Use of metabolic markers in the manufacture of a product for diagnosing asd and adhd

Abstract

This invention discloses the application of a metabolic biomarker in the preparation of products for diagnosing ASD and ADHD, which has the advantages of high sensitivity, high specificity, and high accuracy, and can reflect the severity of the disease and comorbidities. The biomarker is one of the following: linoleamide, valeric acid, androstenedione sulfate, arachidic acid, or uric acid.

Description

Technical Field

[0001] This invention relates to the field of biomedical technology, and more particularly to the application of a metabolic biomarker in the preparation of products for diagnosing ASD and ADHD. Background Technology

[0002] Autism Spectrum Disorder (ASD) and Attention Deficit Hyperactivity Disorder (ADHD) are among the most common neurodevelopmental disorders in childhood, affecting millions worldwide. In recent years, the prevalence of ASD and ADHD has risen rapidly. Both typically appear in early childhood and exhibit overlapping features, such as inattention, hyperactivity, and difficulties in social communication, which can complicate clinical diagnosis. Although ASD and ADHD are often comorbid and share similar behavioral phenotypes, they remain two distinct disorders, differing in core symptoms, neuropsychological characteristics, intervention strategies, and clinical outcomes. Extensive evidence suggests that early and accurate identification and timely intervention can significantly improve developmental trajectories and long-term prognosis. Therefore, reliable biomarkers not only aid in diagnosis but also guide individualized interventions targeting the specific symptom characteristics of each disorder.

[0003] Currently, the clinical diagnosis of ASD and ADHD still relies primarily on behavioral assessments and structured clinical interviews. While these methods are indispensable, they are inherently subjective and can vary depending on the assessor and cultural background, thus limiting the reliability of the diagnosis. This highlights the urgent need for objective, biologically based, and non-invasive biomarkers in research and clinical practice to support early differentiation between ASD and ADHD. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, the technical problem to be solved by the present invention is to provide an application of a metabolic biomarker in the preparation of products for diagnosing ASD and ADHD, which has the advantages of high sensitivity, strong specificity and high accuracy, and can reflect the severity of the disease and comorbidity.

[0005] The technical solution of the present invention is: the application of this metabolic biomarker in the preparation of products for diagnosing ASD and ADHD, wherein the biomarker is at least one of the following: linoleamide, valeric acid, androstenedione sulfate, arachidic acid, and uric acid.

[0006] The beneficial effects of this invention are: it has the advantages of high sensitivity, strong specificity and high accuracy, and can reflect the severity of the disease and comorbidity. Attached Figure Description

[0007] Figure 1The queue construction and marker machine learning process according to the present invention is illustrated.

[0008] Figure 2 The diagnostic potential of different sets of markers in the discovery queue is shown. Figure 2 A represents the diagnostic potential of the combination of linoleamide, androstenone sulfate, and arachidic acid (Panel-3). Figure 2 B represents the diagnostic potential of the combination of linoleamide, valeric acid, androstenedione sulfate, arachidic acid, and uric acid (Panel-5).

[0009] Figure 3 The correlation between three core biomarkers (Panel-3) and clinical characteristics is shown. The core biomarkers are linoleamide, arachidic acid, and androsterone sulfate. The clinical characteristics are mood disorders, infections, constipation, allergies, ABC (Autism Behavior Checklist) scores, the CARS (Childhood Autism Rating Scale), the ATEC scale (ATEC-C for verbal communication, ATEC-S for social skills, ATEC-R for sensory and cognitive abilities, ATEC-M for health-related physiological behaviors, and ATEC-T for total score), and the CLSQ (Clinical Language Function Questionnaire, CLSQ-C for verbal dimension and CLSQ-R for cognitive dimension). Purple items represent comorbidity types, and pink items represent the content of the evaluation scales. Bubble colors indicate the magnitude of the correlation, and asterisks indicate significance.

[0010] Figure 4 The chromatograms of the targeted quantitative substance standard and the extracted ion chromatogram are shown. Figure 4 A is the ion chromatogram of linoleamide. Figure 4 B is the ion chromatogram of arachidic acid. Figure 4 C represents the ion chromatogram of androstenone sulfate.

