lesion and normal tissue. This agrees with the direct measurement results by Nathanson and Nelson (16), though the difference between the malignant and benign lesions was higher in their study. This could be because our study had more cases of fibroadenoma having a higher IFP compared to other benign lesions (16).

The subharmonic amplitude gradient between the malignant lesion and normal tissue was lower than that between the benign lesion and normal tissue. This may imply that IFP in the malignant lesion were higher than that in the benign lesion assuming the MVP in adjacent normal tissue were similar in both lesions. Normal tissue capillary pressure is known to be around 10-22 mmHg (44) and studies with mathematical models of vascular and interstitial fluid dynamics demonstrated that IFP can never exceed the average capillary pressure even in tumors with very leaky vessels (45, 46), though high IFP above 30 mmHg have been reported from breast cancers (16). Thus, the high tumor IFP reflects the high vascular blood flow resistance as well as capillary pressure (45, 46). There have been mixed results on the correlation between IFP and tumor size, as well as between IFP and cell proliferation. Generally, previous studies showed a positive correlation between IFP and tumor size (10, 16, 47, 48), but others found no correlation (49–51). Our study showed weak positive correlation (r = 0.2~0.3) between the IFP and tumor volume. In vitro studies showed that there was positive or negative correlation between cell proliferation and applied hydrostatic pressure depending on cell type (51, 52). However, another study found no relationship between cell proliferation and IFP (53). In this study, the proliferation marker Ki-67 had no or weak (r = 0.3) correlation with the pressure gradient obtained by SHAPE or by direct measurements, respectively.

One limitation of this study was the small sample size, especially for the vascular evaluation. The vascularity assessment is also included in ultrasound BI-RADS and previous studies have suggested CEUS is a more sensitive and dynamic vascular assessment tool for breast lesions (28, 54, 55). Generally, strong internal vascularity was observed in malignant lesions (28, 55). Since this study focused on SHAPE utilizing the infusion of contrast agent, the dynamic perfusion of blood flow was not assessed. Additionally, the vascularity in the lesions was evaluated using conventional second harmonic CEUS, although a previous study demonstrated the superiority of subharmonic imaging for breast cancer characterization (56). This was because our subharmonic imaging settings were optimized for SHAPE. An increase in tumor IFP is directly related to unregulated angiogenesis, which is characterized by structural and functional abnormalities in vascular and lymphatic vessels (57). The weak correlation between IFP and vascularity in this study could be because our vascularity analysis results mainly reflect vascular density. In other studies, IFP and vascular density showed no or little tumor type-specific relationship (53, 58). However, our small sample size could be the cause.

This pilot study indicates a potential for IFP to aid in the characterization of indeterminate breast lesions. While the indirect measurements by SHAPE demonstrated slightly lower diagnostic performance than the invasive measurements, it is nonetheless highly encouraging considering it is a noninvasive technique. In the future, standardization of SHAPE analysis by the assessment of observer variability as well as improvement of SHAPE sensitivity using monodisperse microbubbles will be investigated (59). Also, smaller (< 1 cm) and deeper lesions (> 3 cm) will be included. In this study, there were limitations on lesion size and depth imposed in order to perform the direct pressure measurement safely during the ultrasound-guided core needle biopsy. Combining IFP with other diagnostic markers will also be evaluated for better characterization of breast lesions (54). Thus, further investigation is warranted to confirm that the proper deployment of the SHAPE technology can help reduce unnecessary breast biopsies and improve the positive predicative value of diagnostic breast imaging.

DATA AVAILABILITY

The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.