[Skip to Navigation]
May 2011

Blood-Based Biomarkers in Alzheimer Disease

Author Affiliations

Author Affiliation: Department of Psychiatry, University of Tuebingen, Tuebingen, Germany.

Arch Neurol. 2011;68(5):685-686. doi:10.1001/archneurol.2011.78

O’Bryant and colleagues describe in their recent article1 the identification of a blood-based biomarker panel for classification of patients with Alzheimer disease (AD). If confirmed, this opens an alternative approach to finding useful biomarkers for the early detection of AD using blood as an alternative sampling tissue. A simple blood test could revolutionize AD diagnostics, as it would be widely available, noninvasive, easy to obtain, economic, and rapid to perform. However, there are some major points that have to be considered when interpreting the results by O’Bryant et al. (1) Patients with AD included in the present study were clinically diagnosed according to National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA2) criteria for probable AD, and there were no neuropathologically confirmed cases. However, it has been shown in several studies that the agreement between clinical diagnosis and true pathology can vary significantly and may mask a good agreement between a biomarker result and true pathology.3 The implications of using an imperfect reference standard when assessing biomarker performance have to be taken into consideration. (2) The authors proposed a panel of more than 20 serum markers. Including such a high number of parameters can risk overfitting the data set. Overfitting implies poor generalization to correctly classify new data.4 (3) Patients with other subtypes of dementia should be included in the examination in future studies to test the specificity of the described findings in patients with AD. (4) The authors describe the value of adding clinical variables (age, sex, education, APOE status) to their biomarker data. In this context, adding other parameters such as imaging and neuropsychological data to the classifier should be expected to significantly increase diagnostic accuracy. In conclusion, the test performance should be confirmed in further independent validation studies under consideration of the above mentioned points.

Add or change institution