[0011] Figure 5 The diagnostic efficacy of panel-3 in the targeted validation cohort is shown. From left to right, these represent the diagnostic efficacy of panel-3 in autism-healthy children, autism-ADHD children, and ADHD-healthy children.

[0012] Figure 6 This study demonstrates the validation potential of serum biomarker sets for different groups. Figure 6 A represents the diagnostic efficacy of a set of biomarkers in the serum of autistic-healthy children. Figure 6 B represents the statistical significance of the model; Figure 6C represents the diagnostic efficacy of a set of biomarkers in the serum of children with autism-attention deficit disorder. Figure 6 D represents the statistical significance of the model; Figure 6 Diagnostic efficacy of E biomarker set in serum of children with attention deficit disorder and healthy children. Figure 6 F represents the statistical significance of the model.

[0013] Figure 7 The expression levels of three core biomarkers are shown in urine samples from the untargeted cohort, urine samples from the targeted validation cohort, and blood samples. "Untargeted metabolism in urine" represents the expression level of the core biomarkers in urine samples from the untargeted cohort; "Targeted metabolism in urine" represents the expression level of the core biomarkers in urine samples from the targeted cohort; and "Targeted metabolism in serum" represents the expression level of the core biomarkers in serum samples from the targeted cohort. The core biomarkers are linoleamide, arachidic acid, and androsterone sulfate. Detailed Implementation

[0014] Unless otherwise stated, the technical terms used in this application have the meanings commonly understood by those skilled in the art to which this invention pertains.

[0015] To more clearly and in detail illustrate the present invention, the technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore merely examples, and should not be used to limit the scope of protection of the present invention. Unless otherwise specified, the experimental methods used in the embodiments are conventional methods, and the experimental materials used are all available from commercial companies.

[0016] Metabolomics, as a rapidly developing discipline, offers a powerful opportunity for the discovery of biomarkers for neurodevelopmental disorders. Metabolites are dynamic indicators that reflect the combined effects of genetic, transcriptional, and environmental factors, thus holding particular promise in biomarker research. Importantly, metabolomics profiling has the potential to identify neurodevelopmental disorders before the onset of conventional behavioral symptoms, thereby paving the way for earlier intervention and improved treatment outcomes. Previous studies have shown that abnormalities in amino acid, lipid, and energy metabolism may be involved in the pathophysiological processes of ASD and ADHD. However, existing studies are often limited by small sample sizes, inconsistent methodologies, and a lack of independent cohort validation, making the translational value of metabolite-based biomarkers uncertain.

[0017] To overcome the aforementioned limitations, the inventors employed a multi-platform urinary metabolomics approach, combining liquid chromatography-mass spectrometry (LC-MS) and gas chromatography-mass spectrometry (GC-MS). Urine, as a non-invasive and easily collected biological sample, offers unique advantages in the pediatric population. By combining non-targeted and targeted metabolomics with machine learning analysis, the inventors systematically characterized the urinary metabolome of typical ASD, ADHD patients, and healthy controls (HC). The inventors identified a concise and highly discriminative set of biomarkers and validated their robustness in independent cohorts. The results not only provide a feasible diagnostic tool but also reveal the unique metabolic characteristics of ASD and ADHD, laying the foundation for future precise interventions targeting their potential biochemical pathways.

[0018] The application of this metabolic biomarker in the preparation of products for diagnosing ASD and ADHD, wherein the biomarker is at least one of the following: linoleamide, valeric acid, androstenone sulfate, arachidic acid, and uric acid.

[0019] The beneficial effects of this invention are: it has the advantages of high sensitivity, strong specificity and high accuracy, and can reflect the severity of the disease and comorbidity.

[0020] Preferably, the biomarker is a combination of linoleamide, androstenone sulfate, and arachidic acid.

[0021] Alternatively, the biomarker may be a combination of linoleamide, valeric acid, androstenedione sulfate, arachidic acid, and uric acid.

