On a stained slide, a Lewy body is a dot. A pathologist puts an antibody against alpha-synuclein onto a thin slice of donated brain tissue, the antibody finds its target, and a round brown mark appears inside a neuron. That mark is what confirms a diagnosis of Parkinson's disease or dementia with Lewy bodies after death, and it is what a generation of drug developers has been aiming at. The dot, however, is a summary, not a description. Zoom in far enough and the summary falls apart.

That is what Notash Shafiei and colleagues at EPFL and the University of Lausanne, working with Amsterdam UMC and the Netherlands Brain Bank, set out to do. Writing in Acta Neuropathologica on 8 July 2026 (the work first appeared as a bioRxiv preprint on 22 December 2025), they applied large-scale correlative light and electron microscopy, or CLEM: stain a section for alpha-synuclein, find the glowing inclusion under the light microscope, then image that exact spot with an electron microscope so the chemistry and the architecture line up. The material came from ten brain donors, four with dementia with Lewy bodies, three with Parkinson's disease, one with Alzheimer's disease and two with no neurological disease, all with a post-mortem delay under six hours. They looked at entorhinal cortex, anterior cingulate cortex, the CA2 region of the hippocampus and the substantia nigra. This is autopsy tissue: a still photograph of the end of a long illness, not a film of how it got there.

The pictures do not show one structure repeated. In the entorhinal cortex the inclusions ran across a spectrum of fibrillar densities, from loose filaments with organelles scattered among them through to tightly packed fibrillar masses. In the anterior cingulate cortex, by contrast, the inclusions the team found were uniformly compact. Lewy neurites, the thread-like deposits inside nerve fibres, came in three recognisable forms: some crowded with membranes, some with a fibril-rich core ringed by organelles and vesicles, some fully fibrillar with no internal organelles at all, a few of those with a dense core at the centre. In the substantia nigra of the donors with dementia with Lewy bodies, the authors saw the full range of classical nigral Lewy body shapes that had previously been described only in Parkinson's disease. The direction of travel here is not new so much as sharpened: an overlapping group reported in Nature Neuroscience in 2019 that alpha-synuclein inclusions in Parkinson's tissue were crowded with membranes and organelles rather than being pure filament. The new work maps that crowding region by region.

Then comes the part that reframes the angle. The authors' own summary is that Lewy body ultrastructure varies by brain region "and not disease identity". Where the two diagnoses did part company was outside the inclusion, in the cell around it. Segmenting more than 10,000 mitochondria from more than 115 CLEM images of substantia nigra and entorhinal cortex, the team reported increased mitochondrial density and enlarged mitochondria in the substantia nigra in Parkinson's disease, increased density confined to the entorhinal cortex in dementia with Lewy bodies, and enlargement seen only in Parkinson's. So the lesion looks regional; the cellular distress around the lesion looks disease-specific. That is a different map from the one most people carry in their heads.

For drug development the implication is uncomfortable in a useful way. A therapy described as targeting aggregated alpha-synuclein is, on this evidence, being pointed at objects that are not made of the same stuff in the same proportions from one region to the next: a densely packed fibrillar mass in the cingulate is not, structurally, the membrane-crowded neurite next door. Roche's prasinezumab, the furthest advanced antibody of this kind, enrolled 586 participants in the phase 2b PADOVA trial between May 2021 and March 2023; the results, published in The Lancet in 2026, did not meet the primary endpoint of time to confirmed motor progression, though prespecified exploratory analyses were read as supporting continued study, and a phase 3 trial is running. Nothing in this microscopy explains that result, and the authors do not claim it does. What the microscopy does is complicate the premise beneath it, and it argues, as the paper puts it, for disease- and region-specific models rather than one generic aggregate.

For a family, the practical content today is close to zero, and it should be said plainly: this changes no test, no prescription and no care plan. Its value is in the framing. If someone has been told that the difference between Parkinson's disease dementia and dementia with Lewy bodies is mostly a matter of timing, this is structural evidence that the two conditions can carry lesions that look alike while the neurons around them are struggling in measurably different ways, in different places. The authors are candid that their cohort size was constrained by tissue availability, and that other structures inside these inclusions, lysosomes, endoplasmic reticulum, lipofuscin and autophagic vesicles among them, still await the same quantitative treatment.