Advances and applications of fluids biomarkers in diagnosis and therapeutic targets of Alzheimer's disease

Abstract Aims Alzheimer's disease (AD) is a neurodegenerative disease with challenging early diagnosis and effective treatments due to its complex pathogenesis. AD patients are often diagnosed after the appearance of the typical symptoms, thereby delaying the best opportunity for effective measures. Biomarkers could be the key to resolving the challenge. This review aims to provide an overview of application and potential value of AD biomarkers in fluids, including cerebrospinal fluid, blood, and saliva, in diagnosis and treatment. Methods A comprehensive search of the relevant literature was conducted to summarize potential biomarkers for AD in fluids. The paper further explored the biomarkers' utility in disease diagnosis and drug target development. Results Research on biomarkers mainly focused on amyloid‐β (Aβ) plaques, Tau protein abnormal phosphorylation, axon damage, synaptic dysfunction, inflammation, and related hypotheses associated with AD mechanisms. Aβ42, total Tau (t‐Tau), and phosphorylated Tau (p‐Tau), have been endorsed for their diagnostic and predictive capability. However, other biomarkers remain controversial. Drugs targeting Aβ have shown some efficacy and those that target BACE1 and Tau are still undergoing development. Conclusion Fluid biomarkers hold considerable potential in the diagnosis and drug development of AD. However, improvements in sensitivity and specificity, and approaches for managing sample impurities, need to be addressed for better diagnosis.

gamma 1 monoclonal antibodies selectively binding to Aβ fibrils and soluble oligomers. 6,7 These drugs provide a new treatment strategy, representing a breakthrough in AD treatment.
The above six drugs are the only approved therapeutic drugs against AD. Other than the controversial aducanumab and justapproved lecanemab, the therapeutic results of the other drugs also remained unsatisfactory. Since these drugs could not prevent neuronal degeneration, although they improved the clinical symptoms and quality of life of patients during the initial treatment stages. The biggest challenge is the short-term efficacy of these drugs, which is approximately 1 ~ 3 years, and cannot prevent the disease from progressing. 1 The dismal condition of AD therapy requires attention. Early detection and intervention could be possible solutions. Early diagnosis facilitates the identification of reversible causes, improves symptom management, and provides a future plan. In addition, based on the cost-effectiveness analysis, the cost of early interventions is less than subsequent dementia treatment. 8 Biomarkers have excellent prospects for early diagnosis. They are a class of molecules occurring in biological media, such as fluids or cells, and depict physiological states, pathological processes, or therapeutic agent responses. The significance of the biomarkers lies in: (1) Finding the latent disease to provide an opportunity for early intervention, particularly in patients having neurological disorders. (2) Providing some clues to molecular pathological mechanisms of AD, depending on the correlation between the appearance timing and changes in molecular biomarker levels and disease process. (3) Assisting in the differential diagnosis and determining the causes of the disease to select more targeted treatments, facilitating individualized treatment, and managing AD patients. 9 (4) Developing new drugs based on the mechanisms and targets. AD drugs in the market and under investigation are presented in Figure 1 (developed using BioRe nder.com). Some biomarkers have already been identified and utilized in AD diagnosis, and more candidate biomarkers being investigated.

| P OTENTIAL B I OMARK ER S OF AD
In 2007, cerebrospinal fluid (CSF) biomarkers were first included in the AD diagnostic criteria by the International Working Group, indicating their increasingly important role in diagnosing AD. The diagnosis of AD depends on the images, biomarkers, and scales. 10 Positron emission tomography (PET) is the gold standard for AD diagnostics, and the high cost impedes its popularization. Scale tests are unable to distinguish and predict possible risk promptly. CSF biomarkers involve invasive extraction. In this case, other easily collectible fluids such as blood, urine, and saliva are the first choice to solve the difficulty. Several studies focus on diagnosing and predicting AD through fluid biomarkers, and many relevant test kits have been awaiting approval and marketing. The current study summarizes the most promising AD biomarkers for differentiating AD patients from healthy controls (Figure 2, developed using BioRe nder.com).

