Why do some people with amyloid plaques decline, while others do not? In the May 29 Nature Medicine, researchers led by Tharick Pascoal at the University of Pittsburgh laid some of the blame on reactive astrocytes. Among cognitively healthy people with amyloid plaques, only those whose astrocytes were activated, as judged by plasma GFAP, accumulated tau pathology. The relationship held in three separate cohorts, and when using different ways to measure amyloid and tau. The findings hint that astrocytes drive tangle formation, the authors said. “GFAP positivity could help us identify people with brain amyloid who are fated to develop tau pathology,” Pascoal told Alzforum.
Commenters deemed the evidence convincing for a link between GFAP and p-tau181. “While correlation does not mean causation, the finding … suggests that reactive astrocytes play a central role at the earliest stages of the AD pathophysiological process,” Alberto Serrano-Pozo at Massachusetts General Hospital, Charlestown, wrote to Alzforum. Nonetheless, they said questions remain, such as the direction of that association, the effect of co-morbidities, and the specificity of the marker itself. “More research into this important area is warranted,” said Jeffrey Dage at Indiana University, Indianapolis (comments below).
Scientists have long known that reactive astrocytes appear around amyloid plaques early in AD (Pike et al., 1995; Kumar et al., 2021). More recently, studies have correlated high plasma GFAP with plaques in the brain (Mar 2021 conference news; Chatterjee et al., 2021; Pereira et al., 2021). Curiously, one study reported that GFAP in plasma predicts the presence of brain amyloid more accurately than does GFAP in cerebrospinal fluid (Benedet et al., 2021).
To find out how GFAP relates to other AD biomarkers, first author Bruna Bellaver compared cognitively healthy people in three observational cohorts: the TRIAD study at McGill University, a research cohort at Pittsburgh, and the population-based MYHAT cohort in Pittsburgh. Together, the sample comprised 1,016 people, average of 70 years old. Bellaver and colleagues classified each participant as positive (Ast+) or negative (Ast-) for reactive astrocytes, setting the cutoff for plasma GFAP at two standard deviations above the average in amyloid-negative participants. Among the 294 amyloid-positive participants, 108 were Ast+ by this measure, 186 were Ast-.
In the former, plasma p-tau181 was high, while in the latter, levels were as low as in amyloid-negative controls. Importantly, only in Ast+ people did plaque burden correlate with p-tau181 (see graph above). The effect size was large, with a Cohen’s d of 0.80. Likewise, in a subgroup of 147 participants who had a tau PET scan with the MK6240 tracer, tangle load in the earliest Braak regions correlated with amyloid load only in the Ast+ group, again suggesting that astrocytes mediate this association.
In the Ast+ group, the association between plaque and p-tau181 was stronger in men than women. Previous studies have reported higher overall tangle burden in women than men (Aug 2018 conference news; Feb 2019 news; Nov 2019 news). The new data suggest that reactive astrocytes may play a bigger role in men, while other mechanisms are at work in women, Pascoal noted.
How might astrocytes influence tangle formation? Previous cell culture and animal studies have found that activated astrocytes release factors, such as cytokines and adenosine triphosphatase, that trigger tau phosphorylation in neurons (Garwood et al., 2011; Litvinchuk et al., 2018; Mann et al., 2022). Pascoal plans to probe mechanisms in the human brain by analyzing single-cell RNA-Seq in postmortem tissue to identify genes highly expressed in reactive astrocytes.
Serrano-Pozo noted that in the AD brain, reactive astrocytes wrap their processes around plaques, tangles, and dystrophic neurites containing phosphorylated tau. “Reactive astrocytes are uniquely positioned to link Aβ and p-tau pathologies,” he said. Others noted caveats. Rik Ossenkoppele and Emma Coomans at Amsterdam University Medical Center pointed out that plasma markers often track up or down in concert within individuals, and that more work will be needed to show a causal effect (comment below).
What happens over time? The study had longitudinal data on 71 participants. In them, baseline plaque burden predicted tangle spread over two years only in those who were Ast+ (image at right). This implies that screening for plasma GFAP could enrich clinical trials for people whose disease is most likely to progress, Pascoal suggested. If the findings hold up, he believes reactive astrocytes could eventually be incorporated into the definition of preclinical AD.
Beau Ances at Washington University in St. Louis agreed, noting that the ATN staging scheme could be updated to include an inflammatory marker. On the other hand, Oskar Hansson and Gemma Salvadó at Lund University, Sweden, said that in their hands, plasma GFAP does not improve upon high-performing plasma p-tau assays that detect plaques and tangles. “That said, we think plasma GFAP is a very interesting outcome marker in interventional studies evaluating the effects of new disease-modifying or lifestyle interventions,” they wrote. Others cautioned that GFAP lacks specificity. “This is why it is important that we shift toward the use of bona fide reactive astrocyte markers that are never expressed by astrocytes in a physiological setting, such as Lcn2, Serpina3n, and many others identified by recent sequencing efforts,” wrote Shane Liddelow, New York University (comments below).
Pascoal and colleagues are now broadening their study by including racially diverse cohorts with co-morbidities such as cardiovascular disease. They will also examine later stages of disease to determine how long into the disease this early stage relationship between plaques, tangles, and reactive astrocytes occurs.—Madolyn Bowman Rogers