Cerebrospinal fluid amyloid beta and glial fibrillary acidic protein concentrations in Huntington’s disease

Introduction Huntington’s disease (HD) is a genetic incurable lethal disease. Biomarkers are needed for objective assessment of disease progression. Evidence supports both complex protein aggregation and astrocyte activation in HD. This study assesses the 42 amino acid long amyloid beta (Aβ42) and glial fibrillary acidic protein (GFAP) as potential biomarkers in the cerebrospinal fluid (CSF) of HD mutation carriers. Methods CSF was obtained from manifest HD patients (ManHD), premanifest HD-gene-expansion carriers (PreHD) and gene-negative controls (controls). Disease Burden Score (DBS) and Total Functional Capacity (TFC) were calculated. Protein concentrations were measured by enzyme-linked immunosorbent assays (ELISA) and intergroup differences were analysed using Mann-Whitney U test. Spearman correlations were calculated to assess disease stage association. Age-adjustment was included in the statistical tests. Results The study enrolled 27 ManHD and 13 PreHD subjects. The number of controls differed in the analysis of Aβ42 and GFAP (n = 19, and 8 respectively). Aβ42 levels were higher in ManHD (mean 741 ng/l, SD 361) compared with PreHD (mean 468 ng/l, SD 184) (p = 0.025). Likewise GFAP concentration was higher in ManHD (mean 435 ng/l, SD 255) compared with both PreHD (mean 266 ng/l, SD 92.4)(p = 0.040) and controls (mean 208 ng/l, SD 83.7)(p = 0.011). GFAP correlated with DBS (r = 0.361, p = 0.028), TFC (r = − 0.463, p = 0.005), and 5-year risk of onset in PreHD (r = 0.694, p = 0.008). In contrast, there was no correlation between Aβ42 concentration and DBS, TFC or 5-year risk of onset. Conclusion CSF Aβ42 levels did not correlate with disease stage suggesting no Aβaggregation in HD. GFAP is a potential biomarker in HD with association to disease stage. Validation in larger HD cohorts and potential correlations with clinical phenotype would be of interest.

Huntington's disease (HD) is an autosomal dominant neurodegenerative disorder caused 2 by pathological expansions in the polyglutamine (CAG expansion) region in exone 1 in 3 the huntingtin gene (HTT) on chromosome 4q [1]. CAG encodes for the aminoacid 4 glutamine (Q) and the CAG expansion leads to the formation of a protein with an 5 abnormal poly-Q tail. Mutated HTT (mHTT) interacts with other proteins, 6 accumulates and causes dysfunction and eventually degeneration and death of the 7 neurons [2]. These changes have shown to be most prominent in capsula interna and 8 striatum [3]. Symptoms of HD are progressive and include movement disorders, 9 psychiatric symptoms [4] and cognitive decline [5]. There is no disease-modifying drug 10 (DMD) or curative treatment available for HD so current treatment options are purely 11 symptomatic. Several promising gene-based clinical trials are under way. HTT lowering 12 therapies hold the promise of a disease modifiying therapy for HD. The diagnosis of HD 13 is based on the presence of symptoms, clinical signs and confirmed by genetic testing. 14 In the era of emerging therapies there is a need for biofluid markers to assess the 15 progression of HD and thereby evaluate the utility of DMDs objectively. Some promising 16 candidate biomarkers have recently been studied [6], of which neurofilament light 17 protein (NfL), a broad biomarker for neurodegenerative diseases and neuroinflammation 18 such as multple sclerosis (MS), seems to be the most promising with association to both 19 disease onset and stage in HD [7][8][9], as well as mHTT [10]. Total tau (an axonal 20 microtubule-stabilizing protein), another biomarker for neurodegeneration, has also 21 been associated with the severity of symptoms, especially psychiatric symptoms but the 22 data is less convincing compared with NfL [7,9,[11][12][13][14]. Here we wanted to extend the 23 CSF analysis of neurodegenerative markers with studies on Amyloid β(Aβ) protein and 24 glial fibrillary acidic protein (GFAP) which have not been reported in HD before. 25 Dysregulation of amyloid β(Aβ) protein causes extracellular aggregation of the 42 26 amino acid form of Aβ(Aβ42) in brain tissue. Aβ42 is considered to be the most 27 pathogenic form of Aβ [15]. Aβ42 aggregation leads to neuronal dysfunction mediated 28 by excitotoxicity and neuroinflammation [16][17][18] and cereral hypoperfusion [19]. One 29 form of Aβpathology is cerebral amyloid angiopathy (CAA), where the amyloid 30 accumulates in the vascular walls [20]. Aβ42 aggregation is a hallmark of Alzheimer's 31 disease (AD) [21,22], where CSF Aβ42 levels are reduced [23][24][25][26]. This decrease is likely 32 due to increased amyloid deposition in the brain and decreased clearance into the dementia [32]. CSF Aβ42 has been examined as a biomarker in depression where higher 38 brain Aβburden in PiB-PET was found to be associated with increasing 39 anxious-depressive symptoms over time in cognitively normal older individuals [33]. 40 Studies on MS have found correlation between lower CSF Aβ42 levels and disease 41 progression [34], including cognitive impairment [35], and higher gadolinium-enhancing 42 lesion burden on magnetic resonance imaging [35]. 43 GFAP (glial fibrillary acidic protein) is an intermediary fibrillary protein in the glial 44 cells (astrocytes) and increased GFAP immunoreactivity is associated with gliosis and 45 slowly progressing neuronal damage [36,37]. It is used as a diagnostic marker in gliomas, 46 especially astrocytomas [38,39], is released in blood following ischaemic stroke [40], and 47 is a marker for inflammation and disease stage in, e.g. MS [41][42][43]. Highly increased 48 CSF GFAP concentration is seen in neuromyelitis optica spectrum disorders (NMOSD) 49 due to rapid astrocytic damage [44,45]. A post-mortem study performed on HD 50 patients revealed astrocytosis in moderate, but not the earliest pathological stages of 51 HD [46]. Similarly, trials on the HD mouse model R6/2 have shown increased GFAP 52 levels in astrocytes of moderate and late disease stages. [47]. 53 This study aims to investigate alterations of Aβ42 and GFAP levels as potential 54 biomarkers for disease onset and progression in HD. 55 Materials and methods 56 Definition of participants and clinical assessment 57 The participants were recruited from the HD clinic at Uppsala University Hospital, from 58 Karolinska University Hospital in Stockholm and Sahlgrenska University Hospital, 59 Gothenburg, and were either premanifest gene-expansion carriers or manifest HD was approved by the regional ethical review board in Uppsala, Sweden (DNR 2012/274). 70 All participants signed an informed consent before study entry.