| Biomarkers related to Aβ aggregation
Aβ is a polypeptide with 38-43 amino acids mainly produced by neurons. The most common human subtypes are Aβ 42 with 42 amino acids and Aβ 40 with 40 amino acids. 11 Aβ is considered the initial trigger of AD, with research suggesting that Aβ is upstream of Tau for accelerating age-related tauopathy and neurodegeneration. [11][12][13] However, in 2020, a new study observed that subtle cognitive difficulties could be identified before or during the amyloid plaque deposition in the brain. 14 This indicates that damage to neurons could happen before Aβ. Another likely reason is that brain scans could not detect the initial deposition of Aβ, demanding more evidence for validating the finding.

| Aβ peptide
The amount of Aβ or its fragments can facilitate an early AD diagnosis. 10 Levels of CSF Aβ 42 and Aβ 42 /Aβ 40 both decrease in AD patients. 15 A meta-analysis revealed that the CSF concentration of Aβ 42 was 0.55 times lower in AD cohorts than in healthy controls. 16 Mild cognitive impairment (MCI) and dementia due to AD are associated with lower CSF levels of Aβ 42 . Compared with stable MCI, a 0.67 times reduction (95% CI 0.63-0.73, p < 0.001) was observed. 16,17 The CSF Aβ 42 /Aβ 40 ratio decreased by 40%-60% in patients with cerebral Aβ pathology. 18 Thus, the ratio of Aβ 42 /Aβ 40 could predict the abnormal Aβ inside the brain and differentiate AD patients from normal individuals. Aβ 42 /Aβ 40 holds promise for future applications but requires significant evidence to establish its sensitivity and specificity.
According to several studies, Aβ levels in blood and saliva showed significant changes with inconsistent results. 16,19 This could be due to the analytical sensitivity of the assay, as some early measurements were performed with enzyme-linked immunosorbent assay (ELISA) methods. In contrast, recent studies utilized ultra-sensitive immunoassays or targeted mass spectrometry (MS). Depending on the improvements in diagnostic techniques, more decreases of Aβ 42 and Aβ 40 levels in blood and saliva were observed in AD patients. [20][21][22][23] Aβ discovery offers new insights for developing drugs to combat AD. Anti-Aβ antibody levels are lower than in the healthy population, indicating that the immune system could prevent the occurrence and development of AD. 24,25 Therefore, it becomes the theoretical basis of active immunotherapy. Phase 1 or 2 vaccine clinical trials are being conducted or completed. Unfortunately, no benefits were revealed, and many vaccines have significant adverse reactions.
Passive immunotherapy is promising, depending on current evidence. Aducanumab and lecanemab are humanized antibodies for reducing Aβ plaques and decreasing the risks of developing AD. 26 They are already available in the U.S. amidst controversies.
Dr. Knopman et al. described that no evidence was presented to correlate biomarker changes to cognitive benefits even though Aβ was decreased in aducanumab trials. Post hoc analysis indicates that the efficacy signal of aducanumab was unreliable. 27 Definitive, optimally designed, and adequately powered trials should be undertaken to determine the efficacy of the drugs. On the contrary, the phase 3 double-blinded trial on lecanemab satisfied all the primary and secondary endpoints, demonstrating significant improvements in brain amyloid and disease-related scores. Further post-marketing F I G U R E 1 AD therapeutic agents and mechanisms. Three classes of drugs have been used to treat AD: cholinesterase inhibitors, NMDA receptor antagonist, and anti-Aβ therapy. BACE1 inhibitors, γ-secretase modulators, and anti-Tau therapies are under investigation. APP, amyloid-beta precursor protein; BACE1, beta-secretase 1; Aβ, amyloidβ; NMDA, N-methyl-D-aspartic acid. evaluation of its effectiveness is required. 28 Both active and passive immunotherapeutic drugs targeting Aβ epitopes are represented in Table 1.
Beta-secretase 1 (BACE1) BACE1 is another candidate biomarker for AD. It cleaves amyloidbeta precursor protein in the first step of producing Aβ. CSF BACE1 activity and/or concentrations were significantly elevated in MCI.
Thus, it could indicate an early AD stage, compared with healthy controls and AD. [44][45][46] Cervellati et al. identified that BACE1 in serum changed across groups (p < 0.001) with a 25% increase in late-onset AD against the controls. High BACE1 levels were independently associated with late-onset AD diagnosis (odds ratios 2.79; 95% CI 1. 37-5.67). This characteristic could identify AD during the pre-dementia stage and predict its onset. The evidence supports the hypothesis that BACE1 is critical in AD pathophysiology, emerging as a new drug target in treating early AD to inhibit Aβ production. 47 BACE1 inhibitors can slow down the progress of AD. Besides, some studies have also identified that inhibiting BACE1 stabilizes Tau levels in CSF, 48,49 although this is not its most significant role. Additionally, reduced brain volume and hepatic adverse reactions are common among patients using BACE1 inhibitors (Table 1).

