Use of the term atypical cells in the reporting of ascitic fluid cytology: A caveat.

BACKGROUND

Abdominal paracentesis is a routine diagnostic procedure for assessment of patients with recent onset or worsening of ascites.

OBJECTIVES

The objective of the study is to (1) review clinically confirmed cases of malignancy with negative, atypical, and suspicious cytology reports and provide reasoning for discrepancies and (2) recalculate sensitivity, specificity, and predictive values after review.

MATERIALS AND METHODS

Papanicolaou smears of ascitic fluid paracentesis samples received over one calendar year were reviewed retrospectively by an expert in cytopathology blinded to the final clinical and/or histopathological diagnoses. Cases with discrepancies after review were noted. Sensitivity, specificity, and predictive values were calculated before and after review of slides. Data were analyzed using SPSS version 16.

RESULTS

Malignant etiology was identified in 49/115 cases (42.6%) with female genital tract being the most common site of malignancy (22, 44.8%). The remaining 66 (57.4%) had a benign etiology with hepatic cirrhosis in 42 cases (63.6%). A review revealed discrepancies in five cases, three of which were earlier called negative for malignant cells (one case each of ovarian adenocarcinoma, cecal adenocarcinoma, and cholangiocarcinoma). Two cases of ovarian adenocarcinoma that were reported as atypical/reactive mesothelial hyperplasia showed malignant cells upon review. Sensitivity and specificity after review were 69.4% and 100%, respectively, with 100% positive predictive value.

CONCLUSION

Being a minimally invasive procedure, abdominal paracentesis continues to be an important diagnostic tool in guiding patient management. A proper morphological assessment with adequate clinical information and correlation with other investigations can be used to arrive at a definitive diagnosis in most cases. The term "atypical" can be misleading and is often used for want of clinical information and is best avoided.