Limits...
CSF protein biomarkers predicting longitudinal reduction of CSF β-amyloid42 in cognitively healthy elders.

Mattsson N, Insel P, Nosheny R, Zetterberg H, Trojanowski JQ, Shaw LM, Tosun D, Weiner M, Alzheimer's Disease Neuroimaging Initiati - Transl Psychiatry (2013)

Bottom Line: It is believed to start many years prior to symptoms and is reflected by reduced cerebrospinal fluid (CSF) levels of the peptide Aβ1-42 (Aβ42).There was an overall reduction in Aβ42 from a mean concentration of 211-195 pg ml(-1) after 4 years.We conclude that baseline CSF proteins related to Aβ metabolism, microglia activity or synapses predict longitudinal Aβ42 reduction in cognitively healthy elders.

View Article: PubMed Central - PubMed

Affiliation: Department of Veterans Affairs Medical Center, Center for Imaging of Neurodegenerative Diseases, San Francisco, CA 94121, USA. niklas.mattsson@neuro.gu.se

ABSTRACT
β-amyloid (Aβ) plaque accumulation is a hallmark of Alzheimer's disease (AD). It is believed to start many years prior to symptoms and is reflected by reduced cerebrospinal fluid (CSF) levels of the peptide Aβ1-42 (Aβ42). Here we tested the hypothesis that baseline levels of CSF proteins involved in microglia activity, synaptic function and Aβ metabolism predict the development of Aβ plaques, assessed by longitudinal CSF Aβ42 decrease in cognitively healthy people. Forty-six healthy people with three to four serial CSF samples were included (mean follow-up 3 years, range 2-4 years). There was an overall reduction in Aβ42 from a mean concentration of 211-195 pg ml(-1) after 4 years. Linear mixed-effects models using longitudinal Aβ42 as the response variable, and baseline proteins as explanatory variables (n=69 proteins potentially relevant for Aβ metabolism, microglia or synaptic/neuronal function), identified 10 proteins with significant effects on longitudinal Aβ42. The most significant proteins were angiotensin-converting enzyme (ACE, P=0.009), Chromogranin A (CgA, P=0.009) and Axl receptor tyrosine kinase (AXL, P=0.009). Receiver-operating characteristic analysis identified 11 proteins with significant effects on longitudinal Aβ42 (largely overlapping with the proteins identified by linear mixed-effects models). Several proteins (including ACE, CgA and AXL) were associated with Aβ42 reduction only in subjects with normal baseline Aβ42, and not in subjects with reduced baseline Aβ42. We conclude that baseline CSF proteins related to Aβ metabolism, microglia activity or synapses predict longitudinal Aβ42 reduction in cognitively healthy elders. The finding that some proteins only predict Aβ42 reduction in subjects with normal baseline Aβ42 suggest that they predict future development of the brain Aβ pathology at the earliest stages of AD, prior to widespread development of Aβ plaques.

Show MeSH

Related in: MedlinePlus

Proteins and quartiles of CSF Aβ42 change in healthy controls. The graph includes the baseline proteins that significantly predicted longitudinal reduction in CSF Aβ42 in ROC analysis. For this analysis, subjects were ranked according to relative change in Aβ42 over 4 years, and subjects with most changes (1st quartile of subjects >11.5% reduction, red boxes) were compared with subjects with least changes (4th quartile of subjects <4.0% reduction, blue boxes). Subjects in the 2nd and 3rd quartiles (representing intermediate reductions) were not included in the ROC analysis but are shown for comparison in yellow boxes (for most proteins, subjects in the 2nd and 3rd quartiles had concentrations comparable to subjects in the 1st quartile). Baseline protein levels were centered and standardized.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3756294&req=5

fig3: Proteins and quartiles of CSF Aβ42 change in healthy controls. The graph includes the baseline proteins that significantly predicted longitudinal reduction in CSF Aβ42 in ROC analysis. For this analysis, subjects were ranked according to relative change in Aβ42 over 4 years, and subjects with most changes (1st quartile of subjects >11.5% reduction, red boxes) were compared with subjects with least changes (4th quartile of subjects <4.0% reduction, blue boxes). Subjects in the 2nd and 3rd quartiles (representing intermediate reductions) were not included in the ROC analysis but are shown for comparison in yellow boxes (for most proteins, subjects in the 2nd and 3rd quartiles had concentrations comparable to subjects in the 1st quartile). Baseline protein levels were centered and standardized.