| Biomarkers related to Tau pathology
Tau, or microtubule-associated protein tau (MAPT), is the primary component of the neuroaxonal protein bindings microtubules and stabilizing their assembly. Its function is dependent on the balance between phosphorylation and dephosphorylation. When Tau is hyperphosphorylated, the synthesis and assembly of microtubules are restricted. The aggregation of hyperphosphorylated Tau led to NFTs, inducing neuronal apoptosis. 50,51 Being the most significant characteristic of AD, Tau detection is essential for the early diagnosis of AD. 11 Presently, total Tau (t-Tau) and phosphorylation sites of Tau include threonine 181 (p-Tau 181), serine 199 (p-Tau 199), threonine 217 (p-Tau 217), and threonine 231 (p-Tau 231). 52 The concentration of CSF t-Tau is approximately three times higher in AD patients than in normal aging subjects. 53 In blood, t-Tau and p-Tau also indicated a good association with AD.
The study also observed that p-Tau and t-Tau are correlated with Tau-PET signals within the temporal cortex (r s > 0.56, p < 0.001) and predicted abnormal Tau-PET status and Aβ. 54 Moreover, AD diagnosis accuracy can be improved in combination with other biomarkers.
The sensitivity and specificity were 85% and 95%, 55 respectively, along with Aβ 42 . In saliva, t-Tau and p-Tau were detectable in participants having MCI, AD, and age-matched controls. The p-Tau levels were higher in AD without any statistical significance. 56 The interest in therapeutic strategies targeting Tau is rising. Early anti-Tau therapies were based on inhibiting kinases, Tau aggregation, or stabilizing microtubules. However, these approaches were terminated due to toxicity and/or lack of effectiveness. 59 Besides, Tau immunotherapies are the topical strategy now. Currently, there F I G U R E 2 The biomarkers having a potential diagnostic and therapeutic value in body fluids identified so far have been primarily focused on the Aβ pathway, Tau pathology, inflammation, neuron axon-associated molecules, synaptic dysfunction-related elements, microvascular-associated markers, lipids, and miRNAs. Aβ, amyloidβ; BACE1, beta-secretase 1; GFAP, glial fibrillary acidic protein; hFABP, heart-type fatty acid-binding protein; NFL, neurofilament light; sTREM2, soluble triggering receptor expressed on myeloid cells 2; VILIP-1, visinin-like protein-1; YKL-40/CHI3L1, chitinase 3-like 1.

Crenezumab
Hoffmann-La Roche • Two Phase 3 trials of crenezumab versus placebo in early AD did not show efficacy. 34 • The study about efficacy in AD patients with PSEN1 mutation is underway.

Donanemab
Eli Lilly and Company • Donanemab resulted in a better composite score for cognition and for the ability to perform activities of daily living than placebo in patients with early AD. • A Phase 3 study of donanemab compared with aducanumab is ongoing.

Gantenerumab
Hoffmann-La Roche • Clinical trials stopped for futility.

GSK933776
GlaxoSmithKline • Two Phase 1 trials were conducted but no further studies.
Immune globulin • Participants with mild to moderate AD had no clinical benefit on cognition or function. 35 Only a significant decrease in plasma Aβ 42 was observed.

Lecanemab
Eisai, Biogen • Lecanemab was approved for marketing in the United States in January 2023.
Ponezumab Pfizer • Clinical effect did not meet the expectation 36 and the research on ponezumab was terminated.

Solanezumab
Eli Lilly and Company • Solanezumab did not improve cognition of AD patients in a multicenter Phase 3 trial. 37 UB-311 United Neuroscience • Patients with mild AD had a high responder rate. 38 • The primary outcome of efficacy is not available.

