ABSTRACT
This is the first systematic review and meta-analysis of The International System (TIS) for reporting serous fluid cytopathology. Our aims were to present the pooled malignancy rate of each TIS reporting category and the diagnostic accuracy of cytology using this system. Database search using a predefined strategy was followed by study selection, data extraction, study quality assessment, and statistical analysis. Data derived from 16 eligible studies were pooled. The pooled rates of malignancy were as follows: 27% (95% CI; 16%-41%) for "nondiagnostic" (ND), 11% (95% CI; 7%-18%) for negative for malignancy" (NFM), 49% (95% CI; 37%-61%) for "atypia of undetermined significance" (AUS), 90% (95% CI; 81%-95%) for "suspicious for malignancy" (SFM), and 100% (95% CI; 98%-100%) for "positive for malignancy" (MAL). Studies performed exclusively in cancer hospitals showed higher pooled malignancy rates, compared with academic and community hospitals serving the general population, in the ND [40% (95% CI; 21%-62%) vs. 22% (95% CI; 11%-39%)], NFM [20% (95% CI; 13%-30%) vs. 9% (95% CI; 5%-17%)], and AUS categories [55% (95% CI; 47%-63%) vs. 46% (95% CI; 31%-62%)]. Notably, the difference was significant in the NFM category ( P =0.04). When both SFM and MAL cytology interpretations were considered as malignant outcomes, the pooled sensitivity and specificity were 68.74% (95% CI; 59.90%-76.39%) and 98.81% (95% CI; 98.18%-99.22%), respectively. In addition, the diagnostic odds ratio (DOR) was found to be 170.7 (95% CI; 96.2-303.3). Despite its strengths, our study also had some limitations. Therefore, future large-scale longitudinal studies could strengthen the findings of this review.
