PURPOSE/EXPERIMENTAL DESIGN: Immunocytochemical detection of isolated tumor cells (ITCs) in the bone marrow (BM) is a prognostic factor in breast cancer. However, hematopoietic cells (HCs) can occasionally be stained by the techniques used. Morphological evaluation improves the specificity of ITC detection, but optimal separation of ITCs from false-positive HCs needs to be determined. Here, predetermined morphological categories of immunocytochemically (ICC)-positive cells in the BM and the number of detected ITCs were analyzed for association with clinical outcome in 817 early-stage breast cancer patients (median 49 months of follow-up). All ICC+ cells detected were categorized into one of the following groups: (a) tumor cell (TC); (b) uninterpretable cell (UIC); (c) probable HC; or (d) HC.
Results: Among the TC+ patients, 30.6% and 25.9% experienced systemic relapse (SR) and breast cancer death (BCD), respectively, as compared with 13.3% and 8.5% of patients without TCs in the BM (survival analyses: P < 0.001, log-rank). The SR and BCD rate was 19.7% and 15.8% for TC-/UIC+ patients versus 12.5% and 7.4% for TC-/UIC- patients. Survival analyses confirmed that the UIC+ group contained clinically significant cells (P = 0.018, log-rank). No difference in clinical outcome was observed, regardless of whether probable HCs or HCs were present. Analyzing the number of ITC+ cells, SR and BCD occurred in 12.4% and 7.4% of patients with 0 ITCs present, 21.3% and 18.5% of patients with 1 ITC present, 19.4% and 16.7% of patients with 2 ITCs present, and 42.5% and 32.5% of patients with > or = 3 ITCs present.
Conclusions: Morphological categorization of ICC+ cells improves the clinical value of ITC detection in the BM. The presence of only one ITC reduces survival, and a greater number of ITCs further aggravates the prognosis.