Frontotemporal dementia (FTD) is a heterogeneous group of neurodegenerative conditions affecting the frontal and temporal lobes and, to a varying degree, the subcortical grey matter. In the 1998 consensus document, FTD was divided into a number of clinical subtypes: behavioural variant (bvFTD) and progressive aphasias, semantic dementia (SD) and progressive non-fluent aphasia (PNFA) . There is a close association, both clinically and neuropathologically, with corticobasal degeneration (CBD) and progressive supranuclear palsy (PSP); thus these diagnoses are now generally considered part of the FTD spectrum .
There is considerable overlap between the different subtypes of FTD, both clinically and neuropathologically. A major challenge in diagnosing FTD is the difficulty in predicting a neuropathological subtype based on clinical characteristics. With reference to protein pathology, cases of FTD have previously been recognised as tau-positive or tau-negative. Among the tau-positive pathologies, cases of Pick’s disease and FTD with tau-positive inclusions have been found. The majority of tau-negative cases have been ubiquitin-positive. In 2006, TDP-43 was identified in the ubiquitin-positive inclusions in the majority of these cases  and, recently, FUS-positive pathology was identified in most of the remaining cases [4, 5]. The remaining small group exhibits a spectrum of variable morphological factors [6, 7]. There is not always a clear congruence between the clinical and neuropathological subtypes of FTD. While bvFTD can reflect any of the neuropathological subtypes, SD almost exclusively shows TDP-43 pathology and PNFA most often shows tau-positive inclusions . FTD with concomitant motor neuron disease (FTD-MND) most often exhibits TDP-43 pathology.
In AD, the analysis of cerebrospinal fluid (CSF) biomarkers (with a typical pattern of elevated tau, phosphorylated tau [phospho-tau] and decreased beta-amyloid 42 [Aβ42]) is an acknowledged and promising tool in the diagnosis of Alzheimer’s disease . However, there are no reliable CSF biomarkers available for FTD in general or for the different FTD subtypes. A previous study demonstrated that CSF levels of TDP-43 were slightly increased in amyotrophic lateral sclerosis (ALS) compared with controls , but its role in the clinical diagnosis of FTD is yet to be determined.
Neurofilament proteins, when prevalent in the CSF, are considered to reflect neuronal degeneration. These intermediate filaments form a major cytoskeletal constituent of neurons and they comprise the neurofilament heavy (NFH), intermediate (NFM) and light (NFL) chain proteins, the latter being a non-phosphorylated form.
NFL levels are slightly augmented in healthy aging individuals and have been correlated with increasing age . In acute cerebral infarction, very high NFL levels can be seen. The levels are higher in VaD than in AD . A recent study  showed that NFL levels were significantly higher in atypical parkinsonian disorders (CBD and multiple system atrophy) compared with Parkinson’s disease; whereas NFL levels in Lewy body dementia were intermediate. The suggested explanation for these differences was that patients with atypical parkinsonian disorders experienced a greater loss of neurons over a shorter period of time. NFL levels are also increased in individuals with white matter disease (WMD) . WMD is common in patients with dementia (AD, among other diseases), but can also be seen in healthy older adults . WMD is indicated by hyperintensities visualised during life as punctuate or confluent changes in the periventricular regions or deep white matter on computerized tomography or magnetic resonance imaging scans [15, 16].
Some studies have shown that CSF levels of NFL are higher in FTD compared with both early-onset AD and controls [17, 18]. In ALS it has also been shown that NFL is often considerably increased compared with controls [19, 20], and elevated levels of NFL correlate with shorter survival .
The aim of this study was to investigate whether existing CSF biochemical markers, in particular NFL, can assist in the differential diagnostics of frontotemporal dementias.
Can biomarkers help us differentiate between clinical and/or neuropathological subtypes of frontotemporal dementias? Furthermore, is the severity of the disease post mortem related to in vivo NFL levels?