Women s Imaging Original Research

Transcription

1 Women s Imaging Original Research Park et al. Strain Elastography of Axillary Nodes in Breast Cancer Women s Imaging Original Research Young Mi Park 1,2 Bruno D. Fornage 1,3 Ana Paula Benveniste 1 Patricia S. Fox 4 Roland L. Bassett, Jr. 4 Wei Tse Yang 1 Park YM, Fornage BD, Benveniste AP, Fox PS, Bassett RL Jr, Yang WT Keywords: axilla, breast cancer, elastography, lymph node, metastasis, sonography, ultrasound DOI: /AJR Received December 5, 2013; accepted after revision April 4, Supported in part by the National Cancer Institute (grant P30 CA016672). 1 Department of Diagnostic Radiology, University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd, Unit 1459, Houston, TX Address correspondence to W. T. Yang (wyang@mdanderson.org). 2 Department of Radiology, Busan Paik Hospital, Inje University College of Medicine, Busan, South Korea. 3 Department of Surgical Oncology, University of Texas M. D. Anderson Cancer Center, Houston, TX. 4 Department of Biostatistics, University of Texas M. D. Anderson Cancer Center, Houston, TX. AJR 2014; 203: X/14/ American Roentgen Ray Society Strain Elastography of Abnormal Axillary Nodes in Breast Cancer Patients Does Not Improve Diagnostic Accuracy Compared With Conventional Ultrasound Alone OBJECTIVE. The purpose of this study was to determine the diagnostic value of strain elastography (SE) alone and in combination with gray-scale ultrasound in the diagnosis of benign versus metastatic disease for abnormal axillary lymph nodes in breast cancer patients. SUBJECTS AND METHODS. Patients with breast cancer and axillary lymph nodes suspicious for metastatic disease on conventional ultrasound who underwent SE of the suspicious node before ultrasound-guided fine-needle aspiration biopsy (FNAB) were included in this study. On conventional ultrasound, the long- and short-axis diameters, long-axis to short-axis ratio, cortical echogenicity, thickness, and evenness were documented. The nodal vascularity was assessed on power Doppler imaging. Elastograms were evaluated for the percentage of black (hard) areas in the lymph node, and the SE-ultrasound size ratio was calculated. Two readers assessed the images independently and then in consensus in cases of disagreement. ROC AUCs were calculated for conventional ultrasound, SE, and both methods combined. Interreader reliability was assessed using kappa statistics. RESULTS. A total of 101 patients with 104 nodes were examined; 35 nodes were benign, and 69 had metastases. SE alone showed a significantly lower AUC (62%) than did conventional ultrasound (92%) (p < 0.001). There was no difference between the AUC of conventional ultrasound and the AUC of the combination of conventional ultrasound and SE (93%) (p = 0.16). Interreader reliability was moderate for all variables (κ 0.60) except the SE-ultrasound size ratio (κ = 0.35). CONCLUSION. Added SE does not improve the diagnostic ability of conventional ultrasound when evaluating abnormal axillary lymph nodes. T he most reliable prognostic indicators for disease recurrence and patient survival at the time of the initial diagnosis of breast cancer are the presence and extent of metastasis to the axillary lymph nodes [1]. Traditionally, axillary node dissection at the time of breast surgery has been the method of choice for obtaining histologic information on axillary node status. However, in up to half of the patients, axillary node dissection is associated with morbidity, including lymphedema and upper extremity dysfunction or discomfort [2, 3]. The challenge is thus to devise methods that identify breast cancer patients who can safely be spared axillary surgery. As a result, interest in less invasive alternative staging procedures such as sentinel lymph node biopsy (SLNB) and ultrasound-guided axillary node sampling has increased. SLNB is performed in patients with a low risk of axillary node metastases instead of axillary node dissection. However, false-negative results of SLNB have been reported [4]. The aim of imaging the axilla is to identify patients with nodal metastases for whom axillary node dissection would be appropriate. Among the existing imaging modalities for preoperative axillary node staging, grayscale ultrasound with ultrasound-guided fineneedle aspiration biopsy (FNAB) has emerged as an effective technique, albeit with a substantial false-negative rate [5]. Therefore, any improvement of gray-scale ultrasound s diagnostic accuracy is desirable. Elasticity is a term used to describe the compressibility of a lesion. If a lesion is harder than the surrounding tissues, it will undergo less distortion under compression than the surrounding tissues. Strain elastography (SE) uses ultrasound waves to measure the relative displacements of tissues according to stiffness and to produce an elastographic map. This elastogram can be displayed in AJR:203, December