71
CSF sample collection and handling 72 CSF was collected by lumbar puncture according to standardized protocol for procedure, 73 materials and handling, but the time of the lumbar puncture and relation to meals 74 varied. Polypropylene tubes and collecting vessels were used to avoid protein adsorption. 75 The CSF was put on ice and centrifuged at 4 degrees Celsius and 1300 G for 10 minutes. 76 The acellular proportion was pipetted off for storage at minus 70 degrees Celsius until  using an in-house ELISA, as previously described in detail [48]. Tests for normality of distribution included Shapiro-Wilk and inspection of histogram 84 and the skewness statistics were applied. Age was approximately normally distributed, 85 but the protein levels were not. Intergroup differences in protein levels were tested with 86 non-parametric tests (Mann-Whitney U test). If there was deemed to be an association 87 to age and/sex this variable was included as a covariant in a linear regression analysis 88 model. We performed Spearman rank correlation with all gene-expansion carriers 89 pooled into one group to assess the correlation between the protein levels and TFC as 90 well as DBS. All statistical analyses were performed on cross-sectionally sampled data. 91 The level of significance was defined by p-value less than 0.05. Statistical analyses were 92 performed using SPSS Statistics Subscription Build 1.0.0.1461

94
The study enrolled 59 participants (mean age; standard deviation, range see Table 1 95 and  Table 1 and Table 2.

103
The protein levels were also compared to those of the laboratorys age-stratified 104 reference rates. The group sizes differ because of insufficient amounts of CSF from 105 several individuals, precluding quantitative analysis of both Aβ42 and GFAP. concentration has been noted in healthy individuals [48].

112
Regarding Aβ42, the premanifest and the manifest HD group differed significantly in 113 age (p = 0.001), and regarding GFAP the manifest HD group was significantly older 114 than both the premanifest HD group and control group (p = 0.0003 and p = 0.043 115 respectively), due to the natural course of the disease. To exclude potential confounding 116 age was included as a covariate in the statistical analyses.
117 Figure 1 shows the concentrations of both proteins in the three groups. CSF Aβ42 118 concentration was significantly higher in the manifest HD group (mean 741 ng/l, SD 119 361) compared with the premanifest HD group (mean 468 ng/l, SD 183, (p = 0.025).

120
The difference remained significant after adjustment for age (p = 0.03). There were no 121 significant differences between the gene-negative controls (mean 535 ng/l, SD 238) and 122 the manifest HD group, but a trend towards higher levels in the manifest HD group was 123 noted (p = 0.076, age-adjusted p = 0.112). There were no significant differences  CSF GFAP concentration correlated positively with DBS and inversely to TFC with 141 a significant correlation (r = 0.361, p = 0.028 and r = −0.463, p = 0.005, respectively). 142 Correlation between protein levels and TFC remained statistically significant even after 143 adjustment for age (p = 0.001). DBS was not adjusted for age since this variable is already included in the composite score. GFAP levels correlated with 5-year risk of 145 onset among premanifest gene-expansion carriers (r = 0.694, p = 0.008). 146 We did not find any correlation between CSF GFAP and Aβ42 concentrations (r =-147 0.097, p = 0.522).

