Background
Neurodegenerative disorders (NDDs) such as Alzheimer’s disease (AD), Parkinson’s disease (PD), and dementia with Lewy bodies (DLB) share a central pathogenic theme: the accumulation, in extra- or intracellular deposits, of aggregated and misfolded proteins [
1]. The type of protein, as well as the size, shape, and location of the deposits, is quite typical of each disorder and is used in the pathological examination for characterizing each disease. However, NDDs show remarkable similarities from the clinicopathological point of view, making accurate diagnosis difficult, especially at early stages of the disease [
2,
3]. For instance, the clinical presentation of AD and DLB, two diseases considered the most common neurodegenerative forms of dementia, may overlap significantly, leading to low accuracy of the differential diagnosis [
4]. Cognitive impairment can occur also in patients initially diagnosed with a prototypical movement disorder such as PD, generally at later stages of disease, and often leading to Parkinson’s disease with dementia (PDD). Apart from the temporal difference in the onset of cognitive deficits, PDD is remarkably similar to DLB in clinical terms, showing, beyond extrapyramidal signs, multidomain impairment and visual hallucinations [
5].
Besides clinical similarities, the co-occurrence of different protein aggregates is also a prominent molecular feature of NDDs. On one hand, inclusion bodies composed of α-synuclein (α-syn), representing the major pathological determinants in PD and DLB, can also be detected in AD brains, especially in selected areas such as the amygdala [
6]. On the other hand, tau and amyloid-β (Aβ) aggregation, considered the pathological hallmark of AD, is also found in DLB and PD brains, usually to different degrees [
7‐
9].
The presence of similar molecular signatures across NDDs is also detectable in cerebrospinal fluid (CSF) [
10]. CSF levels of the Aβ peptide 1–42 (Aβ
1–42) are generally reduced in both AD and DLB compared with control subjects [
11]. Patients with PD may show reduced CSF Aβ
1–42 levels as well, a decrease often associated with cognitive decline [
12,
13]. CSF α-syn is currently studied as a biomarker for PD and other synuclein-associated diseases, generally showing lower levels than control subjects and patients with AD [
14‐
18]. Also, tau proteins in CSF may show a partial overlap between AD and DLB, with phosphorylated tau 181 (p-tau) being the most useful for differential diagnosis [
19‐
21].
Other proteins have been evaluated across NDDs for their potential value in differential diagnosis. Among them, several studies underlined the importance of the fatty acid binding protein 3, heart type (FABP3), a small cytosolic protein involved in lipid transport. In the brain, FABP3 regulates the lipid composition of the membrane [
22], entailing possible roles in synapse formation [
23] and in the activity of cholinergic and glutamatergic neurons [
24]. Increased FABP3 levels were found in the CSF of patients with different neurological disorders, including Creutzfeldt-Jakob disease (CJD), AD in both its prodromal and dementia phases, and vascular dementia (VAD) [
25‐
31]. Furthermore, in CSF, FABP3 strongly correlates with tau, the prototypical marker of neurodegeneration [
30]. The role of FABP3 in NDDs related to α-syn aggregation is less defined. Increased levels of FABP3 have been found in the serum of patients diagnosed with DLB and PDD [
26]. Furthermore, FABP3 is highly expressed in mouse dopaminergic neurons, and its overexpression has been linked to α-syn aggregation and PD pathogenesis [
32].
Considering the different roles of FABP3, total α-syn, and AD core biomarkers across the AD-PD spectrum, we hypothesized that their combination would be of value for the differential diagnosis of NDDs. In this study, we measured this biomarker panel in a large, multicentric cohort composed of patients with AD, DLB, PDD, and PD compared with a group of subjects with other neurological diseases without dementia (OND). Additionally, we explored the associations of the CSF biomarker panel with cognitive decline and other clinical scores in the different diagnostic groups.
Discussion
In this study, we show that the combination of CSF biomarkers linked to different aspects of neurodegeneration may improve the characterization of NDDs, namely AD, DLB, PDD, and PD. In particular, we report the following findings:
1.
Tau protein and α-syn levels were significantly increased in patients with AD compared with the other diagnostic groups.
2.
FABP3 CSF levels were increased in the AD and DLB groups compared with the OND and PD groups.
3.
Combination of FABP3 with p-tau showed excellent performance in the discrimination between AD and DLB, whereas the inclusion of α-syn in the logistic models improved the discrimination of PDD and DLB from PD.
4.
FABP3 showed a significant inverse correlation with MMSE score in the whole cohort.
To our knowledge, this is the largest study analyzing the combination of FABP3 and α-syn with AD core biomarkers for differential diagnosis of NDDs, also including patients with PD without dementia. The core CSF AD biomarkers have become an important tool to support the diagnosis of AD and have been included in the new diagnostic criteria for AD [
33,
42]. However, these three biomarkers have not shown enough accuracy in the differential diagnosis of dementias [
43]. This is possibly related to the existence of a disease continuum across the neurodegeneration spectrum, at both the molecular and phenotypical levels [
2].