2 Park et al. gray-scale or color, usually side by side with the standard gray-scale sonogram. SE has been used to evaluate breast, thyroid, and prostate tumors and cervical and inguinal lymphadenopathy [6 10]. The other main commercial ultrasound elastography technique is shear wave elastography (SWE), which documents the speed of a shear wave during acoustic radiation force excitation and provides a quantitative measurement of tissue stiffness [11, 12]. However, SE has rarely been applied to the evaluation of axillary lymph nodes in patients with breast cancer [13, 14]. The purpose of this study was to evaluate the accuracy of SE added to conventional ultrasound in the diagnosis of indeterminate axillary lymph nodes in patients with breast cancer. Subjects and Methods Patients This single-institution study was approved by the institutional review board with a waiver of informed consent. Our institution is a tertiary referral center where patients present for workup of abnormal mammography results and suspicious clinical findings and for staging of known breast cancer. Patients who presented for ultrasound staging of known cancer and patients who had breast ultrasound findings suspicious for breast cancer and an abnormal axillary node or nodes identified on conventional ultrasound with ultrasound-guided FNAB were included in this study. The ultrasound protocol for patients with known breast cancer and for patients with findings suspicious for breast cancer in our institution is real-time whole-breast ultrasound that includes the axillary, infraclavicular, internal mammary, and supraclavicular lymph node basins. Abnormal axillary nodes identified on gray-scale ultrasound based on previously published criteria [15 17] undergo ultrasound-guided FNAB of the highest-order node. SE was performed prospectively but was not used to determine the need for biopsy of axillary nodes. From August 2007 to September 2009, 109 consecutive breast cancer patients with 115 axillary lymph nodes suspicious for metastatic disease on conventional gray-scale ultrasound were evaluated with SE. Of these 115 axillary lymph nodes, seven were excluded for the following reasons: Nodes did not undergo pathologic confirmation or were lost to follow-up (n = 6), and the suspected axillary mass was found to be a primary invasive ductal carcinoma, not a metastatic node (n = 1). In addition, four lymph nodes in three patients were excluded because histopathologic confirmation of breast cancer was not available. A total of 104 lymph nodes in 101 patients were therefore included in the final data analysis. Three patients had bilateral nodes examined on SE. Pathologic diagnoses of the nodes were obtained using ultrasound-guided FNAB and then SLNB or axillary node dissection. Ultrasound Examinations Gray-scale ultrasound, power Doppler ultrasound, and SE of the axilla were performed by one of four ultrasound technologists with a median of 12 years of experience (range, 8 20 years) in breast ultrasound (Antares scanner, Siemens Healthcare) on a unit equipped with a 13-5 MHz linear transducer. For all patients, the ultrasound examination started with conventional gray-scale ultrasound. The positioning of the patients for imaging was identical to that used for standard breast ultrasound that is, the patient was lying in the supine oblique position with the ipsilateral arm elevated. Each abnormal lymph node was documented in two orthogonal planes, and its three longest perpendicular diameters (longitudinal, transverse, and anteroposterior) were measured on gray-scale ultrasound. On power Doppler imaging, the pattern of nodal blood flow was evaluated. Doppler gain was adjusted to avoid background noise and to try to display some hilar vascularity. After identification of the most suspicious node on conventional ultrasound, SE was applied using a freehand compression technique by the same ultrasound technologist who performed the conventional ultrasound. To obtain SE images, a 3.0-cm wide and 2.5-cm deep rectangular box was used to delineate the ROI. Next, the technologist manually applied slight axial compression to the lesion with the ultrasound probe in such a way as to depict the subcutaneous fat layer in gray. Hard lesions (with less strain) would be displayed in black and soft lesions (with more strain), in white. Gray indicated an average strain in the ROI. Images were displayed in a split-screen mode with the gray-scale ultrasound image on the left and the elastogram on the right. The