Advertisement

Breast Cyst Fluid Analysis Correlations with Speed of Sound Using Transmission Ultrasound

Open AccessPublished:June 07, 2018DOI:https://doi.org/10.1016/j.acra.2018.03.027

      Rationale and Objectives

      The purpose of this work is to determine if the speed of sound value of a breast cyst can aid in the clinical management of breast masses. Breast macrocysts are defined as fluid-filled tissue masses >1 cm in diameter and are thought to be aberrations of normal development and involution, often associated with apocrine metaplasia. The benign natural history of breast cysts is well known, and it is important to obtain high specificity in breast imaging to avoid unnecessary biopsies in women who have benign diseases, particularly those with dense breast tissue. Transmission ultrasound is a tomographic imaging modality that generates high-resolution, 3D speed of sound maps that could be used to identify breast tissue types and act as a biomarker to differentiate lesions. We performed this study to investigate the microanatomy of macrocysts observed using transmission ultrasound, as well as assess the relationship of speed of sound to the physical and biochemical parameters of cyst fluids.

      Materials and Methods

      Cyst fluid samples were obtained from 37 patients as part of a case-collection study for ultrasound imaging of the breast. The speed of sound of each sample was measured using a quantitative transmission ultrasound scanner in vivo. Electrolytes, protein, cholesterol, viscosity, and specific gravity were also measured (in the aspirated cyst fluid) to assess their relationship to the speed of sound values obtained during breast imaging.

      Results

      We found positive correlations between viscosity and cholesterol (r = 0.71) and viscosity and total protein × cholesterol (r = 0.78). Additionally, we performed direct cell counts on cyst fluids and confirmed a positive correlation of number of cells with speed of sound (r = 0.74). The speed of sound of breast macrocysts, as observed using transmission ultrasound, correlated with the cytological features of intracystic cell clumps.

      Conclusion

      On the basis of our work with speed as a classifier, we propose a spectrum of breast macrocysts from fluid-filled to highly cellular. Our results suggest high-speed cysts are mature macrocysts with high cell counts and many cellular clumps that correlate with cyst microanatomy as seen by transmission ultrasound. Further studies are needed to confirm our findings and to assess the clinical value of speed of sound measurements in breast imaging using transmission ultrasound.

      Key Words

      INTRODUCTION

      Biology of Breast Cysts

      The normal breast consists of segments (lobes) drained by collecting ducts. The segments consist of lobules composed of terminal ducts, acini and their supporting stroma (
      • Moinfar F
      Essentials of diagnostic breast pathology: a practical approach.
      ). Breast macrocysts are reviewed by Hughes and are defined as fluid-filled tissue masses >1 cm in diameter (
      • Hughes LE
      • Bundred NJ
      Breast macrocysts.
      ). They were first described in 1829 and are thought to be aberrations of normal development and involution, often associated with apocrine metaplasia (
      • Cooper AP
      Illustrations of the diseases of the breast…In two parts.
      ,
      • Hughes LE
      • Mansel RE
      • Webster DJ
      Aberrations of normal development and involution (ANDI): a new perspective on pathogenesis and nomenclature of benign breast disorders.
      ). Macrocysts are seen in 21% of women at postmortem and in over 10,000 breast biopsies for benign disease of which 23% had macrocysts (
      • Davis HH
      • Simons M
      • Davis JB
      Cystic disease of the breast: relationship to carcinoma.
      ,
      • Dupont WD
      • Page DL
      Risk factors for breast cancer in women with proliferative breast disease.
      ). Cysts are manifestations of lobular involution and their origin from the breast lobule has been demonstrated using histochemical techniques (
      • Azzopardi JG
      • Ahmed A
      • Millis RR
      Problems in breast pathology.
      ). They are usually lined by a single layer of epithelium which has proteins found only in apocrine epithelium (
      • Bundred NJ
      • Miller WR
      • Walker RA
      An immunohistochemical study of the tissue distribution of the breast cyst fluid protein, zinc alpha 2 glycoprotein.
      ). Evidence that active secretion is responsible for cyst formation has come from analysis of cyst fluid (
      • Bradlow HL
      • Skidmore FD
      • Schwartz MK
      • et al.
      Cation levels in human breast cyst fluid.
      ). The benign natural history of breast cysts is well known in these women with a low incidental risk for cancer (
      • Dupont WD
      • Page DL
      Risk factors for breast cancer in women with proliferative breast disease.
      ,
      • Bhate RD
      • Chakravorty A
      • Ebbs SR
      Management of breast cysts revisited.
      ). Thus, it is important to have high specificity in breast imaging to avoid unnecessary biopsies in women who have benign diseases, particularly in those women with dense breast tissue (
      • Know Error
      DNA Specimen Provenance Assay - DSPA Testing.
      ,
      • Ong MS
      • Mandl KD
      National expenditure for false-positive mammograms and breast cancer overdiagnoses estimated at $4 billion a year.
      ).

