Integrative multi-omics profiling reveals the molecular subtypes and circulating biomarkers for pediatric mitochondrial disease

Integrative multi-omics profiling reveals the molecular subtypes and circulating biomarkers for pediatric mitochondrial disease | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Integrative multi-omics profiling reveals the molecular subtypes and circulating biomarkers for pediatric mitochondrial disease Jianxin Lyu, Xiaoting Lou, Zhehui Chen, Qiongya Zhao, Yuwei Zhou, and 12 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3389404/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Pediatric mitochondrial disease (PMD) refers the MD happened before 14 years old, which is a collection of rare, heterogenies and lethal syndromes. However, PMD still lacks molecular subtypes and a noninvasive diagnostic biomarker for precise medication and early diagnosis. By using multi-omics analyses for the discovery cohort, the molecular subtypes and robust biomarkers firstly discovered. The biomarkers further validated in an independent cohort. We found multiple energetic pathways altered in the PMD plasma (proteomics and metabolomics) and blood cells (transcriptomes), indicating the qualification of working pipelines. Some pathways were discovered without expectation may provide new insight into PMD pathogenesis. Molecular subtypes modeling revealed that PMD can be calcified into “AA-META”, “LIP-META” and “MIDDLE-META”, interestingly, the “AA-META” correlated with severe symptoms with a higher rate of neurologic and cardiac affected. Based on three machine learning algorithms, we discovered a panel of biomarkers with 13 molecules (1 gene, 2 proteins, and 10 metabolites), including classic (lactate, pyruvate) and novel biomarkers, showed more effective diagnosis rate of PMD (AUC=0.947) than reported ones. Overall, our work defined molecular subtypes of PMD and established a new panel of biomarkers for the precision diagnosis of PMD. Health sciences/Biomarkers/Diagnostic markers Health sciences/Diseases/Metabolic disorders Biological sciences/Molecular biology/Transcriptomics Biological sciences/Molecular biology/Proteomics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introduction Pediatric mitochondrial disease (PMD) refers to the MD happens before 14 years old, which is a collection of rare, heterogenies and lethal syndromes with significant morbidity and mortality( 1 , 2 ). The standard diagnostic procedure for PMD includes a complex combination of clinical assessments, blood metabolic profiles, brain imaging, tissue biopsies for histology or enzymology, and a genetic examination, which is a time-consuming and highly invasive process( 3 ). Also, it still lacks validated biomarkers for the clinical management and rigorous testing of novel therapy. The discovery of noninvasive biomarkers for MD has been the research hot spot, however, mostly studies used the samples of a specific subtype of MD (such MELAS syndrome( 4 ), Leigh syndrome( 5 ), TK2 deficient patients( 6 , 7 ), mtDNA deletions patients( 8 )) or teenager/adult( 9 , 10 ) MD, which it still lacks noninvasive biomarkers with high sensitivity and specificity for PMD regardless of the complexity of clinical phenotypes and genetic backgrounds. The subtypes of MD mostly follow the genetic background (according to the disease-causing genes) or the clinical manifestations, however, there are more than 400 genes across both mitochondrial genome and nuclear genome have been identified as the disease-causing gene of MD, and there are more than 30 clinical syndrome subtypes (such as Leigh syndrome, MELAS syndrome, LHON and so on) of PMD ( 11 ). The lack of molecular subtyping of PMD dramatically hindered the development of PMD precision medicine. Evidence indicates the multi-omics analysis may give new insight into the molecular subtypes of the disease with diverse pathophysiologic mechanisms( 12 – 14 ). However, it still lacks study for the molecular subtype analysis for MD regardless the clinical phenotypes and genetic backgrounds. The developments of technologies especially the omics analysis such as genomics, transcriptomes, metabolomics, proteomics, and so on, gives new opportunities in the disease studies. The proteomics and metabolomics studies of an adult m.3243A > G cohort have been performed. The study validated 1 protein and 19 metabolites which may enable monitoring of mitochondrial disease( 15 ). Considering the diagnosis for PMD is more difficult than adult, here the cohort focus on the pediatrics, the mean age of PMD cohort is 5.6 years old, which range from 0.6 to 15 years old. In this research, we detected the characteristic molecular changes with the PMD blood samples, interrogated the molecular heterogeneity of PMD based on the omics data with 3 subtypes, and discovered the PMD diagnosis panel with high diagnosis efficiency (AUC = 0.947). We hope the efforts give some proof for clinicians on PMD diagnosis and monitoring. Results Overview of PMD cohort for multi-omics analysis We aim to investigate: (1) uncover the potential biomarkers to improve the diagnosis for PMD; (2) define the molecular subtypes of PMD regardless the genetic and clinical backgrounds as shown in Figure 1 . To achieve the two goals, here we included two independent cohorts of PMD: discovery cohort and validation cohort. Individuals for all cohorts are pediatrics, the mean age is around 5 years old with the range from 0.6 to 15 years (details was available in Table 1 and Table S1 ). The discovery phase analyzed 144 whole blood and 144 plasma samples, which included 90 PMD patients, and 54 matched controls. The overall demographic and disease features of both cohorts are listed in Table 1 , and the detailed medical records are available in Table S1 . All individuals including patients and controls were in their baseline state of health at the timepoints of participation. All blood samples were taken at the first morning sample under the status of fasting. Comparing the clinical features across the patients and control, consistent with prior studies, PMD patients had a lower mean body mass index (BMI) than Ctrl cohorts. “Neurologic-affected” (73/90, 81%) was nominated for patients presented with epilepsy, dystonia and so on or Leigh syndrome or stroke-like features on MRI, “Cardiac-affected” (23/90, 26%) was nominated for patients have abnormalities in electrocardiogram or echocardiogram, developmental delay (86/90, 96%), gastrointestinal disturbances (36/90, 40%), psychotic problems (13/90, 14%), and hearing loss (3/90, 3%), are the frequently PMD complaints. The validation cohort was recruited independently, with 48 individuals for 24 PMD and 24 controls, the overall demographic and disease features are listed in Table 1, which is consistent with the discovery cohort., and the detailed medical records are available in Table S1 . 22 (92%) patients are “neurological-affected”, 5 (21%) patients are “cardiac-affected”, 21 (88%) patients presented with developmental delay, 19 (79%) patients suffered for gastrointestinal disturbance, 13 (14%) patients suffered psychiatric conditions, 3 (13%) patients suffered from hearing loss. Detection of whole blood transcriptomic alterations in PMD patients To investigate the pathogenesis and identify the potential biomarkers for PMD, whole blood transcriptome analysis was performed. Here, PAXgene blood RNA system was used for the whole blood transcriptome analysis. The resulting raw data were preprocessed to create normalized data for further analysis. The transcriptome data identified 15353 protein-coding transcripts with available p value (Supplementary Table 2) . The partial least squares discriminant analysis (PLS-DA) algorithms were performed to visualize the overall trend in gene expression levels among all samples (Figure 2A) . The preliminary examination revealed that whole blood gene expression levels differed between PMD patients and controls. In total of 3454 transcripts significantly changed (with adjust p value <0.05), in which 1226 upregulated and 2228 downregulated transcripts in MD patients compared to controls. Among 1226 upregulated-transcripts, 35 with the fold change (FC) more than 1.5 and among 2228 downregulated-transcripts, 130 with the FC less than 0.67 (Figure 2B-C) . The top ranked differentially expressed genes (DEGs) were analyzed accordingly (Figure 2D) . Disease specific DEGs were observed in the top alteration list, genes relevant to immune system such as killer-cell immunoglobulin-like receptors family members (KIR2DL3, KIR2DL4), metalloproteinases (MMP1, MMP8, ADAMTS2), hemoglobin related genes (HBB, HBA1). The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis performed based on the DEGs. The DEGs principally related to pathways of neurodegeneration, amyotrophic lateral sclerosis, Alzheimer disease, also the oxidative phosphorylation (Figure 2E) . All above indicating that whole blood transcripts could pinpoint key genes that are up- and down-regulated in the plasma of PMD patients may give the clues for the pathogenesis of MDs and the data generated is worth further analysis. Metabolic characterization of plasma sample in PMD patients Plasma metabolic profile was performed to further understand the pathogenesis and identify the potential biomarkers for PMD. The resulting raw data were preprocessed to create normalized data for further analysis. The plasma metabolome data identified 914 metabolites (992 metabolites in total, 78 metabolites with more than 50% missing values were eliminated) matched to known annotations (Supplementary Table 3) . PLS-DA algorithms used for the preliminary examination revealed that plasm metabolite levels differed between PMD patients and controls (Figure 3A) . In a total of 185 metabolites significantly changed (with adjust p value <0.05), in which 119 upregulated and 66 downregulated metabolites in MD patients compared to controls. Among 119 upregulated-metabolites, 58 with the FC more than 1.2 and among 66 downregulated- metabolites, 13 with the fold change less than 0.6 (Figure 3B-C) . Top differentially expressed metabolites (DEMs) mostly are disease specific (Figure 3D) . The FC of erythritol in PMD patients compared to controls was only 0.06. Erythritol is a newly recognized human metabolic product of glucose, synthesized through the pentose phosphate pathway (29). The elevated level of erythritol precedes the onset of central adiposity gain in humans. Most pediatric patients suffer for suboptimal growth and weight gain (7), the mean BMI in our PMD cohort (14.82) is also smaller than that in age-matched control cohort (16.78). Nucleotides such inosine and guanosine are decreased in the PMD. Top DEMs also includes lipids, such as 1-stearoyl-2-oleory-GPS (18:0/18:1), dexycarnitine, 3-hydroxybutyrylcarnitine, 3-hydroxyoleoylcarnitine, cis-3,4-methyleneheptanoate, 3-hydroxylaurate, and 3-hydroxymyristate (Figure 3D) . The Metabolite Set Enrichment Analysis (MSEA) pathway analysis performed based on the DEMs. The DEMs principally related to pathways of taurine and hypo-taurine and metabolism, glycerophospholipid metabolism, arginine biosynthesis, alanine, aspartate and glutamate metabolism, and TCA cycle (Figure 3E) . Proteomic characterization of plasma sample in PMD patients Plasma protein profile was performed to further detectthe pathogenesis and findthe usefulbiomarkers for PMD. The resulting raw data were extractedto create normalized data for further analysis. The plasma proteome data quantified 888 proteins (1706 proteins in total, 818 proteins with more than 50% missing values were eliminated) matched to known annotations (Supplementary Table 4) . Firstly, we conducted PLS-DA algorithm analysis (Figure 4A) . The preliminary examination revealed that plasma protein levels differed between PMD patients and controls. In total of 113 proteins significantly changed (with adjust p value <0.05), in which 11 upregulated and 102 downregulated proteins in PMD patients compared to controls. Among 11 upregulated-proteins, 5 with the FC more than 1.1 and among 102 downregulated-proteins, 69 with the fold change less than 0.8 (Figure 4B-C) . Top 10 upregulated differentially expressed proteins (DEPs) listed as: polymeric immunoglobulin receptor (PIGR), chromogranin A (CHGA), limbic system associated membrane protein (LSAMP), paraoxonase 1 (PON1), protein tyrosine phosphatase receptor type J (PTPRJ), collectin subfamily member 10 (COLEC10), reversion inducing cysteine rich protein with kazal motifs (RECK), prostaglandin D2 synthase (PTGDS), intelectin 1 (ITLN1), and sex hormone binding globulin (SHBG) (Figure 4D) . PTGDS is a member of lipocalin superfamily and plays dual roles in prostaglandin metabolism and lipid transport. The upregulation of PTGDS may related to the dysregulated of lipid metabolism in PMD