longest diameters of the node were measured using the built-in electronic calipers on representative static sonograms and elastograms, and the SEultrasound size ratio was calculated. Acquisition of the elastograms took approximately 2 minutes per study. Real-time elastographic studies were then saved as short video files for later review. Image Evaluation Two experienced breast radiologists with 14 and 9 years of experience, respectively, in breast ultrasound who had not performed the SE examination or biopsy analyzed the randomly ordered archived gray-scale and power Doppler images and assigned a probability of malignancy. The readers were blinded to the mammographic and ultrasound appearances of the primary breast cancers and to the pathologic diagnosis of the node. The following gray-scale ultrasound parameters were evaluated: long- and short-axis diameters, long-axis to shortaxis ratio, cortical echogenicity with respect to the surrounding fat tissue (isoechoic, hypoechoic, or mixed echogenicity), maximum cortical thickness, and cortical thickening pattern (even, uneven, or hilar loss). Nodal vascularity patterns on power Doppler imaging were rated as absent, hilar (when flow signals were limited to the hilum), peripheral (when there was increased peripheral blood flow), or penetrating (when a node had transcapsular feeding vessels). In SE, the lesion size (transverse dimension or area) is compared with the corresponding lesion size on the B-mode ultrasound image during dual image capture. Malignant lesions appear larger on elastograms than on B-mode ultrasound images; thus, the SE-ultrasound ratio of malignant lesions is usually greater than 1 (Figs. 1 and 2). Elastograms were evaluated for the proportion of black (hard) areas in the lymph node (absent, or very small black area, < 50%, 50%, peripheral, or nearly 100%) and the SE-ultrasound size ratio (< 1, = 1, or > 1) (Fig. 3). Two readers assessed the images independently with discordant cases solved by consensus. The independent assessment of each reader was used to calculate the interobserver variability, and the consensus opinion was used to calculate the accuracy of elastography features. The diagnostic accuracy of conventional ultrasound alone was compared with that of SE alone and with that of ultrasound combined with SE. Interobserver reliability was evaluated for cortical echogenicity, cortical thickening pattern, nodal vascularity patterns, SEultrasound size ratio, and proportion of black areas. Statistical Analysis The Fisher exact, chi-square, and Wilcoxon rank sum tests were used to analyze variables of interest by lymph node type (metastatic or not). Univariate logistic regression analyses were performed to estimate whether each of the examined variables could predict that a lymph node was metastatic. Multivariate logistic regression was then performed using significant variables from the univariate analyses. ROC curves were generated from the multivariate logistic regression analyses, and the AUCs were calculated for conventional ultrasound, SE, and both methods combined. Three patients had two lesions each; these lesions were treated independently for a lesion-level analysis. Kappa statistics were calculated to assess interobserver reliability; p values < 0.05 were considered statistically significant. All statistical analyses were performed using statistics software for Microsoft Windows (SAS, version 9.2, SAS Institute) AJR:203, December 2014

3 Strain Elastography of Axillary Nodes in Breast Cancer Fig. 1 Reactive axillary lymph node in 57-year-old woman with invasive ductal carcinoma of right breast. In this case, elastogram finding was false-positive. Dotted lines show measurements used to calculate ratio of size of node on strain elastography and ultrasound. A, On gray-scale ultrasound image, lymph node (arrow) appears indeterminate; lymph node shows evenly thickened, isoechoic cortex and has long-axis to short-axis ratio of 2. B, Strain elastogram of node shows black area is covering 100% of lymph node (arrow). Ultrasound-guided fine-needle aspiration biopsy of node and sentinel lymph node biopsy were negative for malignancy. Fig. 2 Metastatic axillary lymph node in 63-year-old woman with invasive ductal carcinoma of left breast. In this case, elastogram finding was true-positive. Dotted lines show measurements used to calculate ratio of size of node on strain elastography and ultrasound. A, Lymph node (arrow) appears suspicious on gray-scale ultrasound; lymph node shows eccentric, hypoechoic cortical thickening. B, Strain elastogram of node shows black area is covering 100% of lymph node (arrow). Ultrasound-guided fine-needle aspiration biopsy of node and sentinel