Background AECG classification criteria are largely used as diagnostic criteria for primary Sjögren’s syndrome (pSS) in the clinical setting, but their diagnostic value has not been studied in cohorts of patients with suspected pSS since their publication in 2002. They do not include salivary gland ultrasonography (SGUS), which is a non-invasive procedure to assess salivary glands involvement during pSS
Objectives The objective of this work was to determine the accuracy of SGUS for diagnosing pSS and to suggest modifications of AECG classification criteria.
Methods We conducted this cross-sectional study in a prospective cohort of patients with suspected pSS established between 2006 and 2011. Both parotid and submandibular glands were examined by ultrasonography, and the echostructure of each gland was rated between 0 and 4. We also evaluated the size of the glands, and parotid blood flow by Döppler waveform analysis. The reference standard was a clinical diagnosis of pSS by a group of experts blinded to the results of SGUS.
Results Of 158 included patients, 78 patients were diagnosed by the experts as having pSS, including 61 (78.2%) who met AECG criteria. All AECG items were significantly associated with a pSS diagnosis except xerostomy and xerophtalmy. Döppler analysis and measurement of gland size did not display good diagnostic properties. By ROC curve analysis of echostructure scores, the best performance was for the maximal score of one of the four glands in each patient. Optimal cut-off was ≥2/4, with 62.8% sensitivity and 95.0% specificity. The diagnostic value of the echostructure score was similar when the patients were stratified according to the disease duration. A weighted score was constructed using the five variables selected by logistic regression analysis: (salivary flow*1.5) + Schirmer*1.5 + (salivary gland biopsy*3) + (SSA/SSB*4.5) + SGUS*2. By ROC curve analysis, a score ≥5/12.5 had 85.7% sensitivity and 94.9% specificity, compared to 77.9% and 98.7% for AECG criteria (figure). Internal validation through bootstrapping analysis showed good performance of this score.
Conclusions SGUS has good diagnostic properties for pSS. Modifications of AECG criteria including the addition of an SGUS score notably improve their performance.
Disclosure of Interest None Declared