This retrospective longitudinal study modeled the temporal ordering of 10 multimodal biomarkers (plasma GFAP, plasma NfL, WMH volume, insular/frontal/gray matter atrophy on MRI, TMT-A, TMT-B, FBI, CDR-SoB) in 489 patients across the sporadic FTLD spectrum (bvFTD, PPA, CBS, PSP, FTD-ALS) over 1,904 patient-visit observations, using sigmoid trajectory modeling.
Plasma GFAP departed from normality earliest (most gradual, pre-symptomatic trajectory), followed by TMT-B (earliest inflection point, θ = 0.3 years; only biomarker significantly abnormal 10–5 years before symptom onset), WMH volume, and plasma NfL. Insula atrophy had the steepest MRI transition (Smax = 0.137 pooled; 0.200 in PPA); CDR-SoB declined most steeply overall (Smax = 0.200). The cascade ordering was robust across 91% of biomarker pairs (precedence probability ≥0.80) and stable in non-carriers of pathogenic variants (Spearman ρ = 0.93).
- Most participants were assessed at or after symptom onset, so pre-symptomatic trajectory estimates are model extrapolations (sigmoid assumptions) with wide confidence intervals, not direct observations. - Single-center design limits generalizability. - CBS, PSP, and FTD-ALS subgroups were too small for standalone cascade analyses; early biomarker positions are most firmly established for cortical phenotypes (bvFTD, PPA).
Combining plasma GFAP with executive function testing (TMT-B) may offer the earliest detectable signal in sporadic FTLD, making them promising screening tools for prodromal disease. Biomarker sensitivity varies by disease stage and phenotype, so clinical trial designs should stratify by FTLD syndrome and target stage-specific endpoints.
