Multiparameter Flow Cytometry (MFC) of Cerebrospinal Fluid (CSF) from Adults Receiving Hypercvad for Acute Lymphoblastic Leukemia/Lymphoma (ALL) Identifies Patients at Higher Risk of Central Nervous System (CNS) Relapse: A Single-Center Retrospective Analysis

BACKGROUND: CNS directed chemotherapy (chemo) reduces the risk of CNS relapse of ALL, but the optimal way to measure baseline CNS involvement is controversial. Baseline CSF testing can provide risk stratification and often dictates CNS prophylaxis (ppx) during initial treatment. Despite advances, up to 10% of adults still suffer CNS relapse (Blood 2008;112:1646, Cancer 2014;120:3660). Testing typically focuses on conventional cytospin (CC) of CSF, but CC is limited by interobserver variability and insensitivity (Blood 2014;124:3799). Alternatively, MFC is more sensitive (ASH 2018, #658). A previous study in children with ALL found leukemia detection in the CNS by CC or MFC at diagnosis to be associated with an increased risk of both isolated bone marrow and any CNS relapse (ASH 2018, #657). To our knowledge, this use of MFC has not specifically been studied in hyperCVAD, a commonly used chemotherapy regimen for the initial treatment of adult ALL. We hypothesized that baseline lymphoblast detection in the CSF by MFC is predictive of CNS relapse.

METHODS: We included data collected from 11/6/07 to 6/22/20. Patients (pts) ≥18 years old who received ≥4 cycles (equivalent of cycle 2B) of hyperCVAD as first-line therapy were eligible. Pts must have had MFC done on CSF at baseline. Pts with mixed phenotype or insufficient follow-up data were excluded. CNS ppx with hyperCVAD historically uses intra-CSF (ie, intrathecal or intraventricular) methotrexate 12 mg on day 2 and cytarabine 100 mg on day 8 of each cycle for 6-12 doses without cranial radiation, though all patients were treated per physician discretion. CC and MFC results were taken from clinical reports and verified by an independent hematopathologist (Dr. Cherian). Baseline CSF positivity (CSF+) and negativity (CSF-) were defined by MFC, with any percentage of lymphoblasts considered CSF+. COG definitions of CNS status by CC were also used (J Clin Oncol 2016;34:2380). Univariate Cox models evaluated associations to CNS relapse. Kaplan-Meier curves were used to estimate the probabilities of overall survival (OS) and cumulative incidence of CNS relapse.

RESULTS: We identified 92 eligible pts. Disease and treatment characteristics are summarized in Table 1, and results of initial CSF testing are shown in Table 2. Twenty-one pts were CSF+ by MFC at baseline: by CC, 6 of these were positive, 6 were negative, 7 were equivocal (ie, morphologic review was unable to rule out blasts), and 2 did not have baseline CC. None were positive by CC but negative by MFC. Eight of 29 (38%) with circulating blasts had a traumatic (>10 RBC/uL of CSF) first lumbar puncture (LP). Of these 8, 4 (50%) were CSF+. Five (5%) pts had an Ommaya reservoir placed during treatment.

Pts received a median of 8 doses of intra-CSF chemo (range, 5-52). All but 7 pts (8%) received a combination of methotrexate and cytarabine like above. Twenty CSF+ pts (95%) achieved CSF- after a median of 1 dose of intra-CSF chemo (range, 1-5). Out of the 71 CSF- pts, none became CSF+ during initial treatment despite being tested a median of 5 times (range, 2-10).

Eight pts (9%) had CNS relapse (3 isolated, 5 concurrent medullary relapse). Comparisons between baseline CSF testing, intra-CSF treatment, and CNS relapse are shown in Table 2. CSF+ pts received significantly greater doses of intra-CSF chemo. Of the 21 CSF+ pts, 5 (24%) had CNS relapse (2 isolated, 3 concurrent), and out of the 71 CSF- pts, 3 (4%) had CNS relapse (1 isolated, 2 concurrent; p=0.014). Among univariate models of CSF+ vs -, traumatic LP vs not, B vs T lineage, LDH, and WBC, only CSF+ was significantly associated with CNS relapse (hazard ratio 4.8, 95% confidence interval [CI] 1.2-20, p=0.031). Estimated 3-year cumulative incidence of CNS relapse for CSF+ was 15% vs 3.3% for CSF- (Figure 1). Median OS for the whole cohort was not reached, with estimated 3-year OS of 74% (95% CI, 65%-84%).

CONCLUSIONS: MFC of CSF is more predictive of CNS relapse risk than CC in the context of front-line hyperCVAD, questioning the use of CC if MFC is performed. Further, CNS relapses were significantly more frequent when CSF+ despite administration of significantly more intra-CSF chemo. Traumatic LP per se did not increase risk of CNS relapse if CSF- by MFC. Surveillance during treatment by MFC in CSF- pts identified no cases of occult CNS relapse, arguing against this routine practice. Future studies should consider incorporating MFC of CSF into risk-adapted CNS-directed treatment strategies.