Latent disease models estimate symptom onset in familial frontotemporal lobar degeneration and enable novel designs for early‐stage clinical trials: Developments in clinical, radiological and biomarker measurements for the detection and monitoring of early‐stage frontotemporal dementia and implications for treatment development. (7th December 2020)

• Record Type: Journal Article
• Authors: Staffaroni, Adam M.; Quintana, Melanie; Wendelberger, Barbara; Heuer, Hilary W.; Rojas, Julio C.; Cobigo, Yann; Goh, Sheng‐Yang Matthew; Wolf, Amy; Onyike, Chiadi U.; Boeve, Bradley F.; Rosen, Howard J.; Boxer, Adam L.
• Is Part Of: Alzheimer's & dementia. Volume 16(2020)Supplement 5
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-12-07
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.045854
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Measuring changes in the earliest phases of familial frontotemporal lobar degeneration (f‐FTLD) and estimating symptom onset has important implications for clinical care. Early stage clinical trials have begun for f‐FTD, but the low prevalence of f‐FTLD mutation carriers and heterogeneous symptomology challenges trial design. Bayesian disease progression models (DPM) incorporate clinical data, neuroimaging metrics, and fluid biomarker concentrations to estimate latent disease age, providing estimates of symptom onset and progression status. Such models may facilitate enrollment in studies (i.e., inclusion of presymptomatic individuals) and provide more powerful analysis methods for trials. Method: A Bayesian latent variable repeated measures model was used to estimate disease progression, conditional on latent disease age (proximity to symptom onset), in 275 participants with mutations in GRN (n=68), MAPT (n=80) and C9orf72 (n=127). Median follow‐up was one year (range: 0‐4). Jointly modeled longitudinal variables included neuropsychological scores, CDR®+NACC‐FTLD box score, 3T MRI frontotemporal brain volumes, and plasma levels of neurofilament light chain (NfL). Sample size estimates for detecting treatment effects were generated by varying thresholds of estimated proximity to symptom onset. Result: The Bayesian DPM provided a clear picture of clinical and neurological progression within each mutation group. Neuroimaging and NfL captured early progression rates (1 SD from controls) 5‐10 years before symptom onset. After symptom onset, CDR®+NACC‐FTLD box score was the most sensitive measure of disease progression. GRN mutation carriers had the largest annual progression rates (and lowest variability) compared to MAPT and C9orf72 . Between 5 years pre‐ and 10 years post‐onset, CDR®+NACC‐FTLD box score progressed by 2 units/year (SD=2.7) in GRN mutation carriers ‐ almost twice the rate of progression in MAPT (M=1.2;SD=2.5) and C9orf72 (M=1.1;SD=2.2) mutation carriers. Compared to a trial enrolling only early symptomatic cases (CDR®+NACC‐FTLD=0.5 or 1), utilizing latent disease age would facilitate enrollment of asymptomatic cases with the same power. Conclusion: Based on an estimation of symptom onset, Bayesian DPMs suggest clinical trials in GRN mutation carriers would require the smallest sample sizes. Furthermore, leveraging these DPMs would allow novel, powerful trial designs that may facilitate greater participant enrollment.