Serum biomarker combination for predicting curative effect of astragaloside on diabetic peripheral neuropathy and kit and application thereof

Abstract

The application belongs to the technical field of biological medicine and in vitro diagnosis, and discloses a serum biomarker combination for predicting the curative effect of astragaloside on diabetic peripheral neuropathy, a kit and an application thereof. The combination is composed of miR-155-5p, neuron-specific enolase, tumor necrosis factor alpha, arginine and tyrosine trans-nucleic acid-protein-metabolism three-layer combination. In the serum before treatment, stem loop RT-qPCR, ELISA and LC-MS / MS are combined for quantitative input into a multivariate discriminant model, and a response index P is output to complete the responder-non-responder stratification by using a threshold Pc. The kit comprises five independent modules of nucleic acid, protein, amino acid, internal reference quality control and data processing, and is used for assisting in screening astragaloside suitable population, dynamically monitoring response stability and stratifying clinical research.

Description

Technical Field

[0001] This invention belongs to the field of biomedicine and in vitro diagnostic reagent technology, specifically relating to a combination of serum biomarkers and their kits and applications for predicting the efficacy of astragaloside A in the treatment of diabetic peripheral neuropathy. Background Technology

[0002] Diabetic peripheral neuropathy is one of the most common and irreversible chronic complications of type 2 diabetes. Clinically, it manifests as symmetrical numbness, tingling, paresthesia, and decreased nerve conduction velocity in the distal lower extremities. Long-term progression can lead to diabetic foot ulcers, gangrene, and even amputation. Epidemiological data shows that the incidence of diabetic peripheral neuropathy exceeds 50% in patients with diabetes for more than 10 years, making it a major driver of disability and healthcare burden due to diabetic complications. Current clinical intervention strategies for this disease mainly include strict glycemic control, antioxidant stress drugs such as alpha-lipoic acid, neurotrophic agents such as mecobalamin, and aldose reductase inhibitors. However, existing interventions have limited ability to reverse existing nerve damage, and different patients respond significantly differently to the same treatment regimen, necessitating new treatment methods and precise stratification tools.

[0003] Astragaloside A is a natural saponin compound isolated and extracted from Astragalus membranaceus Radix, a plant belonging to the genus Astragalus in the legume family. It is the main monomeric active substance in Astragalus membranaceus, with the molecular formula C41H68O14 and a molecular weight of 784.97 g / mol. Basic pharmacological studies over many years have shown that astragaloside A has a significant protective effect against Schwann cell damage, dorsal root ganglion neuronal apoptosis, neuroinflammation, and oxidative stress in diabetic peripheral neuropathy by regulating multiple signaling axes, including the phosphatidylinositol 3-kinase / protein kinase B / target of rapamycin pathway, the silencing signaling regulator 1 / p53 pathway, the Toll-like receptor 4 / nuclear factor κB pathway, and the nuclear factor-related factor 2 / heme oxygenase 1 pathway. In clinical practice, astragaloside A is used in various forms, including as a monomer or in compound traditional Chinese medicine preparations and proprietary Chinese medicines, as an adjunct treatment for patients with diabetic peripheral neuropathy, showing some effectiveness in improving nerve conduction velocity and alleviating sensory abnormalities.

[0004] Chinese patent application CN101181285A discloses the application of astragaloside IV in the preparation of drugs for treating neurodegenerative diseases. This method uses astragaloside IV as the active ingredient to prepare therapeutic drugs for neurodegenerative diseases and demonstrates its protective effect against neurotoxic damage induced by monosodium glutamate and β-amyloid using a PC12 cell model. However, this method only provides protection from the perspective of drug preparation for treatment and does not address the issue of objectively and quantifiable in vitro stratification of patients' potential treatment response before treatment initiation. The method's implementation is based on the assumption that all target populations will produce a treatment response to astragaloside IV, but in clinical practice, significant heterogeneity in astragaloside response exists. Some patients show significant improvement in nerve conduction velocity and symptom scores after standardized treatment, while others show limited improvement or no response, resulting in non-responders incurring unnecessary medication costs and time losses.

[0005] In their paper "Astragaloside IV alleviates Schwanncell injury in diabetic peripheral neuropathy by regulating microRNA-155-mediated autophagy" published in Phytomedicine (Vol. 92, 2021, p. 153749), Yin et al. proposed that astragaloside IV alleviates Schwann cell myelin damage in a rat model of diabetic peripheral neuropathy by downregulating the microRNA miR-155-mediated autophagy pathway. They demonstrated changes in miR-155 expression levels, phosphatidylinositol 3 kinase / protein kinase B / rapamycin target protein pathway activity, and changes in the key autophagy proteins Beclin-1 and LC3 after astragaloside IV intervention in both GK rats and RSC96 cells. However, the study subjects of this scheme are limited to experimental animal models and in vitro cell models, and the changes in indicators used are retrospective measurements after treatment. It does not transform the above molecular markers into response prediction tools that can be detected in vitro in patient serum samples before treatment, nor does it provide a combination of markers and its discrimination rules that can be used to objectively stratify patients' potential treatment response capabilities before treatment initiation.

[0006] In the field of biomarker patents, several serum biomarkers for the diagnosis of diabetic peripheral neuropathy have been reported, such as tumor necrosis factor-α, interleukin-6, eosinophil chemokine-11, neuron-specific enolase, heat shock protein 27, serum uric acid, formic acid, and various amino acid metabolites. However, these reported biomarkers are all aimed at the diagnostic stratification problem of whether or not one has diabetic peripheral neuropathy, rather than the efficacy prediction problem of whether or not one responds to a specific treatment regimen. The biomarker screening criteria, threshold determination methods, discrimination model construction logic, and application scenarios for these two types of problems are fundamentally different. The biomarker system established by the former cannot be directly transferred to solve the latter. In the direction of precision research in traditional Chinese medicine, a few biomarker panels for evaluating the efficacy of traditional Chinese medicine compound treatments for diseases such as cirrhosis and chronic atrophic gastritis have been reported. However, for the specific drug-disease combination of astragaloside A in the treatment of diabetic peripheral neuropathy, there are currently no reports on biomarker combinations and their detection kits specifically used for stratifying pre-treatment responders to non-responders.

[0007] In summary, the core bottleneck of existing technologies is the lack of a laboratory method for objectively, quantifiable, and reproducibly stratifying patients' potential treatment response using a single in vitro serum test before treatment initiation in the clinical practice of astragaloside A treatment for diabetic peripheral neuropathy. This leads to clinicians being unable to predict patients' response tendencies before making medication decisions, resulting in a series of problems such as drug resource misallocation, delays in patient treatment timing, and uncontrolled heterogeneity in clinical research enrollment. The purpose of this invention is to address this core bottleneck by providing a combination of serum biomarkers and its detection kit that can achieve objective stratification before treatment initiation. Summary of the Invention

[0008] The purpose of this invention is to address the core bottleneck in existing technologies regarding the lack of pre-treatment response stratification methods for astragaloside A treatment of diabetic peripheral neuropathy. This invention provides a combination of serum biomarkers for predicting the efficacy of astragaloside A treatment for diabetic peripheral neuropathy, a detection kit containing specific detection reagents of this combination, and provides the application of this combination in the preparation of diagnostic reagents, kits, chips, or detection devices for in vitro screening of suitable populations. The core technical objective of this invention is to output a quantifiable, repeatable, and interpretable response index through a single in vitro combined detection of peripheral serum samples before patients receive astragaloside A treatment, and to achieve objective stratification of responders and non-responders based on a preset threshold, providing laboratory stratification evidence for clinicians to screen suitable populations for astragaloside A treatment.

[0009] To achieve the above-mentioned invention objectives, the technical solution of the present invention adopts a combined detection strategy of five biomarkers across three biological levels of nucleic acids, proteins, and metabolism. The selection of these five biomarkers is not an arbitrary combination, but is based on an accurate mapping of the known mechanism of action of astragaloside IV in diabetic peripheral neuropathy: the first biomarker, miR-155-5p, corresponds to the autophagy-apoptosis pathway of Schwann cells regulated by astragaloside IV, reflecting the cumulative state of autophagic damage in peripheral nerve Schwann cells; the second biomarker, neuron-specific enolase, corresponds to the cumulative damage of dorsal root ganglion neurons, reflecting the degree of substantial damage to peripheral neurons; the third biomarker, tumor necrosis factor α, corresponds to the downstream nerve inflammation pathway inhibited by astragaloside IV, reflecting the systemic nerve inflammation level; the fourth biomarker, arginine, corresponds to the substrate metabolism pathway of nitric oxide synthase, reflecting the state of the peripheral nerve microvascular endothelial function and nitric oxide signaling axis; the fifth biomarker, tyrosine, corresponds to the catecholamine precursor metabolism pathway, reflecting the peripheral nerve mitochondrial energy metabolism and catecholamine synthesis ability. The detection object of these five biomarkers is uniformly an in vitro peripheral serum sample collected before the start of treatment for patients with diabetic peripheral neuropathy.

