Simulating Tangle Spread Along Axons, Scientists Predict Tau PET Patterns

Tangles creep through a person’s brain as Alzheimer’s disease progresses; alas, researchers disagree about what drives the spread of this pathology. In the May 26 Nature Communications, a collaborative group led by Alan Evans at McGill University in Montreal and Oskar Hansson at Lund University, Sweden, present more evidence that tangles propagate through axons.

  • Modeling the spread of tangles through axonal connections fits data from PET scans.
  • Amyloid plaques accelerate tangle spread beyond predictions.
  • Tangles advance through medial temporal lobe even in people without amyloid.

The scientists modeled the spread of tangles across the brain using atlases based on either anatomical or functional connectivity, then compared the resulting patterns with actual data from tau PET scans of 312 people. Both maps fit the data, but the anatomical connectivity model performed best, explaining 70 percent of the observed pattern of tangles within the brain. Notably, brain regions with high amyloid plaque load had more tangles than the model predicted. This reinforces the idea that Aβ accelerates tangle spread, as seen in many previous imaging studies and animal data.

Overall, the findings imply that tangles propagate through the human brain along axons the same way they do in mouse models, noted first author Jacob Vogel at McGill. “We’re extending what we’ve seen in animal models to the human context,” he told Alzforum. The manuscript was posted to bioRxiv last year (Jun 2019 news). 

Axonal Spread? The observed pattern of tangles in tau PET scans (left) matches that predicted by an anatomical model of spread (middle) better than a functional model (right). [Courtesy of Vogel et al., Nature Communications.]

Previous tau PET studies have found that tangles propagate through brain networks and match the progression of tangle pathology outlined by Braak staging (Aug 2015 conference news; Mar 2016 news; Ossenkoppele et al., 2019). A recent longitudinal analysis from Michael Ewers' group at Ludwig Maximilian University confirmed that tau pathology spreads into functionally connected regions, rather than those that are simply nearby, which might occur by simple diffusion through the parenchyma (Jan 2020 news). However, most of these PET studies were correlational, comparing global tau signal with brain-wide connectivity, and did not model spread through anatomical pathways.

Evans and co-author Yasser Iturria-Medina at McGill had previously developed an “epidemic spreading model” to simulate the progression of amyloidosis through the brain. In this in silico approach, researchers place a seed in one virtual brain region, which serves as the epicenter of the model, and then allow that seed to propagate to connected regions in a simulated brain. These regions become secondary seeds that disseminate pathology to additional regions. The scientists repeat this process using different epicenters to find the model that best fits the in vivo data (Nov 2014 news). 

Vogel applied this model to tau pathology, using the entorhinal cortex as the epicenter. He modeled tangle spread based on both diffusion tensor imaging (DTI) data, which tracks axons, and resting-state functional MRI data, which identifies functionally connected brain regions. The two maps are similar, but not identical, since brain regions that work together functionally are not always directly connected by axons. Thus, each map produced a slightly different pattern of tangle spread through the brain.

“The methodological approach is highly original. A strength of the paper is that they use two kinds of connectivity atlases to predict spreading,” said Bernard Hanseeuw at Massachusetts General Hospital, Boston. Beau Ances at Washington University in St. Louis likewise appreciated the use of multiple types of imaging data to make the case. “I don’t know many [propagation] studies that use resting-state fMRI, DTI, amyloid PET, and tau PET. That gives you a much better picture than any one of them alone,” Ances told Alzforum.

Amyloid Boosts Tangles. Areas where the tangle-spreading model underestimated the tau PET signal (green) match brain regions that have a high amyloid plaque load (yellow). [Courtesy of Vogel et al., Nature Communications.]

The researchers compared the predicted tangle distribution pattern from these models to cross-sectional flortaucipir tau PET scans from participants in the ADNI and BioFINDER observational studies. Their average age was 72. Volunteers included 162 cognitively healthy people, 89 with mild cognitive impairment, and 61 with AD. The model based on DTI gave the best fit to their data, predicting 70 percent of the observed pattern. The functional connectivity model explained 58 percent of the observed tau PET pattern (see image above). A model of local spread, on the other hand, gave only 48 percent agreement with scans, suggesting that most propagation occurs through long-range connections rather than locally.

“This lends credence to the idea of tangle pathology spreading through physical [axonal] connections, although we can’t conclude that it’s necessarily tau fibrils moving from cell to cell,” Vogel said. Hanseeuw, however, noted that the data are consistent with a model where misfolded tau passes directly from axons to connecting neurons in a prion-like fashion.

The model diverged most from the in vivo data when people were at advanced AD stages. In particular, the model underestimated tangle pathology in regions with amyloid plaques (see image above). The data imply that regional amyloidosis somehow fuels tangles. Numerous prior studies have found plaques are necessary for tangles to escape the medial temporal lobe and wreak havoc through the brain (Aug 2016 news; Feb 2018 news; May 2019 news). 