patients. The top 10 downregulated DEPs listed as: myosin heavy chain 9 (MYH9), myosin light chain 12B (MYL12B), coronin 1C (CORO1C), C-reactive protein (CRP), myosin light chain 6 (MYL6), PDZ and LIM domain 1 (PDLIM1), actin alpha cardiac muscle 1 (ACTC1), clathrin heavy chain (CLTC), tropomyosin 4 (TPM4), and thymosin beta 4 X-linked (TMSB4X) (Figure 4D) . CRP is a biomarker produced by the liver in response to inflammation, the elevated of plasma CRP in PMD patients in consistent with the status of PMD. Figure 4E shows a visualization of known protein interactions. Protein data was analyzed by using STRING for protein network analysis. The groups of interacting proteins were isolated and individually subjected to gene ontology (GO) analysis using DAVID to determine biological function. Groups deemed significant (Benjamin-corrected p value≤0.05) were assigned to the biological function with the lowest p value. Multiple protein group functions especially the metabolic related groups demonstrated to be changed within PMD patients. GO analysis of plasma in PMD patients demonstrated that DEGs were mainly enriched in pathways of carbon metabolism, glycolysis/ gluconeogenesis, and biosynthesis of amino acids (Figure 4F) . Three molecular subtypes of PMD based on 3-layer omics profiles To comprehensively characterize the molecular features of the PMD, the whole blood transcripts, plasma metabolic profiles and proteomic profiles data from discovery cohort were used for the consensus clustering for subtypes. At first, all three layers of omics data including 15353 protein-coding transcripts, 914 metabolites, and 888 proteins were subjected to further analysis (Supplementary Table 2-4) . The PMD cohort was divided into three distinct clusters (subtype 1, subtype 2, and subtype 3), with an optimal consensus k-value of 3 (Supplementary Figure 2A-B) . The DEGs, DEMs and DEPs analysis were performed for each pair of molecular subtypes, and a total of 2 transcripts, 1 protein, and 152 metabolites were identified to be differentially expressed in the subtypes (Figure 5, Table S5) . Overall, the most different expression molecules among types are lipids. The subtype 1, is almost hypometabolic, especially downregulated in the lipids, and most upregulated in the amino acid, which was annotated for “AA-META”. The subtype 2 is almost upregulated in lipids and downregulated in the amino acid, which was annotated for “LIP-META”. As the subtype 3, most molecules are in the middle of two types, which is annotated for “MIDDLE-META”. Interestingly, when combined with the phenotype, we find the patients of “AA-META” seems more severe: according to Table 1 , the ratio of patients suffering for both cardiac and neurological problem in the discovery cohort is about 18/90 (20%), however, the ratio in “AA-META” type was 14/44 (32%), in “LIP-META” was 4/41 (10%), and in “MIDDLE-META” was 0/5 (0%). Consistently, ratio of patients only suffering for cardiac symptoms in the discovery cohort is about 23/90 (26%), however, the ratio in “AA-META” type was 16/44 (36%), in “LIP-META” was 7/41 (17%), and in “MIDDLE-META” was 0/5 (0%). The ratio of patients suffering only for neurologic problems in the discovery cohort is about 73/90 (81%), however, the ratio in “AA-META” type was 39/44 (89%), in “LIP-META” was 30/41 (73%), and in “MIDDLE-META” was 4/5 (80%) (Figure 5) . Identification of key DEGs, DEMs, and DEPs as biomarkers of PMD based on machine learning algorithms To confirm the significance of key DEGs, DEMs, and DEPs as biomarkers of PMD, we used three different machine learning algorithms to screen. Twenty key biomarkers were identified by RF algorithm, twenty-six key biomarkers were identified by SVM-RFE algorithm, and sixty-one key biomarkers were identified by Boruta algorithm. Thirteen overlapping biomarkers (TAF 15, HBB, ILK, lactate, pyruvate, deoxycarnitine, 2-hydroxyadipate, acetylcarnitine, N-acetylasparagine, 2,4-di-tert-butylphenol, gamma-glutamylserine, 2,3-dihydroxy-2-methybutyrate, and N, N, N-trimethyl-5-aminovalerate) were selected and give the name for “omics. markers” (Figure 6A) . The results of ROC curves indicated that the omics. makers which filtered out by machine learning algorithms have a favorable diagnostic value in the test cohort (the discovery cohort was randomly divided into training cohort and test cohort with the ratio of 0.8:0.2), with the AUC of 0.947 (Figure 6B) . Considering the lactate level, pyruvate level, and lactate/pyruvate (L/P) ratio were normally used for the clinical evaluation of MD, we combined the omics. markers with L/P ratio trying to improve the diagnostic rate, however, the AUC remains the same as 0.947 (Figure 6C) . As FGF-21 and GDF-15 were considered as new MD biomarkers, the AUC of omics. markers combined with FGF21 was about 0.942 (Figure 6D) . The AUC of omics. markers combined with GDF15 was about 0.938 (Figure 6E) . The AUC of omics. markers combined with FGF21 and GDF15 was about 0.940 (Figure 6F) . Finally, we combined omics. markers with FGF21, GDF15, and L/P ratio, got the AUC 0.937 (Figure 6G) . To confirm the results from the discovery cohort data in relation to differentiating a separate set of PMD from controls, we validated the 13 candidate biomarkers in Figure 6A via the independent validation cohort (including 24 PMD patients and 24 pediatric controls). The results confirm that among 13 candidate biomarkers, 10 biomarkers are significant changed in the validation cohort. The mRNA expression level of TAF15 is down regulated in validation cohort, which is consistent with the data in discovery cohort (Figure 7A) . The plasma ILK level also decreased in PMD (Figure 7B) . The classic PMD biomarkers lactate and pyruvate levels both elevated in validation cohort (Figure 7C-D) . 2,4-Di-tert-butylphenol (2,4-DTBP) is a common toxic secondary metabolite which is also elevated in the PMD patients (Figure 7E) . 2,3-dihydroxy-2-methylbutyrate has been considered a significant marker of ECHS1 deficiency(16, 17), here we found also increased in the expanded PMD cohort (Figure 7F) . 2-hydroxyadipate and deoxycarnitine both increased in the PMD in the validation cohort (Figure 7G-H) . Deoxycarnitine (also named for γ-butyrobetaine), a precursor of carnitine, is released in circulation to be taken up by skeletal muscle for storage. It has been reported that the increase of deoxycarnitine in circulation can be caused by the reduced uptake capacity in muscle tissue(18). Blood concentration of acetylcarnitine (C2) reflect intracellular levels and the regulation of acetyl CoA and free CoA via carnitine acetyl-CoA transferase, increased production of C2 represents the down regulated in glucose oxidation or upregulated in β-oxidation, which is consistent with results that the acetylcarnitine is upregulated in the PMD patients as the most PMD patients are deficit in OXPHOS function as hinder the glucose oxidation but enhance mitochondrial β-oxidation (Figure 7I) . Through our data, the N, N, N-trimethyl-5-aminovaleric acid (TMAVA) level is decreased in the pan-PMD patients in both discovery and validation cohort (Figure 7J) . The HBB, N-acetylasparagine, and gamma glutamylserine showed no significant changes in the validation cohort may be due to the patient number (Figure S3A-C) . Also, it may indicate that the biomarkers may be used as a panel rather than used separately. Discussion PMDs are both highly heterogeneity genetically and clinically heterogeneity, which is difficult for diagnosis and treatment. To address these difficulties in the PMD medications, we recruited 114 PMD patients (both clinical and genetically diverse) and 78 matched controls to conduct the whole blood transcriptome profiles, plasma metabolomics profiles, and proteomics profiles to find blood biomarkers help diagnosis and defined the PMDs molecular subtypes help the precise treatment. Previously studies mostly some certain type of PMD, e.g., m.3243A > G associated PMD ( 15 ). Considering the pediatric PMD mostly are syndromic presented with severe symptoms, diverse clinical manifestations, and poor prognosis. Studies based on a large diverse pediatric PMD cohort may be supposed to carry out. We first analyzed the data from the discovery cohort (including 90 pediatric PMD patients and 54 matched controls). Overall, 3454 DEGs, 185 DEMs, and 113 DEPs are first generated. The differential analysis based on transcriptome profiles, plasma metabolomics profiles, and proteomics profiles indicates the blood molecules have the potential to be the biomarkers for the PMDs, as most differential expressed blood molecules are metabolism even correlate with cell bioenergetics. To identify the key differentially expressed molecules which may have the potential to be the diagnosis biomarkers for PMDs, here we used three different machine learning algorithms (including RF algorithm, SVM-RFE algorithm, and Boruta algorithm) to screen key DEGs, DEMs, and DEPs. Thirteen overlapping biomarkers (TAF 15, HBB, ILK, lactate, pyruvate, deoxycarnitine, 2-hydroxyadipate, acetylcarnitine, N-acetylasparagine, 2,4-di-tert-butylphenol, gamma-glutamylserine, 2,3-dihydroxy-2-methybutyrate, and N, N, N-trimethyl-5-aminovalerate) were selected, with the AUC of 0.947, which the efficiency is much better than the traditional biomarkers. Furthermore, the thirteen biomarkers were applied in the independent validation cohort. One transcript biomarker TAF15 still significantly downregulated in PMD, one protein biomarker ILK is also downregulated in PMD in the validation cohort. Classic biomarkers including lactate, pyruvate, and 2,3-dihydroxy-2-methybutyrate all upregulated in PMD patients in the validation cohort. The omics-based biomarkers panel has higher diagnostic efficiency than conventional biomarkers in PMD cohort. The traditional PMD biomarkers can be mostly correlated to the genetic background, the biomarkers applied in diverse PMD mostly lack sensitivity and specificity such as lactate and pyruvate. Our results also indicate the omics biomarkers are more suitable for use as a panel. Convention blood biomarkers for PMD are lack of sensitivity( 19 – 21 ), which also had a bad performance in our PMD cohort: the area under the curve (AUC) of lactate is 0.697, pyruvate is 0.625, lactate/pyruvate ratio is 0.421, combined lactate and pyruvate is 0.740, and combined lactate, pyruvate and lactate/pyruvate ratio is up to 0.752 (Figure S1 A-E). Fibroblast growth factor 21(FGF21) ( 9 , 19 , 22 ) and growth differentiation factor 15 (GDF15) emerged as superior indicators of PMD ( 23 – 27 ), especially the GDF15. In our PMD cohort, the protein biomarkers behave better than conventions: the AUC for FGF21 is 0.675, GDF15 is 0.894, and the combination of FGF21 and GDF15 is 0.889 (Figure S1 F-H) . When combined the conventional metabolic biomarkers with protein biomarkers FGF21 and GDF15, the diagnosis efficiency dramatically increased: when combined the pyruvate with FGF21 and GDF15, the AUC raised to 0.925, which indicating that the combination of omics biomarkers may help the precision diagnosis of PMD (Figure S1 I-L) . Insight into organ system. Despite the omics dramatically improves the diagnosis efficiency for PMD, we still find it difficult in assess the clinical manifestations: considering PMD patients frequently suffers for neurologic problem, the PMD patients suffered from cardiac symptoms are at risk for sudden death or major cardiac event. Here, we analyze the PMD patients under different manifestations. We reviewed the medical record and divided the discovery cohort into three groups (detailed criteria see Table 1 ): “ neuro ” for the individuals suffers from “Neurological-affected” but not “cardiac-affected”, “ both ” for the individuals both “Neurological-affected” and “cardiac-affected”, and “ other ” for the rest. We find there are no significant changes in the transcriptomic profile, 105 proteins changed between 3 groups, and 22 metabolites changed between 3 groups (Figure S4A-B, Table S6) . Interestingly, the differential expressed molecules between diverse PMD, and control are mostly metabolites, the differential expressed molecules between different clinical manifestations mainly proteins ( Fig. 8 A-B ) . All DEM in between three groups are regulated in the “ both ” group, most are key metabolites from bioenergetic pathways, e.g., lactate and branched-chain amino acids (BCAAs). Especially, a group of BCAAs’ 1-carboxyethyl conjugates can be found here. To identify the omics biomarkers in if works in for the identification of different clinical manifestation, we further evaluate the levels of 13 biomarkers in 3 groups. However, it needs to expand the cohort and further studies. Among thirteen biomarkers, seven biomarkers showed no significant changes ( Fig. 8 C-H, Figure S4C-I) . Comparing with former