Serum GFAP and NfL for non-invasive differentiation of low-grade and high-grade gliomas

Abstract Background Gliomas are the most prevalent primary malignant CNS tumors in adults. Accurate glioma grading is crucial for targeted treatment strategies and predicting disease progression. This study systematically evaluated the potential significance of serum neural tissue markers, including GFAP, S100B, NfL, and UCH-L1, for non-invasive LGG/HGG differentiation. Methods This cross-sectional study enrolled 89 glioma patients, 70 with brain metastases, and 81 healthy controls. Serum biomarker levels were quantified using chemiluminescence immunoassay. Diagnostic performance was assessed through ROC analysis, Spearman correlation, logistic regression, and propensity score matching. Results GFAP, S100B, and UCH-L1 were significantly elevated in glioma patients compared to both brain metastases patients and healthy controls (all p < 0.001); NfL was elevated in glioma versus healthy controls ( p < 0.0001) but showed no significant difference between glioma and brain metastases ( p > 0.05). S100B and UCH-L1 showed good discrimination for glioma detection versus healthy controls (AUCs of 0.894 and 0.856, respectively), but demonstrated limited utility for differentiating glioma from brain metastases. GFAP exhibited a moderate positive correlation with tumor volume ( r = 0.54, 95% CI: 0.380–0.676, p < 0.001). GFAP, NfL, and UCH-L1 levels were associated with WHO grade and IDH mutation status. Multivariate logistic regression identified GFAP, NfL, and age as independent predictors of HGG. The integrated GFAP-NfL-age model achieved an AUC of 0.916 (95% CI: 0.859–0.973, p < 0.001) for distinguishing LGG from HGG, with a bootstrap optimism-corrected AUC of 0.893 (95% CI: 0.827–0.962). Conclusions Serum GFAP and NfL, combined with age, provide promising non-invasive diagnostic accuracy for preoperative glioma grading.