For decades, Alzheimer's diagnosis relied on a blunt yes-or-no read of amyloid burden — positive or negative on a PET scan. A perspective published in JAMA argues this binary framework may be leaving critical clinical information on the table, and that standardized quantitative scoring could meaningfully change how patients with mild cognitive impairment (MCI) are assessed and treated.
The Centiloid scale — a standardized 0-to-100 unit system anchored between amyloid-negative young adults and confirmed Alzheimer's patients — allows clinicians to express amyloid burden as a continuous value rather than a categorical cutoff. The JAMA viewpoint contends that Centiloid values carry diagnostic and prognostic weight beyond binary classification: patients hovering near the positivity threshold may have meaningfully different trajectories than those with high amyloid loads, and treatment eligibility decisions for anti-amyloid therapies could be refined accordingly. The authors propose that Centiloid quantification should be integrated into clinical and regulatory frameworks governing diagnostics and therapeutics.
This argument lands at a pivotal moment. The FDA approval of lecanemab and donanemab — both anti-amyloid monoclonal antibodies — has introduced real-world urgency to amyloid measurement precision, since treatment response and safety profiles (particularly ARIA risk) appear to correlate with amyloid burden magnitude, not just its presence. The Centiloid framework has been validated across multiple PET tracers and scanner platforms, which strengthens its case for standardized adoption. However, the primary limitation here is that this is a viewpoint, not a clinical trial or meta-analysis — it is an expert argument rather than empirical evidence of improved patient outcomes. Whether Centiloid-guided thresholds translate into measurably better decisions in real-world clinical settings remains to be demonstrated prospectively. Still, as precision medicine increasingly demands granular biomarker data, the case for retiring simple binary amyloid reads appears substantively compelling.