J Alzheimers Dis. 2015 Nov 27;50(1):111-26.

Characterizing Aging, Mild Cognitive Impairment, and Dementia with Blood-Based Biomarkers and Neuropsychology.
 

Kleinschmidt M1,2, Schoenfeld R3, Göttlich C1,4, Bittner D5, Metzner JE6, Leplow B3, Demuth HU1,2.
  • 1Probiodrug AG, Halle (Saale), Germany.
  • 2Current address: Fraunhofer Institute of Cell Therapy and Immunology, Department of Drug Design and Target Validation, Halle (Saale), Germany.
  • 3Martin-Luther-University Halle-Wittenberg, Department of Psychology, Halle (Saale), Germany.
  • 4Department Tissue Engineering & Regenerative Medicine (TERM), University Hospital Wuerzburg, Germany.
  • 5Clinic for Neurology, Otto-von-Guericke University Magdeburg, Germany.
  • 6Galmed GmbH, Halle (Saale), Germany.

 

Abstract

BACKGROUND: Current treatment in Alzheimer’s disease (AD) is initiated at a stage where the brain already has irreversible structural deteriorations. Therefore, the concept of treatment prior to obvious cognitive deficits has become widely accepted, and simple biochemical tests to discriminate normal aging from prodromal or demented stages are now common practice.

OBJECTIVE: The objective of the study was the differentiation of controls, mild cognitive impairment (MCI) and AD patients by novel blood-based assays in combination with neuropsychological tests.

METHODS: In a cross-sectional study, 143 subjects aged 18 to 85 years were recruited. All participants were classified by a comprehensive neuropsychological assessment. Blood samples were analyzed for several amyloid-β (Aβ) species, pro-inflammatory markers, anti-Aβ autoantibodies, and ApoE allele status, respectively.

RESULTS: Plasma Aβ1-42 was significantly decreased in MCI and AD compared to age-matched controls, whereas Aβ1-40 did not differ, but increases with age in healthy controls. The Aβ1-42 to Aβ1-40 ratio was stepwise decreased from age-matched controls via MCI to AD, and shows a clear correlation with memory scores. Reduced Aβ1-42 and Aβ1-42 to Aβ1-40 ratio have strongly correlated with carrying ApoE ɛ4 allele. Autoantibodies against pyroglutamate-modified Aβ, but only a certain subclass, were significantly decreased in AD compared to MCI and age-matched controls, whereas autoantibodies against the unmodified N-terminus of Aβ did not differ.

CONCLUSION: Comprehensive sample preparation and assay standardization enable reliable usage of plasma Aβ for diagnosis of MCI and AD. Anti-pGlu-Aβ autoantibodies correlate with cognition, but not with ApoE, supporting the associated plasma Aβ analysis with additional and independent information.

KEYWORDS: Aging; Alzheimer’s disease; ApoE; amyloid-β; autoantibodies; blood; mild cognitive impairment; neuropsychology

PMID: 26639953

 

Supplementary

Current treatment of Alzheimer ’s disease (AD) is initiated at stages where the brain has irrevocably lost numerous neurons which cannot be rescued, and neuroinflammation progresses. Since the concept of neurodegeneration treatment prior to obvious cognitive deficits becomes widely accepted, simple biochemical tests to discriminate normal aging from prodromal or demented stages are needed. Current standards identifying preclinical AD are neuroimaging and cerebrospinal fluid (CSF) protein analysis; however, these methods are more cost-effective and more invasive than blood-based biomarker assays. But in contrast to the approved methods quantifying the AD biomarker amyloid β (Aβ), Tau and P-Tau in CSF, detection of these biomarkers in blood is much more difficult due to at least 10-fold lower concentrations, but the 100-fold higher overall protein content leading to massive interference in currently used assay systems. In our study, the different Aβ species were isolated in an initial separation step using a multivalent capture system which exploits avidity effects by interactions of several anti-Aβ antibodies to multiple epitopes of the Aβ molecule. Therefore, all interfering proteins can be removed by slightly destabilizing additives breaking the binding to the captured Aβ peptides, which were subsequently eluted from the capture system and quantified by specific ELISA. However, this procedure is very laborious and has to be standardized minimizing inter-assay variations, which was checked in 25 independent experiments using aliquots of the same plasma sample. We have obtained CVs of ~10% according to an error of mean of ~2%. Furthermore, we have balanced inter-assay variations by normalizing all obtained plasma concentrations to the before mentioned plasma sample, which ran through the complete procedure together with samples from the study in every single measurement series. Beside the standardization of the biomarker analysis also the sample generation was performed under strict regulations at each study site. Biomarker alterations due to circadian rhythms and food intake were minimized by sample taking in the morning after an overnight fasting. Differences in sampling were almost completely excluded performing a standardized protocol with appropriate equipment and usage of exactly the same devices for blood withdrawal, processing and storage.

