Cerebrospinal fluid circulating tumor cells as a predictive biomarker for proton craniospinal irradiation for leptomeningeal metastases.

Abstract

2011 Background: Leptomeningeal metastasis (LM) involves seeding of tumor cells to the cerebrospinal fluid (CSF) and the leptomeninges. Proton craniospinal irradiation (pCSI) has been shown to be potentially effective for patients with solid tumor LM. We evaluated whether CSF circulating tumor cells (CSF-CTC) and neuroimaging correlate with outcomes in patients with LM treated with pCSI. Methods: We reviewed a single-institution retrospective database of patients treated with pCSI for LM between 2018-2020 who had ≥ 3 months (mos.) follow-up and identified 58 patients. Pre-pCSI CSF-CTC using CellSearch and magnetic resonance imaging (MRI) data, and post-pCSI CSF-CTC nadir before initiation of new cancer-directed therapy were assessed. The optimal cutoff for pre-pCSI CSF-CTC was determined using maximally selected rank statistics. Kaplan Meier analysis was used to identify univariate correlates with CNS progression free survival (CNS PFS) and overall survival (OS), calculated from start of pCSI. Multivariate Cox proportional hazards modeling was used to test independence of univariate associations. Results: The median follow-up for patients who were censored (n = 15, 26%) was 15 mos. (interquartile range (IQR): 9 -21). Most patients were diagnosed with lung (n = 27, 47%) or breast cancer (n = 22, 38%). The median CNS PFS and OS were 6 mos. (IQR: 3 – 9) and 8 mos. (IQR: 5 – 18), respectively. Of the 49 patients with pre-pCSI CSF-CTCs analyzed, CSF-CTCs were identified in 43 (88%). Pre-pCSI CSF-CTC< 53/3mL was associated with improved CNS PFS (11.8 vs 6.0 mos., p = 0.01), and a trend toward improved OS (16.7 vs 7.7 mos., p = 0.08). On pre-pCSI MRI, patients with parenchymal brain metastases (n = 33, 57%) had worse OS (6.7 vs 12.7 mos., p = 0.01) but not CNS PFS. Patients with both brain and spine LM (n = 42, 72%) compared to those only one site or no visible disease (n = 16, 28%) showed worse CNS PFS (5.8 vs 7.5 mos., p = 0.03) and OS (7.7 vs 16.7 mos., p = 0.05). In a multivariate model, pre-pCSI CSF-CTC was significantly associated with CNS PFS (p = 0.03) while brain and spine LM on MRI was not (p = 0.20) No patient had an increase in CSF-CTC immediately post-pCSI, and in those with both detectable pre-pCSI CSF-CTCsand a post-pCSImeasurement(n = 29, 50%), the median decrease at nadir was 37/3mL (range: 0-200) occurring at a median of 1.6 mos. (range: 0.5 -5.2). A decrease in CSF-CTC > 37/3mL was associated with improved CNS PFS (7.1 vs 4.4 mos., p = 0.04) but not OS (12.5 vs.7.7 mos., p = 0.2). Conclusions: Proton CSI is an effective treatment for patients with solid tumor LM and can result in prolonged disease control in some patients. Lower CSF-CTC count prior to pCSI and larger changes after pCSI are predictive of survival outcomes, arguing for early pCSI intervention for solid tumor LMD. Early treatment escalation after pCSI can be considered for patients with high pre-pCSI CSF-CTC and a smaller nadir post-pCSI.