Accuracy of endoscopic ultrasound in the diagnosis of pancreas divisum in a busy clinical setting

Background: ERCP is of the gold standard for identifying pancreas divisum (PD), however it is associated with a small risk of procedure-related pancreatitis. EUS is capable of tracing the course of the pancreatic ductal system and may have a role in the diagnosis of PD without such procedure related complications.

Aim: To calculate the sensitivity and specificity of blinded EUS in the diagnosis of PD compared to ERCP using a large sample of consecutive patients with and without PD.

Method: Over 2400 patients have undergone both EUS and ERCP procedure at the MUSC over the last 10 years. The EUS and ERCP databases were cross-referenced and patients who underwent EUS prior to ERCP between November 1994 and October 2005 were included in the study. Endosonographers were therefore, by definition, blinded to ERCP results. Patients were included if their ERCP occurred within a maximum of one-month after their EUS. Patients were excluded if the ERCP did not involve a successful pancreatogram, or if they had a known diagnosis of PD prior to the EUS, a mass, or a non-pancreatic EUS. Sample size calculation revealed 65 cases (PD at ERCP) and controls (no PD at ERCP) were needed.

Results: 77 consecutive patients undergoing both procedures with PD on ERCP (cases), and 77 consecutive patients undergoing both procedures without PD on ERCP (controls) were identified. Common indications were abdominal pain with possible pancreatic origin, and recurrent idiopathic pancreatitis. The radial echoendoscope was used for 73% of cases. The sensitivity of EUS for the detection of PD was 50.6% (95% confidence interval: 39.0% to 62.2%), however the specificity was 94.8% (87.2% –98.6%). Predictive values were not calculated due to the artificially set prevalence of 50% (cases:controls=1:1).

Conclusion: In routine practice, the sensitivity of EUS for PD appears poor, but specificity is high.