CSF FAM3C- a possible biomarker of Alzheimer’s disease

Backgrounnd: The impact of Alzheimer's disease (AD) on the living quality and life expectancy has becoming increasingly severe, but the pathogenesis of AD is still being studied .This study aimed to investigate whether Family with sequence similarity 3 member C (FAM3C), which has been reported to be possibly associated with cognitive function, was associated with the incidence of AD. Methods: A total of 282 participants from Alzheimer’s Disease Neuroimaging Initiative (ADNI) were included. The demographic data, cerebrospinal uid (CSF) FAM3C content, neuropsychological scores, CSF amyloid protein level, CSF tau protein level, and imaged data were collected. One-way analysis of variance and chi-square test were used to compare demographic data, neuropsychological test scores, CSF protein levels, and imaging data of three groups. Partial correlation analysis and multiple linear regression analysis were used to investigate the relationship between FAM3C and AD diagnostic indicators in different dimensions. An ordered multi-classication logistic regression model was established to determine whether FAM3C is related to the onset of AD. Results: Lower levels of FAM3C were observed in AD and mild cognitive impairment (MCI) groups. CSF FAM3C level was related to whole-brain atrophy and temporal lobe atrophy, worse cognitive performance, decreased amyloid β 1-42 (Aβ 1-42 ), and lower regional cerebral glucose metabolism. Conclusion: CSF FAM3C might be a potential diagnostic biomarker of AD. Further study on concrete mechanisms may contribute to early diagnosis and treatment of AD. be that the ratio of APOE4 carriers to non-carriers increased successively in NC group, MCI group and AD group. The MMSE score, MEM score and EF score while the CDR score and ADAS score increased. These neuropsychological results displayed the declined AD and MCI, compared with the normal group. The increase of CSF T-tau, P-tau181 CSF 1−42 MCI groups One-way ANOVA that FAM3C content among the three groups statistically signicant. CSF FAM3C was lower in AD and MCI groups compared to that of NC group. The pairwise comparison results of FAM3C among the three groups were also statistically signicant (NC group versus AD group:P = 0.019, MCI group versus AD group:P = 0.042). Data are presented as mean ± standard deviation or participants. P-values are for one-way analysis of variance or chi-square test (categorical variables).


Introduction
Alzheimer's disease (AD), the most common cause of dementia among the elderly, is an irreversible, progressive neurological disease [1]. At present, there is no cure for AD, only ve prescription drugs were approved to intend for symptom management, with varying degrees of e cacy [2]. With the deepening of studies, some theories, such as amyloid β (Aβ) deposition, tau protein abnormality, axonal transport defect, in ammatory response, and cholinergic damage, etc, have been gradually recognized [3][4][5]. The clinical and genetic trials of more-effective Alzheimer's drugs targeting anti-amyloid, anti-neuroin ammation, anti-tau, or neuroprotection are ongoing. Between 2000 and 2012, more than 400 AD clinical trials were undertaken. Unfortunately, it was suggested that 99.6% of them failed [2]. One potential reason for this result may be that studies were initiated too late in the disease as the pathological changes of AD may appear 10 to 20 years before the appearence of symptoms. Herein lies the promise for more effective AD biomarkers in addition to the well-known Aβ and the tau protein.
Family with sequence similarity 3 member C (FAM3C), also called interleukin-like epithelial-mesenchymal transition inducer (ILEI), is a member of the FAM3 cytokine family which consists of four sequentially similar members (FAM3A, FAM3B, FAM3C and FAM3D) [6]. FAM3C plays an important role in epithelial-mesenchymal transformation (EMT) and tumor metastasis [7,8]. It has been found overexpressed in colorectal cancer, esophageal cancer, breast cancer, hepatocellular cancer, etc, predicting a poor prognosis [9][10][11][12]. Recently, it was indicated that FAM3C may be a risk factor for the development of AD in an animal experiment [13]. However, whether FAM3C is related to AD patients remains to be con rmed.
In this study, we investigated the relationship between AD and cerebrospinal uid (CSF) FAM3C using data obtained from the Alzheimer's Disease Neuroimaging Initiative database (ADNI).

