Diagnostic and Prognostic Value of Circulating Cell-Free DNA in Cerebrospinal Fluid and Plasma in Newly Diagnosed Diffuse Large B Cell Lymphoma

Background:Diffuse large B cell lymphoma (DLBCL) is a clinically and molecularly heterogeneous non-Hodgkin lymphoma. The clinical risk assessment for central nervous system (CNS) recurrence is still challenging in DLBCL due to the limited conventional biomarkers and techniques available. Circulating cell-free DNA (cfDNA), which reflects comprehensive tumor genetic profiles, has emerged as a potential novel strategy in the era of precision medicine. However, the clinical value of cfDNA detected in different samples in newly diagnosed DLBCL in China remains elusive. This study aims to explore the diagnostic and prognostic value of cfDNA detection in cerebrospinal fluid (CSF) and plasma in DLBCL.

Methods:Next-generation sequencing (NGS) was performed on cfDNA derived from CSF, plasma, encompassing single nucleotide variants (SNV), insertion/deletions, copy number variants (CNV), IG related fusions, and clonotypic immunoglobulin gene rearrangements，alongside systemic tumor tissues in newly diagnosed DLBCL patients.

Results:We recruited 30 patients, among whom CSF cfDNA was detected in 19 patients, with 6 (31.6%) being positive for SNV only. The positive rate of CSF cfDNA increased to 36.8% when considering both SNV, IG related fusion, and clonotypic immunoglobulin gene rearrangement. Notably, cfDNA was detected in 100% of CSF samples from 3 patients with known CNS involvement, and in 4 patients without known CNS involvement based on absent evidence of abnormal CSF cytology, flow cytometry, or radiography. The positive rates of SNV, IG related fusion, and clonotypic immunoglobulin gene rearrangement in CSF were 80%, 40%, and 40%, respectively, relatively confirmed by available paired tumor tissues or bone marrow samples. CfDNA was detected in 100% of plasma samples. When accounting for both SNV, IG related fusion, and clonotypic immunoglobulin gene rearrangement, the positive rate increased to 76.7% from 73.3%, which was much higher than paired peripheral blood genomic DNA (20%). The cfDNA detected in plasma exhibited high concordance rates with matched systemic tumor tissues or bone marrow samples, with concordance rates of 100%, 76.9%, and 76.5% for SNV, IG related fusion, and clonotypic immunoglobulin gene rearrangement, respectively. Interestingly, cfDNA was detected in plasma samples of 2 patients while tumor tissues' genomic DNA was negative. In total, 24 gene alterations were identified in CSF, 172 in plasma, and 160 in tumor tissue or bone marrow. Commonly mutated genes in plasma and CSF cfDNA included PIM1 (36.7% vs. 14.2%), MYD88 (33.3% vs. 28.6%), TET2 (30% vs. 28.6%), and CD79B (26.7% vs. 28.6%). The most common types of immunoglobulin gene rearrangement were IGK in different samples, and the common IG related fusions were IGH:: BCL6, IGH:: MYC, and IGH:: BCL2.

Conclusion:Our study suggests that plasma cfDNA exhibits a high concordance rate with tumor tissues and can aid in the precise diagnosis and measurable residual disease monitoring. CSF cfDNA is more sensitive than conventional methods for diagnosing CNS invasion and could serve as a more accurate biomarker for predicting the risk of CNS recurrence among patients with newly diagnosed DLBCL, helping to select patients who may require enhanced CNS prophylaxis. Moreover, synchronous detection of SNV, CNV, IG related fusion, and clonotypic immunoglobulin gene rearrangement could be helpful in improving the positive rate of cfDNA.