Synaptic, axonal damage and inflammatory cerebrospinal fluid biomarkers in neurodegenerative dementias

Alzheimer’s & Dementia - (2019) 1-12

Featured Article

Synaptic, axonal damage and inflammatory cerebrospinal fluid biomarkers in neurodegenerative dementias a  Anna Antonella,*, Adri
a Tort-Merinoa, Jose Rıosb,c, Mircea Balasaa, Sergi Borrego-Ecija , d a a a a Josep M. Auge , Cristina Mu~ noz-Garcıa , Beatriz Bosch , Neus Falg
as , Lorena Rami , a Oscar Ramos-Campoy , Kaj Blennowe,f, Henrik Zetterberge,f,g,h, Jose L. Molinuevoa, Albert Llad oa,1, Raquel Sanchez-Vallea,1 a

Alzheimer’s Disease and Other Cognitive Disorders Unit, Hospital Clınic, Fundacio Clınic per a la Recerca Biom
edica, Institut d’Investigacions Biom
ediques August Pi i Sunyer (IDIBAPS), Universitat de Barcelona, Barcelona, Spain b Medical Statistics Core Facility, Institut d’Investigacions Biom
ediques August Pi i Sunyer (IDIBAPS) and Hospital Clınic, Barcelona, Spain c Biostatistics Unit, Faculty of Medicine, Universitat Aut
onoma de Barcelona, Barcelona, Spain d Biochemistry and Molecular Genetics Department, Biomedical Diagnostic Center, Hospital Clınic, Barcelona, Spain e Department of Psychiatry and Neurochemistry, Institute of Neuroscience and Physiology, The Sahlgrenska Academy at the University of Gothenburg, M€olndal, Sweden f Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, M€olndal, Sweden g Department of Neurodegenerative Disease, University College London, London, UK h UK Dementia Research Institute at UCL, London, UK

Abstract

Introduction: Synaptic damage, axonal neurodegeneration, and neuroinflammation are common features in Alzheimer’s disease (AD), frontotemporal dementia (FTD), and Creutzfeldt-Jakob disease (CJD). Methods: Unicentric cohort of 353 participants included healthy control (HC) subjects, AD continuum stages, genetic AD and FTD, and FTD and CJD. We measured cerebrospinal fluid neurofilament light (NF-L), neurogranin (Ng), 14-3-3, and YKL-40 proteins. Results: Biomarkers showed differences in HC subjects versus AD, FTD, and CJD. Disease groups differed between them except AD versus FTD for YKL-40. Only NF-L differed between all stages within the AD continuum. AD and FTD symptomatic mutation carriers presented differences with respect to HC subjects. Applying the AT(N) system, 96% subjects were positive for neurodegeneration if 14-3-3 was used, 94% if NF-L was used, 62% if Ng was used, and 53% if YKL-40 was used. Discussion: Biomarkers of synapse and neurodegeneration differentiate HC subjects from neurodegenerative dementias and between AD, FTD, and CJD. NF-L and 14-3-3 performed similar to total tau when AT(N) system was applied. Ó 2019 the Alzheimer’s Association. Published by Elsevier Inc. All rights reserved.

Keywords:

Alzheimer’s disease; Frontotemporal dementia; Creutzfeldt-Jakob disease; Biomarker; Cerebrospinal fluid; Neurofilament light; Neurogranin; YKL-40; 14-3-3; Mutation carriers; AT(N) system; Preclinical AD; MCI due to AD

1. Introduction

1

These authors share senior authorship. *Corresponding author. Tel.: 134-932275400 x 4814; Fax: 134932275783. E-mail address: [email protected]

Alzheimer’s disease (AD) is a heterogeneous and pathophysiologically complex disease, polygenic in most cases (sporadic AD) with a minority of monogenic phenotypes (,0.5%, autosomal dominant AD, ADAD) [1]. Early diagnosis and differential diagnosis with other

https://doi.org/10.1016/j.jalz.2019.09.001 1552-5260/Ó 2019 the Alzheimer’s Association. Published by Elsevier Inc. All rights reserved.

2

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

neurodegenerative diseases are crucial for patient management and for the accurate use of future treatments. Cerebrospinal fluid (CSF) amyloid beta isoform 42 (Ab42), total tau (t-tau) protein, and tau protein phosphorylated at Thr-181 (ptau) have been accepted as core biomarkers for detecting AD neuropathologic features in living individuals [2]. CSF Ab42 is considered indicative of amyloid pathology, p-tau of tangle pathology, and t-tau of neurodegeneration. Previous guidelines for AD [3,4] have defined the preclinical AD stage (which includes subjects with normal cognition and abnormal amyloid and tau markers) and the mild cognitive impairment (MCI) stage that represents the early, symptomatic, predementia phase of AD, characterized by an episodic memory deficit and pathologic levels of core AD biomarkers (MCI due to AD). The new National Institute on Aging-Alzheimer’s Association (NIA-AA) Research Framework has proposed a biological definition of AD, using the AT(N) system (amyloid, tau, neurodegeneration), where (N) is open to new biomarkers if supported by the available evidence [5]. Subjects are classified according to their biomarker profile within the Alzheimer’s continuum: Ab biomarkers (A) determine whether an individual is in the Alzheimer’s continuum and pathologic tau biomarkers (T) determine if someone who is in the Alzheimer’s continuum has AD. Core AD biomarkers have demonstrated high diagnostic validity to differentiate AD from healthy control (HC) subjects, but with little added value for prognosis or disease severity staging [6]. Core AD biomarkers have also demonstrated good global accuracy in the differential diagnosis of AD with other neurodegenerative dementias, like frontotemporal dementia (FTD) or Creutzfeldt-Jakob disease (CJD) [7,8]. Other CSF biomarkers are being studied in AD and other neurodegenerative dementias. Neurofilament light (NF-L) is a major cytoskeletal constituent of neuronal cells and can be released into the CSF, detecting significant increases in different neurologic disorders [9–17]. Neurogranin (Ng) is a calmodulin-binding postsynaptic neuronal protein that is abundantly expressed in perikaryal and dendritic cytoplasm [18], involved in modulating synaptic transmission and plasticity mechanisms [19–23]. The 14-3-3 proteins are highly expressed in the brain and some of the 14-3-3 isoforms are particularly enriched in the presynapses, to regulate transmission and plasticity [24]. The 14-3-3 g protein presents increased levels in the CSF of sporadic CJD and is accepted as part of the current diagnostic criteria [25–27]. YKL-40 is a glycoprotein produced by inflammatory cells, mostly astrocytes. Its physiological role is not completely understood but YKL-40 is increased in the brain and CSF in several neurologic and neurodegenerative diseases associated with inflammatory processes [15,28–30]. In this study, we aimed to analyze and compare the diagnostic accuracy and discriminatory properties of four noncore biomarkers that reflect different aspects of the neurodegeneration process: axonal damage, presynapsis and postsynapsis loss, and neuroinflammation (NF-L, 143-3, Ng and YKL-40, respectively) in an unicentric cohort