 

 

Figure 1Figure 1. Correlation between symptoms of depression (BDI, Beck Depression Inventory) and an inflammation (IL-6, blood serum level of interleukin 6). The regression suggests a moderate (13% explained variance) but not negligible effect. Applying biomarkers in Alzheimer’s disease needs to consider even mild symptoms (dashed vertical line) to separate effects of the disease from a depression, which is very likely in older people.

 

Decisive for comparison of biomarker levels between the subjects is good and standardized classification of them. Dementia diagnosis is primarily based on the assessment of cognitive state, emotional-behavioral stability, and activity of daily living [1]. The characterization of subjects becomes a critical issue, especially in cross-sectional studies. Psychological screening tests, such as the Mini-Mental Status Examination (MMSE), have been frequently used for this purpose, but lack reliable psychometric properties and lead to many false negative and false positive results, respectively [2]. Contemporary dementia classifications also consider the concept of mild cognitive impairment (MCI) as a prodromal state of Alzheimer’s dementia [3]. But MCI could also be a neuropsychological impairment of an underlying psychiatric disorder (e.g., depression, anxiety, and compulsive disorders), which can falsify the classification; therefore all subjects known psychiatric disorders were excluded from the study. Furthermore, all included subjects were additionally tested for depression and individuals with positives responses were excluded from evaluation. Figure 1 shows a probable association between inflammation (IL-6 marker) and depressive symptoms in our sample. The effect size indicates a moderate association and demonstrates a source of confounding factors in AD biomarker studies. In our study, subjects with BDI (Beck Depression Inventory) scores above 10 were excluded consequently to avoid misleading interpretations.

Cross-sectional studies examining the association between biomarkers and stages of AD rely on valid classifications of these stages. Healthy aging and the probable AD time course from mild to severe cognitive impairment represent distinct classes for that purpose. We administered a classification scheme that is strongly related to the recommendations of the Consortium to Establish a Registry for Alzheimer ’s disease (CERAD) and highly standardized procedure for diagnosis of AD. Four core domains were assessed: impairments of daily living, a cognitive decline, and the neuropsychological domains learning and memory, executive functions, and attention. Figure 2 illustrates the complete classification scheme. Several (neuro-) psychological constructs and tests were collected for each domain to obtain more reliable values of participants’ real performance. Than the individual test scores were compared against normative and individual standards in order to reach decisions whether a deficit could be present. For instance, verbal learning was assessed with the word list test of the CERAD battery, the narrative test of the Wechsler Memory scale, and the copy condition of the Rey-Complex-Figure test. Three scores were gathered from the word list test and compared against test norms based on the averages of a mild AD population and a population of healthy old age. Narrative and verbal memory were also measured against the performance of a healthy population. We expressed criteria of probable memory deficits in terms of z values and cut-offs (Fig. 2; 4th col.). Applying cut-offs converts all test scores to a dichotomous variable, which were combined by means of logical (and/or) and arithmetic operations (no. of decreased scores) in order to obtain a final decision (Fig. 2; 5th col.). A cognitive decline was detected while measuring the prior level of functioning against the present performance. Here, we administered measures of crystallized intellectual abilities to assess the premorbid level, e.g., by means of a German vocabulary and reading test. If this level were beneath a normal IQ subjects were excluded. In the remaining subjects the level of present functioning was tested by means of a word association and reasoning test. If this level were beneath a normal IQ subjects were identified with a cognitive deterioration.