studies. Vamsi K. Mootha Team( 15 ) conducted a study recruited a big m.3243A > G cohort, conducting the proteomics and metabolomics for the biomarker discovery, which validated classic biomarkers (GDF15, lactate, pyruvate, alanine) and 3 biochemical families (N-lactoyl-amino acids, hydroxy-fatty acids, hydroxyacyl carnitines). Here, in our cohorts we also validated the classic biomarkers, the hydroxyacyl carnitines (including C2:0, C3:0, and C4:0) all significantly upregulated in PMD which is consistent with the former study, however, only the acetylcarnitine and deoxycarnitine (metabolic precursor of carnitine) is recruited in the biomarkers panel after machine learning (Supplementary Table 3) , same situation with the hydroxy-fatty acids. The possible main reasons may list as: 1. The different cohorts, as the metabolites and proteins dramatically changed with age, sex, and BMI, the former cohort is mostly adults, with the mean age of about 40 years old, and the mean age from our cohort is only 5 years old. 2. The former cohort is focus on the m.3243A > G individuals, the cohort here is more diverse. In this study, we hypothesized that pediatric PMD induces characteristic molecular changes that can be detected in the invasive samples including blood. To test the hypothesis, two independents pediatric PMD cohorts including discovery (90 patients and 54 controls) and validation (24 patients and 24 controls) recruited. Firstly, based on the discovery cohort, whole blood transcriptomic profiles, plasma metabolomics profiles, and plasma proteomics profiles were performed. Secondly, the integrative analysis of three-layer omics was performed for molecular subtype taxonomy of pediatric PMD despite of genetic background but based on the differentially expressed plasma metabolites and proteins. Thirdly, all differentially expressed molecules used for the biomarker discovery, an omics biomarker set (two transcripts, one protein, and ten metabolites) was produced, which was validated based on the validation cohort. Finally, in-depth analysis based on neurologic and cardiologic impairment was performed. The observation may represent a step towards PMD diagnosis and therapy. Methods Please see the supplemental material for detailed methods. Study approval All samples and clinical information were collected under the approval of the Ethics Committee of Peking University First Hospital (2017-217) and the Second Affiliated Hospital of Wenzhou Medical University (2021-K-54-02). Declarations Conflict of Interest: The authors have declared that no conflict of interest exists. Authors’ Contributions: J.L, Y.Y, H.F, and X.L were responsible for overarching research goals and aims. J.L, Y.Y, H.F, and X.L conceived and designed the experiments. Z.C, Y.Z, P.L, C.Z, and X.M collected and stored samples from patients and were involved in the translation of results to clinical implications. X.L and Q.Z performed data interpretation of whole blood transcription data, untargeted-metabolomics data, and untargeted plasma proteome data. T.G was responsible for the experimental design of untargeted plasma proteome analysis. Y.Z was responsible for ELISA experiments and analyzed the data. Y.Y, X.L, Z.C, Y.Z, X.Z, Y.X, Y.W, L.Z, Q.D, and X.Y collected and reviewed the medical records from patients. X.L, Z.C, and Y.Z drafted the manuscripts. J.L, Y.Y, and H.F proofread the manuscript. All authors read and approved the manuscript. Authors’ contributions J.L, Y.Y, H.F, and X.L was responsible for overarching research goals and aims. J.L, Y.Y, H.F, and X.L conceived and designed the experiments. Z.C, Y.Z, P.L, C.Z, and X.M collected and stored samples from patients and were involved in the translation of results to clinical implications. X.L and Q.Z performed data interpretation of whole blood transcription data, untargeted-metabolomics data, and untargeted plasma proteome data. T.G was responsible for the experimental design of untargeted plasma proteome analysis. Y.Z was responsible for ELISA experiments and analyzed the data. Y.Y, X.L, Z.C, Y.Z, Y.X, Y.W, L.Z, Q.D, X.Y, and X.Z collected and reviewed the medical records from patients. X.L, Z.C, and Y.Z drafted the manuscripts. J.L, Y.Y, and H.F proofread the manuscript. All authors read and approved the manuscript. Acknowledgments The authors would like to thank all collaborators and patient participants for their contributions to this study. References Schapira AH. Mitochondrial diseases. Lancet. 2012;379(9828):1825–34. Thompson K, Collier JJ, Glasgow RIC, Robertson FM, Pyle A, Blakely EL, et al. Recent advances in understanding the molecular genetic basis of mitochondrial disease. J Inherit Metab Dis. 2020;43(1):36–50. Parikh S, Goldstein A, Karaa A, Koenig MK, Anselm I, Brunel-Guitton C, et al. Patient care standards for primary mitochondrial disease: a consensus statement from the Mitochondrial Medicine Society. Genet Med. 2017;19(12). Esterhuizen K, Lindeque JZ, Mason S, van der Westhuizen FH, Suomalainen A, Hakonen AH, et al. A urinary biosignature for mitochondrial myopathy, encephalopathy, lactic acidosis and stroke like episodes (MELAS). Mitochondrion. 2019;45:38–45. Thompson Legault J, Strittmatter L, Tardif J, Sharma R, Tremblay-Vaillancourt V, Aubut C, et al. A Metabolic Signature of Mitochondrial Dysfunction Revealed through a Monogenic Form of Leigh Syndrome. Cell Rep. 2015;13(5):981–9. Dominguez-Gonzalez C, Badosa C, Madruga-Garrido M, Martí I, Paradas C, Ortez C, et al. Growth Differentiation Factor 15 is a potential biomarker of therapeutic response for TK2 deficient myopathy. Sci Rep. 2020;10(1):10111. Kalko SG, Paco S, Jou C, Rodríguez MA, Meznaric M, Rogac M, et al. Transcriptomic profiling of TK2 deficient human skeletal muscle suggests a role for the p53 signalling pathway and identifies growth and differentiation factor-15 as a potential novel biomarker for mitochondrial myopathies. BMC Genomics. 2014;15:91. Forsström S, Jackson CB, Carroll CJ, Kuronen M, Pirinen E, Pradhan S, et al. Fibroblast Growth Factor 21 Drives Dynamics of Local and Systemic Stress Responses in Mitochondrial Myopathy with mtDNA Deletions. Cell Metab. 2019;30(6):1040-54.e7. Davis RL, Liang C, Edema-Hildebrand F, Riley C, Needham M, and Sue CM. Fibroblast growth factor 21 is a sensitive biomarker of mitochondrial disease. Neurology. 2013;81(21):1819–26. Davis RL, Liang C, and Sue CM. A comparison of current serum biomarkers as diagnostic indicators of mitochondrial diseases. Neurology. 2016;86(21):2010–5. Rahman S. Mitochondrial disease in children. J Intern Med. 2020;287(6):609–33. Neff RA, Wang M, Vatansever S, Guo L, Ming C, Wang Q, et al. Molecular subtyping of Alzheimer's disease using RNA sequencing data reveals novel mechanisms and targets. Sci Adv. 2021;7(2). Greenwood AK, Montgomery KS, Kauer N, Woo KH, Leanza ZJ, Poehlman WL, et al. The AD Knowledge Portal: A Repository for Multi-Omic Data on Alzheimer's Disease and Aging. Curr Protoc Hum Genet. 2020;108(1):e105. Iturria-Medina Y, Adewale Q, Khan AF, Ducharme S, Rosa-Neto P, O'Donnell K, et al. Unified epigenomic, transcriptomic, proteomic, and metabolomic taxonomy of Alzheimer's disease progression and heterogeneity. Sci Adv. 2022;8(46):eabo6764. Sharma R, Reinstadler B, Engelstad K, Skinner OS, Stackowitz E, Haller RG, et al. Circulating markers of NADH-reductive stress correlate with mitochondrial disease severity. J Clin Invest. 2021;131(2). Haack TB, Jackson CB, Murayama K, Kremer LS, Schaller A, Kotzaeridou U, et al. Deficiency of ECHS1 causes mitochondrial encephalopathy with cardiac involvement. Ann Clin Transl Neurol. 2015;2(5):492–509. Peters H, Buck N, Wanders R, Ruiter J, Waterham H, Koster J, et al. ECHS1 mutations in Leigh disease: a new inborn error of metabolism affecting valine metabolism. Brain. 2014;137(Pt 11):2903–8. Saoi M, Li A, McGlory C, Stokes T, von Allmen MT, Phillips SM, et al. Metabolic Perturbations from Step Reduction in Older Persons at Risk for Sarcopenia: Plasma Biomarkers of Abrupt Changes in Physical Activity. Metabolites. 2019;9(7). Suomalainen A, Elo JM, Pietiläinen KH, Hakonen AH, Sevastianova K, Korpela M, et al. FGF-21 as a biomarker for muscle-manifesting mitochondrial respiratory chain deficiencies: a diagnostic study. Lancet Neurol. 2011;10(9):806–18. Koenig MK. Presentation and diagnosis of mitochondrial disorders in children. Pediatr Neurol. 2008;38(5):305–13. Munnich A, and Rustin P. Clinical spectrum and diagnosis of mitochondrial disorders. Am J Med Genet. 2001;106(1):4–17. Crooks DR, Natarajan TG, Jeong SY, Chen C, Park SY, Huang H, et al. Elevated FGF21 secretion, PGC-1α and ketogenic enzyme expression are hallmarks of iron-sulfur cluster depletion in human skeletal muscle. Hum Mol Genet. 2014;23(1):24–39. Riley LG, Nafisinia M, Menezes MJ, Nambiar R, Williams A, Barnes EH, et al. FGF21 outperforms GDF15 as a diagnostic biomarker of mitochondrial disease in children. Mol Genet Metab. 2022;135(1):63–71. Lehtonen JM, Auranen M, Darin N, Sofou K, Bindoff L, Hikmat O, et al. Diagnostic value of serum biomarkers FGF21 and GDF15 compared to muscle sample in mitochondrial disease. J Inherit Metab Dis. 2021;44(2):469–80. Salehi MH, Kamalidehghan B, Houshmand M, Aryani O, Sadeghizadeh M, and Mossalaeie MM. Association of fibroblast growth factor (FGF-21) as a biomarker with primary mitochondrial disorders, but not with secondary mitochondrial disorders (Friedreich Ataxia). Mol Biol Rep. 2013;40(11):6495–9. Montero R, Yubero D, Villarroya J, Henares D, Jou C, Rodríguez MA, et al. GDF-15 Is Elevated in Children with Mitochondrial Diseases and Is Induced by Mitochondrial Dysfunction. PLoS One. 2016;11(2):e0148709. Yatsuga S, Fujita Y, Ishii A, Fukumoto Y, Arahata H, Kakuma T, et al. Growth differentiation factor 15 as a useful biomarker for mitochondrial disorders. Ann Neurol. 2015;78(5):814–23. Table Table 1. Demographics and disease features of the cohorts Discovery cohort Validation cohort PMD Ctrl PMD Ctrl Units （ Range ） Number 90 54 24 24 / Sex, male/female 49/41 30/36 13/11 16/8 / Age, mean (Range, SD) 5.6(0.6-15, 3.22) 6.8(0.6-15, 3.78) 6.09(1.5-16, 4.26) 5.69(2.6-12, 3.32) years BMI, mean (Range, SD) 14.82(10.94-22.71, 2.26) 16.78(12.88-23.46, 2.55) 14.75(10.61-20.55, 2.36) 16.56(12.81-23.73, 2.37) Kg/m 2 Disease features Neurological affected 73(81%) / 22(92%) / Number (%) Cardiac affected 23(26%) / 5(21%) / Growth failure 86(96%) / 21(88%) / Gastrointestinal disturbance 36(40%) / 19(79%) / psychotic problems 13(14%) / 10(42%) / Hearing loss 3(3%) / 3(13%) / Note: The positive inclusion criteria of neurological manifestations were the presence of epilepsy, dystonia and so on or Leigh syndrome or stroke-like features on MRI. The positive inclusion for heart disease is abnormalities in electrocardiogram or echocardiogram. Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryTable1.xlsx Supplementary Table 1 SupplementaryTable2.xlsx Supplementary Table 2 SupplementaryTable3.xlsx Supplementary Table 3 SupplementaryTable4.xlsx Supplementary Table 4 SupplementaryTable5.xlsx Supplementary Table 5 SupplementaryTable6.xlsx Supplementary Table 6 SupplementaryAppendix.docx Supplementary Appendix Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3389404","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":240841438,"identity":"c8264874-e7d9-4e03-8171-f7fad15da991","order_by":0,"name":"Jianxin Lyu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuUlEQVRIiWNgGAWjYBACxgbmBsYPBjVybOztB4jVAtQkUXDMmI/nTALx9jDwfGBOnCfhYECcBuYZiW0SEgZs6W0SDAkMPyq2EWFHz8E2iQIDmdw26cYDjD1nbhOhpb0RbEtum8yBBGbGNmK0NDO2SfAYMKezSSQYEKkFZAtQSwIJWnoONltLGBwzbAMG8kGi/GI4I/ngzQ9/auTl29sPPvhRQYyWBiTOAcLqgUCeKFWjYBSMglEwsgEA0HE5NmxRN/sAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0003-2343-1666","institution":"Wenzhou Medical University","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Jianxin","middleName":"","lastName":"Lyu","suffix":""},{"id":240841439,"identity":"bbd7d86b-5dd0-4fe4-ad9c-ae0586de4c4b","order_by":1,"name":"Xiaoting Lou","email":"","orcid":"","institution":"Zhejiang Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xiaoting","middleName":"","lastName":"Lou","suffix":""},{"id":240841440,"identity":"3e85e954-0312-406b-8177-931d3d7f716b","order_by":2,"name":"Zhehui Chen","email":"","orcid":"https://orcid.org/0000-0002-6139-5347","institution":"","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Zhehui","middleName":"","lastName":"Chen","suffix":""},{"id":240841441,"identity":"12517133-6543-4644-9ea2-4931759c30da","order_by":3,"name":"Qiongya Zhao","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Qiongya","middleName":"","lastName":"Zhao","suffix":""},{"id":240841442,"identity":"663bfb61-195d-4484-a076-6143bf93c23c","order_by":4,"name":"Yuwei Zhou","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yuwei","middleName":"","lastName":"Zhou","suffix":""},{"id":240841443,"identity":"3639c218-88f8-4143-84ae-677828002af7","order_by":5,"name":"Xiaofei