Abstract
We investigated cerebrospinal fluid (CSF) concentrations of the postsynaptic biomarker neurogranin at baseline in cognitively healthy controls (HC) compared to individuals with mild cognitive impairment (MCI), patients with Alzheimer’s disease (AD) dementia, and patients with frontotemporal dementia (FTD). CSF neurogranin was quantified using an in-house immunoassay in a cross-sectional multicenter study of 108 participants [AD dementia (n = 35), FTD (n = 9), MCI (n = 41), cognitively HC (n = 23)]. CSF neurogranin concentrations were significantly higher in AD patients compared with both HC subjects and FTD patients, suggesting that increased CSF neurogranin concentrations may indicate AD-related pathophysiology. CSF neurogranin was independently associated with both total tau and hyperphosphorylated tau proteins, whereas a non-significant correlation with the 42-amino acid-long amyloid-β peptide was evident. CSF neurogranin, however, was not superior to core AD biomarkers in differentiating HC from the three diagnostic groups, and it did not improve their diagnostic accuracy. We conclude that further classification and longitudinal studies are required to shed more light into the potential role of neurogranin as a pathophysiological biomarker of neurodegenerative diseases.
Keywords
INTRODUCTION
Synaptic pathology plays a central role in Alzheimer’s disease (AD) [1, 2] and correlates with cognitive decline [3–6]. Because “synaptic failure” is increasingly being recognized as a core feature of AD pathophysiology [7], synaptic proteins hold promise to serve as novel candidate markers of neurodegenerative disorders. Neurogranin—a postsynaptic protein mainly localized in dendritic spines of neurons within associative cortical areas [8–10]—is involved in synaptic plasticity [11]. Neurogranin expression is significantly lower in the cortex and hippocampus of postmortem AD brains compared with control tissue [1, 12]. Compared with healthy controls (HC), cerebrospinal fluid (CSF) neurogranin concentrations are increased in patients with AD [13–15] and subjects with mild cognitive impairment (MCI) as well as MCI progressors and converters to AD (MCI-AD) [13, 16]. Moreover, neurogranin predicts progression from MCI to AD dementia [13, 18] and rate of cognitive decline [13], and correlates longitudinally with rates of hippocampal atrophy [18, 19] as well as with reduced regional cortical glucose metabolism assessed by 18F-Fluorodeoxyglucose-positron emission tomography (18F-FDG-PET) [18]. Very recently, CSF profiling of the human brain enriched proteome has confirmed a significant association between increased neurogranin concentrations and AD [20].
The aim of this study was two-fold. First, we sought to investigate CSF neurogranin concentrations in different neurodegenerative diseases. Second, we assessed whether CSF neurogranin may be superior to core CSF biomarkers of AD—namely the 42-amino acid-long amyloid-β peptide (Aβ42), total tau (t-tau), and hyperphosphorylated tau (p-tau) proteins—and/or it could improve their diagnostic performance.
MATERIALS AND METHODS
Study participants
The research was a multicenter cross-sectional study of a convenience sample of subjects recruited from three independent European memory clinics. A total of 135 individuals were examined. Of these participants, 27 were excluded due to missing data in one or more CSF biomarkers and the remaining 108 were included in the present study. Specifically, 35 participants were recruited from the Institute of Memory and Alzheimer’s Disease (Institut de la Mémoire et de la Maladie d’Alzheimer, IM2A), a sub-cohort of the Alzheimer Precision Medicine Initiative Cohort Program (APMI-CP) [21], at the Pitié-Salpêtrière University Hospital in Paris (France), 57 from the German Center for Neurodegenerative Diseases (DZNE) in Rostock (Germany), and 16 from the Institute of Neuroscience and Physiology at Sahlgrenska University Hospital in Mölndal (Sweden). The study complied with the tenets of the Declaration of Helsinki and was approved by the local Ethical Committees at each participating university center. All participants or their representatives gave written informed consent for the use of their clinical data for research purposes.
Clinical diagnoses
The clinical diagnosis of AD dementia was performed according to the National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) consensus criteria [22]. The clinical diagnosis of MCI was based on MCI core clinical criteria [23]. The diagnosis of the FTD was performed according to the consensus on clinical diagnostic criteria published in 1998 [24]. HC were individuals who 1) volunteered for a lumbar puncture, 2) showed a negative history of neurological or psychiatric diseases, and 3) had a Mini-Mental State Examination (MMSE) score between 27 and 30. Of the 23 cognitively HC, two individuals from the Gothenburg cohort showed CSF t-tau concentrations higher than the established cut-off value. Being asymptomatic-at-risk of AD [25] or preclinical AD [26], they were excluded from further analyses. The group clinically defined as MCI [23] included 41 participants. Finally, 35 AD dementia [25] and 9 FTD [24] patients were included.
