We sought to determine the performance of real-time sonoelastography in the differential diagnosis of parotid gland tumors.
We sought to determine the performance of real-time sonoelastography in the differential diagnosis of parotid gland tumors.
Between April, 2014, and June, 2015, 54 parotid gland masses were examined by ultrasound and strain sonoelastography in 46 patients. Real-time sonoelastography using the elasticity score (E-index), which gives an absolute value between 0 (softest) and 6 (hardest), was performed. Demographic characteristics, histopathologic examination, and difference in elasticity scores between benign and malignant masses were evaluated.
The mean age of the patients was 60.01 ± 2.97 years, and 56.52% of the patients were male (n = 26). Among the 54 parotid gland masses, 44 (81.5%) were benign and 10 (18.5%) were malignant tumors, 63% (n = 34) of the lesions being on the right side. The diagnoses as benign tumors consisted of Warthin tumor (n = 18, 33.3%), pleomorphic adenoma (n = 8, 14.8%) and other benign tumors (n = 18, 33.3%). The mean elasticity score and the size of all tumors were 2.87 ± 0.96 and 23.68 ± 12.38 mm, respectively. The mean elasticity score for benign tumors was 2.75 ± 0.95, and for malignant tumors it was 3.44 ± 0.85 (P = .034).
According to our results, real-time strain sonoelastography seems to have additional value over routine sonographic evaluation of parotid gland tumors in the differential diagnosis of benign and malignant parotid masses. However, with a small sample of malignant cases and appreciable overlap of the stiffness of benign and malignant masses, caution must be applied because the findings may not be representative of all patients with a parotid gland tumor.
fine-needle aspiration biopsy
region of interest
Parotid glands are the largest of the salivary glands. They are the most frequent site of salivary gland tumors, accounting for approximately 80% to 85% of these tumors.[1, 2] Approximately 80% of parotid lesions are benign, and approximately 20% are malignant.[3, 4] The most common of the benign tumors include pleomorphic adenoma (59%) and Warthin tumor (22%). Although pleomorphic adenoma affects predominantly females (60%), Warthin tumors are found more often in males than in females. The most common malignant tumors of the parotid glands are mucoepidermoid carcinoma and adenoid cystic carcinoma.
Salivary gland tumors vary according to their histopathology, so classification of these tumors is difficult. Although fine-needle aspiration biopsy (FNAB) and imaging are 2 ways to obtain proper diagnosis of a salivary mass before surgical excision, there are some limitations to acquiring accurate information from a tumor. Subtyping of salivary gland tumors and the small samples obtained with FNAB are instances of the limitations.[6, 7] Therefore, novel imaging methods such as elastography are required to obtain accurate information from a salivary gland mass without using invasive methods.
High-resolution grayscale ultrasonography is a sensitive method and is currently the first-choice imaging technique in the evaluation of parotid gland lesions. The normal parotid gland has a homogeneous appearance and increased echogenicity, related to the fat composition of the gland relative to adjacent muscle on ultrasonography. In addition, ultrasonography can facilitate fine-needle aspiration and core-needle biopsy. As a new sonographic method, sonoelastography allows examination of variations in tissue hardness. Strain elastography, which we used in our study, indicates the stiffness of a soft tissue by measurement of the tissue strain induced by manual compression and allows real time imaging.
The aim of this study was to determine the performance of strain-type real-time sonoelastography by using the elasticity score, which ranges from 0 to 6, in the differential diagnosis of parotid gland tumors. To our knowledge, no previous reports have evaluated the E-index, which gives an absolute value between 0 (softest) and 6 (hardest) via strain elastography in parotid gland tumors. The E-index is reflected as a color distribution within the selected region of interest (ROI). Although low E-index indicates that soft tissue and red color is dominant, higher E-index indicates stiff tissue and blue color is dominant.
We performed this study at the Ankara University School of Medicine in the Department of Radiology between April, 2014, and June, 2015. The work was designed as a prospective study. All patients gave informed written consent before interventional procedures (biopsy). All the procedures used were performed in accordance with the Declaration of Helsinki for human subjects, and the Institutional Review Board approved the study.
In total 46 consecutive patients who had been referred to the Department of Radiology for a parotid mass biopsy procedure were included in the study. Demographic characteristics of the patients were noted. Fifty-four parotid gland masses clinically suspected to be tumors were examined by B-mode ultrasonography and sonoelastography prior to biopsy.
