Salivary Gland Ultrasound to Diagnose Sjogren's Syndrome
Salivary Gland Ultrasound to Diagnose Sjogren's Syndrome
The study cohort consisted of 107 subjects, of whom 50 had definitive pSS and 57 had idiopathic sicca syndrome. Table 1 summarizes the characteristics of the patients. The mean age of the pSS group was lower than the non-pSS group (47 (13) vs 53 (12) yrs, p = 0.006). No further differences between the two groups were observed with respect to gender, frequency and duration of dry mouth and dry eye symptoms and objective features of dry eyes.
Salivary Gland Size and Parenchymal Echostructure. Salivary gland size was measured for parotid and submandibular glands in pSS patients and subjects with idiopathic sicca syndrome. There was correlation between the gland surface areas on the right and on the left sides and no significant differences were detected in major salivary gland surfaces between the two groups. The interobserver agreement on hypoechogenicity was good, with a Cohen's kappa value of 0.80. Cohen's kappa values for homogenicity were: 0.80, 0.70 and 0.90 for grades 1, 2 and 3, respectively. Cohen's kappa values for glandular size were 0.82 for parotid glands and 0.93 for submandibular glands. Finally, Cohen's kappa value for the posterior glandular border was 0.73. The echogenicity of the parotid glands was increased in 22/50 pSS patients (44 %) vs 2/57 subjects without pSS (3.5 %), whereas the echogenicity of the submandibular glands was increased in 34/50 of the pSS patients (68 %) vs 4/57 subjects without pSS (7 %). Among the parameters analyzed, inhomogeneity was the most significant in discriminating pSS patients from subjects with idiopathic sicca syndrome. More specifically, abnormal SGUS findings were detected in about 66 % of pSS patients and in fewer than 10 % of controls (p <0.0001).
Table 2 summarizes the inhomogeneity scores for the parotid and submandibular glands in patients with and without pSS. Examples of the parenchymal inhomogeneity grades of both parotid and submandibular glands in patients with pSS and in controls are shown in Figs. 1 and 2. Concordance between the parotid and the submandibular ultrasonographic grades was high with a Cohen's kappa value of 0.764. The overall SGUS score was significantly higher in pSS patients compared to subjects with idiopathic sicca syndrome (2.1 (1.8) vs 0.1 (0.4), p <0.0001). The SGUS score correlated inversely with the USFR (r = −0.37, p <0.0001) and directly with the MSGB focus score (r = 0.39, p <0.0001). More specifically, the SGUS score was significantly higher in patients with a USFR <1.5 ml/15 minutes (2.6 ± 1.7 vs 0.9 ± 1.5, p <0.0001) and in patients with an MSGB FS ≥1 (2.0 ± 1.8 vs 0.1 ± 0.7, p <0.0001).
(Enlarge Image)
Figure 1.
Parenchymal inhomogeneity in the submandibular glands demonstrated by ultrasonography. a Normal submandibular gland (grade 0). b Submandibular gland with evident inhomogeneity (grade 2). c, d Submandibular glands with gross inhomogeneity (grade 3)
(Enlarge Image)
Figure 2.
Parenchymal inhomogeneity in the parotid glands demonstrated by ultrasonography. a Normal parotid gland (grade 0). b Parotid gland with mild unspecific inhomogeneity (grade 1). c Parotid gland with evident inhomogeneity (grade 2). d Parotid gland with gross inhomogeneity (grade 3)
Diagnostic Value of Salivary Gland Echostructure. Consistent with previous studies a cutoff of 2 for the SGUS score allowed us to obtain the best ratio between sensitivity and specificity. The cutoff of 2 was associated with 66 % sensitivity, 98 % specificity, 97 % positive predictive value (PPV) and 77 % negative predictive value (NPV). Table 3 summarizes the diagnostic accuracy of the SGUS in comparison to anti-Ro/SSA, Schirmer's test and the MSGB focus score. With a cutoff of 2 the diagnostic performance of SGUS was very similar to that observed for the presence of anti-Ro/SSA antibodies in terms of fairly good sensitivity and high specificity: The SGUS score was less sensitive, but more specific than Schirmer's test. In comparison to MSGB, the SGUS had similar specificity but lower sensitivity, with moderate agreement between the two procedures (kappa Cohen's value = 0.573). The PPV of the SGUS was higher than for the MSGB alone, which means that when the SGUS results were positive, 33/34 pSS patients (97 %) had been correctly classified. The probability of correctly classifying patients as having pSS was even higher than with the MSGB alone, as only 30/34 pSS patients (88 %) were correctly classified on the basis of the MSGB alone. On the other hand, the NPV of the SGUS score was lower; therefore the probability of correctly classifying the subjects as not having pSS when the SGUS results were negative was only 77 % (i.e., 56/73 subjects with idiopathic sicca syndrome were correctly classified). As a consequence 17/73 pSS patients would have been misclassified on the basis of the SGUS alone without the MSGB. Figure 3 represents the receiver operator characteristic (ROC) curves showing the diagnostic accuracy of the SGUS score (range 0–6) assigned to the parenchymal inhomogeneity (area under the curve (AUC) = 0.82, 95 % CI 0.74, 0.91) and of the MSGB (AUC = 0.96, 95 % CI 0.92, 1) for pSS.
(Enlarge Image)
Figure 3.