      Transmission Ultrasound

      Quantitative transmission (QT) ultrasound is an imaging modality based on tomographic techniques extended to ultrasound. In such a system, images are generated using both reflection and transmission techniques. While transmission ultrasound has been investigated as an adjunct to mammography for quite some time, recent developments in hardware and imaging algorithms have enabled marked improvements in spatial resolution and clinical utility (
      • Glover GH
      Computerized time-of flight ultrasonic tomography for breast examination.
      ,
      • Greenleaf JF
      • Bahn RC
      Clinical imaging with transmissive ultrasonic computerized tomography.
      ). Physically, a transmitter and receiver pair is colocated with multiple reflection transducers with various focal lengths in a U-shaped arrangement (
      • Wiskin J
      • Borup D
      • Johnson S
      • et al.
      Three-dimensional nonlinear inverse scattering: quantitative transmission algorithms, refraction corrected reflection, scanner design and clinical results.
      ,
      • Andre M
      • Wiskin J
      • Borup D
      Clinical results with ultrasound computed tomography of the breast.
      ,
      • Andre M
      • Wiskin J
      • Borup D
      • et al.
      Quantitative volumetric breast imaging with 3D inverse scatter computed tomography.
      ). The transmitter emits a plane wave that is received by the receiver with multiple acquisitions at frequencies ranging from 300 kHz to 1.5 MHz as the U-channel is rotated 360° around the subject. Once acquired, the projection information is reconstructed using nonlinear inverse scattering in 3D (
      • Wiskin J
      • Borup DT
      • Johnson SA
      • et al.
      Non-linear inverse scattering: high resolution quantitative breast tissue tomography.
      ). The result of this reconstruction is a quantitative volume map of speed of sound (measured at 1.5 MHz), with units of meters per second (m/s). In reflection ultrasound imaging, each of the three reflection transducers (4 MHz center frequency) with different focal lengths are alternately fired between transmission measurements in a B-mode acquisition. The resulting images are compounded together and corrected for refraction using the speed map computed in the transmission phase. This compounding produces a nonquantitative image that is proportional to impedance mismatch, referred to simply as reflection units (RUs). The result of each scan is a 3D volume of speed and reflection. These image stacks are precisely coregistered since they are acquired at the same time, so they can be put together to form a 3D view of the object in the field of view. The imaging system can image human breast tissue anatomy with high spatial and contrast resolution, while the speed of sound information corresponds well with tissue specificity (
      • Lenox MW
      • Wiskin J
      • Lewis MA
      • et al.
      Imaging performance of quantitative transmission ultrasound.
      ,
      • Klock JC
      • Iuanow E
      • Malik B
      • et al.
      Anatomy-correlated breast imaging and visual grading analysis using quantitative transmission ultrasoundTM.
      ,
      • Malik B
      • Klock J
      • Wiskin J
      • et al.
      Objective breast tissue image classification using quantitative transmission ultrasound tomography.
      ).

      Mass Visualization Using Transmission Ultrasound

      We have previously shown that the speed of sound, as measured by QT ultrasound, can both define tissue types and distinguish cystic from solid masses with high specificity in a clinical setting (
      • Malik B
      • Klock J
      • Wiskin J
      • et al.
      Objective breast tissue image classification using quantitative transmission ultrasound tomography.
      ,
      • Iuanow E
      • Smith K
      • Obuchowski NA
      • et al.
      Accuracy of cyst versus solid diagnosis in the breast using quantitative transmission (QT) ultrasound.
      ). Traditional reflection ultrasound imaging of cysts reveals either an anechoic interior (no internal reflecting elements visible) for simple cysts or varying degrees of internal reflecting elements visible for complicated or complex cysts. When transmission ultrasound images these same “anechoic” cysts by reflection, there are variations in the internal speed of sound (from 1540 m/s to approximately 1575 m/s) with correlated small (∼200 µm) foci of higher speed areas within the simple cyst. By making an accurate correlation with anatomy shown by transmission ultrasound, physicians can better interpret the type of breast mass visualized (fluid-filled cyst, highly cellular cyst, or solid mass). The current study was designed to aid in the interpretation of “low speed” simple cysts without high-speed foci and “high speed” simple cysts with high-speed foci.