lymph node biopsy revealed metastatic carcinoma. Results The 101 patients included 99 women and two men with a total of 104 nodes. The median age of all patients was 55 years (age range, years). Sixty-nine of 104 (66.3%) nodes were malignant and 35 (33.7%), benign. Ultrasound-guided FNAB was performed after SE on the same day in all nodes. After ultrasound-guided FNAB, surgery was performed on 87 of the 104 axillary nodes. SLNB revealed benign lymph nodes in 17 patients, axillary node dissection performed in eight patients revealed metastases in seven and a benign hyperplastic node in one, and axillary node dissection after neoadjuvant chemotherapy was performed in 57 patients with metastatic lymph nodes. Surgery was not performed after ultrasound-guided FNAB in five nodes: two nodes of benign hyperplasia, two nodes of metastasis that had undergone chemoradiation, and one recurrent metastatic lymph node in a patient who had undergone surgery for breast cancer 8 years earlier. Twelve cases, four cases of benign hyperplasia and eight metastases, were lost to follow-up after ultrasound-guided FNAB. The 101 primary malignancies included invasive ductal carcinoma (n = 80), invasive lobular carcinoma (n = 8), invasive mammary carcinoma (n = 3), ductal carcinoma in situ (n = 3), mucinous carcinoma (n = 1), papillary carcinoma (n = 1), tubular carcinoma (n = 1), metaplastic carcinoma (n = 1), lymphoma (n = 1), malignant phyllodes tumor (n = 1), and leiomyosarcoma (n = 1). Gray-Scale and Power Doppler Ultrasound The characteristics of the examined lymph nodes are listed in Table 1. The longand short-axis diameters of the metastatic nodes (mean ± SD, 2.3 ± 1.1 cm and 1.2 ± 0.5 cm, respectively) differed significantly AJR:203, December

4 Park et al. from those of the benign nodes (1.5 ± 0.5 cm and 0.7 ± 0.2 cm, respectively) (p < for both measurements). There was no significant difference observed in long-axis to short-axis diameter ratios between metastatic and benign nodes (p = 0.15). Metastatic nodes showed significantly greater cortical thickness (0.8 ± 0.4 cm) than benign lymph nodes did (0.4 ± 0.1 cm; p < 0.001). With respect to cortical echogenicity, 49 of 69 (71%) metastatic nodes were hypoechoic or heterogeneous versus only eight of 35 (23%) benign nodes (p < 0.001). Uneven cortical thickening or hilar loss was seen in 58 of 69 (84%) malignant lymph nodes and 14 of 35 (40%) benign lymph nodes (p < 0.001). On power Doppler ultrasound, peripheral or penetrating vessels were seen in 25 of 50 (50%) metastatic nodes and in seven of 29 (24%) benign lymph nodes (p = 0.030). Strain Elastography There was no statistically significant difference in the proportions of black (hard) areas between benign and metastatic nodes (p = 0.10). There was no significant difference in SE-ultrasound size ratios between the two groups (p = 0.33) (Table 1). Univariate logistic regression results modeling the probability of a metastatic lymph node are presented in Table 2. The following gray-scale ultrasound parameters showed significant associations with metastatic disease: long-axis diameter (odds ratio [OR], TABLE 1: Ultrasound and Strain Elastography (SE) Characteristics of 104 Axillary Lymph Nodes 3.30; 95% CI, ; p < 0.001); shortaxis diameter (OR, 56.99; 95% CI, ; p < 0.001); cortical thickness (OR, 2.09; 95% CI, ; p < 0.001); hypoechoic versus isoechoic relative to surrounding fat tissue (OR, 8.27; 95% CI, ; p < 0.001); and hilar loss compared with even cortical thickening (OR, 14.00; 95% CI, ; p = 0.001). On power Doppler imaging, the presence of peripheral or penetrating vessels compared with their absence or the presence of hilar vessels was associated with malignancy (OR, 3.14; 95% CI, ; p = 0.03). On SE, the only parameter associated with malignancy was the presence of a black area of 50% or more of the nodal area (OR, 2.54; Imaging Features Benign Lymph Nodes (n = 35) Metastatic Lymph Nodes (n = 69) p Conventional ultrasound features Long-axis diameter (cm), mean (SD) 1.5 (0.5) 2.3 (1.1) < Short-axis diameter (cm), mean (SD) 0.7 (0.2) 1.2 (0.5) < Ratio of long-axis diameter to short-axis diameter, mean (SD) 2.1 (0.7) 2.0 (0.7) 0.15 Cortical thickness (cm), mean (SD) 0.4 (0.1) 0.8 (0.4) < Cortical echogenicity, no. (%) of nodes < Isoechoic 27 (77) 20 (29) Hypoechoic 8 (23) 46 (67) Heterogeneous 0 (0) 3 (4) Cortical thickening, no. (%) of nodes < Even 21 (60) 11 (16) Uneven 11 (31) 36 (52) Hilar loss 3 (9) 22 (32) Vascularity at power Doppler ultrasound (n = 79), no. (%) of nodes Absent 2 (7) 7 (14) Hilar 20 (69) 18 (36) Peripheral 7 (24) 20 (40) Penetrating 0 (0) 5 (10) SE features Elastography gray-scale ultrasound size ratio, no. (%) of nodes 0.33 < 1 0 (0) 0 (0) = 1 33 (94) 60 (87) > 1 2 (6) 9 (13) Proportion of black area, no. (%) of nodes 0.10 Absent or very small black area 1 (3) 0 (0) Black area < 50% 18 (50) 22 (32) Black area 50% 15 (42) 44 (64) Peripheral 0 (0) 1 (1) Nearly 100% black area 1 (3) 2 (3) Note Boldface indicates that p value is statistically significant AJR:203, December 2014

