engPetit, EmilienCoarelli, GiuliaRosenthal, Liana SSchmahmann, Jeremy DYacoubian, Talene APerlman, Susan LGeschwind, Michael DGomez, Christopher MHawkins, TrevorSubramony, S. H.Shakkottai, Vikram GBushara, Khalaf OMorgan, DavidZesiewicz, Theresa APulst, Stefan MPark, Young WooLenglet, ChristopheKlockgether, ThomasPaulson, Henry LDurr, AlexandraÖz, GülinAshizawa, TetsuoTezenas du Montcel, SophieCunha, PaulinaHurmic, HortenseFaber, JenniferGrobe-Einsler, MarcusRezende, ThiagoKuo, Sheng-HanWilmot, George RPredictive models for ataxia progression and conversion in spinocerebellar ataxia type 1 and 3.info:eu-repo/classification/ddc/610HumansMaleDisease ProgressionFemaleMiddle AgedAdultMagnetic Resonance ImagingMachado-Joseph Disease: diagnostic imagingSpinocerebellar Ataxias: diagnostic imagingNeurofilament Proteins: bloodLongitudinal StudiesAgedbiomarkersbrain MRIclinical scalesnatural history studyspinocerebellar ataxiaNeurofilament Proteinsneurofilament protein LThe READISCA study aims to prepare for clinical trials in spinocerebellar ataxia types 1 and 3 (SCA1 and SCA3). Hence, we searched for predictive variables of ataxia onset (phenoconversion) and progression. Individuals with SCA1 or SCA3 and controls were enrolled from 2018 to 2021 in the USA and Europe. Clinical scores, MRI measures and neurofilament light chain (NfL) levels were assessed annually for 5 years. In the pre-ataxic group at baseline, we compared phenoconverters with non-converters. A Bayesian mixed model was used to model the longitudinal progression of clinical scores and NfL levels. The impact of data-driven selected baseline variables (demographic, clinical and MRI) on the expected Scale for Assessment and Rating of Ataxia progression was tested. Forty-three controls, 55 SCA1 carriers and 124 SCA3 carriers were included; a subset of the cohort (n = 109) had MRI data. Converters from pre-ataxic to ataxic stages represented 5/22 (22%) and 12/38 (32%) for SCA1 and SCA3, respectively. Converters were more depressed (Patient Health Questionnaire 9: 3.9 ± 2.9 versus 2.3 ± 2.6, P = 0.04), had higher plasma NfL levels (17.6 ± 5.7 versus 11.1 ± 5.9 pg/ml, P < 0.0001), more cerebellar white matter atrophy (1.44% ± 0.12% of total intracranial volume versus 1.54% ± 0.16%, P = 0.032) and more Inventory of Non-Ataxia Signs signs (1.8 ± 1.3 versus 0.7 ± 0.8, P = 0.002). All clinical scores except Cerebellar Cognitive Affective Syndrome significantly worsened during the study. NfL levels significantly increased in non-converters and ataxic SCA3 (1.06 ± 0.33 pg/ml/year, P = 0.002 and 0.57 ± 0.21 pg/ml/year, P = 0.01, respectively) but not in controls and ataxic SCA1 (0.31 ± 0.26 pg/ml/year, P = 0.24 and 0.26 ± 0.42 pg/ml/year, P = 0.55, respectively). In the best predictive model of Scale for Assessment and Rating of Ataxia progression after 1 year (R2 = 0.54), factors linked with faster progression were higher functional stage (P < 0.001), higher Composite Cerebellar Functional Score (P = 0.002) and higher total creatine in cerebellar white matter (P = 0.026). Factors significantly linked to conversion, namely NfL levels, depression and lower motor neuron involvement, differ from those driving disease progression. The NfL levels and lower motor neuron signs could be used as predictors of phenoconversion and MRI variables as ataxia progression predictors. Psychological care should be provided in the pre-ataxic phase of the disease.Brain 149(4), 1268 - 1277 (2026). doi:10.1093/brain/awaf408info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionOxford Univ. Press2026info:eu-repo/semantics/closedAccessDEhttps://pub.dzne.de/record/285923https://pub.dzne.de/search?p=id:%22DZNE-2026-00369%22Researchersinfo:eu-repo/semantics/altIdentifier/issn/1460-2156info:eu-repo/semantics/altIdentifier/issn/0006-8950info:eu-repo/semantics/altIdentifier/pmid/pmid:41150672info:eu-repo/semantics/altIdentifier/doi/10.1093/brain/awaf408