Parathyroid carcinoma and atypical parathyroid tumors remain a conundrum to diagnose preoperatively. Lacking efficient methods to identify them before surgery could result in insufficient treatment. In our prospective study, the SWE characteristics of the two entities were described and the differences between them and parathyroid adenomas were demonstrated by values of SWE and the elasticity map. To our knowledge, this is the first study trying to differentiate parathyroid tumors with risk of recurrence from parathyroid adenomas with favorable prognosis using the 2D-SWE method.

We classified parathyroid carcinoma and atypical parathyroid tumor into a group in this study due to the commonly equivocal diagnosis of the two entities except for a long-term follow-up [5]. Meanwhile, the 2022 WHO classification of parathyroid tumors changed the nomenclature of “atypical parathyroid adenoma” into “atypical parathyroid tumor”, which reflected a parathyroid neoplasm of uncertain malignant potential [6]. Along with worrisome clinical and biochemical features, atypical parathyroid tumors could harbor histology aberrances often seen in parathyroid carcinoma including band-like fibrosis, adherence to adjacent structures, and Ki-67 labeling index > 5%. Compared with the malignant lesions with a high risk of recurrence, atypical parathyroid tumors with parafibromin deficiency have a low risk of recurrence and subsequent CDC73 gene sequencing is necessary for this entity [7]. Therefore, it is not only difficult but meaningless to distinguish cancer and atypical tumor before surgery. Preoperative diagnosis of both parathyroid tumors together would be appropriate to guide the choice of surgical approach.

In recent decades, elastographic techniques have experienced rapid developments with device iterations. From strain elastography to acoustic radiation force impulse (ARFI) [8,9,10], from point shear wave elastography to 2D shear wave elastography [11,12,13], previous studies employed these emerging methods to differentiate parathyroid adenoma from parathyroid hyperplasia, reactive lymph nodes or thyroid nodules. Given these researches, some scholars suggested minimally invasive parathyroidectomy for those single parathyroid lesions with high SWV because they are more likely to be parathyroid adenomas rather than parathyroid hyperplasia. However, they have ignored an important clinical scenario that the parathyroid carcinoma and atypical parathyroid tumors could be identified occasionally and they are indeed stiffer than parathyroid adenomas according to our study. Tumor stiffness is proved to correspond with tumor progression and invasiveness in cancers of different origins. The thyroid and breast malignancies have shown greater stiffness than their benign counterparts in many studies [14, 15]. Under such a situation, we firstly reported the cut-off value of 2.35 m/s in SWV and 17.05 kPa in Young’s modulus suggestive of the parathyroid carcinoma and atypical parathyroid tumor using 2D-SWE. Additionally, 2D-SWE has overcome the main limitations of previous studies utilizing operator-dependent strain elastography or small lesion-unfriendly VTQ technique, which may enhance the practicability in the clinical setting [16].