[0010] The specific technical solution is as follows: The biomarker combination includes five biomarkers: miR-155-5p, neuron-specific enolase, tumor necrosis factor α, arginine, and tyrosine; miR-155-5p is detected by stem-loop reverse transcription quantitative polymerase chain reaction, and the relative expression level is normalized with U6 small nuclear RNA as an internal reference; neuron-specific enolase and tumor necrosis factor α are detected by sandwich enzyme-linked immunosorbent assay; arginine and tyrosine are detected by high performance liquid chromatography-tandem mass spectrometry, and absolute quantification is completed using a stable isotope internal standard. The measured values of these five biomarkers are input into a preset multivariate discrimination model after standardization. The model is selected from a Logistic regression model, a random forest model, or a support vector machine model, and the output response index P ranges from 0 to 1; based on a preset threshold Pc, when P≥Pc, it is determined as a responder, and when P<Pc, it is determined as a non-responder; where the value range of Pc is 0.55 to 0.75, preferably 0.60 to 0.70, and most preferably 0.65.

[0011] To ensure the industrial and standardized application of the aforementioned detection method, this invention further provides a detection kit comprising a specific detection reagent for the above-mentioned biomarker combination. The kit comprises five independently packaged functional modules: a nucleic acid detection module containing the stem-loop reverse transcription primer shown in SEQ ID NO:2, the upstream primer shown in SEQ ID NO:3, the universal downstream primer shown in SEQ ID NO:4, and the TaqMan probe shown in SEQ ID NO:5 modified with 5′ FAM and 3′ MGB; a protein detection module containing a sandwich enzyme-linked immunosorbent assay (ELISA) reagent targeting neuron-specific enolases and tumor necrosis factor α; and an amino acid detection module containing high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS / MS) quantitative reagents for arginine and tyrosine. 13 C or 2 The kit includes an H-labeled stable isotope internal standard; the internal standard and quality control module contains the U6 small nuclear RNA detection primer pairs shown in SEQ ID NO:6 and SEQ ID NO:7, as well as standard curves and quality control samples with multiple concentration gradients; the data processing module includes a computer-readable storage medium storing the parameter file and threshold Pc parameter of the multivariate discriminant model. The kit can be sold as a standalone detection product or integrated into hospital in vitro diagnostic systems as a stratification tool accompanying astragaloside A treatment regimens.

[0012] This invention also provides the application of the biomarker combination in the preparation of diagnostic reagents, kits, chips, or detection devices for in vitro screening of patients with diabetic peripheral neuropathy suitable for treatment with astragaloside A. The applications cover various scenarios, including baseline stratification before treatment initiation, dynamic monitoring at weeks 4, 8, and 12 after treatment initiation, dose adjustment decision support during long-term treatment, and stratification of enrolled populations in clinical studies, thus forming a complete protection chain for the biomarker combination from product form to industrial preparation to clinical application.

[0013] Compared with existing technologies, this invention achieves the following beneficial effects. First, this invention provides for the first time a combination of serum biomarkers specifically for predicting the efficacy of astragaloside A in treating diabetic peripheral neuropathy, filling the gap that existing diagnostic stratification biomarkers are all geared towards disease diagnosis and cannot be used for drug response prediction, thus elevating the clinical application of astragaloside A from empirical dosing to precise laboratory-based stratified dosing. Second, the five biomarkers span three biological levels: nucleic acid, protein, and metabolism, and precisely correspond to the five main pharmacodynamic pathways of astragaloside A. A one-to-one mapping mechanism is formed between the biomarker combination and the multiple drug pathways, providing both objective data support and clear biological interpretation for stratification determination, unlike conventional combinations that simply select five DPN-related factors by chance. Third, the five-element combination exhibits a significant synergistic gain in discriminative power compared to any single biomarker or any sub-combination within the same stratum, and the combined response index P consistently outputs reproducible stratification results in both the training and independent validation sets. Fourth, the test kit is modularly designed and individually packaged. The three types of detection—nucleic acid, protein, and amino acid—are performed using mature commercial platforms such as RT-qPCR, ELISA, and LC-MS / MS, respectively. No new, undisclosed instruments or reagents need to be developed, allowing for direct application in the laboratory departments of ordinary tertiary hospitals or independent third-party medical laboratories. Fifth, the application method explicitly limits the test subjects to serum samples that have been removed from the human body, and the output results are for auxiliary stratification rather than diagnostic conclusions or treatment decisions. Attached Figure Description

[0014] Figure 1 This is a comparison graph of the response index P of the five-element biomarker combination described in the embodiments of the present invention and the receiver operating characteristic curves of each single biomarker used to predict the efficacy of astragaloside in the treatment of diabetic peripheral neuropathy. The vertical axis is sensitivity, and the horizontal axis is 1 minus specificity. The larger the area under the curve, the higher the discriminative power.

[0015] Figure 2 This is a box plot showing the horizontal distribution of the five biomarkers described in this invention in the responder group, non-responder group, and healthy control group. The vertical axis represents the standardized level of each biomarker, and the differences between groups were assessed using the nonparametric Mann-Whitney U test.

[0016] Figure 3 This is a schematic diagram of the multivariate discriminant model for predicting the efficacy of astragaloside A in the treatment of diabetic peripheral neuropathy as described in this invention. It shows the complete closed-loop logic from serum sample collection, five biomarker detection, data standardization, response index P calculation to responder-non-responder stratification.

[0017] Figure 4This is a distribution density map of the response index P described in this embodiment of the invention in the training set and the independent validation set of responders and non-responders. The vertical axis is the probability density, and the horizontal axis is the range of the response index P (0~1). The vertical dashed line in the figure marks the optimal threshold Pc.

[0018] Figure 5 The graphs show the changes in sural nerve conduction velocity at weeks 4, 8, and 12 after standardized treatment with astragaloside A in the responder and non-responder populations stratified according to the five-element combination of the present invention, as described in the embodiments of the present invention. The vertical axis represents the improvement value of sural nerve conduction velocity relative to the baseline, and the horizontal axis represents the treatment time.

[0019] Figure 6 This is a schematic diagram of the structure of the detection kit of the present invention, showing the internal composition and synergistic relationship between the five functional modules: nucleic acid detection module, protein detection module, amino acid detection module, internal component and quality control module, and data processing module. Detailed Implementation

[0020] To make the objectives, technical solutions, and effects of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are for illustrative purposes only and do not constitute a limitation on the scope of protection of this invention. Unless otherwise stated, all reagents used in this invention are of analytical grade or higher purity, and all commercial kits used are operated according to the manufacturer's instructions.

[0021] Example 1: Clinical Sample Cohort Design and Sample Collection

[0022] This embodiment describes the clinical sample cohort design and sample processing procedure used for training and validating the biomarker combinations described in this invention. This study was approved by the applicant's institution's ethics committee (ethics approval number SH-Y-2025-082), and all subjects signed written informed consent forms. A total of 250 patients with type 2 diabetes and peripheral neuropathy were included (150 in the training cohort and 100 in the independent validation cohort), as well as 60 age-, sex-, and body mass index-matched healthy controls.

[0023] Patients with diabetic peripheral neuropathy included in the study met the following criteria: diabetes duration ≥ 5 years; glycated hemoglobin (HbA1c) within the range of 6.5%–10.0%; Michigan Neuropathy Screening Scale (MSS) score ≥ 2; sural nerve conduction velocity ≤ 40 m / s. Exclusion criteria included peripheral neuropathy caused by other reasons (such as vitamin B12 deficiency neuropathy, alcoholic neuropathy, chemotherapy-related neuropathy, chronic inflammatory demyelinating polyneuropathy, etc.) and those with severe hepatic or renal insufficiency (alanine aminotransferase or aspartate aminotransferase levels ≥ 3 times the upper limit of normal, estimated glomerular filtration rate < 30 mL / min / 1.73 m³ / min).2 Patients were included in the study. Peripheral venous blood samples were collected from all patients on the day of enrollment before treatment, and they began 12 weeks of standardized treatment with astragaloside A the following day (oral dose of 60 mg / day, 20 mg three times a day for 12 weeks).