ADNI
The ADNI database was launched in 2003 as a public-private partnership, led by Principal Investigator Michael W.
Weiner, MD. The whole project includes ADNI1, ADNIGO, ADNI2 and ADNI3 research stages. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early Alzheimer's disease. For up-to-date information, see www.adni-info.org.

Participants
In this study, we included participants who underwent lumbar puncture for quantitative detection of CSF proteins from baseline ADNI1 cohort. A total of 282 participants were included and the general inclusion criteria were as follows: (1

APOE genotypes
There are three common APOE alleles: ε2, ε3 and ε4. APOE4 gene has been accepted as a risk factor for late-onset and sporadic AD. At baseline, venous blood of all participants was collected and the APOE genotype was determined. In this study, subject possessing one or two APOE4 alleles was considered as an APOE4 carrier.

Neuropsychological assessment
According to AD diagnostic criteria, neuropsychological assessment of AD consists of cognitive impairment, social and daily functioning decline, and psychosocial behavioral symptoms. The decline of cognitive function was mainly determined by various psychological scales in clinical practice. Quali ed ADNI staff conducted more than a dozen scale tests on each participant at the screening stage and then derived composite scores for executive functioning (EF) and memory (MEM) using data from the ADNI neuropsychological battery according to item response theory (IRT). This study included the Mini-Mental State Examination (MMSE), Clinical Dementia Rating (CDR) and Alzheimer's Disease Assessment Scale-Cognitive section (ADAS-Cog) scores, as well as composite scores EF and MEM. Higher MMSE, MEM and EF scores re ect better cognitive function while higher CDR and ADAS-cog scores re ect worse cognitive function.

CSF FAM3C, tau and Aβ
All of the 282 participants included in our study underwent lumbar puncture to obtain CSF baseline data. CSF samples were analyzed by the electrochemiluminescence immunoassays (ECLIA) for CSF total tau (T-tau), phosphorylated tau 181 (P-tau181), and amyloid β 1-42 (Aβ 1−42 ). It should be noted that the elecsys Aβ 1−42 immunoassay in use is an assay that is currently under development and for investigational use only, therefore it is excluded for clinical decision making or derivation of medical decision points. The quanti cation of CSF FAM3C was conducted by the use of liquid chromatography-tandem mass spectrometry with multiple reaction monitoring (LC/MS-MRM) approach. The whole MRM panel consisted of 567 peptides representing 221 proteins, and peptide sequence used for FAM3C was GINVALANGK in our research. The output results are reported in arbitrary signal intensity units on a natural log scale in a form of "log peptide intensity". Detailed data was shown in the Biomarkers Consortium CSF Proteomics MRM data set in the o cial website.

Neuroimaging data
All of the ADNI participants received structural MRI examinations at baseline and quite a few of them had follow-up visits. Tensor-based morphometry (TBM) was used to analyze cross-sectional and longitudinal MRI data of bilateral temporal lobes. Iterative principal component analysis (IPCA) was practiced to show the numerical value of the wholebrain atrophy rate from baseline to 12 months. This study included typical MRI results such as total brain volume, lobe volume, ventricular volume, cortical thickness at baseline as well as follow-up visits. Baseline spatial pattern of abnormalities of recognition of early AD (SPARE-AD) scores which provided a predictive value of cognitive decline and conversion of NC to AD were also obtained. Approximately 50% of participants had a uorodeoxyglucose (FDG) PET scan. Based on coordinates frequently cited in other FDG studies comparing AD, MCI, and NC, FDG PET results of angular gyrus, posterior cingular and inferior temporal gyrus were obtained in this study.

Statistical methods
One-way analysis of variance and chi-square test were used to compare demographic data, neuropsychological test scores, CSF protein levels and imaging data of three groups. Partial correlation analysis and multiple linear regression analysis were used to investigate the relationship between FAM3C and AD diagnostic indicators in different dimensions. An ordered multi-classi cation logistic regression model was established to determine whether FAM3C is related to the occurrence and progression of AD. P-value < 0.05 was statistically signi cant. All statistical analyses were conducted by SPSS 22.0.