[0022] Furthermore, the analytical method for the metabolic biomarker includes the following steps: (1) Non-targeted liquid chromatography-mass spectrometry analysis: An ultra-high performance liquid chromatography system was used to connect the chromatographic column and coupled with a mass spectrometer for liquid chromatography-mass spectrometry analysis. The mobile phase A was an aqueous solution containing 25 mM ammonium acetate and 25 mM ammonia, and the mobile phase B was acetonitrile. The autosampler temperature was 4℃ and the injection volume was 2 μL. The mass spectra were obtained in information-dependent acquisition mode. The full scan parameters were: capillary temperature 350 ℃; primary mass spectrometry resolution 60000, secondary mass spectrometry resolution 7500; NCE mode collision energy 10 / 30 / 60; spray voltage 3.6 kV or –3.2 kV. The raw data were converted to mzXML format by ProteoWizard for peak detection, extraction, alignment and integration. Then, the metabolite annotation and identification were completed using the database. (2) Gas chromatography-mass spectrometry analysis: Gas chromatography-mass spectrometry was used, equipped with a capillary column of 30 m × 250 μm × 0.25 μm; injection volume was 1 μL, split ratio was 5:1; carrier gas was helium, septum purge flow rate was 3 mL / min, column flow rate was 1 mL / min; column temperature program: 50 ℃ for 1 min, ramped up to 310 ℃ at 8 ℃ / min, and held for 11.5 min; EI source ionization energy was –70 eV, and the m / z range of the full scan mode was 50–500; data processing included peak extraction, baseline correction, peak alignment, deconvolution, identification and integration; (3) Targeted metabolomics analysis and validation: Standards were prepared into 10 mmol / L stock solutions with each compound, mixed and serially diluted, and calibration curves were established for qualitative and quantitative analysis; analysis was performed by ultra-high performance liquid chromatography in tandem triple quadrupole mass spectrometry; separation was performed by chromatographic column according to the properties of the compounds, and the parent ion-daughter ion pairs and MRM parameters were optimized for each target metabolite; (4) Model building: In the initial stage, the gold standard dataset contains autism samples, ADHD samples and HC samples. According to the date of subject recruitment, the entire dataset is divided into a discovery cohort and a validation cohort. The discovery cohort is then split into a training set and a test set in a 2:1 ratio. The model performance is evaluated on both the discovery cohort and the validation cohort. A binary classification model is built using the random forest algorithm. In the subsequent stage, autism, ADHD and HC samples are added. The model outputs feature importance scores and ranks all features accordingly. Logistic regression, support vector machine and partial least squares discriminant analysis algorithms are used for comparison. (5) Evaluation of the model: The performance of the model is evaluated by the receiver operating characteristic (ROC) curve, and the area under the curve (AUC) is used as the main classification ability index. At the same time, the accuracy, sensitivity, specificity and precision are calculated. The training process uses 1000 self-sampling and random resampling each time to ensure the repeatability of the results and generate a 95% confidence interval.

[0023] Preferably, in step (1), 100 μL of urine or serum sample is taken, and 400 μL of extraction solution is added. The extraction solution is a mixture of acetonitrile:methanol = 1:1 and an isotope-labeled internal standard. After standing at -20℃ for 1 h to precipitate the protein, the sample is centrifuged at 12000 rpm for 15 min. The supernatant is transferred to a glass sample vial. An equal amount of supernatant from each sample is mixed to prepare a quality control (QC) sample for verifying the repeatability of liquid chromatography-mass spectrometry analysis.

[0024] Preferably, in step (2), 100 μL of urine or serum sample is mixed with 360 μL of pre-cooled methanol and 10 μL of internal standard (0.5 mg / mL adonitol), vortexed, and centrifuged at 12000 rpm for 15 min; 30 μL of each extract is mixed to form a QC sample; 180 μL of supernatant is transferred to a new EP tube and concentrated under vacuum until dry; 80 μL of 20 mg / mL pyridine solution is added to reconstitute the sample, and incubated at 80 °C for 30 min; then 100 μL of derivatization reagent BSTFA is added, and the sample is reacted at 80 °C for 1.5 h; after naturally cooling to room temperature, 5 μL of fatty acid methyl ester is added to the QC sample.

