If you have been reading the research news for a year with a diagnosis in the house, you will know the rhythm by now. One month it is the shape of an amyloid fibril. The next it is a microglial cell that switches on and then exhausts itself. Then a gene nobody had mentioned before, then a receptor, then a complement protein borrowed from the immune system. Each story arrives sealed in its own headline, and none of them tell you how they fit. In July, three researchers at Washington University in St. Louis published the thing that tries to fit them: Michelle Rudman, Jason Ulrich and David Holtzman, of the Hope Center for Neurological Disorders and the Knight Alzheimer's Disease Research Center, writing in Cell on 9 July 2026, volume 189, pages 4193 to 4224. It is a review, which means no new patients and no new experiment; its work is to put the cycle's findings in order and say what they appear to mean together.

The spine of the map is old and the flesh on it is new. Plaques and tangles remain the hallmarks, but the authors are explicit that these are now known to travel with a complex array of other pathologies that culminate in neurodegeneration and cognitive decline. What has changed is the traffic between them. A phrase recurs through the review like a refrain: amyloid-induced tau seeding and spreading. Amyloid, in this reading, is not simply a parallel deposit sitting elsewhere in the brain; it is the thing that gives tau its licence to travel. That single structural claim is why the order of events matters so much to families watching for what a drug can and cannot reach.

The tau chapter is where the review gets uncomfortably concrete. Seeding and propagation, in the model systems it reviews, went with a failing waste-disposal system inside neurons: dysfunctional endolysosomal compartments, an accumulation of multilamellar bodies, the cellular equivalent of bins that stop being emptied. And the process was augmented by increased neuronal activity. Busy circuits, in these models, moved more pathology. This is laboratory work in cells and mice, not a measurement in living people, and it does not mean anyone should be more or less mentally active; it means the researchers now have a lever they can pull in an animal and watch the spread change.

Then the immune chapter, which has quietly become the most consequential. Microglia, the brain's resident clean-up cells, are not bystanders here. The review describes microglial TREM2 promoting the compaction of amyloid plaques, reducing the damaged neurites that crowd around them, reducing amyloid-induced tau seeding and spreading, and pruning overactive synapses by recognising exposed phosphatidylserine. Meanwhile the complement proteins C1q and C3 pile up at synapses and promote synapse loss and neurodegeneration, with TREM2 possibly binding C1q to hold C3 back. The timing detail is the one to hold on to: augmenting TREM2 is described as protective early in amyloid pathology, with limited benefit at later stages, and its effects in pure tauopathy as variable and model dependent. A protective mechanism with an expiry date is a different kind of target than a permanent one.

Genetics closes the circuit rather than sitting beside it. The review's treatment of APOE, the strongest common risk gene, turns on binding strength: isoforms that grip their receptors less tightly may, the authors suggest, be protective against amyloid-induced tau seeding and spreading. The rare Christchurch variant is the worked example. In mouse models it reduced amyloid deposition only modestly, yet strongly increased microglial activation around plaques, and its weaker grip on cell-surface heparan sulfate proteoglycans was linked to greater uptake and degradation of tau by myeloid cells. Read as a chain rather than a list, genetics is not a separate topic from immunity or from tau. It is the setting on the dial.

What does a map buy a family? Mostly it buys reasons. It explains why current disease-modifying treatment works upstream, by removing plaques, and why the authors frame the field's gains as early biomarker-based diagnosis plus amyloid-targeting approaches that are beginning to slow progression in clinical studies rather than stop it. It explains why a drug's window might close, and why two trials of the same mechanism at different disease stages can honestly disagree. And it is candid about the gap between a mouse and a person. On one promising immune target the review says only that "whether TLR inhibition could be leveraged to attenuate disease progression in humans remains to be seen". That sentence is the whole field in miniature: the machinery is better understood than it has ever been, and understanding is not yet the same as interrupting.