Abstract
Background
Sonoelastography has been used to differentiate malignant from benign lesions in numerous types of tissues including breast, prostate, liver, blood vessels, thyroid, musculoskeletal structures, and salivary glands.
Purpose
To evaluate the efficacy and application of real-time qualitative sonoelastography in the differentiation of benign and malignant focal parotid gland lesions.
Material and Methods
A total of 75 patients (36 boys/men, 39 girls/women; age range, 10–83 years) with 81 lesions were evaluated prospectively by sonoelastography performed and interpreted by two expert radiologists. The results of these experts classification and scoring of lesions according to relative stiffness of the mass were compared with each other and with histopathological findings. The interpretation of sonoelastography scores of 1–4 were as follows: 1, soft; 2, mostly soft; 3, mostly stiff; and 4, stiff.
Results
The kappa statistic of 0.508 (P <, 0.001) indicated moderate agreement between the two radiologists. The sonoelastography scores correctly diagnosed 30 of 49 benign tumors (sensitivity, 61.2%) and 19 of 32 malignant tumors (specificity, 59.4%). The area under the receiver-operating characteristic curve was 0.603. The diagnostic value of sonoelastography for evaluating pleomorphic adenomas, Warthin tumors, adenoid cystic carcinoma, and high-grade tumors was low, whereas the diagnostic rates for low-grade tumors such as mucoepidermoid carcinoma, acinic cell carcinoma, and metastases of basal cell carcinoma were better with sonoelastography.
Conclusion
Although sonoelastography seems to be promising in the differentiating of low-grade malignancies, the primary role of radiology is currently limited to determination of localization, size, and morphology of parotid tumors.
The incidence of salivary neoplasia is extremely low, representing about 3% of all tumors of the head and neck, considered as an organ system (1). However, it has one of the greatest diversities of benign and malignant neoplasia. Parotid gland imaging cannot be relied upon to provide a definite histological diagnosis, as the imaging characteristics of parotid gland tumors are non-specific.
Palpation, a subjective method that is heavily dependent on the experience and skill level of the physician as well as the dimensions and territory of the lesion, is the oldest and has been the most common way of identifying parotid gland masses. High-resolution ultrasonography (US), conventional magnetic resonance imaging (MRI), and computed tomography (CT) are the primary imaging techniques for evaluation of masses in the major salivary glands. While these techniques can detect masses with excellent sensitivity, they are less precise for predicting histology because of appreciable overlap in morphological features between various salivary pathological conditions, including between benign and malignant salivary neoplasms (2–5).
One of the key elements of evaluating masses by palpation is the degree of firmness; malignant lesions tend to be firmer than benign ones. Sonoelastography, which is analogous to manual palpation (6), is based on the proposition that pathological processes such as cancer modify the physical characteristics of diseased tissues. Sonoelastography differs from anatomical imaging techniques in that tissue contrast is derived from relative differences in mechanical stiffness – malignant tissues have greater stiffness than their benign counterparts (3, 7). In this regard, as a relatively novel technique first described in 1987 by Krouskop et al. (8), sonoelastography has been used to differentiate malignant from benign lesions in numerous types of tissues including breast, prostate, liver, blood vessels, thyroid, and musculoskeletal structures. Recently, it has been used to evaluate and differentiate the masses of salivary glands (3, 9).
In this study, we aimed to prospectively evaluate the efficacy and application of real-time qualitative sonoelastography in differentiation of benign and malignant focal parotid gland masses.
Material and Methods
Subjects
This prospective study was approved by the Institutional Ethics Committee. All participating patients signed a written informed consent before the study procedures.
Between 2009 and 2011, 78 patients with 85 abnormal focal masses in the parotid glands were examined. Among these patients, three had cystic masses, and one had a lesion in the deep lobe of the parotid gland. Because sono-elastography is not useful in evaluating cystic masses and those in the deep lobe of the parotid gland, these masses were excluded from the study cohort; thus, the study cohort consisted of 75 patients aged between 10 and 83 years (36 boys/men and 39 girls/women; mean age, 44.75 × 18.82 and 49.44 × 13.00, respectively) with 81 consecutive abnormal focal masses in the parotid gland. Among the patients recruited in the study, two patients had bilateral masses, two patients had two masses in a single parotid gland, and one patient had four masses in a single parotid gland (one mass was excluded from the study because it was observed in the deep lobe of the parotid gland).
All patients had undergone sonoelastography prior to biopsy, resection, or US-guided fine-needle aspiration for cytology.
