Background
Several studies illustrate the potential of neurogranin as biomarker for Alzheimer’s disease (AD), the most common cause of dementia. Concentrations of this post-synaptic protein are higher in CSF of AD patients compared with cognitively healthy individuals [
1‐
7] and such increased levels probably reflect degeneration of synapses, which has been linked to the cognitive decline [
8,
9]. Contrary to CSF, neurogranin in plasma does not differ between AD patients and controls, indicating that the protein may not be an AD blood biomarker [
3,
10]. Yet, in conditions where sudden severe neurological damage occurs, such as in the event of traumatic brain injury (TBI), increased blood neurogranin concentrations have been found. As recently reported, serum neurogranin levels were elevated in TBI patients compared to controls [
11]. This urges the need for additional studies examining the potential value of CSF and plasma neurogranin in conditions associated with brain injury, like acute ischemic stroke (AIS). So far, neurogranin has not been investigated in AIS although this condition is also characterized by sudden brain damage [
12].
We therefore studied neurogranin in paired CSF/plasma samples of individuals who have suffered from AIS (
n = 50). We analyzed CSF and plasma levels on admission to the hospital, as well as plasma concentrations of neurogranin at different time points after stroke onset. For each of these time points, the relationship was explored between neurogranin levels and stroke severity (as defined by the National Institutes of Health Stroke Scale (NIHSS) [
13]), as well as long-term stroke outcome (represented by the modified Rankin Scale (mRS) [
14]) and infarct volume. To enable a further assessment of neurogranin as biomarker reflecting stroke characteristics, we performed analysis on protein tau as well. Although, to the best of our knowledge, there have been no studies up to date in paired CSF/plasma samples from AIS patients, the brain injury coherent with stroke is seemingly reflected by elevated levels of tau, in both CSF and plasma [
15‐
20]. In the current study, we verified the correlation between tau levels and stroke characteristics.
Methods
Study population
The study population is part of the Middelheim’s Interdisciplinary Stroke Study, a project on the clinical, biochemical, neuroimaging, electrophysiological, and neuropsychological evaluation of AIS patients. These individuals were admitted to the stroke unit of Ziekenhuis Netwerk Antwerpen (ZNA)-Middelheim hospital between October 2005 and February 2008. All patients with suspicion of a cerebrovascular accident or transient ischemic attack were assessed for eligibility: patients were included if clinical evaluation and neuroimaging of the brain (within 24 h after admission) were consistent with AIS. A transient ischemic attack was defined as a clinical presentation believed to be secondary to focal cerebral ischemia, but with symptoms lasting less than 1 h and without proof of acute infarction on structural neuroimaging [
21,
22]. Biochemical analyses from the Middelheim’s Interdisciplinary Stroke Study project have been reported elsewhere [
23‐
26].
At admission, CSF sampling was performed 8.7 h (6.2 h) (mean (standard deviation)) after onset of stroke symptoms as well as blood collection (5.2 h (5.8 h) after stroke onset). Plasma samples were available at admission and at minimal 3 consecutive time points after stroke onset: at 24 h, 72 h, 7 days, 1 month and/or 3 months post-stroke. CSF samples were obtained by lumbar puncture at the L3/L4 or L4/L5 interspaces, using a 20 Gauge, 3.5 in. Quincke point spinal needle (BD, Belgium), collected in polypropylene cryovials (Nalgene, cat.no.5000–1020 and 5000–0050), immediately frozen in liquid nitrogen and stored at −80 °C until analysis. Venous blood samples were collected in 4.9 mL tubes containing EDTA (Sarstedt Monovette, Germany) and immediately centrifuged at 4000 g during 10 min. Plasma was frozen in liquid nitrogen before storing at −80 °C at the biobank facilities.
