Prospective, blinded assessment of factors influencing the accuracy of biliary cytology interpretation

OBJECTIVES: There is little published data assessing factors that influence the accuracy of biliary cytology. The aim of this study was to (a) prospectively compare interobserver variability among two blinded pathologists interpreting biliary cytology specimens, (b) to describe the predictors of interpathologist agreement, and (c) to characterize the predictors of accurate cytology interpretation. METHODS: In total, 113 consecutive patients undergoing endoscopic retrograde cholangiopancreatography with brushing of suspicious biliary tract strictures were prospectively enrolled to assess routine cytology (RC) accuracy. The initial RC interpretation was performed by the pathologist on duty with the benefit of the patient's clinical information. Subsequent interpretation was performed by two independent pathologists blinded to the patients' clinical details. RESULTS: Of the 113 patients, 67 had malignant strictures and 46 had benign strictures. The sensitivity of RC varied from 9% to 24% (p = 0.02), while the specificity varied from 61% to 100% (p < 0.001). Accuracy varied from 43% to 51% (p = n.s.). The rate of equivocal readings was lowest for the initial interpretation (1.7%), p < 0.0001 versus pathologist 1, p = 0.002 versus pathologist 2. Overall correlation of the blinded pathologists' interpretations was moderate, k = 0.66. Neither cytology accuracy nor interpathologist agreement improved with increasing specimen cellularity. CONCLUSIONS: There is a high rate of interpathologist variation for the biliary cytology interpretation. The knowledge of the patient's clinical information appears to clarify cytology interpretation resulting in fewer equivocal results. We did not detect any reliable predictors of cytology accuracy.