Article Text

Evaluation of CSF biomarkers for axonal and neuronal degeneration, gliosis, and β-amyloid metabolism in Alzheimer's disease
  1. N ANDREASEN
  1. Department of Rehabilitation, Piteå River Valley Hospital, PO Box 715, SE-941 28 Piteå, Sweden
  2. Astra-Zeneca, Mölndal, Sweden
  3. Innogenetics, Ghent, Belgium
  4. Department of Clinical Neuroscience, Section of Experimental Neuroscience, University of Göteborg, Sahlgren's University Hospital, Mölndal, Sweden
  5. Department of Clinical Neuroscience, Section of Neurology, University of Göteborg, Sahlgren's University Hospital, Göteborg, Sweden
  6. The Medical Research Council, Sweden
  1. Dr N Andreasen niels.andreasen{at}nll.se
  1. J GOTTFRIES
  1. Department of Rehabilitation, Piteå River Valley Hospital, PO Box 715, SE-941 28 Piteå, Sweden
  2. Astra-Zeneca, Mölndal, Sweden
  3. Innogenetics, Ghent, Belgium
  4. Department of Clinical Neuroscience, Section of Experimental Neuroscience, University of Göteborg, Sahlgren's University Hospital, Mölndal, Sweden
  5. Department of Clinical Neuroscience, Section of Neurology, University of Göteborg, Sahlgren's University Hospital, Göteborg, Sweden
  6. The Medical Research Council, Sweden
  1. Dr N Andreasen niels.andreasen{at}nll.se
  1. E VANMECHELEN,
  2. H VANDERSTICHELE
  1. Department of Rehabilitation, Piteå River Valley Hospital, PO Box 715, SE-941 28 Piteå, Sweden
  2. Astra-Zeneca, Mölndal, Sweden
  3. Innogenetics, Ghent, Belgium
  4. Department of Clinical Neuroscience, Section of Experimental Neuroscience, University of Göteborg, Sahlgren's University Hospital, Mölndal, Sweden
  5. Department of Clinical Neuroscience, Section of Neurology, University of Göteborg, Sahlgren's University Hospital, Göteborg, Sweden
  6. The Medical Research Council, Sweden
  1. Dr N Andreasen niels.andreasen{at}nll.se
  1. P DAVIDSSON,
  2. K BLENNOW
  1. Department of Rehabilitation, Piteå River Valley Hospital, PO Box 715, SE-941 28 Piteå, Sweden
  2. Astra-Zeneca, Mölndal, Sweden
  3. Innogenetics, Ghent, Belgium
  4. Department of Clinical Neuroscience, Section of Experimental Neuroscience, University of Göteborg, Sahlgren's University Hospital, Mölndal, Sweden
  5. Department of Clinical Neuroscience, Section of Neurology, University of Göteborg, Sahlgren's University Hospital, Göteborg, Sweden
  6. The Medical Research Council, Sweden
  1. Dr N Andreasen niels.andreasen{at}nll.se
  1. L ROSENGREN
  1. Department of Rehabilitation, Piteå River Valley Hospital, PO Box 715, SE-941 28 Piteå, Sweden
  2. Astra-Zeneca, Mölndal, Sweden
  3. Innogenetics, Ghent, Belgium
  4. Department of Clinical Neuroscience, Section of Experimental Neuroscience, University of Göteborg, Sahlgren's University Hospital, Mölndal, Sweden
  5. Department of Clinical Neuroscience, Section of Neurology, University of Göteborg, Sahlgren's University Hospital, Göteborg, Sweden
  6. The Medical Research Council, Sweden
  1. Dr N Andreasen niels.andreasen{at}nll.se
  1. K BLENNOW
  1. Department of Rehabilitation, Piteå River Valley Hospital, PO Box 715, SE-941 28 Piteå, Sweden
  2. Astra-Zeneca, Mölndal, Sweden
  3. Innogenetics, Ghent, Belgium
  4. Department of Clinical Neuroscience, Section of Experimental Neuroscience, University of Göteborg, Sahlgren's University Hospital, Mölndal, Sweden
  5. Department of Clinical Neuroscience, Section of Neurology, University of Göteborg, Sahlgren's University Hospital, Göteborg, Sweden
  6. The Medical Research Council, Sweden
  1. Dr N Andreasen niels.andreasen{at}nll.se

Statistics from Altmetric.com

Although the accuracy rate of the clinical diagnosis of Alzheimer's disease is around 75%–90%, it is probably considerably lower early in the disease course, when symptoms are vague. Therefore, in view of potential future disease modifying compounds there is a great need for reliable diagnostic biochemical markers for Alzheimer's disease in CSF.

Such markers should reflect the central pathogenic processes of the disease—that is, the disturbance in the metabolism of β-amyloid (Aβ) with subsequent Aβ deposition in senile plaques, the hyperphosphorylation of tau protein with subsequent formation of neurofibrillary tangles, neuronal degeneration, and gliosis.

Two promising biomarkers are tau protein (reflecting neuronal and axonal degeneration) and Aβ42 (reflecting disturbances in Aβ metabolism and possibly Aβ deposition in senile plaques). The ability of the combination of CSF tau and CSF Aβ42 to differentiate Alzheimer's disease from normal aging and depression is high, about 85%, also early in the course of the disease.1 Similarly, most degenerative neurological disorders have normal concentrations. However, the specificity against other dementias is not optimal.1 Thus, there is a need for additional CSF biomarkers for Alzheimer's disease, to further increase the diagnostic accuracy.