Zeng","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xiaofei","middleName":"","lastName":"Zeng","suffix":""},{"id":240841444,"identity":"43b3f738-9ba1-4775-beea-ecd0a71d9d9e","order_by":6,"name":"Xue Ma","email":"","orcid":"","institution":"Peking University First Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xue","middleName":"","lastName":"Ma","suffix":""},{"id":240841445,"identity":"9018f90b-988b-4920-9ea4-c4d6373924ce","order_by":7,"name":"Peng Luo","email":"","orcid":"","institution":"The 2nd Affiliated Hospital and Yuying Children's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Peng","middleName":"","lastName":"Luo","suffix":""},{"id":240841446,"identity":"0a40f869-9354-4aa3-a8cd-e6fd27fe3d09","order_by":8,"name":"Chunxia Zhang","email":"","orcid":"","institution":"The 2nd Affiliated Hospital and Yuying Children's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Chunxia","middleName":"","lastName":"Zhang","suffix":""},{"id":240841447,"identity":"c5d502af-25cc-4fd8-b9c8-765312154705","order_by":9,"name":"Luyi Zhang","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Luyi","middleName":"","lastName":"Zhang","suffix":""},{"id":240841448,"identity":"3e11053a-39f1-4849-a729-189711ec7214","order_by":10,"name":"Yaojun Xie","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yaojun","middleName":"","lastName":"Xie","suffix":""},{"id":240841449,"identity":"769ae08e-6ddd-408f-906f-c5071859fc92","order_by":11,"name":"Ya Wang","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Ya","middleName":"","lastName":"Wang","suffix":""},{"id":240841450,"identity":"59461331-ba2d-4903-9070-b12b93f70624","order_by":12,"name":"Qiyu Dong","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Qiyu","middleName":"","lastName":"Dong","suffix":""},{"id":240841451,"identity":"ead3af01-10f5-4ac4-8026-9aafe7872053","order_by":13,"name":"Xiaojie Yin","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xiaojie","middleName":"","lastName":"Yin","suffix":""},{"id":240841452,"identity":"6e6e9608-67e0-4e11-b450-892df2c58d84","order_by":14,"name":"Tiannan Guo","email":"","orcid":"https://orcid.org/0000-0003-3869-7651","institution":"Westlake University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Tiannan","middleName":"","lastName":"Guo","suffix":""},{"id":240841453,"identity":"043da2f2-33f0-4527-ba2d-5b94f3bcc6d7","order_by":15,"name":"Hezhi Fang","email":"","orcid":"","institution":"Wenzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Hezhi","middleName":"","lastName":"Fang","suffix":""},{"id":240841454,"identity":"8e3a73d6-3f3e-4db1-ae03-40a82b4fa864","order_by":16,"name":"Yanling Yang","email":"","orcid":"","institution":"Department of Pediatrics, Peking University First Hospital, Beijing","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yanling","middleName":"","lastName":"Yang","suffix":""}],"badges":[],"createdAt":"2023-09-26 15:35:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3389404/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3389404/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":44856199,"identity":"e7b1c32b-6262-4380-b3de-0adcce1e9028","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":517750,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOverview of PMD cohort for omics analysis.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStudy design.\u003c/p\u003e","description":"","filename":"Binder11.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/488891c5c4f80710c9970979.png"},{"id":44856203,"identity":"fa911baf-3e23-4b9c-ad09-808bbf3f2585","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":245825,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of whole blood transcripts between PMD (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=90) and Ctrl (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=54) by using PAXgene system. (A) \u003c/strong\u003eThe partial least squares discriminant analysis (PLS-DA) algorithms of whole blood transcriptomic profile. \u003cstrong\u003e(B)\u003c/strong\u003e Binary comparison of transcriptome data by volcano plot. The Log2Fold Change (Log2FC) between PMD and Ctrl of relative abundance for transcripts illustrates positive values (pink) indicating upregulation in PMD and negative values (blue) indicating downregulation in PMD. Gray dots represent non-significantly expressed genes. The dashed line indicates the significant threshold. \u003cstrong\u003e(C) \u003c/strong\u003eThe numbers of the differential expression genes (DEGs) between PMD and Ctrl.\u003cstrong\u003e (D) \u003c/strong\u003eThe top 10 upregulated and downregulated DEGs between PMD and Ctrl. \u003cstrong\u003e(E) \u003c/strong\u003eThe Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis based on DEGs.\u003c/p\u003e","description":"","filename":"Binder12.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/5e048ff013f5cf1f2a7dc764.png"},{"id":44858028,"identity":"a648b5ff-8428-4947-8781-72e3e504dc0f","added_by":"auto","created_at":"2023-10-18 14:29:24","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":436335,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of plasma metabolites between PMD (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=90) and Ctrl (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=54) by using untargeted metabolomics platform. (A)\u003c/strong\u003e PLS-DA analysis of plasma metabolites. \u003cstrong\u003e(B)\u003c/strong\u003e Binary comparison of plasma metabolome data by volcano plot. The Log2FC between PMD and Ctrl of relative abundance for transcripts illustrates positive values (pink) indicating upregulation in PMD and negative values (blue) indicating downregulation in PMD. Gray dots represent non-significantly expressed genes. The dashed line indicates the significant threshold.\u003cstrong\u003e (C)\u003c/strong\u003e The numbers of the differential expression metabolites (DEMs) between PMD and Ctrl. \u003cstrong\u003e(D) \u003c/strong\u003eThe top 10 upregulated and downregulated DEMs between PMD and Ctrl. \u003cstrong\u003eE.\u003c/strong\u003e The metabolite set enrichment analysis (MSEA) is based on DEMs.\u003c/p\u003e","description":"","filename":"Binder13.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/fb1dbc2afa203c49c6184721.png"},{"id":44856204,"identity":"3915eb6f-1d56-4e41-a4b3-f3926c17c0a8","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":511648,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of plasma proteins between PMD (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=90) and Ctrl (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=54) by using a tandem mass tag (TMT)-based proteomics platform. (A)\u003c/strong\u003eThe PLS-DA algorithms of plasma proteins. (\u003cstrong\u003eB) \u003c/strong\u003eBinary comparison of plasma proteome data by volcano plot. The Log2FC between PMD and Ctrl of relative abundance for proteins illustrates positive values (pink) indicating upregulation in PMD and negative values (blue) indicating downregulation in PMD. Gray dots represent non-significantly expressed genes. The dashed line indicates the significant threshold.\u003cstrong\u003e (C) \u003c/strong\u003eThe numbers of the differential expression proteins (DEPs) between PMD and Ctrl. \u003cstrong\u003e(D) \u003c/strong\u003eThe top 10 upregulated and top 10 downregulated DEPs between PMD and Ctrl.\u003cstrong\u003e (E)\u003c/strong\u003e STRING interaction network of proteins from DEPs, the size of the circle is proportional to the significance, the shade is indicative of the fold change. \u003cstrong\u003e(F)\u003c/strong\u003e Proteins classification chart. Clusters of proteins were isolated and subjected to KEGG analysis. The rectangular bars of different colors represent different passways of proteins, the X-axis represents the number of protein classifications, and the Y-axis represents the protein classification entries.\u003c/p\u003e","description":"","filename":"Binder14.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/638ae7c9a12e55abd42272a0.png"},{"id":44857656,"identity":"975ccece-e549-4974-8975-9c72588efb15","added_by":"auto","created_at":"2023-10-18 14:21:24","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1920992,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMolecular subtypes for PMD patients based on multi-omics data. \u003c/strong\u003eClustering results of PMD patients and Ctrl. The heatmap displays the normalized expression of clinical factor-relevant molecules. The shades of color represent the relative fold change in abundance, with blue for increases and red for decreases in PMD patients.\u003c/p\u003e","description":"","filename":"Binder15.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/7cd587f932c1ac8f818ee58a.png"},{"id":44856206,"identity":"29d87c66-17b9-4f31-b67a-4e1b81b385a7","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":584851,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMachine learning for the multi-omics biomarkers and the validation of the biomarkers. (A)\u003c/strong\u003e Multi-omics biomarkers discovered by machine learning. \u003cstrong\u003e(B)\u003c/strong\u003e The diagnosis efficiency of omics. markers. \u003cstrong\u003e(C)\u003c/strong\u003e The diagnosis efficiency of omics. markers. combined with lactate/pyruvate (L/P) ratio. \u003cstrong\u003e(D)\u003c/strong\u003e The diagnosis efficiency of omics. markers. combined with FGF21. \u003cstrong\u003e(E)\u003c/strong\u003e The diagnosis efficiency of omics. markers. combined with GDF15. \u003cstrong\u003e(F) \u003c/strong\u003eThe diagnosis efficiency of omics. markers. combined with FGF21 and GDF21. \u003cstrong\u003e(G)\u003c/strong\u003e The diagnosis efficiency of omics. markers. combined with FGF21, GDF21, and L/P ratio.\u003c/p\u003e","description":"","filename":"Binder16.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/22ff3f7d57bfe9a0fe01c9e8.png"},{"id":44857657,"identity":"2ead559f-9637-4203-96a6-2ce7d2d41fe9","added_by":"auto","created_at":"2023-10-18 14:21:24","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":130878,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBiomarkers validated by the validation cohort (including PMD (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=24) and Ctrl (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e=24)). \u003c/strong\u003eThe expression and abundance of 13 alternative biomarkers are displayed in a boxplot of the validation cohort. \u003cstrong\u003e(A-J) \u003c/strong\u003eThe normalized of TAF15 mRNA expression level in whole blood\u003cstrong\u003e, \u003c/strong\u003ethe protein abundance of ILK level in plasma, the lactate abundance in plasma, the pyruvate abundance in plasma, the 2,4-di-tert-butylphenol abundance in plasma, the 2,3-dihydroxy-2-methylbutyrate abundance in plasma, the 2-hydroxyadipate abundance in plasma, the deoxycarnitine abundance in plasma, the acetylcarnitine (C2) abundance in plasma, the N, N, N-trimethyl-5-aminovalerate abundance in plasma between PMD and Ctrl in validation cohort. Significance determined by Student’s \u003cem\u003et\u003c/em\u003etest.\u003c/p\u003e","description":"","filename":"Binder17.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/1a7cc461c2e5b0bc0dd5cda5.png"},{"id":44856208,"identity":"e7535002-e79e-4fc7-8fec-92b5171496d5","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":192881,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIn-depth analysis of clinical manifestation subtypes and multi-omics data in the PMD patients from discovery cohort. \u003c/strong\u003e“\u003cem\u003eneuro\u003c/em\u003e” annotated for the patients suffers for only neurologic symptoms, “\u003cem\u003eboth\u003c/em\u003e” for the patients suffered for both neurologic and cardiac symptoms, and “\u003cem\u003eother\u003c/em\u003e” for the patients of the rest, the detailed criteria for each system can be attached in the Table 1. \u003cstrong\u003e(A-B)\u003c/strong\u003eThe heatmap of proteins and metabolites changed significantly between 3 subgroups of PMD. \u003cstrong\u003e(C-H) \u003c/strong\u003eThe normalized mRNA expression level of HBB, abundance of Q13418_ILK, lactate, 2,3-dihydroxy-e-methylbutyrate, 2-hydroxyadipate, gamma-glutamylserine in plasma between 3 subgroups of PMD.