5 Strain Elastography of Axillary Nodes in Breast Cancer TABLE 2: Odds Ratios (OR) and 95% CIs From Univariate Analyses for the Odds of a Lymph Node Being Metastatic Imaging Features OR 95% CI p Conventional ultrasound features Long-axis diameter (per 1-cm increase) < Short-axis diameter (per 1-cm increase) < Ratio of long-axis diameter to short-axis diameter (per unit increase) Cortical thickness (per 0.1-cm increase) < Cortical echogenicity Isoechoic Hypoechoic or heterogeneous < Pattern of cortical thickening Even Uneven Hilar loss Vascularity at power Doppler ultrasound Absent or hilar Peripheral or penetrating SE features SE-ultrasound size ratio = 1 > Percentage of black area Black area < 50% Black area 50% Note Boldface indicates that p value is statistically significant. Dash ( ) indicates reference category. TABLE 3: Odds Ratios (OR) and 95% CIs From Multivariate Analyses for the Odds of a Lymph Node Being Metastatic Imaging Features OR 95% CI p Conventional ultrasound features Short-axis diameter (per 1-cm increase) Long-axis diameter (per 1-cm increase) Cortical thickness (per 0.1-cm increase) Cortical echogenicity Isoechoic Hypoechoic or heterogeneous Cortical thickening Even Uneven or hilar loss Vascularity on power Doppler ultrasound Absent or hilar Peripheral or penetrating Note Boldface indicates that p value is statistically significant. Dash ( ) indicates reference category. 95% CI, ; p = 0.03) (Table 2 and Figs. 1 and 2). Table 3 presents the multivariate logistic regression results, which include all significant variables from the univariate results given in Table 2. One multivariate model was created for gray-scale and Doppler ultrasound for the six significant univariate gray-scale and Doppler ultrasound variables; an AUC for conventional ultrasound was thus obtained from this model. Because the proportion of black area was the only significant variable for SE, AJR:203, December

6 Park et al. Fig. 3 Reactive axillary lymph node in 41-year-old woman with invasive ductal carcinoma of left breast. In this case, elastogram finding was true-negative. Dotted lines show measurements used to calculate ratio of size of node on strain elastography and ultrasound. A, Lymph node (arrow) appears indeterminate on gray-scale ultrasound; lymph node shows eccentric cortical thickening of cortex. B, Strain elastogram of node shows mixed black-and-white areas (arrow), indicating soft regions with less strain. Ultrasound-guided fine-needle aspiration biopsy of node and sentinel lymph node biopsy were negative for malignancy. an AUC for SE was obtained based on black area alone. A multivariate model was then created for conventional ultrasound combined with SE that included all significant variables from the univariate analyses; an AUC for ultrasound combined with SE was thus obtained from this model. For conventional ultrasound, cortical thickness per 0.1 cm (OR, 1.78; 95% CI, ; p = 0.02) remained as a significant predictor of metastasis in the presence of the other ultrasound variables. Figure 4 presents the ROC curves displaying the AUCs for conventional ultrasound, SE based only on the proportion of black area, and the combination of both. SE alone had a significantly lower AUC (62%) than conventional ultrasound (92%) (p < 0.001). The AUC of SE in combination with conventional ultrasound (93%) was not significantly different from that of conventional ultrasound alone (p = 0.16). Interobserver reliability between the two readers was assessed for several variables. Readings of cortical thickening on gray-scale ultrasound and analysis of vascular signals on power Doppler ultrasound showed moderate interobserver agreement (κ coefficient = 0.60 and 0.68, respectively), whereas evaluations of cortical echogenicity on gray-scale ultrasound and percentages of black areas on SE showed stronger agreement (κ coefficients = 0.73 and 0.76, respectively). The SE-ultrasound size ratio showed the lowest agreement (κ = 0.35). Discussion Gray-scale ultrasound provides morphologic information about masses, and color (power) Doppler ultrasound provides information about vascularity. SE was developed Sensitivity Specificity ROC curve (AUC) Conventional ultrasound (0.9207) SE (0.6159) Combined (0.9290) to offer additional information about tissue elasticity by measuring the relative degree of displacement of tissues when