[0024] The gold standard for responder-nonresponder grouping was determined based on clinical improvement after 12 weeks of treatment. Patients were defined as responders if their sural nerve conduction velocity improved by ≥3 m / s relative to baseline and their Michigan Neuropathy Screening Scale score decreased by ≥1 point relative to baseline after 12 weeks of treatment; otherwise, they were defined as nonresponders. Of the 250 patients, 122 were ultimately identified as responders (response rate 48.8%), and 128 were identified as nonresponders (nonresponse rate 51.2%), which is generally consistent with the reported response rates of traditional Chinese medicine monomers in the treatment of diabetic peripheral neuropathy in the literature.

[0025] The serum sample collection and processing procedure was as follows: 8 mL of venous blood was collected from subjects in the morning between 7:00 and 8:30 AM on an empty stomach using vacuum blood collection tubes without anticoagulants. After collection, the blood was allowed to clot naturally at room temperature (22°C) for 30-60 minutes. The sample was then centrifuged at 3000 rpm for 10 minutes at 4°C. The supernatant serum was aliquoted into 1.5 mL RNase-free centrifuge tubes (200 μL per tube) and immediately transferred to a -80°C freezer for storage. For subsequent testing, each sample was subjected to ≤2 freeze-thaw cycles. All sample collection and processing procedures were performed by standardized training research nurses and laboratory technicians according to standard operating procedures, and the collection and processing times were recorded in the sample log. The clinical baseline characteristics of the two cohorts are compared in Table 1.

[0026] Table 1: Comparison of clinical baseline characteristics between the training cohort and the independent validation cohort

[0027] feature Training queue (n=150) Verification queue (n=100) Healthy control group (n=60) p-value Age (years) 58.3±9.7 57.9±10.2 57.5±8.9 0.842 Male ratio 54.7% 53.0% 55.0% 0.953 Duration of diabetes (in years) 11.2±5.4 10.8±5.1 not applicable 0.567 Glycated hemoglobin (%) 8.1±1.3 8.0±1.2 5.4±0.4 <0.001 Fasting blood glucose (mmol / L) 8.7±2.4 8.5±2.3 5.1±0.5 <0.001 <![CDATA[Body mass index (kg / m 2 )]]> 25.6±3.2 25.4±3.1 24.8±2.7 0.215 MNSI score 4.2±1.6 4.1±1.7 0.3±0.5 <0.001 Surgical nerve conduction velocity (m / s) 34.2±4.8 34.5±4.6 48.7±3.2 <0.001 Respondent ratio 49.3%（74 / 150） 48.0%（48 / 100） not applicable 0.834

[0028] As shown in Table 1, there were no statistically significant differences between the training cohort and the independent validation cohort in baseline characteristics such as age, gender, duration of diabetes, glycated hemoglobin, fasting blood glucose, body mass index, MNSI score, and sural nerve conduction velocity (P>0.05). The proportion of respondents in the two cohorts was also basically the same (49.3% vs 48.0%, P=0.834), proving that the two cohorts were independent samples from the same population and could be used for model training and external independent validation.

[0029] Example 2: Serum miR-155-5p stem-loop RT-qPCR detection

[0030] This embodiment describes the stem-loop reverse transcription quantitative polymerase chain reaction (PCR) assay procedure for serum miR-155-5p, including serum total RNA extraction, stem-loop reverse transcription reaction, quantitative polymerase chain reaction amplification, and relative expression level calculation. The assay system uses seven sequences from SEQ ID NO:1 to SEQ ID NO:7.

[0031] Serum total RNA extraction was performed using the miRN easySerum / Plasma Kit (QIAGEN, catalog number 217184) according to the manufacturer's instructions. The specific procedure was as follows: 200 μL of serum sample was added to 1000 μL of QIAzol lysis buffer, vortexed for 10 s, and incubated at room temperature for 5 min; 200 μL of chloroform was added, vigorously vortexed for 15 s, and incubated at room temperature for 3 min; centrifuged at 12000g for 15 min at 4°C; approximately 600 μL of the upper aqueous phase was carefully transferred to a new centrifuge tube, and an equal volume of anhydrous ethanol was added and gently mixed; the mixture was transferred to an RNeasy MinElute column, and washing and elution were performed according to the manufacturer's instructions. Finally, total RNA was eluted with 14 μL of RNase-free water. During extraction, cel-miR-39 spike-in was added to control the final concentration (1.6 × 10⁻⁶). 8 (Copies / μL) to monitor extraction efficiency.

[0032] The total volume of the stem-loop reverse transcription reaction system was 20 μL, containing 5 μL template RNA, 50 nmol / L stem-loop reverse transcription primers as shown in SEQ ID NO:2, 0.25 mmol / L dNTPs, 50 U SuperScript III reverse transcriptase (Invitrogen, catalog number 18080-093), 4 U RNase inhibitor (Promega, catalog number N2511), 4 μL 5× first-strand synthesis buffer, and 1 μL 0.1 mol / L dithiothreitol. The reverse transcription program was as follows: stem-loop primer annealing and initial extension at 16°C for 30 min; main extension at 42°C for 30 min; reverse transcriptase inactivation at 85°C for 5 min; and storage at 4°C. An internal control U6snRNA reverse transcription reaction (using the reverse primers shown in SEQ ID NO:7 as stem-loop substitute primers) and a template-free control were also included.

[0033] Quantitative polymerase chain reaction (PCR) was performed on a 7500 FastReal-Time PCR System (Applied Biosystems). The total reaction volume was 20 μL, containing 10 μL of TaqMan Universal Master Mix II (UNG-free), 0.4 μmol / L of the upstream primer shown in SEQ ID NO:3, 0.4 μmol / L of the universal downstream primer shown in SEQ ID NO:4, 0.2 μmol / L of the 5′-FAM / 3′-MGB dual-labeled TaqMan probe shown in SEQ ID NO:5, 1.5 μL of reverse transcription product template, and nuclease-free water to a final volume of 20 μL. The amplification program was as follows: 95°C pre-denaturation for 10 min; 40 cycles of 95°C denaturation for 15 s and 60°C annealing extension for 60 s; signal acquisition was performed during the 60°C annealing extension phase. The U6 snRNA internal control was amplified using the primer pairs shown in SEQ ID NO:6 and SEQ ID NO:7, employing the SYBR Green detection system. The amplification program was consistent with that of miR-155-5p. Three technical replicates were performed for each sample, with the relative standard deviation within each group controlled to within 5%.

[0034] The relative expression level of miR-155-5p was calculated using 2 -ΔΔCt Methods. Where ΔCt = Ct(miR-155-5p) − Ct(U6), ΔΔCt = ΔCt(sample) − average ΔCt(healthy control group). The detection limit of miR-155-5p for all samples was less than 10 copies / reaction, and the linear range of the amplification curves covered 5 orders of magnitude (10 to 10^6). 6 (Copy / reaction), amplification efficiency maintained between 95% and 105%. If the Ct value of cel-miR-39spike-in deviates from the average value by ±1 cycle during the detection process, the RNA extraction of that sample is deemed unqualified and must be re-extracted.

[0035] Example 3: Detection of serum neuron-specific enolase and tumor necrosis factor α

[0036] This embodiment describes the detection procedure for a sandwich enzyme-linked immunosorbent assay (ELISA) of serum neuron-specific enolase and tumor necrosis factor α. Neuron-specific enolase was detected using the Human NSE ELISA Kit (R&D Systems, catalog number DY3947), and tumor necrosis factor α was detected using the Human TNF-alpha Quantikine HS ELISA Kit (R&D Systems, catalog number HSTA00E). Both kits are CE / FDA certified and exhibit good intra- and inter-assay reproducibility.

[0037] The procedure for detecting neuron-specific enolase is as follows: Frozen serum is slowly thawed at 4°C for 30 min and diluted 1:50 with kit diluent. 100 μL of the diluted sample is added to a 96-well plate pre-coated with anti-NSE monoclonal antibody and incubated at room temperature for 2 h. After washing four times with washing buffer, 100 μL of biotinylated detection antibody is added and incubated at room temperature for 1 h. After washing, 100 μL of streptavidin-horseradish peroxidase conjugate is added and incubated at room temperature in the dark for 30 min. After washing, 100 μL of 3,3′,5,5′-tetramethylbenzidine substrate solution is added and incubated at room temperature in the dark for 15 min. The reaction is terminated by adding 50 μL of stop solution (2 mol / L sulfuric acid). The absorbance value is read at 450 nm within 30 min, with a reference wavelength of 540 nm used for background subtraction. Two technical replicates were performed for each sample, with the relative standard deviation within each group controlled within 8%. A standard curve with 7 points was set for each plate (concentration gradient of 1.56, 3.13, 6.25, 12.5, 25.0, 50.0, and 100.0 ng / mL) as well as three quality control samples of high, medium, and low (corresponding to 75.0, 25.0, and 5.0 ng / mL, respectively).