\u003c/p\u003e","description":"","filename":"Binder18.png","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/3bd6cbd39784381409ad9dd5.png"},{"id":60977555,"identity":"02ebfeb8-b4d7-4fb0-af74-1dbe77e41380","added_by":"auto","created_at":"2024-07-24 08:33:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5732210,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/a4a4aa09-cdc6-4355-a5a7-4b0f576805e3.pdf"},{"id":44858026,"identity":"e8437cbd-6d80-4988-88d0-52061e323e8c","added_by":"auto","created_at":"2023-10-18 14:29:24","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":18772,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Table 1\u003c/p\u003e","description":"","filename":"SupplementaryTable1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/deff898e8c1b4265d72b180b.xlsx"},{"id":44857653,"identity":"897ef621-7606-49c6-ac87-9dbd2ff5efd8","added_by":"auto","created_at":"2023-10-18 14:21:24","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":1066533,"visible":true,"origin":"","legend":"Supplementary Table 2","description":"","filename":"SupplementaryTable2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/b81bf4f42ded68529c13d8e6.xlsx"},{"id":44858337,"identity":"a853430f-327d-458c-9326-3562260922c4","added_by":"auto","created_at":"2023-10-18 14:37:24","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":97900,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Table 3\u003c/p\u003e","description":"","filename":"SupplementaryTable3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/5a05b40550954752bec77427.xlsx"},{"id":44856196,"identity":"7f507baa-ae5a-46a9-ab46-423d3ae39a80","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":68000,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Table 4\u003c/p\u003e","description":"","filename":"SupplementaryTable4.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/0b94c02c67f48cb9e8614f1e.xlsx"},{"id":44857651,"identity":"b0ebe560-6d13-437b-81d6-0afb6e85be7c","added_by":"auto","created_at":"2023-10-18 14:21:24","extension":"xlsx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":59075,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Table 5\u003c/p\u003e","description":"","filename":"SupplementaryTable5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/3c3756cdd7d60449ba8ad937.xlsx"},{"id":44856210,"identity":"791db43d-b1d0-4f1e-ac75-52ca57786ff6","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"xlsx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":2668287,"visible":true,"origin":"","legend":"Supplementary Table 6","description":"","filename":"SupplementaryTable6.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/67fef919513af79ed65d3e4c.xlsx"},{"id":44856209,"identity":"a65a78a2-b087-4a6e-a075-22681708c278","added_by":"auto","created_at":"2023-10-18 14:13:24","extension":"docx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":1850962,"visible":true,"origin":"","legend":"Supplementary Appendix","description":"","filename":"SupplementaryAppendix.docx","url":"https://assets-eu.researchsquare.com/files/rs-3389404/v1/94cb4cc77f3ab753aa5f29ea.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Integrative multi-omics profiling reveals the molecular subtypes and circulating biomarkers for pediatric mitochondrial disease","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePediatric mitochondrial disease (PMD) refers to the MD happens before 14 years old, which is a collection of rare, heterogenies and lethal syndromes with significant morbidity and mortality(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). The standard diagnostic procedure for PMD includes a complex combination of clinical assessments, blood metabolic profiles, brain imaging, tissue biopsies for histology or enzymology, and a genetic examination, which is a time-consuming and highly invasive process(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Also, it still lacks validated biomarkers for the clinical management and rigorous testing of novel therapy. The discovery of noninvasive biomarkers for MD has been the research hot spot, however, mostly studies used the samples of a specific subtype of MD (such MELAS syndrome(\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e), Leigh syndrome(\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e), TK2 deficient patients(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e), mtDNA deletions patients(\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e)) or teenager/adult(\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e) MD, which it still lacks noninvasive biomarkers with high sensitivity and specificity for PMD regardless of the complexity of clinical phenotypes and genetic backgrounds.\u003c/p\u003e \u003cp\u003eThe subtypes of MD mostly follow the genetic background (according to the disease-causing genes) or the clinical manifestations, however, there are more than 400 genes across both mitochondrial genome and nuclear genome have been identified as the disease-causing gene of MD, and there are more than 30 clinical syndrome subtypes (such as Leigh syndrome, MELAS syndrome, LHON and so on) of PMD (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). The lack of molecular subtyping of PMD dramatically hindered the development of PMD precision medicine. Evidence indicates the multi-omics analysis may give new insight into the molecular subtypes of the disease with diverse pathophysiologic mechanisms(\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). However, it still lacks study for the molecular subtype analysis for MD regardless the clinical phenotypes and genetic backgrounds.\u003c/p\u003e \u003cp\u003eThe developments of technologies especially the omics analysis such as genomics, transcriptomes, metabolomics, proteomics, and so on, gives new opportunities in the disease studies. The proteomics and metabolomics studies of an adult m.3243A\u0026thinsp;\u0026gt;\u0026thinsp;G cohort have been performed. The study validated 1 protein and 19 metabolites which may enable monitoring of mitochondrial disease(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Considering the diagnosis for PMD is more difficult than adult, here the cohort focus on the pediatrics, the mean age of PMD cohort is 5.6 years old, which range from 0.6 to 15 years old.\u003c/p\u003e \u003cp\u003eIn this research, we detected the characteristic molecular changes with the PMD blood samples, interrogated the molecular heterogeneity of PMD based on the omics data with 3 subtypes, and discovered the PMD diagnosis panel with high diagnosis efficiency (AUC\u0026thinsp;=\u0026thinsp;0.947). We hope the efforts give some proof for clinicians on PMD diagnosis and monitoring.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eOverview of PMD cohort for multi-omics analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe aim to investigate: (1) uncover the potential biomarkers to improve the diagnosis for PMD; (2) define the molecular subtypes of PMD regardless the genetic and clinical backgrounds\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eas shown in \u003cstrong\u003eFigure 1\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eTo achieve the two goals, here we included two independent cohorts of PMD: discovery cohort and validation cohort. Individuals for all cohorts are pediatrics, the mean age is around 5 years old with the range from 0.6 to 15 years (details was available in \u003cstrong\u003eTable 1\u003c/strong\u003e and \u003cstrong\u003eTable S1\u003c/strong\u003e). The discovery phase analyzed 144 whole blood and 144 plasma samples, which included 90 PMD patients, and 54 matched controls. The overall demographic and disease features of both cohorts are listed in \u003cstrong\u003eTable 1\u003c/strong\u003e, and the detailed medical records are available in \u003cstrong\u003eTable S1\u003c/strong\u003e. All individuals including patients and controls were in their baseline state of health at the timepoints of participation. All blood samples were taken at the first morning sample under the status of fasting.\u003c/p\u003e\n\u003cp\u003eComparing the clinical features across the patients and control, consistent with prior studies, PMD patients had a lower mean body mass index (BMI) than Ctrl cohorts. \u0026ldquo;Neurologic-affected\u0026rdquo; (73/90, 81%) was nominated for patients presented with epilepsy, dystonia and so on or Leigh syndrome or stroke-like features on MRI, \u0026ldquo;Cardiac-affected\u0026rdquo; (23/90, 26%) was nominated for patients have abnormalities in electrocardiogram or echocardiogram, developmental delay (86/90, 96%), gastrointestinal disturbances (36/90, 40%), psychotic problems (13/90, 14%), and hearing loss (3/90, 3%), are the frequently PMD complaints.\u0026nbsp;The validation cohort was recruited independently, with 48 individuals for 24 PMD and 24 controls, the overall demographic and disease features are listed in\u003cstrong\u003e\u0026nbsp;Table 1,\u0026nbsp;\u003c/strong\u003ewhich is consistent with the discovery cohort., and the detailed medical records are available in \u003cstrong\u003eTable S1\u003c/strong\u003e. 22 (92%) patients are \u0026ldquo;neurological-affected\u0026rdquo;, 5 (21%) patients are \u0026ldquo;cardiac-affected\u0026rdquo;, 21 (88%) patients presented with developmental delay, 19 (79%) patients suffered for gastrointestinal disturbance, 13 (14%) patients suffered psychiatric conditions, 3 (13%) patients suffered from hearing loss.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDetection of whole blood transcriptomic alterations in PMD patients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo investigate the pathogenesis and identify the potential biomarkers for PMD, whole blood transcriptome analysis was performed. Here, PAXgene blood RNA system was used for the whole blood transcriptome analysis. The resulting raw data were preprocessed to create normalized data for further analysis. The transcriptome data identified 15353 protein-coding transcripts with available \u003cem\u003ep\u0026nbsp;\u003c/em\u003evalue \u003cstrong\u003e(Supplementary Table 2)\u003c/strong\u003e. The partial least squares discriminant analysis (PLS-DA) algorithms were performed to visualize the overall trend in gene expression levels among all samples \u003cstrong\u003e(Figure 2A)\u003c/strong\u003e. The preliminary examination revealed that whole blood gene expression levels differed between PMD patients and controls. In total of 3454 transcripts significantly changed (with adjust \u003cem\u003ep\u003c/em\u003e value \u0026lt;0.05), in which 1226 upregulated and 2228 downregulated transcripts in MD patients compared to controls. Among 1226 upregulated-transcripts, 35 with the fold change (FC) more than 1.5 and among 2228 downregulated-transcripts, 130 with the FC less than 0.67 \u003cstrong\u003e(Figure 2B-C)\u003c/strong\u003e. The top ranked differentially expressed genes (DEGs) were analyzed accordingly \u003cstrong\u003e(Figure 2D)\u003c/strong\u003e. Disease specific DEGs were observed in the top alteration list, genes relevant to immune system such as killer-cell immunoglobulin-like receptors family members (KIR2DL3, KIR2DL4), metalloproteinases (MMP1, MMP8, ADAMTS2), hemoglobin related genes (HBB, HBA1). The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis performed based on the DEGs. The DEGs principally related to pathways of neurodegeneration, amyotrophic lateral sclerosis, Alzheimer disease, also the oxidative phosphorylation \u003cstrong\u003e(Figure 2E)\u003c/strong\u003e. All above indicating that whole blood transcripts could pinpoint key genes that are up- and down-regulated in the plasma of PMD patients may give the clues for the pathogenesis of MDs and the data generated is worth further analysis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetabolic characterization of plasma sample in PMD patients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePlasma metabolic profile was performed to further understand the pathogenesis and identify the potential biomarkers for PMD. The resulting raw data were preprocessed to create normalized data for further analysis. The plasma metabolome data identified 914 metabolites (992 metabolites in total, 78 metabolites with more than 50% missing values were eliminated) matched to known annotations \u003cstrong\u003e(Supplementary Table 3)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003ePLS-DA algorithms used for the preliminary examination revealed that plasm metabolite levels differed between PMD patients and controls \u003cstrong\u003e(Figure 3A)\u003c/strong\u003e. In a total of 185 metabolites significantly changed (with adjust \u003cem\u003ep\u003c/em\u003e value \u0026lt;0.05), in which 119 upregulated and 66 downregulated metabolites in MD patients compared to controls. Among 119 upregulated-metabolites, 58 with the FC more than 1.2 and among 66 downregulated- metabolites, 13 with the fold change less than 0.6 \u003cstrong\u003e(Figure 3B-C)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eTop differentially expressed metabolites (DEMs) mostly are disease specific \u003cstrong\u003e(Figure 3D)\u003c/strong\u003e. The FC of erythritol in PMD patients compared to controls was only 0.06. Erythritol is a newly recognized human metabolic product of glucose, synthesized through the pentose