Data-driven mapping of known and novel fluid biomarker progression profiles in genetic frontotemporal lobar degeneration
Abstract
Recent proteomic studies have identified both established and novel proteins in genetic frontotemporal lobar degeneration (FTLD). However, it remains unclear at what point in the disease these proteins deviate from normal levels and how their trajectories relate to one another. Defining the temporal sequence of protein abnormalities could not only improve disease staging but also help identify biomarkers most sensitive to early disease activity in pathogenic variant carriers. We aimed to apply discriminative event-based modelling (DEBM) to characterize the progression profiles of proteins identified in a previous cross-sectional proteomic analysis. Building on our prior cross-sectional CSF proteomic analysis of genetic FTLD using a proximity extension assay, we selected the top ten significant proteins for each genetic group ( C9orf72 , GRN , MAPT ). We then applied DEBM to characterize temporal dynamics of these proteins separately in each genetic group and evaluated their potential as early disease markers. To validate model performance, each individual was assigned a disease stage according to their position along the estimated disease timeline, based on protein levels and independent of clinical labels. Next, we assessed how well these stages discriminated symptomatic from presymptomatic carriers and non-carriers. Across all genetic groups, NfL consistently became abnormal before TPM3, although the earliest abnormal proteins differed between groups. In C9orf72 , ELAVL4 is the first protein to become abnormal; in GRN SEMA3G and GRN , and in MAPT MMP-10. Estimated individual-level disease stage effectively distinguished symptomatic carriers from presymptomatic carriers and non-carriers, demonstrating high diagnostic accuracy (range AUC 0.74–0.98). Our data-driven findings provide a temporal ordering of multiple CSF proteins, highlighting potential early biomarkers and disease dynamics in different forms of genetic FTLD. In addition, the model’s accurate estimation of disease stages underscores the value of DEBM for patient stratification, offering a promising tool to support clinical trial design.