[0038] The tumor necrosis factor α (TNF-α) assay employed a high-sensitivity version of the Quantikine HS system. The detection principle is the same as that of neuron-specific enolase, but the substrate color development utilizes enzyme-amplified fluorescent substrates to enhance detection sensitivity to below 0.5 pg / mL. The specific procedure was as follows: Frozen serum was thawed and used directly without dilution. 100 μL of sample was added to a 96-well plate pre-coated with anti-TNF-α monoclonal antibody and incubated at room temperature for 3 h. After washing, 100 μL of enzyme-linked antibody was added and incubated at room temperature for 2 h. After washing, 100 μL of fluorescence-enhancing substrate solution was added and incubated at 37°C in the dark for 60 min. Fluorescence signals were read using a fluorescence microplate reader with an excitation wavelength of 320 nm and an emission wavelength of 405 nm. Seven standard curves were set for each plate (concentration gradients of 7.81, 15.6, 31.3, 62.5, 125, 250, and 500 pg / mL) and three quality control samples (corresponding to 350, 100, and 20 pg / mL, respectively).

[0039] The test results of all samples were obtained by calculating the standard curve through a four-parameter logistic curve fitting. The correlation coefficient R of the standard curve was... 2All values ​​were not lower than 0.995. The recovery rates of quality control samples were all between 85% and 115%, and samples exceeding this range were retested. The sensitivity of the neuron-specific enolase assay was 0.4 ng / mL (below the ≤0.5 ng / mL threshold), the intra-assay coefficient of variation was 5.7% (below the ≤8% threshold), and the inter-assay coefficient of variation was 9.2% (below the ≤12% threshold); the sensitivity of the tumor necrosis factor α assay was 1.5 pg / mL (below the ≤2 pg / mL threshold), the intra-assay coefficient of variation was 6.3%, and the inter-assay coefficient of variation was 10.8%, both meeting the above-mentioned requirements.

[0040] Example 4: LC-MS / MS Quantitative Detection of Serum Arginine and Tyrosine

[0041] This embodiment describes the quantitative detection procedure of serum arginine and tyrosine by high performance liquid chromatography-tandem mass spectrometry. The detection platform is an Agilent 1290 Infinity II UHPLC system tandem with an Agilent 6470 Triple Quadrupole mass spectrometer, employing electrospray ionization positive ion mode and multiple reaction monitoring (MRM) scanning.

[0042] The sample pretreatment procedure is as follows: Take 50 μL of thawed serum sample and add 200 μL of... 13 C6-arginine (final concentration 10 μmol / L) and 13 The protein solution containing C9-tyrosine (final concentration 10 μmol / L) was precipitated in methanol by vortexing for 30 s and then allowed to stand at 4°C for 20 min to complete the protein precipitation. The solution was then centrifuged at 13000g for 10 min at 4°C. 200 μL of the supernatant was added to 200 μL of an aqueous solution containing 0.1% formic acid for dilution. The solution was filtered through a 0.22 μm nylon filter and transferred to a vial, which was then placed in an autosampler at 4°C for analysis.

[0043] The chromatographic conditions were as follows: a Waters ACQUITY UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 μm) was used, with a column temperature of 40°C; mobile phase A was an aqueous solution containing 0.1% formic acid (v / v), and mobile phase B was an acetonitrile solution containing 0.1% formic acid (v / v); the gradient elution program was as follows: 0–1 min, mobile phase B maintained at 2%; 1–5 min, mobile phase B linearly increased from 2% to 30%; 5–6 min, mobile phase B increased from 30% to 95%; 6–7 min, mobile phase B maintained at 95%; 7–7.1 min, mobile phase B returned to 2%; 7.1–9 min, mobile phase B maintained at 2% to equilibrate the column. The flow rate was 0.4 mL / min, and the injection volume was 2 μL. Under these chromatographic conditions, the retention time of arginine was approximately 0.85 min, and the retention time of tyrosine was approximately 3.42 min, with resolutions greater than 1.5 for both.

[0044] Mass spectrometry conditions were as follows: electrospray ionization positive ion mode, capillary voltage 3500 V, nozzle voltage 500 V, sheath gas temperature 350°C, sheath gas flow rate 11 L / min, drying gas temperature 250°C, drying gas flow rate 8 L / min, and nebulizer gas pressure 35 psi. Multiple reaction monitoring (MRM) ion pairs: quantitative ion pair for arginine m / z 175→70 (collision energy 18 eV), qualitative ion pair m / z 175→116 (collision energy 12 eV). 13 C6-arginine internal standard m / z 181→74 (collision energy 18eV); tyrosine quantitative ion pair m / z 182→136 (collision energy 14eV), qualitative ion pair m / z 182→91 (collision energy 22eV); 13 C9-tyrosine internal standard m / z 191→144 (collision energy 14eV). The residence time for each monitored ion pair was set to 50ms.

[0045] Quantitative curves were constructed using the stable isotope dilution internal standard method. The concentration gradients for arginine were 5, 10, 25, 50, 100, 150, and 200 μmol / L, and the concentration gradients for tyrosine were 5, 10, 25, 50, 75, 100, and 150 μmol / L. The correlation coefficient R0 of the curves was [not specified]. 2 All values ​​were not lower than 0.998; the lower limit of quantification (LOQ) for arginine was 2.5 μmol / L and the upper limit of quantification (LOQ) was 200 μmol / L; the LQ for tyrosine was 2.5 μmol / L and the LQ for tyrosine was 150 μmol / L, all meeting the above requirements. Three quality control levels were set up for each plate (20, 80, and 150 μmol / L for arginine; and 20, 60, and 120 μmol / L for tyrosine). The recoveries of the quality control samples were all between 90% and 110%, and the intra-group relative standard deviation was controlled within 5%.

[0046] Example 5: Construction of Multivariate Discriminant Model and Determination of Threshold

[0047] This embodiment describes the complete process of constructing a multivariate logistic regression discriminant model based on training queue data and determining the optimal threshold Pc. For example... Figure 3 As shown, the workflow of the discriminant model consists of five steps: serum sample collection, joint detection of five biomarkers, data standardization preprocessing, response index P calculation, and stratification of responders and non-responders, forming a complete objective stratification closed loop.

[0048] The raw data preprocessing employed Z-score standardization, i.e., X_standardized = (X − μ_train) / σ_train, where μ_train and σ_train are the mean and standard deviation of each biomarker in the training cohort, respectively. This standardization process eliminates differences in the units and ranges of different biomarkers, ensuring that the five features have comparable initial weights in the Logistic Regression model. The mean and standard deviation parameters from Z-score standardization are fixed as training set parameters and deployed with the model. The validation set and future clinical samples will also be standardized using these fixed parameters.

[0049] The maximum likelihood estimation method was used to estimate the regression coefficients of the Logistic regression model. On a training cohort of 150 cases, the Logit class of the Python statsmodels0.14 package was used to complete the model fitting, and the iteration convergence criterion was set to coefficient change < 10. -6 The maximum number of iterations was 100. The regression equation output by the model was P=1 / (1+exp(−Z)), where Z=β0+β1·X1+β2·X2+β3·X3+β4·X4+β5·X5, and X1 to X5 were the standardized values ​​of miR-155-5p relative expression level, neuron-specific enolase concentration, tumor necrosis factor α concentration, arginine concentration, and tyrosine concentration, respectively. The estimated regression coefficients of the model are shown in Table 2.

[0050] Table 2: Estimation of regression coefficients and significance test of the Logistic regression discriminant model

[0051] variable regression coefficient β Standard error SE Z-statistic p-value 95% confidence interval <![CDATA[Intercept β0]]> 0.124 0.218 0.569 0.569 [-0.303,0.551] <![CDATA[miR-155-5p(X1)]]> -1.247 0.298 -4.184 <0.001 [-1.831,-0.663] <![CDATA[NSE(X2)]]> -0.892 0.241 -3.701 <0.001 [-1.364,-0.420] <![CDATA[TNF-α(X3)]]> -0.751 0.226 -3.323 0.001 [-1.194,-0.308] <![CDATA[Arg(X4)]]> 0.823 0.235 3.502 <0.001 [0.363,1.283] <![CDATA[Tyr(X5)]]> 0.685 0.213 3.216 0.001 [0.268,1.102]

[0052] As shown in Table 2, the regression coefficients of all five biomarkers were statistically significant (P<0.01). The regression coefficients for miR-155-5p, neuron-specific enolase, and tumor necrosis factor α were all negative, indicating that higher baseline levels of these three markers correlated with a greater tendency for non-response in patients. The regression coefficients for arginine and tyrosine were both positive, indicating that higher baseline levels of these two markers correlated with a greater tendency for response in patients. This correlation is entirely consistent with the multi-pathway pharmacodynamic mechanism of astragaloside A: high baseline miR-155-5p reflects long-term high activation of the Schwann cell autophagy-apoptosis pathway, exceeding the pharmacological regulatory threshold of astragaloside A; high baseline neuron-specific enolase reflects irreversible neuronal damage accumulation; high baseline tumor necrosis factor α reflects systemic neuroinflammation exceeding the inhibitory capacity of the drug; high baseline arginine suggests sufficient nitric oxide synthase substrate and room for microvascular function regulation; and high baseline tyrosine suggests good mitochondrial energy metabolism reserves, allowing the drug to amplify its protective effect through metabolic pathways.