with a broad spectrum of clinical diagnoses: the Alzheimer’s continuum (including preclinical phase and genetic cases), FTD (sporadic and genetic), and sporadic CJD. In addition, we investigated the performance of each of these markers as a potential neurodegenerative (N) marker in the recently proposed AT(N) system in AD. 2. Materials and methods 2.1. Participants Unicentric cohort of 353 participants was recruited at the Alzheimer’s Disease and Other Cognitive Disorders Unit, Hospital Clınic, Barcelona, Spain: HC (n 5 50), asymptomatic subjects within the AD continuum (Preclinical AD) (n 5 21), MCI due to AD (n 5 56), AD dementia (n 5 108), sporadic FTD (n 5 34), CJD (n 5 38), ADAD cases (n 5 25, nine asymptomatic, ADaMC/16 symptomatic, ADsMC), and FTD genetic cases (n 5 21, five asymptomatic, FTDaMC/16 symptomatic, FTDsMC). In some of the analysis, sporadic FTD and FTDsMC were pooled together (FTD, n 5 50). All the participants underwent a complete clinical and neuropsychological examination. Longitudinal follow-up consisted in an annual clinical assessment and Mini-Mental State Examination (MMSE). HC subjects had normal cognition defined according to the following criteria: MMSE scores greater than 24, objective cognitive performance within the normal range in all tests from a specific test battery, no significant psychiatric symptoms or previous neurologic disease, and normal CSF core AD biomarkers. All clinical diagnoses were made in accordance with standard criteria [31–34]. Some of the Preclinical AD subjects had only Ab42 biomarker altered, being Alzheimer’s pathologic change biomarker category according to the new Research Criteria for AD [5], and some had Ab42 and tau biomarkers altered, classified as AD biomarker profile. Core AD biomarkers were used as selection criteria with the following own cutoffs: for Ab42 550 pg/mL (CSF samples measured before February 2016) and 750 pg/mL (CSF samples measured after February 2016), for t-tau 385 pg/mL and for p-tau 65 pg/mL. All the genetic participants were recruited through the Information and Counselling Program for Genetic dementias (PICOGEN program) at the Hospital Clınic, Barcelona, Spain [35]. Participants were adult children from families with a known mutation in one of the following genes: PSEN1 or APP gene (for ADAD) and GRN, MAPT, or C9orf72 (for FTD genetic cases). The study was approved by the Ethics Committee of Hospital Clınic of Barcelona. Written informed consent was obtained from all participants. 2.2. CSF collection and protein level determination CSF samples were collected from all participants and centrifuged at 2000g for 10 minutes at 4 C and stored at 280 C in polypropylene tubes until usage.

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

3

Core AD biomarker concentrations were measured with INNOTEST Enzyme-Linked ImmunoSorbent Assays (ELISAs) following manufacturer’s instructions (Fujirebio, Ghent, Belgium). CSF NF-L concentration was measured using the ELISA kit of UmanDiagnostics distributed by IBL International (Hamburg, Germany) [36]. For g-14-3-3 protein, we used the ELISA kit CircuLex 14-3-3 gamma (MBL International Corporation, Woburn, MA, USA) with a CSF sample dilution of 1:5 [37]. CSF YKL-40 concentration was measured with an ELISA from QUIDEL (San Diego, CA, USA), using a CSF sample dilution of 1:2.5. Ng was measured using an in-house ELISA assay based on the monoclonal antibody Ng7 as described previously [19]. The antibodies for the detection of these four noncore biomarkers have been used by us and other authors on previous studies with CSF samples in neurodegenerative dementias. These biomarkers have been also individually studied using other technologies or antibodies [37–40]. All analyses were performed by duplicate and experienced laboratory personnel blinded to clinical diagnosis. We are participants of the Alzheimer’s Association QC program [41], and Ab42, t-tau, and p-tau levels obtained in our laboratory have consistently been within mean 6 2SD. Because of the limited availability of the CSF sample, not all the biomarkers could be studied in all the samples (see Tables 1 and 2).

estimation of area under the curve (AUC) from receiver operating characteristic (ROC) curves. An approach to cutoff for each biomarker was assessed by likelihood ratio positive (LR1) 5 (sensitivity/1 2 specificity) from ROC analysis [42]. For some of these analyses, a group of patients with neurodegenerative diseases was created, which included patients with MCI due to AD, AD dementia, FTD, and CJD. All analyses were performed using SPSS software, version 25.0. (IBM Corp, Armonk, NY) and were performed using a two-sided type I error of 5%. Because of methodological characteristics of this study the P values presented were nominal and were not adjusted for multiplicity.

2.3. APOE genotyping

3. Results

APOE genotype was determined through the analysis of two single nucleotide polymorphisms: rs429358 and rs7412, either using TaqMan genotyping technologies (Thermo Fisher Scientific, Waltham, MA, USA) or alternatively by Sanger sequencing.

3.1. Group characteristics

2.4. Statistical analysis All data are described using median and 25th, 75th percentiles, or absolute frequency and percentages for quantitative and qualitative variables, respectively. Estimation outcomes from statistical models described subsequently were assessed by 95% confidence intervals (95% CIs). Baseline comparisons were made using Mann-Whitney U test or Fisher’s exact test and generalized lineal models for adjusted comparisons by age and gender. Generalized estimating equation (GEE) models, using an autoregressive model (AR(1)) approach to assess intraindividual correlation matrix, were used for longitudinal analyses with baseline result of MMSE as covariate with the aim of evaluating differences between diagnosis groups with or without adjustment for each biomarker. In addition, these GEE models were used to assess the evolution of MMSE in each group. Generalized lineal models and GEE models were performed by nonparametrical approach by mean rank-transformation of dependent variable. Accuracy for prediction of biomarkers between control subjects and neurodegenerative diseases was performed by

2.5. Performance of neurodegeneration biomarkers in the AT(N) system Participants within the AD biological continuum with both altered Ab42 and p-tau (thus, AD according to the NIA-AA Research Framework) were analyzed to evaluate the effect of the different neurodegeneration biomarkers on the AT(N) classification system. Biomarker results were dichotomized as positive or negative according to the established cutoff best discriminating between HC subjects and AD dementia in the present work. The percentage of positivity in CSF of the five potential biomarkers of neurodegeneration (t-tau, 14-3-3, NF-L, Ng and YKL-40) was calculated.

