A Systematic Review of the Diagnostic Accuracy of Anal Cytology in the Screening for Anal Cancer

Abstract

Purpose: Current anal cancer screening programs include anal cytology followed by High Resolution Anoscopy (HRA)-guided biopsy. However, the utility of anal cytology as the initial screening tool is unclear. This systematic review aims to analyze the effectiveness of anal cytology in screening for anal cancer and its precursor lesions. Methods: We searched Medline (1966-2011) and Scopus (1960-2011) and the reference lists of the selected articles for review. We used the MeSH terms: anal cytology, cytology, anal cancer screening, and anoscopy. Only studies comparing anal cytology/cytology with standard histopathology were included. All authors independently performed data extraction using a standard protocol. Results: We retrieved 84 and selected 15 studies, totaling 3962 subjects. Four studies analyzed results from prior anal cytology and HRA-guided biopsies retrospectively. HRA was performed following abnormal anal cytology in three prospective studies only. Both anal cytology and HRA were performed at the onset in the remaining 8 prospective studies. Only 6 studies included female subjects, comprising 3-15% (n=137) of the total subjects recruited. Five studies enrolled HIV-infected subjects alone. Majority of the studies utilized the 2001 Bethesda system for classifying anal intraepithelial neoplasia (AIN). The sensitivity of anal cytology for histology of any grade ranged from 61-98%. With high-grade disease (AIN II, III, squamous cell carcinoma), the sensitivity of anal cytology was considerably lower, varying from 16-81%. The specificity of anal cytology for histology of any grade ranged from 14-67%. With high-grade disease, the specificity of anal cytology was higher but varied widely from 22% to 97%. Three studies evaluated anal cytology separately among HIV-positive (n=654) and HIV-negative groups (n=492). Generally, anal cytology has a higher sensitivity (sensitivity HIV+ vs HIV-) among HIV-positive patients and a higher specificity among HIV-negative patients. The positive predictive value of anal cytology ranged from 61-98% for histology of any grade and 30-81% for high-grade disease. The negative predictive value of anal cytology varied from 11-84% for histology of any grade and 65-91% for high-grade disease. In four studies, anal cytology showed poor correlation with HRA-guided histology, with a weighted kappa statistic of 0.03-0.36. Conclusion: The estimates of sensitivity, specificity and predictive values for anal cytology have considerable variation. Current studies have limited data for women. Based on our review, anal cytology has poor correlation with HRAbiopsy derived histopathology. HRA performance at the onset may improve the accuracy of anal cancer screening.