[0053] Based on the response index P distribution of the training queue, the Youden index (Youdenindex = sensitivity + specificity − 1) is used to determine the optimal threshold Pc. The Youden index is calculated every 0.01 within the range of P (0~1), finally reaching its maximum value at Pc = 0.65 (Youden index = 0.624), corresponding to a sensitivity of 81.1% and a specificity of 81.6%. This invention limits the value range of Pc to 0.55~0.75 (covering the clinically adjustable range near the optimal threshold), preferably 0.60~0.70 (covering the core range where both sensitivity and specificity are higher than 75%), and most preferably 0.65 (the precise value based on maximizing the Youden index in this embodiment). These three limits converge layer by layer, forming a complete three-piece protection for the optimal threshold.

[0054] Example 6: Diagnostic efficacy evaluation of various biomarkers in the training set

[0055] This embodiment describes receiver operating characteristic (ROC) curve analysis of the pentagonal combined response index P and the discriminative power of each individual biomarker in a training cohort. For example... Figure 1 As shown, the receiver operating characteristic (ROC) curve of the five-factor combined response index P on the training set deviates significantly from the diagonal, and its area under the curve (AUC) is significantly higher than that of any single biomarker. Specific ROC curve analysis results are shown in Table 3.

[0056] Table 3: Comparison of the pentagonal combination response index P and the diagnostic efficacy of each individual biomarker in the training cohort.

[0057] index AUC 95% CI Sensitivity Specificity PPV NPV Yoden Index Five-element combination response index P 0.892 [0.842,0.942] 81.1% 81.6% 80.0% 82.7% 0.624 miR-155-5p single marker 0.745 [0.668,0.822] 70.3% 69.7% 67.5% 72.5% 0.394 NSE single marker 0.728 [0.649,0.807] 67.6% 68.4% 66.2% 69.7% 0.358 TNF-α single marker 0.701 [0.620,0.782] 64.9% 65.8% 63.5% 67.1% 0.305 Arg single marker 0.685 [0.602,0.768] 63.5% 63.2% 60.8% 65.8% 0.265 Tyr single marker 0.672 [0.589,0.755] 62.2% 61.8% 59.7% 64.5% 0.240

[0058] As shown in Table 3, the AUC of the pentagonal combined response index P on the training set reached 0.892, with a 95% confidence interval of [0.842, 0.942], which is significantly higher than the AUC of any single biomarker (the highest single biomarker miR-155-5p AUC is only 0.745, and the combined AUC is 0.147 higher than it). The core indicators such as sensitivity (81.1%), specificity (81.6%), positive predictive value (80.0%), and negative predictive value (82.7%) are all at a high level, which significantly exceeds the clinical usability threshold of AUC ≥ 0.85 required for the adequacy of biomarker patent disclosure.

[0059] like Figure 2As shown, all five biomarkers exhibited significant inter-group differences (Mann-Whitney U test, P < 0.001) among the responders (n = 74), non-responders (n = 76), and healthy controls (n = 60). Specifically, miR-155-5p, neuron-specific enolase, and tumor necrosis factor α levels were significantly higher in the non-responders than in the responders, reflecting a high baseline indicating poor response; arginine and tyrosine levels were significantly higher in the responders than in the non-responders, reflecting a high baseline indicating good response. The specific levels of each biomarker in the responders, non-responders, and healthy controls are shown in Table 4.

[0060] Table 4: Levels of various biomarkers in the training cohort in the responding group, non-responding group, and healthy control group.

[0061] markers Response group (n=74) No response group (n=76) Healthy control group (n=60) P-value (response vs. no response) miR-155-5p (relative expression fold) 1.85±0.62 3.94±1.13 1.00±0.21 <0.001 NSE (ng / mL) 15.4±4.7 24.8±6.3 8.2±2.1 <0.001 TNF-α (pg / mL) 32.6±11.4 58.9±18.2 12.3±3.7 <0.001 Arg (μmol / L) 95.3±18.7 68.4±15.9 102.6±14.3 <0.001 Tyr (μmol / L) 78.2±14.5 58.7±12.6 82.4±11.8 <0.001

[0062] Example 7: External Validation of Independent Validation Queues

[0063] This embodiment describes the external validation of a trained pentagonal combined response index P model on an independent validation cohort (n=100). The validation process is as follows: the raw measurements of the five biomarkers of the validation cohort samples are standardized using the Z-score standardization parameters (μ_train, σ_train) fixed in the training set; the response index P for each patient is calculated using the regression coefficients (β0 to β5) fixed in the training set; and the optimal threshold Pc=0.65 determined in the training set is used as the stratification criterion.

[0064] like Figure 4 As shown, the response index P of the respondent group (n=48) in the validation set is generally right-skewed, with the distribution center located at P≈0.78; the response index P of the non-respondent group (n=52) is generally left-skewed, with the distribution center located at P≈0.42. The two distributions form a clear separation boundary at the threshold of Pc=0.65, and the overlapping area is narrower than that of the training set, indicating that the model maintains good discriminative ability on independent cohorts. The diagnostic efficacy metrics of the validation set are shown in Table 5.

[0065] Table 5: Comparison of diagnostic performance between the training queue and the independent validation queue

[0066] index Training set (n=150) Validation set (n=100) difference AUC 0.892 0.876 -0.016 95% CI [0.842,0.942] [0.812,0.940] — Sensitivity 81.1% 79.2% -1.9 percentage points Specificity 81.6% 78.8% -2.8 percentage points Positive predictive value 80.0% 77.6% -2.4 percentage points Negative predictive value 82.7% 80.4% -2.3 percentage points Yoden Index 0.624 0.580 -0.044

[0067] As shown in Table 5, the AUC of the validation set (0.876) decreased by only 0.016 compared to the training set (0.892). The differences in sensitivity, specificity, positive predictive value, and negative predictive value were all less than 3 percentage points, demonstrating that the model has stable generalization ability on independent cohorts and shows no signs of overfitting on the training set. The lower limit of the 95% confidence interval for the validation set AUC is 0.812, which is still significantly higher than the AUC ≥ 0.7 threshold for higher accuracy in diagnostic tests.

[0068] Example 8: Comparison of clinical outcomes between responders and non-responders based on the stratification of this invention

[0069] This embodiment describes a comparison of the improvement curves of sural nerve conduction velocity between responders and non-responders during 12 weeks of standardized astragaloside A treatment, after stratification based on the five-element combined response index P of this invention. All 250 patients completed baseline stratification before treatment initiation and then received the same astragaloside A treatment regimen. Sural nerve conduction velocity was remeasured at weeks 4, 8, and 12 of treatment. The results are as follows: Figure 5 As shown.

[0070] like Figure 5 As shown, the sural nerve conduction velocity in the responder population (n=122) improved from week 4 of treatment (average increase of 1.2 m / s), reaching 2.7 m / s by week 8 and 4.5 m / s by week 12, with the improvement curve exhibiting a stable monotonically increasing trend. In contrast, the sural nerve conduction velocity in the non-responder population (n=128) improved by only 0.3 m / s on average at week 4, 0.6 m / s at week 8, and 0.9 m / s at week 12, consistently failing to reach the gold standard response threshold (≥3 m / s). The difference in sural nerve conduction velocity improvement between the two populations at week 12 reached 3.6 m / s (4.5 m / s improvement in responders vs. 0.9 m / s improvement in non-responders), with a t-test P<0.001, constituting direct evidence of the clinical translational value of the stratified scheme of this invention.

[0071] Example 9: Validation of the cellular mechanism of biomarker combinations

[0072] This embodiment describes the validation of the dose-response relationship between the miR-155-5p biomarker and astragaloside A intervention at the cellular level, supporting the biological rationale for using biomarker combinations to predict therapeutic efficacy from a mechanistic perspective. miR-155-5p was selected as the representative biomarker for this validation in cellular experiments because it has the largest absolute regression coefficient in the combination of this invention (|β1|=1.247) and exhibits a clear direct regulatory relationship with astragaloside A.

