Association of Cerebrospinal Fluid Neurofilament Light Protein With Risk of Mild Cognitive Impairment Among Individuals Without Cognitive Impairment | Dementia and Cognitive Impairment | JAMA Neurology | JAMA Network
[Skip to Navigation]
Sign In
Table 1.  Baseline Participant Characteristics
Baseline Participant Characteristics
Table 2.  Continuous CSF Measures and Risk of Mild Cognitive Impairment
Continuous CSF Measures and Risk of Mild Cognitive Impairment
Table 3.  CSF Measures in Quartiles and Risk of Mild Cognitive Impairment
CSF Measures in Quartiles and Risk of Mild Cognitive Impairment
Table 4.  Combination of CSF Neurofilament Light Protein and T-tau for Risk of Mild Cognitive Impairment
Combination of CSF Neurofilament Light Protein and T-tau for Risk of Mild Cognitive Impairment
1.
Scheltens  P, Blennow  K, Breteler  MM,  et al.  Alzheimer’s disease.  Lancet. 2016;388(10043):505-517. doi:10.1016/S0140-6736(15)01124-1PubMedGoogle ScholarCrossref
2.
Blennow  K, Hampel  H, Weiner  M, Zetterberg  H.  Cerebrospinal fluid and plasma biomarkers in Alzheimer disease.  Nat Rev Neurol. 2010;6(3):131-144. doi:10.1038/nrneurol.2010.4PubMedGoogle ScholarCrossref
3.
Lewczuk  P, Riederer  P, O’Bryant  SE,  et al; Members of the WFSBP Task Force Working on this Topic: Peter Riederer, Carla Gallo, Dimitrios Kapogiannis, Andrea Lopez Mato, Florence Thibaut.  Cerebrospinal fluid and blood biomarkers for neurodegenerative dementias: an update of the Consensus of the Task Force on Biological Markers in Psychiatry of the World Federation of Societies of Biological Psychiatry.  World J Biol Psychiatry. 2018;19(4):244-328. doi:10.1080/15622975.2017.1375556PubMedGoogle ScholarCrossref
4.
Norgren  N, Rosengren  L, Stigbrand  T.  Elevated neurofilament levels in neurological diseases.  Brain Res. 2003;987(1):25-31. doi:10.1016/S0006-8993(03)03219-0PubMedGoogle ScholarCrossref
5.
Hoffman  PN, Cleveland  DW, Griffin  JW, Landes  PW, Cowan  NJ, Price  DL.  Neurofilament gene expression: a major determinant of axonal caliber.  Proc Natl Acad Sci U S A. 1987;84(10):3472-3476. doi:10.1073/pnas.84.10.3472PubMedGoogle ScholarCrossref
6.
Zetterberg  H, Skillbäck  T, Mattsson  N,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Association of cerebrospinal fluid neurofilament light concentration with Alzheimer disease progression.  JAMA Neurol. 2016;73(1):60-67. doi:10.1001/jamaneurol.2015.3037PubMedGoogle ScholarCrossref
7.
Kvartsberg  H, Portelius  E, Andreasson  U,  et al.  Characterization of the postsynaptic protein neurogranin in paired cerebrospinal fluid and plasma samples from Alzheimer’s disease patients and healthy controls.  Alzheimers Res Ther. 2015;7(1):40. doi:10.1186/s13195-015-0124-3PubMedGoogle ScholarCrossref
8.
Mattsson  N, Insel  PS, Palmqvist  S,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Cerebrospinal fluid tau, neurogranin, and neurofilament light in Alzheimer’s disease.  EMBO Mol Med. 2016;8(10):1184-1196. doi:10.15252/emmm.201606540PubMedGoogle ScholarCrossref
9.
Pereira  JB, Westman  E, Hansson  O; Alzheimer’s Disease Neuroimaging Initiative.  Association between cerebrospinal fluid and plasma neurodegeneration biomarkers with brain atrophy in Alzheimer’s disease.  Neurobiol Aging. 2017;58:14-29. doi:10.1016/j.neurobiolaging.2017.06.002PubMedGoogle ScholarCrossref
10.
Kester  MI, Teunissen  CE, Crimmins  DL,  et al.  Neurogranin as a cerebrospinal fluid biomarker for synaptic loss in symptomatic Alzheimer disease.  JAMA Neurol. 2015;72(11):1275-1280. doi:10.1001/jamaneurol.2015.1867PubMedGoogle ScholarCrossref
11.
Wellington  H, Paterson  RW, Portelius  E,  et al.  Increased CSF neurogranin concentration is specific to Alzheimer disease.  Neurology. 2016;86(9):829-835. doi:10.1212/WNL.0000000000002423PubMedGoogle ScholarCrossref
12.
Portelius  E, Olsson  B, Höglund  K,  et al.  Cerebrospinal fluid neurogranin concentration in neurodegeneration: relation to clinical phenotypes and neuropathology.  Acta Neuropathol. 2018;136(3):363-376. doi:10.1007/s00401-018-1851-xPubMedGoogle ScholarCrossref
13.
Roberts  RO, Geda  YE, Knopman  DS,  et al.  The Mayo Clinic Study of Aging: design and sampling, participation, baseline measures and sample characteristics.  Neuroepidemiology. 2008;30(1):58-69. doi:10.1159/000115751PubMedGoogle ScholarCrossref
14.
St Sauver  JL, Grossardt  BR, Yawn  BP,  et al.  Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system.  Int J Epidemiol. 2012;41(6):1614-1624. doi:10.1093/ije/dys195PubMedGoogle ScholarCrossref
15.
Ivnik  RJ, Malec  JF, Smith  GE,  et al.  Mayo’s older Americans normative studies: updated AVLT norms for ages 56 to 97.  Clin Neuropsychol. 1992;6(suppl):83-104. doi:10.1080/13854049208401880Google ScholarCrossref
16.
Petersen  RC.  Mild cognitive impairment as a diagnostic entity.  J Intern Med. 2004;256(3):183-194. doi:10.1111/j.1365-2796.2004.01388.xPubMedGoogle ScholarCrossref
17.
American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). 4th ed. Washington, DC: American Psychiatric Association; 1994.
18.
Gaetani  L, Höglund  K, Parnetti  L,  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):8. doi:10.1186/s13195-018-0339-1PubMedGoogle ScholarCrossref
19.
Portelius  E, Zetterberg  H, Skillbäck  T,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Cerebrospinal fluid neurogranin: relation to cognition and neurodegeneration in Alzheimer’s disease.  Brain. 2015;138(pt 11):3373-3385. doi:10.1093/brain/awv267PubMedGoogle ScholarCrossref
20.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8PubMedGoogle ScholarCrossref
21.
Rocca  WA, Yawn  BP, St Sauver  JL, Grossardt  BR, Melton  LJ  III.  History of the Rochester Epidemiology Project: half a century of medical records linkage in a US population.  Mayo Clin Proc. 2012;87(12):1202-1213. doi:10.1016/j.mayocp.2012.08.012PubMedGoogle ScholarCrossref
22.
Crowson  CS, Atkinson  EJ, Therneau  TM.  Assessing calibration of prognostic risk scores.  Stat Methods Med Res. 2016;25(4):1692-1706. doi:10.1177/0962280213497434PubMedGoogle ScholarCrossref
23.
Zetterberg  H.  Tauomics and kinetics in human neurons and biological fluids.  Neuron. 2018;97(6):1202-1205. doi:10.1016/j.neuron.2018.02.030PubMedGoogle ScholarCrossref
24.
Skillbäck  T, Rosén  C, Asztely  F, Mattsson  N, Blennow  K, Zetterberg  H.  Diagnostic performance of cerebrospinal fluid total tau and phosphorylated tau in Creutzfeldt-Jakob disease: results from the Swedish Mortality Registry.  JAMA Neurol. 2014;71(4):476-483. doi:10.1001/jamaneurol.2013.6455PubMedGoogle ScholarCrossref
25.
van Harten  AC, Smits  LL, Teunissen  CE,  et al.  Preclinical AD predicts decline in memory and executive functions in subjective complaints.  Neurology. 2013;81(16):1409-1416. doi:10.1212/WNL.0b013e3182a8418bPubMedGoogle ScholarCrossref
26.
Vos  SJ, Xiong  C, Visser  PJ,  et al.  Preclinical Alzheimer’s disease and its outcome: a longitudinal cohort study.  Lancet Neurol. 2013;12(10):957-965. doi:10.1016/S1474-4422(13)70194-7PubMedGoogle ScholarCrossref
27.
Okonkwo  OC, Alosco  ML, Griffith  HR,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Cerebrospinal fluid abnormalities and rate of decline in everyday function across the dementia spectrum: normal aging, mild cognitive impairment, and Alzheimer disease.  Arch Neurol. 2010;67(6):688-696. doi:10.1001/archneurol.2010.118PubMedGoogle ScholarCrossref
28.
Buchhave  P, Minthon  L, Zetterberg  H, Wallin  AK, Blennow  K, Hansson  O.  Cerebrospinal fluid levels of β-amyloid 1-42, but not of tau, are fully changed already 5 to 10 years before the onset of Alzheimer dementia.  Arch Gen Psychiatry. 2012;69(1):98-106. doi:10.1001/archgenpsychiatry.2011.155PubMedGoogle ScholarCrossref
29.
Roe  CM, Fagan  AM, Grant  EA,  et al.  Amyloid imaging and CSF biomarkers in predicting cognitive impairment up to 7.5 years later.  Neurology. 2013;80(19):1784-1791. doi:10.1212/WNL.0b013e3182918ca6PubMedGoogle ScholarCrossref
30.
van Rossum  IA, Vos  SJ, Burns  L,  et al.  Injury markers predict time to dementia in subjects with MCI and amyloid pathology.  Neurology. 2012;79(17):1809-1816. doi:10.1212/WNL.0b013e3182704056PubMedGoogle ScholarCrossref
31.
Headley  A, De Leon-Benedetti  A, Dong  C,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Neurogranin as a predictor of memory and executive function decline in MCI patients.  Neurology. 2018;90(10):e887-e895. doi:10.1212/WNL.0000000000005057PubMedGoogle ScholarCrossref
32.
Jack  CR  Jr, Bennett  DA, Blennow  K,  et al; Contributors.  NIA-AA Research Framework: toward a biological definition of Alzheimer’s disease.  Alzheimers Dement. 2018;14(4):535-562. doi:10.1016/j.jalz.2018.02.018PubMedGoogle ScholarCrossref
Original Investigation
November 12, 2018

