Comparison of ThinPrep® and conventional preparations in pancreatic fine-needle aspiration biopsy

Use of ThinPrep(R) preparation for fine-needle aspiration biopsy (FNA) is gaining popularity. However, there may be a difference in the morphology and the operating characteristics between ThinPrep and conventional methods. The objective of this study was to compare the accuracy of the two methods and to address the pitfalls of ThinPrep preparation in pancreatic FNA. A computer search identified 67 pancreatic FNAs with both conventional smears and ThinPrep preparation during a 19-mo period. These cases, obtained under endoscopic ultrasound-guidance, consisted of 47 malignant neoplasms (44 ductal carcinomas, two mucinous neoplasms, and one islet cell tumor) and 20 benign lesions. Direct smears were prepared first and the remaining material was then put into PreservCyt Solution for ThinPrep slides. All slides were reviewed and the cytologic diagnoses were correlated with histologic and clinical follow-up. Five conventional and 16 ThinPrep specimens were unsatisfactory due to insufficient cellularity. These cases were excluded from the analysis. Among the 62 cases evaluated by conventional preparation, 77% (34) were diagnosed as positive and 14% (seven) atypical/suspicious by conventional smears. For the 51 ThinPrep specimens, 58% (22) were interpreted as positive and 31% (12) atypical/suspicious. The sensitivity, specificity, and accuracy of diagnosing a malignancy were 77%, 100%, and 84% for conventional smears and 58%, 100%, and 67% for ThinPrep preparation, respectively. There were no false positives with either method. However, three benign lesions were interpreted as atypical/suspicious with ThinPrep preparation because of the presence of single atypical cells with distinct nucleoli. One of the two mucinous neoplasms was incorrectly diagnosed with ThinPrep preparation because of lack of mucin. The diagnostic accuracy of pancreatic FNA using ThinPrep is inferior to that of conventional smears. This may be partly due to the use of split sample technique resulting in scant cellularity in ThinPrep preparation and partly due to the differences in morphology between the two preparations. Therefore, the current morphologic criteria may need modification for ThinPrep preparation in pancreatic FNA. Diagn. (C) 2004 Wiley-Liss, Inc.