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.
This population-based study examines the association of cerebrospinal fluid neurofilament light protein and neurogranin levels with the risk of mild cognitive impairment among participants in the Mayo Clinic Study of Aging.
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 Ng8,9,10 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).
Table 1. Baseline Participant Characteristics.
| Characteristics | Total No. | Median (IQR) |
|---|---|---|
| Age, y | 648 | 72.3 (63.4 to 78.3) |
| Male, No. (%) | 648 | 366 (56.5) |
| Education | 648 | 14.0 (12.0 to 16.0) |
| APOE E4, No. (%) | 647 | 172 (26.6) |
| Hypertension, No. (%) | 642 | 386 (60.1) |
| BMI | 642 | 28.2 (25.2 to 31.6) |
| Charlson comorbidity index | 642 | 2.0 (1.0 to 4.0) |
| Cognitive z scores | ||
| Global | 626 | 0.05 (−0.65 to 0.73) |
| Memory | 645 | 0.05 (−0.76 to 0.69) |
| Visual spatial | 632 | 0.03 (−0.60 to 0.69) |
| Attention | 635 | 0.09 (−0.68 to 0.73) |
| Language | 638 | 0.11 (−0.62 to 0.73) |
| CSF measures | ||
| NfL, pg/mL | 639 | 484.7 (359.5 to 700.9) |
| Neurogranin, pg/mL | 647 | 164.7 (131.0 to 217.5) |
| Total tau | 647 | 211.3 (166.1 to 269.7) |
| Phosphorylated tau | 647 | 17.8 (14.1 to 23.3) |
| Aβ42 | 647 | 1105.0 (786.2 to 1537.0) |
Abbreviations: Aβ42, amyloid-β; APOE, apolipoprotein E; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CSF, cerebrospinal fluid; IQR, interquartile range; NfL, neurofilament light.
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.
Table 2. Continuous CSF Measures and Risk of Mild Cognitive Impairment.
| z Log CSF | Model 1a | Model 2b | ||||||
|---|---|---|---|---|---|---|---|---|
| Total No. | Events | HR (95% CI) | P Value | Total No. | Events | HR (95% CI) | P Value | |
| NfL | 633 | 94 | 1.32 (1.08-1.60) | .01 | 632 | 94 | 1.33 (1.09-1.62) | .01 |
| Neurogranin | 641 | 95 | 1.03 (0.83-1.29) | .79 | 640 | 95 | 1.03 (0.83-1.27) | .81 |
| Total tau | 641 | 95 | 1.07 (0.85-1.34) | .58 | 640 | 95 | 1.06 (0.85-1.33) | .59 |
| P-tau | 641 | 95 | 1.12 (0.90-1.39) | .30 | 640 | 95 | 1.10 (0.89-1.37) | .37 |
| Aβ42 | 641 | 95 | 0.74 (0.61-0.89) | .001 | 640 | 95 | 0.80 (0.66-0.97) | .02 |
Abbreviations: Aβ42, amyloid-β; APOE, apolipoprotein E; CSF, cerebrospinal fluid; HR, hazard ratio; NfL, neurofilament light; P-tau, phosphorylated tau.
Model 1 with age as time scale and adjusted for sex, education, and the Charlson Comorbidity index.
Model 2 with age as time scale and adjusted for the variables in model 1 and APOE genotype
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).
Table 3. CSF Measures in Quartiles and Risk of Mild Cognitive Impairment.