Association of Cerebrospinal Fluid Neurofilament Light Protein With Risk of Mild Cognitive Impairment Among Individuals Without Cognitive Impairment

Author Affiliations
  • 1Department of Health Sciences Research, Mayo Clinic, Rochester, Minnesota
  • 2Department of Psychiatry and Neurochemistry, Institute of Neuroscience and Physiology, the Sahlgrenska Academy at the University of Gothenburg, Mölndal, Sweden
  • 3University of Gothenburg, Mölndal, Sweden
  • 4Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Mölndal, Sweden
  • 5Neuropsychiatric Epidemiology Unit, Department of Psychiatry and Neurochemistry, Institute of Neuroscience and Physiology, the Sahlgrenska Academy at the Institute of Neurology, University College London, Queen Square, London, England
  • 6United Kingdom Dementia Research Institute at University College London, London, England
  • 7Department of Neurology, Mayo Clinic, Rochester, Minnesota
  • 8Alzheimer Center, VU University Medical Center, Amsterdam, the Netherlands
  • 9Department of Radiology, Mayo Clinic, Rochester, Minnesota
JAMA Neurol. 2019;76(2):187-193. doi:10.1001/jamaneurol.2018.3459
Key Points

Question  Are cerebrospinal fluid (CSF) neurofilament light protein (NfL) and neurogranin (Ng) levels associated with the risk of incident mild cognitive impairment (MCI), and how do these markers compare with CSF total (T-tau) or phosphorylated tau (P-tau) for risk of MCI in the general community?