phosphate pathway (29). The elevated level of erythritol precedes the onset of central adiposity gain in humans. Most pediatric patients suffer for suboptimal growth and weight gain (7), the mean BMI in our PMD cohort (14.82) is also smaller than that in age-matched control cohort (16.78). Nucleotides such inosine and guanosine are decreased in the PMD. Top DEMs also includes lipids, such as 1-stearoyl-2-oleory-GPS (18:0/18:1), dexycarnitine, 3-hydroxybutyrylcarnitine, 3-hydroxyoleoylcarnitine, cis-3,4-methyleneheptanoate, 3-hydroxylaurate, and 3-hydroxymyristate \u003cstrong\u003e(Figure 3D)\u003c/strong\u003e. The Metabolite Set Enrichment Analysis (MSEA) pathway analysis performed based on the DEMs. The DEMs principally related to pathways of taurine and hypo-taurine and metabolism, glycerophospholipid metabolism, arginine biosynthesis, alanine, aspartate and glutamate metabolism, and TCA cycle\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003cstrong\u003e(Figure 3E)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProteomic characterization of plasma sample in PMD patients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePlasma protein profile was performed to further detectthe pathogenesis and findthe usefulbiomarkers for PMD. The resulting raw data were extractedto create normalized data for further analysis. The plasma proteome data quantified 888 proteins (1706 proteins in total, 818 proteins with more than 50% missing values were eliminated) matched to known annotations \u003cstrong\u003e(Supplementary Table 4)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eFirstly, we conducted PLS-DA algorithm analysis \u003cstrong\u003e(Figure 4A)\u003c/strong\u003e. The preliminary examination revealed that plasma protein levels differed between PMD patients and controls. In total of 113 proteins significantly changed (with adjust \u003cem\u003ep\u003c/em\u003e value \u0026lt;0.05), in which 11 upregulated and 102 downregulated proteins in PMD patients compared to controls. Among 11 upregulated-proteins, 5 with the FC more than 1.1 and among 102 downregulated-proteins, 69 with the fold change less than 0.8 \u003cstrong\u003e(Figure 4B-C)\u003c/strong\u003e. Top 10 upregulated differentially expressed proteins (DEPs) listed as: polymeric immunoglobulin receptor (PIGR), chromogranin A (CHGA), limbic system associated membrane protein (LSAMP), paraoxonase 1 (PON1), protein tyrosine phosphatase receptor type J (PTPRJ), collectin subfamily member 10 (COLEC10), reversion inducing cysteine rich protein with kazal motifs (RECK), prostaglandin D2 synthase (PTGDS), intelectin 1 (ITLN1), and sex hormone binding globulin (SHBG) \u003cstrong\u003e(Figure 4D)\u003c/strong\u003e. PTGDS is a member of lipocalin superfamily and plays dual roles in prostaglandin metabolism and lipid transport. The upregulation of PTGDS may related to the dysregulated of lipid metabolism in PMD patients. The top 10 downregulated DEPs listed as: myosin heavy chain 9 (MYH9), myosin light chain 12B (MYL12B), coronin 1C (CORO1C), C-reactive protein (CRP), myosin light chain 6 (MYL6), PDZ and LIM domain 1 (PDLIM1), actin alpha cardiac muscle 1 (ACTC1), clathrin heavy chain (CLTC), tropomyosin 4 (TPM4), and thymosin beta 4 X-linked (TMSB4X) \u003cstrong\u003e(Figure 4D)\u003c/strong\u003e. CRP is a biomarker produced by the liver in response to inflammation, the elevated of plasma CRP in PMD patients in consistent with the status of PMD. \u003cstrong\u003eFigure 4E\u003c/strong\u003e shows a visualization of known protein interactions. Protein data was analyzed by using STRING for protein network analysis. The groups of interacting proteins were isolated and individually subjected to gene ontology (GO) analysis using DAVID to determine biological function. Groups deemed significant (Benjamin-corrected \u003cem\u003ep\u003c/em\u003e value\u0026le;0.05) were assigned to the biological function with the lowest \u003cem\u003ep\u003c/em\u003e value. Multiple protein group functions especially the metabolic related groups demonstrated to be changed within PMD patients. GO analysis of plasma in PMD patients demonstrated that DEGs were mainly enriched in pathways of carbon metabolism, glycolysis/ gluconeogenesis, and biosynthesis of amino acids\u003cstrong\u003e\u0026nbsp;(Figure 4F)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThree molecular subtypes of PMD based on 3-layer omics profiles\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo comprehensively characterize the molecular features of the PMD, the whole blood transcripts, plasma metabolic profiles and proteomic profiles data from discovery cohort were used for the consensus clustering for subtypes. At first, all three layers of omics data including 15353 protein-coding transcripts, 914 metabolites, and 888 proteins were subjected to further analysis \u003cstrong\u003e(Supplementary Table 2-4)\u003c/strong\u003e. The PMD cohort was divided into three distinct clusters (subtype 1, subtype 2, and subtype 3), with an optimal consensus k-value of 3 \u003cstrong\u003e(Supplementary Figure 2A-B)\u003c/strong\u003e. The DEGs, DEMs and DEPs analysis were performed for each pair of molecular subtypes, and a total of 2 transcripts, 1 protein, and 152 metabolites were identified to be differentially expressed in the subtypes \u003cstrong\u003e(Figure 5, Table S5)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eOverall, the most different expression molecules among types are lipids. The subtype 1, is almost hypometabolic, especially downregulated in the lipids, and most upregulated in the amino acid, which was annotated for \u0026ldquo;AA-META\u0026rdquo;. The subtype 2 is almost upregulated in lipids and downregulated in the amino acid, which was annotated for \u0026ldquo;LIP-META\u0026rdquo;. As the subtype 3, most molecules are in the middle of two types, which is annotated for \u0026ldquo;MIDDLE-META\u0026rdquo;.\u003c/p\u003e\n\u003cp\u003eInterestingly, when combined with the phenotype, we find the patients of \u0026ldquo;AA-META\u0026rdquo; seems more severe: according to \u003cstrong\u003eTable 1\u003c/strong\u003e, the ratio of patients suffering for both cardiac and neurological problem in the discovery cohort is about 18/90 (20%), however, the ratio in \u0026ldquo;AA-META\u0026rdquo; type was 14/44 (32%), in \u0026ldquo;LIP-META\u0026rdquo; was 4/41 (10%), and in \u0026ldquo;MIDDLE-META\u0026rdquo; was 0/5 (0%). Consistently, ratio of patients only suffering for cardiac symptoms in the discovery cohort is about 23/90 (26%), however, the ratio in \u0026ldquo;AA-META\u0026rdquo; type was 16/44 (36%), in \u0026ldquo;LIP-META\u0026rdquo; was 7/41 (17%), and in \u0026ldquo;MIDDLE-META\u0026rdquo; was 0/5 (0%). The ratio of patients suffering only for neurologic problems in the discovery cohort is about 73/90 (81%), however, the ratio in \u0026ldquo;AA-META\u0026rdquo; type was 39/44 (89%), in \u0026ldquo;LIP-META\u0026rdquo; was 30/41 (73%), and in \u0026ldquo;MIDDLE-META\u0026rdquo; was 4/5 (80%) \u003cstrong\u003e(Figure 5)\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIdentification of key DEGs, DEMs, and DEPs as biomarkers of PMD based on machine learning algorithms\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo confirm the significance of key DEGs, DEMs, and DEPs as biomarkers of PMD, we used three different machine learning algorithms to screen. Twenty key biomarkers were identified by RF algorithm, twenty-six key biomarkers were identified by SVM-RFE algorithm, and sixty-one key biomarkers were identified by Boruta algorithm. Thirteen overlapping biomarkers (TAF 15, HBB, ILK, lactate, pyruvate, deoxycarnitine, 2-hydroxyadipate, acetylcarnitine, N-acetylasparagine, 2,4-di-tert-butylphenol, gamma-glutamylserine, 2,3-dihydroxy-2-methybutyrate, and N, N, N-trimethyl-5-aminovalerate) were selected and give the name for \u0026ldquo;omics. markers\u0026rdquo; \u003cstrong\u003e(Figure 6A)\u003c/strong\u003e. The results of ROC curves indicated that the omics. makers which filtered out by machine learning algorithms have a favorable diagnostic value in the test cohort (the discovery cohort was randomly divided into training cohort and test cohort with the ratio of 0.8:0.2), with the AUC of 0.947 \u003cstrong\u003e(Figure 6B)\u003c/strong\u003e.\u0026nbsp;Considering the lactate level, pyruvate level, and lactate/pyruvate (L/P) ratio were normally used for the\u0026nbsp;clinical\u0026nbsp;evaluation of MD, we combined the omics. markers with L/P ratio trying to improve the diagnostic rate, however, the AUC remains the same as 0.947 \u003cstrong\u003e(Figure 6C)\u003c/strong\u003e. As FGF-21 and GDF-15 were considered as new MD biomarkers, the AUC of omics. markers combined with FGF21 was about 0.942\u003cstrong\u003e\u0026nbsp;(Figure 6D)\u003c/strong\u003e. The AUC of omics. markers combined with GDF15 was about 0.938 \u003cstrong\u003e(Figure 6E)\u003c/strong\u003e. The AUC of omics. markers combined with FGF21 and GDF15 was about 0.940 \u003cstrong\u003e(Figure 6F)\u003c/strong\u003e. Finally, we combined omics. markers with FGF21, GDF15, and L/P ratio, got the AUC 0.937 \u003cstrong\u003e(Figure 6G)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eTo confirm the results from the discovery cohort data in relation to differentiating a separate set of PMD from controls, we validated the 13 candidate biomarkers in \u003cstrong\u003eFigure 6A\u0026nbsp;\u003c/strong\u003evia the independent validation cohort (including 24 PMD patients and 24 pediatric controls). The results confirm that among 13 candidate biomarkers, 10 biomarkers are significant changed in the validation cohort. The mRNA expression level of TAF15 is down regulated in validation cohort, which is consistent with the data in discovery cohort \u003cstrong\u003e(Figure 7A)\u003c/strong\u003e. The plasma ILK level also decreased in PMD \u003cstrong\u003e(Figure 7B)\u003c/strong\u003e. The classic PMD biomarkers lactate and pyruvate levels both elevated in validation cohort \u003cstrong\u003e(Figure 7C-D)\u003c/strong\u003e. 2,4-Di-tert-butylphenol (2,4-DTBP) is a common toxic secondary metabolite which is also elevated in the PMD patients \u003cstrong\u003e(Figure 7E)\u003c/strong\u003e. 2,3-dihydroxy-2-methylbutyrate has been considered a significant marker of ECHS1 deficiency(16, 17), here we found also increased in the expanded PMD cohort \u003cstrong\u003e(Figure 7F)\u003c/strong\u003e. 2-hydroxyadipate and deoxycarnitine both increased in the PMD in the validation cohort \u003cstrong\u003e(Figure 7G-H)\u003c/strong\u003e. Deoxycarnitine (also named for \u0026gamma;-butyrobetaine), a precursor of carnitine, is released in circulation to be taken up by skeletal muscle for storage. It has been reported that the increase of deoxycarnitine in circulation can be caused by the reduced uptake capacity in muscle tissue(18). Blood concentration of acetylcarnitine (C2) reflect intracellular levels and the regulation of acetyl CoA and free CoA via carnitine acetyl-CoA transferase, increased production of C2 represents the down regulated in glucose oxidation or upregulated in \u0026beta;-oxidation, which is consistent with results that the acetylcarnitine is upregulated in the PMD patients as the most PMD patients are deficit in OXPHOS function as hinder the glucose oxidation but enhance mitochondrial \u0026beta;-oxidation \u003cstrong\u003e(Figure 7I)\u003c/strong\u003e. Through\u0026nbsp;our data, the N, N, N-trimethyl-5-aminovaleric acid (TMAVA) level is decreased in the pan-PMD patients in both discovery and validation cohort \u003cstrong\u003e(Figure 7J)\u003c/strong\u003e. The HBB, N-acetylasparagine, and gamma glutamylserine showed no significant changes in the validation cohort may be due to the patient number \u003cstrong\u003e(Figure S3A-C)\u003c/strong\u003e. Also, it may indicate that the biomarkers may be used as a panel rather than used separately.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003ePMDs are both highly heterogeneity genetically and clinically heterogeneity, which is difficult for diagnosis and treatment. To address these difficulties in the PMD medications, we recruited 114 PMD patients (both clinical and genetically diverse) and 78 matched controls to conduct the whole blood transcriptome profiles, plasma metabolomics profiles, and proteomics profiles to find blood biomarkers help diagnosis and defined the PMDs molecular subtypes help the precise treatment.