Receiver operator characteristic curves show the diagnostic accuracy of the salivary gland ultrasonography (SGUS) score assigned to the parenchymal inhomogeneity (area under the curve (AUC) = 0.82, 95 % CI 0.74, 0.91) and of the minor salivary gland biopsy (AUC = 0.96, 95 % CI 0.92, 1) for primary Sjögren's syndrom
Results
Study Cohort
The study cohort consisted of 107 subjects, of whom 50 had definitive pSS and 57 had idiopathic sicca syndrome. Table 1 summarizes the characteristics of the patients. The mean age of the pSS group was lower than the non-pSS group (47 (13) vs 53 (12) yrs, p = 0.006). No further differences between the two groups were observed with respect to gender, frequency and duration of dry mouth and dry eye symptoms and objective features of dry eyes.
Ultrasonography Results
Salivary Gland Size and Parenchymal Echostructure. Salivary gland size was measured for parotid and submandibular glands in pSS patients and subjects with idiopathic sicca syndrome. There was correlation between the gland surface areas on the right and on the left sides and no significant differences were detected in major salivary gland surfaces between the two groups. The interobserver agreement on hypoechogenicity was good, with a Cohen's kappa value of 0.80. Cohen's kappa values for homogenicity were: 0.80, 0.70 and 0.90 for grades 1, 2 and 3, respectively. Cohen's kappa values for glandular size were 0.82 for parotid glands and 0.93 for submandibular glands. Finally, Cohen's kappa value for the posterior glandular border was 0.73. The echogenicity of the parotid glands was increased in 22/50 pSS patients (44 %) vs 2/57 subjects without pSS (3.5 %), whereas the echogenicity of the submandibular glands was increased in 34/50 of the pSS patients (68 %) vs 4/57 subjects without pSS (7 %). Among the parameters analyzed, inhomogeneity was the most significant in discriminating pSS patients from subjects with idiopathic sicca syndrome. More specifically, abnormal SGUS findings were detected in about 66 % of pSS patients and in fewer than 10 % of controls (p <0.0001).
Table 2 summarizes the inhomogeneity scores for the parotid and submandibular glands in patients with and without pSS. Examples of the parenchymal inhomogeneity grades of both parotid and submandibular glands in patients with pSS and in controls are shown in Figs. 1 and 2. Concordance between the parotid and the submandibular ultrasonographic grades was high with a Cohen's kappa value of 0.764. The overall SGUS score was significantly higher in pSS patients compared to subjects with idiopathic sicca syndrome (2.1 (1.8) vs 0.1 (0.4), p <0.0001). The SGUS score correlated inversely with the USFR (r = −0.37, p <0.0001) and directly with the MSGB focus score (r = 0.39, p <0.0001). More specifically, the SGUS score was significantly higher in patients with a USFR <1.5 ml/15 minutes (2.6 ± 1.7 vs 0.9 ± 1.5, p <0.0001) and in patients with an MSGB FS ≥1 (2.0 ± 1.8 vs 0.1 ± 0.7, p <0.0001).
(Enlarge Image)
Figure 1.
Parenchymal inhomogeneity in the submandibular glands demonstrated by ultrasonography. a Normal submandibular gland (grade 0). b Submandibular gland with evident inhomogeneity (grade 2). c, d Submandibular glands with gross inhomogeneity (grade 3)
(Enlarge Image)
Figure 2.
Parenchymal inhomogeneity in the parotid glands demonstrated by ultrasonography. a Normal parotid gland (grade 0). b Parotid gland with mild unspecific inhomogeneity (grade 1). c Parotid gland with evident inhomogeneity (grade 2). d Parotid gland with gross inhomogeneity (grade 3)
Diagnostic Value of Salivary Gland Echostructure. Consistent with previous studies a cutoff of 2 for the SGUS score allowed us to obtain the best ratio between sensitivity and specificity. The cutoff of 2 was associated with 66 % sensitivity, 98 % specificity, 97 % positive predictive value (PPV) and 77 % negative predictive value (NPV). Table 3 summarizes the diagnostic accuracy of the SGUS in comparison to anti-Ro/SSA, Schirmer's test and the MSGB focus score. With a cutoff of 2 the diagnostic performance of SGUS was very similar to that observed for the presence of anti-Ro/SSA antibodies in terms of fairly good sensitivity and high specificity: The SGUS score was less sensitive, but more specific than Schirmer's test. In comparison to MSGB, the SGUS had similar specificity but lower sensitivity, with moderate agreement between the two procedures (kappa Cohen's value = 0.573). The PPV of the SGUS was higher than for the MSGB alone, which means that when the SGUS results were positive, 33/34 pSS patients (97 %) had been correctly classified. The probability of correctly classifying patients as having pSS was even higher than with the MSGB alone, as only 30/34 pSS patients (88 %) were correctly classified on the basis of the MSGB alone. On the other hand, the NPV of the SGUS score was lower; therefore the probability of correctly classifying the subjects as not having pSS when the SGUS results were negative was only 77 % (i.e., 56/73 subjects with idiopathic sicca syndrome were correctly classified). As a consequence 17/73 pSS patients would have been misclassified on the basis of the SGUS alone without the MSGB. Figure 3 represents the receiver operator characteristic (ROC) curves showing the diagnostic accuracy of the SGUS score (range 0–6) assigned to the parenchymal inhomogeneity (area under the curve (AUC) = 0.82, 95 % CI 0.74, 0.91) and of the MSGB (AUC = 0.96, 95 % CI 0.92, 1) for pSS.
(Enlarge Image)
Figure 3.
Receiver operator characteristic curves show the diagnostic accuracy of the salivary gland ultrasonography (SGUS) score assigned to the parenchymal inhomogeneity (area under the curve (AUC) = 0.82, 95 % CI 0.74, 0.91) and of the minor salivary gland biopsy (AUC = 0.96, 95 % CI 0.92, 1) for primary Sjögren's syndrom
Source...