Identification of m6A-associated diagnostic biomarkers and subtypes for osteoporosis's disease diagnosis and risk prediction based on machine learning

Identification of m6A-associated diagnostic biomarkers and subtypes for osteoporosis's disease diagnosis and risk prediction based on machine learning | 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 Identification of m6A-associated diagnostic biomarkers and subtypes for osteoporosis's disease diagnosis and risk prediction based on machine learning Shengzhong Luo, Jiahao Liu, Xinxin Miao, Jianjian Deng, Tianlong Wu, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4153446/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Osteoporosis(OP) is a systemic skeletal dysfunction disorder that occurs in large numbers worldwide. The aim of our study was to screen potential m6A-related diagnostic markers and establish diagnostic predictive models to provide a reference framework for the diagnosis and treatment of OP. Methods GSE56815 and GSE35956 were retrieved from GEO. The m6A-related diagnostic markers for OP were identified through random forest algorithm screening of GSE56815, and subsequently validated for diagnostic efficacy in GSE35956. The mechanism exploration of diagnostic markers was conducted via GO and KEGG analyses based on GSEA. Immune cell infiltration estimation was performed using ssGSEA, with correlation analysis between immune cells and diagnostic markers. Consensus clustering produced two OP patient subtypes, which were compared using m6A scores. Results 4 m6A-related diagnostic markers for OP have been selected by random forest. Utilizing these 4 diagnostic markers, an OP diagnostic model was successfully established, and its diagnostic value was validated using independent external data. The immune infiltration analysis of ssGSEA revealed that CD56 dim natural killer cell significantly infiltrated the OP samples, while the diagnostic markers were found to possess regulatory effects on diverse immune cells. OP samples were classified into two m6A subtypes through the concordance clustering, with type A having a higher m6A score than type B. The classification can provide more instructive assistance for the diagnosis of OP. Conclusions A diagnostic model for OP was constructed based on four m6A-related genes in the study, which provides significant references for diagnosis of OP and holds practical significance. osteoporosis m6A diagnostic biomarkers diagnostic predictive model machine learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introdution Osteoporosis (OP) is a prevalent degenerative disease associated with aging, symbolic of diminished bone density that leads to an elevated risk of fractures. A highly heritable trait, bone mineral density (BMD), represents the utmost clinically significant risk factor for fracture diagnosis in OP [ 1 ]. Each year, a total of 8.9 million osteoporotic fractures are incurred globally [ 2 ]. Among postmenopausal women, osteoporotic fractures are suffered by around half [ 3 ]. The health of elderly individuals is adversely impacted by osteoporotic fractures [ 4 ]. After hip fractures, a significant correlation exists between OP and notable morbidity and mortality rates [ 5 ].As a result of the accelerated aging of the global population, an escalation of OP rates has been observed, posing a considerable challenge to public healthcare [ 6 ]. Based on its etiology, OP can be classified into two categories: primary OP, which comprises of postmenopausal, senile, and juvenile OP, and secondary OP, which stems mainly from prolonged drug usage [ 7 ]. At an early stage, the diagnosis of osteoporosis remains a challenge despite improvements in therapeutic strategies, and the prognosis is poor [ 8 ]. Hence, there is an imperative necessity to enhance our comprehension of the pathogenesis of OP, with the aim of identifying innovative biomarkers and formulating novel therapeutic mechanisms for both the prevention and treatment of this disease. m6A, a reversible and dynamic modification of RNA, is governed by m6A WERs, commonly known as "writers" ,"erasers" ,and "readers"[ 9 ]. The participation of m6A methylation in RNA biology in terms of mRNA splicing, translation, and decay is apparent through an expanding list of components such as m6A readers, m6A writer-complex, and erasers[ 10 ]. The implication of m6A in the onset of diverse human diseases has been documented[ 11 ]. It has been reported that m6A, an RNA epigenetic modification, plays a crucial role in the regulation of osteogenic differentiation[ 12 ]. For example, ALKBH5 has been demonstrated to regulate osteogenic differentiation of MSCs by means of m6A demethylation activity[ 13 ]. The