The GRAD (Gatekeeper-Reflex for Alzheimer's Disease) two-stage plasma algorithm was developed in 145 cognitively impaired ADNI participants and externally validated in 1,644 cognitively unimpaired A4/LEARN participants to reduce amyloid PET use by improving Aβ-positivity classification in the p-Tau217 'gray zone' (indeterminate probability 0.25–0.75), using additional plasma biomarkers (GFAP, Aβ42/40), APOE ε4 status, and age in Stage 2.
In the external gray zone (n=630), Stage 2 improved AUC over p-Tau217 alone (0.793 vs. 0.719, p<0.001), raised correct classification without PET from 41.4% to 56.2%, and cut PET referral from 35.9% to 24.3%; overall, 90.7% of A4/LEARN participants were classified without PET, and modeled costs were 44–52% lower than p-Tau217 screening with PET for all gray-zone cases (~$863 vs. $1,639 per correct diagnosis).
Both cohorts are predominantly White (89–94%) and highly educated, limiting generalizability to diverse populations; the development gray zone is small (n=49) and the Stage 2 AUC gain was non-significant internally (p=0.54); calibration did not transfer to the external cohort (intercept shift +0.93), meaning predicted probabilities may require recalibration before clinical use; renal function and comorbid proteinopathies—known confounders of plasma biomarker levels—were not accounted for.
For patients with indeterminate p-Tau217 results, a reflex multimarker panel (adding Aβ42/40 and APOE ε4 status) can reclassify over half without needing amyloid PET, potentially cutting imaging costs by 44–52%; however, GRAD requires prospective validation in diverse clinical populations and should currently be used as a triage and trial-enrichment aid rather than a replacement for confirmatory Aβ-PET.
Explore related topics