CSF sampling
A diagnostic lumbar puncture was performed in all participants. All CSF samples included in the three study cohorts were collected in polypropylene tubes, centrifuged at 1000 g for 10 min at +4°C (samples collected at IM2A in Paris), 1500 g for 10 min at +4°C (samples collected at DZNE in Rostock), 1800 g for 10 min at +4°C (samples collected at Clinical Neurochemistry Laboratory in Mölndal). The collected supernatant was aliquoted and stored at –80°C pending biochemical analysis.
Immunoassays for CSF core biomarkers
For the Paris cohort, CSF analyses of the biomarkers Aβ42, t-tau, and p-tau were performed at the Laboratory of Biochemistry, Unit of Biochemistry of Neurometabolic diseases, Pitié-Salpêtrière University Hospital of Paris. For the Rostock cohort, CSF analyses were executed in two different units: the Institute of Clinical Chemistry and Laboratory Medicine, Rostock University Medical Center, as of June 2012, and the Laboratory of Neurochemistry, Department of Neurology, Göttingen University Medical Center, before June 2012. For the Gothenburg cohort, CSF analyses were executed at the Clinical Neurochemistry Laboratory at the Sahlgrenska University Hospital, Mölndal. CSF Aβ42, t-tau, and tau phosphorylated at threonine 181 (p-tau181) concentrations were measured using established sandwich ELISA methods, INNOTEST β-AMYLOID(1–42) [27], INNOTEST hTAU-Ag [28], and INNOTEST Phospho-Tau[181P] [29] (Fujirebio Europe NV, Gent, Belgium), respectively. All analyses were performed by board-certified laboratory technicians blinded to clinical information. CSF biomarkers abnormalities were defined based on reference threshold cut-off values currently utilized in each memory clinic: at IM2A in Paris, Aβ42 <500 pg/mL, t-tau>450 pg/mL, p-tau181 >60 pg/mL; at DZNE in Rostock, Aβ42 <567 pg/mL, t-tau >512 pg/mL, p-tau181 >66 pg/mL for the CSF samples measured before 06/2012 and Aβ42 <450 pg/mL, t-tau >450 pg/mL, p-tau181 >62 pg/mL for the CSF samples measured after 06/2012; at Clinical Neurochemistry Laboratory in Mölndal, Aβ42 <550 pg/mL, t-tau >400 pg/mL, p-tau181 >80 pg/mL.
Immunoassay for CSF neurogranin
All CSF neurogranin analyses were performed at the Clinical Neurochemistry Laboratory at the Sahlgrenska University Hospital (Mölndal, Sweden) using a previously described analytical methodology [13]. In short, CSF neurogranin was measured using an in-house ELISA assay based on the monoclonal antibody Ng7 (epitope including amino acids 52–65 on neurogranin) for capture, a polyclonal neurogranin anti-rabbit antibody (ab23570; Upstate Biotechnology, Lake Placid, NY, USA) for detection, and full-length neurogranin protein as calibrator. All analyses were performed in a randomized fashion by laboratory personnel blinded to clinical data. The detection limit of the assay was 125 pg/mL. The intra- and inter-assay coefficients of variations were 6% and 9%, respectively.
Statistical analysis
Differences in neurogranin concentrations according to different diagnostic categories were tested with the nonparametric Kruskal-Wallis test followed by post-hoc Conover’s test for multiple comparisons. All neurogranin values were initially adjusted for age, sex, and study site with nonparametric regression. The associations of neurogranin with core biomarkers in the entire study cohort were initially investigated with Spearman’s correlation coefficients and subsequently by multivariable linear regression analysis. The discriminatory ability of neurogranin to correctly allocate participants to different diagnostic groups was investigated by comparisons of areas under the receiver operating characteristic (AUROCs) curves. The higher the AUROCs values the better was the discriminatory ability, as follows: “excellent” (AUROC 0.90–1.00), “good” (AUROC 0.80–0.89), “fair” (AUROC 0.70–0.79), “poor” (AUROC 0.60–0.69), or “fail”/no discriminatory capacity (AUROC 0.50–0.59) [30]. We initially calculated and compared the AUROCs for neurogranin and single core biomarkers separately in relation to the different clinical diagnosis. We subsequently compared the AUROCs for the single core biomarkers in relation to the different clinical diagnosis, either with or without the addition of neurogranin. All statistical analyses were performed in the R statistical environment, version 3.2.3 (R Foundation for Statistical Computing, Vienna, Austria; https://www.R-project.org/). Two-tailed p values <0.05 were considered statistically significant. A correction for multiple testing was not performed because of the exploratory nature of the study.