The NIHSS was used to assess the neurological deficit by qualified stroke physicians at admission. A NIHSS cut-off score of ≤7 and >8 was used to dichotomize the study population based on stroke severity. To estimate the infarct volume, all patients underwent magnetic resonance imaging (MRI) of the brain at admission or during the first week after stroke onset. The infarct volume was assessed on diffusion weighted MRI by two independent raters as described before [
27], using the public domain software Image J (NIH, Bethesda, Md,
http://rsb.info.nih.gov/ij). All lesion areas were manually outlined on a slice-by-slice basis and volumes were automatically produced by the multiplication of the total lesion area times the sum of slice thickness and interslice gap. The final infarct volume was calculated as the mean of the infarct volumes obtained by the two raters. The distance between brain infarct and the lateral ventricles was assessed as well. Long-term functional outcome of the AIS was evaluated using the mRS. A favorable outcome was defined by mRS 0 to 3 and a poor outcome as mRS 4 to 6, which is a commonly used and accepted method for dichotomization of the mRS score [
28]. From all but one patient, data were available on the CSF/serum albumin ratio, calculated as previously described [
29].
Neurogranin in CSF and plasma
To measure neurogranin in both CSF and plasma, we used a research ELISA described in our previous study [
30]. This assay, based on two mouse monoclonal antibodies (mAbs), i.e. ADx403 (clone ADxNGCI2) and ADx451 (clone ADxNGCT1), quantifies specifically neurogranin fragments that are C-terminally truncated at P75. This type of neurogranin is the most abundant form circulating in CSF and plasma based on mass spectrometry data [
10]. CSF samples, in duplicate, were analyzed undiluted, whereas plasma samples, also in duplicate, were diluted 1:20 in sample diluent preceding analysis (15 μL). The serial plasma samples were analyzed anonymously, i.e. samples were randomized. Final concentrations of neurogranin in both CSF and plasma were intrapolated (log(X); 5PL) based on a synthetic peptide calibrator. For the analysis of CSF, both the intra-assay (
n = 2) and inter-plate (
n = 2) variability of the ELISA was 5% coefficient of variation (CV), based on three run validation (RV) samples with different neurogranin concentrations. For plasma, RV samples with low, mid and high concentrations were quantified, resulting in an intra-assay (
n = 2) variability of 3%, 2% and 5%CV, and an inter-plate (
n = 9) precision of 8%, 10% and 14%.
Total-tau in CSF and plasma
To measure total protein levels of tau in CSF, we used the total-tau ELISA by Euroimmun (Lübeck, Germany). The kit includes lyophilized, ready-to-use calibrators, a standardized protocol, and was performed according to the kit-insert. All CSF samples were run in duplicate. Intra-assay (n = 2) variability was 4%CV for RV samples with a low or a high tau concentration. Inter-plate (n = 2) precision was 3% and 4%CV respectively.
For quantification of total-tau levels in plasma samples, an in-house research prototype ELISA was used [
3]. Undiluted plasma samples were analyzed in duplicate. Tau levels were calculated via intrapolation (log(X); 5PL) based on the calibrator, i.e.
E.coli recombinant tau 441 (rPeptide, USA), but the quantifications were expressed as arbitrary units (AU) since this concerns a not yet fully developed research assay. The research ELISA enabled detection of tau in 92% of the plasma samples. Two RV samples with different tau levels were run in parallel with the patient samples, resulting in an intra-plate (
n = 2) variability of 5% and 3%CV, and inter-plate (
n = 9) precision of 8% and 5%.
Tau plasma levels were also measured using the Simoa assay by Quanterix (Boston, USA), as described elsewhere [
31,
32], for which samples, in duplicate, were diluted 4-fold. Based on two samples with different tau concentrations, intra-assay (
n = 2) variability was 11% and 3%CV, whereas inter-mediate (
n = 4) precision was 11% and 5%CV. Tau could be quantified in all plasma samples.
As in case of the analysis of neurogranin, tau levels were quantified in randomized plasma samples, blinded from the test performer.
Statistical analysis
GraphPad Prism 6 was used for statistical analyses and figures. To test normality of data, a Shapiro-Wilk test was performed. In case of a non-normal distribution, all subsequent analyses were based on log-transformed data. To compare quantitative data between two groups, a Mann-Whitney test was applied, whereas correlations were determined using Pearson’s correlation analysis. Results were considered significant for P-values <0.05.
Discussion
Although progressive brain damage, as in case of AD, is clearly not reflected in plasma levels of neurogranin, it remains to be investigated how acute brain injury affects the blood levels. In the current study, we focused on acute ischemic stroke and observed a relationship between increased plasma neurogranin and infarct volume. Yet, neither stroke severity nor long-term outcome were reflected by neurogranin in plasma, or CSF. Levels of axonal protein tau, on the other hand, were significantly associated with stroke severity and long-term outcome, indicating that different neuronal proteins may entail different kinds of information, at least in AIS.