osteogenesis of BMSCs is reported to be regulated by METTL14 through the induction of autophagy via the m6A/IGF2BPs/Beclin-1 signal axis[ 14 ]. The effect of m6A methylation on osteoclasts has also been observed, with previous studies demonstrating that osteoblast exosomal Circ_0008542 promotes bone resorption induced by osteoclasts through m6A methylation[ 15 ]. Bone density loss can be clinically manifested as osteopenia or OP due to the dysfunction of either osteoclasts or osteoblasts, which results in perturbation[ 16 ]. Therefore, the m6A methylation modification is anticipated to significantly impact the pathogenesis of OP, and an investigation into the association of m6A with OP would aid in comprehending the pathophysiological mechanisms underlying its development, thereby providing clinical utility. The popularity of machine learning (ML) has experienced a rapid growth Over the last several years, as evidenced by its increasing usage[ 17 ]. Utilization of various machine learning algorithms was employed to Discriminate diagnostic feature biomarkers[ 18 ]. In this study, transcriptome sequencing data of OP blood samples was collected from the GEO database. Machine learning algorithms were utilized to select m6A-related diagnostic markers of OP. Subsequently, a diagnostic predictive model was established based on four selected diagnostic markers. Additionally, OP was clustered, which laid the groundwork for early OP diagnosis and demonstrated practical applicability. Materials and methods Datasets and data pre-processing The RNA-seq profiles were searched in the GEO database using the keywords "osteoporosis" to retrieve raw gene expression profiling data. Two datasets, namely GSE56815 and GSE35956, were obtained from the search. The former dataset consisted of 40 people with low hip bone mineral density (BMD) and 40 subjects with high hip BMD. The latter dataset had 5 patients with osteoporosis (OP) and 5 normal samples, which were utilised as external validation datasets. The R package "limma" was utilised to do background correction and data normalisation for the aforementioned two datasets. Identification of m6A-related DEGs and functional annotation From GSEA ( https://www.gsea-msigdb.org/gsea ), a total of 26 m6A-related genes were obtained, followed by the extraction of m6A-related gene expression data from the GSE56815 dataset. Utilizing the R package "limma", m6A-related DEGs between healthy controls and OP samples were identified with a screening criterion that included a Wilcoxon test and a p value < 0.05. The functional enrichment analysis of genes is frequently conducted using the widely used bioinformatics technique known as Gene Ontology (GO) enrichment analysis. Additionally, to comprehend biological mechanisms, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis has gained significant traction among researchers. To present the results of GO and KEGG pathway analyses, the application of the GOplot program package is commonly adopted. Screening diagnostic markers via machine learning algorithm The caret package-generated RF and SVM algorithm models underwent residual distribution analysis, which entailed their evaluation via the model performance function of the "DALEX" package in R. Employing box plots and reverse cumulative distribution plots enabled the visualization of the residual distribution. The RF analyses, on the other hand, employed the "randomForest" R packages. Development and verification of the diagnostic model The determination of risk coefficients for all diagnostic markers was achieved through the utilization of logistic regression.The Nomogram predicting the likelihood of OP was created via the "rms" toolkit in R. Statistical significance was established for distinctions in prospective diagnostic genes between OP and controls with a threshold of p < 0.05. The predictive efficacy of probable diagnostic genes in the validation set was evaluated using ROC analysis from the "pROC" package. GSEA of diagnostic markers To explore the potential roles of diagnostic markers in OP, GO and KEGG based on gene set enrichment analysis (GSEA) were conducted. Enriched functional categories and pathways were identified by utilizing the R package "clusterprofiler". The reference gene sets, c5.go.symbols.gmt and c2.cp.kegg.symbols.gmt were selected and a cut-off criteria of p adjusted value < 0.05 was applied. Evaluation and correlation analysis of immune cell infiltrating To investigate the degree of immune cell infiltration based on the expression profiling of 28 immunity-relevant signatures and gain a better understanding of infiltrating immune cells between OP and control groups, single-sample gene set enrichment analysis (ssGSEA) was