BACKGROUND

Pancreatic ductal adenocarcinoma (PDAC) has a dismal prognosis in part due to delayed diagnosis. Even with advances in cross-sectional imaging, small pancreatic malignancies can be missed. We sought to determine the performance of endoscopic ultrasound (EUS) in those without an obvious mass on multi-detector CT scan (MDCT), but with clinical suspicion for pancreatic malignancy.

METHODS

Multiple databases were systematically searched to identify studies that assessed the diagnostic performance of EUS after negative or inconclusive pancreatic protocol MDCT for detection of pancreatic malignancy when clinically suspected. A total of four studies met the inclusion criteria. The point estimates in each study were compared to the summary pooled estimates of sensitivity and specificity with the aid of forest plots. Funnel plots and Egger's test were employed to evaluate possible publication bias.

RESULTS

EUS-guided fine needle aspiration was performed in all studies. EUS was performed in 206 subjects with a clinical suspicion of a pancreatic mass but with an indeterminate MDCT. A pancreatic mass (mean size 21 ± 1.2 mm) was identified in 70% (n = 144) of the subjects, and 42.2% (n = 87) were diagnosed with PDAC. The pooled estimates of EUS for diagnosing pancreatic malignancy in the setting of an indeterminate MDCT were a sensitivity of 85% (95% CI 69-94%), specificity of 58% (95% CI 40-74%), positive predictive value of 77% (69-84%), negative predictive value of 66% (95% CI 53-77%), and an accuracy of 75% (95% CI 67-82). The summary area under the ROC curve was 0.80 (95% CI 0.52-0.89). The funnel plots and Egger's test did not show a significant publication bias.

CONCLUSIONS

The yield of EUS is comparatively higher for the diagnosis of a pancreatic malignancy in patients with suspected cancer, but a non-diagnostic MDCT. Importantly, the majority of the lesions missed on CT represent PDAC, in which early diagnosis is essential.