Current methods for detection of CNS involvement in lymphoma (CSF cytology, flow cytometry) have very limited sensitivity, particularly in cases of parenchymal brain involvement. Early detection is critical to institute CNS-directed therapy and avert dismal outcomes of overt CNS recurrence. Clonotype-specific cfDNA can be detected in the plasma of patients with lymphoma using next-generation sequencing (NGS), and cfDNA-based minimal residual disease (MRD) assay can predict impending recurrence (Roschewski, Lancet Oncol 2015). cfDNA has not been systematically evaluated in the CSF, yet it may hold promise as a sensitive and specific method to detect CNS invasion. To evaluate the ability of an NGS-MRD assay in CSF to detect CNS invasion, we examined CSF and plasma samples from patients with aggressive lymphomas who either had overt CNS disease or who were without known CNS invasion, but at high clinical risk. Genomic DNA from primary tumors was analyzed for tumor-specific clonotype using NGS of rearranged IGK, IGH (VJ or DJ), or IGL loci (Adaptive Biotechnologies; Carlson, Nat Commun 2013). Tumor-specific clonotypes from each case were selected for subsequent tracking by NGS-MRD in CSF and plasma samples. Clonotype copy numbers are expressed per mL for acellular CSF, and clonotype frequency per all B cells. NGS identified median 3 (range, 2-7) dominant immunoglobulin sequences in each primary lymphoma (N=16), with median dominant clonotype frequency 50.3% (range, 26.8-9.28%). In the CSF, the NGS-MRD assay detected the dominant clonotype in 9 out of 16 samples, including all (N=4) with overt CNS invasion (sensitivity=100%), of which 2 had parenchymal disease only with negative CSF cytology, flow cytometry, or IGH PCR. Median detectable cfDNA clonotype in the CSF was 1,077 copies per mL (range, 2-5,620), with median clonotype frequency of 28.4% (range, 0.1-98.5%). cfDNA copy counts were significantly higher in cases with positive CSF cytology than those with parenchymal or clinically occult disease (P=.0016). We observed no significant correlation between the red blood cell count in the CSF and the cfDNA clonotype concentration (P=0.73) or frequency (P=0.62), suggesting that the presence of cfDNA in the CSF was not due to contamination by blood plasma. There was also no evident correlation between cfDNA in plasma and CSF. Within median 11 months of follow-up, 1 of 4 patients (25%) with a positive CSF NGS-MRD assay and no CNS disease developed a fatal CNS recurrence. Our results suggest that NGS-MRD assay for cfDNA in the CSF can identify intraparenchymal or leptomeningeal CNS invasion with high sensitivity, including cases not identifiable by traditional methods. Prognostic significance of detecting lymphoma-specific cfDNA in the CSF of high-risk patients without overt CNS disease will be explored in a larger sample. Pretreatment NGS-MRD assay could be prospectively tested to predict the risk of CNS recurrence and potentially enable more accurate selection of patients for CNS prophylaxis therapy.

Citation Format: Adam J. Olszewski, Anna Chorzalska, Habibe Kurt, Thomas A. Ollila, Diana O. Treaba, Andrew Hsu, Adam Zayac, John L. Reagan, Ilyas Sahin, William Rafelson, Pamela C. Egan, Jordan Robison, John Vatkevich, Chelsea D. Mullins, Max Petersen, Patrycja M. Dubielecka. Clonotypic cell-free DNA (cfDNA) in the cerebrospinal fluid (CSF) of patients with aggressive lymphomas [abstract]. In: Proceedings of the AACR Virtual Meeting: Advances in Malignant Lymphoma; 2020 Aug 17-19. Philadelphia (PA): AACR; Blood Cancer Discov 2020;1(3_Suppl):Abstract nr PO-27.