Findings  In this population-based study, compared with the bottom quartile, the top quartile of CSF NfL was associated with a 3-fold increased risk of MCI. Neither CSF T-tau, P-tau, nor Ng levels were associated with risk of MCI.

Meaning  Elevated CSF NfL levels, but not T-tau, P-tau, or Ng levels, are associated with risk of MCI in a community population.

Abstract

Importance  Accumulating data suggest that elevated cerebrospinal fluid (CSF) neurofilament light (NfL) and neurogranin (Ng) levels are associated with cognitive decline and may be useful markers of neurodegeneration. However, to our knowledge, previous studies have not assessed these CSF markers in the community, evaluated them with regards to risk of mild cognitive impairment (MCI), or compared their prognostic value with CSF total tau (T-tau) or phosphorylated tau (P-tau).

Objective  To determine (1) whether CSF NfL and Ng levels were associated with risk of MCI, (2) the effect size of these markers compared with CSF T-tau or P-tau for risk of MCI, and (3) whether CSF amyloid-β (Aβ42) modified these associations.

Design, Setting and Participants  The analyses included 648 participants without cognitive impairment who were enrolled into the prospective population-based Mayo Clinic Study of Aging between January 2004 and December 2015 with available CSF data and at least 1 follow-up visit. Participants were followed up for a median of 3.8 years (interquartile range, 2.6-5.4 years). The CSF NfL and Ng levels were measured using an in-house sandwich enzyme-linked immunosorbent assay. The CSF Aβ42, T-tau, and P-tau levels were measured with automated electrochemiluminescence immunoassays. Cox proportional hazards models, with age as the timescale, were used to assess the association between CSF NfL, Ng, Aβ42, T-tau, or P-tau with risk of MCI after adjusting for sex, education, apolipoprotein E genotype, and the Charlson comorbidity index. To examine CSF Aβ42 as an effect modifier, it was categorized into tertiles; the bottom tertile was defined as having elevated brain amyloid.

Main Outcomes and Measures  Risk of MCI.

Results  At baseline, the median age of the 648 participants without cognitive impairment was 72.3 years (range, 50.7-95.3 years) and 366 (56.5%) were men; 96 (14.8%) developed incident MCI. Compared with the bottom quartile, the top quartile of CSF NfL was associated with a 3.1-fold increased risk of MCI (hazard ratio, 3.13; 95% CI, 1.36-7.18) in multivariate models. Neither CSF T-tau, P-tau, nor Ng was associated with risk of MCI. There was no interaction between Aβ42 and CSF NfL for risk of MCI.

Conclusions and Relevance  Elevated CSF NfL levels but not CSF T-tau, P-tau or Ng are a risk factor for MCI in a community population and are independent of brain amyloid.

Introduction

Alzheimer disease (AD) is a progressive neurodegenerative disease that is characterized by the accumulation of amyloid-β (Aβ) and tau pathologies, neurodegeneration, and cognitive decline.1 Cerebrospinal fluid (CSF) Aβ42, total tau (T-tau), and phosphorylated tau (P-tau) are established diagnostic and/or prognostic biomarkers for AD.2 More recently, 2 promising biomarkers for neurodegeneration, neurofilament light protein (NfL) and neurogranin (Ng), have emerged.3 Neurofilament light protein functions to determine the axonal caliber, which is crucial for morphological integrity and conduction velocity.4,5 Therefore, NfL is a putative marker of subcortical large-caliber axonal degeneration and has recently been highlighted for its potential as a biomarker of AD progression.6 In contrast, Ng is a postsynaptic protein that is enriched in dendritic spines and therefore could be a potential biomarker for detecting synaptic dysfunction and/or loss in AD.7 Recent studies have reported that elevated CSF NfL8,9 and/or Ng8-10 levels are associated with a greater risk of progression from mild cognitive impairment (MCI) to AD dementia or with cognitive decline. Notably, unlike CSF NfL levels, which have been elevated in multiple neurodegenerative diseases, elevations of CSF Ng levels may be specific to AD,11 a finding that was recently verified in a large study that also included neuropathologically confirmed cases.12

Studies that examine CSF NfL and Ng have primarily used memory clinic patients (eg, the Amsterdam Dementia cohort study) or individuals who were screened to exclude cerebrovascular disease or other forms of dementia (eg, Alzheimer’s Disease Neuroimaging Initiative [ADNI]). Thus, the prognostic performance of CSF NfL and Ng in a community-based population is not well understood. In this study, we determined whether CSF NfL and CSF Ng were risk factors for MCI and compared their strengths of association for MCI risk with CSF T-tau and P-tau in a population-based cohort of participants without cognitive impairment (CU). We also examined whether a combination of elevated CSF NfL and T-tau levels was more strongly associated with risk of MCI than either alone, and whether CSF Aβ42 modified the association between CSF NfL or Ng and risk of MCI.

