Plasma GFAP May Enrich Patient Selection, Reduce PET Scans Required for Alzheimer Trials

Bruna Bellaver, PhD

In a recently published analysis, findings suggest that Alzheimer disease (AD) clinical trials enriched with cognitively unimpaired (CU) patients who are positive for glial fibrillary acidic protein (GFAP) and amyloid-ß (Aß) status require a reduced sample size compared with Aß+ only. In addition, investigators concluded that CU GFAP+/Aß+ enrichment reduces the number of Aß PET scans required, as well as allows for better selection of individuals at the earliest stages of AD continuum.¹

Published in Alzheimer’s & Dementia, study authors estimated longitudinal progression, effect size, and costs of hypothetical trials designed to test an estimated 25% drug effect on reducing tau PET accumulation in the medial temporal lobe (MTL) and temporal neocortical region (NEO-T). Overall, 218 CU participants (mean age, 69.0 years [±6.4]) with plasma GFAP, Aß PET, as well as longitudinal tau PET (follow-up duration, 2.48 years [±0.84]) from 2 centers were included. Of these, 28 were GFAP+/Aß+, 41 were GFAP–/Aß+, 38 GFAP+/Aß–, and 105 GFAP–/Aß–.

Led by Bruna Bellaver, PhD, a research assistant professor of psychiatry at the University of Pittsburgh, both the GFAP–/Aβ+ and GFAP+/Aβ+ groups showed a significant longitudinal increase in tau PET_MTL uptake, with the GFAP+/Aβ+ group exhibiting a higher annual rate of change (0.75) compared to the Aβ+ only group (0.66), and both groups showing greater increases than the GFAP–/Aβ– group (0.12). In tau PET_NEO-T, only the GFAP+/Aβ+ group showed a significant annual rate of change, which was higher than both the GFAP–/Aβ+ group and the GFAP−/Aβ− group, with effect sizes of 0.56, 0.35, and −0.08, respectively.

In the analysis, investigators found that trials aiming to detect changes in tau PET_MTL as a secondary outcome using this enriched population would require 433 individuals per study arm, whereas using only CU Aß+ only patients would require 507 individuals per arm. Overall, this translated to a 15% reduction in the number of patients needed. Even larger differences were observed with PET_NEO-T, as trials using CU GFAP+/Aß+ individuals would require 891 individuals per study arms vs 2068 individuals for CU Aß+ only, resulting in a 57% reduction.

Enriching trials with CU GFAP+/Aß+ individuals also led to reduced resource use. When examining trials targeting tau PET_MTL, enriching the population lowered number of Aß PET scans necessary by 49%, from 1619 to 866. Similarly, trials targeting tau PET_NEO-T, the GFAP+ prescreening step reduced the total number of individuals required to be recruited (5937 vs 6892) and the number of Aß PET scans required by up to 74% (1781 vs 6892 if using only Aß only). Notably, the estimated cost reduction for clinical trials using the GFAP+/Aβ+ strategy compared to Aβ PET only was 28% for tau PET_MTL trials and 64% for tau PET_NEO-T trials.

Study authors compared two strategies for selecting individuals at the earliest stages of AD: Aβ+ plus GFAP+ versus Aβ+ plus p-tau+. GFAP+/Aβ+ individuals had lower Aβ pathology than p-tau+/Aβ+ individuals, though plasma p-tau217 levels were not significantly different overall, except when an outlier was removed, showing higher p-tau217 in the p-tau+/Aβ+ group. There were no significant differences in the annual rate of change in tau PET_MTL or tau PET_NEO-T between the two groups, though the effect size in tau PET_MTL was higher for p-tau+/Aβ+ (0.93 vs. 0.77). After accounting for Aβ levels, GFAP+/Aβ+ showed a higher effect size (0.49 vs. 0.40). In tau PET_NEO-T, GFAP+/Aβ+ also had a numerically higher effect size (0.57 vs. 0.52), with GFAP+/Aβ+ maintaining a greater effect size even when continuous Aβ levels were considered.

READ MORE: Frontotemporal Dementia Insights From Baptist Health’s Brain and Spine Symposium

When focusing on individuals with Centiloid levels between 12 and 50, GFAP+/Aβ+ individuals showed higher effect sizes in both tau PET_MTL (0.95 vs. 0.84) and tau PET_NEO-T (0.48 vs. 0.29) compared to p-tau+/Aβ+ individuals. These findings suggest that GFAP+/Aβ+ individuals may better represent early-stage AD based on tau PET changes, with less influence from p-tau217 levels than p-tau+/Aβ+ individuals.

"This finding may have important implications for clinical trial participant selection," the study authors noted. "Our results suggest that selecting GFAP+/Aβ+ individuals may help identify individuals who are more likely to progress in the disease but have a lower Aβ burden (mean Centiloid of 46) compared to the selection of p-tau+/Aβ+ individuals (mean Centiloid of 65). This approach is particularly relevant for trial enrichment if we consider the TRAILBLAZER-ALZ study, which demonstrated that individuals with lower baseline Aβ levels were more likely to achieve complete Aβ clearance."

Overall, study authors concluded that the data suggest using GFAP+ rather than p-tau+ in combination with Aß PET could help select individuals most likely to benefit from anti-Aß therapies. The analysis did have some limitations, including the fact that the observed population was highly educated, mostly White participants, which does not represent a more diverse general world population. In addition, the effect sizes and samples sizes determined may not be generalizable to other tau PET tracers that present distinct intrinsic characteristics and off-target binding, which could result in varying rates of changes between tracers.

REFERENCES
1. Bellaver B, Povala G, Ferreira PCL, et al. Plasma GFAP for populational enrichment of clinical trials in preclinical Alzheimer's disease. Alzheim & Dement. 2025;21(5):e70209. doi:10.1002/alz.70209