In the present study, cells from body cavity fluids were examined for nuclear DNA and protein content, nuclear volume and receptor expression by multiparametric flow cytometry (FCM), immunocytochemistry (ICC) utilizing epithelial membrane antigen (EMA), and by conventional cytology. Out of 130 pleural and peritoneal fluids examined, 30 were identified by cytology to contain malignant or “suspicious” cells. 12 of these (40.0%) had DNA index (DI) of >1.2 (henceforth referred to as aneuploid). Eight samples with aneuploid cells were from patients with cirrhosis or end stage liver disease who did not have any malignancy. Analysis of nuclear volume vs. DNA content or protein content vs. DNA content could differentiate between subpopulations with diploid DNA content. Ninety-six samples in which we had data from cytology, ICC and FCM were used for statistical analysis. Data from cytology and ICC had the best agreement with a moderate kappa value (overall agreement relative to expected agreement by chance alone) of 0.585. Agreement was low between cytology and FCM (kappa = 0.200) and between ICC and FCM (kappa = 0.205). ICC had significantly higher sensitivity than cytology (79.4% v. 58.8%, p=0.016) or FCM (79.4% v. 38.2%, p=0.001). Cytology had significantly higher specificity than FCM (96.8% vs. 82.3%, p=0.012). The differences in specificity between IHC and FCM (87.1% v. 82.3%, p=0.607) or ICC and cytology (87.1% v.96.8%, p=0.109) were not statistically significant. Sensitivity increased significantly for the combinations cytology + ICC (79.4%, p=0.016) and cytology + FCM (Se=73.5%, p=0.004) as compared to cytology alone (Se= 58.8%). Also, the combination cytology + FCM had higher sensitivity than FCM alone (73.5% v. 38.2%, p<0.0001). By using high resolution flow cytometric analysis of DNA aneuploidy, nuclear volume and expression of markers such as epithelial membrane antigen (EMA), it may be possible to decrease the 50% false negative rate of conventional cytology for detection of malignant cells in body cavity fluids. In addition we may be able to determine the site of common primary tumors such as breast and lung by using ICC and FCM for estrogen receptor and thyroid transcription factor-1(TTF-I) expression, respectively. Grant support NIH CA09733.

[Proc Amer Assoc Cancer Res, Volume 47, 2006]