Introduction Imaging diagnostic criteria for chronic pancreatitis (CP) have long been debated. Patients with CP frequently undergo Endoscopic Ultrasound (EUS) as part of their diagnostic pathway, but it is unclear if should it be used as an adjunct for diagnosis or to simply exclude more sinister pathology.
Aim To assess the effectiveness of EUS in diagnosis in patients with CP when comparing resection specimens as the gold standard.
Methods All patients who had undergone EUS within 1 year prior to pancreatic resection between 2008 and 2011 at the Royal Liverpool University Hospital were identified from a prospectively maintained database. Resection specimens were compared to EUS and fine needle aspiration (FNA) findings. Agreement was assess using κ statistic and groups were compared using χ2.
Results In the cohort of 93 patients (42% male; median [IQR] age 60 [51–72]) that underwent pancreatic resection, 60 (65%) patients had a clinical suspicion of CP. In the resection specimens CP was determined in 47 (51%) cases; 29 of which had CP alone and 18 with associated neoplasia. EUS correctly reported CP in 28 (60%) patients; 15 alone & 13 with a mass or cyst. EUS falsely identified three as CP, and missed the diagnosis in 19 thus giving a sensitivity of 60% and specificity of 93% (positive predictive value 90%; negative predictive value 70%). Overall agreement EUS and resection histology was moderate (κ 0.53). In the 29 resection specimens where CP was the sole histological diagnosis, there was good agreement with EUS which identified this as the sole pathology in 18 (62%) cases (κ 0.62). However, in these cases, EUS reported CP in association with other pathology incorrectly in 11/29 (38%) of cases, querying neoplasia in 9/29(31%). The sensitivities of radial EUS (50) & linear EUS with FNA (43) were similar (54% & 65%; p=0.47). FNA cytology only confirmed the diagnosis of CP in 2/11(15%) were CP was the sole diagnosis, both of which were already been identified on sonography. The EUS detection of neoplasia in patients with CP was only 53% (9/17).
Conclusion Overall EUS only showed moderate agreement with histology in diagnosing CP, which improved when CP was the sole diagnosis. There was no difference between linear and radial EUS, and FNA did not enhance the diagnosis of CP. This study shows that in context of CP, the EUS diagnosis of a concurrent neoplasia remains to prove difficult. Further detailed evaluation between clinical course, histological and EUS findings may be helpful.
Competing interests None declared.