[0025] Preferably, in step (3), the targeted metabolite extraction process is as follows: Take 150 μL of sample, add 600 μL of pre-cooled extraction solution (acetonitrile-methanol 1:1, -40 ℃), vortex mix, sonicate in an ice-water bath for 15 min, stand at -40 ℃ for 1 h, and centrifuge at 12000 rpm for 15 min at 4 ℃; take 650 μL of supernatant, vacuum dry, and redissolve in 100 μL of reconstitution solution (acetonitrile-methanol-water 2:2:1), centrifuge again at 12000 rpm for 15 min, and take the supernatant for analysis.

[0026] Preferably, the application is for differential diagnosis of autism-healthy children, autism-attention deficit disorder children, and attention deficit disorder-healthy children.

[0027] Preferably, the application is used to predict the severity of clinical symptoms and comorbidities in autistic-healthy children, autistic-attention deficit disorder children, and attention deficit disorder-healthy children.

[0028] Preferably, the biological sample is the subject's serum or urine.

[0029] The embodiments of the present invention will be described in detail below.

[0030] To screen for potential diagnostic biomarkers for typical autism and ADHD, machine learning strategies were employed for feature selection and predictive model construction. For example... Figure 1 As shown in the flowchart, based on the recruitment date, the typical autism group (autism) and the healthy control group (HC) were divided into a discovery cohort (autism=47, HC=34) and a validation cohort (autism=63, HC=51). Due to the limited sample size of ADHD (ADHD=24), a separate model was not built. Instead, the biomarker combinations selected from the ASD-vs-HC screening were directly tested to determine whether ADHD-vs-HC and ADHD-vs-autism could be distinguished. For a detailed machine learning workflow, see [link to machine learning documentation]. Figure 1 .

[0031] The model was first constructed in the discovery cohort, and the Random Forest (RF) algorithm was used to screen for the optimal features that distinguish typical autism from HC, ultimately identifying five potential biomarkers: linoleamide, valeric acid, androsterone sulfate, arachidic acid, and uric acid (Table 1). Univariate ROC analysis was used to assess their diagnostic potential across groups, as shown in Table 2.

[0032] Furthermore, the diagnostic efficacy of a 3-marker combination (Panel-3: linoleamide + archidic acid + androsterone sulfate) and a 5-marker combination (Panel-5: linoleamide + archidic acid + androsterone sulfate + valeric acid + uric acid) was constructed and compared using machine learning. Figure 2 ).

[0033] In both the discovery and validation queues, Panel-3 demonstrated robustness: after 1000 RF simulations, the average AUCs were 0.88 and 0.923, respectively; Panel-5's corresponding AUCs were 0.861 and 0.937, but the improvement was not significant. For ADHD-vs-HC, the AUCs of Panel-3 and Panel-5 were 0.811 and 0.789, respectively; for typical autism-vs-ADHD, they were 0.787 and 0.766, respectively. To verify robustness, Panel-3 was also tested using SVM, PLS-DA, and LR algorithms (Table 3). The AUCs obtained by all algorithms were comparable to or even better than RF, especially in the ADHD-vs-HC and typical autism-vs-ADHD tasks, where Panel-3 outperformed RF, highlighting its consistency and robustness. In summary, Panel-3, containing only three metabolites, already possesses superior diagnostic performance, enabling accurate and efficient differentiation between typical autism and ADHD.

[0034] Further correlation analysis was conducted between biomarkers and clinical features (comorbidities, developmental scores, etc.) (Table 4). A bubble chart was also created for the three core Panel-3 biomarkers (Figure 3), with bubble size and color intensity reflecting correlation strength, and asterisks indicating statistical significance. The results suggest that some metabolites can not only serve as diagnostic biomarkers but may also reflect disease severity and comorbidities.

[0035] To verify the reliability and diagnostic potential of the above biomarkers, targeted metabolomics was used for replication in an independent cohort. This cohort consisted of 48 cases: 27 with typical autism, 9 with ADHD, and 10 with HC. Panel-3 (linoleamide, androsterone sulfate, arachidic acid) was selected for quantitative validation, showing the best performance in the target-free phase. The chromatographic peaks and extracted ion chromatograms (EIC) of the standards are shown in Figure 4.