      Cyst Fluid Analysis by Others

      It is difficult to find studies of the physical properties of human breast cyst fluid (ie, specific gravity or viscosity), but a number of studies measuring electrolytes, proteins, and hormones can be found (
      • Bundred NJ
      • Miller WR
      • Walker RA
      An immunohistochemical study of the tissue distribution of the breast cyst fluid protein, zinc alpha 2 glycoprotein.
      ,
      • Sartorius OW
      The biochemistry of breast cyst fluids and duct secretions.
      ,
      • Tsung JS
      • Wang TY
      • Wang SM
      • et al.
      Cytological and biochemical studies of breast cyst fluid.
      ,
      • Dixon JM
      • Miller WR
      • Scott WN
      • et al.
      The morphological basis of human breast cyst populations.
      ,
      • Sánchez LM
      • Díez-Itza I
      • Vizoso F
      • et al.
      Cholesterol and apolipoprotein D in gross cystic disease of the breast.
      ,
      • Dogliotti L
      • Orlandi F
      • Torta M
      • et al.
      Cations and dehydroepiandrosterone-sulfate in cyst fluid of pre- and menopausal patients with gross cystic disease of the breast. Evidence for the existence of subpopulations of cysts.
      ). Variation in Na+, K+, and Na+/K+ ratios has been shown to correlate with typing by histologic examination of cells lining the cysts and correlate with a degenerative rather than secretory process (
      • Sartorius OW
      The biochemistry of breast cyst fluids and duct secretions.
      ). To our knowledge, no measurements of human breast cyst fluid that correlate with speed of sound have been published. The current prospective study was an extension of our retrospective investigation into the specificity of the speed of sound in transmission ultrasound of the breast for determining the presence of cystic or solid mass. Additionally, it was done to further clarify the radiologist's interpretation of cysts with high-speed foci observed by transmission ultrasound (
      • Iuanow E
      • Smith K
      • Obuchowski NA
      • et al.
      Accuracy of cyst versus solid diagnosis in the breast using quantitative transmission (QT) ultrasound.
      ).

      METHODS

      Samples

      In this prospective study, we collected cyst fluid from patients in order to assess the relationship of speed of sound to the physical and biochemical parameters of the cyst fluids. All cyst fluid samples were obtained as part of a case-collection study at two academic institutions: George Washington University in Washington DC and Elizabeth Wende Cancer Center in Rochester, New York. The study was approved by the western IRB and is registered with ClinicialTrials.gov (https://clinicaltrials.gov/NCT02133417). Inclusion criteria consisted of any patient with an abnormality on their screening mammogram that was considered by the interpreting breast radiologist to be a mass. There were 207 cases collected: 38 fibroadenomas, 57 cysts, 46 cancers, and 63 that turned out to be normal. However, not all patients with cysts agreed to have cyst aspiration and in some cases the collected fluid was not sufficient for a full analysis. Therefore, cyst fluids from only 37 patients were available for analysis.
      The following chemical tests were done on each cyst fluid sample: sodium, potassium, total protein, and cholesterol. Specific gravity and viscosity were also performed on each sample. Thin preparations were made of the cyst fluid for cell counts and morphology. Speed of sound for each cystic mass was measured using the QT Viewer workstation, a proprietary viewer that is a part of the QT Ultrasound Breast Scanner-1 (QT Ultrasound, Novato, California). The viewer has been independently validated for accuracy of the speed measurement.

      Chemical Analysis

      All samples were appropriately preserved and shipped to Strong Memorial Hospital University of Rochester Medical Center in Rochester, New York, for analysis. Most of the chemistry assays were performed on a Roche Cobas 8000 Modular Analytics System (

      Cobas 8000 modular analyzer series [cited 2017 October 9] Available at: https://usdiagnostics.roche.com/en/core_laboratory/instrument/cobas-8000-analyzer-series.html.

      ).
      Total protein was performed according to the in vitro test for the quantitative determination of total protein in human serum and plasma (). Cholesterol was measured using the Cholesterol Gen.2 assay (

      In vitro test for the quantitative determination of cholesterol in human serum and plasma on Roche/Hitachi cobas c systems. [cited 2017 October 9] Available at: https://usdiagnostics.roche.com/products/03039773190/PARAM41/overlay.html.

      ). Specific gravity was measured using the specific gravity test module (
      • Sánchez LM
      • Díez-Itza I
      • Vizoso F
      • et al.
      Cholesterol and apolipoprotein D in gross cystic disease of the breast.
      ). Viscosity was measured using the capillary method (
      • Cooke BM
      • Stuart J
      Automated measurement of plasma viscosity by capillary viscometer.
      ).

      Cytological Analysis

      Slide preparation of the cyst fluid was done using the ThinPrep Processor (

      ThinPrep Pap Test [cited 2017 October 9] Available at: https://healthdxs.com/en/thinprep/.

      ). The ThinPrep Processor processes all samples in a similar fashion and deposits cells from a similar sample size within a 20-mm circle on the microscope slide, permitting sample comparisons using uniform methods of cell deposition. Slide staining was done according to the method of Papanicolaou (
      • Papanicolaou GN
      • Traut HF
      Diagnosis of uterine cancer by the vaginal smear.
      ). Sodium and potassium analysis was done using standard ion-specific electrodes on the ISE Module of the Roche Cobas 8000 system (

      510(k) Summary: Cobas 8000 ISE Module, Urine Sample Type [cited 2017 October 9] Available at: https://www.accessdata.fda.gov/cdrh_docs/pdf12/K123726.pdf.