carried out. The correlation between diagnostic markers and immune infiltrating cells was examined and visualized using the "limma", "reshape2", "tidyverse", and "ggplot2" packages. Subclusters analysis To identify the molecular subtypes of OP, the OP patients were partitioned into distinct subclusters with the aid of the R package "ConsensusClusterPlus". The DEGs of the two subtypes were subjected to GO enrichment analysis and Transcription factor targets enrichment analysis by utilizing the online analytical tool Metascape ( https://metascape.org/gp/index.html ). Statistical Analysis Data processing and analysis were conducted using R software (version 4.2.3). The Wilcoxon test was used to assess the disparities between independent variables and non-normally distributed variables. Spearman correlation analysis was performed to compute the correlation coefficients among various genes, the correlation between genes and immune cells, and the relationships between modules and traits. A p-value less than 0.05 was deemed to be statistically significant. Results A flow chart illustrating the study is presented Our analysis flow chart is shown in the figure above ( Fig. 1 ). Screening and functional analysis of m6A-related DEGs In GSE56815, four m6A-related genes, which exhibited differential expression between OP patients and controls, were detected ( Fig. 2 A,B ). GO functional enrichment analysis showed that the terms of biological function(BP) of m6A-related DEGs were mainly enriched in regulation of mRNA stability and RNA stability. The major enriched terms in the cellular component (CC) were nuclear speck and cytoplasmic stress granule. For molecular function (MF), terms of protein phosphatase activator activity and insulin − like growth factor binding were enriched ( Fig. 2 C ). Additionally, KEGG paythway enrichment analysis highly indicated that m6A-related DEGs were mainly enriched in terms of p53 signaling pathway and growth hormone synthesis, secretion and action ( Fig. 2 D ). Identification of diagnostic markers for OP by machine learning algorithm The residual boxplots and reverse cumulative distribution plots were analyzed to compare the performance of RF and SVM in screening diagnostic markers of OP. It was observed that the residual values of the models built using RF were found to be smaller than those obtained from SVM. These results indicate that the RF is better suited for screening diagnostic markers of OP ( Fig. 3 A,B ). Meanwhile, the AUC value of RF was shown to be 1 according to the ROC curve, while the AUC value of SVM was only 0.819, indicating that the diagnostic efficiency of RF was higher ( Fig. 3 C ). The 4 DEGs were initially filtered using a random forest algorithm. Subsequently, the optimal number of variables in a binary tree node (mtry) was determined, and the model's recurrent random forest classification was performed on all possible values of 1–4 variables. The model's average error rate was assessed to determine the optimal variable number parameter, which was determined to be 3 while keeping the out-of-band error and number of variables to a minimum. Furthermore, By employing 500 trees as the model's parameters, a correlation between the number of decision trees and the model's error has been established ( Fig. 3 D ). In order to ensure the reliability of our findings, we employed the Gini coefficient approach to assess the significance of each variable throughout the model building process, taking into account both accuracy and mean square error. Diagnostic markers for OP were identified by selecting four genes (YTHDF2, IGFBP3, FTO, and CBLL1) with an importance greater than 2 ( Fig. 3 E ). Establishment of a diagnostic markers-based nomogram for OP In order to enhance the efficiency of biomarkers in diagnosis, a novel diagnostic risk score was formulated by performing multivariate logistic regression analysis to determine the corresponding coefficient for each gene, which was then multiplied by its respective gene expression. The resulting diagnostic score is as follows: Z= (-3.514* expression level of CBLL1) + (-2.247 * expression level of YTHDF2) + (3.880 * expression level of IGFBP3) + (5.483 * expression level of FTO).Optimal diagnostic risk model = 1/(1 + e^-Z). A diagnostic tool for osteoporosis was developed using diagnostic markers, with the construction of a nomogram. Each diagnostic marker was assigned a corresponding score in the nomogram, and the total score was computed by summation of all scores. The total score corresponded to varying risks of OP ( Fig. 3 F ) . Evaluating the accuracy of nomograms for estimating the occurrence of OP using calibration curves showed that the model had high precision