Sonographic evaluations were performed prior to biopsy by a radiologist having 2 years of experience in elastography (N.K.A.). On grayscale ultrasound, the images of the lesions were obtained in 2 perpendicular planes. Lesion features including the dimensions and echogenicity were evaluated. Cystic degeneration and calcification in lesions on grayscale sonography and vascularization of the masses on power Doppler sonography were evaluated and noted. The masses were considered vascularized when more than 3 vessels could be identified and nonvascularized when only 1 or 2 vessels could be identified. Grayscale sonography, power Doppler sonography, and real-time sonoelastography examinations were performed with a Logiq S7 Expert machine (GE Healthcare, Milwaukee, WI) equipped with a 9L-D linear-array probe. Mild repetitive manual compression and decompression were applied in the evaluation of elastography. The elastographic images were obtained with appropriate compression. The quality of compression was indicated by a bar scale ranging from 1 to 7 on the screen and images were obtained only when the optimal compression bar was in the range of 5 to 7. To obtain correct elastographic maps, the evaluation was repeated until images with adequate quality were achieved. The elastogram was displayed as a real-time color map of the relative elasticity, superimposed on the B-mode image. Elastography and grayscale sonography images were simultaneously presented as a 2-panel image. The ROI was placed on the mass via free-hand drawing tool according to the lesion contour. E-index was obtained by elastographic quantitative analysis. This is a strain-type elastography method, which provides relative information about stiffness compared with shear wave elastography measuring absolute stiffness of the target area (in kPa). The sonography machine automatically gives an absolute value between 0 (softest) and 6 (hardest). The E-index is reflected as a color distribution within the selected ROI. Although low E-index indicates soft tissue and red color is dominant, higher E-index indicates stiff tissue and blue color is dominant. Elasticity scores from real-time strain elastography were noted. The elasticity score is a value expressed from 0 to 1.0 for softer than average, and from 1.0 to 6.0 for stiffer than average, assuming that the average strain in the displayed ROI is 1.0 on the GE system. The difference in E-index between benign and malignant masses was evaluated.
Ultrasound-guided FNAB was performed in all cases following sonographic examinations with elastography. All procedures were performed in the sonography room under aseptic conditions, with local anesthesia. Three to five samples were obtained with a 21-gauge FNAB needle for pathologic examinations. Samples were obtained from all patients and were collected for histopathologic examination in a tube with formalin. Cytological examination was performed at the time of biopsy by a cytologist. Then, collected material (cell block) was used for histological evaluation of microtissue fragments. Cytological and histological results were used as the reference standard.
SPSS 15.0 software for Windows (IBM Corporation, Armonk, NY) was used for statistical evaluations. Descriptive statistics were used to summarize the characteristics of the study group, including mean ± SD and median (minimum–maximum). To compare the groups Mann-Whitney U test was used for nonnormally distributed data to test the difference between benign and malignant groups. Categorical variables were tested by the Fisher exact test. Two-tailed P < .05 was considered statistically significant.
Fine-needle aspiration biopsy was performed with 26 male (56.52%) and 20 female (43.47%) patients. There were a total of 54 lesions. The mean age of the study group was 60.01 ± 12.97 years (range 28–86 years). The mean size of the lesions was 23.68 ± 12.38 mm (range 6–55 mm).
Results of the FNABs are presented below in Table 2. Among the 54 parotid gland masses, 44 (81.5%) were benign and 10 (18.5%) were malignant tumors (Figures 1, 2). Additionally, 63% (n = 34) of the lesions were on the right side, whereas 37% (n = 20) were on the left side.
The diagnosis as benign tumor consisted of Warthin tumor (n = 18, 33.3%, in 13 patients), pleomorphic adenoma (n = 8, 14.8%, in 8 patients), and other benign tumors (n = 18, 33.3%, in 16 patients). For malignant tumors, there were only 10 masses (18.5%, in 9 patients; Table 1). Patients, diagnosed as having Warthin tumor (n=3), malignant melanoma (n=1), and granulomatous inflammation (n=1), had 2 lesions on the same side of the parotid gland. There were 2 lesions on both sides of the gland in patients with reactive lymphoid hyperplasia (n = 1; Figure 3) and Warthin tumor (n = 2).
