Peripheral blood stem cells were mobilised with G-CSF from steady-state hae
mopoiesis after previous anthracyclin-containing standard dose chemotherapy
in patients with high-risk breast cancer. 48 samples were obtained from pa
tients with stage II-III breast cancer and greater than or equal to 10 lymp
h nodes, 15 samples from patients with chemotherapy sensitive metastatic di
sease, and 13 samples from women with inflammatory breast cancer. 44 sample
s were first or single leukaphereses and 32 samples were second or third ha
rvests. Aliquots were searched for contaminating tumour cells by immunocyto
chemistry (IC) and cytokeratin-19 reverse transcriptase polymerase chain re
action (rtPCR). The median count of MNCs examined by IC was 2 x 10(6); cDNA
prepared from 2 x 10(7) cells was subjected to PCR. Fifty-nine samples wer
e examined by immunocytochemistry, 36 samples by rtPCR, and 19 samples by b
oth techniques. Samples investigated by IC and rtPCR were judged as positiv
e if there was at least one positive test. On the whole, 42/79 (55.3%) of t
he samples were positive with an insignificant trend to a higher positivity
rate in second or subsequent leukaphereses (52.3% vs 59.3%). The median tu
mour cell load per 10(6) MNCs was low with 0.5 (0-7) cells in all, and a to
tal of 2.2 (0.5-7) cells in positive specimen. Differences in the cancer ce
ll load of first and subsequent leukaphereses and between subgroups of pati
ents were not found. PCR and IC gave consistent results in 63.2%. This phen
omenon can be explained by the greater sensitivity of the molecular method
and by a Poisson distribution of coharvested tumour cells in samples. Tumou
r cell contamination in G-CSF mobilised stem cells from patients with breas
t cancer from steady state haemopoiesis after preceding anthacyclin-contain
ing chemotherapy is frequent, but the tumour cell load is low. To allow a c
omparison of different studies dealing with cancer cell contamination in st
em cells, standardisation of assays is necessary.