 

 

Figure 2

Figure 2. Classifying participants by their cognitive state needs to examine at least four domains (I – IV a/b). A wide spread selection of tests and norms is necessary for the neuropsychological assessment (III and IV a/b). Our classification scheme relies on many evaluation items to minimize false-negative as well as false positive judgments. Note a Mini Mental State Examination (MMSE) is not sufficient in that case.

 

Questions often have been raised about neuropsychological assessment concerning its ecological validity. Ecological validity means how well neuropsychological data predicts the behavior, e.g., in daily living. The predictive accuracy of neuropsychological measures has been shown to fail under a variety of “real” behavioral conditions. We targeted this issue in our study by additionally applying a behavioral task. Based on the Morris water maze [4] we developed a virtual reality task that tests spatial learning and navigation skills in humans [5]. That means participants perform the task on a computer screen and navigate by means of a joystick. They had to find a hidden box on a circular island that was surrounded by four landmarks (Fig. 3). First, they perform twelve trials starting from different locations to acquire knowledge how to get to the box as fast as possible. The distance moved and the elapsed time of each trial was measured. Three trials with the box removed were administered after a delay of 30 minutes. The time that participants perseverate at the trained location of the box served as a predictor for long-term spatial memory performance.

By applying the maze paradigm we are not targeting the issue of ecological validity only; deteriorations in spatial memory and navigation performance are often reported as signs of very early AD stages. It is known from animal and human research that deficits in spatial behavior are strongly related to hippocampal [6, 7] and entorhinal [8, 9] violations. Both regions have been also identified as vulnerable against AD pathology in its initial stages (cf. stages I-III in Braak’s staging scheme for AD [10, 11]). We hoped to establish the water maze task as a diagnostic tool of early cognitive deteriorations that ought to be associated with the underlying AD pathology.

 

 

Figure 3-2Figure 3. Configuration and look inside the virtual water maze (upper panel). Landmarks are distal cues to guide participants’ navigation across the island. A hidden box is located in the upper left quadrant that is used as navigation target. 12 consecutive trials were given to acquire the location of the target. Delayed probe and control sessions were given to tests spatial memory and visual motor skills. The probe trials contain no target, and the frequency of search within the upper left (marked) quadrant was measured.

 

To analyze spatial performance, first, all individuals older than 65 years were assigned to a group of probably mild demented (AD), or mild cognitive impaired with spared (non-amnestic; naMCI) or impaired (amnestic; aMCI) memory function. Subjects combining both (non-amnestic and amnestic) deficits were associated with the MCI group; and subjects with no deteriorations at all were assigned to an age matched control group (ctrls). Figure 4 summarized the results of the virtual water maze task. Learning curves from the acquisition (trials 1-9) were shown on the left. Here, the latency (time in seconds participants searching for the hidden box) served as the dependent measure. Healthy controls and naMCIs were on average significantly faster than ADs and aMCIs. Furthermore naMCI and controls showed similar learning curves and no differences in learning performance. Subjects of the aMCI performed worse in the initial trial, came up with a slight improvement over time (trials2-4), but end up with the same learning difficulties shown by the AD group. We observed the same pattern of results after the delay of 30 minutes (Fig. 4; right panel). Here, the first probe trial (with box removed) served as a delayed free recall, and the relative dwell times for each quadrant served as dependent measure. The relative dwell time of the trained quadrant, that former located the box, is used to quantify the amount of spatial memory. Controls and naMCI spent on average more than 25% of their total trial time to search the trained quadrant. These results indicated that controls and naMCIs had some kind of idea where they suggest the hidden box, whereas AD and aMCIs had not. Reduced dwell times in trained quadrant were also associated with increased blood-serum Aβ1-40 levels (r=-0.4, p<.05). One could argue that the group differences were related rather to differences of visual and motor skills that were associated with computer tasks and joystick proficiency. We did not find such differences between our cognitive groups in the maze control condition where the box was made visible. Furthermore, all participants were given preparation trials to get familiar with the computer task and the joystick navigation. See also the individual path plots in Figure 4 (lower panel) for that argument. We selected one subject from each group to present some trials from each part of the maze experiment. It can be seen that all these subjects were able to directly navigate to the box when it was visible (control condition). A further typical pattern observed is circling that is used whenever participants could not perform spatial navigation. Note that the healthy control and the naMCI participant used this pattern only in the beginning of the task, whereas aMCI and AD relapsed to it in many trials.

