Objective: The purpose of our study was to determine whether gadolinium is necessary in the follow-up evaluation of pancreatic cystic lesions.
Materials and methods: Fifty-six patients with pancreatic cystic lesions detected on initial MRI and who underwent follow-up MRI were identified. Mean cyst size was 1.9 cm, and mean follow-up was 9.1 months. MRI included multiacquisition T1- and T2-weighted sequences before contrast administration and 3D fat-suppressed T1-weighted images before and after gadolinium administration. Two radiologists independently reviewed the entire initial examination and follow-up MRI using only unenhanced T1- and T2-weighted sequences from the second examination. Each radiologist made one of three recommendations: 1, no follow-up necessary or follow-up imaging in 6-12 months; 2, cyst aspiration; or 3, cyst resection. Four weeks later, imaging studies were reevaluated with the contrast-enhanced images from the second examination. A second recommendation using the same outcomes was made. Interobserver and intraobserver variations for the same patient were summarized in terms of kappa coefficients and the percentage of times the decisions were concordant. A 95% CI for the percentage of times management decisions would change without and with gadolinium was calculated.
Results: Concordance between the two different readers for the interpretations (when using the same MRI interpretation technique for follow-up surveillance) was 87.5% with a kappa coefficient to assess interobserver variation of 0.075, suggesting only slight agreement between the two readers. However, treatment recommendations provided by a single reader with and without information from the contrast-enhanced images were discordant only 4.5% of the time. Recommendations were concordant without and with gadolinium 95.5% (107/112; kappa=0.67) of the time, suggesting substantial agreement. A retrospective consensus review of the five cases in which gadolinium effected a change in the observer's recommendation was performed. There was nothing on the gadolinium-enhanced sequences that would specifically alter a change in a management decision, and it is likely that the changes in management decisions in these five cases were simply related to expected variations in categorizing lesions rather than to the use of gadolinium.
Conclusion: The use of gadolinium has minimal impact in the follow-up MR assessment of pancreatic cystic lesions.