Methods
Study Participants

The Mayo Clinic Study of Aging is a prospective population-based study aimed at characterizing the incidence and prevalence of MCI in Olmsted County, Minnesota.13 In 2004, Olmsted County residents between age 70 and 89 years were enumerated using the Rochester Epidemiology Project medical records linkage system.14 An age- and sex-stratified random sampling design was used to ensure that men and women were equally represented in each 10-year age strata. The study was extended to include those 50 years and older in 2012. These analyses included 648 CU participants with available CSF NfL, Ng, Aβ42, T-tau, or P-tau measures, cognitive testing, and at least 1 follow-up visit with cognitive testing.

Protocol Approvals Standard, Registrations, and Patient Consents

The study was approved by Mayo Clinic and Olmsted Medical Center institutional review boards. Written informed consent was obtained from all participants.

Participant Assessment

Mayo Clinic Study of Aging visits occurred every 15 months and included a physician examination, an interview by a study coordinator, and neuropsychological testing that was administered by a psychometrist.13 The neuropsychological battery included 9 tests that covered 4 domains: (1) memory (Auditory Verbal Learning Test Delayed Recall Trial, Wechsler Memory Scale-Revised Logical Memory II, and Visual Reproduction II), (2) language (Boston Naming Test and Category Fluency), (3) executive function (Trail Making Test B, Wechsler Adult Intelligence Scale-Revised [WAIS-R] DigitSymbol subtest), and (4) visuospatial (WAIS-R Picture Completion, Block Design subtests). We calculated sample-specific z scores for all cognitive tests and created domain scores by averaging the z scores within each domain. We created a global cognitive score using the z transformation of the average of the 4 domains.

MCI and Dementia Diagnostic Determination

Clinical diagnoses were determined by a consensus committee, including the neurologist, neuropsychologist, and the study coordinator, who evaluated each participant. Performance in a cognitive domain was compared with the age-adjusted scores of CU individuals that were previously obtained using Mayo’s Older American Normative Studies.15 This approach relied on prior normative work in an independent sample from the same population. Participants with scores around 1.0 SD below the age-specific mean in the general population were considered for possible cognitive impairment. The operational definition of MCI was based on clinical judgment, including a history from the patient and informant. Published criteria were used for the diagnosis: cognitive complaint, cognitive function not normal for age, essentially normal functional activities, and no dementia.16 A final decision about impairment in a cognitive domain was made after considering education, occupation, visual or hearing deficits, and reviewing all other participant information. The diagnosis of dementia was based on published criteria.17 Participants who performed in the normal range and did not meet criteria for MCI or dementia were deemed CU. Neither CSF results nor neuroimaging were considered in determining the clinical diagnoses of MCI or dementia and the consensus committee was masked to these findings.

Lumbar Punctures and CSF Measurements

Fasting lumbar punctures were performed early in the morning in the lateral decubitus position using a 20- or 22-gauge needle (Quincke). Two cubic centimeters of CSF were used to evaluate routine markers (glucose level, protein level, and blood cell count). The remainder was divided into 0.5-cm3 aliquots and stored at −80°C for future analyses avoiding freeze-thaw cycles before the current analyses.

The CSF Aβ42, T-tau, and P-tau levels were measured with automated electrochemiluminescence Elecsys immunoassays (Roche Diagnostics) at Mayo Clinic Rochester. The CSF NfL and Ng levels were measured in the Clinical Neurochemistry Laboratory at the University of Gothenburg. For NfL, an in-house sandwich enzyme-linked immunosorbent assay (ELISA) with capture and detection antibodies that were directed against the central rod domain of the protein (NfL 21 and NfL 23, respectively) was used.18 As published, this ELISA had within-plate and interplate variations of below 8% and 13%, respectively, and a strong correlation (r = 0.998; P < .001) with CSF values that were analyzed using an Uman Diagnostics ELISA.18 Nine of the 648 participants (1.4%) had CSF NfL levels that were more than 20 000 ng/L, which was more than 3 SDs from the mean (ie, >4723.5 ng/L). These 9 values were recoded as missing. For CSF Ng, an in-house ELISA method was used.19 The Ng level of 1 participant was below the detection limit and recoded as missing.

Assessment of Covariates

Participant demographics (eg, age, sex, and years of education) were ascertained at the in-clinic examination. Apolipoprotein E (APOE) ε4 genotyping was performed from blood drawn at the in-clinic examination. Medical conditions and the Charlson Comorbidity Index20 were determined for each participant by medical record abstraction using the Rochester Epidemiology Project medical records linkage system.14,21

Statistical Analysis

We examined the associations between CSF NfL or Ng levels and CSF Aβ 42, T-tau and P-tau using the Spearman rank correlation. All CSF variables were z–log transformed to normalize the distributions and to use the same units for each variable to adequately compare the effect size across markers. We used Cox proportional hazard regression models to determine whether the baseline CSF markers, continuous and in quartiles, were associated with risk of MCI among CU participants. Age was used as the timescale. Participants were followed up until a diagnosis of MCI/dementia, death, or last follow-up visit. The event date for those who developed MCI or dementia was defined as the midpoint between a participant’s last visit defined as CU and first visit defined as MCI or dementia. Four participants went from a diagnosis of CU at one visit to a diagnosis of dementia at a subsequent visit. The analyses were repeated including and excluding these participants and the results remained the same. Thus, they were included in the final models.

Multivariable models adjusted for sex, educational, and Charlson comorbidity index. Analyses were repeated after additional adjustment for APOE ε4 status. To determine whether a combination of elevated CSF NfL and T-tau levels was more strongly associated with MCI risk compared with either alone, we developed a categorical variable with NfL only in the highest quartile, T-tau only in the highest quartile, NfL and T-tau in the highest quartile, and neither NfL nor T-tau in the highest quartile. We then used Cox proportional hazard models to assess the association between this variable and MCI risk. We also examined whether there was an interaction between CSF Aβ42 and CSF NfL or Ng for risk of MCI. The proportional hazard assumption was tested using the Schoenfeld residuals test for each measure. The test was nonsignificant, indicating that the assumption was valid. We used a model-based framework for calibration that extended to survival data22 and receiver-operating characteristics curve analyses for discrimination. Statistical analyses were completed using SAS, version 9.4 (SAS Institute). A 2-tailed P < .05 was considered significant.

