[Auxiliary Diagnostic Value of Tumor Markers in the Cerebrospinal Fluid and Blood for Leptomeningeal Metastasis from Non-small Cell Lung Cancer].

BACKGROUND Leptomeningeal metastasis (LM) is one of the most common causes of death in patients with advanced non-small cell lung cancer (NSCLC), which is defined as malignant cells spreading to meninges and cerebrospinal fluid (CSF). Therefore, early diagnosis and timely treatment are essential. CSF cytology is the gold standard for LM diagnosis, however, it has a low sensitivity for diagnosis and can't be used to evaluate the treatment effect. The aim of this study was to assess the clinical value of serum and CSF tumor markers (TM) in the diagnosis and treatment of NSCLC patients with LM. METHODS Nineteen patients with NSCLC-LM and 27 patients with nonmalignant neurological diseases (NMNDs) were included. We tested the levels and positive rates of carbohydrate antigen (CEA), carbohydrate antigen-125 (CA125), cytokeratin 19 fragments (CYFRA21-1) and neurone specific enolase (NSE) in CSF and serum, compared the sensitivity and specificity in the diagnosis of LM between different groups, and analyzed the correlation of detection between serum and CSF. Finally, we measured serum and CSF TM dynamically in 2 patients with NSCLC developing LM in an attempt to correlate these with the treatment response of extracranial and intracranial, respectively. RESULTS The levels and positive rates of TM in CSF and serum in LM group were higher than those in NMNDs (P 0.05). In CSF, CYFRA21-1 has the highest sensitivity (88.2%) and CEA has the best specificity (92.3%) to distinguish patients between LM and NMNDs. For combined detection of CEA, CA125, CYFRA 21-1 and NSE in CSF, when at least CEA or NSE was positive in patients with LM, the sensitivity and negative predictive value were 100.0%, and the specificity was 74.1%. When both CYFRA21-1 and NSE were positive, the specificity and positive predictive value were 100.0%, and the sensitivity was 78.9%. Furthermore, subgroup analysis showed that the detection rates of TM in CSF cytology positive population was higher than that in typical abnormalities magnetic resonance imaging population, but there was no statistical difference (P>0.05). The detection of TM between serum and CSF in LM patients had no significant correlation. Moreover, biochemical properties of CSF from ventricle and lumbar puncture are similar, therefore evaluating the levels of TM in serum and CSF dynamically can be used to assess the extracranial and intracranial treatment effect, respectively. CONCLUSIONS Our study demonstrates that Serum and CSF TM can work as an auxiliary clinical diagnostic tool, which has a potential value in early diagnosis of NSCLC patients with LM. Serial measurement of TM may play an important role in the clinical management of NSCLC patients with LM, which is worthy of further promotion and clinical application.