Our results show that tau proteins are fundamental biomarkers, not only for distinguishing patients with AD from those with OND but also in differential diagnosis across dementias. In our cohort, tau proteins showed a different degree of increase in dementia groups (AD > DLB > PDD). This trend has been found previously, with patients with AD generally showing higher CSF levels of tau proteins than patients with DLB and patients with PDD [
43‐
45]. On one hand, our data show a high discriminative power of p-tau in distinguishing patients with AD from those with DLB and patients with PDD, confirming previous results [
46,
47]. On the other hand, researchers in some studies have found substantial overlap of the CSF AD profile between subjects with DLB and AD [
11,
44,
48‐
50]. Because of the current uncertainty about the real potential of classical AD biomarkers in the differential diagnosis of NDDs, the inclusion of additional biomarkers with the core AD panel is mandatory in order to improve neurochemical dementia diagnostics.
α-Syn is currently studied for its possible value as a PD biomarker and in the differential diagnosis of NDDs [
51‐
53]. Researchers in previous studies found that α-syn species, also in combination with tau proteins, may be useful in improving diagnostic accuracy across dementia disorders, especially for DLB [
14,
49,
50]. In our cohort, total α-syn levels were not significantly changed between the OND group and the Lewy body disorders groups, showing only a trend toward reduction in patients with PDD and patients with DLB. However, we found significantly increased levels of α-syn in patients with AD, and the final models for differential diagnosis of PD and PDD vs. AD included α-syn. This finding confirms the potential role of α-syn as a biomarker also in AD, where α-syn CSF levels are usually increased compared with those of control subjects and patients with parkinsonism. The alteration of α-syn in AD has been linked to synaptic damage [
54] or to an underlying Lewy body pathology [
55,
56]. In our cohort, the strong correlation between α-syn and tau proteins without specific group differences may underline an association with neuronal damage, as recently reported [
57].
FABP3 measurement in serum and CSF has previously been tested as a biomarker for the differential diagnosis of NDDs, including DLB and PDD, in relatively small-scale studies [
26,
58]. In the present study, increased FABP3 CSF levels were linked to AD and DLB, whereas patients with PD showed levels similar to those of subjects with OND. The value of FABP3 as an AD biomarker was moderate compared with the core AD biomarkers, confirming the results of a recent meta-analysis [
59]. Previous studies have shown a high correlation between FABP3 and tau proteins in CSF [
30,
31], supporting the role of FABP3 as a neurodegeneration biomarker. This is confirmed by the parallel increase in FABP3 and tau in other conditions, such as CJD [
25], subarachnoid hemorrhage [
60], and VAD [
61]. However, some findings may endorse specific roles of FABP3 in AD dementia development, because elevated CSF FABP3 levels have been shown to correlate with atrophy of the entorhinal cortex and amyloid pathology in AD-vulnerable brain regions [
62]. The association with amyloid pathology was also found in our study, where CSF levels of FABP3 and tau proteins were increased in Aβ
1–42-positive subjects, similarly to a recent report [
63].
Patients with PD without dementia had levels of FABP3 similar to those of the OND group and were characterized only by reduced t-tau CSF levels. In a recent study, Bäckström and colleagues found that high levels of FABP3 and neurofilament light chain protein, together with low Aβ
1–42, were significantly associated with the development of dementia after 5–9 years of follow-up in a large cohort of patients with PD [
64]. In our study, FABP3 CSF levels were inversely correlated with MMSE scores in the whole cohort. The difference from the above-mentioned study may be due to the shorter follow-up available for patients with PD in our cohort (mean 5.2 months, maximum 1 year). Also, although dementia usually occurs in advanced phases [
65], not all patients with PD develop dementia along the disease course. In patients with PDD, CSF FABP3 levels showed a trend toward an increase and were inversely correlated with MMSE score. This evidence supports the idea that FABP3 is linked to the neurodegeneration process and cognitive impairment occurring at later stages [
26] and is more evident in patients with PDD. Furthermore, the lack of any association with motor and progression scores in patients with PD and patients with PDD may indirectly support the hypothesis of FABP3 as a degenerative marker not linked to pathogenic mechanisms specific to PD.
Despite the high predictive value of the biomarker combinations for the differentiation between AD and the other dementias, the five biomarkers we tested did not improve the distinction between patients with DLB and patients with PDD, with the best biomarker, α-syn, showing a relatively low discriminatory capability (AUC 0.63). This finding, together with the high correlation between the CSF profile of the five biomarkers between DLB and PDD (
r = 0.99), confirms the molecular and clinical similarities between these two conditions, which can be considered as a continuum across the pathogenesis of Lewy body disorders [
66].
Our study has some limitations. Some of the diagnostic groups were enrolled only in one center (AD, OND, and PD in Perugia), possibly introducing a source of variability in the results linked to CSF processing. However, the three clinics are experienced reference centers for CSF biomarker measurement and follow international guidelines for CSF collection [
37]. Another limitation is the heterogeneity of the control group, which was composed of different neurological disorders and not of healthy control subjects. Nonetheless, the OND group represents a population ordinarily assayed for CSF biomarkers in a neurology and memory clinic, thus exemplifying the use of CSF biomarkers in routine clinical practice. A third limitation might be related to the disease stage, because most of the patients included in this study were enrolled at quite advanced stages of the disease. Therefore, the value of this panel of biomarkers at early stages of neurodegeneration remains to be determined, even though FABP3 has already shown some diagnostic value in early AD [
30,
31].
Acknowledgements
We thank Cristiano Spaccatini for his excellent technical support.