submitted to external compression for example, by the ultrasound probe. The differential diagnosis of benign versus malignant lymph nodes on conventional ultrasound is based on the assessment of size, margins, long-axis to short-axis ratio, cortical echogenicity, cortical thickness, presence or loss of echogenic hilum, and vascularity on color (power) Doppler ultrasound [15 17]. In our study, gray-scale ultrasound findings showed significant associations with the probability of metastatic disease, which is consistent with Fig. 4 ROC curves for conventional ultrasound, strain elastography (SE), and combination of both. AUC for SE (62%) is significantly lower (p < 0.001) than that of conventional ultrasound and there is no difference between AUCs of conventional ultrasound alone (92%) and of combination of conventional ultrasound and SE (93%) (p = 0.16). Diagonal line indicates no discrimination line. previous reports [15, 16]. However, a metaanalysis of conventional ultrasound of the axillary nodes revealed moderate sensitivity for the diagnosis of metastases and revealed that negative ultrasound findings do not exclude the presence of axillary lymph node metastases [18]. The results of our study show that the accuracy of SE was lower than that of gray-scale ultrasound and that SE did not add any performance increase when combined with conventional ultrasound for the evaluation of abnormal axillary nodes in patients with suspicious findings on conventional ultrasound. Elastography scoring systems of axillary lymph 1376 AJR:203, December 2014

7 Strain Elastography of Axillary Nodes in Breast Cancer nodes have been reported [13, 14]. Choi et al. [13] classified axillary lymph nodes using a 4-point scoring system based on the percentage of blue (hard) area in the lymph node. They found that the relative proportion of hard area in the lymph node was 45% or greater in 80.6% of metastatic nodes, whereas 66.7% of benign nodes had a proportion of hard area of less than 45%. Taylor et al. [14] used a fourpattern scoring system based on the amount of black (hard) area within the lymph node. In their study, the sensitivity, specificity, positive predictive value, and negative predictive value were 76%, 78%, 70%, and 81%, respectively, for conventional ultrasound and were 90%, 86%, 83%, and 93%, respectively, for SE [14]. Our results are discordant with these published results. Our review of the literature showed that SE is not assessed in a standardized manner across studies, which is a limitation that leads to different conclusions regarding whether SE is useful. In our study, we used a threshold based on 50% of the total black area within a lymph node in differentiating benign from malignant lymph nodes, whereas Choi et al. used a 45% criterion and Taylor et al. used both the distribution of and percentage of (50%) black area within the node. Consistent methods to score SE should be explored. In addition, visual estimation of the total high elasticity area within a lymph node on elastography is subjective and may be difficult to reproduce accurately. An objective method of estimating the percentage of hard area is therefore needed. Elastography by acoustic radiation force impulse [19] or real-time shear velocity [20], also known as SWE, may be an alternate solution [21]. When abnormal stiffness is identified in a specific area of interest on SWE, values of maximum and average stiffness and SD can be measured [22]. SWE provides quantitative measures of tissue stiffness with reduced interobserver variability. Color-coded assessment of maximum elasticity using SWE are displayed as shear wave velocity (in meters per second) or elasticity (in kilopascals) in the ROI acquired. The positive predictive value for malignancy increases with increasing elasticity [23]. It is however important to note that after the initial reports claiming outstanding diagnostic accuracy values for SWE in the diagnosis of breast and head and neck cancers and axillary lymph nodes [23 26], recent better controlled and non industrysponsored studies have shown that elastography did not improve the diagnostic accuracy of conventional (gray-scale and power Doppler) ultrasound studies [27, 28]. SE has been reported to be helpful in the diagnosis of metastatic disease in cervical and inguinal lymphadenopathy [8 10]. Alam et al. [9] found the diagnostic accuracy values of gray-scale ultrasound, SE, and combined methods in the differential diagnosis of enlarged cervical lymph nodes to be 84%, 89%, and 93%, respectively. In another study, Lyshchik et al. [8] showed that SE was more accurate in the differentiation of benign from metastatic cervical lymph nodes than was gray-scale sonography (92% vs 