Multi-Institutional Assessment of Circulating Cell-Free DNA in Cerebrospinal Fluid Facilitates Central Nervous System Lymphoma Diagnosis and Treatment Initiation

Neurosurgery 99:40–49, 2026

This multi-institutional clinical study evaluates a CLIA-certified rapid PCR assay detecting the MYD88 L265P variant in cell-free DNA from cerebrospinal fluid to diagnose central nervous system (CNS) lymphoma. The assay demonstrated 100% specificity, 40% sensitivity, and facilitated earlier treatment initiation, sometimes obviating the need for risky CNS tissue biopsy.

The report details prospective implementation across 19 hospitals, methods for CSF processing and qPCR, diagnostic performance metrics, clinical trajectories of MYD88-positive patients, and implications for using CSF liquid biopsy to accelerate safe, targeted CNS lymphoma therapy.

Clinical need CNS lymphoma diagnosis has historically relied on CNS tissue biopsy, which can delay treatment and carries neurological morbidity risk.

Assay approach A CLIA-certified rapid PCR test was implemented to detect the MYD88 L265P variant in cell-free DNA (cfDNA) from CSF as a minimally invasive diagnostic method.

Deployment scale Prospective testing was conducted over 16 months across 19 hospitals: 201 samples from 184 patients; 19 samples (18 patients) were MYD88 L265P positive, with 2 test failures from inadequate DNA.

Performance characteristics In patients with available records, the assay showed specificity 100%, sensitivity 40%, PPV 100%, NPV 83% (positive LR ∞; negative LR 0.6).

Clinical utility Positive MYD88 results enabled CNS lymphoma–directed treatment initiation, including cases treated without CNS tissue confirmation, with 100% concordance between CSF and CNS tissue biopsy among contemporaneous paired cases (N=8).

Time impact MYD88-positive results often returned before biopsy (median 5.5 vs 10.5 days from admission), and time to treatment was shorter when LP/CSF testing avoided CNS biopsy (7 vs 9 days, P=.048).

Predictors/limitations Detection was more likely with leptomeningeal disease (multivariable P=.017) and DLBCL histology (multivariable P=.027); sensitivity remained modest, implying negative tests still require further workup such as biopsy.

Practice caveat (annotation) Because of modest sensitivity and concerns about spectrum bias/loss to follow-up, MYD88 CSF testing can supplement or obviate biopsy in selected cases, but generally does not replace the need for tissue-based molecular profiling.