Parotid sialography for diagnosing Sjögren syndrome.

OBJECTIVE: Despite the availability of many new imaging procedures, sialography has, after decades of use, maintained its status as the imaging procedure of choice for evaluating the oral component of Sjögren syndrome (SS). In this study, the clinical value of sialography as a diagnostic tool in SS was explored by assessing its diagnostic accuracy, observer bias, and staging potential.

METHODS: One hundred parotid sialograms were interpreted independently in a blinded fashion by 2 trained and 2 expert observers. Sialograms were derived from a group of consecutive patients referred for diagnostics of SS. Patients were categorized as SS and non-SS by the revised European classification criteria.

RESULTS: Trained observers reached a sensitivity of 95 and a specificity of 33% for SS by sialogram, whereas expert observers reached a sensitivity of 87 and a specificity of 84%. There was only "fair" interobserver agreement between trained and expert observers, whereas both expert observers showed "good" agreement with one another, according to Cohen's kappa. Intraobserver agreement was "good" to "very good" for all observers. The 4 different gradations of sialectasia, ie, punctate, globular, cavitary, and destructive, showed a weak but significant correlation with the duration of oral symptoms.

CONCLUSIONS: This study markedly shows that the diagnostic value of parotid sialography for diagnosing SS greatly depends on the skills of the observer, implying that sialography lacks general applicability as a diagnostic tool in SS and requires specific expertise. Nevertheless, given its potentially high sensitivity and specificity in diagnosing SS as well as its useful staging potential, sialography still has its use in the evaluation of the oral component of SS.