[0073] A high glucose injury model was established using the rat Schwann cell line RSC96 (Shanghai Cell Bank, Chinese Academy of Sciences, catalog number SCSP-606). RSC96 cells were cultured in DMEM high glucose medium containing 10% fetal bovine serum, 100 U / mL penicillin, and 100 μg / mL streptomycin, and passaged at 37°C and 5% CO2. Cells in the logarithmic growth phase were cultured at 5 × 10⁻⁶ cells / year. 4 Cells were seeded in 6-well plates and, after overnight adhesion, were divided into 5 groups: a normal control group (5.5 mmol / L glucose), a high glucose injury group (30 mmol / L glucose), and three concentration gradient groups of high glucose plus astragaloside IV (30 mmol / L glucose + astragaloside IV at concentrations of 5, 20, and 80 μmol / L). Total RNA was collected from each group after 48 h of treatment, and the relative expression level of miR-155-5p was detected using the stem-loop RT-qPCR procedure described in Example 2, with three biological replicates per group.

[0074] The results showed that the relative expression level of miR-155-5p in RSC96 cells of the high glucose-induced injury group was 3.74 times higher than that in the normal control group (P<0.001), confirming that the high glucose environment induced the upregulation of miR-155-5p in Schwann cells. After intervention with astragaloside A, the expression level of miR-155-5p decreased in a dose-dependent manner, with a decrease of 18.7% in the 5 μmol / L astragaloside A group (P<0.05), 41.2% in the 20 μmol / L astragaloside A group (P<0.001), and 62.5% in the 80 μmol / L astragaloside A group (P<0.001). The dose-response curve showed a typical S-shaped saturation pattern, and the estimated half-maximal effective concentration (EC50) was [value missing]. 50 The concentration was approximately 24.3 μmol / L. This result confirms at the molecular level the direct regulatory capacity of astragaloside A on miR-155-5p and provides a mechanistic basis for using baseline miR-155-5p levels before treatment to predict response. The higher the baseline level in patients, the greater the regulatory potential that astragaloside A needs to overcome, and the more limited the treatment response.

[0075] Comparative Example 1: Discrimination power of five-element combinations lacking any single marker

[0076] To demonstrate the synergistic effect and irreplaceability among the markers in the five-element combination of this invention, this comparative example uses the leave-one-out method to evaluate the discriminative power of the training queue after sequentially deleting each marker from the five-element combination. Specifically, one marker is sequentially deleted from the five-element combination, and the remaining four markers are used to refit the Logistic regression model on the training queue. The AUC of the training set is then calculated, and the results are shown in Table 6.

[0077] Table 6: AUC variation of the five-element combination leave-one-out method (training set n=150)

[0078] Comparison Combinations AUC AUC decrease The percentage point decrease relative to the complete portfolio Complete 5-element combination (benchmark) 0.892 — — Missing miR-155-5p (NSE+TNF-α+Arg+Tyr only) 0.788 -0.104 -11.7% NSE deficiency (only miR-155+TNF-α+Arg+Tyr) 0.815 -0.077 -8.6% TNF-α deficiency (miR-155+NSE+Arg+Tyr only) 0.832 -0.060 -6.7% Arg deficiency (only miR-155+NSE+TNF-α+Tyr) 0.836 -0.056 -6.3% Tyr deficiency (only miR-155+NSE+TNF-α+Arg) 0.841 -0.051 -5.7%

[0079] As shown in Table 6, deleting any one biomarker significantly reduced the AUC. The largest decrease was observed in the combination lacking miR-155-5p (AUC = 0.788, a decrease of 11.7%), while the smallest decrease was observed in the combination lacking Tyr (AUC = 0.841, a decrease of 5.7%). The AUC of all combinations lacking one biomarker was lower than that of the complete quintuple combination, demonstrating that the five biomarkers described in this invention do not exhibit redundancy within the combination and that the quintuple combination possesses an inherent synergistic effect and is irreplaceable. This result further illustrates that the biomarker selection in this invention is not a simple aggregation of any five DPN-related factors, but rather an indivisible synergistic unit formed after precise screening.

[0080] Comparative Example 2: A comparison of the discriminative power of single-layer marker combinations and cross-layer combinations.

[0081] To demonstrate the synergistic advantage of the three-layer combination across nucleic acid, protein, and metabolism in this invention compared to any single-layer combination, this comparative example further fitted logistic regression models on the training cohort for three single-layer three-marker combinations: nucleic acid layer only (containing miR-155-5p and two other DPN-related miRNAs reported in the literature: miR-21-5p and miR-146a-5p), protein layer only (containing NSE, TNF-α, and IL-6), and metabolism layer only (containing Arg, Tyr, and Glu), and compared the AUC. The results showed that the training set AUC of the nucleic acid layer only combination was 0.812, the AUC of the protein layer only combination was 0.804, and the AUC of the metabolism layer only combination was 0.798. The AUC of the three single-layer combinations was significantly lower than the AUC of the five-element combination across the three layers in this invention (0.892), with a decrease of more than 8 percentage points.

[0082] The comparative results confirm that the biomarker selection strategy across three biological levels—nucleic acid, protein, and metabolism—of this invention has a significant advantage over any single-layer selection strategy. This advantage stems from a key insight in the design principle of this invention: the multi-pathway pharmacodynamic spectrum of astragaloside A requires the biomarker combination to simultaneously map the drug's regulatory network at multiple biological levels. A single-layer biomarker combination, regardless of its quantity, cannot comprehensively reflect the drug's overall regulatory capacity on the patient's body. Therefore, cross-layer combination is a necessary, not optional, condition for achieving high discriminative efficacy. This comparative example further verifies at the molecular mechanism level that the selected pentagonal combination precisely covers all major pharmacodynamic pathways of astragaloside A, without redundancy (loss of any one pathway would reduce efficacy) or deficiency (single-layer combination is insufficient).

[0083] Example 10: Structural composition and usage of the detection kit

[0084] This embodiment describes the specific structural composition and usage method of the detection kit of the present invention. For example... Figure 6 As shown, the kit is packaged independently according to the detection modules, and includes five functional modules: nucleic acid detection module, protein detection module, amino acid detection module, internal component and quality control module, and data processing module.

[0085] The nucleic acid detection module comprises three independent sub-components: the first sub-component is the stem-loop reverse transcription primer lyophilized powder shown in SEQ ID NO:2, 100 pmol per tube, reconstituted with 100 μL of nuclease-free water before use; the second sub-component is a mixed lyophilized powder of the upstream primer shown in SEQ ID NO:3 and the universal downstream primer shown in SEQ ID NO:4, 200 pmol of each primer per tube; the third sub-component is the 5′-FAM / 3′-MGB dual-labeled TaqMan probe lyophilized powder shown in SEQ ID NO:5, 100 pmol per tube, stored protected from light. All lyophilized powders should be stored at -20°C and have a shelf life of at least 18 months. The protein detection module comprises two sets of pre-coated 96-well microplates (targeting neuron-specific enolase and tumor necrosis factor α, respectively) and accompanying detection antibodies, enzyme conjugates, substrate solutions, and stop solutions, stored at 2°C to 8°C and have a shelf life of at least 12 months. The enzyme-linked immunosorbent assay (ELISA) reagent for neuron-specific enolase has a detection range of 1.56 ng / mL to 100 ng / mL and a sensitivity ≤0.5 ng / mL; the ELISA reagent for tumor necrosis factor-α has a detection range of 7.81 pg / mL to 500 pg / mL and a sensitivity ≤2 pg / mL; the intra-assay coefficient of variation for the aforementioned ELISA reagents is ≤8%, and the inter-assay coefficient of variation is ≤12%. The amino acid detection module includes... 13 C6-arginine and 13 Stable isotope internal standard solutions of C9-tyrosine (concentration 100 μmol / L, 100 μL per tube) should be stored at -80°C and have a shelf life of at least 24 months. The internal standard and quality control module includes the U6snRNA detection primer pairs shown in SEQ ID NO:6 and SEQ ID NO:7, as well as 7-point concentration gradient standard curves covering the expected detection range of various protein and amino acid biomarkers, and high, medium, and low-grade quality control samples. The data processing module includes a computer-readable storage medium (USB flash drive). The storage medium stores the parameter file (including Z-score standardized parameters and regression coefficients) and threshold Pc parameter of the multivariate discriminant model in encrypted form. The file includes a version checksum to prevent accidental modification of the parameter file.