Results

The baseline characteristics of the 648 CU participants are shown in Table 1. Their median age was 72.3 years (range, 50.7-95.3 years) and 366 (56.5%) were men. The median education was 14 years (range, 8-20), 172 (26.6%) had an APOE ɛ4 allele, and 96 (14.8%) progressed to MCI. Participants were followed up a median of 3.8 years (range,0.5-10.1). Compared with the participants who remained CU, those who progressed to MCI had a similar frequency of women but were more frequently carriers of an APOE ɛ4 allele (36.5% vs 24.9%, P = .02), had less education (14.0 years vs 15.0 years; P = .004), and had a higher Charlson Comorbidity Index score (3.0 vs 2.0; P < .001). Cerebrospinal fluid NfL and Ng were moderately correlated (Spearman ρ = 0.30; P < .001). The CSF Ng levels were moderately to highly correlated with CSF Aβ42 (Spearman ρ = 0.40; P < .001), P-tau (Spearman ρ = 0.81, P < .001) and T-tau (Spearman ρ = 0.82; P < .001). In contrast, there were only moderate to low correlations between CSF NfL and CSF Aβ42 (CSF NfL Spearman ρ = 0.12; P = .002), P-tau (NfL Spearman ρ = 0.46; P < .001), and T-tau (NfL Spearman ρ = 0.47; P < .001).

Continuous CSF Measures and Risk of MCI

Each z log unit increase in CSF NfL was associated with a 1.32-fold increased risk of MCI (95% CI, 1.08-1.60) after adjusting for sex, education, and the Charlson comorbidity index (Table 2). This association was almost identical after additional adjustment for APOE. In contrast, neither CSF Ng, T-tau, nor P-Tau were associated with risk of MCI (Table 2). However, each z log unit increase in CSF Aβ42 was associated with a decreased risk of MCI in multivariable models, both when excluding (hazard ratio [HR], 0.74; 95% CI, 0.61-0.89) and including (HR, 0.80; 95% CI, 0.66-0.97) adjustment for APOE. Thus, low CSF Aβ42 was associated with an increased risk of MCI.

CSF Measures, in Quartiles, and Risk of MCI

Compared with the lowest quartile of CSF NfL, the highest quartile was associated with a 2.9-fold increased risk of MCI (95% CI, 1.26-6.67; Table 3). The associated risk was even larger after adjustment for APOE (HR, 3.13; 95% CI, 1.36-7.18). The second and third quartiles of CSF NfL were not associated with increased risk. There were no associations between CSF T-tau, P-tau, or Ng in quartiles and risk of MCI (Table 3). Elevated CSF Aβ42 levels were associated with decreased risk of MCI (quartile 3 vs 1: HR, 0.41; 95% CI, 0.23-0.75; quartile 4 vs 1: HR, 0.47; 95% CI, 0.27-0.81). However, after adjustment for APOE, this association was only significant for quartile 3 (HR, 0.48; 95% CI, 0.26-0.88).

Risk of MCI Using a Combination of Elevated CSF NfL and T-tau Levels in Quartiles

The CSF NfL and T-tau levels may contribute divergent information regarding neurodegeneration,23 and our previously described findings align with this hypothesis. Given the potential of each as unique neurodegenerative measures, we next determined whether the combination of CSF NfL and T-tau was associated with a greater risk of MCI compared with either alone. We created a categorical variable for CSF NfL and T-tau in the highest quartile (n = 68), only CSF NfL in the highest quartile (n = 90), only CSF T-tau in the highest quartile (n = 89), and neither CSF NfL nor T-tau in the highest quartile (n = 391). Participants with only CSF NfL in the highest quartile (HR, 2.24; 95% CI, 1.31-3.83) and with both CSF NfL and T-tau in the highest quartile had a similarly increased risk of MCI (HR, 2.29; 95% CI, 1.28-4.09) compared with participants without both markers in the highest quartile. Participants who only had T-tau in the highest quartile did not have an increased risk of MCI (Table 4). The associations did not change after adjusting for APOE.

Assessing Effect Modification by Elevated Brain Amyloid Levels

In additional analyses, we examined the interactions of CSF Aβ42 with CSF NfL, Ng, T-tau, and P-tau for risk of MCI. The CSF Aβ42 levels were not an effect modifier in any analyses. Adjusting for CSF Aβ42 also did not alter any of the results.

Assessing Discrimination and Calibration

A Cox proportional hazards model that used only sex, education, APOE, and Charlson comorbidity index score to predict the risk of MCI with age as the time scale gave a C-statistic of 0.65. Only the addition of CSF NfL (C-statistic = 0.69), but not CSF total tau (C-statistic = 0.64), enhanced the predictive value of the model. Using a model-based framework for calibration extended to survival data, a nonsignificant P value was found (favoring acceptable calibration) for only the model in which CSF NfL was added to the previously mentioned covariates.

Discussion

In this study, we examined and compared established CSF (T-tau) and newer (NfL and Ng) measures of neurodegeneration for the risk of MCI among a community-based population of CU participants. Elevated CSF NfL levels were strongly associated with risk of MCI. There were no associations between CSF Ng, T-tau, or P-tau and risk of MCI. The combination of elevated CSF NfL and T-tau levels for risk of MCI was not better than CSF NfL alone. As expected, low CSF Aβ42 levels were associated with an increased risk of MCI. However, the associations between CSF NfL and risk for MCI were independent of CSF Aβ42. Collectively, these results suggest that in a community-based sample of CU participants, CSF NfL is more strongly associated with risk of MCI compared with other CSF markers of neurodegeneration (eg, T-tau and Ng).

