Introduction

Sjögren’s syndrome is a systemic autoimmune condition that typically damages exocrine glands with resulting symptoms of dryness in the mouth, eyes, and other body regions. In about a third of affected individuals, Sjögren’s also affects extra-exocrine tissues causing arthritis, Raynaud’s, nerve damage, renal tubular acidosis, interstitial lung disease, vasculitis, as well as other possible manifestations [1]. Diagnosing this condition can be challenging since it shares many of its features with other autoimmune diseases on one hand, and can itself be a feature of other autoimmune diseases—so called secondary Sjögren’s syndrome. Furthermore, dryness of the eyes and mouth, also called sicca syndrome, can be caused by many conditions other than Sjögren’s syndrome. The research classification of Sjögren’s syndrome has passed through many iterations but typically depends on a combination of symptoms of dryness in the eyes and mouth, clinical exam features objectively confirming exocrine dryness, auto-antibody test abnormalities such as anti-SSA/SSB, and salivary gland biopsy features of focal lymphocytic sialadenitis [2]. However, the auto-antibody tests miss 30% of patients with Sjögren’s [3], while the biopsy of a minor salivary gland can be as sensitive as 82% [4], but is invasive.

Due to the challenges in diagnosing Sjögren’s syndrome, ultrasonography has been extensively investigated as a diagnostic tool since 1988 [5]. The most characteristic ultrasound feature of salivary glands affected by Sjögren’s syndrome is a diffuse cyst-like heterogeneity with involvement of both parotid and submandibular glands. Glands can be evaluated sonographically for homogeneity, echogenicity, hypoechoic areas, hyperechoic areas, and border clarity. Based on these features, numerous ultrasound scoring systems have been developed, and systematic reviews have estimated their testing characteristics. Recently, ultrasound scoring has been proposed as part of the Sjögren’s classification criteria [6•]. Doppler and elastography are recent additions which may have a role in salivary gland investigation. We will review the state of the ultrasound features and scoring systems for salivary glands as well as the potential mimics of Sjögren’s syndrome and whether ultrasound can help distinguish these conditions.

Sonographic Features of Salivary Glands in Sjögren’s Syndrome

The most characteristic feature of Sjögren’s syndrome in the salivary gland on ultrasound are hypoechoic or anechoic lesions producing tissue inhomogeneity [7] (Fig. 1). The cause of these changes has been proposed to be either foci of lymphocytic infiltrates or due to ductal dilatation. Ductal dilation is a typical histologic finding in chronic obstructive submandibular sialadenitis, but not in Sjögren’s syndrome [8•], where as lymphoplasmacytic infiltrate and acinar atrophy are typical.

Fig. 1
figure 1

Ultrasound images (in gray scale) of submandibular glands from different patients. Panel (a) shows a normal submandibular gland. Panel (b) shows a submandibular gland from patient with Sjögren’s syndrome. Small arrowheads point to small hypoechoic lesions with hazy margins which represent plasmalymphocytic infiltrates. Panel (c) shows a submandibular gland from a patient with IgG4-related disease. Note the larger “nodal” pattern producing a bulging on the surface of the gland (larger arrowhead). Panel (d) is a gland from a patient with HIV. The cystic lesions are significantly larger, with septation (arrow) and producing through transmission (asterisk)

Full size image

Similarly, Hashimoto’s thyroiditis results in focal deposits throughout the thyroid parenchyma or within germinal centers producing a “giraffe pattern” on ultrasound [9]. Since the salivary gland echogenicity and homogeneity is graded in reference to the thyroid tissue, it is also important to ensure that the thyroid tissue is in fact normal as Hashimoto’s thyroiditis is a co-morbid condition with Sjögren’s syndrome in a third of cases [10].

Investigators have found that the salivary gland ultrasound score (SGUS) correlates with the focus score [11] in minor salivary gland biopsies, as well as in the parotid [12]. In contrast, there are no reports of ductal dilatation on gland biopsy of Sjögren’s syndrome. Furthermore, some research studies describe improvement in parotid echostructure and a trend toward improvement in submandibular echostructure after rituximab therapy [13]. If cystic changes were due to ductal dilation, there would be less reason to expect the lesions to improve with B cell depletion than if the lesions were due to lymphocytic infiltration.