performance ( Fig. 3 G ). Nomogram could be beneficial clinically to patients diagnosed with OP, as demonstrated by decision curve analysis and clinical impact curve (CIC) analysis ( Fig. 3 H, I ). External independent dataset validate differential expression and diagnostic efficiency of diagnostic markers Differential expression analysis of diagnostic markers was conducted in the GSE35956 dataset, revealing three genes (YTHDF2, FTO, and CBLL1) that showed a consistent trend of expression in the osteoporotic group as compared to GSE56815 ( Fig. 4 A-D ). The ROC curve displayed a high diagnostic yield for OP using these markers, with AUC values of 0.72 for CBLL1, 0.82 for YTHDF2, 0.7 for IGFBP3, and 0.68 for FTO ( Fig. 4 E ). GSEA of diagnostic markers In the disease group FTO and IGFBP3 were highly expressed, whereas CBLL1 and YTHDF2 were lowly expressed. In OP, GO functions of the FTO high expression group are enriched in inner mitochondrial membrane protein complex ( Fig. 5 A ). KEGG pathways of FTO high expression group are mainly enriched in proteasome and proteasome ( Fig. 5 B ). GO functions of the IGFBP3 high expression group are enriched in ATP dependent activity acting on dna ( Fig. 5 C ). KEGG pathways of IGFBP3 high expression group are mainly enriched in mismatch repair and mismatch repair ( Fig. 5 D ). GO functions of the CBLL1 low expression group are enriched in external encapsulating structure and G-protein coupled receptor activity ( Fig. 5 E ). KEGG pathways of cbll1 low expression group are mainly enriched in neuroactive ligand receptor interaction and primary immunodeficiency ( Fig. 5 F ). GO functions of the YTHDF2 low expression group are enriched in gated channel activity and signaling receptor regulator activity (Fig. 5 G). KEGG pathways of YTHDF2 low expression group are mainly enriched in focal adhesion and focal adhesion ( Fig. 5 H ). Immunologic infiltration analysis of OP Through differential analysis of immune cells, high infiltration of CD56 dim natural killer cells was observed in osteoporotic patients ( Fig. 6 A, B ). Furthermore, correlation analysis revealed that immune cell infiltration is associated with three diagnostic markers, namely YTHDF2, IGFBP3, and FTO, indicating their involvement in the regulation of immune cell infiltration in OP ( Fig. 6 C ). Construction of two m6A subtypes of OP based on diagnostic markers By utilizing the consensus clustering approach, OP was clustered built on the expression profiling of four m6A-related diagnostic markers. Two distinct subtypes (named A and B) were identified ( Fig. 7 A,B,C ). Differential expression analysis demonstrated significant differences in three diagnostic markers between the two subtypes ( Fig. 7 D,E ). Remarkable differences were also observed between the subtypes using principal component analysis (PCA) ( Fig. 7 F ). Additionally, a significant disparity was discovered in the M6A score between the two subtypes ( Fig. 7 G ). Differential genes and functional enrichment analysis of two subtypes A heat map was generated to display the differential gene expression between two subtypes ( Fig. 8 A ). The results of the GO analysis of DEGs were presented with a focus on the regulation of hematopoietic stem cell differentiation, anatomical structure size regulation, and response to lithium ion ( Fig. 8 B ). The transcription factor targets enrichment analysis revealed that these genes were predominantly regulated by TATA 01 and NRSF 01 transcription factors ( Fig. 8 C ). Discussion Due to inadequate early diagnostic methods and tools, OP patients have been increasingly susceptible to developing pathological fractures resulting in the need for surgical intervention and internal fixation. These interventions incur substantial physical, emotional, and financial burdens on the patients[ 19 ]. In this study, differential expression analysis was initially conducted on m6A-related genes in GSE56815 dataset, resulting in identification of four statistically significant genes. Subsequently, RF and SVM residual values were compared to create a diagnostic model, which demonstrated that the RF algorithm yielded higher diagnostic ability. Following this, four diagnostic markers (YTHDF2, IGFBP3, FTO, and CBLL1) were selected via the RF algorithm. Ultimately, diagnostic risk assessment model for OP patients was successfully developed using the aforementioned diagnostic markers and validated in the independent dataset GSE35956, which showed that the model exhibited favorable diagnostic efficacy. The four diagnostic markers related to m6A have not been described in OP so far.. The recognition of m6A sites on target mRNAs and the regulation of their fate are performed by YTHDF2, which