Neurofilament Light as a Dementia Biomarker

A Compelling Need

The success in generating therapies for neurodegeneration is contingent upon identifying effective diagnostic and prognostic biomarkers. Therapy discovery in dementia and the majority of neurodegenerative diseases has been challenged by difficulty confirming a change to underlying pathologic processes. Without biomarkers, there is no ability to monitor if a drug is reaching its target and exerting its hypothesized effect. Blood (ie, serum or plasma) biomarkers are desirable tools in dementia research and clinical care because, compared with established neuroimaging tools and cerebrospinal fluid (CSF) assays, they are more feasible to collect for screening and serial evaluation. Blood testing is also less expensive and more easily deployable compared with neuroimaging and CSF analysis, which could affect earlier disease detection and development of surrogate markers of pharmacodynamics.

Although blood biomarkers have been identified as a compelling need for a long time, sensitivity limitations of available analytic techniques have been an obstacle to their development (See Blood Tests for Alzheimer Disease in this issue). A recently approved tau PET may also be useful as an AD staging tool (See Molecular Imaging Biomarkers in Dementia in this issue).¹ APOE genotype testing may be useful to counsel individuals regarding risks of AD and the use of antiamyloid immunotherapies.² Gene testing is also available for the rare autosomal dominant forms of AD and frontotemporal dementia (FTD). These tools are insufficient, however, because they are either invasive, expensive, or not sensitive enough. It is expected that the nascent field of therapeutics in dementia will benefit from the availability of blood biomarkers, similar to what has occurred in other medical fields, in which blood biomarkers aid in risk determination, diagnosis, staging, therapy response prediction, treatment efficacy measurement, progression and recurrence monitoring, and treatment adherence assessment (Table).

Neurofilament Light

Neurofilament light chain (NfL) is a neuron-specific intermediate filament protein that has recently emerged as a biomarker of neuronal injury with enormous clinical potential. Neurofilaments are the most abundant structural proteins in vertebrate myelinated axons,³ determining axon caliber. Neurofilaments consist of a head, central rod region, and tail. The size of the tail classifies the neurofilament as light, medium, or heavy chain. Neurofilament polymers provide cylindrical scaffolding along axon segments,⁴ and dynamically interact with other cellular components involved in processes that are important to neuronal development and degeneration.

Early neuropathologic evidence linked neurofilament abnormalities to neurologic disease. Mutations in the neurofilament gene cause type 1 and 2 forms of hereditary axonopathies (eg, Charcot-Marie-Tooth disease) that feature giant axons, organelle accumulations, and disorganized neurofilament aggregates.⁵ Toxicologic animal models also suggested that an accumulation of disorganized neurofilaments underlies the chronic sensorimotor toxic axonopathies associated with organic solvent exposure in humans.⁶ In amyotrophic lateral sclerosis (ALS), motoneurons accumulate neuroaxonal spheroids rich in intermediate filament proteins.⁷ Similarly, neuronal intermediate filament inclusion disease was identified as the neuropathologic substrate in some cases of FTD with pyramidal or extrapyramidal features; this was in the absence of tau of TDP-43 immunoreactivity, later linked to immunoreactivity to fused-in-sarcoma.⁸

NfL elevation was initially observed in CSF of people with dementia, including AD, FTD, and vascular dementia.⁹ A small case series suggested CSF NfL could help distinguish AD from FTD, especially when combined with traditional CSF AD biomarkers.¹⁰ A single analyte immunoassay for CSF NfL was commercialized for research purposes, and subsequent and more sophisticated studies confirmed that CSF NfL correlates with disease severity and brain injury detected by neuroimaging. These studies also confirmed that CSF NfL is elevated in degenerative and nondegenerative neurologic disorders. A revolution in the field occurred with the introduction of an ultrasensitive immunoassay that allowed detection of sub-fentomolar concentrations of proteins in serum and plasma.¹¹ Measurements of NfL concentrations in animal models and in humans affected by different neurologic conditions (eg, stroke, traumatic brain injury, HIV infection, autoimmune inflammatory polyradiculoneuropathy, multiple sclerosis [MS], AD, FTD, Huntington disease [HD], and prion disease) confirmed blood and CSF NfL concentrations are highly correlated (ie, correlation coefficients ranging from 0.6 to 0.8), and NfL is a sensitive, but nonspecific marker of neuronal injury.¹²

