The present study showed that CSF GAP-43 was elevated in the clinical AD dementia group and in tau positive individuals. CSF GAP-43 was correlated with CSF p-tau, CSF t-tau and plasma p-tau. CSF GAP-43 levels were much higher in the tau positive group compared to the tau negative group, which distinguished tau positive status from tau negative status. The 2018 AD research framework proposed a biomarker classification system called the Aβ/tau/neurodegeneration (AT(N)) system, which defined AD as an AD continuum based on core neuropathological changes in vivo.21 If only an Aβ abnormality was present (A + T-N-) this would put the individual at a place on the AD continuum called AD pathological change. If both Aβ and tau abnormalities were present, but neurodegeneration was absent (A + T + N-), the individuals could be considered as having AD even without exhibiting signs of neurodegeneration. If all three domains of Aβ, tau and neurodegeneration were present (A + T + N+), this would reflect a more advanced stage on the AD continuum compared to an A + T + N- profile. In our study, we demonstrated that CSF GAP-43 levels increased along the AD continuum. Besides, CSF GAP-43 levels were also associated with cognitive deficits and neuroimaging finds both at baseline and during longitudinal follow up.
There is much evidence that synaptic loss is correlated with cognitive decline in AD, and synaptic dysfunction is one of the earliest detectable changes in many neurodegenerative diseases, which may appear even before neuronal loss.28 The significant role of synaptic dysfunction in the pathology of AD promotes the analysis and quantification of synaptic proteins. GAP-43 is a synaptic membrane protein which plays an important role in the regulation of synaptic plasticity, learning and memory functionality.11 Previous studies reported that the concentration of CSF GAP-43 was increased in AD,13–16 and our results are in line with these reports. Our study demonstrated that CSF GAP-43 was correlated with CSF p-tau, CSF t-tau and plasma p-tau, which might reflect a common pathogenic process between GAP-43 and tau pathology. Our study also demonstrated that CSF GAP-43 could efficiently discriminate tau pathology status. A + T- represented a stage of amyloid pathological change and A + T + indicated an advanced stage with both amyloid and tau pathologies present. CSF GAP-43 was particularly pronounced in A + T + compared to A + T- individuals, thus CSF GAP-43 could efficiently discriminate between A + T- and A + T + stages. It has been suggested that high concentrations of CSF t-tau represent axonal degeneration and high concentrations CSF p-tau represent the increased formation of neurofibrillary tangles, and that these two events are associated.29,30 As CSF GAP-43 was highly correlated with CSF p-tau and CSF t-tau, this may indicate that increased CSF GAP-43 concentration is associated with the degeneration of axons or presynaptic terminals, or the regeneration of axons and/or synapses.31
According to the amyloid cascade hypothesis, the soluble oligomer Aβ initiates tau pathology, and tau pathology leads to neuronal dysfunction and cell loss.32 A previous study utilized synaptosomes from the cortex of postmortem human subjects and transgenic rat models of AD to elucidate the time sequence of Aβ and tau pathology in synaptic terminals. This study demonstrated that Aβ accumulated in synaptic terminals in the early stages of AD, and these changes appeared before the accumulation of synaptic p-tau, the accumulation of p-tau in synaptic terminals occurred in the late stages of AD.33 Aβ initiated synaptic dysfunction via tau pathology, and without the presence of tau pathology there would not be resulting synaptic dysfunction and memory impairment.34 In our study we showed that CSF GAP-43 were correlated with CSF p-tau and plasma p-tau, and there was no correlation between CSF GAP-43 and CSF Aβ42. This could be partially explained by CSF GAP-43 being a biomarker of synaptic dysfunction, and the accumulation of tau pathology being correlated with synaptic dysfunction and memory impairment. The accumulation of Aβ42 initiated all of these events.
Aβ pathology and tau pathology are two major pathological hallmarks of AD, and tau aggregation is the primary pathological feature of a category of clinically heterogeneous neurodegenerative disorders termed tauopathies.35 Tauopathies include but are not limited to AD, progressive supranuclear palsy (PSP), corticobasal syndrome (CBS) and some types of frontotemporal lobar degeneration (FTD), and pathogenic tau aggregation has different characteristics among these disorders.36 A study of 662 individuals found that CSF GAP-43 concentration was increased significantly in clinical AD and CBS patients, while there were no concentration differences between both the control and clinical PSP groups, and behavioral variant FTD (bvFTD), and amyotrophic lateral sclerosis (ALS) with FTD groups.15 The differences in the structural characteristics of tau aggregation and cellular localization among those disordered might lead to heterogeneous concentrations of CSF GAP-43 in comparison to the control group.
In this study we also found that CSF GAP-43 was associated with several cognitive and neuroimaging hallmarks of AD at baseline and over time. Specifically, high CSF GAP-43 levels were associated with MMSE, ADAS-COG 11, and CDR-SB scores at baseline as well as with accelerated deterioration of those cognitive measures longitudinally. Synapses are important for cognitive function, and synaptic loss is one of the pathologic features of AD.2 Synaptic dysfunction has been associated with cognitive impairment in AD.4 Our study showed that on neuroimaging, high CSF GAP-43 levels were associated with smaller hippocampus and medial temporal lobe volume, lower FDG-PET values at baseline, and accelerated reduction in hippocampus and medial temporal volume, greater decline of FDG-PET values over time. A neuroimaging study reported that synaptic density in the hippocampus area was decreased in AD patients.37 Synaptic activity could be measured by cerebral glucose metabolism, and FDG-PET measures low cerebral glucose metabolism in regions of interest in AD patients, which associated with concurrent cognitive decline.27
The present study is limited by lacking data from other neurodegenerative disorders, which prevents our ability to test the disease specificity of CSF GAP-43 for AD, and whether CSF GAP-43 is associated with other neuropathological biomarkers such as α-synuclein, TDP-43. Another limitation of this study is its lack of evaluation of other tauopathies such as FTD, PSP, and CBS, which limits analysis of association between CSF GAP-43 and tau pathology in other tauopathies. For tau pathology assessment, in this study due to lack of tau PET finings, we used p-tau in CSF to quantify tau pathology, which was associated with CSF GAP-43 levels. However, previous studies have demonstrated that findings of tau PET neuroimaging were more closely associated with neurodegeneration than CSF tau.38 Our study also lacks analysis of the longitudinal changes of CSF GAP-43 levels in AD, and longitudinal changes of CSF GAP-43 levels would better depict the features of CSF GAP-43 in AD.
In conclusion, the CSF concentration of synaptic membrane protein GAP-43 was specifically elevated in tau positive individuals and could be used a biomarker for synaptic dysfunction. In addition to Aβ/tau/neurodegeneration (AT(N)) neuropathological biomarkers, CSF GAP-43 could be included as another neuropathological measurement of synaptic dysfunction in AD. In clinical study scenarios of AD, CSF GAP-43 could be used as another outcome measurement to predict longitudinal disease progression.