Even in cognitively healthy people with no amyloid buildup, the epidemic spreading model predicted the pattern of tau-tracer uptake. These individuals typically had subthreshold tracer uptake in medial temporal lobe regions, corresponding to Braak stages I, II, or III. Vogel thinks this accumulation might represent primary age-related tauopathy (PART). He noted that the data question whether PART follows the same progression as Alzheimer’s disease, with the only difference being that the addition of amyloid plaques allows tangles to run wild. Ances was surprised that tangles spread throughout the medial temporal lobe and into the inferior and lateral temporal cortex even in healthy people who were aging well. “That hasn’t been shown much before,” he said.

Another surprise was that tangles often spread asymmetrically in vivo. PET scans of most participants fit a model of progression that started in the right entorhinal cortex, and thus had more pathology on the right side; however, a smaller group fit a model that started on the left side. Those with a right-side epicenter tended to be older, and had more tau-tracer uptake in frontal regions than did left-siders. The reason for this is unclear. “It warrants further study, because it might tell us more about spreading mechanisms,” Vogel noted.

In ongoing work, Vogel and colleagues are investigating sources of individual variation in tau PET scans that are not predicted by the model. For clues, they are studying variant forms of Alzheimer’s disease, such as hippocampal-sparing, posterior cortical atrophy, and logopenic progressive aphasia. “We want to see if we can improve on the epidemic spreading model by incorporating this heterogeneity, and maybe even predict the spread of tau at the individual patient level,” Vogel said.

Vogel also plans to test the model’s predictions against longitudinal scans. That could make tau PET more useful in the clinic, and as an outcome measure in trials.—Madolyn Bowman Rogers

Comments

  1. This is another valuable study adding to the growing body of work implicating brain networks in the pathophysiologic mechanisms of Alzheimer’s disease (AD). This is a very important line of investigation that has implications for prevention, treatment, and monitoring AD.

    There are several prominent theories as to how brain networks are involved in the pathogenesis of AD. Reductionist models begin with molecular pathology and some microscale spreading shaped by neuronal connections in various ways. Complex systems models, such as the cascading network failure model, recast neural networks as active players in the pathophysiology that both leads and responds to protein misfolding. Mathematical models on average connectivity matrices from control populations, like the ones used in this study, cannot differentiate these models, and future work should focus on longitudinal multimodal neuroimaging studies that include measures of brain networks in the same group of individuals that are imaged with amyloid- and tau-PET. These studies should test explicit predictions made by these models regarding the relationship between amyloid, tau, and brain networks across the disease continuum.

    View all comments by David Jones
    • User Profile Image Marcia Ratner
      Boston University Chobanian & Avedisian School of Medicine
    • Posted:

    These are exciting findings. It has been suggested previously that tau pathology begins in locus coeruleus, from which it spreads in a connectome‐dependent pattern during the progression of Alzheimer's from the subclinical to clinically overt stages of the disease. It therefore will be both interesting and important to extend these observations to future investigations looking at the spread of tauopathy from the locus coeruleus to the hippocampus.

    References:

    Satoh A, Iijima KM. Roles of tau pathology in the locus coeruleus (LC) in age-associated pathophysiology and Alzheimer's disease pathogenesis: Potential strategies to protect the LC against aging. Brain Res. 2017 Dec 21; PubMed.

    Olivieri P, Lagarde J, Lehericy S, Valabrègue R, Michel A, Macé P, Caillé F, Gervais P, Bottlaender M, Sarazin M. Early alteration of the locus coeruleus in phenotypic variants of Alzheimer's disease. Ann Clin Transl Neurol. 2019 Jul;6(7):1345-1351. Epub 2019 Jun 23 PubMed.

    View all comments by Marcia Ratner

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Serial PET Nails It: Preclinical AD Means Amyloid, Tau, then Cognitive Decline
  2. New Imaging Data Tells Story of Travelling Tau
  3. Tau PET Aligns Spread of Pathology with Alzheimer’s Staging
  4. Connectivity, Not Proximity, Predicts Tau Spread
  5. The Epidemic in Your Head? New Model Casts Amyloid as Intra-Brain Contagion
  6. Brain Imaging Suggests Aβ Unleashes the Deadly Side of Tau
  7. Imaging Clinches Causal Connections between Aβ, Tau, Circuitry, and Cognition
  8. Longitudinal Tau PET Links Aβ to Subsequent Rise in Cortical Tau

Paper Citations

  1. Ossenkoppele R, Iaccarino L, Schonhaut DR, Brown JA, La Joie R, O'Neil JP, Janabi M, Baker SL, Kramer JH, Gorno-Tempini ML, Miller BL, Rosen HJ, Seeley WW, Jagust WJ, Rabinovici GD. Tau covariance patterns in Alzheimer's disease patients match intrinsic connectivity networks in the healthy brain. Neuroimage Clin. 2019;23:101848. Epub 2019 May 2 PubMed.

Further Reading

Primary Papers

  1. Vogel JW, Iturria-Medina Y, Strandberg OT, Smith R, Levitis E, Evans AC, Hansson O, Alzheimer’s Disease Neuroimaging Initiative, Swedish BioFinder Study. Spread of pathological tau proteins through communicating neurons in human Alzheimer's disease. Nat Commun. 2020 May 26;11(1):2612. PubMed.

Annotate

To make an annotation you must Login or Register.