Data-driven neuroanatomical subtypes of primary progressive aphasia
Beatrice Taylor,Martina Bocchetta,Cameron Shand,Emily G Todd,Anthipa Chokesuwattanaskul,Sebastian J Crutch,Jason D Warren,Jonathan D Rohrer,Chris J D Hardy,Neil P Oxtoby
DOI: https://doi.org/10.1093/brain/awae314
IF: 14.5
2024-10-09
Brain
Abstract:The primary progressive aphasias are rare, language-led dementias, with three main variants: semantic, non-fluent/agrammatic, and logopenic. Whilst semantic variant has a clear neuroanatomical profile, the non-fluent/agrammatic and logopenic variants are difficult to discriminate from neuroimaging. Previous phenotype-driven studies have characterised neuroanatomical profiles of each variant on MRI. In this work we used a machine learning algorithm known as SuStaIn to discover data-driven neuroanatomical "subtype" progression profiles and performed an in-depth subtype–phenotype analysis to characterise the heterogeneity of primary progressive aphasia. Our study included 270 participants with primary progressive aphasia seen for research in the UCL Queen Square Institute of Neurology Dementia Research Centre, with follow-up scans available for 137 participants. This dataset included individuals diagnosed with all three main variants (semantic: n=94, non-fluent/agrammatic: n=109, logopenic: n=51) as well as individuals with un-specified primary progressive aphasia (n=16). A data set of 66 patients (semantic n=37, non-fluent/agrammatic: n=29) from the ALLFTD North American cohort study, was used to validate our results. MRI scans were segmented and SuStaIn was employed on 19 regions of interest to identify neuroanatomical profiles independent of the diagnosis. We assessed the assignment of subtypes and stages, as well as their longitudinal consistency. We discovered four neuroanatomical subtypes of primary progressive aphasia, labelled S1 (left temporal), S2 (insula), S3 (temporoparietal), S4 (frontoparietal), exhibiting robustness to statistical scrutiny. S1 correlated strongly with semantic variant, while S2, S3, and S4 showed mixed associations with the logopenic and non-fluent/agrammatic variants. Notably, S3 displayed a neuroanatomical signature akin to a logopenic only signature, yet a significant proportion of logopenic cases were allocated to S2. The non-fluent/agrammatic variant demonstrated diverse associations with S2, S3, and S4. No clear relationship emerged between any of the neuroanatomical subtypes and the unspecified cases. At first follow up 84% of patients' subtype assignment was stable, and 91.9% of patients' stage assignment was stable. We partially validated our findings in the ALLFTD dataset, finding comparable qualitative patterns. Our study, leveraging machine learning on a large primary progressive aphasia dataset, delineated four distinct neuroanatomical patterns. Our findings suggest that separable spatio-temporal neuroanatomical phenotypes do exist within the PPA spectrum, but that these are noisy, particularly for nfvPPA and lvPPA. Furthermore, these phenotypes do not always conform to standard formulations of clinico-anatomical correlation. Understanding the multifaceted profiles of the disease, encompassing neuroanatomical, molecular, clinical, and cognitive dimensions, holds potential implications for clinical decision support.
neurosciences,clinical neurology