Potential Clinical Value of NfL

The Effect of Age

NfL can be detected in CSF and blood in individuals without cognitive impairment, and there is a strong linear relationship between measurable concentrations in biofluids and age.¹³ Normal NfL range values are not well established. Studies of the diagnostic value of NfL in CSF and blood have generated cohort-specific range values, but the utility of these ranges in the general population requires dedicated studies. NfL concentrations in CSF of individuals without cognitive impairment range from 150 pg/mL around age 20, to almost 1,000 pg/mL around age 80. In plasma, NfL concentrations range from 5 pg/mL around age 30, to 35 pg/ mL around age 80. Similar trends are observed when NfL is measured in serum, although plasma concentrations are about 75% of those measured in serum for the same person.¹⁴ Thus, plasma and serum NfL concentrations are highly correlated but not equivalent. A recent meta-analysis concluded there may be a need to establish normal concentration values for different age ranges.¹³ Importantly, the association between age and NfL concentrations is also observed in people with mild cognitive impairment (MCI), AD, dementia with Lewy bodies (DLB), and FTD, although levels typically become more elevated at earlier ages.^13,15 The relationship between NfL concentrations and age is relevant because although the general trend is a positive relationship between age and NfL concentrations, at the individual level, some with neurodegenerative diseases show decreases in NfL concentration over time. In addition, the relationship between age and NfL concentrations could be strong in early stages of the disease (ie, steeper slopes reflecting faster rates of increase in MCI and early AD compared with controls), or could plateau or be lost or even reversed (ie, increased age relating to lower NfL concentrations) in persistent conditions that lead to a critical state of neuron loss (eg, untreated relapsing-remitting MS, HIV central nervous system [CNS] infection, or advanced neurodegenerative diseases). The relationship between age and NfL concentrations in neurodegenerative conditions needs further exploration, and current comparisons between group NfL concentrations in neurodegenerative diseases should correct for age. A better understanding of the trajectory of blood NfL levels during the natural history of a disease is crucial to establish normal values and make use of potential clinical applications.

Risk Determination

In the future, blood NfL concentrations may be useful for detecting dementia risk. A hint to this potential application was first observed in people with MS. A study in military personnel showed individuals who eventually developed MS had higher median baseline serum NfL concentrations as long as 6 years before disease onset, compared with individuals who did not develop MS. Within-person increases in presymptomatic serum NfL concentrations of 5 pg/mL or more were associated with 7.5 times higher MS risk, whereas serum NfL concentrations of 25 pg/mL or more were associated with actual clinical onset.¹⁶ A key study in carriers of a rare autosomal dominant mutation in CHMPB2 that causes ALS showed that asymptomatic carriers of the mutation had higher CSF NfL concentrations compared to noncarriers from the same family, whereas symptomatic mutation carriers had the highest concentrations.¹⁷ High blood NfL concentrations were subsequently confirmed in asymptomatic PSEN1 mutation carriers at risk of autosomal dominant AD,¹⁸ and of more common mutations that cause FTD (eg, C9orf72, GRN, and MAPT) in people at high short-term risk of disease conversion, compared with noncarriers from the same families.¹⁹

Differential Diagnosis

Increasing evidence supports that blood NfL has great potential diagnostic utility in a number of clinical scenarios, including discriminating AD from FTD, Parkinson disease (PD) from other forms of atypical parkinsonism, and primary psychiatric disease from FTD or early-onset AD with psychiatric presentations.

AD vs FTD. Blood NfL is an excellent discriminator between AD or FTD and the absence of cognitive impairment. In plasma, a cutoff value of 20.3 pg/mL and 22.0 pg/ mL identified people with AD and FTD, respectively, compared with people without cognitive impairment. For AD, this level had 76.7% sensitivity and 95.5% specificity and the level for FTD had 87.6% sensitivity and 100% specificity. A cutoff value of 38.6 pg/mL discriminated AD from FTD with 76.7% sensitivity and 88.4% specificity. Although, discrimination between AD and FTD is less effective (sensitivity 70%, specificity 60%), the combination of NfL with other biomarkers could enhance the distinction between these 2 entities.²⁰ For example, an early study suggested that, in combination with currently available CSF AD biomarkers, CSF NfL could aid in distinguishing FTD and early-onset AD (EOAD) with those with EOAD having relatively lower NfL and high tau vs those with FTD having relatively higher NfL and lower tau.¹⁰