functions as an m6A 'reader' in this process [ 20 – 22 ]. The dynamic m6A modification, which affects the translation status and lifespan of mRNA, has been reported to be recognized by YTHDF2 [ 23 , 24 ]. In this study, a decreased expression level of YTHDF2 was observed among patients with osteoporosis, which led to a hypothesis that the reduced YTHDF2 expression might have contributed to destabilization of mRNA, thereby leading to abnormal cell function and accelerated development of osteoporosis. IGFBP3 is a member of the IGFBP family, which possesses high affinity binding ability to specific peptide growth factors (such as IGF-I and -II), resulting in the regulation of their biological activity, including metabolic activity, cellular differentiation, and proliferation, by preventing the activation of the IGF-I receptor[ 25 , 26 ]. The modulation of IGFBP3 mainly involves inducing apoptosis and/or inhibiting cellular proliferation, which has been thoroughly investigated. The effects are mediated through diverse effectors, including the transforming growth factor (TGF)-βV receptor, tumor necrosis factor (TNF)-α receptor, retinoid X receptor (RXR)-α, and nuclear factor-κB (NF-κB) cascades, as well as by IGF-independent attenuation of IGF-I's interaction with its receptor[ 27 , 28 ]. In the GSE56815 dataset, IGFBP3 exhibited significantly high expression levels in OP samples, which suggests that IGFBP3 may have involvement in apoptosis in patients with osteoporosis. The existing function of this gene supports this idea. However, the external validation set GSE35956 showed a tendency for under-expression in disease groups. This could be due to sampling error as a result of the small sample size of the GSE35956 dataset, necessitating further validation in osteoporotic samples. FTO, a well-known m6A demethylase, is recognized to have a critical role in m6A methylation[ 29 ]. FTO has been identified as the first demethylase of N6-methyladenosine (m6A) in RNA, which can oxidatively convert m6A to adenosine. This suggests that FTO may participate in lipid metabolism through m6A-mediated epigenetic regulation[ 30 ]. It has been suggested by earlier research that lipogenesis is modulated by m6A levels via FTO regulation[ 31 ]. The reciprocal differentiation of BMSCs gives rise to both osteogenic and adipogenic cells, indicating a closely interconnected relationship exists between the two processes[ 32 ]. It can be speculated that the indirect or direct impact of FTO on osteogenesis may contribute to the development of OP. Additionally, CBLL1, an E3 ubiquitin ligase, has been repeatedly suggested as a possible component of the m6A methyltransferase complex, although its specific function has yet to be definitively determined [ 33 ]. In summary, the importance of m6A-related genes as important contributors to the development of OP has been highlighted; with particular emphasis on the significance of the expression levels of YTHDF2, IGFBP3, FTO, and CBLL1. Two subtypes of OP were constructed based on m6A diagnostic markers, whereby type A had a higher m6A score in comparison to type B. The methylation pattern of m6A could be fully analyzed using the m6A score [ 34 ]. The m6A landscape of OP could be reflected by our clustering study, potentially facilitating the early diagnosis and intervention of the disease. It should be noted that some limitations existed in our study, although machine learning algorithms identified m6A diagnostic markers associated with OP and their diagnostic efficacy was confirmed in external datasets, further experimental validation through RT-qPCR, western blotting, and immunohistochemical analysis is necessary to comprehensively understand the role of these diagnostic markers and their potential regulatory mechanisms associated with OP. Conclusions Using machine learning methods, this study identified 4 m6A-related diagnostic markers significantly associated with OP. Additionally, a m6A-related genetic diagnostic prediction model for OP was developed, providing a necessary foundation for the exploration of molecular mechanisms underlying OP biomarkers and therapeutic target selection. However, larger biological studies are required to validate the diagnostic ability of these genes for OP before clinical application. Declarations Author contributions Shengzhong Luo and Jiahao Liu: Methodology, Visualization, Investigation, Writing - Original Draft, Writing – review & editing. Xinxin Miao: Methodology, Writing - Original Draft. Jianjian Deng: Methodology, Writing-Review and Editing. Tianlong Wu: Writing - Review and Editing. Xigao Cheng: conceptualization, writing - review and editing, supervision. All authors have read and agreed to