The demographics and biomarker results are given in Tables 1 and 2 and Supplementary Table 1. There were no significant differences in gender proportions between groups with the exception of HC and FTD comparison (male predominance in FTD). When comparing all the participants pooled, we found statistical significant differences in NF-L between genders (male . female, P value .002) and in Ng (male , female, P value .015). These differences were not observed when we restricted the analysis to the HC group. Age was significantly different between all the comparisons except for “Preclinical AD versus MCI due to AD” and “AD dementia versus CJD.” 3.2. Group comparisons CSF biomarker results were compared between HC and each group of neurodegenerative dementias studied (AD dementia, FTD and CJD), presenting in all of them statistically significant differences with respect to control subjects. When different clinical groups were compared, AD dementia and FTD differed in all the biomarkers except for YKL-40, and CJD differed from both AD dementia and FTD in NF-L and Ng (Table 3; Fig. 1A). The groups could be sorted according to their concentration, from lower to higher values for each noncore biomarker: NF-L, HC , AD dementia , FTD , CJD; Ng, FTD , HC , AD

4

Table 1 Demographic, clinical characteristics, and CSF biomarkers levels Control subjects

Preclinical AD

MCI due to AD

AD dementia

FTD

CJD

Sex, n (F/M) Age at LP (y, IQR) APOEε4 2/1 (ε41 %) Basal MMSE Ab42 (pg/mL) T-tau (pg/mL) P-tau (pg/mL) NF-L (pg/mL) YKL-40 (ng/mL) Ng (pg/mL)

50

21 (14/7)

56 (36/20)

108 (68/40)

50 (24/26)

38 (20/18)

69.4 (62.9; 75.2)

67.0 (62.8; 74.6)

63.6 (57.7; 68.4)

62.9 (55.7; 67.2)

68.0 (61.0; 72.0)

11/10 (47.6)

19/36 (65.5)

52/53 (50.5)

40/10 (20.0)

6/0 (0)

28.0 [27.0; 29.0]

26.0 [24.0; 27.0]

21.50 [18.0; 25.0]

24.00 (37) [21.0; 26.0]

NA

416.00 [313.50; 482.35]

385.66 [337.00; 503.27]

372.98 [311.90; 446.47]

683.00 [544.70; 867.45]

NE

258.00 [147.05; 482.00]

719.64 [565.08; 940.95]

765.11 [543.13; 1060.65]

276.45 [208.72; 400.00]

61.30 [37.50; 86.00]

108.43 [93.49; 122.74]

98.46 [79.58; 134.44]

45.10 [34.00; 57.10]

3802.64 (28) [1303.17; 8631.83] NE

534.65 [428.32; 942.68]

942.68 [743.53; 1129.08]

1020.71 [815.28; 1273.10]

2469.35 [1439.09; 3942.00]

325.14 [237.24; 363.73]

347.24 [257.71; 397.06]

305.12 [247.33; 396.56]

315.28 [213.31; 387.28]

159.10 [104.90; 224.10]

250.00 [216.64; 320.08]

252.40 [189.80; 314.20]

135.30 [101.70; 170.80]

NE

5096.34 [4443.38; 5842.29]

5187.65 (45) [3973.76; 5963.26]

3833.63 [2968.15; 4758.52]

14-3-3 (AU)

(34/16) 57.0 (48.1; 65.4) 38/12 (24.0) 29.00 [28.0; 30.0] 789.25 [676.93; 984.28] 233.05 [167.57; 267.00] 49.64 [40.53; 57.10] 410.22 [312.40; 496.48] 217.23 [153.86; 271.37] 173.00 [132.30; 210.62] 2595.64 (45) [2306.13; 3010.51]

5045.66 [3266.22; 7114.03] NE 587.84 [335.50; 1079.17] NE

NOTE. The CSF biomarker level values presented are medians [IQR]. In parenthesis next to the median biomarker value the N evaluated when this is different from the N of the entire group. Abbreviations: CSF, cerebrospinal fluid; FTD, frontotemporal dementia; CJD, Creutzfeldt-Jakob disease; LP, lumbar puncture; NA, not available; NE, not evaluated; Ab42, amyloid b isoform 42; t-tau, total tau protein; p-tau, phosphorylated tau at Threonine 181; NF-L, neurofilament light; Ng, neurogranin; AU, arbitrary units; IQR, interquartile range; MCI, mild cognitive impairment.

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

Variable

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

5

Table 2 Demographic, clinical characteristics, and CSF biomarkers levels of subjects with ADAD and genetic FTD

Sex, n (F/M) Age at LP (y, IQR) APOEε4 2/1 (ε41 %) Basal MMSE Ab42 (pg/mL) T-tau (pg/mL) P-tau (pg/mL) NF-L (pg/mL) YKL-40 (ng/mL) Ng (pg/mL) 14-3-3 (AU)

ADaMC

ADsMC

FTDaMC

FTDsMC

9 (3/6) 37.4 [33.5; 45.0] 8/1 (11.1) 29.0 [29.0; 30.0] 788.00 [609.32; 1268.20] 233.10 [195.50; 297.50] 51.25 [42.75; 58.03] 288.14 [255.92; 307.75] 148.34 [148.34; 167.18] 147.50 [135.80; 208.95] 2680.25 [2381.05; 3009.20]

16 (6/10) 49.6 [45.0; 57.1] 15/1 (6.3) 20.5 [19.0; 24.0] 374.61 [199.70; 460.00] 849.50 [464.00; 1532.30] 117.67 [72.39; 172.20] 1008.96 [896.97; 1278.31] NE 271.55 [142.00; 543.85] 4805.95 [3126.18; 8383.87]

5 (3/2) 40.9 [34.4; 58.1] 2/2 (50.0) 30.0 [29.0; 30.0] 881.50 [842.50; 1205.50] 203.00 [131.00; 301.00] 49.50 [31.50; 60.00] 545.19 [388.03; 565.78] 289.96 [140.20; 374.41] NE 2067.05 [1759.61; 2378.79]

16 (6/10) 59.6 [53.7; 65.5] 10/6 (37.5) 24.0 [23.0; 26.0] 719.33 [567.70; 1048.28] 414.00 [211.42; 605.95] 45.55 [38.00; 58.90] 3770.36 [1674.79; 5715.4] 317.77 [168.82; 380.97] 151.90 [113.10; 186.90] 4297.20 [3598.81; 5379.76]

NOTE. The CSF biomarker level values presented are medians [IQR]. Abbreviations: AD, Alzheimer’s disease; ADAD, autosomal dominant AD; aMC, asymptomatic mutation carrier; CSF, cerebrospinal fluid; FTD, frontotemporal dementia; NE, not evaluated; Ab42, amyloid beta isoform 42; t-tau, total tau protein; p-tau, phosphorylated tau at Threonine 181; NF-L, neurofilament light; Ng, neurogranin; IQR, interquartile range; MMSE, Mini-Mental State Examination; sMC, symptomatic mutation carrier.

