Subscribe to eTOC

PET Tau Scans Predict Longitudinal Atrophy in Alzheimer's Disease

Article In Brief

A longitudinal study found that buildup of toxic tau proteins in the brain, detected by PET, can predict the location and severity of brain atrophy in Alzheimer's disease later on.


PET scans of patients with early AD showed that binding of a marker of tau protein correlated to regions of atrophy in the brain (colored areas).

The signal from a PET radiotracer that measures phosphorylated tau (p-tau) in the brains of people in the early clinical stages of Alzheimer's disease (AD) was strong enough to predict the rate and location of brain atrophy (measured on a MRI scan) more than a year later, according to a study by scientists at University of California, San Francisco (UCSF).

The stronger the tau-PET signal, the more atrophy was detected on a subsequent MRI scan.

The findings are part of a wealth of new evidence that tau “is driving the bus in the progression of the disease,” said Gil D. Rabinovici, MD, FAAN, professor of neurology at UCSF and senior author of the study published on January 1 in Science Translational Medicine. PET amyloid scans did not have the same power to offer such predictive information.

“This information provides more personalized prognosis for patients and will also be useful in clinical trials to enhance our power to predict a drug effect in a patient-specific way,” said Dr. Rabinovici.

The US Food and Drug Administration is still weighing evidence on the first of several tau_PET radiotracers in development. So far, it remains a research tool. This study opens up the possibility that tau PET could be useful in gaining more information on how and where the pathology will spread in a given individual, and how it might inform clinical symptoms. Tau-PET is now being used as a biomarker in several clinical studies.

Study Design, Findings

PET radiotracers that measure amyloid-beta (Abeta) plaques and tau-containing neurofibrillary tangles are allowing scientists to visualize these hallmark pathological findings in living patients and in those who carry the rare mutations for AD who are destined to develop the disease. These landmark developments allow them to study the relationship between protein aggregation, neurodegeneration, and cognitive impairment, Dr. Rabinovici said.

Cross-sectional data and autopsy studies had already set the stage for what the investigators wanted to do in this prospective study. There is growing evidence that tau is a stronger marker for neurodegeneration and disease progression than amyloid. Therefore, they were hoping to show that a higher baseline (18 F) flortaucipir (FTP) FTP-PET would track with higher atrophy and that the location of the signal would predict a pattern of subsequent brain atrophy. And they were hoping to predict that on a single patient level.

The research team also wanted to see whether FTP-PET could offer up a crystal ball into a patient's clinical deterioration as measured on the Clinical Dementia Rating-Sum of Boxes (CDR-SB) test that assesses disease severity based on a patient's functional decline. They were also interested in figuring out whether these younger patients somehow had a purer form of Alzheimer's that could be measured by a greater tau burden and faster atrophy rates than older patients diagnosed with Alzheimer's. The UCSF Memory and Aging Center has a unique stream of patients coming in for evaluation. Many have rare dementias and others arrive in their 50s and 60s and are ultimately diagnosed with mild cognitive impairment or early AD. Most have no family history of AD.


“This information provides more personalized prognosis for patients and will also be useful in clinical trials to enhance our power to predict a drug effect in a patient-specific way.”—DR. GIL D. RABINOVICI

This study looked at baseline data on Abeta and tau-PET burden using Pittsburgh compound B (PIB) to measure Abeta deposition and FTP to measure phosphorylated tau in 32 patients in the early clinical stages of AD. The group members ranged in age from 49 to 80, and 63 percent of them were under 65. They all had a positive PIB-PET scan before entering the study. Six of the patients had a language-based dementia, logopenic variant primary progressive aphasia, and three had a visual and spatial dementia, posterior cortical atrophy. Everyone in the study had baseline MRI and PET tau and amyloid scans. Another MRI scan was taken 15 months after baseline tests were done.

The two MRI scans provided a three-dimensional map of the progression of atrophy over the year. The researchers first created a group average of PIB, FTP, and the longitudinal atrophy, and then they did the analysis on every patient. Both PET Abeta and p-tau binding showed up in different areas: PIB binding in medial areas, the prefrontal and posterior cingulate/precuneus regions and lateral frontal and temporo-parietal cortices. By contrast, the FTP binding signal was dominant at the temporo-parietal junction and the posterior cingulate/precuneus. There was also a moderate signal in dorsal frontal, occipital, and infero-medial temporal cortices. At a group level, they found that the patients with high FTP-PET binding at baseline developed more severe cortical atrophy over the 15-month period, explained Renaud La Joie, PhD, a UCSF postdoctoral fellow in neurology and first author on the paper. The atrophy was more strongly correlated with the FTP-PET than the PIB-PET.

