[CANCER RESEARCH 51. 2706-2709, May 15, 1991] Sensitivity of Immunocytochemical Detection of Breast Cancer Cells in Human Bone Marrow1 Michael P. Osborne,2 George Y. Wong, Shirin Asina, Lloyd J. Old, Richard J. Cote, and Paul P. Rosen Breast Cancer Research Laboratory, Department of Surgery [M. P. O., 5. A.], Department of Epidemiology and Biostatistics [G. Y. W.], Human Tumor Immunology Laboratory [L. J. O.J, and Department of Pathology ¡K. J. C., P. P. R.J. Memorial Sloan-Kettering Cancer Center. New York, New York 10021 ABSTRACT We have previously shown that occult micrometastases can be detected in the bone marrow of breast cancer patients, at the time of initial treatment, using a panel of epithelial specific monoclonal antibodies indirectly labeled with fluorescein. These monoclonal antibodies permit us to detect cancer cells at a concentration of two/million normal bone marrow cells. Immunofluorescence carries the disadvantage that detailed morphological examination of detected cells cannot be accomplished. A modification of the alkaline phosphatase anti-alkaline phosphatase method has been used to detect cancer cells and to observe their mor phology in human bone marrow. The sensitivity of this method has been examined using an established human metastatic breast cancer cell line (MCF-7) mixed with normal bone marrow cells at various dilutions from 400 cancer cells/106 marrow cells to 10 cancer cells/10' marrow cells. The number of immunocytochemically stained MCF-7 cells counted at each concentration was related to the concentration by a simple nonlinear statistical model. At a concentration of 10 cancer cells/106 bone marrow cells, the model shows that this method has the sensitivity to detect between four and six MCF-7 cells 95% of the time. Extrapolation, using this model, predicts that at the very low concentration of one cancer cell/ IO6marrow cells, there is a 95% chance of detecting the cancer cell. This assay may be a very sensitive method for detecting cancer cells in the bone marrow of breast cancer patients. INTRODUCTION The long-term survival statistics for breast cancer patients are approximately 79% for Stage I (T,NoM0)' (1), 83% for Stage IIA (T2N0Mo) (2), 73% for Stage IIB (T, and T,N,M0) (2), and 20% for Stage III (T.,N,M0) (3). The commonest site for distant metastatic disease is the skeletal system, resulting from tumor cell dissemination to the bone marrow (4). Periop erative staging at the time of initial treatment, such as biochem ical measurement of alkaline phosphatase, bone scanning, skel etal radiography, and routine cytological examination of bone marrow, fails to identify those patients who will relapse. Pre diction of relapse currently rests on the determination of prog nostic parameters in the primary tumor or regional lymph nodes (5, 6). Improved methods for detecting tumor cells in the bone marrow of patients with early stage disease at the time of diagnosis may allow a more accurate assessment of prognosis and aid in selecting candidates for adjuvant systemic therapy. Prior studies using polyclonal antibodies to an epithelial cell membrane antigen (7-12) or monoclonal antibodies to epithe lium (13-17) have shown that occult cancer cells are present in Received 5/21/90; accepted 3/1/91. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1This project was supported by an American Cancer Society Clinical Research Award (PDT-367), the Society of Sloan-Kettering, the Charles and Helen Lazarus Charitable Foundation, and the Iris and B. Gerald Cantor Foundation. 2To whom requests for reprints should be addressed, at Breast Cancer Re search Laboratory', Department of Surgery, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021. 3The abbreviations used are: TNM, tumor, nodes, métastasesclassification; Mab, monoclonal antibody; APAAP. alkaline phosphatase anti-alkaline phospha tase; PBS, phosphate-buffered saline; TBS, Tris-buffered saline; BSA, bovine serum albumin; NLS, nonlinear least squares; WNLS, weighted nonlinear least squares; EMA, epithelial membrane antigen; CI, confidence interval. the bone marrow of patients with breast cancer at the time of diagnosis. In our previous studies we have used Mabs that recognize membrane (C26, T16) and cytoskeletal (AE1) anti gens expressed by epithelial derived cells in an immunofluores- cent assay (18) to detect cancer cells in the bone marrow aspirates of primary breast cancer patients (15). The sensitivity of this assay was shown statistically to be capable of detecting cancer cells at a concentration of 2/106 normal bone marrow cells (18). However, the immunofluorescent method has the disadvantage that the cells observed cannot be studied morpho logically to determine whether they are consistent with cancer cells. Morphological study may be important to exclude false positive cells and tumor cells that are not labeled by the anti body. In order to overcome this problem, we have used a nonfluorescent immunocytochemical method. This technique has been evaluated in a model system using a cell line derived from human metastatic breast cancer cells (MCF-7) mixed with normal human bone marrow. MATERIALS AND METHODS MCF-7 Breast Cancer Cells. The test cells used in this study were MCF-7 human breast cancer cells (Michigan Cancer Foundation, De troit, MI). They were grown in Eagle's minimal essential medium at 37°C,supplemented with Hanks' buffered salts, nonessential amino acids, sodium pyruvate (100 ¿ig/ml),L-glutamine (2 m\i). gentamicin (50 Mg/ml), penicillin (100 lU/ml), streptomycin (100 ¿ig/ml),Fungi- zone (2.5 Mg/ml), bovine insulin (6.6 ^g/ml), and 7% fetal calf serum. The cells were seeded into Petri dishes (60 x 15 mm) at a density of 0.5 x IO6cells/dish. The medium was changed every 3 days. Cells were obtained by scraping from the culture dishes; they were then suspended in RPMI 1640 and filtered through double 30-¿imnylon mesh. Bone Marrow Aspirates. Bone marrow samples from normal volun teer donors were suspended in 2x volumes of RPMI 1640. The suspen sion was layered over a Ficoll-Hypaque solution and centrifuged at 400 x g for 20 min. The resultant interface contained nucleated bone marrow cells and the pellet contained RBC and damaged cells. The interface was collected and washed in RPMI 1640. Test Specimens. MCF-7 cells were serially diluted and added to normal nucleated bone marrow cells. The marrow cells were maintained at a constant concentration (1 x IO6); suspensions of MCF-7 cells at 10, 25, 50, 100, 200, 300, and 400 cancer cells/IO6 marrow cells were made. These specimens were suspended in 50 n\ PBS; thin smears were prepared, fixed in 100% ethanol, and stored at -20°C.The experiments were repeated 6 times for each dilution of MCF-7 cells in normal nucleated bone marrow cells. In addition, six cytospin preparations were made for each of two low concentrations (10 and 25 cancer cells/ IO6 marrow cells) to test immunocytochemistry and immunofluores- cence on the same batches of cells in order to make a direct comparison. Monoclonal Antibodies. Three monoclonal antibodies, C26,4 T16 (19), and AE1 (20-23) (Boehringer Mannheim, Indianapolis, IN), were studied in combination. Each monoclonal antibody reacts with distinct epithelial specific antigens; C26 and T16 react with dimeric cell surface glycoproteins with molecular weights of 40,000/28,000 and 48,000/ 42,000, respectively, while anticytokeratin (AE1) reacts with acidic cytokeratin cytoskeletal antigens. All are epithelial cell specific and 4 Manuscript in preparation. 2706