BACE1
Atabecestat Janssen • Liver enzymes elevated and cognition declined in early AD patients treated with atabecestat. 39 Elenbecestat Eisai, Biogen • People with elenbecestat had more loss of brain volume, loss of weight, liver enzymes elevation, neuropsychiatric and skin rashes adverse events than those on placebo. 40 • Elenbecestat resulted in a statistically significant reduction of brain amyloid. 40 Lanabecestat AstraZeneca, Eli Lilly and Company • Clinical trials stopped for futility.

LY2811376
Eli Lilly and Company • A Phase 1 trial was conducted but no further studies.

LY2886721
Eli Lilly and Company • The study was terminated due to abnormal liver biochemical tests in some participants. are two vaccine regimens, AADvac1 and ACI-35, under clinical active immunization trials (Table 1). AADvac1 binds a six-amino-acid sequence found in each of the microtubule-binding repeats of Tau.
Rats with tauopathy produced Tau antibodies after AADvac1 administration which reduced the level of pathological Tau, thereby improving the sensorimotor function scores in rats. 60 In phase I trials, AADvac1 revealed good tolerance and safety without encephalitis or edema, 43

| Biomarkers related to neuron axon
Neurofilaments (NFs) comprise three subunits: neurofilament light (NFL), medium, and heavy, and they are involved in axon formation.
They maintain the integrity of neurons and conduct nerve impulses.
When the axon is damaged, NFs are released into the CSF and eventually enter the bloodstream, specifically indicating axonal damage.
NFL has been reported as a potential biomarker in multiple sclerosis, AD, frontotemporal dementia, ALS, atypical Parkinson's disease (PD), and brain injury. 62,63 In AD patients, NFL levels are higher than in age-matched controls in CSF and blood. 63,64 In addition to CSF and blood, the saliva NFL levels were determined by a single molecule array in a large study of 152 patients, with negative data.
As with saliva Tau, NFL was detectable in MCI, AD, and healthy controls, which did not differ significantly. 65 NFL is a sensitive but not a specific marker of axonal damage.
NFL is unable to differentiate nervous system disorders character- Other synaptic-related proteins reflect the physiological synaptic functions. These include human synaptosomal-associated protein-25 (SNAP-25), growth-associated protein-43 (GAP-43), and synaptotagmin, as potential AD biomarkers. A correlation between these modules and AD was found without much specificity. 76 More studies are required to validate the relationship between synapticrelated proteins and AD.

| Biomarkers related to inflammation
Aβ activates various cell receptors and signaling pathways causing inflammation. Microglia and astrocytes are active in neuroinflammation. Thus, several biomolecules generated by these cells could become promising AD biomarkers.

| Triggering receptor expressed on myeloid cells 2 (TREM2)
TREM2 is a transmembrane glycoprotein, highly expressed in microglia and strongly associated with Tau and inflammation. 77 80 Thus, sTREM2 appears to be applied as a prognosis biomarker and provides a perspective for therapeutic drug targets on AD immune response.
Several studies also observed a moderate elevation in CSF sTREM2 levels in AD patients. 81 The peak occurred in the early symptomatic phase, with a slight decline in the clinical dementia stage of AD. 82 A meta-analysis revealed that CSF sTREM2 levels in AD patients were higher than in healthy subjects, suggesting that it could be utilized as a biomarker to diagnose or determine the clinical stages of AD. 83 High levels of sTREM2 can also be observed in other neurodegenerative diseases despite the apparent increase of sTREM2 in AD.
Clarifying its role in AD or complementarity with other specific biomarkers could facilitate its application in AD.

| Glial Fibrillary Acidic Protein (GFAP)
GFAP is a specific brain protein expressed by astrocytes and involved in their structure and movement. Some research demonstrated that astrocytes could respond to neuroinflammation and GFAP has a concomitant change. 84 An elevated GFAP in CSF has been observed in AD, PD, Lewy body dementia (LBD), frontotemporal dementia, and prion disease. 85,86 Moreover, there is an increase in plasma GFAP in AD, PD dementia, and LBD. However, more studies could not show the prominent differences between neurological disorders. The average ratio of AD to controls in CSF was 1.12 (95% CI 0.58-2.15, p = 0.736). 16 Thus, GFAP is not a promising CSF diagnostic biomarker for AD. The inconsistent GFAP changes may be associated with the different affected brain regions and astrocytes distribution. The sensitivity and specificity of serum GFAP differentiating AD patients from healthy age-matched controls are high, reaching 77% and 93%, respectively. These two GFAP indexes are even higher than Aβ 42 and slightly lower than Tau. 85 88 This finding indicated that salivary GFAP could be an excellent biomarker for discriminating controls from MCI or AD patients.
The poor specificity in CSF limits GFAP usage as a biomarker.
However, GFAP is a promising candidate peripheral biomarker providing disease-specific information, considering blood and saliva. 89 Several large-scale studies are required to validate this conclusion.