\u003c/p\u003e \u003cp\u003ePreviously studies mostly some certain type of PMD, e.g., m.3243A\u0026thinsp;\u0026gt;\u0026thinsp;G associated PMD (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Considering the pediatric PMD mostly are syndromic presented with severe symptoms, diverse clinical manifestations, and poor prognosis. Studies based on a large diverse pediatric PMD cohort may be supposed to carry out. We first analyzed the data from the discovery cohort (including 90 pediatric PMD patients and 54 matched controls). Overall, 3454 DEGs, 185 DEMs, and 113 DEPs are first generated. The differential analysis based on transcriptome profiles, plasma metabolomics profiles, and proteomics profiles indicates the blood molecules have the potential to be the biomarkers for the PMDs, as most differential expressed blood molecules are metabolism even correlate with cell bioenergetics.\u003c/p\u003e \u003cp\u003eTo identify the key differentially expressed molecules which may have the potential to be the diagnosis biomarkers for PMDs, here we used three different machine learning algorithms (including RF algorithm, SVM-RFE algorithm, and Boruta algorithm) to screen key DEGs, DEMs, and DEPs. Thirteen overlapping biomarkers (TAF 15, HBB, ILK, lactate, pyruvate, deoxycarnitine, 2-hydroxyadipate, acetylcarnitine, N-acetylasparagine, 2,4-di-tert-butylphenol, gamma-glutamylserine, 2,3-dihydroxy-2-methybutyrate, and N, N, N-trimethyl-5-aminovalerate) were selected, with the AUC of 0.947, which the efficiency is much better than the traditional biomarkers. Furthermore, the thirteen biomarkers were applied in the independent validation cohort. One transcript biomarker TAF15 still significantly downregulated in PMD, one protein biomarker ILK is also downregulated in PMD in the validation cohort. Classic biomarkers including lactate, pyruvate, and 2,3-dihydroxy-2-methybutyrate all upregulated in PMD patients in the validation cohort.\u003c/p\u003e \u003cp\u003e \u003cb\u003eThe omics-based biomarkers panel has higher diagnostic efficiency than conventional biomarkers in PMD cohort.\u003c/b\u003e The traditional PMD biomarkers can be mostly correlated to the genetic background, the biomarkers applied in diverse PMD mostly lack sensitivity and specificity such as lactate and pyruvate. Our results also indicate the omics biomarkers are more suitable for use as a panel. Convention blood biomarkers for PMD are lack of sensitivity(\u003cspan additionalcitationids=\"CR20\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e), which also had a bad performance in our PMD cohort: the area under the curve (AUC) of lactate is 0.697, pyruvate is 0.625, lactate/pyruvate ratio is 0.421, combined lactate and pyruvate is 0.740, and combined lactate, pyruvate and lactate/pyruvate ratio is up to 0.752 (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA-E). Fibroblast growth factor 21(FGF21) (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e) and growth differentiation factor 15 (GDF15) emerged as superior indicators of PMD (\u003cspan additionalcitationids=\"CR24 CR25 CR26\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e), especially the GDF15. In our PMD cohort, the protein biomarkers behave better than conventions: the AUC for FGF21 is 0.675, GDF15 is 0.894, and the combination of FGF21 and GDF15 is 0.889 \u003cb\u003e(Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eF-H)\u003c/b\u003e. When combined the conventional metabolic biomarkers with protein biomarkers FGF21 and GDF15, the diagnosis efficiency dramatically increased: when combined the pyruvate with FGF21 and GDF15, the AUC raised to 0.925, which indicating that the combination of omics biomarkers may help the precision diagnosis of PMD \u003cb\u003e(Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eI-L)\u003c/b\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eInsight into organ system.\u003c/b\u003e Despite the omics dramatically improves the diagnosis efficiency for PMD, we still find it difficult in assess the clinical manifestations: considering PMD patients frequently suffers for neurologic problem, the PMD patients suffered from cardiac symptoms are at risk for sudden death or major cardiac event. Here, we analyze the PMD patients under different manifestations. We reviewed the medical record and divided the discovery cohort into three groups (detailed criteria see Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e): \u0026ldquo;\u003cem\u003eneuro\u003c/em\u003e\u0026rdquo; for the individuals suffers from \u0026ldquo;Neurological-affected\u0026rdquo; but not \u0026ldquo;cardiac-affected\u0026rdquo;, \u0026ldquo;\u003cem\u003eboth\u003c/em\u003e\u0026rdquo; for the individuals both \u0026ldquo;Neurological-affected\u0026rdquo; and \u0026ldquo;cardiac-affected\u0026rdquo;, and \u0026ldquo;\u003cem\u003eother\u003c/em\u003e\u0026rdquo; for the rest. We find there are no significant changes in the transcriptomic profile, 105 proteins changed between 3 groups, and 22 metabolites changed between 3 groups \u003cb\u003e(Figure S4A-B, Table S6)\u003c/b\u003e. Interestingly, the differential expressed molecules between diverse PMD, and control are mostly metabolites, the differential expressed molecules between different clinical manifestations mainly proteins \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e8\u003c/span\u003eA-B\u003cb\u003e)\u003c/b\u003e. All DEM in between three groups are regulated in the \u0026ldquo;\u003cem\u003eboth\u003c/em\u003e\u0026rdquo; group, most are key metabolites from bioenergetic pathways, e.g., lactate and branched-chain amino acids (BCAAs). Especially, a group of BCAAs\u0026rsquo; 1-carboxyethyl conjugates can be found here. To identify the omics biomarkers in if works in for the identification of different clinical manifestation, we further evaluate the levels of 13 biomarkers in 3 groups. However, it needs to expand the cohort and further studies. Among thirteen biomarkers, seven biomarkers showed no significant changes \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e8\u003c/span\u003eC-H, \u003cb\u003eFigure S4C-I)\u003c/b\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eComparing with former studies.\u003c/b\u003e Vamsi K. Mootha Team(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e) conducted a study recruited a big m.3243A\u0026thinsp;\u0026gt;\u0026thinsp;G cohort, conducting the proteomics and metabolomics for the biomarker discovery, which validated classic biomarkers (GDF15, lactate, pyruvate, alanine) and 3 biochemical families (N-lactoyl-amino acids, hydroxy-fatty acids, hydroxyacyl carnitines). Here, in our cohorts we also validated the classic biomarkers, the hydroxyacyl carnitines (including C2:0, C3:0, and C4:0) all significantly upregulated in PMD which is consistent with the former study, however, only the acetylcarnitine and deoxycarnitine (metabolic precursor of carnitine) is recruited in the biomarkers panel after machine learning \u003cb\u003e(Supplementary Table\u0026nbsp;3)\u003c/b\u003e, same situation with the hydroxy-fatty acids. The possible main reasons may list as: 1. The different cohorts, as the metabolites and proteins dramatically changed with age, sex, and BMI, the former cohort is mostly adults, with the mean age of about 40 years old, and the mean age from our cohort is only 5 years old. 2. The former cohort is focus on the m.3243A\u0026thinsp;\u0026gt;\u0026thinsp;G individuals, the cohort here is more diverse.\u003c/p\u003e \u003cp\u003eIn this study, we hypothesized that pediatric PMD induces characteristic molecular changes that can be detected in the invasive samples including blood. To test the hypothesis, two independents pediatric PMD cohorts including discovery (90 patients and 54 controls) and validation (24 patients and 24 controls) recruited. Firstly, based on the discovery cohort, whole blood transcriptomic profiles, plasma metabolomics profiles, and plasma proteomics profiles were performed. Secondly, the integrative analysis of three-layer omics was performed for molecular subtype taxonomy of pediatric PMD despite of genetic background but based on the differentially expressed plasma metabolites and proteins. Thirdly, all differentially expressed molecules used for the biomarker discovery, an omics biomarker set (two transcripts, one protein, and ten metabolites) was produced, which was validated based on the validation cohort. Finally, in-depth analysis based on neurologic and cardiologic impairment was performed. The observation may represent a step towards PMD diagnosis and therapy.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003ePlease see the supplemental material for detailed methods.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStudy approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll samples and clinical information were collected under the approval of the Ethics Committee of Peking University First Hospital (2017-217) and the Second Affiliated Hospital of Wenzhou Medical University (2021-K-54-02).\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConflict of Interest:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have declared that no conflict of interest exists.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; Contributions:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.L, Y.Y, H.F, and X.L were responsible for overarching research goals and aims. J.L, Y.Y, H.F, and X.L conceived and designed the experiments. Z.C, Y.Z, P.L, C.Z, and X.M collected and stored samples from patients and were involved in the translation of results to clinical implications. X.L and Q.Z performed data interpretation of whole blood transcription data, untargeted-metabolomics data, and untargeted plasma proteome data. T.G was responsible for the experimental design of untargeted plasma proteome analysis. Y.Z was responsible for ELISA experiments and analyzed the data. Y.Y, X.L, Z.C, Y.Z, X.Z, Y.X, Y.W, L.Z, Q.D, and X.Y collected and reviewed the medical records from patients. X.L, Z.C, and Y.Z drafted the manuscripts. J.L, Y.Y, and H.F proofread the manuscript. All authors read and approved the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.L, Y.Y, H.F, and X.L was responsible for overarching research goals and aims. J.L, Y.Y, H.F, and X.L conceived and designed the experiments. Z.C, Y.Z, P.L, C.Z, and X.M collected and stored samples from patients and were involved in the translation of results to clinical implications. X.L and Q.Z performed data interpretation of whole blood transcription data, untargeted-metabolomics data, and untargeted plasma proteome data. T.G was responsible for the experimental design of untargeted plasma proteome analysis. Y.Z was responsible for ELISA experiments and analyzed the data. Y.Y, X.L, Z.C, Y.Z, Y.X, Y.W, L.Z, Q.D, X.Y, and X.Z collected and reviewed the medical records from patients. X.L, Z.C, and Y.Z drafted the manuscripts. J.L, Y.Y, and H.F proofread the manuscript. All authors read and approved the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to thank all collaborators and patient participants for their contributions to this study.\u003cstrong\u003e\u003cbr\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSchapira AH. Mitochondrial diseases. Lancet. 2012;379(9828):1825\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThompson K, Collier JJ, Glasgow RIC, Robertson FM, Pyle A, Blakely EL, et al. Recent advances in understanding the molecular genetic basis of mitochondrial disease. J Inherit Metab Dis. 2020;43(1):36\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParikh S, Goldstein A, Karaa A, Koenig MK, Anselm I, Brunel-Guitton C, et al. Patient care standards for primary mitochondrial disease: a consensus statement from the Mitochondrial Medicine Society. Genet Med. 2017;19(12).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEsterhuizen K, Lindeque JZ, Mason S, van der Westhuizen FH, Suomalainen A, Hakonen AH, et al. A urinary biosignature for mitochondrial myopathy, encephalopathy, lactic acidosis and stroke like episodes (MELAS). Mitochondrion. 2019;45:38\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThompson Legault J, Strittmatter L, Tardif J, Sharma R, Tremblay-Vaillancourt V, Aubut C, et al. A Metabolic Signature of Mitochondrial Dysfunction Revealed through a Monogenic Form of Leigh Syndrome. Cell Rep. 2015;13(5):981\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDominguez-Gonzalez C, Badosa C, Madruga-Garrido M, Mart\u0026iacute; I, Paradas C, Ortez C, et al. Growth Differentiation Factor 15 is a potential biomarker of therapeutic response for TK2 deficient myopathy. Sci Rep. 2020;10(1):10111.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalko SG, Paco S, Jou C, Rodr\u0026iacute;guez MA, Meznaric M, Rogac M, et al. Transcriptomic profiling of TK2 deficient human skeletal muscle suggests a role for the p53 signalling pathway and identifies growth and differentiation factor-15 as a potential novel biomarker for mitochondrial myopathies. BMC Genomics. 2014;15:91.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eForsstr\u0026ouml;m S, Jackson CB, Carroll CJ, Kuronen M, Pirinen E, Pradhan S, et al. Fibroblast Growth Factor 21 Drives Dynamics of Local and Systemic Stress Responses in Mitochondrial Myopathy with mtDNA Deletions. Cell Metab. 