Mentions: For the ROC statistics, we compared subjects showing the most Aβ42 reduction (1st quartile >11.5% reduction from baseline) with subjects showing the least reduction (4th quartile <4.0% reduction from baseline). Note that the 11.5% cutoff for Aβ42 reduction exceeds the within-run measurement variability of the xMAP AlzBio3 Aβ42 assay and thereby very likely reflects a biological change.27 By these definitions, 12 subjects had Aβ42 reduction (four APOE ɛ4+, six female subjects, mean age 80 years (range 73–88, mini mental state examination 29.3 (28–30), ADAS-Cog 5.7 (1.7–15), ADAS-MOD 8.9 (1.7–18)) and 12 subjects had stable Aβ42 (one APOE ɛ4+, six female subjects, aged 78 years (72–87), mini mental state examination 29.6 (20–30), ADAS-Cog 6.8 (2.3–10), ADAS-MOD 10 (4.0–15)). In the ROC analysis, 11 proteins significantly predicted Aβ42 reduction (Figure 3 and Supplementary Table 4). These proteins largely overlapped with the proteins identified by the linear mixed effect analysis described above. Highest AUC were seen for ACE (AUC=0.87 (95% CI 0.72–1.00)), B2M (AUC=0.83 (95% CI 0.67–0.98)) and CgA (AUC=0.83 (95% CI 0.67–0.98)). All identified significant proteins were higher at baseline in subjects with longitudinal Aβ42 reduction, except Cystatin C, which was lower in subjects with longitudinal Aβ42 reduction.


CSF protein biomarkers predicting longitudinal reduction of CSF β-amyloid42 in cognitively healthy elders.

Mattsson N, Insel P, Nosheny R, Zetterberg H, Trojanowski JQ, Shaw LM, Tosun D, Weiner M, Alzheimer's Disease Neuroimaging Initiati - Transl Psychiatry (2013)

Proteins and quartiles of CSF Aβ42 change in healthy controls. The graph includes the baseline proteins that significantly predicted longitudinal reduction in CSF Aβ42 in ROC analysis. For this analysis, subjects were ranked according to relative change in Aβ42 over 4 years, and subjects with most changes (1st quartile of subjects >11.5% reduction, red boxes) were compared with subjects with least changes (4th quartile of subjects <4.0% reduction, blue boxes). Subjects in the 2nd and 3rd quartiles (representing intermediate reductions) were not included in the ROC analysis but are shown for comparison in yellow boxes (for most proteins, subjects in the 2nd and 3rd quartiles had concentrations comparable to subjects in the 1st quartile). Baseline protein levels were centered and standardized.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3756294&req=5