Scoring Systems

The hypo/anechoic glandular lesions tend to have indistinct borders, no through transmission, to be spread diffusely through all glandular areas, and to be small, typically less than 6 mm. Scoring systems of Salafii [14] and Jousse-Joulin [15] both utilize lesion size as part of their scoring system with measurable lesions of < 2 mm resulting in a grade of 2, lesions from 2 to 6 mm resulting in grade 3, and those > 6 mm resulting in the highest grade of 4. Presence of echogenic bands (thought to represent fibrotic septa) tissue calcifications, and posterior gland border visibility also increase the ultrasound score (0–16 total score). The scoring system of Milic [16] is the most simplified, grading the glands from 0 to 3 based on degree of homogeneity only (0–12 total score). While the scoring system on Hočevar [17] is the most complicated, grading on parenchymal echogenicity, homogeneity, hypoechoic areas (without specific measurements), hyperechoic reflections, and glandular border clarity (0–48 total score). These comprise the most commonly used scoring systems, although many more have been described. These four scoring systems were compared in the same cohort of patients with primary Sjögren’s, secondary Sjögren’s, and non-autoimmune sicca controls resulting in similar ranges of sensitivity and specificity with areas under the curve of 0.915 (Salaffi), 0.897 (Joussee-Joulin), 0.891 (Hočevar), and 0.885 (Milic) for primary Sjögren’s and 0.851 (Joussee-Joulin), 0.844 (Salaffi), 0.824 (Hočevar), and 0.808 (Milic) for secondary Sjögren’s [18•].

Doppler

Sjögren’s syndrome salivary glands show significantly more vascularity than healthy controls. Abnormal vascularity has been found to correlate with minor salivary gland histopathological grades, and Doppler grading improved the sensitivity, specificity, and accuracy of SGUS from 44%, 97%, and 65%, to 63%, 90%, and 74%, respectively [19]. However, increased Doppler has been described in many other pathologic processes that affect the salivary glands [20], and some have shown that Doppler does not discriminate between various pathological salivary gland lesions [21]. Doppler signal may also vary with disease duration, higher than normal signal in early active disease and lower than normal in late “burned out” disease [22].

Elastography

One of the newest ultrasonographic measures applied to salivary glands is elastography, which can measure the stiffness of a structure. In fact, glandular stiffness [23] in patients with Sjögren’s syndrome was determined to be significantly higher than in patients with non-autoimmune sicca. Furthermore, in cases where gray scale ultrasound was inconclusive in distinguishing patients with primary Sjögren’s syndrome from sicca controls, elastographic measurements resulted in a sensitivity of 67% and specificity of 86% for distinguishing the groups [24].

Gray-Scale Ultrasound Sensitivity and Specificity for Sjögren’s Syndrome

Through 2019, 47 studies have reported ultrasound test characteristics for salivary glands in Sjögren’s syndrome (Fig. 2). These studies have varied in many respects, including the gold standard for comparison, with most using one of the validated classification criteria and some using a biopsy result. In reference to classification criteria, parotid gland biopsy has been found to have a sensitivity of 75% and specificity of 88%, almost identical to that for labial salivary gland biopsy with sensitivity of 72% and specificity of 85%. The difference in gold standard may not matter much as optimal Hočevar ultrasound score agreed equally with parotid biopsy and classification criteria [12].

Fig. 2
figure 2

Sensitivity (a), specificity (b), and positive likelihood ratio (c) estimates for salivary gland ultrasound assessment of Sjögren’s syndrome. Circles represent point estimates reported by 47 studies on this topic published through 2019 ordered by study size, with circle areas representing relative study size

Full size image

However, these results do not mean that ultrasound can reasonably substitute for salivary gland biopsy. In 22 patients with sicca and US score < 15 as well as negative anti-SSA, five (23%) still had positive labial salivary gland biopsy, and five patients (11%) fulfilled the AECG criteria [12]. Thus, a negative US, even if the serology is also negative, will miss ACR-EULAR criteria positive Sjögren’s patient 11% of the time. However, a combination of a positive serology with a positive ultrasound predicts the fulfillment of the ACR-EULAR classification criteria 97% of the time. Van Nimwegan et al. [6•] have showed that the validity of the ACR-EULAR criteria remains high after incorporation of SGUS, but that substitution of SGUS score for either minor salivary gland biopsy, parotid gland biopsy, or anti-SSA antibody testing hinders test accuracy mainly through effects on sensitivity. Despite this caveat, Cornec et al. [4] found that inclusion of SGUS in the ACR/EULAR criteria improves its sensitivity from 64.4% to 84.4%, without changing its specificity (89.3% vs. 91.0%).