2019;30(6):1040-54.e7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavis RL, Liang C, Edema-Hildebrand F, Riley C, Needham M, and Sue CM. Fibroblast growth factor 21 is a sensitive biomarker of mitochondrial disease. Neurology. 2013;81(21):1819\u0026ndash;26.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavis RL, Liang C, and Sue CM. A comparison of current serum biomarkers as diagnostic indicators of mitochondrial diseases. Neurology. 2016;86(21):2010\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahman S. Mitochondrial disease in children. J Intern Med. 2020;287(6):609\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNeff RA, Wang M, Vatansever S, Guo L, Ming C, Wang Q, et al. Molecular subtyping of Alzheimer's disease using RNA sequencing data reveals novel mechanisms and targets. Sci Adv. 2021;7(2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGreenwood AK, Montgomery KS, Kauer N, Woo KH, Leanza ZJ, Poehlman WL, et al. The AD Knowledge Portal: A Repository for Multi-Omic Data on Alzheimer's Disease and Aging. Curr Protoc Hum Genet. 2020;108(1):e105.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIturria-Medina Y, Adewale Q, Khan AF, Ducharme S, Rosa-Neto P, O'Donnell K, et al. Unified epigenomic, transcriptomic, proteomic, and metabolomic taxonomy of Alzheimer's disease progression and heterogeneity. Sci Adv. 2022;8(46):eabo6764.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSharma R, Reinstadler B, Engelstad K, Skinner OS, Stackowitz E, Haller RG, et al. Circulating markers of NADH-reductive stress correlate with mitochondrial disease severity. J Clin Invest. 2021;131(2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHaack TB, Jackson CB, Murayama K, Kremer LS, Schaller A, Kotzaeridou U, et al. Deficiency of ECHS1 causes mitochondrial encephalopathy with cardiac involvement. Ann Clin Transl Neurol. 2015;2(5):492\u0026ndash;509.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeters H, Buck N, Wanders R, Ruiter J, Waterham H, Koster J, et al. ECHS1 mutations in Leigh disease: a new inborn error of metabolism affecting valine metabolism. Brain. 2014;137(Pt 11):2903\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaoi M, Li A, McGlory C, Stokes T, von Allmen MT, Phillips SM, et al. Metabolic Perturbations from Step Reduction in Older Persons at Risk for Sarcopenia: Plasma Biomarkers of Abrupt Changes in Physical Activity. Metabolites. 2019;9(7).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuomalainen A, Elo JM, Pietil\u0026auml;inen KH, Hakonen AH, Sevastianova K, Korpela M, et al. FGF-21 as a biomarker for muscle-manifesting mitochondrial respiratory chain deficiencies: a diagnostic study. Lancet Neurol. 2011;10(9):806\u0026ndash;18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoenig MK. Presentation and diagnosis of mitochondrial disorders in children. Pediatr Neurol. 2008;38(5):305\u0026ndash;13.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMunnich A, and Rustin P. Clinical spectrum and diagnosis of mitochondrial disorders. Am J Med Genet. 2001;106(1):4\u0026ndash;17.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCrooks DR, Natarajan TG, Jeong SY, Chen C, Park SY, Huang H, et al. Elevated FGF21 secretion, PGC-1α and ketogenic enzyme expression are hallmarks of iron-sulfur cluster depletion in human skeletal muscle. Hum Mol Genet. 2014;23(1):24\u0026ndash;39.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRiley LG, Nafisinia M, Menezes MJ, Nambiar R, Williams A, Barnes EH, et al. FGF21 outperforms GDF15 as a diagnostic biomarker of mitochondrial disease in children. Mol Genet Metab. 2022;135(1):63\u0026ndash;71.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLehtonen JM, Auranen M, Darin N, Sofou K, Bindoff L, Hikmat O, et al. Diagnostic value of serum biomarkers FGF21 and GDF15 compared to muscle sample in mitochondrial disease. J Inherit Metab Dis. 2021;44(2):469\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSalehi MH, Kamalidehghan B, Houshmand M, Aryani O, Sadeghizadeh M, and Mossalaeie MM. Association of fibroblast growth factor (FGF-21) as a biomarker with primary mitochondrial disorders, but not with secondary mitochondrial disorders (Friedreich Ataxia). Mol Biol Rep. 2013;40(11):6495\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMontero R, Yubero D, Villarroya J, Henares D, Jou C, Rodr\u0026iacute;guez MA, et al. GDF-15 Is Elevated in Children with Mitochondrial Diseases and Is Induced by Mitochondrial Dysfunction. PLoS One. 2016;11(2):e0148709.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYatsuga S, Fujita Y, Ishii A, Fukumoto Y, Arahata H, Kakuma T, et al. Growth differentiation factor 15 as a useful biomarker for mitochondrial disorders. Ann Neurol. 2015;78(5):814\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Table","content":"\u003cp\u003e\u003cstrong\u003eTable 1. Demographics and disease features of the cohorts\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"605\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"20.826446280991735%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"32.231404958677686%\" colspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eDiscovery cohort\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"32.231404958677686%\" colspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eValidation cohort\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.710743801652892%\" colspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"20.75782537067545%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePMD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCtrl\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePMD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCtrl\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.662273476112027%\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eUnits\u003c/strong\u003e\u003cstrong\u003e（\u003c/strong\u003e\u003cstrong\u003eRange\u003c/strong\u003e\u003cstrong\u003e）\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"20.75782537067545%\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.662273476112027%\" colspan=\"2\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"20.75782537067545%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSex, male/female\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e49/41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e30/36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e13/11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e16/8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.662273476112027%\" colspan=\"2\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"20.75782537067545%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge, mean (Range, SD)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e5.6(0.6-15, 3.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e6.8(0.6-15, 3.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e6.09(1.5-16, 4.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e5.69(2.6-12, 3.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.662273476112027%\" colspan=\"2\"\u003e\n \u003cp\u003eyears\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"20.75782537067545%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBMI, mean (Range, SD)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e14.82(10.94-22.71, 2.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e16.78(12.88-23.46, 2.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e14.75(10.61-20.55, 2.36)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e16.56(12.81-23.73, 2.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.662273476112027%\" colspan=\"2\"\u003e\n \u003cp\u003eKg/m\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"99.33993399339934%\" colspan=\"6\"\u003e\n \u003cp\u003e\u003cstrong\u003eDisease features\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.6600660066006601%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"20.75782537067545%\"\u003e\n \u003cp\u003e\u003cstrong\u003eNeurological affected\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e73(81%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e22(92%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.14497528830313%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.662273476112027%\" colspan=\"2\" rowspan=\"6\"\u003e\n \u003cp\u003eNumber (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.324324324324323%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCardiac affected\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e23(26%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e5(21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.324324324324323%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGrowth failure\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e86(96%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e21(88%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.324324324324323%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGastrointestinal disturbance\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e36(40%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e19(79%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.324324324324323%\"\u003e\n \u003cp\u003e\u003cstrong\u003epsychotic problems\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e13(14%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e10(42%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.324324324324323%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHearing loss\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e3(3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e3(13%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.91891891891892%\"\u003e\n \u003cp\u003e/\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNote: The positive inclusion criteria of neurological manifestations were the presence of epilepsy, dystonia and so on or Leigh syndrome or stroke-like features on MRI. The positive inclusion for heart disease is abnormalities in electrocardiogram or echocardiogram.\u003c/p\u003e\n"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":false,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3389404/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3389404/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePediatric mitochondrial disease (PMD) refers the MD happened before 14 years old, which is a collection of rare, heterogenies and lethal syndromes. However, PMD still lacks molecular subtypes and a noninvasive diagnostic biomarker for precise medication and early diagnosis. By using multi-omics analyses for the discovery cohort, the molecular subtypes and robust biomarkers firstly discovered. The biomarkers further validated in an independent cohort. We found multiple energetic pathways altered in the PMD plasma (proteomics and metabolomics) and blood cells (transcriptomes), indicating the qualification of working pipelines. Some pathways were discovered without expectation may provide new insight into PMD pathogenesis. Molecular subtypes modeling revealed that PMD can be calcified into “AA-META”, “LIP-META” and “MIDDLE-META”, interestingly, the “AA-META” correlated with severe symptoms with a higher rate of neurologic and cardiac affected. Based on three machine learning algorithms, we discovered a panel of biomarkers with 13 molecules (1 gene, 2 proteins, and 10 metabolites), including classic (lactate, pyruvate) and novel biomarkers, showed more effective diagnosis rate of PMD (AUC=0.947) than reported ones. Overall, our work defined molecular subtypes of PMD and established a new panel of biomarkers for the precision diagnosis of PMD.\u003c/p\u003e","manuscriptTitle":"Integrative multi-omics profiling reveals the molecular subtypes and circulating biomarkers for pediatric mitochondrial disease","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-10-18 14:13:19","doi":"10.21203/rs.3.rs-3389404/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"7024dd57-ef07-4917-9ebc-28425953fe4d","owner":[],"postedDate":"October 18th, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":25450490,"name":"Health sciences/Biomarkers/Diagnostic markers"},{"id":25450491,"name":"Health sciences/Diseases/Metabolic disorders"},{"id":25450492,"name":"Biological sciences/Molecular biology/Transcriptomics"},{"id":25450493,"name":"Biological sciences/Molecular biology/Proteomics"}],"tags":[],"updatedAt":"2026-07-21T06:41:39+00:00","versionOfRecord":[],"versionCreatedAt":"2023-10-18 14:13:19","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3389404","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3389404","identity":"rs-3389404","version":["v1"]},"buildId":"rHA-KDH7Qsr4HCuvH75dn","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}