Suddenly, phospho-tau217 looks to be the most robust plasma biomarker for Alzheimer’s disease yet. That’s the general—and enthusiastic—consensus from two papers and several presentations at this year’s virtual Alzheimer’s Association International Conference, being held July 27 to 31.
- Plasma p-tau217 was measured in several cohorts by two types of test.
- It differentiates AD from controls and from related diseases.
- It outperforms plasma p-tau181, neurofilament light, and imaging markers.
In one study, the marker identified people with the highest likelihood of Alzheimer’s disease, as judged by neuropathology of plaques and tangles, with an AUC or 0.98, i.e., almost 100 percent accuracy. Postmortem neuropathology remains the gold standard for AD diagnosis. In differential diagnosis, plasma p-tau217 also distinguished AD from other neurodegenerative diseases, notably including tauopathies, with high accuracy, beating out other plasma markers including neurofilament light and p-tau181.
“I am tremendously excited about the potential for plasma p-tau217 to advance research, drug development, and care,” Eric Reiman, Banner Alzheimer’s Institute, Phoenix, told Alzforum. Reiman was a senior author on one of the studies, which was led by Oskar Hansson, Lund University, Sweden.
The new data comes from several labs using different cohorts and two types of test—ELISA and mass spectrometry. Hansson’s group, as well as researchers in Adam Boxer’s lab at the University of California, San Francisco, used an immunoassay developed by Jeffrey Dage at Eli Lilly, Indianapolis. Nicolas Barthélemy and colleagues in Randall Bateman’s lab at Washington University, St. Louis, used MS spectrometry to detect p-tau217 after concentrating plasma samples up to 800-fold. All told Alzforum they were extremely encouraged by their results and by the general agreement between their labs.
“From a clinical perspective, p-tau217 assays will be very useful in the future. I actually wish I could use it now, because it would make diagnosis so much easier,” said Boxer. Hansson expressed similar sentiments. “Given it is so accurate, a p-tau217 assay could easily be used to complement cognitive screening in memory clinicals that lack PET or CSF,” he said, adding, “It will be really useful in primary care to aid in diagnosis of people who never get to see a memory disorder specialist.”
The high accuracy of the plasma assay does not come as a complete surprise since, earlier this year, the Hansson and Bateman labs independently reported that p-tau217 in the cerebrospinal fluid outperformed other markers in diagnosing AD (Apr 2020 news). The question was whether p-tau217 in the plasma would do the same.
Using the Dage assay, Hansson’s group measured the marker in three different cohorts totaling 1,402 cases and controls. Co-first authors Sebastian Palmqvist and Shorena Janelidze tested 699 volunteers in BioFINDER-2, a Swedish prospective observational study; a neuropathology cohort comprising 81 AD cases and controls from the Arizona Study of Aging and Neurodegenerative Disorders/Brain and Body Donation Program at Banner Health, Phoenix; and 622 carriers and noncarriers of the presenilin 1 E280A Paisa mutation in Colombia. The data are reported in the July 28 JAMA online.
In the neuropathology cohort of 34 AD patients and 47 controls, plasma p-tau217 correlated tightly with neurofibrillary tangle density (see image below). It identified those with an intermediate-to-high likelihood of having had AD, as judged by CERAD criteria, with an AUC of 0.89. This compares with AUCs of 0.72 and 0.50 for plasma p-tau181 and neurofilament light (NfL), respectively.
The AUC, or area under the curve, reflects both the specificity and sensitivity of a test, with a value of 1.0 being perfection. When the authors considered only those brain samples that had a high likelihood of AD, i.e., with moderate to frequent plaques and Braak stage V/VI neurofibrillary tangles, the AUC for p-tau217 was 0.98.
Tangle Telltale. In a neuropathology series, plasma p-tau217 tightly correlated with the density of neurofibrillary tangles in people who had had Alzheimer’s, but not in controls.
Reiman was amazed at this level of accuracy. “Consider that this is only in those with amyloid plaques, and that this marker also distinguishes amyloid-related from primary tauopathies. What we have is a measure of amyloid-mediated tau pathophysiology, essentially an indication of plaques and tangles in one marker,” he said. To a large majority of Alzheimer’s researchers, amyloid-mediated tau pathophysiology defines Alzheimer’s disease.
The blood marker achieved similarly high accuracy when used to identify people who were still living with a clinical diagnosis of AD. In BioFINDER-2, p-tau217 distinguished 121 AD patients from 224 controls with an AUC of 0.98.
It differentiated those same AD patients from 99 other people with a different neurodegenerative disease with an AUC of 0.96. They included 12 people with vascular dementia, 45 with Parkinson’s disease or multiple system atrophy, 21 people with behavioral variant frontotemporal dementia (bvFTD) or primary progressive aphasia (PPA), and 21 with progressive supranuclear palsy (PSP) or corticobasal syndrome (CBS). Again, in this differential diagnosis, plasma p-tau217 outperformed plasma p-tau181, plasma NfL, and two imaging markers—cortical thickness and hippocampal volume (see image below).