| Chitinase 3-like 1 (CHI3L1/YKL-40)
YKL-40 is a 4 kDa glycoprotein, also called CHI3L1. This glycoprotein is secreted by microglia; its CSF content is positively associated with t-Tau and p-Tau, 90 while it is negatively correlated with AD progression. 91 YKL-40 increases in CSF and decreases in the serum of AD patients. 90,92 YKL-40, along with sTREM2 in young subjects, revealed better results in predicting AD. 93 Therefore, YKL-40 may be a potential biomarker to predict AD occurrence.

| Biomarkers related to microvascular injury
Cerebrovascular disease was reported more commonly in AD than in other neurodegenerative disorders. 94 Vascular dysregulation plays an important role in AD, even though the molecular pathogenesis of vascular alterations remains undetermined. Some proofs suggested that microvascular injury was associated with disease onset and progression, reducing oxygen and nutrient supply to the brain. 95 This led to subsequent neurotoxic effects, such as inflammation, oxidative stress, and nitric oxide disorder. 96 Vascular dysregulation is related to abnormally high heart-type fatty acid-binding protein (hFABP) levels. 97 It is a potential cardiac damage biomarker for predicting acute myocardial infarction, congestive heart failure, or pulmonary embolism. 98 However, the hFABP levels also increased in CSF and plasma in AD. Therefore, it can be a potential biomarker for predicting and differentiating against AD. 99 Chiasserini et al. demonstrated that the combination of hFABP with p-Tau led to high accuracy in the differential diagnosis of AD from LBD with 76% specificity and 95% sensitivity (AUC = 0.92). 100 Lehallier et al. observed that a combination of hFABP, fibroblast growth factor 4, calcitonin, and tumor necrosis factor-related apoptosis-inducing ligand receptor 3 in CSF, apolipoprotein A-II and cortisol levels in plasma enhanced the reliability of disease status prediction with 80% classification accuracy, 88% sensitivity, and 70% specificity. 101 More combinations and values remain to be explored in the future.

| Biomarkers related to heredity
Amyloid-beta precursor protein, presenilin 1, and presenilin 2 are genes correlated with early-onset AD, especially familial ones.
In contrast, the late-onset of AD is more closely associated with other genes, including apolipoprotein E (ApoE), bridging integra-tor1 (BIN1), and cluster protein (CLU). The ApoE-ε4 allele is the most promising candidate gene biomarker and possesses serious genetic risk factors for AD. 15 Astrocytes in the central nervous system produce ApoE. The more copy numbers of the ApoE-ε4 gene, the higher the susceptibility to AD. 102 The increase in ApoE impairs cholesterol transmission between astrocytes and neurons. This led to cholesterol accumulation in the neurons, affecting the clearance rate of Aβ and increasing its production. 15 ApoE plays various biochemical roles in the brain, including regulating immune cells and Tau tangles. However, the specific mechanism of ApoE has not been fully elucidated in this context. 102

