Diagnostic Value of Cerebrospinal Fluid Neurofilament Light Protein in Neurology:A Systematic Review and Meta-analysis

Key PointsQuestionHow do levels of neurofilament light in cerebrospinal fluid (cNfL) compare between neurological conditions and with healthy controls? FindingsAmong 10 059 individuals in this systematic review and meta-analysis, cNfL was elevated in most neurological conditions compared with healthy controls, and the magnitude of the increase varies extensively. Although cNfL overlaps between most clinically similar conditions, its distribution did not overlap in frontotemporal dementia and other dementias or in Parkinson disease and atypical parkinsonian syndromes. MeaningThe cNfL is a marker of neuronal damage and may be useful to differentiate some clinically similar conditions, such as frontotemporal dementia from Alzheimer disease and Parkinson disease from atypical parkinsonian syndromes.This systematic review and meta-analysis assesses the associations of age, sex, and diagnosis with neurofilament light in cerebrospinal fluid and evaluates its potential in discriminating clinically similar conditions.ImportanceNeurofilament light protein (NfL) is elevated in cerebrospinal fluid (CSF) of a number of neurological conditions compared with healthy controls (HC) and is a candidate biomarker for neuroaxonal damage. The influence of age and sex is largely unknown, and levels across neurological disorders have not been compared systematically to date. ObjectivesTo assess the associations of age, sex, and diagnosis with NfL in CSF (cNfL) and to evaluate its potential in discriminating clinically similar conditions. Data SourcesPubMed was searched for studies published between January 1, 2006, and January 1, 2016, reporting cNfL levels (using the search terms neurofilament light and cerebrospinal fluid) in neurological or psychiatric conditions and/or in HC. Study SelectionStudies reporting NfL levels measured in lumbar CSF using a commercially available immunoassay, as well as age and sex. Data Extraction and SynthesisIndividual-level data were requested from study authors. Generalized linear mixed-effects models were used to estimate the fixed effects of age, sex, and diagnosis on log-transformed NfL levels, with cohort of origin modeled as a random intercept. Main Outcome and MeasureThe cNfL levels adjusted for age and sex across diagnoses. ResultsData were collected for 10059 individuals (mean [SD] age, 59.7 [18.8] years; 54.1% female). Thirty-five diagnoses were identified, including inflammatory diseases of the central nervous system (n=2795), dementias and predementia stages (n=4284), parkinsonian disorders (n=984), and HC (n=1332). The cNfL was elevated compared with HC in a majority of neurological conditions studied. Highest levels were observed in cognitively impaired HIV-positive individuals (iHIV), amyotrophic lateral sclerosis, frontotemporal dementia (FTD), and Huntington disease. In 33.3% of diagnoses, including HC, multiple sclerosis, Alzheimer disease (AD), and Parkinson disease (PD), cNfL was higher in men than women. The cNfL increased with age in HC and a majority of neurological conditions, although the association was strongest in HC. The cNfL overlapped in most clinically similar diagnoses except for FTD and iHIV, which segregated from other dementias, and PD, which segregated from atypical parkinsonian syndromes. Conclusions and RelevanceThese data support the use of cNfL as a biomarker of neuroaxonal damage and indicate that age-specific and sex-specific (and in some cases disease-specific) reference values may be needed. The cNfL has potential to assist the differentiation of FTD from AD and PD from atypical parkinsonian syndromes.