Overall, the new marker was able to distinguish AD from any of these other diseases, or from people who had a mild cognitive impairment but whose Aβ levels were normal, with AUCs that were 0.92 or higher.
Differential Diagnosis. In BioFINDER-2, plasma p-tau217 accurately distinguishes Alzheimer’s from other neurodegenerative diseases, outperforming other plasma and imaging markers.
Elisabeth Thijssen, who worked in Boxer’s lab but is now at Amsterdam University Medical Centers, painted a similar picture of differential diagnosis using plasma donated by 210 people who sought care at the USCF Memory and Aging Center or enrolled in the Advancing Research and Treatment for Frontotemporal Lobar Degeneration (ARTFL) study. These samples comprised 37 healthy controls, 39 with a clinical diagnosis of AD, 35 with mild cognitive impairment, and 99 who had been diagnosed with some form of FTLD. The latter included 33 each with CBS or PSP, 23 with bvFTD, seven with non-fluent PPA, and three with semantic-variant PPA.
Using the Dage assay, Thijssen found that plasma p-tau217 separated AD from controls and from FTLD with AUCs of 0.92 and 0.86, respectively. This was slightly more accurate than p-tau181, which returned AUCs of 0.86 and 0.84. Among 76 volunteers who had had a flortaucipir PET scan, their plasma p-tau217 and p-tau181 values correlated with tracer uptake with AUCs of 0.96 and 0.91, respectively. “Both markers show good results, but I believe plasma p-tau217 is incrementally better,” said Boxer.
For his part, Barthélemy at WashU also found that the p-217 isoform outperformed p-181. He analyzed samples from two cohorts—a discovery cohort comprising 34 people who had enrolled in a stable-isotope-labeling kinetic study to measure Aβ production and clearance, and a validation cohort that included an additional 92 people enrolled in the same study. These data were published in the July 28 Journal of Experimental Medicine.
In the discovery cohort, both plasma p-tau217 and plasma p-tau181 correlated tightly with the respective concentrations of these same markers in the CSF—but only in people who had a positive amyloid PET scan. The p-217 and p-181 tau isoforms identified 15 Aβ-positive samples from 19 negatives with AUCs of 0.98 and 0.95, respectively. In the validation cohort, which had 50 Aβ-positive and 42 negative samples, the AUC for plasma p-tau217 fell slightly to 0.93, but here p-tau181 did poorly, with an AUC of 0.72.
Why does p-tau217 seem to be better than even p-tau181? For one, its dynamic range is luxuriously large. In BioFINDER-2, plasma p-tau217 was seven times higher in AD than in controls. That is fold. Not percent. Thijssen found it to be 5.7-fold higher in AD, while p-tau181 was 4.5-fold higher. These are expansive, newly comfortable ranges to work with for Alzheimer’s researchers, who are more used to wondering whether a 5 or 10 percent reduction in CSF tau in a therapeutic trial might be meaningful, or how robust an Aβ42/40 ratio change within a 15 percent range can be.
Barthélemy believes p-tau217 is the better marker because very little of it is present in control plasma. It is barely detectable in controls, whereas small amounts of p-tau181 are detectable. Herein lies the beauty of mass-spectrometry analysis. It detects both phosphorylated and unphosphorylated fragments, allowing Barthélemy and colleagues to calculate the occupancy of phospho groups at the various amino acids. What they found was telling.
In control CSF, about 2.6 percent of all tau fragments containing the 217 site were phosphorylated at that position. In control plasma, only 0.6 percent of the fragments were phosphorylated. Barthélemy thinks this is because the plasma contains some tau that originates from the body’s peripheral organs and is not phosphorylated at 217. In essence, in the blood, peripheral tau dilutes the tau that seeps out of the brain. For p-tau181, the scenario was slightly different. In the CSF, 13.5 percent was phosphorylated, while in the plasma 6.8 percent was. In essence, it seems that plasma contains about half as much p-tau181 as does CSF, but 4.3 times less p-tau217. “We drew the conclusion that some other source of tau in the plasma has a different phosphorylation status than does CNS tau,” said Barthélemy. “P-tau217 is highly specific to the CNS, explaining why plasma p-tau217 outperforms plasma p-tau181.”
Early Diagnostic Marker?
How soon in the trajectory of this decades-long disease can the increase in plasma p-tau217 be picked up? Clues to this question came from Hansson’s analysis of the Colombia cohort samples. Mean concentrations were 1.9 pg/mL in 257 members of the kindred who did not carry the Paisa mutation, 4.5 pg/mL in 259 cognitively unimpaired mutation carriers, and 16.8 pg/mL in 106 cognitively impaired carriers. In carriers, plasma p-tau217 rose over time, such that by 25 years of age, they had significantly higher levels than noncarriers. That’s about 20 years before the age of symptom onset (see image below). The data suggest that, like p-tau181, p-tau217 is detectable in the blood years before a clinical diagnosis and even before a positive tau PET test. Hansson told Alzforum that, in a follow-up manuscript, Janelidze describes plasma p-tau217 increasing before tau can be detected by PET, even in the entorhinal cortex, an early region for tangles. “Just as for CSF p-tau217, the plasma level increases first, showing it is a much earlier marker than PET.”