      ). Cell counting on the thin-prep and stained slides was done in a semiautomated method, using open-source image analysis software DotCount v1.2 and ImageJ (NIH, Bethesda, Maryland) (

      Image Analysis DotCount v1.2 [cited 2017 October 9] Available at: http://reuter.mit.edu/software/dotcount/.

      ). For each slide, nine nonoverlapping images were acquired, which covered over 80% of the sample area. The images were stitched before cell counting was performed. A representative single image is shown in Figure 1.
      Parameters for the software were similar for all counting sessions and the scores results of the software were used for the analysis. All results of the chemical testing were provided by a spreadsheet from University of Rochester Medical Center and validated by an independent clinical studies team.

      Statistical and Correlation Analysis

      Correlation analysis was performed between speed of sound values and multiple chemical concentrations within the cyst aspirate. In addition, correlation analysis was performed between speed of sound and cell count performed on ThinPrep slides. In all instances, “correlation” is defined as the Pearson product-moment correlation coefficient (r), which measures the strength of the linear relationship between two variables. We also tested whether speed of sound shows any relationship as a function of cyst size. We defined macrocysts as larger than 10 mm in diameter. We performed the nonparametric Mann–Whitney U test to assess if speed of sound can be used as a predictor of cyst size. In addition, we performed linear discriminant analysis with leave-one-out cross validation scheme to test if cysts can be classified as large or small based on the speed of sound. All analysis was performed using JMP and Microsoft Excel software. Graphing was performed with OriginPro software.

      Case Selection and Exclusion

      A flow chart of the case selection process is shown in Figure 2. Cytological preparation was complete and available for visualization and cell counts on 82 cases from the case-collection study. Out of these 82 cases, fluid for 37 cysts were collected for the study. Thirty cysts had the entire panel of chemical tests done, but in seven cases the amount of fluid for testing was not sufficient to perform all tests. In order to look for a correlation between the speed of sound value of the cyst interior and the presence of punctate, high-speed foci within the cyst fluid, we selected cysts that were large enough to measure the speed of sound with statistical accuracy. For correlating speed of sound with cyst size, we selected a total of 52 cases based on the availability of size information and on the clarity of QT images to allow clear demarcation of the size of the cyst. Within these 52 cases, 41 cases were identified to have large macrocysts (size >10 mm), and 11 cases were identified to have small cysts (size <10 mm). For correlative analysis of speed of sound with cell counts, we selected 33 of these 52 cases of cyst samples. Cases were excluded due to the presence of large sized clumps of cells in the ThinPrep slides and/or due to the presence of more than one slide for one study where we were unable to confirm the validity of the data point.
      Fig 2
      Figure 2Panel A (left) showing exfoliated cyst lining epithelial cells and panel B (right) showing clumps of exfoliated cyst epithelial cells.

      RESULTS

      Chemical Correlations

      The chemical analysis results from 30 patient samples are shown in Table 1. Figure 3 summarizes the correlations between these various chemical parameters, and Table 2 shows the respective correlation coefficients for the scatterplots shown in Figure 3.
      Table 1Chemical analysis results from 30 patient samples
      CystSodiumPotassiumNa/K RatioTotal ProteinCholesterolViscositySp. Gravity
      11394.332.83.36911.711.033
      22299.60.22.83421.361.039
      34486.20.52.86561.61.039
      41364.431.22.813692.921.035
      524980.21.7489QNS1.033
      6<20140.2<0.12.3291.391.035
      71404354.13801.551.032
      83988.70.41.23761.21.03
      91404.134.54.76371.961.036
      10<18110.2<0.22.5559QNS1.038
      1165770.81.77662.431.033
      1224107.10.21.35631.171.03
      132294.10.21.7695QNS1.035
      1413715.19.11.6606QNS1.03
      158565.71.32.88101.611.04
      1647108.90.42.14421.431.035
      171373.934.63.44991.231.031
      182487.90.31.98412.491.038
      191363.934.728611.891.028
      20311170.31.46531.281.032
      2123119.70.22.4886QNS1.041
      22137622.91.57281.551.026
      234087.10.50.82851.251.024
      241354.430.84.413697.971.042
      2510638.12.81.97441.371.03
      2694>54.6>1.72.19683.121.036
      278363.91.32.46652.521.033
      284991.30.528053.641.034
      29<20120.9<0.22.18203.81.035
      3014014.89.41.48832.621.03
      Sodium/potassium = mmol/L; Protein = g/L; Cholesterol = mg/dL; Viscosity = centipoise; Specific Gravity = g per volume of cyst fluid/1 g per volume of water. QNS = the quantity provided was not sufficient to perform the analysis.
      Fig 3
      Figure 3Correlation plots between the various chemical parameters. The data points correspond to the data in . In addition to the parameters in , the parameter of “Protein × Cholesterol” has also been included. Each of the plots includes 95% density ellipses as well, which means that the ellipses enclose approximately 95% of the points. The narrowness of the ellipses is a reflection of the degree of correlation of the variables. (Color version of figure is available online.)
      Table 2Values of Pearson Product-Moment Correlation Coefficients (r) for Scatterplots Shown in Figure 3 and Data in Table 1. An Additional Variable “Protein × Cholesterol” was Added, Which is the Product of Protein and Cholesterol
      SodiumPotassiumNa/K ratioTotal ProteinCholesterolViscositySp. GravityProt. × Chol.
      Sodium1.00
      Potassium−0.981.00
      Na/K ratio0.83−0.871.00
      Total protein0.44−0.450.631.00
      Cholesterol0.39−0.390.320.221.00
      Viscosity0.12−0.130.180.360.711.00
      Sp. gravity−0.260.26−0.150.540.350.481.00
      Prot. × chol.0.47−0.480.550.710.790.780.541.00
      In general, we observed no strong meaningful correlations between the individual chemical parameters, except for sodium and potassium concentrations that are known to have a strong inverse relationship. We also observed a correlation between cholesterol and viscosity, and between total protein × cholesterol and viscosity.