Technique and image interpretation
Sonoelastographic imaging was performed by an expert radiologist via a real-time, free-hand technique using a Siemens S2000 US scanner (Siemens Medical Solutions, Inc., Malvern, PA, USA) integrated with sonoelastography software and a 13-MHz probe. Image video clips were reviewed by another expert radiologist, who was blind to the pathological findings. The findings of the two radiologists were statistically compared. In the case of different scores, a consensus was reached based on evaluation of the lesions to determine the final score. Consensus sonoelas-tography scores (CSESs) were compared with the histo-pathological findings.
Patient and transducer positioning for sonoelastography were identical to those in conventional US. The monitor was set to a dual display showing elastograms alongside corresponding grey scale US in real time. Using the B-mode display for guidance, a region of interest (ROI) that included the mass and adjacent salivary gland parenchyma was selected, and intermittent free-hand transducer compression was applied along the beam axis, while avoiding out-of-plane motion of the transducer because the latter causes artifacts. Real-time US elastograms were color-coded graphic representation of the relative stiffness of structures within the selected ROI such that blue indicated soft tissue, green indicated intermediate stiffness, and red indicated stiff tissue. An appropriate degree of compression was determined by manual adjustments such that normal adjacent salivary parenchyma appeared predominantly green on the elastograms. Four to eight cine loops lasting at least 5 s in sonoelastography were acquired for subsequent analysis. Pilot quality factor (QF) contributes to the real-time prediction of the accuracy of the amount of compression based on a 0–100 Likert scale. The score determined by sonoelastography is considered valid if it is not below 40. We used scores with an average QF value of 60.
Lesions were classified into four different patterns according to the relative stiffness of the mass, and then, by viewing cine loops of the elastograms, the stiffness of lesions was scored using a four-point scale (CSES 1–4) (Table 1, Figs. 1–4). This simplified scoring system was based on preliminary studies by Bhatia et al. (3) and on parotid and submandibular gland sonoelastography.
A large, lobulated, well-defined, slightly heterogeneously hypoechoic mass (arrows) with necrotic areas. The lesion had no calcifications. The consensus elastography score (CES) was 1 for this lesion (open arrows), i.e. mostly elastic A small, superficial, homogeneously hypoechoic mass with a well-defined regular contour (arrows). The lesion had no calcifications or necrotic areas. The CES of this lesion was 2 (open arrows), i.e. mostly elastic, but with slightly stiff areas A hypoechoic, well-defined mass with a regular contour (arrows). The lesion has a slightly heterogeneous internal structure without calcification or areas of necrosis. The CES was 3 for this lesion, which shows a mosaic patern with more stiffness and less elasticity (open arrows). The histopathological diagnosis was pleomorphic adenoma A large, slightly lobulated, heterogeneously hypoechoic mass (arrows) with internal fluid (necrotic) areas (stars). The lesion has no calcifications. This lesion presented mostly stiff areas and had an CES of 4 (open arrows). The histopathological diagnosis was mucoepidermoid carcinoma
The consensus sonoelastography scores (CSES) for stiffness of focal salivary gland masses
Following sonoelastography, 73 masses underwent surgical biopsy or resection, six underwent US-guided fine-needle aspiration for cytology, and two underwent both. Histopathology was used as the gold standard (n = 75), if available; otherwise, the reference standard was fine-needle aspiration for cytology. Of the 81 masses, 49 (60.5%) were diagnosed as benign lesions, and 32 (39.5%) as malignant lesions. The histological diagnoses of the 49 benign lesions were pleomorphic adenoma (n = 28), Warthin tumor (n = 10), lymphadenopathy (n = 9), and cystic adenoma/ infected cyst rupture (n = 2). The histological diagnoses of the 32 malignant tumors were lymphoma (n = 9), muco-epidermoid carcinoma (n = 6), adenoid cystic carcinoma (n = 5), metastases (n = 3), a myoepithelial malignant tumor (n = 1), and two each of pleomorphic adenocarcinoma, salivary duct carcinoma, acinic cell carcinoma, and basal cell carcinoma.
Statistical analysis
Statistical analysis was performed using SPSS version 20.0 (Statistical Package for Social Sciences, SPSS Inc., Chicago, IL, USA). The kappa statistic was used to evaluate the level of agreement between the two radiologists in sonoelas-tography scores. The CSESs reached by mutual agreement of the two radiologists were grouped into two categories. We grouped sonoelastography scores of “soft” and “mostly soft” into one category as “soft”, and grouped “mostly stiff” and “stiff” into another category as “stiff”. After sub-grouping, the diagnostic accuracy of CSES in diagnosing benign and malignant tumors was evaluated by receiver-operating characteristic (ROC) analysis. The area under the ROC curve was calculated.
