Concordance of the histological diagnosis of type 1 autoimmune pancreatitis and its distinction from pancreatic ductal adenocarcinoma with endoscopic ultrasound-guided fine needle biopsy specimens: an interobserver agreement study

Abstract

The histological diagnosis of type 1 autoimmune pancreatitis (AIP) based on the findings obtained by an endoscopic ultrasound-guided fine needle biopsy (EUS-FNB) is feasible, but the diagnostic consistency of this method has not been confirmed. We determined the interobserver agreement among 20 pathologists regarding the diagnosis of type 1 AIP, including the distinction from pancreatic ductal adenocarcinoma (PDAC) using large tissue samples obtained by EUS-FNB. After guidance for diagnosing AIP with biopsy tissues was provided, a round 2 was performed. The median sensitivity and specificity for diagnosing PDAC vs. non-neoplastic diseases were 95.2% and 100%, respectively. In groups of specialists (n = 7) and the generalists (n = 13), Fleiss' к-values increased from 0.886 to 0.958 and from 0.750 to 0.816 in round 2. The concordance was fair or moderate for obliterative phlebitis and storiform fibrosis but slight for ductal lesion of type 1 AIP. Discordant results were due to ambiguous findings and biopsy tissue limitations. Among the specialists, the ratio of cases with perfect agreement regarding the presence of storiform fibrosis increased in round 2, but agreement regarding obliterative phlebitis or ductal lesions was not improved. Although the histological definite diagnosis of type 1 AIP was achieved by most observers in > 60% of the cases, the confidence levels varied. Because some ambiguities exist, the histological diagnostic levels based on the diagnostic criteria of type 1 AIP should not be taken for granted. Guidance is effective for improving accurate PDAC diagnoses (notably by recognizing acinar-ductal metaplasia) and for evaluating storiform fibrosis.