Diagnostic performance of EUS for chronic pancreatitis: a comparison with histopathology.

Background: EUS has been proposed as a minimally invasive and accurate test to detect chronic pancreatitis (CP).
Objective: To investigate the correlation between EUS criteria and histopathology grading in patients with presumed CP.
Design: Retrospective study.
Setting and Patients: Patients who received pancreatic surgery according to presumed CP from the Medical University of South Carolina surgical database between 1995 and 2003 were identified and included if EUS was performed within 1 year before surgery. The number of EUS criteria for CP was compared with a histologic fibrosis score (FS).
Main Outcome Measurements: Sensitivity and specificity of number of EUS criteria compared with FS.
Results: Seventy-one patients were identified (38 women). Median FS was 7 (range, 0-12). Of the patients with calcifications: calcifications were detected by EUS in 30 (42%), 14 (47%) had calcifications missed by other imaging modalities, and 28 (93%) were confirmed to have abnormal histology (FS > or = 2). Of the patients without calcifications: in the 41 patients without calcifications on EUS, 36 (88%) had FS > or = 2; median FS was 5 (range, 0-12); the correlation between the number of EUS criteria and FS was low but statistically significant (r = 0.40; P = .01). Three or more EUS criteria provided the best balance of sensitivity (83.3%) and specificity (80.0%) for predicting abnormal histology.
Limitations: Retrospective study. All patients were believed to need surgery.
Conclusions: A threshold of 3 or more EUS criteria provides the best balance of sensitivity and specificity for histologic pancreatic fibrosis. Calcifications seen by EUS but missed by other imaging are common in this group of patients.