| Biomarkers associated with lipid metabolism
Lipids, including phospholipids, cholesterol, sphingolipids, and related molecules, are the primary component of cell membranes, and the brain is enriched in lipids. 103 Sixty percent of the non-aqueous portion of the brain comprises cerebral lipids. Thus, changes in lipid metabolism could influence brain function. 104 Alterations of fatty acids and cerebral lipid peroxidation were observed in the early AD stage. A higher level of lipid particles in glial cells was seen in the brain of AD patients, indicating abnormal lipid metabolism. 103 Several genes, like ApoE, are also involved in lipid homeostasis. 102 Lipid evaluation analyzes body fluids using highly sensitive MS platforms and high-throughput technologies. Therefore, it has become one of the interests in the search for AD biomarkers.
A few studies determined the lipid changes in CSF and blood.
Several lipid families (fatty acid, sphingolipids, glycerophospholipids, and lipid peroxidation compounds) demonstrated impairments at early AD stages. In contrast, ceramides were elevated compared with healthy age-matched controls. 105 Many studies showed that some kinds of miRNAs are significantly deregulated in disease groups than in control groups. 108 A systematic review summarized a panel of 10 miRNAs, including miR-107, miR-125b, miR-146a, miR-26b, miR-200c, miR-210, miR-30e, miR-34a, miR-485, and miR-34c, which were associated with early AD. 109 However, further research is needed to determine the application of miRNA biomarkers due to conflicting miRNA variations or inconsistency in reported results. 110,111 The main biomarkers of AD are summarized in Table 2.  Table 3.
Bioinformatics combines information software and assay technology for researchers to store, retrieve, visualize, predict, and analyze biomolecular data, significantly improving research efficiency.
Omics, including proteomics, lipidomics, and RNomics were established based on bioinformatics. They enable the identification of the underlying molecules of the disease with increased comprehension, leading the way toward diagnosis and precision medicine. 140,141 Future studies must exploit biomarkers discovered by bioinformatics to achieve clinical translation.

| CON CLUS IONS
Biomarkers exploration is essential in detecting diseases because it can help predict, diagnose, identify, treat diseases, and understand disease processes. In 2021, Aβ 42 , t-Tau, p-Tau, and NFL were incorporated into the diagnostic criteria of AD, marking a significant breakthrough in studying biomarkers for neurological disorders. Other biomarkers, including inflammation-related and synaptic functionrelated biomarkers, are promising for future clinical implications due to their confirmed pathologies in AD. However, since the pathological phenomenon behind microvascular damage is not fully established in AD and could only be a secondary lesion in the late stage, its significance as a biomarker for early diagnosis may be minimal. 96 Similar changes occur in many neurological diseases, and the specificity is not high in AD, even though lipid metabolites may be a new direction. 135 Thus, the possibility of utilizing them as early diagnostic markers for AD is doubtful. MiRNA has been a hot area recently since its length is too short and its nucleotide sequences are not specific. 107 However, the accuracy of precise identification of one particular RNA needs to be enhanced, and the combined detection results of multiple molecules would probably be more reliable. Nevertheless, the application prospect of such biomarkers is also worrying. The common issues with many of them require an urgent solution.
First, some studies on the same potential biomarkers indicated divergent or opposite results (Table 2). Therefore, more in-depth studies are needed to determine the reliability of novel biomarkers.
Second, the sensitivity and specificity of many candidate biomarkers have not been satisfactory enough. To address this issue, scientists have actively experimented strategies that use multiple potential biomarkers, evaluating the relationship between various biomarkers and diseases and gauging whether diagnostic accuracy can be improved using multiple biomarkers. Some results have been achieved. For instance, the ratio of Aβ 42 /Aβ 40 has superior performance than Aβ 42 alone. 115 The increase in the neurogranin/BACE1 ratio predicts cognitive decline. 70,152 The concentrations of YKL-40 and sTREM inflammatory biomarkers are closely associated with the changes in brain imaging. Thus, combining inflammatory biomarkers with imaging can improve the diagnostic yield among subjects. 93 A good combination of biomarkers can enhance the specificity and sensitivity of diseases, thereby compensating for the shortcomings of a single biomarker in practical application.
The third problem is sampling. CSF and blood mainly represent current samples for biomarkers. CSF sampling is more complex, expensive, and invasive than blood, making it unsuitable for repeat operations in a short time. In contrast, blood sampling avoids the above problems with higher clinical feasibility. Inevitably, blood samples can be contaminated with trace amounts of protein or lipids. Hence, these impurities should be accommodated when setting the range for abnormal values. Nevertheless, exploring blood biomarkers in neurological diseases should be the focus of future research. 153 In addition, the discovery of biomarkers also provides direction and ideas for developing new drugs. Targeted agents based on Aβ, BACE1, and Tau are in clinical trials, and aducanumab and lecanemab are already available in the market. These compounds have significant potential to become new drugs for AD treatment.
Although there is little convincing evidence that current biomarkers are sufficiently specific and sensitive to diagnose AD, the potential socioeconomic and medical benefits of developing drugs to treat AD still make the search for biomarkers a fruitful area of study.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors have no conflicts of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing does not apply to this article, and this research does not involve the analysis and innovation of new data.