Iron, Not Prions

01The evidence

• Ultra-high-field MRI detects widespread brain iron deposition in neuroferritinopathy, showing that a single genetic lesion in iron handling produces a distributed anatomical pattern rather than a focal one.Cambridge repository
• Neurodegeneration with brain iron accumulation produces characteristic imaging patterns, including the comet-tail sign in VAC14-associated disease, which suggests the pattern is set by regional metal handling.Neurology, teaching neuroimage

02The tension

Connectome-spread models are the dominant quantitative account of how pathology propagates, and they fit atrophy maps only moderately well. Every attempt to improve the fit adds parameters. Iron maps predict regional vulnerability strongly, but iron is usually treated as a bystander rather than as the landscape.

03The theory

Iron catalyses the Fenton reaction, and lipid peroxidation products propagate damage locally. A region with high iron has a lower threshold for the same insult, so the same molecular burden produces pathology earlier there. The result is a susceptibility map, not a wiring diagram.

Spread then has two components: local amplification in high-iron regions and slow diffusion of seeds between adjacent regions. Diffusion on a gradient produces a different spatial signature from transport along a graph. It produces smooth fronts that respect anatomical boundaries rather than jumping along long-range connections.

The strongest version of the claim is that iron explains the residuals of connectome models. If you take the regions where connectome predictions fail, they should be systematically the high-iron regions, and adding iron as a susceptibility field should absorb the misfit without adding free parameters per region.

This does not deny that misfolded proteins move between cells. It denies that the movement is the rate-limiting step. If susceptibility is what decides where damage appears, then the transport mechanism is almost incidental, and the therapeutic target is metal handling rather than propagation.

04What this is not

05Predictions

• P1Atrophy patterns should fit an iron-diffusion model better than a connectome model, by a decisive information criterion margin, without per-region tuning.
• P2Iron-lowering interventions should slow spread in a pattern predicted by the iron map, not uniformly.
• P3Baseline regional iron should predict future atrophy better than baseline connectivity.

06What would refute it

If connectome models retain their advantage after controlling for regional iron, the landscape account adds nothing and connectivity is doing the work.

07What it would change

It would make iron a staging biomarker and a therapeutic target, and change how spread models are built across neurodegenerative disease.