Abstract

Introduction: Central nervous system involvement (CNSI) in hematologic malignancies confers poor prognosis and is difficult to diagnose requiring clinical, radiological and cytological correlates. Neuroimaging studies generally lack specificity and sensitivity, and although cerebral spinal fluid (CSF) cytology is considered the gold standard, it has a very low sensitivity. Several CSF proteins have been studied as possible biomarkers for CNSI to no avail. CSF levels of beta-amyloid peptides (Ab42) and the proteins S100B, Tau, 14-3-3 and Hexosaminidase B are known to be altered by neurodegenerative disease or neuronal injury. Tau levels have also been shown to increase after intrathecal chemotherapy (ITC) in children with lymphoid malignancies. Because of the absence of a reliable biomarker for CNSI, we initiated a study of patients’ CSF to screen for candidate markers of neuronal damage or inflammation.

Materials and Methods: Fifty-eight adult patients with hematological malignancies (31 high-grade non-Hodgkin’s lymphoma, 22 acute lymphoblastic leukemia, 1 post-transplant lymphoproliferative disorder, 2 chronic myelogenous leukemia in accelerated phase, and 2 acute myelogenous leukemia) and 6 normal control patients undergoing spinal anaesthesia were included. CSF samples were obtained and frozen at −80°C until analysis. For patients receiving ITC, the CSF sample was withdrawn prior to ITC administration. Clinical information collected included: age, sex, diagnosis, CNS irradiation, ITC/systemic chemotherapy, and presence or absence of CNS disease. Suspected CNS disease was defined as the presence of focal neurological signs or symptoms consistent with leptomeningeal or parenchymal disease, or radiographic evidence of CNSI. Proven CNS disease was defined by the finding of malignant cells in the CSF by cytology or flow cytometry. The control population consisted of patients undergoing spinal anaesthesia with no clinical or pathologic evidence of hematological malignancy or CNS disease. Tau, S100B and Ab42 were quantified by ELISA. 14-3-3 was assayed by Western blotting. Hexosaminidase B was assayed using a fluorogenic substrate. Variables potentially influencing the levels or presence of markers were explored using Mann-Whitney U tests, Student’s t-test, simple linear regression, ANOVA or Pearson ?2 tests, as appropriate. To test correlation with CNSI multiple logistic regression models were constructed. Receiver operating characteristic (ROC) curves and test performance statistics were generated when feasible.

Results: One hundred and twenty-eight samples were analyzed from 58 patients and 6 controls. Whereas no difference was observed for S100B or Ab42, increased levels of Tau or positivity for 14-3-3 were associated with diagnosis, CNSI and probably radiotherapy or ITC (Table 1). Multivariate analysis showed that Tau level or 14-3-3 positivity were associated with CNSI after adjusting for confounders (Table 2). For Tau the area under the ROC curve was 0.791. Sensitivity and specificity were 90 and 33% for a cut-off of 200 pg/mL, and 63% and 90% for a cut-off of 500 pg/ml, respectively.

Conclusions: Our results suggest that elevated CSF Tau levels and 14-3-3 positivity may correlate with CNSI in patients with hematologic malignancies and might be useful for diagnosis in suspected cases. These findings warrant further prospective studies.

Table 1: Univariate analysis exploring variables potentially influencing Tau levels or presence of 14-3-3 protein

 P value (2 tailed) for differences in Tau levels P value (2-tailed) for differences in expression of 14-3-3 protein 
Age 0.992 0.792 
Sex 0.372 0.715 
Systemic chemotherapy 0.485 0.588 
Intrathecal Chemotherapy 0.084 0.191 
CNS Radiotherapy 0.163 0.022 
Diagnosis <0.001 <0.001 
Suspected/confirmed CNS infiltration <0.001 <0.001 
 P value (2 tailed) for differences in Tau levels P value (2-tailed) for differences in expression of 14-3-3 protein 
Age 0.992 0.792 
Sex 0.372 0.715 
Systemic chemotherapy 0.485 0.588 
Intrathecal Chemotherapy 0.084 0.191 
CNS Radiotherapy 0.163 0.022 
Diagnosis <0.001 <0.001 
Suspected/confirmed CNS infiltration <0.001 <0.001 

Table 2. Logistic regression models exploring CNSI with CSF markers

 Tau level (Per decile increase) 14-3-3 (Positivity)   
 OR (95% CI) P OR (95% CI) P 
Unadjusted 1.51 (1.29–1.78) <0.001 7.23 (3.12,–16.76) <0.001 
Adjusted for CNS radiotherapy and diagnosis 1.40 (1.17–1.68) <0.001 3.85 (1.48–9.99) 0.006 
 Tau level (Per decile increase) 14-3-3 (Positivity)   
 OR (95% CI) P OR (95% CI) P 
Unadjusted 1.51 (1.29–1.78) <0.001 7.23 (3.12,–16.76) <0.001 
Adjusted for CNS radiotherapy and diagnosis 1.40 (1.17–1.68) <0.001 3.85 (1.48–9.99) 0.006 

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author