Diagnostic Performance of CSF Interleukin-10 in Primary Central Nervous System Lymphoma: A Retrospective Study.

BACKGROUND: Primary central nervous system lymphoma (PCNSL) remains a diagnostic challenge due to its radiological overlap with other brain lesions and limitations of stereotactic biopsy, particularly following corticosteroid exposure. Interleukin-10 (IL-10) is a cytokine frequently elevated in PCNSL and has emerged as a potential diagnostic biomarker, yet reported cut-off values vary widely, limiting clinical application. OBJECTIVE: To determine the diagnostic accuracy of CSF IL-10 for differentiating PCNSL from other brain lesions by estimating the optimal cut-off, sensitivity, specificity, and area under the receiver operating characteristic (ROC) curve (AUC), and to explore its potential prognostic value. MATERIALS AND METHODS: We conducted a retrospective observational study of 115 patients who underwent lumbar puncture for diagnostic workup of brain lesions at a tertiary referral center between 2015 and 2020. CSF IL-10 levels were measured using a standardized enzyme-linked immunosorbent assay (ELISA) protocol. Patients with prior corticosteroid use, HIV infection, insufficient CSF, or lacking a definitive diagnosis were excluded. Diagnostic performance was assessed via ROC analysis. The association between CSF IL-10 and progression-free survival (PFS) was explored in patients with histologically confirmed PCNSL receiving standard therapy. RESULTS: Sixty-three patients were diagnosed with PCNSL, and 52 with alternative pathologies, including gliomas, demyelinating diseases, and infections. The mean CSF IL-10 level was significantly higher in PCNSL patients (109.9 pg/mL) compared to non-PCNSL (12.6 pg/mL). Using a cut-off of 20.05 pg/mL, CSF IL-10 showed a sensitivity of 93.7% (95% CI: 84.8-97.5%) and specificity of 88.5% (95% CI: 77.0-94.6%), with an AUC of 0.95 (95% CI: 0.91-0.99). The positive predictive value was 90.8% and the negative predictive value was 92.0%. Higher IL-10 levels were modestly correlated with shorter PFS (R(2) = 0.315, p < 0.015). CONCLUSIONS: CSF IL-10 quantification may serve as a minimally invasive, high-yield adjunct in the diagnosis of PCNSL, particularly when biopsy is delayed or contraindicated. However, as both the diagnostic cut-off and prognostic correlation were derived from the same retrospective cohort, prospective validation is essential prior to clinical adoption.