Early Marker. In people who carry the E280A presenilin 1 mutation, p-tau217 rises in the plasma about 25 years before symptom onset (left) and better distinguishes carriers from controls than does serum NfL (right).
Will p-tau217 become the first approved blood test for AD? Researchers agreed that more work remains to be done. Hansson said it is important to see if the test works in a primary-care setting and has begun to test this. A few months ago, he set up a prospective study, much like the IDEAS study for amyloid PET, to see if p-tau217 assays can improve routine AD diagnosis or change how physicians treat patients (Apr 2019 news). He had recruited about 60 volunteers in southern Sweden before COVID-19 slowed the study down. “This study represents an important step and we hope to pick this back up now,” said Hansson.
At the same time, he thinks the test soon can be implemented in clinical trials and even in retrospective testing of completed or ongoing studies. Banked samples could be tested quickly, including those from large epidemiological studies and clinical trials. Boxer said the plasma assay will be easier and more efficient to use than current PET- and CSF-based diagnostics. “It could lead to more creative and pragmatic trial approaches. The hope is it would help recruit more minority and underserved populations who maybe don’t have the resources to take part in studies that require more invasive tests,” he said.
Along the same line, Suzanne Schindler from WashU presented preliminary data from the Study to Evaluate Amyloid in Blood and Imaging Related to Dementia at AAIC. SEABIRD was designed to reflect the demographics of the St. Louis metropolitan area, improving racial, ethnic, educational, and socioeconomic representation in this research. Schindler reported that in blood samples from this population, plasma p-tau217 more accurately reflected a person’s amyloid status than did plasma p-tau181, suggesting that the 217 isoform performs well in this community setting.
Even so, it will take some time to get a plasma tau test ready for regulatory approval as a diagnostic. The assay needs to be validated in additional cohorts and between labs. Cutoffs need to be set, protocol standardized, certified reference materials to calibrate commercial assays need to be made. This kind of applied preapproval research takes at least a year. Additionally, the current immunoassay has a lower limit of detection of 0.48 pg/mL, which is not sensitive enough to measure the extremely low p-tau217 levels found in some control samples. “Ideally, we’d want to have some confidence that a test was negative because the amount of p-tau217 was really low and not because we couldn’t detect it,” said Barthélemy. With a lower limit of detection of 0.05 pg/mL, mass spec detected p-tau217 in all samples tested, but that required at least 4 mL of blood, which is orders of magnitudes more than routinely collected in certain studies, such as clinical trials.
It is possible that the sensitivity of the immunoassay might be improved. There is now a separate, new, single-molecule array assay for plasma p-tau181 developed by Kaj Blennow, Nicholas Ashton, and colleagues at the University of Gothenburg, Sweden, that seems more sensitive than the Dage assay for this form of tau (Apr 2020 news). In an upcoming paper, the Gothenburg group reports that plasma p-tau181, when measured up to eight years before a person’s death, predicts postmortem plaque pathology better than does a clinical diagnosis of AD, and that it distinguishes AD from other neurodegenerative disease with an AUC of 0.97. “We need some additional studies before we can say how much of a difference there is between these various p-tau biomarkers,” Blennow wrote to Alzforum. “Given that plasma p-tau181 performs quite well to identify AD and differentiate it from other diseases, I think differences will be relatively minor.”
Blennow further emphasized that the field needs to learn more about what phosphorylation of tau at different sites means in terms of AD pathophysiology. For example, are tau fragments of different sizes phosphorylated on different amino acids? If so, why? Do some have more than one phosphate and others none, and what would that mean?
Bateman thinks it’s important to better understand what the timing of phosphorylation implies about plaques and tangles. In the DIAN cohort, p-tau217 increases two decades before tau PET can detect neurofibrillary tangles. “Does this indicate that p-tau217 is a reaction to amyloid plaques that precedes tau pathology? Or is tau pathology beginning two decades before it is detectable by tau PET?” he asked (see comment below). Bateman also noted that p-tau217 decreases after symptom onset, while the tau PET signal and clinical dementia continue to increase. “This evidence strongly suggests that p-tau217 (and p-tau181) are not direct measures of tau pathology, but rather a reaction to amyloid plaques that later is associated to tau pathology,” he wrote. —Tom Fagan
No Available Further Reading
- Palmqvist S, Janelidze S, Quiroz YT, Zetterberg H, Lopera F, Stomrud E, Su Y, Chen Y, Serrano GE, Leuzy A, Mattsson-Carlgren N, Strandberg O, Smith R, Villegas A, Sepulveda-Falla D, Chai X, Proctor NK, Beach TG, Blennow K, Dage JL, Reiman EM, Hansson O. Discriminative Accuracy of Plasma Phospho-tau217 for Alzheimer Disease vs Other Neurodegenerative Disorders. JAMA. 2020 Aug 25;324(8):772-781. PubMed.
- Barthélemy NR, Horie K, Sato C, Bateman RJ. Blood plasma phosphorylated-tau isoforms track CNS change in Alzheimer's disease. J Exp Med. 2020 Nov 2;217(11) PubMed.