Two previous studies that used ADNI data reported that elevated CSF NfL levels were associated with a faster cognitive decline among patients with MCI.6,8 Another study based on the Swedish Dementia Registry showed that higher CSF NfL levels were associated with a shorter survival among patients with AD dementia,24 again suggesting that CSF NfL is a marker of clinical progression in symptomatic patients. Our results extend this finding earlier in the clinical disease spectrum by showing that high CSF NfL levels are also strongly associated with risk of MCI in a community-based study of CU participants. Our findings are also consistent with ADNI results suggesting that CSF NfL was associated with cortical and subcortical brain atrophy,6,9 which should translate into cognitive progression.

Low CSF Aβ42 levels were associated with risk for MCI, which has been shown before in participants with preclinical AD stage 1 National Institute on Aging–Alzheimer Association criteria.25,26 The association between elevated CSF NfL levels and MCI was independent of CSF Aβ42. Thus, CSF NfL is a risk factor for cognitive impairment for those on the AD pathway (with low CSF Aβ42 levels) and for those who are not. Our results are also consistent with studies that have shown that elevated CSF NfL levels reflect neurodegeneration in AD6,9,24 and in many other neurological diseases.24 Thus, CSF NfL is a nonspecific marker of neurodegeneration.

When we compared elevated CSF NfL levels with T-tau and P-tau, NfL was associated with risk of MCI whereas CSF T-tau and P-tau were not. In addition, CSF NfL was the best fit model. In fact, there were no associations between T-tau or P-tau and risk of MCI, a finding that contradicts previous studies.26,27 A potential explanation for the lack of an association could be the population-based study design in contrast to including CU clinical patients or at-risk populations because of a significant family history. Further, the combination of elevated CSF T-tau and NfL levels was not better than NfL alone for predicting risk of MCI. These findings suggest that CSF NfL is a stronger predictor of cognitive deterioration than T-tau in a population-based sample of CU participants. In addition, our results support the hypothesis that CSF NfL and CSF T-tau convey divergent information regarding neurodegeneration. Cerebrospinal fluid NfL is a measure of axonal injury, whereas CSF T-tau probably reflects the temporal course and severity of neuronal injury at a given point.28-30

We did not find any associations between CSF Ng and risk for MCI. This parallels findings from the memory clinic–based Amsterdam Dementia Cohort, in which Ng did not predict progression from CU to MCI but did predict progression from MCI to AD dementia.10 Another recent study conducted in ADNI showed that elevated Ng levels were associated with cognitive decline in participants with MCI but not among CU participants.31 These findings, together with our finding that CSF Ng was not associated with risk for MCI, suggest that CSF Ng levels may be elevated later in the disease process after cognitive symptoms become apparent. Notably, we found a strong correlation between CSF Ng and T-tau and P-tau, which has been described previously.8 This strong correlation may explain the lack of association between CSF Ng and risk for MCI because we also found that CSF T-tau and P-Tau were not associated with risk of MCI.

Strengths and Limitations

The strengths of this study include the community-based population and the large number of CU participants. However, there are some limitations. First, we could not differentiate between cognitive subtypes of MCI because the subgroups were quite small. Second, all CSF markers were measured at 1 point. Thus, we could not incorporate time-varying measures and the association between intraindividual change in these markers and cognition could not be assessed. Third, when examining CSF Aβ42 as an effect modifier, we considered the lowest tertile as indicative of elevated brain amyloid levels. It is possible that the subtle effects of emerging Aβ pathology in participants with slightly higher CSF Aβ42 levels (ie, in the second tertile) have been overlooked.

Conclusions

Taken together, our findings suggest that CSF NfL is a valuable biomarker of early neurodegeneration but not specific for AD pathobiology. In the newly published revisions of the 2018 National Institute on Aging–Alzheimer Association criteria,32 NfL is discussed as a neurodegeneration marker in the N+ group. These data suggest that NfL may be the preferred N+ CSF biomarker in the A/T/N classification.

Back to top
Article Information

Accepted for Publication: August 24, 2018.

Corresponding Author: Michelle M. Mielke, PhD, Division Epidemiology, Department of Health Science Research, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (mielke.michelle@mayo.edu).

Published Online: November 12, 2018. doi:10.1001/jamaneurol.2018.3459

Author Contributions: Dr Mielke had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Zetterberg, Blennow, Jack, Petersen, Mielke.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Kern, Mielke.

Critical revision of the manuscript for important intellectual content: Kern, Syrjanen, Blennow, Zetterberg, Skoog, Waern, Hagen, Van Harten, Knopman, Jack, Petersen.

Statistical analysis: Kern, Syrjanen, Hagen, Mielke.

Obtained funding: Blennow, Zetterberg, Waern, Jack, Petersen, Mielke.

Administrative, technical, or material support: Zetterberg, Blennow, Jack, Petersen, Mielke.

Supervision: Blennow, Skoog, Mielke.