      Speed of Sound Correlations

      We observed a correlation between speed of sound in the cyst in vivo and the specific gravity of the aspirated cyst fluid, as shown in Figure 4.
      Fig 4
      Figure 4Plot showing correlation between speed of sound as measured by QT and specific gravity of cyst fluid. The value of r denotes the Pearson correlation coefficient.
      Because of the correlations observed between specific gravity and the product of protein and cholesterol (ie, cell membrane components) and between specific gravity and speed, we reasoned that cell count could be a factor in the in vivo cyst speed. As shown in Figures 5 and 6, protein × cholesterol correlated with cyst cell count as did cell number and speed.
      Fig 5
      Figure 5Plot of correlation between cell count and ‘protein × cholesterol’. The value of r denotes the Pearson correlation coefficient.
      Fig 6
      Figure 6Plot of speed of sound as measured by QT as a function of cell count. The correlation coefficient "r" was calculated to be 0.74.
      In Figure 6 a moderate to high linear relationship between speed of sound and the cell count (r = 0.74) is seen.

      Cyst Classification

      We observed that the larger macrocysts had on average a lower mean speed of sound value (1550.6 ± 8.7 m/s) in comparison to smaller cysts (1566.0 ± 22.2 m/s). The nonparametric Mann–Whitney U test showed significant difference (p < 0.05) between the speed of sound values of the two groups based on size. The fluid in the larger macrocysts (>10 mm diameter) had a nonhomogeneous or “speckled” appearance (Fig 7), whereas the smaller cysts (<10 mm diameter) had a more homogeneous appearance (Fig 8).
      Fig 7
      Figure 7Sagittal view speed of sound image showing (with crosshairs) a macrocyst (cyst# 9 in ) with an average speed value of 1554 m/s. (Color version of figure is available online.)
      Fig 8
      Figure 8Coronal view of a speed of sound image showing a macrocyst (∼7 mm) with an average speed value of 1538 m/s (cyst#15 in ). (Color version of figure is available online.)
      We then tested the ability of speed of sound to predict cyst size. As mentioned previously, we used linear discriminant analysis with a leave-one-out cross validation scheme. The confusion matrix is shown in Table 3. Overall, the classifier showed an accuracy of 80.3%.
      Table 3Classification Summary (Confusion Matrix) as Generated by Linear Discriminant Analysis Using Leave-One-Out Cross Validation Scheme
      Predicted Group
      Large CystsSmall CystsAccuracy
      Actual groupLarge cysts36588%
      Small cysts3873%
      As mentioned previously, the cysts with larger clumps were excluded from the speed of sound versus cell count correlation. The reason for this exclusion is given in the following discussion. However, speed of sound of cysts with clumps was found to be significantly different than cysts with regularly spread of cells. Specifically, the average speed of sound of cysts with regularly spread of cells was measured to be 1551.9 ± 12.7 m/s, whereas cysts with clumps were measured to be 1560.4 ± 15.4 m/s.