the published version of the manuscript. All authors reviewed and approved the final version of the manuscript Funding The present study was supported by the National Natural Science Foundation of China (grant no. 82060403), the Thousand Talents Program of Jiangxi Province, China (grant no. JXSQ2019201026), and the Natural Science Foundation of Jiangxi Province of China (grant no.20224BAB216032). Data availability All data were available in the GEO database (https://www. ncbi.nlm.nih.gov /geo/query/acc.cgi?acc=GSE56815 and https://www.ncbi.nlm.nih.gov /geo/query/acc.cgi? acc=GSE35956 ). All the data analyzed and displayed in the present manuscript are available from the corresponding author upon reasonable request. Conflict of interest The authors declare that they have no conflicts of interest with the contents of this article. Acknowledgments The authors express their appreciation for the data provided by participants in GEO, including patients and researchers. References Morris JA, Kemp JP, Youlten SE, Laurent L, Logan JG, Chai RC, Vulpescu NA, Forgetta V, Kleinman A, Mohanty ST, Sergio CM, Quinn J, Nguyen-Yamamoto L, et al. An atlas of genetic influences on osteoporosis in humans and mice. Nat Genet. United States; 2019; 51: 258–66. Zhang W, Zhou X, Hou W, Chen E, Ye C, Chen M, Lu Q, Yu X, Li W. Reversing the imbalance in bone homeostasis via sustained release of SIRT-1 agonist to promote bone healing under osteoporotic condition. Bioact Mater. China; 2023; 19: 429–43. Hansen KE, Johnson RE, Chambers KR, Johnson MG, Lemon CC, Vo TNT, Marvdashti S. Treatment of Vitamin D Insufficiency in Postmenopausal Women: A Randomized Clinical Trial. JAMA Intern Med. United States; 2015; 175: 1612–21. Jin F, Li J, Zhang Y-B, Liu X, Cai M, Liu M, Li M, Ma C, Yue R, Zhu Y, Lai R, Wang Z, Ji X, et al. A functional motif of long noncoding RNA Nron against osteoporosis. Nat Commun. England; 2021; 12: 3319. Gowen M, Stroup GB, Dodds RA, James IE, Votta BJ, Smith BR, Bhatnagar PK, Lago AM, Callahan JF, DelMar EG, Miller MA, Nemeth EF, Fox J. Antagonizing the parathyroid calcium receptor stimulates parathyroid hormone secretion and bone formation in osteopenic rats. J Clin Invest. United States; 2000; 105: 1595–604. Wang Y, Deng P, Liu Y, Wu Y, Chen Y, Guo Y, Zhang S, Zheng X, Zhou L, Liu W, Li Q, Lin W, Qi X, et al. Alpha-ketoglutarate ameliorates age-related osteoporosis via regulating histone methylations. Nat Commun. England; 2020; 11: 5596. Di Iorgi N, Rosol M, Mittelman SD, Gilsanz V. Reciprocal relation between marrow adiposity and the amount of bone in the axial and appendicular skeleton of young adults. J Clin Endocrinol Metab. United States; 2008; 93: 2281–6. Yao X, Liu M, Jin F, Zhu Z. Comprehensive Analysis of Differentially Expressed Circular RNAs in Patients with Senile Osteoporotic Vertebral Compression Fracture. Biomed Res Int. United States; 2020; 2020: 4951251. Li T, Hu P-S, Zuo Z, Lin J-F, Li X, Wu Q-N, Chen Z-H, Zeng Z-L, Wang F, Zheng J, Chen D, Li B, Kang T-B, et al. METTL3 facilitates tumor progression via an m(6)A-IGF2BP2-dependent mechanism in colorectal carcinoma. Mol Cancer. England; 2019; 18: 112. Yu J, Chen M, Huang H, Zhu J, Song H, Zhu J, Park J, Ji S-J. Dynamic m6A modification regulates local translation of mRNA in axons. Nucleic Acids Res. England; 2018; 46: 1412–23. Yi D, Wang R, Shi X, Xu L, Yilihamu Y, Sang J. METTL14 promotes the migration and invasion of breast cancer cells by modulating N6‑methyladenosine and hsa‑miR‑146a‑5p expression. Oncol Rep. Greece; 2020; 43: 1375–86. Li D, Cai L, Meng R, Feng Z, Xu Q. METTL3 Modulates Osteoclast Differentiation and Function by Controlling RNA Stability and Nuclear Export. Int J Mol Sci. Switzerland; 2020; 21. Li Z, Wang P, Li J, Xie Z, Cen S, Li M, Liu W, Ye G, Zheng G, Ma M, Wang S, Yu W, Wu Y, et al. The N(6)-methyladenosine demethylase ALKBH5 negatively regulates the osteogenic differentiation of mesenchymal stem cells through PRMT6. Cell Death Dis. England; 2021; 12: 578. He M, Lei H, He X, Liu Y, Wang A, Ren Z, Liu X, Yan G, Wang W, Wang Y, Li G, Wang T, Pu J, et al. METTL14 Regulates Osteogenesis of Bone Marrow Mesenchymal Stem Cells via Inducing Autophagy Through m6A/IGF2BPs/Beclin-1 Signal Axis. Stem Cells Transl Med. England; 2022; 11: 987–1001. Wang W, Qiao S-C, Wu X-B, Sun B, Yang J-G, Li X, Zhang X, Qian S-J, Gu Y-X, Lai H-C. Circ_0008542 in osteoblast exosomes promotes osteoclast-induced bone resorption through m6A methylation. Cell Death Dis. England; 2021; 12: 628. Mansky KC. Aging, human immunodeficiency virus, and bone health. Clin Interv Aging. New Zealand; 2010; 5: 285–92. Yang W, Fidelis TT, Sun W-H. Machine Learning in Catalysis, From Proposal to Practicing. ACS Omega. United