We therefore examined whether the addition of other CSF biomarkers (two neuronal and two glial proteins) would add further to the diagnostic ability to identify Alzheimer's disease. The neuronal proteins were neurofilament protein light subunit (NFL), the major protein component of neurofilaments (probably reflecting degeneration of myelinated axons) and neuron specific enolase (NSE), a neuronal glycolytic enzyme (probably reflecting degeneration of neuronal cell bodies). The glial proteins were glial fibrillary acidic protein (GFAP), an astrocyte specific protein considered to be the major component of glial filaments in reactive astrocytes, and S-100β, a calcium binding protein also found in astrocytes (both reflecting gliosis).

From the longitudinal geriatric population study in Piteå, Sweden2 we studied 35 patients with Alzheimer's disease, mean age 72.1 (SD 5.9) years, duration of disease of 48.9 (SD 32.0) months, and with MMSE scores of 23.5 (SD 4.7). The control group consisted of 19 subjects, mean age 71.2 (SD 7.3) years, without symptoms or signs of brain disorders, all with MMSE scores above 28.

The ethics committees at the universities of Umeå and Göteborg approved the study, conducted in accordance with the provisions of the Helsinki Declaration.

Analyses of CSF were performed using enzyme linked immunosorbent assays (ELISAs) as described previously in detail for total tau, Aβ42,1 NFL,3 GFAP,4 and S-100β.5 The NSE in CSF was determined using a commercial ELISA from AB Sangtec Medical, Bromma, Sweden.

The Mann-Whitney U test was used for group comparisons and the Pearson correlation coefficient for correlations. The dataset was also investigated by principal component analysis using the SIMCA-S software (Umetri AB, Umeå, Sweden), and by partial least squares with cross validation as a validation tool for multivariate correlations between CSF biomarkers and diagnosis.

When comparing CSF biomarkers between patients with Alzheimer's disease and controls (values given as means (SD)), there was a significant increase in CSF tau (634 (288) v375 (171) pg/ml; p<0.0001), and in CSF NFL (615 (456)v 295 (194) pg/ml; p=0.002). There was also a significant decrease in CSF Aβ42 in patients with Alzheimer's disease compared with controls (748 (297) v1623 (429) pg/ml; p<0.0001) and a slight but significant decrease in CSF S-100β (1.8 (0.9) v 2.5 (0.9) μg/l; p=0.014). By contrast, there were no significant differences in CSF-NSE (7.4 (2.7) v 6.9 (2.1) μg/l; p=0.568) or CSF GFAP (860 (297) v 717 (250) ng/l; p=0.097).

The combination of CSF tau and CSF Aβ42 gave at best a sensitivity of 32/35 (91.4%) and a specificity of 17/19 (89.5%). A partial least squares analysis with all CSF biomarkers and clinical groups (Alzheimer's disease and controls), showed a relation between a diagnosis of Alzheimer's disease and high CSF tau, high CSF NFL, high CSF GFAP, and low CSF Aβ42 concentrations (fig 1). The NSE and S-100β in CSF showed no discriminative power so these additional biomarkers gave no further aid in the discrimination between Alzheimer's disease and controls. The sensitivity using all CSF biomarkers was 34/35 (97.1%) and the specificity was 17/19 (89.5%).

Figure 1

Top: CSF biomarker intercorrelation for principal components 1 and 2 from partial least squares-discriminant analysis. The discriminant regressor was Alzheimer's disease (filled circle) or healthy control (open square). Highest correlation was found for Aβ42 with Alzheimer's disease. Tau and NFL also correlated with Alzheimer's disease; however, including some structure not related to the disease (loading in principal component 2). Bottom: Interindividual scores of included study objects for principal components 1 and 2 from partial least squares-DA. The principal components were derived by a projection of assessed CSF protein concentrations (included protein assessment as in the figure above). Filled circles=Alzheimer's disease; open squares=healthy controls.

In agreement with previous findings, increased CSF tau and decreased CSF Aβ421 was found in Alzheimer's disease, resulting in a good sensitivity and specificity for discriminating Alzheimer's disease from controls. As the ability of these CSF biomarkers to discriminate Alzheimer's disease from other dementia disorders is less than optimal, we tested whether the combined analysis of additional biomarkers for axonal degeneration (CSF NFL), neuronal degeneration (CSF NSE), and gliosis (CSF GFAP and CSF S-100β) resulted in any further increase in the diagnostic sensitivity or specificity. However, there was only a marginal increase in sensitivity (from 91.4% to 97.1%) whereas the specificity was unchanged (89.5%). Therefor we conclude that these biomarkers have little additional value as diagnostic biochemical markers for Alzheimer's disease.

We hypothesise that other biomarkers more specifically related to Alzheimer's disease pathogenesis, such as hyperphosphorylated tau, synapse specific proteins (for example, rab3a, synaptotagmin), or APP isoforms (for example, α-secretase or β-secretase cleaved APP), may have a larger potential as CSF biomarkers for Alzheimer's disease.

Acknowledgments

This work was supported by grants from the Swedish Medical Research Council (grants 12103 and 11560); Alzheimerfonden, Lund, Sweden; Stiftelsen för Gamla Tjänarinnor, Stockholm, Sweden; Tore Nilssons Fond för Medicinsk Forskning, Stockholm, Sweden; Norrbottens Läns Landstings FoU Fond, Sweden; Svenska Läkaresällskapet, Stockholm, Sweden; and Åke Wibergs Stiftelse, Stockholm, Sweden. We are grateful to Mrs Christina Sjödin and Mrs. Maria Lindbjer for skillful technical assistance.

References

View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.