      Discussion

      In our chemical analysis of cyst fluid, we found positive correlations between viscosity and cholesterol (r = 0.71) and viscosity and total protein × cholesterol (r = 0.78). In our cytological analysis of breast cyst fluid, we showed a positive correlation between direct cell counts on cyst fluids and speed of sound. Furthermore, the speed of sound of breast macrocysts, as observed using transmission ultrasound, correlated well with the cytological features of intracystic cell clumps.
      Our results of cyst fluid Na+, K+ and Na+/K+ ratios are consistent with other studies (
      • Tsung JS
      • Wang TY
      • Wang SM
      • et al.
      Cytological and biochemical studies of breast cyst fluid.
      ,
      • Dixon JM
      • Miller WR
      • Scott WN
      • et al.
      The morphological basis of human breast cyst populations.
      ,
      • Sánchez LM
      • Díez-Itza I
      • Vizoso F
      • et al.
      Cholesterol and apolipoprotein D in gross cystic disease of the breast.
      ,
      • Dogliotti L
      • Orlandi F
      • Torta M
      • et al.
      Cations and dehydroepiandrosterone-sulfate in cyst fluid of pre- and menopausal patients with gross cystic disease of the breast. Evidence for the existence of subpopulations of cysts.
      ,

      Cobas 8000 modular analyzer series [cited 2017 October 9] Available at: https://usdiagnostics.roche.com/en/core_laboratory/instrument/cobas-8000-analyzer-series.html.

      , ), 14 of the 30 samples had a high Na+/K+ ratio. The cholesterol levels in cyst fluid in this study (17.9 ± 6.8 mmol/L) are also consistent with other studies (15 ± 6 mmol/L) (
      • Dogliotti L
      • Orlandi F
      • Torta M
      • et al.
      Cations and dehydroepiandrosterone-sulfate in cyst fluid of pre- and menopausal patients with gross cystic disease of the breast. Evidence for the existence of subpopulations of cysts.
      ).
      The correlation of cell count with speed of sound in macrocysts is consistent with the presence of cell number observed cytologically. This is not surprising given the high-fidelity nature and spatial resolution of transmission ultrasound, and its ability to provide quantitative measurements (
      • Lenox MW
      • Wiskin J
      • Lewis MA
      • et al.
      Imaging performance of quantitative transmission ultrasound.
      ,
      • Klock JC
      • Iuanow E
      • Malik B
      • et al.
      Anatomy-correlated breast imaging and visual grading analysis using quantitative transmission ultrasoundTM.
      ). By way of theory, closely packed structures (such as cells in a suspension) exhibit higher effective refractive index (
      • Ishimaru A
      Wave propagation and scattering in random media.
      ). Hence, higher cell count result in higher value of refractive index, with a consequent increase in speed of sound within the cyst. We would like to highlight the importance of this finding since no other mesoscopic imaging modality, without the use of a contrast mechanism, has the ability to capture such quantitative variation as a function of cell count.
      Very little is known about the speed of sound of human breast tissue subtypes. Also, any system for doing so, must account for acoustic impedance, temperature and the fluid medium used to conduct the sound to allow comparison to other published values. Ophir and Jaeger examined the speed of sound in 10 human tissues in polyethylene glycol–ethanol–water solution at 21.5°C and compared the speed with their densities (
      • Ophir J
      • Jaeger P
      A ternary solution for independent acoustic impedance and speed of sound matching to biological tissues.
      ). Although there was no measurement of breast tissue, splenic tissue was the closest to our measurements for breast parenchyma. That study also found a correlation between density and speed with similar value ranges to our analysis. This is also in line with the fact that higher density usually results in higher refractive index, thereby increasing the speed of sound, as noted in our study.
      In a previous publication, using a validated system and speed calculation methodology, we have measured the speed of sound in breast tissue subtypes (
      • Malik B
      • Klock J
      • Wiskin J
      • et al.
      Objective breast tissue image classification using quantitative transmission ultrasound tomography.
      ). Using discriminant analysis, we have shown that speed of sound was the most important contributor toward the classification: with an accuracy rate of >85% when distinguishing between the five tissue classes in the breast (glands, ducts, skin, connective tissue, and fat). In the current study, we have extended this analysis to use linear discriminant analysis as a predictor of cyst size and type. When using speed of sound as a predictor of cyst size (to differentiate between two groups: cysts >10 mm and cysts <10 mm), the classifier performed at an overall accuracy of >80%. We then tested the ability of speed of sound to predict the type of cyst (regular vs clump). While the speed of sound values showed statistically significant difference, the discriminant analysis or logistic regression classifiers were not able to differentiate between the two classes with high accuracy; accuracy was calculated to be 60.1%.
      When comparing the speed of sound in cysts with a “regular” spread of cells in the ThinPrep slides with those with cell “clumps”, we must be careful in interpreting the results from slides with clumps. ThinPrep machine collects the cells on the surface of a membrane with small pores (
      • Kalinicheva T
      • Frisch N
      • Giorgadze T
      • et al.
      Etiologic factors related to unsatisfactory ThinPrep® cervical cytology: evaluation and potential solutions to improve.
      ). The pores are sufficiently small to aspirate the liquid from vial while trapping the cells on its surface. Debris material such as inflammatory cells, large clumps, and contaminants can block the holes thus preventing the collection of enough epithelial cells (used for diagnostics) onto the filter membrane. Such blockage of the membrane can interfere with adequate collection consistency, which can greatly impact the variance of cell count in slides with clumps. Therefore, we did not include the cell count from slides with clumps in our speed of sound correlation. However, this did not prevent us from pooling together the speed of sound values from clumps to gather valuable information on the relationship of the number of cells with higher speed of sound. As noted previously, there was significant difference in the speed of sound values of cysts with “regular” spread of cells (ie, fluid cysts) and in cysts showing clumping of cells.