|The Results of FNABa||Frequency (n, %)||Type of Neoplasm||Calsification %||Cystic Degeneration %||Vascularity %||P|
|Warthin tumor||18 (33.3)||Benign||11.4||27.3||65.9||>.05|
|Pleomorphic adenoma||8 (14.8)|
|Reactive lymphoid hyperplasia||6 (11.1)|
|Granulomatous inflammation||2 (3.7)|
|Basal cell adenoma||2 (3.7)|
|Lymphoepithelial cyst||1 (1.9)|
|Sialadenitis and abscess||1 (1.9)|
|Primary benign oncocytic neoplasm||1 (1.9)|
|Rosai-Dorfman disease||1 (1.9)|
|Malignant melanoma||3 (5.6)||Malignant||10.0||40||70||>.05|
|Salivary duct carcinoma||2 (3.7)|
|Squamous cell carcinoma||1 (1.9)|
|Adenoid cystic carcinoma||1 (1.9)|
|Low-grade adenocarcinoma||1 (1.9)|
|Mucoepidermoid carcinoma||1 (1.9)|
|Merkel cell carcinoma||1 (1.9)|
The mean elasticity score of all tumors was 2.87 ± 0.96. Although the mean elasticity score for benign tumors was 2.75 ± 0.95, it was 3.44 ± 0.85 for malignant tumors (P = .034). When elastography values of benign and malignant lesions were analyzed, contrary to expectations, higher elastography scores in benign lesions could be observed (Figure 4); malignant lesions had lower scores (Figure 5, Table 2).
|Number of Patient|
|Distribution of Elastography Scores|
|Diagnosis||Frequency (n, %)||0-1||1-2||2-3||3-4||4-5||5-6|
|Benign||Warthin tumor||18 (33.3)||2||10||5||1|
|Pleomorphic adenoma||8 (14.8)||2||3||2||1|
|Reactive lymphoid hyperplasia||6 (11.1)||2||1||1||1||1|
|Granulomatous inflammation||2 (3.7)||1||1|
|Basal cell adenoma||2 (3.7)||1||1|
|Lymphoepithelial cyst||1 (1.9)||1|
|Sialadenitis and abscess||1 (1.9)||1|
|Primary benign oncocytic neoplasm||1 (1.9)||1|
|Rosai-Dorfman disease||1 (1.9)||1|
|Malignant||Malignant melanoma||3 (5.6)||1||1||1|
|Salivary duct carcinoma||2 (3.7)||1||1|
|Squamous cell carcinoma||1 (1.9)||1|
|Adenoid cystic carcinoma||1 (1.9)||1|
|Low-grade adenocarcinoma||1 (1.9)||1|
|Mucoepidermoid carcinoma||1 (1.9)||1|
|Merkel cell carcinoma||1 (1.9)||1|
Cystic degeneration and calcification in lesions on grayscale sonography and vascularization of the masses on power Doppler sonography were also evaluated. Cystic degeneration was demonstrated in 27.3% of benign lesions and 40% of malignant lesions (P = .459). Calcification was found in 11.4% of benign lesions and 10% of malignant lesions (P ≥ .99). Although 65.9% of benign lesions were vascular, this ratio was 70% for malignant lesions (P ≥ .99). According to the findings for lesions on grayscale and power Doppler sonographic imaging, there was no statistically significant difference between benign and malignant lesions (P > .05; Table 1).No patients had complications resulting from the interventions.
Because parotid glands are superficial glands, it is easy to determine masses sonographically. However, it is difficult to perform characterization of the lesions. This study, has investigated the effect of a novel method, real-time sonoelastography, in lesion characterization. Our results indicate that sonoelastography could provide useful information additional to that of grayscale sonography in terms of the differentiation of benign and malignant lesions. Nevertheless, diversification of salivary gland tumors could not be achieved by this method.
Most parotid tumors are benign (approximately 80%), and the most common of the benign tumors is pleomorphic adenoma.[1-4] Our study included 10 (18.5%) malignant tumors, in accordance with the frequency in the literature. However, the most common diagnosis was Warthin tumor (18 masses [33.3%] in 13 patients). Pleomorphic adenoma was 14.8% (8 masses in 8 patients).
Sonoelastography is a newly introduced sonographic technique for assessing tissue elasticity, being qualitative and semiquantitative or quantitative.[10, 13-15] The technique is based on the fact that the stiffness of malignant neoplasms is greater than that of benign tumors because of the desmoplastic reactions and areas of fibrosis that they contain. The degree of tissue hardness or softness can be observed in real time in different color codes, and qualitative scoring can be visually performed with a qualitative method after grayscale imaging. In the semiquantitative method, it is possible to derive strain indices on the elasticity maps obtained by comparing strain levels of different normal-appearing areas with the strain level of the lesion.[10, 16] Shear wave ultrasound elastography is an operator-independent method for measuring tissue elasticity, in which a quantitative estimate indicates the stiffness of the soft tissue.[12, 17] We performed this study with strain-type real-time sonoelastography using the elasticity score, which ranged from 0 to 6, for the evaluation of parotid gland lesions. The ratio between the normal-appearing areas and the lesion was not used, because it could cause a second changeable parameter that might result in error.