[0036] The diagnostic efficacy was further evaluated using ROC curves. As shown in Figure 5, Panel-3 achieved an AUC of 0.84 in the independent cohort distinguishing between typical autism and HC, 0.80 for ADHD-vs-HC, and 0.70 for typical autism-vs-ADHD. After switching to different modeling algorithms, Panel-3's diagnostic performance remained stable (Table 4), further confirming its reliability.

[0037] In addition to urine, serum samples from the same cohort were also subjected to targeted quantification. As shown in Figure 6, the predictive power of blood was generally weaker than that of urine: the AUC for serum to distinguish autism-vs-HC was 0.886, which was comparable to the urine result; however, the AUCs for autism-vs-ADHD and ADHD-vs-HC decreased to 0.595 and 0.599, respectively, suggesting that serum markers still have moderate discriminative power for autism, but limited effect on ADHD.

[0038] Scatter plots were used to show the expression levels of three core biomarkers in urine from the non-targeted cohort and in urine and blood from the targeted validation cohort. Figure 7 As shown.

[0039] Table 1. Marker Identification Information

[0040] Table 2 Univariate ROC and Diagnostic Potential of Each Component

[0041] Table 3. Diagnostic potential of different biomarker sets (panel3, panel-5) in different patient groups in the cohort.

[0042] Table 4. Validation of the diagnostic efficacy of the Panel-3 biomarker set in the targeted cohort.

[0043] Table 5. Correlation between core biomarkers and clinical symptoms

[0044] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. Use of a metabolic marker in the manufacture of a product for the diagnosis of ASD and ADHD, characterized in that: The biomarker is at least one of the following: linoleamide, valeric acid, androstenedione sulfate, arachidic acid, and uric acid.

2. The application of the metabolic biomarker according to claim 1 in the preparation of products for diagnosing ASD and ADHD, characterized in that: The biomarker is a combination of linoleamide, androstenone sulfate, and arachidic acid.

3. The application of the metabolic biomarker according to claim 1 in the preparation of products for diagnosing ASD and ADHD, characterized in that: The biomarkers are a combination of linoleamide, valeric acid, androstenedione sulfate, arachidic acid, and uric acid.

4. The use of the metabolic biomarker according to any one of claims 1-3 in the preparation of products for diagnosing ASD and ADHD, characterized in that: The analytical method for the metabolic biomarkers includes the following steps: (1) Non-targeted liquid chromatography-mass spectrometry analysis: An ultra-high performance liquid chromatography system was used to connect the chromatographic column and coupled with a mass spectrometer for liquid chromatography-mass spectrometry analysis. The mobile phase A was an aqueous solution containing 25 mM ammonium acetate and 25 mM ammonia, and the mobile phase B was acetonitrile. The autosampler temperature was 4 ℃ and the injection volume was 2 μL. The mass spectra were obtained in information-dependent acquisition mode. The full scan parameters were: capillary temperature 350℃; primary mass spectrometry resolution 60000, secondary resolution 7500; NCE mode collision energy 10 / 30 / 60; spray voltage 3.6 kV or –3.2 kV. The raw data were converted to mzXML format by ProteoWizard for peak detection, extraction, alignment and integration. Then, the metabolite annotation and identification were completed using the database. (2) Gas chromatography-mass spectrometry analysis: Gas chromatography-mass spectrometry was used, equipped with a capillary column of 30 m × 250 μm × 0.25 μm; injection volume was 1 μL, split ratio was 5:1; carrier gas was helium, septum purge flow rate was 3 mL / min, column flow rate was 1 mL / min; column temperature program: 50 ℃ for 1 min, ramped up to 310 ℃ at 8 ℃ / min, and held for 11.5 min; EI source ionization energy was –70 eV, and the m / z range of the full scan mode was 50–500; data processing included peak extraction, baseline correction, peak alignment, deconvolution, identification and integration; (3) Targeted metabolomics analysis and validation: Standards were prepared into 10 mmol / L stock solutions with each compound, mixed and serially diluted, and calibration curves were established for qualitative and quantitative analysis; analysis was performed by ultra-high performance liquid chromatography in tandem triple quadrupole mass spectrometry; separation was performed by chromatographic column according to the properties of the compounds, and the parent ion-daughter ion pairs and MRM parameters were optimized for each target metabolite; (4) Model building: In the initial stage, the gold standard dataset contains autism samples, ADHD samples and HC samples. According to the date of subject recruitment, the entire dataset is divided into a discovery cohort and a validation cohort. The discovery cohort is then split into a training set and a test set in a 2:1 ratio. The model performance is evaluated on both the discovery cohort and the validation cohort. A binary classification model is built using the random forest algorithm. In the subsequent stage, autism, ADHD and HC samples are added. The model outputs feature importance scores and ranks all features accordingly. Logistic regression, support vector machine and partial least squares discriminant analysis algorithms are used for comparison. (5) Evaluation of the model: The performance of the model is evaluated by the receiver operating characteristic (ROC) curve, and the area under the curve (AUC) is used as the main classification ability index. At the same time, the accuracy, sensitivity, specificity and precision are calculated. The training process uses 1000 self-sampling and random resampling each time to ensure the repeatability of the results and generate a 95% confidence interval.