dementia , CJD; YKL-40, HC , AD dementia 5 FTD; and 14-3-3, HC , FTD , AD dementia (Table 1). Group comparisons were also performed in the groups within the AD continuum: HC, Preclinical AD, MCI due to AD, and AD dementia. NF-L levels differed between all the stages. In contrast, the other noncore biomarkers analyzed YKL-40, 14-3-3, and Ng, although they were different between HC and MCI due to AD and HC and AD dementia, and showed no differences between HC and Preclinical AD or between MCI due to AD and AD dementia (Table 4; Fig. 1B). Pairwise comparisons in genetic AD and FTD cases showed that NF-L and 14-3-3 differed between HC and symptomatic mutation carriers (sMCs), with no differences with respect to asymptomatic mutation carriers (aMC). Ng levels differed in the comparison between HC and ADsMCs. No differences were found in YKL-40 levels in genetic FTD (ADsMC not analyzed). There were significant differences in all four biomarkers between aMC and sMC in both diseases. Patients with sporadic FTD were divided in three subtypes for pairwise comparisons: behavioral variant FTD (bvFTD, n 5 8), nonfluent variant primary progressive aphasia (nfvPPA, n 5 11), and semantic variant PPA (svPPA, n 5 15) (see Supplementary Table 1 for demographics). No significant differences were found in the comparisons between bvFTD, svPPA, and nfvPPA subtypes for any of the four biomarkers. When comparing HC with every subtype, significant comparisons were as follows: NF-L in all the three comparisons, Ng between HC and svPPA, and 14-3-3 biomarker between HC and nfvPPA and between HC and bvFTD. 3.3. ROC analyses for differential diagnosis of HC and neurodegenerative diseases All the patients with a neurodegenerative disease in our cohort (MCI due to AD, AD dementia, FTD, and CJD) were grouped to calculate a cutoff to differentiate them from HC (Table 5).

An LR1 value .10, that would indicate a good validity of the test, was obtained for all the biomarkers except for YKL-40. The best cutoff for NF-L was .608 pg/mL, with an AUC of 0.97 (95% CI, 0.94; 0.99) and a sensitivity of 96% and specificity of 92%. The other biomarkers showed a slightly lower AUC, with similar specificities but lower sensitivities. 14-3-3 cutoff of .3598 AU/mL had an AUC of 0.91 (95% CI, 0.86; 0.95), with a sensitivity of 77.5% and a specificity of 93.3%. Ng cutoff was .250 and YKL40 cutoff was .545 (data shown in Table 5). The same ROC analyses were also performed comparing HC versus AD dementia, HC versus FTD, and HC versus CJD. For NF-L the same cutoff worked for the different Table 3 Summary of pairwise comparison of P values. Comparisons among HC subjects and neurodegenerative diseases HC NF-L HC AD FTD CJD Ng HC AD FTD CJD YKL-40 HC AD FTD CJD 14-3-3 HC AD FTD CJD

,.001 ,.001 ,.001 ,.001 .024 ,.001

AD dementia

FTD

CJD

,.001

,.001 ,.001

,.001 ,.001 .030

,.001 ,.001 ,.001 ,.001 ,.001 .010

.010 .019 NE

,.001 ,.001 NE

.945 NE ,.001 ,.001 NE

.030 .024 ,.001 ,.001 .019 .945

,.001 ,.001 ,.001

NE NE NE

NE ,.001 ,.001

NE NE NE

NE

Abbreviations: AD, Alzheimer’s disease; CJD, Creutzfeldt-Jakob disease; FTD, frontotemporal dementia; HC, healthy control; NE, not evaluated; NF-L, neurofilament light; Ng, neurogranin.

6

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

Fig. 1. Pairwise comparisons of NF-L, Ng, 14-3-3, and YKL-40 CSF biomarkers. Scatterplots displaying CSF NF-L, Ng, 14-3-3, and YKL-40 concentrations. (A) In neurodegenerative diseases Ctrl, n 5 50; AD dementia, n 5 108; FTD, n 5 50; and CJD, n 5 38. (B) In AD continuum Ctrl, n 5 50; Preclinical AD, n 5 21; MCI due to AD, n 5 56; and AD dementia, n 5 108. The middle line shows the median. The lower and upper lines correspond to interquartile range. The differences between the groups were assessed using generalized lineal models adjusting for age and gender. A P value ,.05 was considered statistically significant. *P , .05, **P , .01, ***P , .001. Abbreviations: NF-L, neurofilament light; Ng, neurogranin; 14-3-3, 14-3-3 g; Ctrl, healthy control subjects; AD, Alzheimer’s disease dementia; CSF, cerebrospinal fluid; MCI, MCI due to AD; FTD, frontotemporal dementia; CJD, Creutzfeldt-Jackob disease.

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12 Table 4 Summary of pairwise comparison of P values. Comparisons among HC subjects and AD continuum HC NF-L HC Preclinical AD MCI AD dementia Ng HC Preclinical AD MCI AD dementia YKL-40 HC Preclinical AD MCI AD dementia 14-3-3 HC Preclinical AD MCI AD dementia

Preclinical AD .003

.003 ,.001 ,.001

,.001 ,.001 .908

.908 .025 ,.001

,.001 ,.001 .726

.726 .025 .010

.131 .109 NE

NE ,.001 ,.001

NE NE

MCI due to AD ,.001 ,.001

AD dementia ,.001 ,.001 .008

.008 .025 ,.001

,.001 ,.001 .321

.321 .025 .131

.010 .109 .980

.980 ,.001 NE

,.001 NE .557

.557

Abbreviations: AD, Alzheimer’s disease; HC, healthy control; MCI, mild cognitive impairment; NE, not evaluated; NF-L, neurofilament light; Ng, neurogranin.

disease groups (.570 pg/mL), and for 14-3-3 a slightly different cutoffs but in the same direction were obtained. For Ng different cutoffs were necessary as patients with FTD presented lower Ng levels with respect to HC, and AD dementia and CJD higher levels. For YKL-40, a cutoff was obtained for FTD although with a not very good LR1, sensitivity, and specificity, and for AD dementia it was not possible to calculate a cutoff (Table 6; Fig. 2). 3.4. Performance of neurodegeneration biomarkers using the AT(N) system We used the five potential neurodegeneration (N) CSF biomarkers to evaluate neuronal damage in AD subgroup (A1 T1). All these subjects had clinical diagnoses falling within the AD continuum (Preclinical AD, MCI due to AD, and AD dementia, including ADsMC). Percentage of (N) positive (N1) classified subjects for each marker for t-tau was 98%, for 14-3-3 was 96%, and for NF-L was 94%. In contrast, only 62% and 53% subjects would be classified as N1 if Ng and YKL-40 were used, respectively. Afterward, the percentage of positivity for each marker of neurodegeneration within the different clinical diagnoses of the AD subgroup (A1 T1) was analyzed. T-tau was positive in 87.5% of Preclinical AD, 100% of MCI due to AD, and 97.9% of individuals with AD dementia. NF-L positivity was 69.5% in Preclinical AD, 89.8% in MCI due to AD, and 97.9% in AD dementia. 14-3-3 was positive in 97.3% of MCI due to AD and 94.7% of AD dementia clinical diag-