Demographic and clinical data
The information of age, gender, education, AOPE4, neuropsychological test scores, CSF T-tau, P-tau181, Aβ 1−42 and FAM3C content are shown in Table 1. There was no statistical difference in age, sex and education years among three groups. In Table 1, "yes" means the subject was an APOE4 carrier. It can be seen that the ratio of APOE4 carriers to noncarriers increased successively in NC group, MCI group and AD group. The MMSE score, MEM score and EF score decreased while the CDR score and ADAS score increased. These neuropsychological results displayed the declined cognitive function in patients with AD and MCI, compared with the normal group. The increase of CSF T-tau, P-tau181 and decrease of CSF Aβ 1−42 in AD and MCI groups was consistent with numerous previous studies. One-way ANOVA results demonstrated that FAM3C content among the three groups was statistically signi cant. CSF FAM3C concentration was lower in AD and MCI groups compared to that of NC group. The pairwise comparison results of FAM3C among the three groups were also statistically signi cant (NC group versus AD group:P = 0.019, MCI group versus AD group:P = 0.042).

FAM3C and MRI results
Craniocerebral MRI examination indexes including volume of total brain parenchyma, ventricles and hippocampus, the thickness of temporal lobe, fusiform gyrus and entorhinal cortex were collected. Longitudinal data showed the rates of total brain atrophy, temporal lobe atrophy and composite prediction index SPARE-AD were also obtained. The above indicators were signi cantly different in NC, MCI and AD groups, revealing that the brain structure atrophy was worse in AD and MCI patients than in normal aging. Partial correlation analysis and multiple linear regression analysis results between FAM3C and structure MRI data are shown in Table 2. Right hippocampus volume, bilateral temporal lobe thickness except for left middle temporal, right fusiform gyrus thickness, bilateral entorhinal cortex thickness were found to have a positive linear correlation with CSF FAM3C content at baseline. Total ventricle volume and lateral ventricle volume had a negative linear correlation with CSF FAM3C content at baseline. In this study, no signi cant correlation was found between FAM3C and total brain parenchyma volume as well as left hippocampus volume. In the longitudinal view, FAM3C was negatively correlated with the 12-month whole brain atrophy rate represented by IPCA in Table 2. Moreover, the follow-up temporal lobe volume ratios to baseline were calculated after one year (adjusted R 2 = 0.330) and two years (adjusted R 2 = 0.411), and it was found that both of them had a positive linear relationship with FAM3C, that is, the reduction of FAM3C was moderately correlated with temporal lobe atrophy. In predicting the likelihood of AD, the SPARE-AD score was negatively related to baseline FAM3C content. In general, the content of baseline FAM3C in this study was negatively correlated with brain atrophy, especially bilateral temporal lobe atrophy. Abbreviations: N, number of participants; R, partial correlation coe cient; P*, P value of partial correlation analysis; Adjusted R2, value of adjusted R2 in multiple linear regression analysis; F, F value of multiple linear regression analysis; B, coe cient B value of multiple linear regression analysis; P**, P value of B in multiple linear regression analysis; 95% CI, 95% con dence interval of B; IPCA, percent annualized whole brain atrophy detected by Iterative Principal component Analysis; TEMPVOLRATIO1, bilateral temporal lobe volume to baseline volume ratio one year later; TEMPVOLRATIO2, bilateral temporal lobe volume to baseline volume ratio two years later; SPARE-AD, spatial pattern of abnormalities of recognition of early AD. Abbreviations: N, number of participants; R, partial correlation coe cient; P*, P value of partial correlation analysis; Adjusted R2, value of adjusted R2 in multiple linear regression analysis; F, F value of multiple linear regression analysis; B, coe cient B value of multiple linear regression analysis; P**, P value of B in multiple linear regression analysis; 95% CI, 95% con dence interval of B; IPCA, percent annualized whole brain atrophy detected by Iterative Principal component Analysis; TEMPVOLRATIO1, bilateral temporal lobe volume to baseline volume ratio one year later; TEMPVOLRATIO2, bilateral temporal lobe volume to baseline volume ratio two years later; SPARE-AD, spatial pattern of abnormalities of recognition of early AD.