States; 2020; 5: 83–8. Zhang H, Hu J, Zhu J, Li Q, Fang L. Machine learning-based metabolism-related genes signature and immune infiltration landscape in diabetic nephropathy. Front Endocrinol (Lausanne). Switzerland; 2022; 13: 1026938. Wang X, Pei Z, Hao T, Ariben J, Li S, He W, Kong X, Chang J, Zhao Z, Zhang B. Prognostic analysis and validation of diagnostic marker genes in patients with osteoporosis. Front Immunol. Switzerland; 2022; 13: 987937. Nguyen LH, Robinton DA, Seligson MT, Wu L, Li L, Rakheja D, Comerford SA, Ramezani S, Sun X, Parikh MS, Yang EH, Powers JT, Shinoda G, et al. Lin28b is sufficient to drive liver cancer and necessary for its maintenance in murine models. Cancer Cell. United States; 2014; 26: 248–61. Wang X, He C. Reading RNA methylation codes through methyl-specific binding proteins. RNA Biol. United States; 2014; 11: 669–72. Huang H, Weng H, Sun W, Qin X, Shi H, Wu H, Zhao BS, Mesquita A, Liu C, Yuan CL, Hu Y-C, Hüttelmaier S, Skibbe JR, et al. Recognition of RNA N(6)-methyladenosine by IGF2BP proteins enhances mRNA stability and translation. Nat Cell Biol. England; 2018; 20: 285–95. Wang X, Zhao BS, Roundtree IA, Lu Z, Han D, Ma H, Weng X, Chen K, Shi H, He C. N(6)-methyladenosine Modulates Messenger RNA Translation Efficiency. Cell. United States; 2015; 161: 1388–99. Shi H, Wei J, He C. Where, When, and How: Context-Dependent Functions of RNA Methylation Writers, Readers, and Erasers. Mol Cell. United States; 2019; 74: 640–50. Firth SM, McDougall F, McLachlan AJ, Baxter RC. Impaired blockade of insulin-like growth factor I (IGF-I)-induced hypoglycemia by IGF binding protein-3 analog with reduced ternary complex-forming ability. Endocrinology. United States; 2002; 143: 1669–76. Baxter RC, Dai J. Purification and characterization of the acid-labile subunit of rat serum insulin-like growth factor binding protein complex. Endocrinology. United States; 1994; 134: 848–52. Jogie-Brahim S, Feldman D, Oh Y. Unraveling insulin-like growth factor binding protein-3 actions in human disease. Endocr Rev. United States; 2009; 30: 417–37. Yamada PM, Lee K-W. Perspectives in mammalian IGFBP-3 biology: local vs. systemic action. Am J Physiol Cell Physiol. United States; 2009; 296: C954-976. Yi D, Wang Q, Zhao Y, Song Y, You H, Wang J, Liu R, Shi Z, Chen X, Luo Q. Alteration of N (6) -Methyladenosine mRNA Methylation in a Rat Model of Cerebral Ischemia-Reperfusion Injury. Front Neurosci. Switzerland; 2021; 15: 605654. Sun D, Zhao T, Zhang Q, Wu M, Zhang Z. Fat mass and obesity-associated protein regulates lipogenesis via m(6) A modification in fatty acid synthase mRNA. Cell Biol Int. England; 2021; 45: 334–44. Zhao X, Yang Y, Sun B-F, Shi Y, Yang X, Xiao W, Hao Y-J, Ping X-L, Chen Y-S, Wang W-J, Jin K-X, Wang X, Huang C-M, et al. FTO-dependent demethylation of N6-methyladenosine regulates mRNA splicing and is required for adipogenesis. Cell Res. England; 2014; 24: 1403–19. Xu H, Yang Y-J, Yang T, Qian H-Y. Statins and stem cell modulation. Ageing Res Rev. England; 2013; 12: 1–7. Coker H, Wei G, Brockdorff N. m6A modification of non-coding RNA and the control of mammalian gene expression. Biochim Biophys Acta Gene Regul Mech. Netherlands; 2019; 1862: 310–8. Jiang F, Hu Y, Liu X, Wang M, Wu C. Methylation Pattern Mediated by m(6)A Regulator and Tumor Microenvironment Invasion in Lung Adenocarcinoma. Oxid Med Cell Longev. United States; 2022; 2022: 2930310. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted 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-4153446","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":290719082,"identity":"6557d788-1ff9-486f-85a7-c4346d0a0f65","order_by":0,"name":"Shengzhong Luo","email":"","orcid":"","institution":"Department of Orthopaedics, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, 330006, China","correspondingAuthor":false,"prefix":"","firstName":"Shengzhong","middleName":"","lastName":"Luo","suffix":""},{"id":290719083,"identity":"c90d97b5-a0a1-4ed5-970b-eb5ebccf3ebe","order_by":1,"name":"Jiahao Liu","email":"","orcid":"","institution":"Department of Orthopaedics, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, 330006, China","correspondingAuthor":false,"prefix":"","firstName":"Jiahao","middleName":"","lastName":"Liu","suffix":""},{"id":290719084,"identity":"039790cb-e41a-4fc6-9f53-fe9bd14e8a76","order_by":2,"name":"Xinxin Miao","email":"","orcid":"","institution":"Department of Orthopaedics, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, 330006, China","correspondingAuthor":false,"prefix":"","firstName":"Xinxin","middleName":"","lastName":"Miao","suffix":""},{"id":290719085,"identity":"57a6e099-4269-40a0-94cd-9447e8ce815a","order_by":3,"name":"Jianjian