Finding Order in Alzheimer's Biomarkers

Abstract & Commentary

By Norman Relkin, MD, PhD, Director, Memory Disorders Program, Associate Professor of Clinical Neurology, Weill Cornell Medical Center. Dr. Relkin receives research/grant support from the NIH and Baxter and is a consultant to Eisai and Intellect Neuroscience.

Synopsis: Measurement of biomarkers, including Ab42 and tau protein in the CSF along with volumetric measurements of the hippocampus, may help in the diagnosis of early or presymptomatic Alzheimer's disease.

Source: Jack CR, et al, for the Alzheimer's Disease Neuroimaging Initiative. Evidence for ordering of CSF biomarkers. Arch Neurol 2011 Aug 8 doi:10.1001/archneurol.2011.183

Three validated biomarkers for alzheimer's disease (AD) become abnormal at different times during the early development of the disease, according to a new study involving more than 400 participants in the Alzheimer's Disease Neuroimaging Initiative (ADNI). Understanding the temporal sequence in which biomarker abnormalities arise may be critical to the success of early detection of AD using new diagnostic criteria that incorporate CSF biomarker and imaging criteria.

Clifford Jack and colleagues previously hypothesized that levels of the 42 peptide form of beta amyloid (Ab42 ) in cerebrospinal fluid (CSF) of persons destined to develop AD would decrease before CSF tau protein becomes elevated and hippocampal volume decreases. To test this hypothesis, the investigators determined thresholds for abnormalities in these biomarkers in independent cohorts of autopsy-proven AD cases and controls. The resulting cutoff scores were applied to CSF and imaging outcomes in a validation cohort of 401 ADNI research volunteers, 298 of whom had sequential biomarker measurements made at least 12 months apart. The subjects were divided into three groups defined as cognitively normal (CN), mild cognitive impairment (MCI), and AD. A clinical diagnosis, Mini-mental State Score, and the outcomes of all three biomarker determinations were available for analysis in all cases.

Across the three diagnostic groups, abnormal CSF Ab42 levels were detectable before abnormalities in CSF tau and head-size adjusted hippocampal volume. In the subset of subjects with measurements at two time points, the percentage with abnormal Ab42 levels did not change over time. Abnormal tau levels increased from baseline to 12 months in cognitively normal individuals but not in subjects with MCI or AD. Extent of hippocampal atrophy changed more in the MCI and symptomatic dementia stages than did the other two biomarkers. The authors concluded that among these three biomarkers, CSF Ab42 levels become abnormal earliest in AD, followed by abnormalities in CSF tau and hippocampal volume loss respectively.


The concept that biomarker abnormalities evolve temporally during the early stages of AD is not new but until now empiric validation has been lacking. The authors of this study did a commendable job in converting a complicated biomarker dataset into workable test of this hypothesis. The study provides reasonable evidence that emergence of biomarker abnormalities follows a temporal sequence in AD. The investigators were forced to confine their analysis to just three biomarkers owing to lack of availability of a more complete set of validated markers in the entire ADNI cohort. This is unfortunate since FDG PET imaging shows regional cerebral hypometabolism quite early in the course of AD and becomes increasingly abnormal as the disease progresses. It is anticipated that amyloid PET imaging will become available for clinical use in the near future and it, too, can show abnormalities in presymptomatic stages of AD.

What are the clinical implications of these findings? The considerable overlap in biomarker profiles across the three diagnostic groups (normal, MCI, AD) suggests that the use of these markers in individual patients for staging purposes may be complicated. Since the temporal sequence of appearance of biomarker abnormalities is not considered in current diagnostic recommendations, the possibility of drawing incorrect conclusions from biomarker findings, particularly in presymptomatic stages of the disease, must be acknowledged. Future guidelines that incorporate biomarkers into the diagnostic criteria for presymptomatic AD will likely need to take into account the temporal sequence of appearance of biomarker abnormalities. It may take several years before adequate longitudinal data are collected to rigorously stage prodromal AD in this manner. However, biomarkers that have already been validated for diagnosis of symptomatic AD, such as the combination of Ab42, tau and phospho-tau in CSF, remain valid and can add to differential diagnostic certainty in those cases in which symptoms of dementia already have become evident.