7

noses (no data available for Preclinical AD). Ng was positive in 37.5% of Preclinical AD, 67.3% of MCI due to AD, and 61.4% of AD dementia subjects. YKL-40 was positive in 100% of subjects (eight subjects) with Preclinical AD, 56.5% of MCI due to AD, and only 47.3% of AD dementia diagnoses. 3.5. Model for MMSE inference We assessed a model for the longitudinal MMSE value to be predicted by the group, baseline MMSE, follow-up time from baseline, and interaction between group and followup time. We added biomarker levels separately to check the effect of the biomarker for the longitudinal prediction of MMSE value. For intragroup comparisons, we did not find differences in statistical significance between the crude model (without biomarker) and the other models adding biomarkers. For intergroup comparisons, changes in statistical significance were obtained in the following comparisons: AD versus FTD at baseline (Ng, 14-3-3, Ab42, t-tau, and p-tau biomarkers separately) and HC versus Preclinical AD at baseline (Ab42 biomarker).

4. Discussion In this study, biomarkers of synapse loss, axonal damage, and astroglial inflammation differentiate HC from neurodegenerative dementias (AD dementia, FTD, and CJD). When assessing their utility in the differential diagnosis between neurodegenerative dementias, NF-L and Ng showed statistically significant differences between all three groups (AD dementia, FTD, and CJD), and 14-3-3 between AD dementia and FTD (CJD not analyzed). YKL-40 showed no differences between AD dementia and FTD. These differences, albeit statistically significant, have to be interpreted with caution because of the existing overlap between groups. In concordance to previously published data, we have found an increase in NF-L in all types of neurodegenerative dementia evaluated, suggesting that CSF NF-L is a non–disease-specific biomarker [10,14]. In agreement with previous studies, CSF NF-L was higher in FTD than in AD [16,17]. The analysis of the biomarkers studied within the AD continuum shows that NF-L is the only biomarker that tracks disease progression and does not reach a plateau in the clinical phases of AD (MCI due to AD and AD dementia) and differentiates severity stages within the AD continuum, including HC versus Preclinical AD. Results obtained in the different subtypes of FTD are similar to those published before, with higher CSF NF-L levels in bvFTD, nfvPPA, and svPPA in comparison with AD dementia and HC [10,11]. In the ROC analysis comparing HC and neurodegenerative diseases, NF-L showed the higher AUC, with good specificity and sensitivity. NF-L levels also performed with high accuracy discriminating HC from AD dementia, FTD, and CJD.

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

8

Table 5 ROC analysis between healthy control subjects and neurodegenerative disease group (MCI 1 AD 1 FTD 1 CJD) Variable

Group

Median (IQR)

Valid N

AUC (95% CI)

Ab42

Control Neurodegenerative disease Control Neurodegenerative disease Control Neurodegenerative disease Control Neurodegenerative disease Control Neurodegenerative disease Control Neurodegenerative disease Control Neurodegenerative disease

789.25 [676.93; 984.28] 424.50 [340.13; 544.85]

50 216

0.907 (0.873; 0.942)

233.05 [167.57; 267.00] 641.37 [431.00; 1025.00]

50 241

49.64 [40.53; 57.10] 92.41 [65.84; 118.28]

T-tau

P-tau

NF-L

Ng

14-3-3

YKL-40

Positive if

Sensitivity

Specificity

LR(1)

597

0.824

0.936

12.9

0.923 (0.893; 0.953)

.327

0.859

0.920

10.7

50 216

0.842 (0.796; 0.888)

.71

0.713

0.940

11.9

410.22 [312.40; 496.48] 1188.63 [868.44; 2469.35]

50 252

0.967 (0.941; 0.993)

.608

0.960

0.920

12

173.00 [132.30; 210.62] 244.00 [170.20; 330.60]

47 239

0.729 (0.669; 0.790)

.250

0.477

0.957

11.1

2595.64 [2306.13; 3010.51] 4534.41 [3766.19; 5628.63]

45 138

0.905 (0.860; 0.950)

.3598

0.775

0.933

11.6

217.23 [153.86; 271.37] 310.71 [241.25; 394.79]

47 208

0.749 (0.669; 0.829)

.545

0.067

0.979

3

Abbreviations: ROC, receiver operating characteristics; AUC, area under the curve; LR, Likelihood ratio; MCI, mild cognitive impairment due to AD; AD, Alzheimer’s disease dementia; FTD, frontotemporal dementia; CJD, Creutzfeldt-Jakob disease; Ab42, amyloid beta isoform 42; t-tau, total tau protein; p-tau, phosphorylated tau at Threonine 181; NF-L, neurofilament light; Ng, neurogranin, IQR, interquartile range; CI, confidence interval.

An increase in CSF Ng concentration was reported to be specific for AD [19–21,23]; however, our results show that patients with CJD also have increased CSF Ng levels. On the other hand, we observed a decrease in Ng levels in patients with FTD, as reported previously [23]. When analyzing the different FTD subtypes, the only statistically significant comparison compared with HC was the lower levels in svPPA and, with respect to AD dementia, all the FTD subtypes presented statistically significant lower levels. These data are different from previously published data showing that AD presented no differences with respect to semantic variant FTD [21]. In the ROC analysis of HC versus

neurodegenerative diseases, Ng was also a good predictor, with a good specificity but a poor sensitivity. A possible reason for this fact in Ng could be explained by the inverse behavior of Ng in AD dementia and FTD compared with HC (higher in AD dementia and lower in FTD). The detection of 14-3-3 protein levels was performed using an optimized commercial ELISA for the detection of 14-3-3 g protein [37]. There is only one previous study that describes 14-3-3 levels in AD and HC with the same method, but with a smaller sample size and without comparison to an HC group [26]. To our best knowledge, this is the first study of 14-3-3 levels in patients with FTD, and FTD presented

Table 6 ROC analysis between healthy control subjects and AD, FTD, and CJD Variable

Disease

AUC (95% CI)

Positive if

Sensitivity

Specificity

LR(1)

NF-L

AD FTD CJD AD FTD CJD AD

0.98 (0.94; 1) 0.97 (0.93; 1) 0.998 (0.99; 1) 0.85 (0.77; 0.93) 0.69 (0.58; 0.8) 0.86 (0.75; 0.96) 0.94 (0.88; 1)

FTD CJD AD FTD CJD

0.80 (0.71; 0.89) NC 0.69 (0.58; 0.81) 0.69 (0.58; 0.81) NC

.570 .570 .570 .230 ,83 .230 .3210 .3400 .3410

1.00 0.96 1.00 0.64 0.11 0.83 0.90 0.90 0.63

0.80 0.81 0.81 0.88 0.98 0.85 0.83 0.88 0.88

5.13 5.02 5.22 5.26 5.11 5.56 5.26 7.36 5.30

NC .310

0.54

0.83

3.26

Ng

14-3-3

YKL-40

Abbreviations: ROC, receiver operating characteristics; AUC, area under the curve; LR, likelihood ratio; AD, Alzheimer’s disease dementia; FTD, frontotemporal dementia; CJD, Creutzfeldt-Jakob disease; Ab42, amyloid beta isoform 42; t-tau, total tau protein; p-tau, phosphorylated tau at Threonine 181; NF-L, neurofilament light; Ng, neurogranin; CI, confidence interval; NC, not calculable, because of LR(1) , 3.

