MRI Trajectories Separate Lewy Body Diseases Into Four Biological Subtypes
Lewy body diseases have long resisted clean clinical sorting. Parkinson’s disease, dementia with Lewy bodies, and prodromal isolated REM sleep behaviour disorder can share the same α-synuclein pathology while diverging sharply in cognition, motor function, hallucinations, and overall tempo of decline. A new data-driven MRI analysis reframes that heterogeneity as a problem of biology rather than labels alone, identifying four transdiagnostic patterns of brain atrophy across 833 individuals.
The study used the Subtype and Stage Inference algorithm to infer spatiotemporal trajectories of atrophy from structural MRI. By focusing on how damage spreads over time rather than where a person happens to sit on a diagnostic spectrum at a single visit, the approach aimed to resolve a central weakness in Lewy body disease classification: the same clinical diagnosis can mask very different underlying courses. The resulting patterns were not merely descriptive. Each subtype mapped to distinct combinations of cognitive, motor, and psychiatric features.
One subtype stood out immediately. The early cortico-limbic, late basal ganglia pattern tracked with a dementia-prone clinical profile across diagnoses. Limbic involvement aligned with the emergence of visual hallucinations, a finding that fits the known role of limbic and associative networks in perception, memory, and reality testing. In practical terms, a patient whose imaging shows early cortical and limbic loss may not simply be “more advanced”; the architecture of decline itself appears different, with implications for prognosis and trial enrollment.
A second subtype followed the opposite sequence, with early basal ganglia and later limbic involvement. That pattern is biologically plausible for a more motor-weighted course, because early basal ganglia disruption would be expected to drive parkinsonism before the spread of atrophy reaches regions more tightly linked to cognition and neuropsychiatric symptoms. The third subtype, early temporo-limbic with late basal ganglia involvement, further complicated the picture by placing memory- and emotion-linked regions at the front of the atrophic cascade. The fourth subtype, early basal ganglia-cingulate with late cortical involvement, suggests a route through motor and salience-related circuitry before more widespread cortical spread.
A Single Diagnosis Can Hide Multiple Disease Engines
The most important message is not that one subtype is “better” or “worse” in a generic sense. It is that the same Lewy body disease label can conceal different neuroanatomical engines of decline. In one patient, the earliest burden may sit in cortico-limbic systems that are tightly tied to cognition and psychiatric symptoms; in another, the basal ganglia may dominate the earliest injury pattern, with dementia arriving later or not in the same way. This kind of stratification matters because a therapy aimed at slowing network spread may perform differently depending on where the disease begins to move through the brain.
The cohort included people with Parkinson’s disease, dementia with Lewy bodies, and prodromal isolated REM sleep behaviour disorder, which gives the analysis a useful transdiagnostic range. Rather than reinforcing a rigid boundary between disorders, the MRI trajectories cut across those diagnostic categories and exposed biologically coherent groupings. That is a major conceptual shift for a field in which clinicians often think in syndromes while pathology moves along networks.
Visual hallucinations emerged as a particularly informative clinical signal. Their association with limbic involvement suggests that hallucinations are not just a late psychiatric complication or a nonspecific marker of worsening illness. They may identify a subtype in which limbic and connected cortico-limbic structures are engaged early, making hallucinations a bedside clue to the underlying anatomy of progression.
The same logic applies to cognitive variability. Dementia risk in Lewy body disease has been difficult to predict because motor symptoms and cognitive decline do not advance in lockstep. A model based on atrophy trajectories can explain why two patients with similar motor severity may diverge markedly in memory, attention, or neuropsychiatric burden. The pattern of spread appears to matter as much as the amount of tissue loss at one time point.
Why Trajectory-Based Imaging Could Change Trial Design
Precision medicine in Lewy body disease has been hampered by noisy phenotypes. Clinical diagnosis alone groups together people whose brain degeneration may be following different routes, which can dilute treatment effects in therapeutic studies and blur prognostic counseling at the bedside. A transdiagnostic subtype framework gives investigators a way to sort participants by biological course rather than by diagnosis label only.
The potential trial use case is straightforward. If a disease-modifying therapy is most effective before limbic spread, a cohort enriched for early cortico-limbic involvement might be the group most likely to show signal. If another intervention is better suited to basal ganglia-dominant disease, then mixing subtypes would create avoidable statistical noise. The study’s interpretation points to exactly this sort of refinement, with MRI serving as a biomarker for hidden progression biology.
There is also a clinical communication benefit. Families often ask why one person with Lewy body disease becomes hallucination-prone and cognitively impaired early while another remains primarily motor-dominant for longer. A trajectory model offers a more concrete answer than the usual appeal to variability alone. The disease is heterogeneous because it follows different anatomic routes through the brain.
The data-driven framework does not eliminate the complexity of Lewy body disease, but it organizes that complexity into interpretable patterns. Four atrophy trajectories, each with distinct sequencing of cortical, limbic, basal ganglia, cingulate, and temporal involvement, explained much of the observed clinical heterogeneity. For neurologists, that means MRI may eventually do more than confirm neurodegeneration; it may identify which biological version of Lewy body disease is already underway.
The study was supported by an Ignition grant from the University of Sydney and University College London Global Engagement Fund. The next step is not to assume every patient fits neatly into one of these categories, but to test how stable these subtypes are over time and whether they can improve prognosis, stratify trials, and guide treatment selection in routine practice.