149
In this exploratory CSF study we found significantly increased CSF Aβ42 concentration 150 in the manifest HD group compared with the premanifest gene-expansion carriers.

151
There was a non-significant tendency to lower levels of Aβ42 in the premanifest HD 152 group compared with the healthy controls. However, Aβ42 concentration did not 153 correlate with Disease Burden Score, Total Functional Capacity, or 5-year risk of onset. 154 A recent study demonstrated an association between lower levels of CSF amyloid 155 precursor protein (APP) and worse clinical phenotype and lower cognitive performance 156 in HD patients [52], the strongest relationship observed with composite UHDRS score. 157 APP is a transmembrane protein with multiple physiological functions, including 158 regulating brain iron homeostasis [53], and it is cleaved by beta-and gamma-secretase 159 to form Aβ peptides [54,55]. Our previous finding of decreased CSF transthyretin in 160 HD patients [56] has also been reported in AD, where it possibly contributes to the 161 failure of cerebral amyloid clearance [57]. This might allude to similarly decreased Aβ42 162 levels in HD, as noted in several other neurodegenerative disorders, most notably in 163 AD [24][25][26][27][28][29]. On the contrary, we found the highest levels of Aβ42 in the manifest HD 164 group. The mechanism underlying this finding is unknown. Nevertheless, the data 165 corroborate that Aβ aggregation is not a feature of HD [58].

166
The present study found higher CSF GFAP concentration in the manifest HD group 167 compared with both premanifest gene-expansion carriers and healthy controls. There 168 was a non-significant trend towards higher CSF GFAP concentration in the premanifest 169 group compared with controls. Further, GFAP concentration correlated positively with 170 Disease Burden score and inversely with the Total Functional Capacity in the pooled 171 gene-expansion carrier group. The strongest correlation was found between GFAP levels 172 and 5-year risk of onset among the premanifest gene-expansion carriers. As GFAP is a 173 marker of astrogliosis and degenerative process [36,37], as well as a known marker for 174 neuroinflammation [36,37,[41][42][43], and/or astrocyte damage [44,45], the interpretation 175 of elevated GFAP levels is not so straightforward as to what kind of underlying 176 pathological processes might be involved. The levels of GFAP also tend to rise with 177 ageing in healthy individuals, most probably as a sign of astroglial filament formation in 178 the CNS [48]. Still, our findings are in line with pathology studies of HD patients that 179 have described astrocytosis in moderate, but not the earliest pathological grades of HD 180 in humans [46], as well as findings in the R6/2 mouse, where GFAP was elevated in 181 astrocytes of moderate and late disease stages as a sign of classical astrocyte 182 activation [47]. Previous CSF studies in HD patients have shown increased CSF levels 183 of YKL-40, another astrocytic activation marker, as a late feature of HD [9,59], and a 184 correlation to several markers of neurodegeneration [59]. YKL-40 is secreted by 185 astrocytes and is increased in many inflammatory CNS disorders [60]. Evidence from 186 CSF studies in HD mainly supports activation of the innate immune system of the CNS. 187 This may reflect inflammation caused by neurodegeneration in the later stages of the 188 disease [9]. However, an early increase of IL-6 and IL-8 levels in both CSF and plasma 189 suggest an innate immune response both centrally and peripherally in HD [61]. There is 190 also evidence of T-cell mediated inflammation ahead of disease onset [59]. The tendency 191 to elevation of GFAP in premanifest gene-expansion carriers together with a strong and this could be a reason that GFAP was not clearly elevated in this group.

195
To the best of our knowledge, this is the first CSF study performed on humans to 196 assess the role of Aβ and GFAP as potential biomarkers in HD. However astrocyte 197 activation in the form of astrocytosis and neuroinflammation has previously been linked 198 to the pathophysiology of HD, especially in the late disease stages so this is in line with 199 the finding of elevated GFAP levels amongst the manifest HD patients. Part of the include the exploratory nature and a small sample size. Different ages between groups 207 was of concern and the fact that the groups differed in size. The medications used by 208 the patients and their potential effect on the protein levels was not assessed. 209 Nonetheless, we believe that these findings may be of relevance regarding the 210 involvement of astrocyte activation, neuroinflammation and gliosis in HD. GFAP may 211 have a role in assessing the severity of HD, and could potentially also serve as a 212 surrogate end-point in clinical trials. Before taking these findings into clinical practice 213 there is a need for validation in a larger HD cohort and assessment of correlations with 214 clinical phenotype would be of interest.