Due to the high plasma concentrations of neurogranin in normal individuals, i.e. more than 1000 pg/mL, small increases are possibly not detected in plasma in case of slow progressive neurodegeneration like AD [
3,
10]. Only in conditions of acute severe brain trauma, such as TBI, neurogranin may serve as potential plasma biomarker. Serum neurogranin concentrations in TBI cases were recently reported as significantly higher than in controls [
11]. In our study on AIS, we monitored plasma neurogranin from admission to the hospital until 3 months post-stroke. This aspect was suggested in the recent study [
11] on neurogranin in TBI since this could offer insights in the temporal profile of the protein in blood, following acute brain damage. We observed a trend in increasing plasma levels of the protein, with the highest concentration at 3 m. Blood brain barrier dysfunction has been reported in stroke [
33], which could allow progressive leakage of brain-derived neurogranin into the plasma. However, this contradicts the observation of the lowest plasma neurogranin levels at 72 h being inversely related to a high CSF/serum albumin ratio, suggesting that lower neurogranin levels may be linked to a dysfunctional blood brain barrier. To the best of our knowledge, this is the first report on neurogranin in AIS and further studies are needed, preferentially in moderate or even severe conditions of brain damage. Based on the relatively low NIHSS scores, the patients suffered from a mild stroke in the current study, which is corroborated by the rather low CSF tau values at admission. CSF tau has been reported to substantially increase in case of stroke, i.e. about a 4-fold increase in case of higher NIHSS values [
15]. Hence, more severe stroke may result in higher neurogranin levels. Nonetheless, our current dataset suggests that neurogranin may not have great value as plasma biomarker, or as CSF biomarker, for acute brain injury in case of stroke. Besides only observing a trend in increasing levels in function of time, plasma levels were inversely related to the distance between the infarct and the ventricles, which represents a confounding factor. In case the infarct occurs at a larger distance from the ventricles, this could lead to a possible underestimate of the infarct volume. Moreover, even though levels in plasma and CSF of neurogranin were significantly associated with infarct volume, levels were not related to stroke severity, neither to long-term outcome, which are clinical relevant characteristics. Perhaps neurogranin could still have value as biomarker, reflecting infarct volume, when combined with other proteins, such as GFAP, and UCH-L1, which show potential in case of TBI [
34,
35]. Or, it could even be combined with tau since our findings are strongly indicative for a high performance of tau, even as single analyte, as measure for stroke severity and outcome in AIS.
Plasma tau significantly peaked 7 days after stroke, where levels were linked to stroke severity, i.e. NIHSS, long-term outcome, represented by the mRS, as well as infarct volume, which confirms prior reports [
19,
20]. No relationship with the distance between the infarct and the ventricles, nor with the CSF/serum albumin ratio was seen at that time point. Similarly, CSF levels of tau, which correlated to the plasma levels of the protein, were related to NIHSS assessment at admission and mRS rating 12 months after stroke. This result was in line with previously published studies on tau in CSF from AIS patients [
15,
17,
36]. Interestingly, CSF tau at admission was not associated with infarct volume unlike plasma tau. Since plasma tau only peaked 7 days after stroke, it is plausible that CSF tau does not adequately reflects infarct volume at admission, but rather 72 h to 1 week after stroke onset. This was actually suggested by Hesse and colleagues, who demonstrated increasing CSF tau levels over time that peaked one or 3 weeks after stroke onset. Only at maximum CSF tau concentrations, there was a correlation with infarct volume [
16,
17]. This corroborates earlier findings on higher CSF tau levels in patients with large infarcts in the subacute stage, i.e. less than 1 month after stroke onset, compared to patients in the acute stage, i.e. less than 1 week [
37]. Importantly, the latter report mentioned that even small infarct volumes resulted in elevated CSF tau levels several days after stroke.
Although plasma tau was the best candidate biomarker in the current study, compared to neurogranin, the search continues for non-invasive AIS biomarkers providing an early evaluation of the stroke. Ideally, physicians can gauge the severity and long-term outcome of the acute brain damage on admission to the hospital. Most likely this will be enabled by analysis of a panel of proteins, possibly including plasma tau.