RESULTS
CSF neurogranin concentrations according to clinical diagnoses
Table 1 summarizes the concentrations of CSF neurogranin and core AD biomarkers (Aβ42, t-tau, and p-tau) in the four study groups. CSF neurogranin concentrations were significantly higher in AD dementia patients than both HC (p = 0.004) and FTD patients (p = 0.004) (Fig. 1). No other significant intergroup differences were found.
Summary of demographic, clinical, and biomarker data
Data are presented as median values with 25th and 75th quartiles, except for n. For statistical comparisons, the above MMSE, neurogranin, Aβ42, t-tau, and p-tau comparisons were adjusted for age, sex, and site. Aβ42, 42-amino acid-long amyloid-β peptide; AD, Alzheimer’s disease; CSF, cerebrospinal fluid; HC, cognitively healthy controls; F, female; FTD, frontotemporal dementia; LP, lumbar puncture; M, male; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; p-tau, hyperphosphorylated tau; t-tau, total tau. ap < 0.05 versus HC; bp < 0.05 versus MCI; cp < 0.05 versus FTD; dp < 0.001 versus HC; ep < 0.001 versus MCI.
CSF neurogranin concentrations according to diagnostic categories. Boxplots showing CSF neurogranin concentrations (adjusted for sex, age, and site) in AD dementia patients, FTD patients, MCI subjects, and cognitively HC. The lower, upper, and middle lines correspond to the 25th centile, 75th centile, and median, respectively. The whiskers extend to the minimum and maximum neurogranin data points. Dark circles represent outliers. Group-wise comparisons of neurogranin values (adjusted for sex, age, and site) were conducted through nonparametric Kruskal-Wallis tests followed by pairwise comparison (Conover’s-test for multiple comparisons). ADD, Alzheimer’s disease dementia; CSF, cerebrospinal fluid; FTD, frontotemporal dementia; HC, healthy controls; MCI, mild cognitive impairment; p-tau, hyperphosphorylated tau; t-tau, total tau.
Associations of CSF neurogranin concentrations with core biomarkers
In the entire study cohort, CSF neurogranin concentrations were significantly correlated with p-tau (rho = 0.808, p < 0.001) and t-tau (rho = 0.830, p < 0.001) but not with Aβ42 (rho = –0.171, p = 0.078) after adjustment for age, sex, and site (Fig. 2). Multivariable linear regression analyses showed that neurogranin was independently associated with both p-tau (regression coefficient = 1.79, p = 0.0171) and t-tau concentrations (regression coefficient = 0.645, p < 0.001, respectively).
Scattergrams and regression lines showing the associations of CSF neurogranin with core biomarkers. Spearman rank-order correlations of CSF neurogranin with (A) Aβ42, (B) p-tau, (C) t-tau in the entire study cohort are shown. Aβ42, 42-amino acid-long amyloid-β peptide; ADD, Alzheimer’s disease dementia; CSF, cerebrospinal fluid; FTD, frontotemporal dementia; HC, healthy controls; MCI, mild cognitive impairment; p-tau, hyperphosphorylated tau; t-tau, total tau.
Accuracy of CSF neurogranin and core biomarkers for different clinical diagnoses
Table 2 summarizes the accuracy of CSF neurogranin and core biomarkers in discriminating HC versus the three diagnostic groups (MCI, AD dementia, and FTD). Particularly, the performance of CSF neurogranin in discriminating clinical AD dementia from HC was fair (AUROC = 0.72), whereas the ability in differentiating FTD from HC was poor (AUROC = 0.68). Finally, CSF neurogranin was unable to distinguish MCI from HC (AUROC = 0.54). Table 3 depicts the comparisons between the AUROCs for the core biomarkers according to the different clinical diagnoses, either with or without the addition of neurogranin. With regard to the HC versus MCI comparison, the differences between the AUROCs of p-tau versus p-tau + neurogranin and t-tau versus t-tau + neurogranin were significant (p = 0.029 and p = 0.010, respectively). As far as HC versus FTD were concerned, the differences between the AUROCs of p-tau versus neurogranin and t-tau versus neurogranin were significant (p = 0.002 in both cases); in addition, the AUROCs comparisons p-tau versus p-tau + neurogranin and t-tau versus t-tau + neurogranin were significant (p = 0.011 and p < 0.001, respectively). With regard to HC versus AD dementia, we found a p = 0.011 for the AUROCs comparison of t-tau versus neurogranin.
Accuracy of CSF neurogranin compared with the core biomarkers Aβ42, t-tau, and p-tau in discriminating MCI versus HC, AD dementia versus HC, and FTD versus HC
Aβ42, 42-amino acid-long amyloid-β peptide; ADD, Alzheimer’s disease dementia; AUROC, area under the receiver operating characteristic curve; CSF, cerebrospinal fluid; FTD, frontotemporal dementia; HC, cognitively healthy controls; MCI, mild cognitive impairment; p-tau, hyperphosphorylated tau; t-tau, total tau.