      Limitations of the Study

      The main limitations of the study include the possibility of a bias due to the small number of cases studied. Although there were 37 cases with chemical analysis and 33 cases with cytologic analysis, the exclusions were for technical reasons (inadequate sample volume, very small cysts, and ThinPrep technical issues). Nevertheless, the statistical analysis confirmed the sample size was adequate. Since much of the data was visually analyzed, there is a risk of reader bias, however all cytological and image viewing was done blindly (without knowledge of any information from the nonvisual datasets).

      CONCLUSIONS

      The speed of sound of breast macrocysts as observed using transmission ultrasound correlates well with biological and cytological features consistent with the spectrum of cystic masses observed clinically. Diagnostic breast radiologist guidelines would include that cysts with “high-speed” foci will have a higher speed of sound and that those foci are likely caused by clumps of cells. To our knowledge, this is the first study to correlate speed of sound as gathered from transmission ultrasound as a function of cyst type and size. Further studies are needed to confirm our findings and to determine the clinical value of speed of sound measurements in breast imaging using transmission ultrasound.
      The clinical relevance of our work is that transmission ultrasound breast imaging will observe a spectrum of “low-speed” (1540–1569 m/s) cystic masses within the breast that can show (1) homogeneous interiors with low speed (fluid-filled cysts), (2) a homogeneous interior with high speed (cysts containing free-floating cells), and/or (3) a heterogeneous interior with many punctuate areas of high speed (cysts with large cell clumps). The present work will allow the breast imaging radiologists to have a basis for describing breast macrocysts and for following changes in these masses over time.

      ACKNOWLEDGMENTS

      This work was supported in part by a grant from the National Cancer Institute (National Institutes of Health grant R01 CA138536). We would like to thank Dr. Elaine Iuanow and Alyson Terry for technical assistance and helpful discussions.