[0086] The method of using the kit is as follows: (1) Use the nucleic acid detection module to complete the serum miR-155-5p detection according to the procedure described in Example 2; (2) Use the protein detection module to complete the serum neuron-specific enolase and tumor necrosis factor α detection according to the procedure described in Example 3; (3) Use the amino acid detection module to complete the serum arginine and tyrosine detection according to the procedure described in Example 4; (4) Use the internal reference quality control module to synchronously perform normalization correction and quality monitoring for each detection; (5) Input the detection results of each marker into the discriminant model parameter file in the data processing module, output the response index P, and complete the stratification of responders and non-responders based on Pc=0.65. The entire process from serum sample to stratification result can be completed within 8 hours in the laboratory department of a general tertiary hospital or a third-party independent medical laboratory. The cost of a single sample test is about 600~800 yuan, which has good clinical scalability.

[0087] Example 11: Stability and Precision Validation of the Detection Kit

[0088] This embodiment describes a systematic investigation of the performance retention of the test kit under accelerated stability and long-term stability conditions, as well as the verification results of the intra-batch and inter-batch precision of the detection system. The purpose of the stability study is to provide shelf-life support data for the production, storage, transportation, and sales of the test kit, ensuring that the test kit can still stably output repeatable test results within its calibrated shelf life.

[0089] Accelerated stability assays were designed according to ICH Q1A(R2) guidelines. The primer and probe lyophilized powders for the nucleic acid detection module were stored at two accelerated temperatures of 25°C and 40°C for 0, 1, 3, and 6 months, respectively. At each time point, the relative expression level of miR-155-5p in a group of 10 serum samples was detected according to the procedure described in Example 2, and paired t-tests were performed with the results from the initial 0-month time point. The results showed that after 6 months of acceleration at 25°C, the change in the Ct value of miR-155-5p was less than 0.5 cycles (P>0.05), and the amplification efficiency remained within the range of 95%~105%. After 6 months of acceleration at 40°C, the change in the Ct value was also less than 1.0 cycle, demonstrating that the primer and probe lyophilized powders have good thermal stability and support a shelf life of no less than 18 months when stored at -20°C. The pre-coated ELISA plates for the protein detection module were accelerated at 25°C and 30°C for 1, 2, and 3 months, respectively. The correlation coefficient R of the standard curve was... 2 The concentration was consistently maintained above 0.995, and the recovery rate of the quality control samples remained between 88% and 112%, supporting a shelf life of no less than 12 months when stored at 2°C to 8°C. The stable isotope internal standard of the amino acid detection module showed a change of less than 5% in mass spectrometry response intensity after 12 months of accelerated storage at -20°C, demonstrating sufficient evidence for the requirement that its shelf life be no less than 24 months under long-term storage conditions at -80°C.

[0090] Intra-assay precision validation employed a design that involved six parallel assays of the same batch of reagents on the same day for the same group of quality control samples. Results showed that the intra-assay relative standard deviation (RSD) for miR-155-5p detection was within the range of 3.2%–4.8% (all below the upper limit of 5%), the intra-assay RSD for neuron-specific enolase detection was within the range of 4.5%–6.7% (all below the upper limit of 8%), the intra-assay RSD for tumor necrosis factor α detection was within the range of 5.1%–7.2% (all below the upper limit of 8%), and the intra-assay RSDs for arginine and tyrosine detection were both within the range of 3.0%–4.5% (all below the upper limit of 5%). Inter-batch precision validation employed a design where three independent production batches of reagents were tested on the same set of quality control samples by different operators on different dates. Results showed that the inter-batch relative standard deviation (RSD) for miR-155-5p detection was 8.3% (below the upper limit of 12%), for neuron-specific enolase detection it was 9.6% (below the upper limit of 12%), for tumor necrosis factor-α detection it was 10.8% (below the upper limit of 12%), and for arginine and tyrosine detection it was 7.4% and 8.1% respectively (both below the upper limit of 10%). All precision data met the strict requirements of intra-batch coefficient of variation ≤8% and inter-batch coefficient of variation ≤12%.

[0091] Example 12: Clinical application case of the detection kit of the present invention

[0092] This embodiment selects 6 typical cases from the validation cohort to describe the complete application process and stratification results of the test kit of the present invention in a real clinical scenario, further illustrating the clinical practical value of the test kit of the present invention. All 6 patients met the inclusion criteria described in Example 1, had signed informed consent forms, and had completed 12 weeks of standardized astragaloside A treatment and post-treatment retesting of sural nerve conduction velocity.

[0093] Case 1: A 58-year-old female patient with a 12-year history of type 2 diabetes, a glycated hemoglobin level of 8.4%, a baseline sural nerve conduction velocity of 32.5 m / s, and a Michigan Neuropathy Screening Scale (MSS) score of 4. Pre-treatment serum test results showed a miR-155-5p relative expression fold of 1.92, neuron-specific enolase (NSA) of 14.8 ng / mL, tumor necrosis factor-α (TNF-α) of 0.5 pg / mL, arginine of 98.2 μmol / L, and tyrosine of 80.1 μmol / L. Substituting these values ​​into the discriminant model of this invention yielded a response index (P=0.81), exceeding the threshold (Pc=0.65), predicting her as a responder. After 12 weeks of treatment, the patient's sural nerve conduction velocity improved to 37.8 m / s (an improvement of 5.3 m / s, reaching the gold standard response threshold), and her MSS score decreased to 2, clinically confirming her as a responder, consistent with the stratification results of this invention.

[0094] Case 2: A 62-year-old male patient with a 15-year history of type 2 diabetes, a glycated hemoglobin level of 9.2%, a baseline sural nerve conduction velocity of 30.2 m / s, and a Michigan Neuropathy Screening Scale score of 5. Pre-treatment serum test results showed a miR-155-5p relative expression fold of 4.21, neuron-specific enolase 26.5 ng / mL, tumor necrosis factor α 62.4 pg / mL, arginine 65.1 μmol / L, and tyrosine 56.3 μmol / L. Substituting these values ​​into the discriminant model of this invention yielded a response index P=0.28, below the threshold Pc=0.65, predicting a non-responder. After 12 weeks of treatment, the patient's sural nerve conduction velocity only improved to 31.0 m / s (an improvement of 0.8 m / s, not reaching the gold standard), clinically confirming a non-responder, consistent with the stratification results of this invention. Based on this prediction, the attending physician will consider adjusting the patient's treatment plan to a combination of astragaloside A and aldose reductase inhibitor, or switching to other treatment strategies.

[0095] The detection and prediction processes in Cases 3 to 6 were similar, covering four typical scenarios: high P-value (P=0.78) for responders, moderate P-value (P=0.71) for responders, moderate P-value (P=0.41) for non-responders, and low P-value (P=0.22) for non-responders. Of the six patients, five (including Cases 1 and 2 above) showed stratification errors (the response index P=0.71 predicted a response, but clinical confirmation of no response, indicating a false positive). Further analysis of this false positive case showed early response signals (increased sural nerve conduction velocity of 1.8 m / s and 2.4 m / s in weeks 4 and 8 of treatment, respectively). However, due to an acute upper respiratory tract infection and acute increase in systemic inflammation in week 10, the improvement in nerve conduction velocity regressed to 2.5 m / s by week 12, failing to reach the gold standard. This suggests that the response prediction results of this invention may be biased in special circumstances with unexpected comorbidities, requiring combined clinical dynamic monitoring for decision-making. This case study also suggests that future research could consider incorporating acute inflammatory markers into the five-factor combination to improve predictive robustness in complex scenarios.

[0096] Through the aforementioned clinical application cases, the core value of the diagnostic kit of this invention in real-world clinical scenarios has been validated. From objective stratification before treatment initiation and dynamic monitoring during treatment, to assisting in strategy adjustment when treatment fails and supporting dosage decisions in long-term treatment, the diagnostic kit of this invention plays a supporting role in laboratory stratification throughout the entire treatment cycle of astragaloside IV in diabetic peripheral neuropathy, providing objective evidence for precision medication in clinical practice. The clinical translation path of this invention is clear, and it can be directly implemented in the clinical laboratories of ordinary tertiary hospitals or independent third-party medical laboratories. It is expected to significantly reduce the burden of ineffective medication for non-responding patients, shorten the trial-and-error cycle for clinicians, and improve the level of evidence-based support for astragaloside IV in clinical guidelines for diabetic peripheral neuropathy, providing a replicable paradigm reference for the application of precision medicine in the field of endocrinology and metabolism.