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

9

Fig. 2. ROC curves of CSF biomarkers NF-L, Ng, 14-3-3, and YKL-40, differentiating healthy control subjects from (A) Alzheimer’s disease dementia, (B) Creutzfeldt-Jackob disease, and (C) frontotemporal dementia. Abbreviations: CSF, cerebrospinal fluid; NF-L, neurofilament light; Ng, neurogranin; ROC, receiver operating characteristics.

significant differences in 14-3-3 levels compared with HC and AD dementia. In the ROC analysis, comparing HC and neurodegenerative diseases, 14-3-3 resulted to be also a good predictor, with a good specificity but a slightly worse sensitivity than NF-L. For the discriminating capacity of HC-AD dementia and HC-FTD, with a very similar cutoff, 14-3-3 showed slightly better AUC and sensitivity for AD dementia than FTD. The opposite behavior of 14-3-3 and Ng levels in FTD, both considered markers of synaptic loss (although 14-3-3 is considered mainly presynaptic and Ng postsynaptic), is intriguing and would deserve further studies. With respect to YKL-40, in this study, we have significantly increased the sample published in a previous work [28], confirming results and providing new data about patients with FTD. Increased levels of YKL-40 in AD and FTD have been previously reported [29,30], although no differences between AD and FTD were observed [30]. In the ROC analysis, on comparing HC and neurodegenerative diseases, we found a low predictive capacity of YKL40, and it

was not possible to calculate a cutoff in the comparison between HC and AD dementia because of the high variability of the biomarker values in HC. In the analyses of genetic cases of AD and FTD, aMC did not show differences in any of the markers studied with respect to HC. However, the lack of differences between HC and aMC may be because of the limited sample size. When comparing the two groups of mutation carriers (asymptomatic vs. symptomatic) we confirmed previous results described in genetic FTD for NF-L [12]. A model for the prediction of longitudinal MMSE values in the AD continuum and FTD patients was assessed. In most of the cases, results indicated that the biomarkers do not explain the evolution of MMSE better than the group classification itself. Other studies have tried to use these noncore AD biomarkers, mainly Ng, to predict future cognitive impairment or cognitive decline [19,20]. Here, we found an effect in MMSE inference when comparing AD and FTD for Ng, 14-3-3, Ab42, t-tau, and p-tau CSF biomarkers, indicating that these biomarkers can explain differences in

10

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

evolution of MMSE between patients with AD and FTD. On comparing HC with Preclinical AD, Ab42 is related with MMSE and group classification, which could be explained by the fact that Ab42 has been used to classify subjects in this group. To date, few studies have reported results on multi-CSF biomarker results (core AD biomarkers, NF-L, Ng, 14-3-3, and YKL-40) within the same cohort, comparing biomarkers levels among groups and sorting the magnitude of the differences in each disease [43,44]. In this study, we have also analyzed the performance of each biomarker in the application of the AT(N) biological definition of AD from the NIA-AA [5]. Our data suggest a relatively similar performance of t-tau, 14-3-3, and NF-L as neurodegeneration (N) CSF biomarkers within the AD participants, pointing out the possibility to use either of these three biomarkers for AT(N) classification purposes. In contrast, YKL-40 and Ng showed worse results, suggesting that these biomarkers differed from the others for this purpose. Analyzing the dynamics of these biomarkers along the AD clinical continuum different patterns emerged: increase in positivity from preclinical to dementia phase (NF-L), increase in positivity in the transition from preclinical to symptomatic AD with little differences once the symptoms have appeared (Ng), or decrease in the percentage of positivity from preclinical to symptomatic phases (YKL40). In the NIA-AA Research Framework [5], the diagnosis of AD in living individuals is based on biomarkers, shifting toward a biological construct of AD. This proposal organizes biomarkers by the measurement of distinct pathologic processes. Although A represents Ab deposition and T pathologic tau deposition, N represents neurodegeneration in the AT(N) system. However, neurodegeneration in AD is a complex process with many contributing factors and complex interactions. In this sense, different biomarkers could be included as “N” markers. As the Editorial accompanying the NIA-AA Research Framework stated [45] “At present, the major barriers limiting advances in this arena are due to the limitations of present knowledge about the neurobiology of Alzheimer’s disease as a neuropathologic entity and its temporal relationship to cognitive/behavioral impairment.” As previously mentioned, we investigate the frequency of the positivity for different markers for neurodegeneration in the preclinical and clinical stages of AD, considering that each of these markers reflects a distinct neurodegenerative process and pathologic change involved in AD. As this is a cross-sectional study, we can only infer the temporal relationship of the positivity of each biomarker to the onset of symptoms and to the positivity of other biomarker based on frequencies. Future longitudinal studies analyzing the positivity of each biomarker at each time point would be needed to evaluate their temporal relationship. On the other hand, in our study, as in most of previous studies, AD is considered a homogeneous disease in terms of biomarkers’ behavior. However, we cannot exclude that

the predominant neurodegenerative processes vary in each individual. In this sense, different neurodegenerative (N) biomarkers could identify subtypes of patients with AD (related to a genetic predisposition, linked to risk factors or both), for example, Ng or YKL-40 positivity could identify patients with marked synaptic loss or astroglial neurodegeneration, and thus, better candidates for synaptic neuroprotective therapies or anti-inflammatory therapies, respectively. In this respect, clinicopathologic studies in patients with different biomarker patterns will be needed to determine if the different percentages of the positivity for N biomarkers reflect differences in the brain of the subjects or if it is just only caused by a difference in sensibility of each biomarker for the detection of the neurodegenerative process in AD. Any of the N biomarkers predicted the longitudinal MMSE; however, it is possible that the clinical progression of the disease could be better explained by the longitudinal change in a concrete biomarker than the basal level [46]. In this sense, again, longitudinal data in repeated samples could provide further evidence of the relationship of biomarkers and cognitive impairment. As a limitation for the study, the individual sample group sizes are relatively small despite being similar to other published cohorts, especially in genetic subjects and FTD subtypes, and not all the biomarkers were available in all the samples because of limited quantity of CSF available. Considering that the cohort is unicentric, confirmation of our findings in other independent cohorts would be needed to evaluate the significance of our findings in clinical practice. To sum up, biomarkers of synapse loss and axonal damage differentiate HC from neurodegenerative dementias and between HC and AD, FTD, and CJD, even if the relevant overlap observed limit their use in the differential diagnosis between different clinical phenotypes. In FTD, different neurodegeneration biomarkers showed different direction of changes, suggesting that these markers are not interchangeable and suggest different underlying mechanisms. For the application of the AT(N) system in the AD continuum, NF-L and 14-3-3 biomarkers perform similar to t-tau as markers for neurodegeneration (N). Acknowledgments The authors are grateful to all the participants for their collaboration in research. This work was developed at Centre de Recerca Biom
edica Cellex, Barcelona, Spain. The authors are indebted to the Biostatistics core facility of the IDIBAPS, CERCA Programme/Generalitat de Catalunya. A.A. is funded by the Departament de Salut de la Generalitat de Catalunya, Pla estrat
egic de recerca i innovaci o en salut (PERIS) 2016-2020 (SLT002/16/00329). H.Z. is a Wallenberg Academy Fellow supported by grants from the Swedish Research Council and the European Research Council. This work has received support from the EU/