5. The application of the metabolic biomarker according to claim 4 in the preparation of products for diagnosing ASD and ADHD, characterized in that: In step (1), take 100 μL of urine or serum sample, add 400 μL of extraction solution, which is acetonitrile:methanol = 1:1 and contains an isotope-labeled internal standard mixture, let stand at -20℃ for 1 h to precipitate protein, and then centrifuge at 12000 rpm for 15 min; take the supernatant and transfer it to a glass sample vial; take equal amounts of supernatant from each sample and mix them to prepare a quality control QC sample for verifying the repeatability of liquid chromatography-mass spectrometry analysis.

6. The application of the metabolic biomarker according to claim 5 in the preparation of products for diagnosing ASD and ADHD, characterized in that: In step (2), 100 μL of urine or serum sample is mixed with 360 μL of pre-cooled methanol and 10 μL of internal standard (0.5 mg / mL adonitol). After vortexing, the mixture is centrifuged at 12000 rpm for 15 min. 30 μL of each extract is mixed to form a QC sample. 180 μL of supernatant is transferred to a new EP tube and concentrated under vacuum until dry. 80 μL of 20 mg / mL pyridine solution is added to redissolve the sample, and the mixture is incubated at 80 °C for 30 min. Then, 100 μL of derivatization reagent BSTFA is added, and the mixture is reacted at 80 °C for 1.5 h. After naturally cooling to room temperature, 5 μL of fatty acid methyl ester is added to the QC sample.

7. The application of the metabolic biomarker according to claim 6 in the preparation of products for diagnosing ASD and ADHD, characterized in that: In step (3), the targeted verification metabolite extraction process is as follows: Take 150 μL of sample, add 600 μL of pre-cooled extraction solution (acetonitrile-methanol 1:1, -40 ℃), vortex mix, sonicate in an ice-water bath for 15 min, stand at -40 ℃ for 1 h, and centrifuge at 12000 rpm for 15 min at 4 ℃; take 650 μL of supernatant, vacuum dry, and redissolve in 100 μL of reconstitution solution (acetonitrile-methanol-water 2:2:1), centrifuge again at 12000 rpm for 15 min, and take the supernatant for analysis.

8. The use of the metabolic biomarker according to any one of claims 1-3 in the preparation of products for diagnosing ASD and ADHD, characterized in that: The application is for the differential diagnosis of autism-healthy children, autism-attention deficit disorder children, and attention deficit disorder-healthy children.

9. The use of the metabolic biomarker according to any one of claims 1-3 in the preparation of products for diagnosing ASD and ADHD, characterized in that: The application is used to predict the severity of clinical symptoms and comorbidities in autistic-healthy children, autistic-attention deficit disorder children, and attention deficit disorder-healthy children.

10. Use of the metabolic markers according to claim 1 for the manufacture of a product for the diagnosis of ASD and ADHD, characterized in that: Biological samples are the subject's serum or urine.

Images