Claims

1. A combination of serum biomarkers for predicting the efficacy of astragaloside A in the treatment of diabetic peripheral neuropathy, characterized in that: The biomarker combination is composed of the following five biomarkers in combination: The first biomarker is the microRNA miR-155-5p with a mature sequence as shown in SEQ ID NO: 1, serving as a nucleic acid biomarker for characterizing the regulation status of the Schwann cell autophagy-apoptosis pathway; The second biomarker is neuron-specific enolase, serving as a protein biomarker for characterizing the cumulative degree of peripheral neuron damage; the third biomarker is tumor necrosis factor α, serving as a cytokine biomarker for characterizing the level of neuroinflammation; the fourth biomarker is arginine, serving as an amino acid biomarker for characterizing the status of the nitric oxide synthase substrate metabolism pathway; the fifth biomarker is tyrosine, serving as an amino acid biomarker for characterizing the status of the catecholamine precursor metabolism pathway; the five biomarkers are all obtained through in vitro detection from an in vitro peripheral serum sample collected from a type 2 diabetic peripheral neuropathy patient before treatment; The application method of the biomarker combination is: input the five measured values of the relative expression level of miR-155-5p normalized by U6 small nuclear RNA, the concentration of neuron-specific enolase, the concentration of tumor necrosis factor α, the concentration of arginine, and the concentration of tyrosine into a preset multivariate discrimination model, and the preset multivariate discrimination model outputs a response index P with a value range of 0 to 1, and the patients are stratified into responders and non-responders based on a preset threshold Pc.

2. The biomarker combination according to claim 1, characterized in that: The value range of the preset threshold Pc is 0.55 to 0.75; when the response index P ≥ Pc, it is determined as a responder; When the response index P < Pc, it is determined as a non-responder.

3. The biomarker combination according to claim 1, characterized in that: The preset multivariate discrimination model is selected from a Logistic regression model, a random forest model, or a support vector machine model; when the preset multivariate discrimination model is a Logistic regression model, the response index P is calculated by the formula P = 1 / (1 + exp(-Z)), where Z = β0 + β1×X1 + β2×X2 + β3×X3 + β4×X4 + β5×X5, X1 to X5 are the relative expression level of miR-155-5p, the concentration of neuron-specific enolase, the concentration of tumor necrosis factor α, the concentration of arginine, and the concentration of tyrosine in sequence, β0 is the intercept term, and β1 to β5 are the regression coefficients of each biomarker, and the regression coefficients are determined in advance by the maximum likelihood estimation method using the training set clinical data.

4. The biomarker combination according to claim 1, characterized in that: The population of diabetic peripheral neuropathy patients simultaneously meets the following conditions: diabetes duration ≥ 5 years, glycated hemoglobin is within the range of 6.5% to 10.0%, Michigan Neuropathy Screening Instrument score ≥ 2 points, sural nerve conduction velocity ≤ 40 m / s, and patients with peripheral neuropathy caused by other reasons and those with severe hepatic and renal insufficiency are excluded.

5. The biomarker combination according to claim 1, characterized in that: The ex vivo peripheral serum samples were prepared as follows: 8 mL of fasting venous blood was collected from the patient before treatment, incubated at 4°C for 30 to 60 minutes, centrifuged at 3000 r / min for 10 minutes, and the supernatant serum was collected. The samples were aliquoted and stored at -80°C, with each sample subjected to ≤2 freeze-thaw cycles. The detection method for miR-155-5p was stem-loop reverse transcription quantitative polymerase chain reaction. The detection methods for neuron-specific enolase and tumor necrosis factor α were enzyme-linked immunosorbent assay (ELISA). The detection methods for arginine and tyrosine were high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS / MS).

6. A test kit for predicting the efficacy of astragaloside A in the treatment of diabetic peripheral neuropathy, characterized in that: The kit comprises specific reagents for detecting each biomarker in the biomarker combination of claim 1, specifically including: (a) a nucleic acid detection module, comprising a stem-loop reverse transcription primer as shown in SEQ ID NO:2, an upstream primer as shown in SEQ ID NO:3, a universal downstream primer as shown in SEQ ID NO:4, and a TaqMan probe as shown in SEQ ID NO:5 with a FAM fluorescent group attached to the 5′ end and an MGB non-fluorescent quenching group attached to the 3′ end; (b) a protein detection module, comprising an enzyme-linked immunosorbent assay (ELISA) reagent for the neuron-specific enolase and an ELISA reagent for the tumor necrosis factor α, wherein each ELISA reagent comprises a coating antibody, a biotinylated detection antibody, a streptavidin-horseradish peroxidase conjugate, a 3,3′,5,5′-tetramethylbenzidine substrate solution, and a stop solution; (c) an amino acid detection module, comprising a high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS / MS) quantitative reagent for arginine and tyrosine, wherein the quantitative reagent comprises... 13 C mark or 2 (d) An internal standard containing H-labeled arginine stable isotope and tyrosine stable isotope; (e) an internal control module containing U6 small nuclear RNA detection primer pairs with sequences as shown in SEQ ID NO:6 and SEQ ID NO:7, and standard curves with 5 to 7 concentration gradients covering the expected detection range of each protein and amino acid marker, as well as high, medium, and low quality control products; (f) a data processing module containing a computer-readable storage medium storing a parameter file of a preset multivariate discriminant model and a threshold Pc parameter; the kit is used as follows: by sequentially performing the tests (a) to (c) on the patient's pre-treatment serum sample, normalization correction and quality monitoring are performed using the internal control products in the internal control module, and responder-non-responder stratification is performed by the data processing module.

7. The detection kit according to claim 6, characterized in that: In the nucleic acid detection module, the stem-loop reverse transcription primer is 50 nt in length, with a 6 nt 3′ end that is inversely complementary to the 3′ end of the mature form shown in SEQ ID NO: 1; the upstream primer is 22 nt in length, with a 5 nt non-template anchoring sequence at the 5′ end to adjust the annealing temperature to 60°C; the TaqMan probe is 16 nt in length, spanning the stem-loop-miRNA linker region of the reverse transcription product, and has a detection limit of no more than 10 copies / reaction; in the protein detection module, the enzyme-linked immunosorbent assay (ELISA) reagent for neuron-specific enolase has a detection range of 1.56 ng / mL to 100 ng / mL and a sensitivity ≤0.5 ng / mL; the ELISA reagent for tumor necrosis factor α has a detection range of 7.81 pg / mL to 500 pg / mL and a sensitivity ≤2 pg / mL; the intra-assay coefficient of variation for the aforementioned ELISA reagents is ≤8%, and the inter-assay coefficient of variation is ≤12%.

8. The detection kit according to claim 6, characterized in that: The chromatographic column of the amino acid detection module is an octadecylsilane-bonded silica column, with a column temperature of 40°C. Mobile phase A is an aqueous solution containing 0.1% (v / v) formic acid, and mobile phase B is an acetonitrile solution containing 0.1% (v / v) formic acid. The flow rate is 0.4 mL / min. The mass spectrometry uses electrospray ionization in positive ion mode. Multiple reaction monitoring (MRM) is used for arginine with a mass-to-charge ratio of 175→70, and for tyrosine with a mass-to-charge ratio of 182→136. The kit is individually packaged according to the detection module. The primer and probe lyophilized powder of the nucleic acid detection module has a shelf life of no less than 18 months when stored at -20°C. The pre-coated ELISA plate of the protein detection module has a shelf life of no less than 12 months when stored at 2°C to 8°C. The stable isotope internal standard of the amino acid detection module has a shelf life of no less than 24 months when stored at -80°C. The parameter file of the data processing module is stored in encrypted form on a computer-readable storage medium and includes a version check code to prevent the parameter file from being erroneously modified.

9. The use of the biomarker combination of claim 1 in the preparation of diagnostic reagents, kits, chips, or detection devices for in vitro screening of patients with diabetic peripheral neuropathy suitable for treatment with astragaloside A, characterized in that: The application includes the combined quantitative detection of the five biomarkers in ex vivo peripheral serum samples collected from the patients before treatment. The quantitative results are input into a preset multivariate discriminant model to obtain the response index P. The patients are then stratified into responders and non-responders based on a preset threshold Pc, thereby providing clinicians with objective laboratory stratification criteria before initiating astragaloside A treatment.

10. The application according to claim 9, characterized in that: The administration form of the astragaloside A is selected from oral preparations of monomeric astragaloside A, injectable preparations of monomeric astragaloside A, compound traditional Chinese medicine preparations containing astragaloside A, or traditional Chinese medicine preparations containing astragaloside A; the application includes at least one of the following: (a) performing a single baseline stratification test to screen the applicable population before the initiation of treatment in the patients; (b) performing dynamic monitoring at weeks 4, 8, and 12 after the initiation of treatment in the patients to help assess response stability; (c) assisting in decision-making on astragaloside A dose adjustment or combination therapy with other drugs during the long-term treatment of the patients; (d) serving as an inclusion stratification index in clinical studies evaluating the efficacy of astragaloside A in the treatment of diabetic peripheral neuropathy.

Images