Summary of the results of the AUROC comparisons for the core biomarkers according to the different clinical diagnoses, either with or without the addition of neurogranin
Aβ42, 42-amino acid-long amyloid-β peptide; ADD, Alzheimer’s disease dementia; AUROC, area under the receiver operating characteristic curve; CSF, cerebrospinal fluid; FTD, frontotemporal dementia; HC, cognitively healthy controls; MCI, mild cognitive impairment; p-tau, hyperphosphorylated tau; t-tau, total tau.
DISCUSSION
The main results of our study indicate that: 1) CSF neurogranin concentrations are significantly higher in AD dementia patients compared with both HC subjects and FTD patients, indicating that increased CSF neurogranin concentrations may present a core pathophysiological feature related to AD; 2) CSF neurogranin was independently associated with both t-tau and p-tau proteins; and 3) CSF neurogranin was not superior to core AD biomarkers in differentiating HC subjects from the three diagnostic groups, neither it improved in a clinically significant manner their diagnostic accuracy.
Our data of increased CSF neurogranin concentrations in AD dementia patients are in agreement with previous reports [13, 32]. Although our study cannot provide pathophysiological insights, it seems plausible that increased CSF neurogranin concentrations originate from the extracellular release resulting from synaptic loss. Our data further support previously reported elevated CSF neurogranin concentrations in AD dementia compared with FTD patients [31]. However, our results on FTD should be interpreted with caution owing to the small size of included participants with the diagnosis (n = 9).
Another interesting observation is the independent and positive association of neurogranin with both p-tau and t-tau, which is line with recent findings reported by Mattsson and colleagues [33]. In this context, it seems plausible that the observed alterations in CSF neurogranin concentrations might reflect the occurrence of neurofibrillary degeneration. CSF neurogranin was not superior to core AD biomarkers in differentiating HC from the three diagnostic groups and it did not improve their diagnostic accuracy in a clinically significant manner. Accordingly, it seems currently fair to conclude that CSF neurogranin is unlikely to provide optimized diagnostically relevant information when added to the established validated AD CSF biomarker tests in clinical routine.
Our findings have a number of significant limitations. Owing to the cross-sectional nature of the study, it was not possible to differentiate stable-MCI subjects from those progressing and converting into dementia. Further studies are needed to confirm the potential value of neurogranin in predicting conversion from MCI to AD [17, 18]. Extensive psychometric data were not available in our study, thus preventing the study of CSF neurogranin concentrations in relation to different cognitive dimensions. Moreover, the quantification of core AD CSF biomarkers was not performed in a centralized laboratory (which seemed acceptable, since it reflects real life and given the established wide clinical use of these analyses worldwide) and, while we controlled for center effects in our statistical analysis, additional inter-laboratory variability cannot be completely ruled out [34, 35]. Another caveat of this study is the relatively modest convenience sample of patients and controls, which limits the generalizability of our conclusions. Moreover, we recognize that we did not perform an a priori power analysis because of the lack of sufficient data upon which to base estimates of variance. Another limitation of the present study is the use of an in-house ELISA assay which is not yet commercially available. Finally, the low number of FTD patients does not permit to derive definite statements in this subset of patients from our current analyses. Future independent validation of our data in larger cohorts of FTD cases is needed to confirm and expand our pilot data.
In conclusion, our cross-sectional study confirms and expands previous findings on the role of CSF neurogranin as a biomarker in different neurodegenerative diseases. Our data indicate that CSF concentration seems to be related to AD-characteristic pathophysiology. However, CSF neurogranin was not able to significantly improve the diagnostic accuracy in differentiating HC from MCI individuals, AD dementia, and FTD patients. Further studies are needed to investigate whether CSF neurogranin might be valuable as a predictor of progression and conversion to clinical milestones, such as MCI (in asymptomatic-at-risk individuals) and dementia (in mildly cognitive impaired subjects).
Footnotes
ACKNOWLEDGMENTS
This work was supported by the AXA Research Fund, the “Fondation Université Pierre et Marie Curie” and the “Fondation pour la Recherche sur Alzheimer”, Paris, France. Ce travail a bénéficié d’une aide de l’Etat “Investissements d’avenir” ANR-10-IAIHU-06 (Harald Hampel). The research leading to these results has received funding from the program “Investissements d’avenir” ANR-10-IAIHU-06 (Agence Nationale de la Recherche-10-IA Agence Institut Hospitalo-Universitaire-6) (Harald Hampel).
Henrik Zetterberg is a Wallenberg Academy Fellow. Kaj Blennow holds the Torsten Söderberg Professorship of Medicine.