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

EFPIA Innovative Medicines Initiative Joint Undertaking (AETIONOMY, grant no. 115568) including in-kind contributions from the EFPIA members involved. This work has also received support from the Spanish Ministry of Economy and Competitiveness-Instituto de Salud Carlos III and Fondo Europeo de Desarrollo Regional (FEDER), Uni on Europea, “Una manera de hacer Europa” (PI14/00282 to A.L. and PI16/000235 to R.S.-V.), and PERIS 2016-2020 (SLT002/ 16/00408 to R.S.-V.). Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.jalz.2019.09.001.

RESEARCH IN CONTEXT

1 Systematic review: Neurogranin, 14-3-3, neurofilament light (NF-L) and YKL-40 have been proposed as surrogate markers for synaptic loss, axonal damage, and neuroinflammation, respectively, in neurodegenerative dementia. The authors reviewed the literature about these biomarkers using traditional sources (e.g., PubMed). 2 Interpretation: The results of these four noncore cerebrospinal fluid biomarkers differentiate healthy control subjects from neurodegenerative dementias and between different diseases, even if the relevant overlap observed limit their use. In frontotemporal dementia, neurodegeneration biomarkers showed different direction of changes. For the application of the AT(N) system in the Alzheimer’s continuum NF-L and 14-3-3 perform similar to total tau protein as markers for neurodegeneration (N). 3 Future directions: The use of NF-L and 14-3-3 as neurodegenerative markers (N) in the AT(N) system should be evaluated in other independent cohorts. Moreover, the opposite behavior of synaptic loss markers observed in frontotemporal dementia deserves further studies.

References [1] Dubois B, Feldman HH, Jacova C, Cummings JL, DeKosky ST, Barberger-Gateau P, et al. Revising the definition of Alzheimer’s disease: a new lexicon. Lancet Neurol 2010;9:1118–27. [2] Blennow K, Zetterberg H. The past and the future of Alzheimer’s disease fluid biomarkers. J Alzheimers Dis 2018;62:1125–40. [3] Dubois B, Hampel H, Feldman HH, Scheltens P, Aisen P, Andrieu S, et al. Preclinical Alzheimer’s disease: definition, natural history, and diagnostic criteria. Alzheimers Dement 2016;12:292–323.

11

[4] Sperling RA, Aisen PS, Beckett LA, Bennett DA, Craft S, Fagan AM, et al. Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:280–92. [5] Jack CR Jr, Bennett DA, Blennow K, Carrillo MC, Dunn B, Budd Haeberlein S, et al. NIA-AA Research Framework: toward a biological definition of Alzheimer’s disease. Alzheimers Dement 2018; 14:535–62. [6] Hampel H, B€urger K, Teipel SJ, Bokde ALW, Zetterberg H, Blennow K. Core candidate neurochemical and imaging biomarkers of Alzheimer’s disease. Alzheimers Dement 2008;4:38–48. [7] Llorens F, Schmitz M, Karch A, Cramm M, Lange P, Gherib K, et al. Comparative analysis of cerebrospinal fluid biomarkers in the differential diagnosis of neurodegenerative dementia. Alzheimers Dement 2016;12:577–89. [8] Abu-Rumeileh S, Mometto N, Bartoletti-Stella A, Polischi B, Oppi F, Poda R, et al. Cerebrospinal fluid biomarkers in patients with frontotemporal dementia spectrum: a single-center study. J Alzheimers Dis 2018;66:551–63. [9] Norgren N, Rosengren L, Stigbrand T. Elevated neurofilament levels in neurological diseases. Brain Res 2003;987:25–31. [10] Landqvist Wald€o M, Frizell Santillo A, Passant U, Zetterberg H, Rosengren L, Nilsson C, et al. Cerebrospinal fluid neurofilament light chain protein levels in subtypes of frontotemporal dementia. BMC Neurol 2013;13:1–8. [11] Scherling CS, Hall T, Berisha F, Klepac K, Coppola G, Kramer JH, et al. CSF neurofilament concentration reflects disease severity in frontotemporal degeneration. Ann Neurol 2014;75:116–26. [12] Meeter LH, Dopper EG, Jiskoot LC, Sanchez-Valle R, Graff C, Benussi L, et al. Neurofilament light chain: a biomarker for genetic frontotemporal dementia. Ann Clin Transl Neurol 2016;3:623–36. [13] Pijnenburg YAL, Verwey NA, van der Flier WM, Scheltens P, Teunissen CE. Discriminative and prognostic potential of cerebrospinal fluid phosphoTau/tau ratio and neurofilaments for frontotemporal dementia subtypes. Alzheimers Dement Diagn Assess Dis Monit 2015;1:505–12. [14] Zerr I, Schmitz M, Karch A, Villar-Pique A, Kanata E, Golanska E, et al. Cerebrospinal fluid neurofilament light levels in neurodegenerative dementia: evaluation of diagnostic accuracy in the differential diagnosis of prion diseases. Alzheimers Dement 2018; 14:751–63. [15] Illan-Gala I, Alcolea D, Montal V, Dols-Icardo O, Mu~noz L, de Luna N, et al. CSF sAPPb, YKL-40, and NfL along the ALS-FTD spectrum. Neurology 2018;91:e1619–28. [16] Sj€ogren M, Rosengren L, Minthon L, Davidsson P, Blennow K, Wallin A. Cytoskeleton proteins in CSF distinguish frontotemporal dementia from AD. Neurology 2000;54:1960–4. [17] Skillback T, Farahmand B, Bartlett JW, Rosen C, Mattsson N, Nagga K, et al. CSF neurofilament light differs in neurodegenerative diseases and predicts severity and survival. Neurology 2014; 83:1945–53. [18] Thorsell A, Bjerke M, Gobom J, Brunhage E, Vanmechelen E, Andreasen N, et al. Neurogranin in cerebrospinal fluid as a marker of synaptic degeneration in Alzheimer’s disease. Brain Res 2010; 1362:13–22. € [19] Kvartsberg H, Duits FH, Ingelsson M, Andreasen N, Ohrfelt A, Andersson K, et al. Cerebrospinal fluid levels of the synaptic protein neurogranin correlates with cognitive decline in prodromal Alzheimer’s disease. Alzheimers Dement 2015;11:1180–90. [20] Tarawneh R, D’Angelo G, Crimmins D, Herries E, Griest T, Fagan AM, et al. Diagnostic and prognostic utility of the synaptic marker neurogranin in Alzheimer disease. JAMA Neurol 2016; 73:561–71. [21] Wellington H, Paterson RW, Portelius E, T€ornqvist U, Magdalinou N, Fox NC, et al. Increased CSF neurogranin concentration is specific to Alzheimer disease. Neurology 2016;86:829–35.