      References

        • Moinfar F
        Essentials of diagnostic breast pathology: a practical approach.
        (with ... 6 tables) Springer, Berlin; Heidelberg2007
        • Hughes LE
        • Bundred NJ
        Breast macrocysts.
        World J Surg. 1989; 13: 711-714
        • Cooper AP
        Illustrations of the diseases of the breast…In two parts.
        (Pt. 1. pp. 89. pl. IX) Longman, Rees, Orme, Brown & Green, London1829
        • Hughes LE
        • Mansel RE
        • Webster DJ
        Aberrations of normal development and involution (ANDI): a new perspective on pathogenesis and nomenclature of benign breast disorders.
        Lancet. 1987; 2: 1316-1319
        • Davis HH
        • Simons M
        • Davis JB
        Cystic disease of the breast: relationship to carcinoma.
        Cancer. 1964; 17: 957-978
        • Dupont WD
        • Page DL
        Risk factors for breast cancer in women with proliferative breast disease.
        N Engl J Med. 1985; 312: 146-151
        • Azzopardi JG
        • Ahmed A
        • Millis RR
        Problems in breast pathology.
        Major Prob Pathol. 1979; : 11
        • Bundred NJ
        • Miller WR
        • Walker RA
        An immunohistochemical study of the tissue distribution of the breast cyst fluid protein, zinc alpha 2 glycoprotein.
        Histopathology. 1987; 11: 603-610
        • Bradlow HL
        • Skidmore FD
        • Schwartz MK
        • et al.
        Cation levels in human breast cyst fluid.
        Clin Oncol. 1981; 7: 388-390
        • Bhate RD
        • Chakravorty A
        • Ebbs SR
        Management of breast cysts revisited.
        Int J Clin Pract. 2007; 61: 195-199
        • Know Error
        DNA Specimen Provenance Assay - DSPA Testing.
        ([cited 2017 October 9] Available at:)
        • Ong MS
        • Mandl KD
        National expenditure for false-positive mammograms and breast cancer overdiagnoses estimated at $4 billion a year.
        Health Affairs. 2015; 34: 576-583
        • Glover GH
        Computerized time-of flight ultrasonic tomography for breast examination.
        Ultrasound Med Biol. 1977; 3: 117-127
        • Greenleaf JF
        • Bahn RC
        Clinical imaging with transmissive ultrasonic computerized tomography.
        IEEE Trans Biomed Eng. 1981; 28: 177-185
        • Wiskin J
        • Borup D
        • Johnson S
        • et al.
        Three-dimensional nonlinear inverse scattering: quantitative transmission algorithms, refraction corrected reflection, scanner design and clinical results.
        Proc Meetings Acoust. 2013; 19075001
        • Andre M
        • Wiskin J
        • Borup D
        Clinical results with ultrasound computed tomography of the breast.
        in: Quantitative Ultrasound in Soft Tissues, Part IV: Ultrasound Computer Tomography. Chap. 15. Springer, New York2013: 395-432
        • Andre M
        • Wiskin J
        • Borup D
        • et al.
        Quantitative volumetric breast imaging with 3D inverse scatter computed tomography.
        Conf Proc IEEE Eng Med Biol Soc. 2012; 2012: 1110-1113
        • Wiskin J
        • Borup DT
        • Johnson SA
        • et al.
        Non-linear inverse scattering: high resolution quantitative breast tissue tomography.
        J Acoust Soc Am. 2012; 131: 3802-3813
        • Lenox MW
        • Wiskin J
        • Lewis MA
        • et al.
        Imaging performance of quantitative transmission ultrasound.
        Int J Biomed Imaging. 2015; 2015454028
        • Klock JC
        • Iuanow E
        • Malik B
        • et al.
        Anatomy-correlated breast imaging and visual grading analysis using quantitative transmission ultrasoundTM.
        Int J Biomed Imaging. 2016; 2016: 9
        • Malik B
        • Klock J
        • Wiskin J
        • et al.
        Objective breast tissue image classification using quantitative transmission ultrasound tomography.
        Sci Rep. 2016; 6: 38857
        • Iuanow E
        • Smith K
        • Obuchowski NA
        • et al.
        Accuracy of cyst versus solid diagnosis in the breast using quantitative transmission (QT) ultrasound.
        Acad Radiol. 2017; 24: 1148-1153
        • Sartorius OW
        The biochemistry of breast cyst fluids and duct secretions.
        Breast Cancer Res Treat. 1995; 35: 255-266
        • Tsung JS
        • Wang TY
        • Wang SM
        • et al.
        Cytological and biochemical studies of breast cyst fluid.
        Breast. 2005; 14: 37-41
        • Dixon JM
        • Miller WR
        • Scott WN
        • et al.
        The morphological basis of human breast cyst populations.
        Br J Surg. 1983; 70: 604-606
        • Sánchez LM
        • Díez-Itza I
        • Vizoso F
        • et al.
        Cholesterol and apolipoprotein D in gross cystic disease of the breast.
        Clin Chem. 1992; 38: 695-698
        • Dogliotti L
        • Orlandi F
        • Torta M
        • et al.
        Cations and dehydroepiandrosterone-sulfate in cyst fluid of pre- and menopausal patients with gross cystic disease of the breast. Evidence for the existence of subpopulations of cysts.
        Eur J Cancer Clin Oncol. 1986; 22: 1301-1307
      1. Cobas 8000 modular analyzer series [cited 2017 October 9] Available at: https://usdiagnostics.roche.com/en/core_laboratory/instrument/cobas-8000-analyzer-series.html.

      2. Total Protein Gen. 2 [cited 2017 October 9] Available at: https://usdiagnostics.roche.com/products/03183734190/PARAM66/overlay.html.

      3. In vitro test for the quantitative determination of cholesterol in human serum and plasma on Roche/Hitachi cobas c systems. [cited 2017 October 9] Available at: https://usdiagnostics.roche.com/products/03039773190/PARAM41/overlay.html.

        • Cooke BM
        • Stuart J
        Automated measurement of plasma viscosity by capillary viscometer.
        J Clin Pathol. 1988; 41: 1213-1216
      4. ThinPrep Pap Test [cited 2017 October 9] Available at: https://healthdxs.com/en/thinprep/.

        • Papanicolaou GN
        • Traut HF
        Diagnosis of uterine cancer by the vaginal smear.
        Commonwealth Fund, New York1949
      5. 510(k) Summary: Cobas 8000 ISE Module, Urine Sample Type [cited 2017 October 9] Available at: https://www.accessdata.fda.gov/cdrh_docs/pdf12/K123726.pdf.

      6. Image Analysis DotCount v1.2 [cited 2017 October 9] Available at: http://reuter.mit.edu/software/dotcount/.

        • Ishimaru A
        Wave propagation and scattering in random media.
        IEEE Press, Piscataway, NJ2005
        • Ophir J
        • Jaeger P
        A ternary solution for independent acoustic impedance and speed of sound matching to biological tissues.
        Ultrason Imaging. 1982; 4: 163-170
        • Kalinicheva T
        • Frisch N
        • Giorgadze T
        • et al.
        Etiologic factors related to unsatisfactory ThinPrep® cervical cytology: evaluation and potential solutions to improve.
        CytoJournal. 2015; 12: 21