| Quartiles of Log CSF | Model 1a | Model 2b | ||||||
|---|---|---|---|---|---|---|---|---|
| Total No. | Events | HR (95% CI) | P Value | Total No. | Events | HR (95% CI) | P Value | |
| NfL | ||||||||
| 2 vs 1 | 633 | 94 | 1.52 (0.64-3.64) | .35 | 632 | 94 | 1.69 (0.71-4.03) | .24 |
| 3 vs 1 | 1.43 (0.60-3.42) | .42 | 1.58 (0.66-3.75) | .30 | ||||
| 4 vs 1 | 2.90 (1.26-6.67) | .01 | 3.13 (1.36-7.18) | .01 | ||||
| Neurogranin | ||||||||
| 2 vs 1 | 641 | 95 | 0.82 (0.45-1.47) | .50 | 640 | 95 | 0.82 (0.45-1.48) | .51 |
| 3 vs 1 | 0.79 (0.43-1.43) | .44 | 0.75 (0.41-1.37) | .35 | ||||
| 4 vs 1 | 0.89 (0.50-1.59) | .70 | 0.88 (0.49-1.56) | .65 | ||||
| Total tau | ||||||||
| 2 vs 1 | 641 | 95 | 1.08 (0.54-2.16) | .82 | 640 | 95 | 1.19 (0.59-2.38) | .63 |
| 3 vs 1 | 0.71 (0.35-1.44) | .34 | 0.77 (0.38-1.57) | .48 | ||||
| 4 vs 1 | 1.13 (0.58-2.20) | .71 | 1.14 (0.59-2.21) | .70 | ||||
| P-tau | ||||||||
| 2 vs 1 | 641 | 95 | 1.88 (0.87-4.10) | .11 | 640 | 95 | 1.96 (0.90-4.27) | .09 |
| 3 vs 1 | 1.56 (0.72-3.36) | .26 | 1.62 (0.75-3.50) | .22 | ||||
| 4 vs 1 | 1.72 (0.81-3.68) | .16 | 1.67 (0.78-3.55) | .19 | ||||
| Aβ42 | ||||||||
| 2 vs 1 | 641 | 95 | 0.65 (0.37-1.13) | .13 | 640 | 95 | 0.74 (0.42-1.30) | .30 |
| 3 vs 1 | 0.41 (0.23-0.75) | .003 | 0.48 (0.26-0.88) | .02 | ||||
| 4 vs 1 | 0.47 (0.27-0.81) | .01 | 0.58 (0.33-1.02) | .06 | ||||
Abbreviations: Aβ42, amyloid-β; APOE, apolipoprotein E; CSF, cerebrospinal fluid; HR, hazard ratio; NfL, neurofilament light; P-tau, phosphorylated tau.
Model 1 with age as time scale and adjusted for sex, education, and the Charlson Comorbidity index.
Model 2 with age as time scale and adjusted for the variables in model 1 and APOE genotype.
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.
Table 4. Combination of CSF Neurofilament Light Protein and T-tau for Risk of Mild Cognitive Impairment.
| Quartiles of Log CSF | Model 1 (n = 632)a | Model 2 (n = 631)b | ||||
|---|---|---|---|---|---|---|
| Events | HR (95% CI) | P Value | Events | HR (95% CI) | P Value | |
| Neither T-tau nor NfL in the top quartile | 94 | 1 [Reference] | NA | 94 | 1 [Reference] | NA |
| T-tau in the top quartile | 1.28 (0.68-2.41) | .44 | 1.23 (0.65-2.31) | .53 | ||
| NfL in the top quartile | 2.24 (1.31-3.83) | .003 | 2.29 (1.34-3.94) | .003 | ||
| NfL and T-tau in the top quartile | 2.29 (1.28-4.09) | .01 | 2.12 (1.18-3.81) | .01 | ||
Abbreviations: APOE, apolipoprotein E; CSF, cerebrospinal fluid; HR, hazard ratio; NA, not applicable; NfL, neurofilament light; T-tau, total tau.
Model 1 with age as time scale and adjusted for sex, education, and the Charlson Comorbidity index.
Model 2 with age as time scale and adjusted for the variables in model 1 and APOE genotype.
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.28,29,30
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.
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)01124-1 [DOI] [PubMed] [Google Scholar]
- 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.2010.4 [DOI] [PubMed] [Google Scholar]
- 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: 10.1080/15622975.2017.1375556 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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)03219-0 [DOI] [PubMed] [Google Scholar]
- 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.84.10.3472 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.2015.3037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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/s13195-015-0124-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.201606540 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.2017.06.002 [DOI] [PubMed] [Google Scholar]
- 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.2015.1867 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.0000000000002423 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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: 10.1159/000115751 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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/dys195 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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: 10.1080/13854049208401880 [DOI] [Google Scholar]
- 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.x [DOI] [PubMed] [Google Scholar]
- 17.American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). 4th ed Washington, DC: American Psychiatric Association; 1994. [Google Scholar]
- 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/s13195-018-0339-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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/awv267 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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)90171-8 [DOI] [PubMed] [Google Scholar]
- 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.2012.08.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Crowson CS, Atkinson EJ, Therneau TM. Assessing calibration of prognostic risk scores. Stat Methods Med Res. 2016;25(4):1692-1706. doi: 10.1177/0962280213497434 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zetterberg H. Tauomics and kinetics in human neurons and biological fluids. Neuron. 2018;97(6):1202-1205. doi: 10.1016/j.neuron.2018.02.030 [DOI] [PubMed] [Google Scholar]
- 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.2013.6455 [DOI] [PubMed] [Google Scholar]
- 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.0b013e3182a8418b [DOI] [PubMed] [Google Scholar]
- 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)70194-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.2010.118 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.2011.155 [DOI] [PubMed] [Google Scholar]
- 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.0b013e3182918ca6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.0b013e3182704056 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.0000000000005057 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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.2018.02.018 [DOI] [PMC free article] [PubMed] [Google Scholar]