The investigators also wanted to know whether the location of the PIB and FTP signals could predict the subsequent atrophy on an individual level. Again, FTP-PET did, and the results were striking, said Dr. La Joie. The baseline FTP-PET predicted atrophy 15 months later and that was independent of the baseline MRI volume. That means that the baseline FTP-PET is offering clues as to what is coming. PIB-PET did not correlate with the atrophy measured on the second MRI.

“The younger patients had a higher tau burden and more rapid atrophy,” added Dr. Rabinovici. The spatial similarities of the maps to patterns of atrophy at the second MRI scan were lower in the older patients in the study.

Interestingly, the neuropsychological testing (the CDR-SB) was correlated with subsequent precuneus/posterior cingulate atrophy. But so far, the cognitive decline was not correlated with the baseline FTP-PET. The investigators will continue to scan patients to follow the clinical and pathological trajectories.

“The predictive value of the baseline tau-PET pattern on future atrophy remained substantial even after adjusting for baseline cortical thickness, with tau-PET explaining approximately 40 percent of unique variance in longitudinal atrophy,” the study authors wrote.

The tau-PET predicted the severity of the progression and where the atrophy would be. Scientists could use this information to inform patient prognosis, and in clinical trials to detect treatment effects over time, Dr. Rabinovici said.

Expert Commentary

“The finding that the local distribution of tau pathology, as detected with tau-PET, can predict the pattern of subsequent cortical atrophy is truly novel and important,” said Oskar Hansson, MD, PhD, professor of neurology and head of the clinical memory research unit at Lund University in Sweden. “It opens up the possibility to use tau-PET to predict the clinical outcome of individual patients, especially if future studies can show that the local tau-PET can predict the development of different clinical symptoms at a patient level.”

Dr. Hansson added that the “results are definitely congruent with the hypothesis that accumulation of phosphorylated tau aggregates can cause neurodegeneration and be a driver of the disease progression in Alzheimer's. However, we need drug trials evaluating the clinical effects of anti-tau therapies to be able to settle this issue.” Ultimately, he said that he believes that tau-PET will be more useful than amyloid-PET in patients with established cognitive impairment. “However, amyloid-PET is probably better at detecting Alzheimer's during the early preclinical stages of the disease.”


“This is a solid and important set of findings. Even with a small number of research participants they were able to predict how much brain loss an individual will have going forward linked with tau pathology.”—DR. DAVID M. HOLTZMAN

“This is a solid and important set of findings,” added David M. Holtzman, MD, FAAN, chair of neurology at Washington University in St. Louis. “Even with a small number of research participants they were able to predict how much brain loss an individual will have going forward linked with tau pathology.”

Dr. Holtzman said that his group has seen very similar findings in animal models. “There is brain atrophy that occurs after tau accumulates. Hyper-phosphorylated, aggregated tau is either directly or indirectly causing damage to the brain.”


Dr. Rabinovici receives research support from Avid Radiopharmaceuticals, GE Healthcare, and Life Molecular Imaging; he has received consulting fees or speaking honoraria from Axon Neurosciences, Roche, Eisai, Genentech, and Merck. Dr. Holtzman has served on the scientific advisory board or consulted for Genentech, Eli Lilly, and AbbVie.

Link Up for More Information

• La Joie R, Visani A, Baker SL, et al. Prospective longitudinal atrophy in Alzheimer's disease correlates with the intensity and topography of baseline tau-PET Sci Transl Med 2020; 12(524):eaau5732.
• Bejanin A, Schonhaut DR, La Joie R, et al. Tau pathology and neurodegeneration contribute to cognitive impairment in Alzheimer's disease Brain 2017;140(12):3286–3300.
• Ossenkoppele R, Schonhaut DR, Schöll M, et al. Tau PET patterns mirror clinical and neuroanatomical variability in Alzheimer's disease Brain 2016;139(5):1551–1567.