Objective: We aimed to explore the role of diffusion-weighted imaging (DWI) in the discrimination of pancreatic lesions through meta-analysis.
Methods: The MEDLINE, EMBASE, Cancerlit, and Cochrane Library databases, from January 2001 to August 2011, were searched for studies evaluating the diagnostic performance of DWI in the discrimination of pancreatic lesions. We determined sensitivities and specificities across studies, calculated positive and negative likelihood ratios (LR+ and LR-), and constructed summary receiver operating characteristic curves.
Results: A total of 11 studies with 586 patients, who fulfilled all of the inclusion criteria, were considered for the analysis. No publication bias was found. The pooled sensitivity of DWI was 0.86 [95% (confidence interval (CI), 0.78, 0.91] and the pooled specificity was 0.91 (95% CI, 0.81, 0.96). Overall, LR+ was 9.8 (95% CI, 4.1, 23.3) and LR- was 0.15 (95% CI, 0.09, 0.26). The area under the curve of the summary receiver operating characteristic was 0.94 (95% CI, 0.91-0.96). In subgroup analysis, prospectively designed studies had the highest pooled sensitivity (0.87, 95% CI 0.75, 0.94) and specificity (0.96, 95% CI 0.91, 0.99) (P<0.05). Study sensitivity was not correlated with the prevalence of pancreatic lesions (R=0.1076, P=0.3247).
Conclusion: A limited number of small studies suggest that DWI is a potentially technically feasible measure to differentiate malignant from benign pancreatic lesions. However, it is still controversial and is limited in that it can only distinguish certain lesions. High-quality prospective studies on DWI for the discrimination of pancreatic lesions still need to be conducted.