Four studies using the same Supersonic Imagine Aixplorer System to evaluate the parathyroid lesions were identified in the PubMed database. In a study carried out by Golu et al., the authors found that the mean elasticity indexes of parathyroid lesions were 10.2 kPa, which is comparable to our measurement of 10.89 kPa for parathyroid adenoma [17]. Stangierski et al. reported that the mean elasticity of parathyroid adenoma (5.2 ± 7.2 kPa) was significantly lower than benign thyroid nodules(24.3 ± 33.8 kPa), and they suggested that the negative predictive value of low elasticity was high enough to exclude suspicion of parathyroid adenomas [18]. Accordingly, we found parathyroid adenomas were more elastic than their counterparts, and the mean Young’s modulus > 17.05 kPa was reliable to diagnose suspicious parathyroid malignancies with excellent specificity. Amzar et al. found that the cut-off values of mean SWE confirmed for parathyroid adenoma and parathyroid tissue were 5.96 kPa and 9.58 kPa, respectively [19]. The elasticity of parathyroid tissue (parathyroid lesions in primary and secondary hyperparathyroidism) was significantly lower than thyroid or muscle tissue in their research. Cotoi et al. reported a cut-off value below 7 kPa to diagnose the parathyroid adenoma, and color maps in strain elastography rather than strain ratios were useful in identifying parathyroid adenomas [20]. In our study, parathyroid carcinoma and atypical parathyroid tumor showed significantly larger stiffness than parathyroid adenoma given the mean and max SWE velocities as well as the corresponding Young’s modulus. The mean Young’s modulus and shear wave velocity demonstrated the best diagnostic efficacy based on the area under the ROC curve (0.813, 0.852, respectively). Nevertheless, for achieving a balanced diagnostic performance, the max Young’s modulus was advised to differentiate the two group lesions with a sensitivity of 0.81 and specificity of 0.75. Moreover, the correlation between the max SWV/Young’s modulus and PTH may be explained by the ultrasound echogenicity features investigated by Li et al. [21] The authors reported that the median serum PTH level of the hypoechoic parathyroid lesions was higher than those of the iso-hyperechogenic group. They ascribed the finding to the histopathologic components in parathyroid tumors. The iso-hyperechogenic areas mainly correspond to non-functioning lipocytes, loose edema, connective tissues, or normal parathyroid tissues by pathology, which are usually soft tissues with relatively low SWE measuring values. In contrast, the PC and APT are mostly full of actively proliferated chief cells and fibrous bands whose elasticities are commonly high [22]. The max SWV/Young’s modulus could embody the histopathologic structures that are strongly associated with a high PTH level.

The elastogram of a parathyroid lesion reflects the tumor stiffness more comprehensively by containing the minimum, mean and maximum value of SWE information all together in a color-coded image. It is expected that the negative pattern was more common in the parathyroid adenoma group, which may due to the homogeneous and soft nature of parathyroid adenomas. The components of this benign parathyroid neoplasm are relatively simple, with high tumor cellularity, low amounts of connective tissue, fibrosis and necrosis. The “Stiff rim” sign was emphasized by Zhou et al. in differentiating between benign and malignant breast lesions [23]. The mechanism behind this qualitative SWE feature may lie in the infiltration of cancer cells into the surrounding tissues. “Colored lesion” on elastogram often indicates the increased stiffness around and within the targeted lesions. BE1 multinational study had found that this sign could improve the diagnostic performance by downgrading or upgrading breast lesions classified into 4a or 3 grades [24]. Accordingly, the pathologic diagnosis of PC should include the unequivocal infiltration of adjacent structures, and vascular or neural tissues, while the atypical parathyroid tumors are invariably characterized by thickened connective tissues, adherence to adjacent structures, or band-like fibrosis [25]. These pathological features are closely associated with our findings of prevailed stiff rim and colored lesion patterns in the PC/APT group. The signal void pattern did not show diagnostic value in this study. This phenomenon often derives from liquid formation or extremely rigid areas inside the tumor that hinders shear wave propagation. The liquid formation could be cystic degeneration, necrosis, or bleeding in pathology and is often displayed as anechoic areas on ultrasonography [26]. According to our previous study, these changes are associated with the size of parathyroid tumors and do not show a difference between parathyroid carcinoma and parathyroid adenoma/hyperplasia [27]. The extremely rigid areas usually found in malignant tumors are confirmed as the presence of intralesional dense collagen deposits in previous breast cancer research [28]. Although similar histopathological features are also reported in PC/APT, the signal void pattern was only observed once in our cohort. We assume that, in most cases, the collagen deposits and fibrosis in PC/APT are not so dense as to cause an extremely rigid area that invalid the SWE analysis.

This study has some limitations. Firstly, the small scale of cases restricts further statistical analysis for exploring independent risk factors of SWE parameters. The improved diagnostic performance of adding SWE to conventional US and clinical features could not be estimated too. Secondly, quantitative values of SWE are believed not possible to extrapolate between two different devices. Thirdly, the interobserver variability in the evaluation of parathyroid lesions using 2D-SWE was not assessed, which requires further research.