Distinct brain atrophy progression subtypes underlie phenoconversion in isolated REM sleep behaviour disorder

Stephen Joza,Aline Delva,Christina Tremblay,Andrew Vo,Marie Filiatrault,Max Tweedale,John-Paul Taylor,John T. O'Brien,Michael Firbank,Alan Thomas,Paul C. Donaghy,Johannes Klein,Michele Hu,Petr Dusek,Stanislav Marecek,Zsoka Varga,Stephane Lehericy,Isabelle Arnulf,Marie Vidailhet,Jean-Christophe Corvol,Jean-Francois Gagnon,Ronald B. Postuma,Alain Dagher,Richard Camicioli,Howard Chertkow,Simon Lewis,Elie Matar,Kaylena A. Ehgoetz Martens,Lachlan Churchill,Michael Sommerauer,Sinah Rottgen,Per Borghammer,Karoline Knudsen,Allan K. Hansen,Dario Arnaldi,Beatrice Orso,Pietro Mattioli,Luca Roccatagliata,Oury Monchi,Shady Rahayel

DOI: https://doi.org/10.1101/2024.09.05.24313131

2024-09-06

Abstract:Background: Synucleinopathies manifest as a spectrum of disorders that vary in features and severity, including idiopathic/isolated REM sleep behaviour disorder (iRBD) and dementia with Lewy bodies. Patterns of brain atrophy in iRBD are already reminiscent of what is later seen in overt disease and are related to cognitive impairment, being associated with the development of dementia with Lewy bodies. However, how brain atrophy begins and progresses remains unclear. Methods: A multicentric cohort of 1,134 participants, including 538 patients with synucleinopathies (451 with polysomnography-confirmed iRBD and 87 with dementia with Lewy bodies) and 596 healthy controls, was recruited from 11 international study centres and underwent T1-weighted MRI imaging and longitudinal clinical assessment. Scans underwent vertex-based cortical surface reconstruction and volumetric segmentation to quantify brain atrophy, followed by parcellation, ComBAT scan harmonization, and piecewise linear z-scoring for age and sex. We applied the unsupervised machine learning algorithm, Subtype and Stage Inference (SuStaIn), to reconstruct spatiotemporal patterns of brain atrophy progression and correlated the distinct subtypes with clinical markers of disease progression. Results: SuStaIn identified two unique subtypes of brain atrophy progression: 1) a "cortical-first" progression subtype characterized by atrophy beginning in the frontal lobes followed by the temporal and parietal areas and remaining cortical areas, with the involvement of subcortical structures at later stages; and 2) a "subcortical-first" progression subtype, which involved atrophy beginning in the limbic areas, then basal ganglia, and only involving cortical structures at late stages. Patients classified to either subtype had higher motor and cognitive disease burden and were more likely to phenoconvert to overt disease compared with those that were not classifiable. Of the 84 iRBD patients who developed overt disease during follow-up, those with a subcortical-first pattern of atrophy were more likely to phenoconvert at earlier SuStaIn stages, particularly to a parkinsonism phenotype. Conversely, later disease stages in both subtypes were associated with more imminent phenoconversion to a dementia phenotype. Conclusions: Patients with synucleinopathy can be classified into distinct patterns of atrophy that correlate with disease burden. This demonstrates insights into underlying disease biology and the potential value of categorizing patients in clinical trials.