In an article by Dumitriu et al, the authors evaluated 74 salivary gland tumors using qualitative real-time elastography and obtained higher elastography scores in malignant tumors compared with benign neoplasms. The same statistically significant results were also documented in our study. However, Dumitriu et al did not report any differences between malignant lesions and pleomorphic adenomas or between Warthin tumors and pleomorphic adenomas. There was appreciable overlap between the stiffness of benign lesions and that of malignant masses. Our study demonstrates the same overlap between benign tumors and malignant neoplasms. The elastography scoring results varied in a wide range from 1–2 to 5–6 in our scoring system (Table 2). A higher elastography score could be detected in reactive lymphoid hyperplasia, whereas a lower score could also be detected in squamous cell carcinoma. Therefore, reaching a decision according to individual elastography scores could be misleading. Further investigation, such as FNAB, generally should be used to obtain an accurate diagnosis before treatment planning and surgery, although FNAB is an invasive method.[6, 7]
In a recent study using real-time qualitative sonoelastography with histopathological correlation in 81 parotid gland masses, the authors found that the diagnostic value of sonoelastography for evaluating pleomorphic adenomas (17/28), Warthin tumors (4/10), adenoid cystic carcinoma (1/5), and high-grade tumors was low, whereas the diagnostic rates for low-grade tumors such as mucoepidermoid carcinoma (6/6), acinic cell carcinoma (2/2), and metastases of basal cell carcinoma were better with sonoelastography. According to their results, the elastography scores correctly diagnosed 30 of 49 (sensitivity = 61.2%) benign tumors and 19 of 32 (specificity = 59.4%) malignant tumors. In our study, the true diagnostic scores of sonoelastography for the differentiation between malignant and benign lesions were as follows; 5 of 8 for pleomorphic adenomas, 12 of 18 for Warthin tumors, 3 of 6 for reactive lymphoid hyperplasia, 1 of 1 for adenoid cystic carcinoma, 1 of 1 for mucoepidermoid carcinoma, and 0 of 1 for low-grade adenocarcinoma. We were not able to identify the accuracy of sonoelastography scores in low-grade malignant tumors because of the small sample of malignant subgroups in our study. Depending on the borderline elasticity (cutoff point = 3), the sensitivity of E-index in predicting benign tumors was 70% (29 of 44 benign tumors), and elasticity scores correctly diagnosed 7 of 10 malignant tumors (specificity = 65.91%) in our study group.
Twelve lesions had an elasticity score of 2.80 to 3.20 in our group. Two of twelve lesions were noted as malignant (salivary duct carcinoma [n = 1], infiltration of malignant melanoma [n = 1]), and 10 were recorded as benign (Warthin tumor [n = 7], pleomorphic adenoma [n = 1], lipoma [n =1], granulomatous inflammation [n = 1]). Radiological evaluation via grayscale sonography and elastography were performed prior to the biopsy, and management of the patient was determined according to the pathology result. As a result, the final management was not according only to elastography scores. All patients who had a score of 2.80 to 3.20, except those with a diagnosis of lipoma or granulomatous inflammation, were offered surgery by an otolaryngologist.
In an article by Wierzbicka et al, the authors suggested that a cutoff point at the level of ES4 (high scores) provided the best differentiation between benign and malignant lesions (sensitivity = 40%, specificity = 97%). They added that, although sonoelastography overlaying elasticity on the grayscale images supported additional information, the large degree of uncertainty with the method should also be remembered. In our study, if the cutoff point of 4 was accepted, sensitivity of 30% and specificity of 88.64% were calculated, so a higher cutoff value with greater specificity was found. However, it should be remembered that the sensitivity is decreased with the higher cutoff point.
In addition to this article, Bhatia et al evaluated salivary gland masses using shear-wave sonoelastography. They found overlap in elastic moduli between the stiffness of benign lesions and that of malignant masses. The authors concluded that there was no clinically useful cutoff value and that elastography was unsuitable for ruling out malignant lesions.
In a recent review, Ariji et al reported that sonographic elastography seemed to be suboptimal for salivary gland malignancies because there were many pathological types and much overlap between pleomorphic adenoma and malignant tumors. In light of the literature, it seems that sonographic elastography is suboptimal for detection of malignancy in salivary gland lesions.
The limitation of our study was the small sample of malignant lesions. The total number of malignant cases should be increased. However, because salivary gland malignancies are relatively rare malignancies in routine clinical practice, it seems difficult to increase the number of malignant cases.
In conclusion, real-time strain elastography can be performed easily to detect parotid masses accompanied by conventional sonography. It seems to have additional value at this site compared with conventional sonography according to the statistical results. However, with a small sample of malignant cases and an appreciable overlap between benign and malignant masses, caution must be applied; the findings may not be representative of all patients with a parotid gland tumor.