| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
Department of Clinical Biochemistry, Hillerød Hospital, Helsevej 2, DK-3400 Hillerød, Denmark
2
Department of Oncology, Ålborg Hospital, DK-9000 Ålborg, Denmark
aauthor for correspondence: fax 45-48-24-00-67; e-mail geso{at}fa.dk
In cases of breast cancer, the diagnoses of parenchymal brain metastases and leptomeningeal carcinomatosis are often delayed (1). New inexpensive biochemical diagnostic initiatives have suggested measurement of cytokeratin tumor markers in the cerebrospinal fluid (CSF) (2)(3). Each type of epithelial cell can be characterized by its cytokeratin polypeptide content because the expression pattern varies with the type of epithelium (4)(5)(6). Malignant epithelial tumors and their metastases maintain the cytokeratin expression typical for the tissue of origin (4)(7)(8). Cytokeratins 8, 18, and 19 are abundantly expressed in adenocarcinoma of the breast, only weakly expressed in epithelial cells of the choroid plexus, and not expressed in nerve, glial, ependymal, and meningeal cells (9). Because neither healthy nor malignant epithelial cells are known to secrete cytokeratins, how they appear in the CSF remains unexplained. Some authors have suggested that cytokeratins are released from the large fraction of dead and dying cells in growing tumors (5), whereas others have suggested that cytokeratins are markers of cell proliferation released during mitosis (10). In either case, release may indicate cell turnover in central nervous system (CNS) metastases from adenocarcinoma. Previous reports have addressed the diagnostic performance of the tissue polypeptide antigen (TPA) assay system in the CSF (2). The system uses polyclonal antibodies, and data indicate that they recognize the rod domain of the cytokeratin 8, 18, and 19 fragments; however, the exact structure of the binding epitope is unknown. Recently, another cytokeratin-based assay system that measures tissue polypeptide-specific antigen (TPS) has become commercially available. The TPS assay is a monoclonal test specific to the M3-binding region on cytokeratin 18. The clinical relevance of using assay systems based on monoclonal antibodies compared with the polyclonal TPA assay is undetermined. We therefore compared the diagnostic accuracy of the monoclonal TPS and the polyclonal TPA tests in terms of identifying and excluding CNS metastases among breast cancer patients, following recent guidelines for reporting studies of diagnostic accuracy of medical tests (11).
A prospective cohort of consecutive ambulatory or hospitalized breast cancer patients suspected of increasing intracranial pressure (headache, dizziness, and disturbances of vision and the coordination of extremities) were allocated to the tumor marker study. The cytokeratin analyses were performed without knowledge of the clinical conclusions, and the physicians attending the patients had no knowledge of the cytokeratin data. Informed consent for enrollment was obtained from all patients. The study complied with the Helsinki II Declaration and was approved by the Scientific Ethics Committee of Copenhagen County. If possible, an autopsy included a postmortem examination of the brain. The cerebrum was cut into 0.5- to 1-cm thick frontal slices. The cerebellum, brainstem, and pituitary gland were cut into horizontal sections. Macroscopically suspicious areas were examined histologically. On the basis of data from computer tomography (CT) scans, CSF cytology, and autopsy findings, the patients were divided into the following three groups: (a) patients with parenchymal brain metastases (i.e., a positive CT scan or a positive autopsy <6 months after a negative CT scan); (b) patients with leptomeningeal carcinomatosis (i.e., a positive CSF cytology or an autopsy with leptomeningeal carcinomatosis <6 months after a negative CSF cytology); and (c) patients without CNS metastases (i.e., a negative CT scan and CSF cytology and the absence of CNS metastases within the following 6 months). Patients with both parenchymal and meningeal metastases were allocated to the group with leptomeningeal carcinomatosis. Both the TPS and the TPA tests were performed on RIA systems purchased from Beki Diagnostics AB and Byk-Sangtec. To investigate the reproducibility of the evaluated tests, we included control samples in each assay. The collected CSF samples were stored at -80 °C until assayed in duplicate. The analytical imprecision was 8.6% at 77 units/L for TPS and 9.4% at 71 units/L for TPA.
The diagnostic accuracy of TPS and TPA was investigated for any CNS metastases, parenchymal brain metastases, and for leptomeningeal carcinomatosis. For classification of patients, we implemented the terminology described by Altman (12). The categorical variables true-positive (TP) marker test results, false-negative marker test results, false-positive (FP) marker test results, and true-negative marker test results were compared with Fisher’s test. Variables were statistically comparable between the single markers of TPA and TPS but were not statistically comparable between a single marker and the combination of TPA and TPS. A P value <0.05 was considered statistically significant. The confidence intervals for frequencies were calculated according to Wulff and Schlichting (13). The terminology of Zweig and Robertson (14) was used for calculating the ROC curves.
All 70 enrolled patients were eligible for the study, and no subjects
were excluded. ROC curves of the correlated TP and FP rates for a
series of cutoff points for each marker are provided in Fig. 1
. We considered 65 and 95 units/L as tentative optimal cutoff
values for TPS and TPA, respectively. When we used these values, the TP
rates were 85% (TPS) and 74% (TPA), and the FP rate was 0% for both
markers (Table 1
).
|
|
TPS was determined in CSF samples from 60 patients, TPA was determined
in CSF samples from 59 patients, and both TPS and TPA were determined
in CSF samples from 49 patients. With regard to the number of patients
without metastases (FP plus true-negative marker test results), the
numbers of individuals shown for specificity vary among the three
columns in Table 1
, not because of different cutoff values but because
of the procedure used to calculate data for specificity. When
investigating the accuracy of the markers to identify and exclude any
CNS metastases, we classified increased marker concentrations
associated with parenchymal brain metastases, as well as with
leptomeningeal carcinomatosis as TP test results. When investigating
the accuracy of the markers to identify and exclude parenchymal brain
metastases, we classified increased marker concentrations associated
with leptomeningeal carcinomatosis as FP test results. When
investigating the accuracy of the markers to identify and exclude
leptomeningeal carcinomatosis, we classified increased marker
concentrations associated with parenchymal brain metastases as FP
results. Neither TPS nor TPA was able to discriminate breast cancer
metastases involving the meninges from those confined to the brain
parenchyma (Table 1
). With regard to any CNS metastases, TPS and TPA
supplied similar diagnostic information (P >0.1, Fisher’s
test), and each marker identified 
80% of patients with CNS
metastases.
We found that there was no diagnostic gain when we combined TPS and TPA because none of the patients had increased concentrations of TPS when they had TPA concentrations within reference values, or vice versa; therefore, only one of these markers should be measured. Given the high positive predictive value, the confidence intervals, and the low costs of a cytokeratin test, our data support the view that cytokeratin measurements may be of use as part of a diagnostic protocol for patients suspected of CNS metastases. Thus, breast cancer patients with increased cytokeratin concentrations associated with healthy CNS scan results and the absence of tumor cells in the CSF may have to be evaluated in terms of initiating or continuing further systemic treatment to avoid unnecessary toxicity associated with an ineffective chemotherapy. On the basis of the presented results, we find the data sufficiently encouraging to further investigate the use of cytokeratins as part of a diagnostic strategy in patients suspected of CNS metastases from breast cancer.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, TPS; 
, TPA. TP rates of 85% (TPS) and 74%
(TPA) and a corresponding FP rate of 0% were obtained at cutoff
concentrations of 65 and 95 units/L, respectively.