# Use of a multimarker blood test in the heterogenous memory clinic setting: an updated and simplified interpretation tool

## TL;DR

This paper introduces a simplified tool to interpret blood test results for dementia diagnosis, using multiple biomarkers to improve accuracy in memory clinics.

## Contribution

The paper presents an updated and simplified multi-marker interpretation tool for blood-based biomarkers in dementia diagnosis.

## Key findings

- The combination of p-tau217 and age-corrected NfL provided AUCs ranging from 0.85 to 0.93 across clinical questions.
- The visualization tool helps clinicians interpret plasma biomarker results and determine etiological diagnoses.
- Grey zones for result interpretation were generally small, increasing confidence in using the tool in routine practice.

## Abstract

Patients referred to specialized memory clinics present with different symptoms and etiologies. A blood test could streamline establishing etiological diagnoses. Given the diversity in patients seen at memory clinics, relying on a single marker, e.g. phosphorylated tau217 (pTau217), is insufficient. However, multi‐marker interpretation in the clinical context is challenging. We aim to provide a comprehensive multi‐marker interpretation tool.

We follow up on our published work (Verberk, Alz&Dem, 2024) in which we presented a result visualization approach for p‐tau181, glial fibrillary acidic protein (GFAP) and neurofilament light (NfL). We now extended our cohort to n = 1773 Amsterdam Dementia Cohort participants with diagnosis established by clinical consensus according to applicable criteria (table 1) and measured p‐tau217 (Simoa Janssen) in addition to amyloid‐beta (Abeta)42/40, GFAP and NfL (age‐corrected). Starting point of the data analyses were clinically relevant questions: 1) identify Amyloid+ across all syndromes SCD, MCI and dementia, 2) discriminate FTD from AD, 3) discriminate FTD from any SCD (Amyloid+/‐), 4) predict if symptoms are due to a neurodegenerative disease (any dementia vs any SCD). We performed logistic regression analyses with 100‐fold Monte‐Carlo cross‐validation across 70%‐30% train‐test splits for all plasma marker combinations, and selected the simplest and most effective combination, based on number of markers, AUC, marker importance and %individuals falling in the grey zone (bounded by 90%‐sensitivity and 90%‐specificity thresholds). We subsequently developed a visualization tool.

The simplest and most effective combination of plasma biomarkers across the clinical questions was p‐tau217 with age‐corrected NfL, with AUCs ranging from 0.85 (FTD/SCD) to 0.93 (Amyloid+/Amyloid‐). %individuals in the grey zones ranged from 8.9% (Amyloid+/Amyloid‐) to 37.5% (FTD/SCD). Figure 1 shows our visualization tool: plasma values of our tested population, the size of the grey zones, and the influence of both plasma markers on the outcome prediction (angle of the grey zone) can be appreciated, per question. Plasma results of new patients are plotted on these graphs, to inform on their etiological diagnosis.

Plasma marker results of new patients are easily interpretable with our visualization tool. Grey zones are generally small, benefitting the confidence of plasma biomarker use in routine dementia practise.