Idiopathic PD vs Atypical Parkinsonism. PD may be difficult to differentiate from atypical parkinsonism, especially early in the disease course, with a rate of clinical misdiagnosis as high as 15%. A consistent finding in several studies is that CSF and blood NfL concentrations in PD (and likely also in LBD, unless AD copathology exists) are not different from those of people without PD, whereas NfL concentrations are elevated in atypical forms of parkinsonism. In a recent study, a serum NfL cutoff value of 14.8 pg/ mL discriminated PD from atypical parkinsonisms (ie, multiple systems atrophy [MSA] and progressive supranuclear palsy [PSP]) with 86% sensitivity and 85% specificity, after adjustment for age differences between groups.²¹

Primary Psychiatric Disease vs Neurodegenerative Diseases With Psychiatric Presentations. CSF and blood NfL concentrations in primary psychiatric disease (eg, adjustment disorders, anxiety, conversion disorders, schizophrenia, major depressive disorder) are typically similar to age-matched people without psychiatric disease, but tend to be elevated in people with neurodegenerative diseases with predominant behavioral presentations (eg, AD, behavioral variant FTD, vascular dementia). The effects seem to be independent of age, considering that differences are also present in people with early-onset dementia, compared to those with no cognitive disorders.²²

Progression Monitoring

Evidence supporting the value of blood NfL to monitor disease progression is derived from several large observational cohorts of people with dementia. This evidence is supportive of stepwise increases in NfL concentrations from asymptomatic phase, to prodromal disease, to early dementia. In the Alzheimer’s Disease Neuroimaging Initiative (ADNI) cohort, for example, plasma NfL concentrations show trends for longitudinal intraindividual increases over time that correlate with declines in cognitive impairment, even after correction for age. People with amyloid-positive status have larger increases than those with amyloid-negative status, and the magnitude of increases in plasma NfL is contingent upon the phenotype at baseline. Larger increases in plasma NfL are present in AD, compared with MCI and in both AD and MCI compared with people without cognitive problems.¹⁵ Similarly, in HD, plasma NfL concentrations are higher in individuals with manifest HD, compared to premanifest HD and in manifest and premanifest HD compared with individuals without the huntinin mutation that causes HD.²³ In genetic forms of FTD, serum NfL concentrations are higher in fully symptomatic individuals, compared to those with prodromal disease or asymptomatic carriers, and mutation carriers show longitudinal increases in serum NfL that are more significant compared to those of noncarriers from the same families.¹⁹

Treatment Efficacy Measurement

Although there are no effective therapies against neuro-degenerative diseases yet, compelling evidence suggests plasma NfL concentrations can be lowered upon effective treatment and thus, NfL could constitute a marker of treatment response (ie, a theragnostic marker) when effective treatments against dementia are available. This application of blood NfL is of particular value in experimental neurotherapeutics for dementia. The most compelling evidence has emerged from MS research. For example, a reduction of about 50% in plasma NfL concentrations is observed after MS treatment with dimethyl fumarate, with reductions that are evident as early as 6 months after treatment begins.¹⁴ In HIV infection, effective treatment with antiretroviral therapy over 2 years is not only associated with decreased HIV viral load and increased CD4⁺ count but also with a steady decrease in CSF NfL.²⁴ Recent reports have also shown that effective treatment of spinal muscle atrophy with nusinersen was associated with clinical benefits and rapid decreases in CSF NfL concentrations.²⁵ The prospect of using NfL as a marker of therapy response in AD, FTD, HD, and other conditions in which treatments are actively being developed is very encouraging.

Conclusion

NfL measurement in plasma for clinical applications in neurodegenerative diseases is feasible. NfL assays are commercially available for research purposes only, although clinical use in the US via tests with Clinical Laboratory Improvement Amendments (CLIA) certification may be available soon. NfL is a sensitive, although nonspecific marker of neuronal injury. Considering that NfL is significantly elevated in both acute and chronic neurologic conditions compared with healthy individuals, that the age-concentration relationship and other clinical variables change with disease stage, and concentrations decrease upon effective treatment, NfL may be conceptualized as a marker of neuronal damage severity. NfL is a promising tool that may significantly affect dementia clinical care and research. Further research is needed, and novel and specific fluid biomarkers should be identified.