Serum neurofilament light is superior to glial fibrillary acidic protein to distinguish sporadic frontotemporal dementia from late-onset primary psychiatric disorders: a retrospective DIPPA-FTD study.

Background: Sporadic behavioural variant frontotemporal dementia (bvFTD) is often misdiagnosed as late-onset primary psychiatric disorder (PPD). Previous research in small sample sizes has shown that neurofilament light (NfL) and glial fibrillary acidic protein (GFAP) are promising biomarkers to distinguish FTD from PPD. We aimed to investigate the discriminative value of NfL and GFAP in a multicentre cohort of sporadic bvFTD and late-onset PPD. Methods: In total, n=275 sporadic bvFTD and n=82 PPD were included from our DIPPA-FTD study. Baseline serum NfL and GFAP levels were measured using Simoa. Biomarker levels were compared between groups. The effect of age and sex on NfL and GFAP was measured using linear regression models. Discriminative accuracies were assessed using logistic regression models and receiver operating characteristic curves, corrected for age and sex. Within a subset of bvFTD patients who were deceased, the prognostic value of biomarkers was assessed by correlating disease duration (age at death minus age at blood sampling) with biomarker levels. Results: Significantly higher serum median NfL and GFAP levels were found in sporadic bvFTD (NfL 33.3 pg/mL, IQR (19.6-49.6); GFAP 124.5 pg/mL, IQR (83.5-181.6)) compared with PPD (NfL 12.2 pg/mL, IQR (9.8-17.9); GFAP 68.9 pg/mL, IQR (50.6-95.0), both p<0.001). Discriminative performance was AUC=0.872 for NfL, AUC=0.787 for GFAP and AUC=0.878 for NfL+GFAP (DeLong's p for NfL+GFAP versus NfL AUCs: p=0.286). A shorter disease duration was significantly correlated with higher NfL, but not GFAP. Conclusion: Our study found that serum GFAP does not provide additional value as a discriminative marker compared with serum NfL alone when differentiating sporadic bvFTD from late-onset PPD.