Conflict of Interest Disclosures: Dr Blennow has served as a consultant on advisory boards for Alzheon, BioArctic, Biogen, Eli Lilly, Fujirebio Europe, IBL International, Merck, Novartis, Pfizer, and Roche Diagnostics and is a cofounder of Brain Biomarker Solutions in Gothenburg AB, a GU Venture–based platform company at the University of Gothenburg. Dr Zetterberg has served on advisory boards for Eli Lilly, Roche Diagnostics, and Wave; has received travel grants from Teva; and is a cofounder of Brain Biomarker Solutions. Dr Skoog has been an advisor and speaker for Takeda. Dr Knopman serves on a data safety monitoring board for Lundbeck Pharmaceuticals and for the Dominantly Inherited Alzheimer Network study; is an investigator in clinical trials sponsored by Biogen, Lilly Pharmaceuticals, and the Alzheimer’s Disease Cooperative Study; and receives research support from the National Institutes of Health (NIH). Dr Jack has provided consulting services for Eli Lilly and receives research funding from the NIH (grants R01 AG011378, U01 HL096917, U01 AG024904, R01 AG041851, R01 AG037551, R01 AG043392, and U01 AG006786) and the Alexander Family Alzheimer's Disease Research Professorship of the Mayo Clinic. Dr Petersen is a consultant for Roche Inc, Merck Inc, Biogen Inc, Genentech, Inc, and GE Healthcare. Dr Mielke served as a consultant to Eli Lilly and Lysosomal Therapeutics, Inc and receives research support from the NIH (grants R01 AG49704, P50 AG44170, U01 AG06786, and RF1 AG55151), the US Department of Defense (grant W81XWH-15-1), and unrestricted research grants from Biogen and Lundbeck.

Funding/Support: This study was supported by funding from the NIH/National Institute on Aging grants U01 AG006786, R01 AG011378, R01 AG041851, and R01 AG049704; the GHR Foundation; and Roche Diagnostics Ltd. The study was made possible by the Rochester Epidemiology Project (grant R01 AG034676). The cerebrospinal fluid neurofilament light protein and neurogranin analyses were supported by grants from The Swedish Research Council (2013-2546), the European Research Council (681712), the Knut and Alice Wallenberg Foundation, Torsten Söderbergs Foundation at the Royal Swedish Academy of Sciences, Sahlgrenska University Hospital (ALF), Sahlgrenska University Hospital (ALFGBG-716681, ALFGBG-715841, ALFGBG-65930, and Higher Medical ALF-position), Torsten Söderberg Foundation, Swedish Alzheimer Foundation, and Hjärnfonden.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

