# Temporal Validation of a Plasma Diagnosis Approach for Early Alzheimer Disease Diagnosis in a Cognitive Disorder Unit

## TL;DR

This study validates a blood-based test for early Alzheimer's diagnosis in a clinic, showing high accuracy with minimal invasiveness.

## Contribution

The study temporally validates a plasma biomarker-based diagnostic model for early Alzheimer's disease in a clinical setting.

## Key findings

- The one-cut-off approach achieved 74.3% sensitivity and 89.5% specificity for Alzheimer's diagnosis.
- The two-cut-off approach showed higher specificity (99.9%) but slightly lower sensitivity (66.7%).
- Both approaches demonstrated satisfactory performance, with the two-cut-off method better identifying non-Alzheimer's cases.

## Abstract

Background: Nowadays, there is a lack of reliable and minimally invasive diagnosis methods for the early detection of Alzheimer’s disease. The development and validation of such tools could significantly reduce the dependence on more invasive and costly confirmatory procedures, such as cerebrospinal fluid biomarkers analysis and neuroimaging techniques. Objectives: The main objective of this study is to validate the clinical performance of a previously developed diagnosis model based on plasma biomarkers from patients in a cognitive disorder unit. Methods: A new cohort of patients was recruited from the same cognitive disorder unit (n = 93). Specifically, demographic data (gender, age, and educational level), plasma biomarkers levels, and genotype (glial fibrillary acidic protein, phosphorylated Tau 181, amyloid-beta42/amyloid-beta40, apolipoprotein E) were collected to evaluate both approaches of the previous diagnosis model (one-cut-off, two-cut-off). Results: The one-cut-off approach showed a sensitivity of 74.3%, a specificity of 89.5%, and an area under the curve of 0.888, while the values for the two-cut-off approach were sensitivity of 66.7%, specificity of 99.9%, and area under the curve of 0.867. Conclusions: A multivariate diagnostic tool was temporally validated for implementation in a clinical unit. In fact, satisfactory results were obtained from both approaches (one-cut-off, two-cut-offs), but the two cut-offs approach was more consistent in correctly identifying non-Alzheimer’s disease cases, allowing us to identify a large number of cases with high specificity.