The spatial and navigational deficits observed in AD as well as in aMCI give further hints toward considering aMCI as a probable prodromal stage of the disease. When we assume strong contributions of the hippocampus to maze task performance then aMCI and AD likely share the same localization of pathology. Increased Aβ levels in the hippocampal region could lead to the memory deficits observed early in aMCI and later on in full demented patients with Alzheimer’s disease.

 

 

Figure 4

Figure 4. Learning curves for four groups of cognitive state – healthy controls (ctrls), non-amnestic (naMCI) or amnestic mild cognitive impairment (aMCI), probable Alzheimer dementia (AD). Group means of the first nine acquisition trials are shown. Means of the control condition were averaged across three trials. Relative dwell time is shown as measure of search frequency within the trained quadrant during two minutes of the 1st probe trial. 25% relative dwell time mark frequencies above the level of chance, which was reached by the control and the naMCI group only. Individual path data (lower panel).

 

References

[1]       Reisberg B (2001) Dementia: Overview. In International Encyclopedia of the Social & Behavioral Sciences, Smelser NJ, Baltes PB eds. Oxford: Pergamon, 3389-3396.

[2]       Folstein MF, Folstein SE, McHugh PR (1975) “Mini-Mental State”: A practical method for grading the cognitive state of patients for the clinician. J Psych Res 12(3), 189-198.

[3]       Petersen RC (2004) Mild cognitive impairment as a diagnostic entity. J Intern Med 256(3), 183-194.

[4]       Morris RG, Garrud P, Rawlins JN, O’Keefe J. Place navigation impaired in rats with hippocampal lesions. Nature. 1982;297(5868):681–683. doi:10.1038/297681a0.

[5]       Schoenfeld R, Moenich N, Mueller F, Lehmann W, Leplow B. Search strategies in a human water maze analogue analyzed with automatic classification methods. Behav Brain Res. 2010;208(1):169–177. doi:10.1016/j.bbr.2009.11.022.

[6]       Bartsch T, Schönfeld R, Müller FJ, et al. Focal Lesions of Human Hippocampal CA1 Neurons in Transient Global Amnesia Impair Place Memory. Science (80- ). 2010;328(5984):1412–1415. doi:10.1126/science.1188160.

[7]       Broadbent, N. J., Squire, L. R., & Clark RE. Spatial memory , recognition memory , and the hippocampus. Proc Natl Acad Sci U S A. 2004;101(40):14515–14520.

[8]       Moser EI, Moser M-B. A metric for space. Hippocampus. 2008;18(12):1142–56. doi:10.1002/hipo.20483.

[9]       Ekstrom A, Viskontas I, Kahana M, et al. Contrasting roles of neural firing rate and local field potentials in human memory. Hippocampus. 2007;17(8):606–617. doi:doi 10.1002/hipo.20300.

[10]     Braak H, Braak E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol. 1991;82(4):239–259. Available at: http://www.springerlink.com/index/mm41klt118937548.pdf. Accessed November 16, 2011.

[11]     Braak H, Alafuzoff I, Arzberger T, Kretzschmar H, Del Tredici K. Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry. Acta Neuropathol. 2006;112(4):389–404. doi:10.1007/s00401-006-0127-z.

 

 

Multiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier SchönmannMultiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier Schönmann