DIAGNOSTIC-VALUE OF CEREBROSPINAL-FLUID CYTOLOGY IN COMPARISON WITH TUMOR-MARKER ACTIVITY IN CENTRAL-NERVOUS-SYSTEM METASTASES SECONDARY TOBREAST-CANCER
F. Bach et al., DIAGNOSTIC-VALUE OF CEREBROSPINAL-FLUID CYTOLOGY IN COMPARISON WITH TUMOR-MARKER ACTIVITY IN CENTRAL-NERVOUS-SYSTEM METASTASES SECONDARY TOBREAST-CANCER, Cancer, 72(8), 1993, pp. 2376-2382
Background. Central nervous system (CNS) metastases occur in approxima
tely 35% of patients with breast cancer. Parenchymal brain metastases
(MET) remain undetected in a large proportion of patients, and only 50
% of patients with leptomeningeal carcinomatosis (MC) are diagnosed in
vivo. Methods. Cytology and activity of the tumor markers tissue poly
peptide antigen (TPA) and creatine kinase-BB isoenzyme (CK-BB) were ev
aluated in the cerebrospinal fluid (CSF) in 71 consecutive patients wi
th breast cancer suspected for CNS metastases. Results. Forty-three pa
tients had no CNS metastases, 12 patients had MET, 5 patients had both
MET and MC, and 11 patients had MC alone. Seven of the patients with
MC had an intracerebroventricular (ICV) reservoir inserted, and an add
itional 70 ICV CSF samples from these patients were obtained. In CSF o
btained by lumbar puncture, 11% of the samples were classified as ''su
spicious for malignancy,'' but a very limited interobserver variabilit
y was demonstrated (Kappa test value, 0.81; 95% confidence limits, 0.6
7-0.95%). Fifty-one percent of the ICV CSF samples were classified as
''suspicious for malignancy'' (Kappa test value, 0.58; 95% confidence
limits, 0.34-0.82%). TPA and CK-BB were both measured in 101 CSF speci
mens (61 from lumbar puncture and 40 ICV samples). The differences bet
ween patients with and without CNS metastases were significantly diffe
rent according to TPA (P < 0.00001) and CK-BB (P < 0.00003; Mann-Whitn
ey test). The sensitivity and predictive value of a negative test for
having any CNS metastases (in case of elevated values of either TPA or
CK-BB or both) were 85% (95% confidence limits, 65-96%) and 90% (95%
confidence limits, 76-97%), respectively. In addition, a significant c
orrelation between TPA and CK-BB was demonstrated in CSF from lumbar p
uncture (Spearmans-Rho, 0.49; P < 0.0001) and ICV (Spearmans-Rho, 0.37
; P < 0.02). Conclusions. Cytologic evaluation of CSF obtained by lumb
ar punctures is a reliable procedure. In CSF from ICV reservoirs, cyto
logic evaluation is of limited use, but CK-BB and TPA is of potential
value.