Project Summary/Abstract To enhance the clinical trial readiness of the ongoing READISCA project, this Administrative Supplement application requests approval and funding for two longitudinal biomarker studies: 1) determination of the plasma level of neurofilament light chain (NfL), and 2) additional analyses of the volumetric and diffusion MRI data obtained from subjects who carry a mutation of SCA1 or SCA3. 1) Plasma samples obtained at baseline visits showed that the NfL level can distinguish pre-ataxia samples from control and early-ataxia samples. NfL levels in the follow-up samples are expected to provide the evolution of the NfL level from the pre-ataxia range to the early ataxia range, the resoponsiveness of the plasma NfL level with indices such as the standardized response mean (SRM), and the utility of plasma NfL level in predicting the age at onset in pre-ataxia subjects. All samples to be analyzed are currently stored at the BioSEND repository as a part of Aim1. The SIMOA analysis of the plasma samples will determine the NfL level. The plasma NfL level will be correlated with clinical outcome assessment (COA) data. 2) To augment the value of standard region-of-interest (ROI) based volumetric analyses of the structural MR data and the simple diffusion tensor imaging (DTI) analyses in Aim 2 of READISCA, we will perform add voxel-based analyses for structural MR data and use higher-order models and fixel-based analysis. Our hypothesis is that there are critical thresholds for the evolving plasma NfL level and MR parameters at which the phenoconversion (i.e., ataxia disease onset) occurs in pre-ataxia subjects who carry the SCA1 or SCA3 mutation. We further postulate that plasma and imaging biomarkers together provide the prediction of the conversion and the age at onset stronger than individual biomarkers. We will use the SARA total score ≥3 as the phenoconversion indicator. For statistics, we plan to use survival models to identify predictors of conversion. Joint models that combine survival models and linear mixed models will be used to look for the influence of time varying covariates. The best model will be selected using C statics criteria.