Two-color multiparametric method for flow cytometric DNA analysis of carcinomas using staining for cytokeratin and...

Solid tumors contain heterogenous cell populations, resulting in flow cytometric (FCM) DNA quantitations of a mixture of tumor and host cells. Such mixed populations can result in dilution of the tumor cells by the host cells, in difficulty defining the diploid reference mean and in histogram peak overlap, precluding cell-cycle analysis. In this study, epithelial (tumor) cells and contaminating host cells in 100 consecutively accessioned human mammary and colorectal carcinomas were segregated in a multiparametric two-color FCM DNA analysis of intact, ethanol-fixed cells. These two carcinomas and bladder carcinomas contain a cytoskeleton of simple epithelium that is selectively stained with an FITC-labeled monoclonal antibody (MAb) to cytokeratin (CK: CAM 5.2-FITC). This MAb detects the CK 8, CK 18 and CK 19 consistently present in all layers of normal and neoplastic urothelium, colonic epithelium and mammary epithelium. Gating on CK in these tumors enables the nonstaining leukocytes, stromal fibroblasts and endothelial cells to be excluded from DNA analysis. A separate aliquot of each tumor evaluated was labeled with an MAb to leukocyte-common antigen (LCA-FITC) to serve as a patient-specific intrinsic diploid reference standard. Both the CK-labeled and LCA-labeled cells were then dual labeled for DNA with propidium iodide. This method (1) correctly identified the intrinsic diploid (LCA-positive) channel, allowing an accurate definition of normal cell DNA content for calculation of the DNA index; and (2) resulted in an increased sensitivity in the identification of both diploid and abnormal hyperdiploid tumor cell populations. It also (3) limited DNA cell cycle analysis to urothelial, colonic and mammary epithelial cells, the majority of which were neoplastic in carefully selected tumor samples. In addition, this method (4) clarified near-tetraploid populations that overlap the normal nonepithelial G2M region by diminishing the normal G2M peak and accentuating the aneuploid tetraploid G0G1 peak and (5) deconvoluted overlapping histograms composed of normal host and diploid-range or aneuploid tumor cells by gating on tissue-specific markers. This exclusion of host cells in both classes of tumors resulted in more accurate cell-cycle calculations in the former and allowed calculation of the S-phase fractions in the latter.