fig3: Proteins and quartiles of CSF Aβ42 change in healthy controls. The graph includes the baseline proteins that significantly predicted longitudinal reduction in CSF Aβ42 in ROC analysis. For this analysis, subjects were ranked according to relative change in Aβ42 over 4 years, and subjects with most changes (1st quartile of subjects >11.5% reduction, red boxes) were compared with subjects with least changes (4th quartile of subjects <4.0% reduction, blue boxes). Subjects in the 2nd and 3rd quartiles (representing intermediate reductions) were not included in the ROC analysis but are shown for comparison in yellow boxes (for most proteins, subjects in the 2nd and 3rd quartiles had concentrations comparable to subjects in the 1st quartile). Baseline protein levels were centered and standardized.
Mentions: For the ROC statistics, we compared subjects showing the most Aβ42 reduction (1st quartile >11.5% reduction from baseline) with subjects showing the least reduction (4th quartile <4.0% reduction from baseline). Note that the 11.5% cutoff for Aβ42 reduction exceeds the within-run measurement variability of the xMAP AlzBio3 Aβ42 assay and thereby very likely reflects a biological change.27 By these definitions, 12 subjects had Aβ42 reduction (four APOE ɛ4+, six female subjects, mean age 80 years (range 73–88, mini mental state examination 29.3 (28–30), ADAS-Cog 5.7 (1.7–15), ADAS-MOD 8.9 (1.7–18)) and 12 subjects had stable Aβ42 (one APOE ɛ4+, six female subjects, aged 78 years (72–87), mini mental state examination 29.6 (20–30), ADAS-Cog 6.8 (2.3–10), ADAS-MOD 10 (4.0–15)). In the ROC analysis, 11 proteins significantly predicted Aβ42 reduction (Figure 3 and Supplementary Table 4). These proteins largely overlapped with the proteins identified by the linear mixed effect analysis described above. Highest AUC were seen for ACE (AUC=0.87 (95% CI 0.72–1.00)), B2M (AUC=0.83 (95% CI 0.67–0.98)) and CgA (AUC=0.83 (95% CI 0.67–0.98)). All identified significant proteins were higher at baseline in subjects with longitudinal Aβ42 reduction, except Cystatin C, which was lower in subjects with longitudinal Aβ42 reduction.

Bottom Line: It is believed to start many years prior to symptoms and is reflected by reduced cerebrospinal fluid (CSF) levels of the peptide Aβ1-42 (Aβ42).There was an overall reduction in Aβ42 from a mean concentration of 211-195 pg ml(-1) after 4 years.We conclude that baseline CSF proteins related to Aβ metabolism, microglia activity or synapses predict longitudinal Aβ42 reduction in cognitively healthy elders.

View Article: PubMed Central - PubMed

Affiliation: Department of Veterans Affairs Medical Center, Center for Imaging of Neurodegenerative Diseases, San Francisco, CA 94121, USA. niklas.mattsson@neuro.gu.se

ABSTRACT
β-amyloid (Aβ) plaque accumulation is a hallmark of Alzheimer's disease (AD). It is believed to start many years prior to symptoms and is reflected by reduced cerebrospinal fluid (CSF) levels of the peptide Aβ1-42 (Aβ42). Here we tested the hypothesis that baseline levels of CSF proteins involved in microglia activity, synaptic function and Aβ metabolism predict the development of Aβ plaques, assessed by longitudinal CSF Aβ42 decrease in cognitively healthy people. Forty-six healthy people with three to four serial CSF samples were included (mean follow-up 3 years, range 2-4 years). There was an overall reduction in Aβ42 from a mean concentration of 211-195 pg ml(-1) after 4 years. Linear mixed-effects models using longitudinal Aβ42 as the response variable, and baseline proteins as explanatory variables (n=69 proteins potentially relevant for Aβ metabolism, microglia or synaptic/neuronal function), identified 10 proteins with significant effects on longitudinal Aβ42. The most significant proteins were angiotensin-converting enzyme (ACE, P=0.009), Chromogranin A (CgA, P=0.009) and Axl receptor tyrosine kinase (AXL, P=0.009). Receiver-operating characteristic analysis identified 11 proteins with significant effects on longitudinal Aβ42 (largely overlapping with the proteins identified by linear mixed-effects models). Several proteins (including ACE, CgA and AXL) were associated with Aβ42 reduction only in subjects with normal baseline Aβ42, and not in subjects with reduced baseline Aβ42. We conclude that baseline CSF proteins related to Aβ metabolism, microglia activity or synapses predict longitudinal Aβ42 reduction in cognitively healthy elders. The finding that some proteins only predict Aβ42 reduction in subjects with normal baseline Aβ42 suggest that they predict future development of the brain Aβ pathology at the earliest stages of AD, prior to widespread development of Aβ plaques.

Show MeSH
Related in: MedlinePlus