12

A. Antonell et al. / Alzheimer’s & Dementia - (2019) 1-12

[22] Portelius E, Olsson B, H€oglund K, Cullen NC, Kvartsberg H, Andreasson U, et al. Cerebrospinal fluid neurogranin concentration in neurodegeneration: relation to clinical phenotypes and neuropathology. Acta Neuropathol 2018;136:363–76. [23] Janelidze S, Hertze J, Zetterberg H, Landqvist Wald€o M, Santillo A, Blennow K, et al. Cerebrospinal fluid neurogranin and YKL-40 as biomarkers of Alzheimer’s disease. Ann Clin Transl Neurol 2016; 3:12–20. [24] Foote M, Zhou Y. 14-3-3 Proteins in neurological disorders. Int J Biochem Mol Biol 2012;3:152–64. [25] Llorens F, Karch A, Golanska E, Schmitz M, Lange P, Sikorska B, et al. Cerebrospinal fluid biomarker-based diagnosis of sporadic Creutzfeldt-Jakob disease: a validation study for previously established cutoffs. Dement Geriatr Cogn Disord 2017;43:71–80. [26] Humpel C, Benke T. Cerebrospinal fluid levels of 14-3-3 gamma: what does it tell us about sporadic Creutzfeldt-Jakob disease? Pharmacology 2017;100:243–5. [27] Burkhard PR, Sanchez JC, Landis T, Hochstrasser DF. CSF detection of the 14-3-3 protein in unselected patients with dementia. Neurology 2001;56:1528–33. [28] Antonell A, Mansilla A, Rami L, Llado A, Iranzo A, Olives J, et al. Cerebrospinal fluid level of YKL-40 protein in preclinical and prodromal Alzheimer’s disease. J Alzheimers Dis 2014;42:901–8. [29] Llorens F, Th€ une K, Tahir W, Kanata E, Diaz-Lucena D, Xanthopoulos K, et al. YKL-40 in the brain and cerebrospinal fluid of neurodegenerative dementias. Mol Neurodegener 2017; 12:1–21. [30] Alcolea D, Irwin DJ, Illan-Gala I, Mu~noz L, Clarimon J, Mcmillan T, et al. Elevated YKL-40 and low sAPPb:YKL-40 ratio in antemortem cerebrospinal fluid of patients with pathologically confirmed FTLD. J Neurol Neurosurg Psychiatry 2019;90:180–6. [31] Mckhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Kawas CH, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:263–9. [32] Albert MS, Dekosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:270–9. [33] Gorno-Tempini ML, Hillis AE, Weintraub S, Kertesz A, Mendez M, Cappa SF, et al. Classification of primary progressive aphasia and its variants. Neurology 2011;76:1006–14.

[34] Rascovsky K, Hodges JR, Knopman D, Mendez MF, Kramer JH, Neuhaus J, et al. Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia. Brain 2011;134:2456–77. [35] Fortea J, Llado A, Clarimon J, Lleo A, Oliva R, Peri J, et al. PICOGEN: experiencia de 5 a~nos de un programa de asesoramiento genetico en demencia. Neurologia 2011;26:143–9. [36] Petzold A, Altintas A, Andreoni L, Bartos A, Berthele A, Blankenstein MA, et al. Neurofilament ELISA validation. J Immunol Methods 2010;352:23–31. [37] Schmitz M, Ebert E, Stoeck K, Karch A, Collins S, Calero M, et al. Validation of 14-3-3 protein as a marker in sporadic CreutzfeldtJakob disease diagnostic. Mol Neurobiol 2016;53:2189–99. [38] Gaetani L, H€oglund K, Parnetti L, Pujol-Calderon F, Becker B, Eusebi P, et al. A new enzyme-linked immunosorbent assay for neurofilament light in cerebrospinal fluid: analytical validation and clinical evaluation. Alzheimers Res Ther 2018;10:1–13. [39] Willemse EAJ, De Vos A, Herries EM, Andreasson U, Engelborghs S, Van Der Flier WM, et al. Neurogranin as cerebrospinal fluid biomarker for Alzheimer disease: an assay comparison study. Clin Chem 2018; 64:927–37. [40] Nordengen K, Kirsebom BE, Henjum K, Selnes P, Gısladottir B, Wettergreen M, et al. Glial activation and inflammation along the Alzheimer’s disease continuum. J Neuroinflammation 2019;16:1–13. [41] Mattsson N, Andreasson U, Persson S, Carrillo MC, Collins S, Chalbot S, et al. CSF biomarker variability in the Alzheimer’s Association quality control program. Alzheimers Dement 2013;9:251–61. [42] McGee S. Simplifying likelihood ratios. J Gen Intern Med 2002; 17:646–9. [43] Mattsson N, Insel PS, Palmqvist S, Portelius E, Zetterberg H, Weiner M, et al. Cerebrospinal fluid tau, neurogranin, and neurofilament light in Alzheimer’s disease. EMBO Mol Med 2016;8:1184–96. [44] Hampel H, Toschi N, Baldacci F, Zetterberg H, Blennow K, Kilimann I, et al. Alzheimer’s disease biomarker-guided diagnostic workflow using the added value of six combined cerebrospinal fluid candidates: Ab 1–42, total-tau, phosphorylated-tau, NFL, neurogranin, and YKL-40. Alzheimers Dement 2018;14:492–501. [45] Khachaturian AS, Hayden KM, Mielke MM, Tang Y, Lutz MW, Gustafson DR, et al. Future prospects and challenges for Alzheimer’s disease drug development in the era of the NIA-AA Research Framework. Alzheimers Dement 2018;14:532–4. [46] Preische O, Schultz SA, Apel A, Kuhle J, Kaeser SA, Barro C, et al. Serum neurofilament dynamics predicts neurodegeneration and clinical progression in presymptomatic Alzheimer’s disease. Nat Med 2019;25:277–83.