The most recent meta-analysis by Carottii et al. in 2019 assessed 37 studies and found a pooled specificity 91% (CI 88–93) and a pooled sensitivity 83% (CI 78–87) [25•]. Other recent meta-analyses found similar results without significant difference in sensitivity or specificity based on scoring system used (75% sensitivity for 0–4 and 0–48 point systems vs. 84% for 0–16 point system, while specificity was 93% for 0–4, 88% for 0–16, and 95% for 0–48) [26]. However, differences in study populations and control groups as well as publication bias [27] may falsely increase the ultrasound test performance characteristics. Figure 2 demonstrates that positive likelihood ratio for salivary gland ultrasound testing ranges from 4 to 12 in the five largest studies to date, and thus, a positive salivary gland ultrasound substantially increases the probability of a patient having Sjögren’s syndrome.

Reliability

Inter-rater reliability for salivary gland scoring has generally been good to excellent, with Hočevar reporting an overall kappa of 0.9, with kappa in the 0.88 to 0.9 range for echogenicity, inhomogeneity, and presence of hypoechoic areas and lower kappa of 0.5–0.52 for gland boarders and hyperechoic foci [28]. Since then, a number of other investigators have also reported an overall salivary gland ultrasound score kappa in the 0.8 to 0.95 range [29, 30]. A more recent study from 2018 [31] also showed an overall inter-rater kappa in the 0.7–0.84 range, confirming that glandular homogeneity and hypoechoic areas were much more reliable than assessments of glandular border, hyperechoic areas, or echogenicity. The number of hypoechoic/anechoic areas had inter-observer reliability of 0.53 in the submandibular gland and 0.74 in the parotid. Abnormal lymph nodes, hyperechoic bands, calcifications, and posterior border visibility showed low inter observer reliability (kappa = 0.38–0.01) [32].

Discriminating Sjögren’s from Other Diseases of the Salivary Glands

ACR-EULAR classification criteria specifically exclude patients with conditions that can be confused with Sjögren’s, namely AIDS, Hepatitis C, amyloidosis, sarcoidosis, IgG4-related disease, graft-versus-host disease, history of head and neck radiation treatment, etc. Thus, it is of particular interest whether ultrasound assessment can differentiate Sjögren’s from these conditions as well as from other autoimmune conditions that affect the salivary glands.

A study comparing salivary gland ultrasound findings in patients with systemic sclerosis in comparison to primary Sjögren’s and healthy controls found abnormal ultrasound scores in 75% of 48 patients with Sjogren’s, 28% of 25 patients with systemic sclerosis, and 9% of 35 healthy controls [33]. The abnormalities were not different in patients with systemic sclerosis than in Sjogren’s; thus, the authors conclude that ultrasound can detect Sjögren’s overlap with systemic sclerosis, rather than a salivary gland fibrotic disease specific to systemic sclerosis. A similar study compared Sjögren’s with other connective tissue diseases including SLE, systemic sclerosis, mixed connective tissue disease, and undifferentiated connective tissue disease and found an ultrasound score ≥ 2 in 78% of the Sjögren’s patients compared to 28% of the connective tissue disease patients. While the Sjögren’s patients had a similar degree of involvement of parotid and submandibular glands (62% and 64%, respectively), the non-Sjögren’s connective tissue disease cohort had a higher proportion of submandibular gland involvement than parotid involvement (28% and 14%, respectively) [34].

Sarcoidosis and amyloidosis are infiltrative diseases that have been reported to affect the salivary glands. However, there have been few descriptions of the ultrasound appearance of these diseases. A recent study [21] comparing cohorts of patients with Sjögren’s, AL amyloidosis, sarcoidosis, and healthy controls did not detect a “nodal” pattern of involvement between the groups as there is between Sjögren’s and IGG4. The overall median Hočevar US score was higher in Sjögren’s than in the other groups, and both the amyloid and sarcoid groups had higher median scores than the healthy control group. Notably, 27% of AL amyloidosis and 19% of sarcoidosis groups scored above the ultrasound score previously described as being specific for Sjögren’s syndrome. Despite prior studies suggesting a greater degree of parotid than submandibular gland involvement in sarcoidosis, this study did not confirm such a pattern.

