Intra-observer agreement rates between the repeat diagnoses were very high. Kappa values for selecting the same diagnosis from normal, IgG4-DS or SS were 0.83 (range, 0.76–0.91) on sonography, 0.69 (range, 0.5–0.90) on FDG-PET/CT, 0.70 (range, 0.49–0.80) on CT and 0.80 (range, 0.64–0.87) on MRI. Moreover, average kappa values for each finding were 0.62 for sonography, 0.61 for FDG-PET/CT, 0.51 for CT and 0.54 for MRI; no findings showed an inverted order of scores the second time. Therefore, we have shown the results of the first diagnosis in the relevant figures.
Analysis of sonographic findings
Figure 3 shows the results of sonographic analysis. In all findings, significant differences (p ≤0.0093) were observed between any two of the three diagnoses. The parotid glands of patients with SS mainly exhibited multiple hypoechoic areas (median score 4) and hyperechoic lines and/or spots (median score 4) (Fig. 3a and b), while the reticular pattern in the parotid glands showed overlap between IgG4-DS (median score 1) and SS (median score 2) (Fig. 3c). In the submandibular glands, overlaps were seen between IgG4-DS and SS regarding the findings of multiple hypoechoic areas (median score 4 for both IgG4-DS and SS) and hyperechoic lines and/or spots (median score 3 for IgG4-DS and 4 for SS), although significant differences (p = 0.0093, p <0.0001, respectively) were seen between the two (Fig. 3d and e). Obscuration of the submandibular gland configuration was mainly observed in SS (median score 3) (Fig. 3f). In contrast, IgG4-DS mainly exhibited reticular and nodal patterns (median score 5), and separation between IgG4-DS and the other conditions was especially good concerning the nodal pattern (median score 1 for normal glands and 2 for SS) (Fig. 3g and h). Intra-observer agreement rates between the repeat diagnoses were very high (kappa values, 0.67–0.87), which showed the nodal pattern could be easily detected.
Each case was diagnosed as normal, IgG4-DS or SS based on all of the sonographic findings. Sonographic sensitivity for detection of IgG4-DS and of SS, specificity and accuracy were 0.85, 0.80, 0.84 and 0.83, respectively.
Analysis of FDG-PET/CT findings
FDG-PET/CT involving patients with IgG4-DS showed a tendency for abnormal accumulation of 18F-FDG and swelling of both the parotid (median scores 2 and 4, respectively) and submandibular glands (median scores 5 and 4, respectively) (Fig. 4a–d). Separation between IgG4-DS and the other conditions was particularly good regarding the abnormal accumulation of 18F-FDG in the submandibular glands (median score 2 for normal glands and 1 for SS) (Fig. 4c). In relation to this finding, significant differences were observed between IgG4-DS and SS (p <0.0001), and between patients with IgG4-DS and those with normal glands (p <0.0001); however, this was not seen between patients with normal glands and SS (p = 0.1180). Intra-observer agreement rates between the repeat diagnoses were high (kappa values, 0.51–0.89). Regarding all other findings, significant differences (p ≤0.0049) were observed between any two of three diagnoses.
Each case was diagnosed as normal, IgG4-DS or SS based on all of the findings from FDG-PET/CT. The sensitivity for detection of IgG4-DS and of SS, specificity and accuracy using FDG-PET/CT were 0.79, 0.59, 0.81 and 0.73, respectively.
Analysis of CT findings
Figure 5 shows the results of CT findings concerning the parotid glands (Fig. 5a–c), and the submandibular glands (Fig. 5d and e). Patients with IgG4-DS had large parotid (median score 4) and submandibular glands (median score 4) (Fig. 5a and d). Although there were significant differences between normal glands (parotid glands, median score 3, p = 0.0255; submandibular glands, median score 3, p <0.0001), there seemed to be a similar tendency between these two. Conversely, patients with SS had small parotid (median score 3) and submandibular glands (median score 2), especially submandibular glands. Patients with SS had a high score regarding the finding of a salt-and-pepper appearance and/or dot-like calcification (median score 4) (Fig. 5b). Regarding this finding, significant differences were observed between the other two conditions (median score 2 for both normal glands and IgG4-DS) (p <0.0001), although this was not the case for between patients with normal glands and IgG4-DS (p = 0.7945). Superficial enhancement of the parotid glands (median score 2 for IgG4-DS) (Fig. 5c) and the presence of a septum-like structure in the submandibular glands (median score 2 for IgG4-DS) (Fig. 5e), which we considered were some of the characteristic findings of IgG4-DS, showed significant differences between the other two conditions (p <=0.0316); however, overlap was also observed and separation was not so good.
Each case was diagnosed as normal, IgG4-DS or SS based on all of the findings from CT. The sensitivity for detection of IgG4-DS and of SS, specificity and accuracy by CT were 0.64, 0.73, 0.70 and 0.70, respectively.
Analysis of MRI findings
Figure 6 shows the results of MRI evaluation of the parotid glands (Fig. 6a–c), and the submandibular glands (Fig. 6d). Patients with IgG4-DS had large parotid (median score 4) and submandibular glands (median score 5) (Fig. 6a and d). These findings differed significantly between the other two conditions for both the parotid and submandibular glands (p <0.0001). Conversely, patients with SS had small parotid (median score 3) and submandibular glands (median score 2), especially submandibular glands; there was a very clear separation between the other two conditions. Patients with SS had a high score regarding the finding of a salt-and-pepper appearance and/or multiple cystic areas in the parotid glands (median score 4.5) (Fig. 6b); there was also a very clear separation between the other two conditions (median score 1.5 for normal glands and 2 for IgG4-DS). Intra-observer agreement rates between the repeat diagnoses were high (kappa values, 0.54–0.79) regarding this finding. In relation to this finding, significant differences were observed between the other two conditions (p <0.0001), although a significant difference was not seen between patients with normal glands and IgG4-DS (p = 0.4163). Fatty degeneration in the parotid glands, which we considered to be one of the characteristic findings of SS, showed significant differences between the other two conditions (p <=0.0061); however, overlap was also observed and separation was not so good.
Each case was diagnosed as normal, IgG4-DS or SS based on all of the findings. The sensitivity for detection of IgG4-DS and of SS, specificity and accuracy using MRI were 0.80, 0.67, 0.81 and 0.76, respectively.
Comparison of the four imaging modalities
Figure 7 shows the sensitivity, specificity and accuracy of the four imaging modalities; sonography showed the highest levels. Regarding the sensitivity of IgG4-DS, there were significant differences between CT and the other three modalities (p <0.001). Concerning the sensitivity of SS, there were significant differences between sonography and FDG-PET/CT (p = 0.0004) and between CT and FDG-PET/CT (p = 0.0180). In relation to specificity, there were significant differences between sonography and CT (p = 0.0463). Regarding the accuracy, there were significant differences between sonography and CT (p = 0.0001) and between sonography and FDG-PET/CT (p = 0.0058); however, there were no significant differences between sonography and MRI (p = 0.0796).