Prospective cohort study comparing transient EUS guided elastography to EUS-FNA for the diagnosis of solid pancreatic mass lesions.

Abstract

Semiquantitative EUS-elastography has been introduced to distinguish between malignant and benign pancreatic lesions. This study investigated whether semiquantitative EUS-guided transient real time elastography increases the diagnostic accuracy for solid pancreatic lesions compared to EUS-FNA. This single centre prospective cohort study included all patients with solitary pancreatic lesions on EUS during one year. Patients underwent EUS-FNA and semiquantitative EUS-elastography during the same session. EUS and elastography results were compared with final diagnosis which was made on the basis of tissue samples and long-term outcome. 91 patients were recruited of which 68 had pancreatic malignancy, 17 showed benign disease and 6 had cystic lesions and were excluded from further analysis. Strain ratios from malignant lesions were significantly higher (24.00; 8.01-43.94 95% CI vs 44.00; 32.42-55.00 95% CI) and ROC analysis indicated optimal cut-off of 24.82 with resulting sensitivity, specificity and accuracy of 77%, 65% and 73% respectively. B-mode EUS and EUS-FNA had an accuracy for the correct diagnosis of malignant lesions of 87% and 85%. When lowering the cut-off strain ratio for elastography to 10 the sensitivity rose to 96% with specificity of 43% and accuracy of 84%, resulting in the least accurate EUS-based method. This was confirmed by pairwise comparison. Semiquantitative EUS-elastography does not add substantial value to the EUS-based assessment of solid pancreatic lesions when compared to B-mode imaging.