Cerebrospinal Fluid Correlates Of L-Asparaginase-Related Thrombotic Events

Thrombosis is a well-described complication of L-asparaginase (asp) treatment in patients with acute lymphoblastic leukemia (ALL), with an incidence of 5-10%. Although asp-associated thrombi can occur anywhere, most are venous and 50% of venous thromboses occur in the central nervous system (CNS); 18% of these are associated with cerebral infarction or stroke. Ninety percent of thrombotic events occur early in the treatment of ALL, during induction and consolidation, consistent with the predominant use of asp during these phases of therapy. Morbidity and mortality have been reported in as many as 50% of patients. Although the mechanism by which asp causes thromboses is thought to be inhibition of protein synthesis resulting in decreased levels of plasma coagulation factors, specific patterns of plasma coagulation factors (including decreases in antithrombotic proteins such as ATIII, protein S, C, and components of the fibrinolytic pathway) have not been good predictors of thrombosis. Risk factors for CNS thrombosis include older age at diagnosis and high risk (HR) disease. Although quantification of proteins in the cerebrospinal fluid (CSF) has been used to identify markers of other diseases, CSF proteomics has not been studied in ALL. We hypothesize that changes in some proteins in the CSF will occur after asp, will be more prominent in patients on HR ALL protocols than in those on standard risk (SR) protocols, and will anticipate CNS thrombosis. In this study, we pilot feasibility and describe serial proteomic analysis of CSF in newly diagnosed patients with ALL or lymphoblastic lymphoma (LL).

Sequential patients, ages 0-30 years, with B-cell ALL, T-cell ALL, or LL diagnosed August 2012-August 2013 at the University of North Carolina and treated according to risk stratified Children’s Oncology Group protocols and the US Intergroup C10403 protocol for adolescents and young adults were offered participation. All patients were on a treatment protocol which utilized PEG-asp on Day 4 of therapy. Following consent approved by our Institutional Review Board, 1 mL of CSF was collected at the time of scheduled lumbar punctures (LP) on Days 0, 8, and 29 of induction. Samples were centrifuged at 1500rpm for 10 minutes within an hour of LP, stored in 50mcL aliquots at -80⁰C. and transported on dry ice to the Duke Proteomics Core Facility. Because of cost ($1000/sample), 5 patients (15 samples) were selected for initial study, including the one patient who had a CNS thrombosis on day 22 of consolidation. Thawed batched duplicate samples were immunodepleted using a MARS-14 LC column (Agilent) and Agilent 1000 HPLC to remove high abundance plasma proteins, and quantitative MS analysis was performed in a label-free fashion using one-dimensional liquid-chromatography tandem mass spectrometry on a Synapt G2 HDMS system (Waters Corporation). Data Analysis was performed in Rosetta Elucidator v3.3 (Rosetta Biosoftware).

Samples were obtained on each of 22 patients (11 male, 11 female). Characteristics of the 5 patient samples set chosen for analysis are shown in Table 1. 636 proteins were identifiable in each sample including several that have potential roles in coagulation. Thirteen proteins met a cut-off of ±1.5-fold change and p<0.05 for unpaired t-test for differential expression, including ADAMTS9, actin, protein Z-dependent protease inhibitor, SERPINA5 and SERPIND1.

In the comparative analysis between the 4 patient sample sets and our index case with clot (subject 13), 28 proteins were differentially expressed in subject 13, on days 8 and 29, including lower levels of anti-thrombin III and plasminogen.

CSF proteomic analysis appears to be feasible and reproducible. These preliminary data suggest that sequential changes in CSF ATIII and plasminogen may predict CNS thrombosis. Reduced assay costs may be possible by targeting proteins of interest for analysis.

No relevant conflicts of interest to declare.