79%, respectively). Superficial organs such as the breast, thyroid, salivary glands, and lymph nodes are the best candidates for SE because they lie within the depth of tissue deformation that can be produced by manual compression [29]. A recent study by Chang et al. [30] investigated factors influencing the quality of SE in the evaluation of suspicious breast masses and found breast thickness at the location of the lesion to be the most important influencing factor. Cervical and inguinal lymph nodes are located more superficially than axillary lymph nodes and are thought to be more suitable for SE. Malignant lesions in the breast have been reported to appear larger on elastograms than on the corresponding conventional sonograms [31 33]. This finding has been attributed to the desmoplastic tissue response associated with some breast cancers. Our results, however, showed no significant difference in apparent size on ultrasound and SE between benign and metastatic lymph nodes. In benign lymph nodes with elastic properties similar to those of normal surrounding tissue, lesion borders are sometimes difficult to determine and may be overmeasured on elastograms because they intermingle with the normal surrounding tissue [33]. One possible explanation is that, unlike breast malignancies, axillary lymph node metastases rarely induce a desmoplastic reaction. Our study showed that interobserver agreement was moderate to substantial for all variables (κ 0.60) except the SE-ultrasound size ratio (κ = 0.35). Yoon et al. [34] reported significant interobserver variability in the assessment of elastograms, with real-time SE (i.e., observing the real-time images) showing fair agreement (κ = 0.28) and static SE (i.e., interpreting the still elastograms) showing moderate agreement (κ = 0.46). Those investigators suggested that the significant variability in real-time SE may be related to inadequate SE data acquisition rather than inaccurate interpretation. The relatively good interobserver agreement in this study is probably due to the fact that the two readers did not perform SE but instead analyzed the static elastograms. Our study has some limitations. First, only patients with abnormal-appearing lymph nodes suspicious for metastatic disease on conventional ultrasound were included in the study. SE data were not collected of lymph nodes that appeared normal on conventional ultrasound. This bias in the selection of patients enabled precise radiologic-pathologic correlation of lymph nodes and allowed the determination of the possible additive value of SE with conventional gray-scale ultrasound in selected patients with abnormal findings on conventional gray-scale ultrasound but did not allow the evaluation of SE for normal axillary nodes in breast cancer patients. Second, a single SE variable versus multiple gray-scale variables was used to assess abnormal axillary nodes in our study. The number of variables may impact the AUC in ROC curve analyses. The different techniques of elastography coupled with the lack of standardization in the interpretation of elastography images is a limitation that does not permit meaningful comparison of data when comparing studies in the published literature. Third, patients included in this study were known to have breast malignancy. Although the readers were blinded to the mammographic and ultrasound appearances of the breast lesions and to the pathologic diagnosis of the node, they may have overestimated their assessment in nodes with borderline imaging features. Fourth, the readers did not perform SE. The diagnostic accuracy of SE might have been improved if the readers had performed the SE examination themselves. However, in most practices in the United States, ultrasound examinations are performed by technologists, and it should be assumed that, if proven of value, SE would also be performed by technologists, with the breast imagers reviewing the archived static elastograms. In conclusion, the results of this study show that the addition of SE offers no significant benefit over conventional ultrasound alone in the evaluation of abnormal axillary lymph nodes in patients with breast cancer. This study also highlights the difficulty in interpretation of elastography images and data because of the wide range of techniques available and the lack of standardization of data capture and data analysis. AJR:203, December