The cut-off value of the sonoelastographic score was determined with a ROC curve in which area under the curve was 0.603 for score 2 (1–2 benign/3–4 malign), 0.511 for score 1 (1 benign/2–3–4 malign), and 0.410 for score 3 (1–2–3 benign/4malign). The best value with highest area under the curve which is score 2 (i.e. between scores 2 and 3) was set in the study.
All tests used a statistical significance level of 0.05.
Results
The CSESs for 75 patients with 81 parotid masses by histo-logical type are provided in Table 2. The kappa statistic was 0.508 (P < 0.001), indicating moderate agreement between the two radiologists. The scores given by two radiologists are shown in Table 3.
The number of patients with respect to the consensus sonoelastography scores (CSES) of the 81 parotid masses in 75 patients
The number of patients with respect to the scores given by two radiologists
The sensitivity and specificity of sonoelastography scores for malignant and benign tumors are summarized in Table 4. The sonoelastography scores correctly diagnosed 30 of 49 benign tumors (sensitivity = 61.2%) and 19 of 32 malignant tumors (specificity = 59.4%). The positive and negative predictive values were 50% and 69.8%, respectively. The area under the ROC curve is 0.603 (P = 0.119) (Fig. 5).
Receiver-operating characteristic (ROC) curves for different sonoelas-tographic score cut-off values. The area under the curve (AUC) is 0.603 (95% CI, 0.4766–0.730, P = 0.119) for score 1–2/3–4, 0.511 (0.381–0.641, P = 0.870) for score 1/2–3–4, and 0.410 (0.280–0.541, P = 175) for score 1–2–3/4
The sonoelastography scores vs. pathological diagnosis of the 81 parotid masses
Kappa = 0.200, P = 0.069
The true diagnostic rates of sonoelastography for differentiating between malignant and benign lesions in subgroups were as follows: 1/5 for adenoid cystic cancer, 2/2 for acinic cell cancer, 2/2 for basal cell carcinoma, 1/1 for infected epi-dermoid cyst, 7/9 for lymphadenopathy, 4/9 for lymphoma, 3/3 for metastases, 6/6 for mucoepidermoid cancer, 0/1 for myoepithelial malignant cancer, 17/28 for pleomorphic adenomas, 4/10 for Warthin tumors, 1/1 for cystadenoma, 1/2 for salivary duct carcinoma, and 1/2 for pleomorphic adenocancer lesions.
Discussion
In this study, we evaluated the efficacy and application of real-time qualitative sonoelastography in differentiation of benign and malignant focal parotid gland masses. We found that although sonoelastography seems to be promising in the differentiating of low-grade malignancies, the primary role of radiology is currently limited to determination of localization, size, and morphology of parotid tumors.
In previous studies, sonoelastography could predict malignancy at several sites (10–18). To the best of our knowledge, only two studies have been performed to date to evaluate the efficacy of sonoelastography in differentiating masses of the salivary glands, those by Bhatia et al. (3) and Dimitriu et al. (9).
Our study group consisted of 81 parotid gland masses with 14 different histopathologies (49 benign, 32 malignant). Our group comprised a wide range of different histopathologies, including malignant lesions. According to our results, pleomorphic adenomas and Warthin tumors both showed a wide range of sonoelastography scores. The diagnostic value of sonoelastography for evaluating pleomorphic adenomas, Warthin tumors, adenoid cystic carcinoma, and high-grade tumors was low, whereas low-grade malignant tumors such as mucoepidermoid carcinoma, acinic cell carcinoma, and metastases of basal cell carcinoma had better diagnostic rates with sonoelastography. However, a larger number of study groups are needed to support this hypothesis.
The largest group of lesions (n = 28) enrolled were pleomorphic adenomas, which contain different amounts of epithelial and mesenchymal structures (19). They showed variable CSES values. The diagnostic accuracy of the sonoelastography method for 28 pleomorphic adenoma lesions was 60%. Pleomorphic adenomas contain variable proportions of myxoid and/or chondroid matrix (20). The myxoid and chondroid matrix of these tumors contain different amounts of fluid (21), which is probably the cause of this wide range of CSESs.