References
1.
Scheltens  P, Blennow  K, Breteler  MM,  et al.  Alzheimer’s disease.  Lancet. 2016;388(10043):505-517. doi:10.1016/S0140-6736(15)01124-1PubMedGoogle ScholarCrossref
2.
Blennow  K, Hampel  H, Weiner  M, Zetterberg  H.  Cerebrospinal fluid and plasma biomarkers in Alzheimer disease.  Nat Rev Neurol. 2010;6(3):131-144. doi:10.1038/nrneurol.2010.4PubMedGoogle ScholarCrossref
3.
Lewczuk  P, Riederer  P, O’Bryant  SE,  et al; Members of the WFSBP Task Force Working on this Topic: Peter Riederer, Carla Gallo, Dimitrios Kapogiannis, Andrea Lopez Mato, Florence Thibaut.  Cerebrospinal fluid and blood biomarkers for neurodegenerative dementias: an update of the Consensus of the Task Force on Biological Markers in Psychiatry of the World Federation of Societies of Biological Psychiatry.  World J Biol Psychiatry. 2018;19(4):244-328. doi:10.1080/15622975.2017.1375556PubMedGoogle ScholarCrossref
4.
Norgren  N, Rosengren  L, Stigbrand  T.  Elevated neurofilament levels in neurological diseases.  Brain Res. 2003;987(1):25-31. doi:10.1016/S0006-8993(03)03219-0PubMedGoogle ScholarCrossref
5.
Hoffman  PN, Cleveland  DW, Griffin  JW, Landes  PW, Cowan  NJ, Price  DL.  Neurofilament gene expression: a major determinant of axonal caliber.  Proc Natl Acad Sci U S A. 1987;84(10):3472-3476. doi:10.1073/pnas.84.10.3472PubMedGoogle ScholarCrossref
6.
Zetterberg  H, Skillbäck  T, Mattsson  N,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Association of cerebrospinal fluid neurofilament light concentration with Alzheimer disease progression.  JAMA Neurol. 2016;73(1):60-67. doi:10.1001/jamaneurol.2015.3037PubMedGoogle ScholarCrossref
7.
Kvartsberg  H, Portelius  E, Andreasson  U,  et al.  Characterization of the postsynaptic protein neurogranin in paired cerebrospinal fluid and plasma samples from Alzheimer’s disease patients and healthy controls.  Alzheimers Res Ther. 2015;7(1):40. doi:10.1186/s13195-015-0124-3PubMedGoogle ScholarCrossref
8.
Mattsson  N, Insel  PS, Palmqvist  S,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Cerebrospinal fluid tau, neurogranin, and neurofilament light in Alzheimer’s disease.  EMBO Mol Med. 2016;8(10):1184-1196. doi:10.15252/emmm.201606540PubMedGoogle ScholarCrossref
9.
Pereira  JB, Westman  E, Hansson  O; Alzheimer’s Disease Neuroimaging Initiative.  Association between cerebrospinal fluid and plasma neurodegeneration biomarkers with brain atrophy in Alzheimer’s disease.  Neurobiol Aging. 2017;58:14-29. doi:10.1016/j.neurobiolaging.2017.06.002PubMedGoogle ScholarCrossref
10.
Kester  MI, Teunissen  CE, Crimmins  DL,  et al.  Neurogranin as a cerebrospinal fluid biomarker for synaptic loss in symptomatic Alzheimer disease.  JAMA Neurol. 2015;72(11):1275-1280. doi:10.1001/jamaneurol.2015.1867PubMedGoogle ScholarCrossref
11.
Wellington  H, Paterson  RW, Portelius  E,  et al.  Increased CSF neurogranin concentration is specific to Alzheimer disease.  Neurology. 2016;86(9):829-835. doi:10.1212/WNL.0000000000002423PubMedGoogle ScholarCrossref
12.
Portelius  E, Olsson  B, Höglund  K,  et al.  Cerebrospinal fluid neurogranin concentration in neurodegeneration: relation to clinical phenotypes and neuropathology.  Acta Neuropathol. 2018;136(3):363-376. doi:10.1007/s00401-018-1851-xPubMedGoogle ScholarCrossref
13.
Roberts  RO, Geda  YE, Knopman  DS,  et al.  The Mayo Clinic Study of Aging: design and sampling, participation, baseline measures and sample characteristics.  Neuroepidemiology. 2008;30(1):58-69. doi:10.1159/000115751PubMedGoogle ScholarCrossref
14.
St Sauver  JL, Grossardt  BR, Yawn  BP,  et al.  Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system.  Int J Epidemiol. 2012;41(6):1614-1624. doi:10.1093/ije/dys195PubMedGoogle ScholarCrossref
15.
Ivnik  RJ, Malec  JF, Smith  GE,  et al.  Mayo’s older Americans normative studies: updated AVLT norms for ages 56 to 97.  Clin Neuropsychol. 1992;6(suppl):83-104. doi:10.1080/13854049208401880Google ScholarCrossref
16.
Petersen  RC.  Mild cognitive impairment as a diagnostic entity.  J Intern Med. 2004;256(3):183-194. doi:10.1111/j.1365-2796.2004.01388.xPubMedGoogle ScholarCrossref
17.
American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). 4th ed. Washington, DC: American Psychiatric Association; 1994.
18.
Gaetani  L, Höglund  K, Parnetti  L,  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):8. doi:10.1186/s13195-018-0339-1PubMedGoogle ScholarCrossref
19.
Portelius  E, Zetterberg  H, Skillbäck  T,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Cerebrospinal fluid neurogranin: relation to cognition and neurodegeneration in Alzheimer’s disease.  Brain. 2015;138(pt 11):3373-3385. doi:10.1093/brain/awv267PubMedGoogle ScholarCrossref
20.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8PubMedGoogle ScholarCrossref
21.
Rocca  WA, Yawn  BP, St Sauver  JL, Grossardt  BR, Melton  LJ  III.  History of the Rochester Epidemiology Project: half a century of medical records linkage in a US population.  Mayo Clin Proc. 2012;87(12):1202-1213. doi:10.1016/j.mayocp.2012.08.012PubMedGoogle ScholarCrossref
22.
Crowson  CS, Atkinson  EJ, Therneau  TM.  Assessing calibration of prognostic risk scores.  Stat Methods Med Res. 2016;25(4):1692-1706. doi:10.1177/0962280213497434PubMedGoogle ScholarCrossref
23.
Zetterberg  H.  Tauomics and kinetics in human neurons and biological fluids.  Neuron. 2018;97(6):1202-1205. doi:10.1016/j.neuron.2018.02.030PubMedGoogle ScholarCrossref
24.
Skillbäck  T, Rosén  C, Asztely  F, Mattsson  N, Blennow  K, Zetterberg  H.  Diagnostic performance of cerebrospinal fluid total tau and phosphorylated tau in Creutzfeldt-Jakob disease: results from the Swedish Mortality Registry.  JAMA Neurol. 2014;71(4):476-483. doi:10.1001/jamaneurol.2013.6455PubMedGoogle ScholarCrossref
25.
van Harten  AC, Smits  LL, Teunissen  CE,  et al.  Preclinical AD predicts decline in memory and executive functions in subjective complaints.  Neurology. 2013;81(16):1409-1416. doi:10.1212/WNL.0b013e3182a8418bPubMedGoogle ScholarCrossref
26.
Vos  SJ, Xiong  C, Visser  PJ,  et al.  Preclinical Alzheimer’s disease and its outcome: a longitudinal cohort study.  Lancet Neurol. 2013;12(10):957-965. doi:10.1016/S1474-4422(13)70194-7PubMedGoogle ScholarCrossref
27.
Okonkwo  OC, Alosco  ML, Griffith  HR,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Cerebrospinal fluid abnormalities and rate of decline in everyday function across the dementia spectrum: normal aging, mild cognitive impairment, and Alzheimer disease.  Arch Neurol. 2010;67(6):688-696. doi:10.1001/archneurol.2010.118PubMedGoogle ScholarCrossref
28.
Buchhave  P, Minthon  L, Zetterberg  H, Wallin  AK, Blennow  K, Hansson  O.  Cerebrospinal fluid levels of β-amyloid 1-42, but not of tau, are fully changed already 5 to 10 years before the onset of Alzheimer dementia.  Arch Gen Psychiatry. 2012;69(1):98-106. doi:10.1001/archgenpsychiatry.2011.155PubMedGoogle ScholarCrossref
29.
Roe  CM, Fagan  AM, Grant  EA,  et al.  Amyloid imaging and CSF biomarkers in predicting cognitive impairment up to 7.5 years later.  Neurology. 2013;80(19):1784-1791. doi:10.1212/WNL.0b013e3182918ca6PubMedGoogle ScholarCrossref
30.
van Rossum  IA, Vos  SJ, Burns  L,  et al.  Injury markers predict time to dementia in subjects with MCI and amyloid pathology.  Neurology. 2012;79(17):1809-1816. doi:10.1212/WNL.0b013e3182704056PubMedGoogle ScholarCrossref
31.
Headley  A, De Leon-Benedetti  A, Dong  C,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Neurogranin as a predictor of memory and executive function decline in MCI patients.  Neurology. 2018;90(10):e887-e895. doi:10.1212/WNL.0000000000005057PubMedGoogle ScholarCrossref
32.
Jack  CR  Jr, Bennett  DA, Blennow  K,  et al; Contributors.  NIA-AA Research Framework: toward a biological definition of Alzheimer’s disease.  Alzheimers Dement. 2018;14(4):535-562. doi:10.1016/j.jalz.2018.02.018PubMedGoogle ScholarCrossref
×