8 Park et al. References 1. Fisher B, Bauer M, Wickerham DL, Redmond CK, Fisher ER. Relation of number of positive axillary nodes to the prognosis of patients with primary breast cancer. Cancer 1983; 52: Duff M, Hill AD, McGreal G, Walsh S, McDermott EW, O Higgins NJ. Prospective evaluation of the morbidity of axillary clearance. Br J Surg 2001; 88: Kwan W, Jackson J, Weir LM, et al. Chronic arm morbidity after curative breast cancer treatment: prevalence and impact on quality of life. J Clin Oncol 2002; 20: Guenther JM. Axillary dissection after unsuccessful sentinel lymphadenectomy for breast cancer. Am Surg 1999; 65: Bedrosian I, Bedi D, Kuerer HM, et al. Impact of clinicopathological factors on sensitivity of axillary ultrasonography in the detection of axillary nodal metastases in patients with breast cancer. Ann Surg Oncol 2003; 10: Cochlin DL, Ganatra RH, Griffiths DF. Elastography in the detection of prostatic cancer. Clin Radiol 2002; 57: Ginat DT, Destounis SV, Barr RG, Castaneda B, Strang JG, Rubens DJ. US elastography of breast and prostate lesions. RadioGraphics 2009; 29: Lyshchik A, Higashi T, Asato R, et al. Cervical lymph node metastases: diagnosis at sonoelastography initial experience. Radiology 2007; 243: Alam F, Naito K, Horiguchi J, Fukuda H, Tachikake T, Ito K. Accuracy of sonographic elastography in the differential diagnosis of enlarged cervical lymph nodes: comparison with conventional grayscale sonography. AJR 2008; 191: Aoyagi S, Izumi K, Hata H, Kawasaki H, Shimizu H. Usefulness of real-time tissue elastography for detecting lymph-node metastases in squamous cell carcinoma. Clin Exp Dermatol 2009; 34:e744 e Barr RG. Sonographic breast elastography: a primer. J Ultrasound Med 2012; 31: Itoh A, Ueno E, Tohno E, et al. Breast disease: clinical application of US elastography for diagnosis. Radiology 2006; 239: Choi JJ, Kang BJ, Kim SH, et al. Role of sonographic elastography in the differential diagnosis of axillary lymph nodes in breast cancer. J Ultrasound Med 2011; 30: Taylor K, O Keeffe S, Britton PD, et al. Ultrasound elastography as an adjuvant to conventional ultrasound in the preoperative assessment of axillary lymph nodes in suspected breast cancer: a pilot study. Clin Radiol 2011; 66: Esen G. Ultrasound of superficial lymph nodes. Eur J Radiol 2006; 58: Ojeda-Fournier H, Nguyen JQ. Ultrasound evaluation of regional breast lymph nodes. Semin Roentgenol 2011; 46: Fornage BD. Ultrasound evaluation of the lymphatic spread of breast cancer. In: Kuerer HM, ed. Kuerer s breast surgical oncology. New York, NY: McGraw-Hill, 2010: 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 2006; 186: Nightingale K, Bentley R, Trahey G. Observations of tissue response to acoustic radiation force: opportunities for imaging. Ultrason Imaging 2002; 24: Hoyt K, Parker KJ, Rubens DJ. Real-time shear velocity imaging using sonoelastographic techniques. Ultrasound Med Biol 2007; 33: Goddi A, Bonardi M, Alessi S. Breast elastography: a literature review. J Ultrasound 2012; 15: Evans A, Whelehan P, Thomson K, et al. Quantitative shear wave ultrasound elastography: initial experience in solid breast masses. Breast Cancer Res 2010; 12:R Berg WA, Cosgrove DO, Doré CJ, et al. Shearwave elastography improves the specificity of breast US: the BE1 multinational study of 939 masses. Radiology 2012; 262: Athanasiou A, Tardivon A, Tanter M, et al. Breast lesions: quantitative elastography with supersonic shear imaging preliminary results. Radiology 2010; 256: Tourasse C, Dénier JF, Awada A, Gratadour AC, Nessah-Bousquet K, Gay J. Elastography in the assessment of sentinel lymph nodes prior to dissection. Eur J Radiol 2012; 81: Evans A, Rauchhaus P, Whelehan P, et al. Does shear wave ultrasound independently predict axillary lymph node metastasis in women with invasive breast cancer? Breast Cancer Res Treat 2014; 143: Bhatia KS, Yuen EH, Cho CC, Tong CS, Lee YY, Ahuja AT. A pilot study evaluating real-time shear wave ultrasound elastography of miscellaneous non-nodal neck masses in a routine head and neck ultrasound clinic. Ultrasound Med Biol 2012; 38: Bhatia KS, Tong CS, Cho CC, Yuen EH, Lee YY, Ahuja AT. Shear wave elastography of thyroid nodules in routine clinical practice: preliminary observations and utility for detecting malignancy. Eur Radiol 2012; 22: Varghese T. Quasi-static ultrasound elastography. Ultrasound Clin 2009; 4: Chang JM, Moon WK, Cho N, Kim SJ. Breast mass evaluation: factors influencing the quality of US elastography. Radiology 2011; 259: Garra B, Cespedes E, Ophir J, et al. Elastography of breast lesions: initial clinical results. Radiology 1997; 202: Hall TJ. Topics in US: beyond the basics elasticity imaging with US. RadioGraphics 2003; 23: Barr RG. Real-time ultrasound elasticity of the breast: initial clinical results. Ultrasound Q 2010; 26: Yoon JH, Kim MH, Kim EK, Moon HJ, Kwak JY, Kim MJ. Interobserver variability of ultrasound elastography: how it affects the diagnosis of breast lesions. AJR 2011; 196: AJR:203, December 2014