Characterization of Pancreatic Tumors with Quantitative Perfusion Analysis in Contrast- Enhanced Harmonic Endoscopic Ultrasonography.

Objectives: This study evaluated whether quantitative perfusion analysis with contrast-enhanced harmonic (CH) endoscopic ultrasonography (EUS) characterizes pancreatic tumors, and compared the hemodynamic parameters used to diagnose pancreatic carcinoma. Methods: CH-EUS data from pancreatic tumors of 76 patients were retrospectively analyzed. Time-intensity curves (TIC) were generated to depict changes in signal intensity over time, and 6 parameters were assessed: baseline intensity, peak intensity, time to peak, intensity gain, intensity at 60 s (I60 ), and reduction rate. These parameters were compared between pancreatic carcinomas (n = 41), inflammatory pseudotumors (n = 14), pancreatic neuroendocrine tumors (n = 14), and other tumors (n = 7). All 6 TIC parameters and subjective analysis for diagnosing pancreatic carcinoma were compared. Results: Values of peak intensity and I60 were significantly lower and time to peak was significantly longer in the groups with pancreatic carcinomas than in the other 3 tumor groups (p < 0.05). Reduction rate was significantly higher in pancreatic carcinomas than in pancreatic neuroendocrine tumors (p < 0.05). Areas under the receiver-operating characteristic curves for the diagnosis of pancreatic carcinoma using subjective analysis, baseline intensity, peak intensity, intensity gain, I60, time to peak, and reduction rate, were 0.817, 0.664, 0.810, 0.751, 0.845, 0.777, and 0.725, respectively. I60 was the most accurate parameter for differentiating pancreatic carcinomas from the other groups, giving values of sensitivity/specificity of 92.7/68.6% when optimal cutoffs were chosen. Conclusions: In pancreatic carcinomas, TIC patterns were markedly different from the other tumor types, with I60 being the most accurate diagnostic parameter. Quantitative perfusion analysis is useful for differentiating pancreatic carcinomas from other pancreatic tumors [ABSTRACT FROM AUTHOR]