Exclusion of underlying malignancy drives some salivary gland biopsies in patients with Sjögren’s, and ultrasound can increase suspicion of underlying MALT lymphoma. Sjögren’s patients with MALT lymphoma had average US scores almost twice as high as Sjögren’s patients without MALT lymphoma or at high risk factors for MALT lymphoma [35]. In a case series of MALT lymphomas of the head and neck, 7 of 15 cases affected the salivary glands while 8 affected the thyroid gland [36]. Of the 7 salivary gland cases, all but one affected the parotid glands. The typical sonographic pattern was that of either “linear echogenic strands pattern” also referred to as “multiple small hypoechoic nodules” or “tortoiseshell pattern”, or “segmental pattern”/ “multiple larger hypoechoic masses”. These patterns may resemble IgG4-related disease, or advanced Sjögren’s. The authors also describe diffuse large B cell lymphoma in 12 cases, where the glands are typically diffusely hypoechoic and have associated lymph node abnormalities.

Similar to the ultrasound results for MALT lymphoma, IgG4-related disease also produced higher salivary gland ultrasound scores on the Hočevar 0–48 point system than matched Sjögren’s patients (26 for IgG4-related disease compared to 21.5 for Sjögren’s group). The difference was accounted for by higher scores in the submandibular glands (18 vs 11), while in the parotid glands, the scores were essentially the same. They also found a correlation between serum IgG4 levels and SGUS in the IgG4-related disease patients (r = 0.331, p < 0.05) [37]. In a separate study of 30 patients with IgG4-related disease compared to 38 with Sjögren’s and 36 healthy controls, a reticular pattern was found in both IgG4 and Sjögren’s, but a “nodal” pattern was found in the submandibular glands in IgG4 much more commonly than Sjögren’s or controls (Fig. 1). Unfortunately, the authors did not present data on nodal pattern specificity in distinguishing the two conditions [38]. The “nodal” pattern was defined in another article as hypoechoic, homogenous areas with relatively high vascularization, and bulging from the surface of the submandibular glands [39]. This article also found submandibular gland “nodal” regions in 8 of 9 patients with IgG4-related disease, but in none of the parotid glands. Similarly, other authors described a “nodal” pattern in 10/15 cases [40], 31/42 cases [41], and 25/30 cases [42]. There are three retrospective, case-control trials, and four case series of ultrasound use for IgG4-related disease of the salivary glands which comprise a total of 160 patients. In 108 cases where submandibular glands involvement was assessed specifically, there were ultrasound abnormalities in 99 (92%) [20, 38, 40,41,42,43,44]. This is in distinction to the parotid glands where 20 out of 60 (33%) were affected. Submandibular glands are also typically longer and thicker in IgG4-related disease than in controls and tend to have rough, irregular contour [20].

Unlike IgG4-related disease, which tends to target submandibular glands more than parotid glands, parotid involvement in HIV has been reported to occur in 6–10% of cases and increases to 51% in AIDS [45,46,47]. Of 200 patients in Uganda with HIV presenting for hospital care, 195 had parotid abnormalities by ultrasound. Forty two percent of the patients had lymphoepithelial cysts, 20% had fatty aggregates defined as whole gland hypoechoic appearance with posterior attenuation, while another 20% had lymphocytic aggregates, and 16% had lymphadenopathy alone [48]. As previously noted, lymphocytic aggregates have a size usually less than 6 mm, ill-defined margins, and lack posterior acoustic enhancement, while lymphoepithelial cysts are the opposite in these three respects and tend to also have internal septations (Fig. 1). Unfortunately, the degree of submandibular involvement was not described, but others have noted that the submandibular glands are usually spared [49].

Conclusions

Ultrasound detects small hypo/anechoic lesions spread throughout the major salivary glands. These findings strongly correlate with histology findings and are sensitive and specific for discriminating Sjögren’s from sicca symptoms due to medication or age-related causes. Salivary gland ultrasound findings may increase diagnostic certainty when other items in the ACR-EULAR classification criteria are equivocal, but similar findings can be encountered in other conditions that can affect the salivary glands such as IgG4-related disease, lymphoma, sarcoidosis, amyloidosis, and HIV. Detection of a nodal pattern of involvement or whole gland hypoechogenicity with gland surface bulging suggests IgG4-related disease in the submandibular glands and lymphoma or HIV in the parotid glands, while large cystic lesions in the parotid glands would be most typical of HIV. Ultrasound also holds a tantalizing opportunity for identification of both early gland involvement via Doppler imaging and detection of ultrastructural glandular healing in response to therapy.