FAM3C and cerebral glucose metabolism
In this study, 37 normal controls, 72 MCI patients, and 36 AD patients accepted FDG PET scan at baseline. The results showed that mean value of glucose metabolism decreased in MCI group and further decreased in AD group (Fig. 4). In

Discussion
In the present study, we analyzed the data from ADNI database and determined the relationship between cerebrospinal uid FAM3C level and Alzheimer's Disease. The current study provides novel evidence that decreased CSF FAM3C level in AD and MCI patients, in relation to the extent of whole-brain atrophy and temporal lobe atrophy, and positively relating to CSF Aβ 1−42 and regional brain glucose metabolism, may contribute to worsening cognitive function of the AD patients.
As a novel marker, the role of FAM3C has not been fully elucidated. It was demonstrated that FAM3C played an important role in bone metabolism, via modulating osteogenic differentiation by down-regulating Runx2 [10]. Moreover, the single-nucleotide polymorphism in gene Fam3c was found to be associated with bone mineral density at the wrist and other sites in the skeleton [14]. Recently, there has been increasing evidence that FAM3C is an important regulator of glucose and lipid metabolism. It was shown that FAM3C suppressed hepatic gluconeogenesis and lipogenesis through activating the HSF1-CaM-Akt pathway and repressing the mTOR-SREBP1-FAS pathway [15]. In obese diabetic mice, FAM3C expression was reduced in the liver [16]. Furthermore, FAM3C genetic variants were found to be associated with dyslipidemia and lipid traits among Chinese children [17]. What's more, FAM3C has been shown to be strongly upregulated in several cancers including gastric cancer, melanoma, colon and breast cancer, oral and esophageal squamous cell carcinoma, essential for EMT and metastasis formation, and alter subcellular localization strongly correlating with poor survival [8,9,12,18,19]. It has been indicated that the FAM3C-YY1-HSF1 pathway exerted an important role in TGF β-triggered proliferation and migration of human breast cancer cells [11,20]. In addition, FAM3C is involved in retinal laminar formation in vertebrates and may relate to cell differentiation and proliferation during inner ear embryogenesis [21,22].
Some studies of FAM3C concentrating on brain were performed and FAM3C was found to be widely expressed in mice brain including cortex, hippocampus, amygdala, striatum, thalamus, hypothalamus, cerebellum and brainstem [13]. Autopsy and immunohistochemical results of the human brain and monkey brain showed that FAM3C was widely distributed in neurons and ependymal cells, but not in glial cells and vascular endothelial cells. FAM3C levels in CSF peaked after birth and then decreased with age, but signi cantly decreased under the pathological condition of AD. It was identi ed that FAM3C could inhibit the deposition of Aβ. And transgenic overexpression of FAM3C signi cantly reduced the brain Aβ burden and ameliorated the memory de cit in AD model mice [13]. The study is consistent with ours, which indicates that FAM3C may be a probable target for AD therapy.
As for the diagnostic criteria for AD, the latest ones are the 2011 NIA-AA criteria, published by the National Institute of Aging (NIA) and the Alzheimer's Association (AA). The new standard includes biomarkers in the diagnostic criteria for AD, such as changes in CSF Aβ, T-tau, P-tau and the amyloid changes shown in PET, decreased uptake of 18FDG in the temporal-parietal cortex, and atrophy in the medial temporal lobe shown by structural MRI [23]. These markers are mainly related to two major pathological features of AD -neuro brillary tangles (NFT) and senile plaques (SP). The NFT is accumulated by the intraneuronal accumulation of tau proteins in phosphorylated form and the accumulation and deposition of Aβ are involved in the formation of senile plaque [24]. Although still controversial, these markers provide some supporting roles in the diagnosis of AD. But the diagnosis of MCI and AD is still based more on neuropsychological tests and neurological examination. Memory impairment is often the rst chief complaint of AD.
After the onset of AD, the hippocampus and entorhinal cortex were initially involved, followed by the medial parietal cortex, the lateral temporal cortex, and the frontal lobe. Finally, the lesion developed to the entire cerebral cortex.
Cortical atrophy was mainly manifested in narrowed gyrus, widened sulcus, thinning cortex, varying degrees of atrophy of white matter and enlargement of lateral ventricle [25]. In this study, FAM3C showed a certain degree of association with the above characteristics of AD, which may be involved in the pathogenesis of AD.
Due to the invasive operation of lumbar puncture and the time cost of imaging examination, the sample size of this study was relatively limited. Moreover, it was only an observational study and long-term clinical studies and experimental studies are needed in the future.

Conclusion
In summary, this study has provided the new evidence that FAM3C might be a potential point of penetration of AD