Our study group included 10 Warthin tumors. The morphological aspects of Warthin tumors resembled those of malignant tumors (22). These tumors consist of lymphatic and cellular components, usually have a peripheral capsule, and can have cystic areas containing mucous or fluid (23–25). Warthin tumors can be solid, solid and cystic, or completely cystic. This tumor group has variable cellular matrix and extracellular compartments. Warthin tumors had a variety of sonoelastography scores resembling pleomorphic adenomas, which may be due to the variable features of these tumors.
Adenoid cystic cancer is a low-grade tumor that originates from the peripheral parotid duct. One of five (20%) adenoid cystic cancer cases could be truly diagnosed by sonoelastography. The least diagnostically accurate results of sonoelastography were for the adenoid cystic cancer group. These cancers contain an abundance of extracellular matrix that consists of mucinous, myxoid, and hyaline components (26). These extracellular components are rich in free fluid. The existence of extracellular matrix and its components may directly affect sonoelastography scoring, which explains why sonoelastography showed the lowest diagnostic accuracy in adenoid cystic cancers.
Mucoepidermoid cancer is a low-grade cancer originating from the parotid duct epithelium. It is the most common malignant parotid tumor (27) The six mucoepidermoid cancer lesions were all diagnosed as malignant by sonoelas-tography. The two cases of acinic cell cancer, a low-grade cancer (28), were also diagnosed as malignant by sonoelastography.
The three cases of squamous cell carcinoma metastases were all diagnosed as malignant by sonoelastography. According to the outcomes of our study, sonoelastography seems to be more accurate in differentiating malignancy in low-grade tumors. Our study group had nine secondary lymphoma lesions, four of which were diagnosed accurately (44%), and most were high-grade malignancies. This seems to support the opinion that sonoelastography has low accuracy in high-grade malignancies.
One patient had an infected epidermoid cyst, which was diagnosed as benign by sonoelastography.
Seven of the nine lymphadenopathy cases were diagnosed as benign according to the CSES. One myoepithelial malignant tumor was misdiagnosed by sonoelastography. One of two pleomorphic adenocarcinomas was diagnosed correctly, for an accuracy of 50% (Fig. 6). One cystadenoma was diagnosed as benign by sonoelastography scoring. Finally, one of two (50%) salivary duct carcinoma cases could be accurately diagnosed according to CSES.
(a) Two lesions. The small one (thin arrow) is a well-defined, homogeneously hypoechoic lesion with a regular contour. The larger one (thick arrow) is a lobulated, ill-defined heterogeneously hypoechoic lesion. (b) The CSES scores were 1 for the small lesion, which was elastic (thin arrow), and 4 for the larger lesion, which was entirely stiff (thick arrow). The histopathological diagnosis was a lymph node for the small lesion and carcinoma ex pleomorphic adenoma for the larger lesion
Dimitriu et al. (9) studied a group of 70 salivary gland masses with 11 different histopathologies (53 benign, 17 malignant). They reported that Warthin tumors and pleo-morphic adenomas mostly showed heterogeneous elasticity, whereas more extensive areas of stiffness were detected in malignant tumors. They also concluded that differential diagnosis among the many types of salivary gland masses were not improved by sonoelastography.
Bhatia et al. (3) studied a group of 65 salivary gland masses with eight different histopathologies (59 benign, 6 malignant). Their findings indicated that the stiffness differences between salivary gland masses were detectable, and although overlap existed between them, pleomorphic adenomas were generally firmer than Warthin tumors. They also suggested that sonoelastography could not reliably differentiate pleomorphic adenomas from malignant lesions of the salivary glands.
Our study had some limitations. Elastographic studies of the parotids were reported to have certain technical limitations related to location of these glands, in particular their proximity to the mandibles, the mastoid processes of the temporal bones, and the auricles of the ears. These structures can interfere with attempts to compress the parotids. In fact, it has been suggested that sonoelastography probably could be used correctly only for assessment of the superficial lobe of the parotid (29). Nevertheless, lack of quantification of the compression applied and the resultant strain image, as well as presence of elastographic artifacts, were reported to be major hurdles that need to be addressed before qualitative elastography can be adopted into routine clinical practice (3).
In conclusion, early results of our study showed that sonoelastography was of no value in distinguishing pleomorphic adenomas from other parotid masses, particularly from malignant tumors. Differentiation of malignant lesions would be very helpful in clarifying optimal surgical approaches. However, the primary role of radiology is currently limited to determination of localization, size, and morphology of the lesion.