The CentiMarker Project: Standardizing Quantitative Alzheimer's disease Fluid Biomarkers for Biologic Interpretation

Abstract

Introduction: Biomarkers have been essential to understanding Alzheimer's disease (AD) pathogenesis, pathophysiology, progression, and treatment effects. However, each biomarker measure is a representation of the biological target, the assay used to measure it, and the variance of the assay. Thus, biomarker measures are difficult to compare without standardization, and the units and magnitude of effect relative to the disease are difficult to appreciate, even for experts. To facilitate quantitative comparisons of AD biomarkers in the context of biologic and treatment effects, we propose a biomarker standardization approach between normal ranges and maximum abnormal AD ranges, which we refer to as CentiMarker, similar to the Centiloid approach used in PET.

Methods: We developed a standardization scale that creates percentile values ranging from 0 for a normal population to 100 for the most abnormal measures across disease stages. We applied this scale to CSF and plasma biomarkers in autosomal dominant AD, assessing the distribution by estimated years from symptom onset, between biomarkers, and across cohorts. We then validated this approach in a large national sporadic AD cohort.

Results: We found the CentiMarker scale provided an easily interpretable metric of disease abnormality. The biologic changes, range, and distribution of several AD fluid biomarkers including amyloid-β, phospho-tau and other biomarkers, were comparable across disease stages in both early onset autosomal dominant and sporadic late onset AD.

Discussion: The CentiMarker scale offers a robust and versatile framework for the standardized biological comparison of AD biomarkers. Its broader adoption could facilitate biomarker reporting, allowing for more informed cross-study comparisons and contributing to accelerated therapeutic development.

Figure 1:

Illustration of CentiMarkers Approach. The rectangles indicate the CentiMarker 0 and CentiMarker 100 for each biomarker

Figure 2:

Comparison of different CentiMarkers in the same study by stage of disease comparing normal controls (green circles) to mutation carriers (color triangles). The mutation carriers included only the baseline data from the treatment groups and all the data from the placebo group in the DIAN-TU-001 trial. The x-axis represents the Estimated Years to Symptom Onset (EYO) based on mutation information, with zero indicating the expected onset of symptoms, covering a range of 25 years to represent disease progression.

Figure 3:

Utilizing CentiMarkers facilitates the interpretation and comparison across biomarkers by converting them to a similar scale, while using raw values is more difficult to compare as units are different and not scaled to disease ranges. Estimated mean change from baseline in CentiMarkers with 95% confidence intervals for the gantenerumab and placebo groups using MMRM analyses in the DIAN-TU-001 trial. These results demonstrate the magnitude of disease normalization compared to normal (CM 0) vs. fully abnormal (CM 100) states. Raw values are in the unit of ng/mL

Figure 4:

Illustration of treatment effects over EYO relative to mutation non-carriers (green line). Participants treated with gantenerumab (blue line) showed slower progression than those non-gantenerumab treated (purple line) across multiple biomarkers. The relative differences of treatment effect indicate that although there were improvements in multiple biomarkers, the measures did not reach normal CentiMarker levels for some measures.

Figure 5:

Illustration of CentiMarkers in ADNI. EYO is calculated by using the individual’s own age of symptom onset as EYO of 0. These distributions replicate increasing CSF tau and p-tau181 by stage of disease.