P1‐247: CEREBROSPINAL FLUID NEUROFILAMENT LIGHT PROTEIN AS A DIFFERENTIAL DIAGNOSIS BIOMARKER IN NEUROLOGICAL DISEASES: A SYSTEMATIC REVIEW AND METAANALYSIS

Multiple studies report elevated levels of neuron-specific cytoskeletal protein neurofilament light (NfL) in cerebrospinal fluid (CSF) of individuals with a variety of neurological conditions compared to healthy controls (HC). In addition, a positive association between CSF NfL and age has been reported in HC. These findings raise the question of the clinical potential of CSF NfL to discriminate individual neurological conditions, and whether this potential is affected by age or gender. We aimed to investigate the potentially modifying influence of age and gender on CSF NfL levels and to study its relevance as a differential diagnosis biomarker. CSF NfL studies listed in PubMed from January 1st 2006 to December 31st 2015. Raw data for 9513 participants from 47 datasets were pooled and classified according to diagnosis, resulting in 35 diagnostic groups including HC. Fixed effects of diagnosis, age and gender and their interactions on CSF NfL levels were assessed using mixed effects models. CSF NfL levels increase with age in HC (b=0.0338, p<0.0001) and a subset of diagnoses including Parkinson's disease (PD, b=0.0374, p<0.0001), Alzheimer's disease (AD, b=0.0179, p<0.0001), vascular dementia (VaD, b=0.0241, p=0.0003), stroke (b=0.0512, p=0.0085), optic neuritis (ON, b=0.0190, p=0.0063), inflammatory neurological diseases of the CNS other than multiple sclerosis (MS) (IND, b=0.0296, p<0.0001) and bipolar disorder (BD, b=0.0346, p<0.0001). The highest mean CSF NfL levels compared to HC were observed in amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), relapsing remitting multiple sclerosis (RRMS) with or without treatment, AD, VaD, multiple system atrophy, progressive supranuclear palsy, and corticobasal syndrome (CBS) (p<0.0001). The distribution of CSF NfL according to age in FTD segregated from other dementias including AD in individuals younger than 75 years old. The positive association between CSF NfL levels and age observed in HC indicates age-specific reference values will be needed. CSF NfL distributions overlap in several clinically-related diseases. Exceptions include the distributions of CSF NfL in FTD compared to other dementias and atypical parkinsonian syndromes compared to PD, suggesting high potential for CSF NfL as a differential diagnosis biomarker in these situations.