A European interlaboratory testing of three well-known procedures for immunocytochemical detection of epithelial cells in bone marrow. Results from analysis of normal bone marrow

Background: This investigation intended to study the unspecific background to be expected in normal bone marrow (BM), comparing three well recognized protocols for immunocytochemical detection of disseminated carcinoma cells. The interlaboratory variation in screening and evaluation of stained cells was analyzed and different screening methods were compared.

Methods: BM mononuclear cells (BM MNC) from 48 normal BMs were immunostained in parallel by three participating laboratories. The protocols, based on three different anti-cytokeratin antibodies, have all been in common use for detection of disseminated carcinoma cells: the A45-B/B3 protocol (Hamburg), the CK2 protocol (Augsburg) and the AE1AE3 protocol (Oslo). For all protocols, the immunostained cells were visualized by the same alkaline-phosphatase (AP) detection system (APAAP) followed by detection of the cells by manual screening and by two different automated screening systems (ACIS from Chromavision and MDS1 from Applied Imaging). Detected AP-visualized cells were morphologically classified into unambiguous hematopoietic (Uhc) and questionable cells (Qc, potentially interpreted as tumor cells).

Results: Seven of 48 BMs (15%) harbored > or = 1 AP-visualized cell(s) among 1 x 10(6) BM MNC, both for the A45-B/B3- and for the AE1AE3 protocol, while for CK2 a higher proportion of BMs (21 BMs; 44%) harbored AP-visualized cells (P < 0.01, McNemar's test). The number of Qc was, for all protocols, 1 log lower than the total number of AP-visualized cells. On average, the frequency of Qc was 0.04, 0.08, and 0.02 per 10(6) BM MNC with A45-B/B3, CK2 and AE1AE3, respectively, and the number of Qc-positive BMs 1, 4, and 1. The MDS1 screening sensitivity was similar to manual screening, while ACIS detected fewer cells (P < 0.001, McNemar's test).

Conclusions: All protocols resulted in AP-visualization of occasional hematopoietic cells. However, morphological classification brings the specificity to a satisfactory high level. Approximately 10% of AP-visualized cells were categorized "questionable". The CK2 protocol turned out less specific than the A45-B/B3 and AE1AE3 protocols.