Deng","email":"","orcid":"","institution":"Department of Orthopaedics, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, 330006, China","correspondingAuthor":false,"prefix":"","firstName":"Jianjian","middleName":"","lastName":"Deng","suffix":""},{"id":290719086,"identity":"fd85c315-aaec-40fd-a3e9-cbe0241d7781","order_by":4,"name":"Tianlong Wu","email":"","orcid":"","institution":"Department of Orthopaedics, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, 330006, China","correspondingAuthor":false,"prefix":"","firstName":"Tianlong","middleName":"","lastName":"Wu","suffix":""},{"id":290719087,"identity":"9f1ff744-aa42-479c-a1ed-00cce3792ad7","order_by":5,"name":"Xigao Cheng","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+UlEQVRIiWNgGAWjYDACZjB5AMQ6BmaysROvhS2NgSEBSDETZxdIC48ZWAsDIS0Gx9kvfvhQc0fOnH/Ntwcff2yT52NmYPzwMQe3FslmnmLJGceeGVvOeLvdcEbCbcM2ZgZmyZnbcGvhZ+ZJkOZtOJy44cbZbdI8CbcZgVrYmHnxaGFj5kn+DdRSv+HGmWcgLfYEtfAzsx8D2ZJgcL6HDaQlkaAWoF/YLGccO2y44QabmeSMtNvJbcyMzXj9YnD++OMbH2oOyxucP/xM4oPNbdv57c0HP3zEowUYHQYQWiIBJsLYgE89ELA/gPrqAAGFo2AUjIJRMGIBAD37UUIkAzvEAAAAAElFTkSuQmCC","orcid":"","institution":"Department of Orthopaedics, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, 330006, China","correspondingAuthor":true,"prefix":"","firstName":"Xigao","middleName":"","lastName":"Cheng","suffix":""}],"badges":[],"createdAt":"2024-03-23 08:29:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4153446/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4153446/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":55058748,"identity":"515327ea-db62-4761-9af2-099e2140daca","added_by":"auto","created_at":"2024-04-22 02:04:22","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":278341,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eA flow chart illustrating the study is presented.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure.1.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/70f2fd04147a8c09647ecac9.png"},{"id":55058749,"identity":"73b2dcf2-446c-4009-bf5d-39ccca26cfac","added_by":"auto","created_at":"2024-04-22 02:04:22","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1235885,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifferential expression and enrichment analysis of m6A-related genes in GSE56815.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Box plots of differential expression of m6A-related genes in GSE56815. (B) Heat map showing four differentially expressed m6A-related genes in GSE56815. (C) GO functional enrichment analysis of differential genes. (D) KEGG pathway enrichment analysis of differential genes.\u003c/p\u003e","description":"","filename":"figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/f69e4b84c92b7c5623fcf52b.png"},{"id":55057721,"identity":"9bbde95d-90ce-4037-8585-924d148f3ef7","added_by":"auto","created_at":"2024-04-22 01:56:22","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":819160,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentify diagnostic markers by machine learning and build diagnostic model.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Boxplots of RF and SVM residuals. (B) Reverse cumulative distribution plots of RF and SVM residuals. (C) ROC curves demonstrate the diagnostic yield of RF and SVM. (D) The effect of the decision tree number on the error rate. On the x-axis, there is a depiction of the quantity of decision trees, whereas the error rate is shown on the y-axis. (E) The results of Gini coefficient method in a random forest classifier. The genetic variable is displayed on the x-axis, while the significance index is shown on the y-axis .(F) Establishment of a nomogram integrating diagnostic markers for predicting occurrence of OP.(G) The prediction accuracy of the nomogram can be assessed through the utilization of a calibration curve .(H) Clinical impact curve for this diagnostic model. (I) The clinical benefits of the nomogram are demonstrated through the utilization of decision curve analysis.\u003c/p\u003e","description":"","filename":"figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/a7f40a310cbbbe54c187b1f4.png"},{"id":55058747,"identity":"dfd50ded-efbd-42e9-b708-5d58fa48980d","added_by":"auto","created_at":"2024-04-22 02:04:22","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":235008,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVerification of diagnostic markers.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A-D)The exprssion level of diagnostic markers in GSE35956. (E) ROC curve calculates the diagnostic efficiency for each diagnostic gene.\u003c/p\u003e","description":"","filename":"figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/18922c8609d3ed1c8d18d7a2.png"},{"id":55057726,"identity":"a2a2d0f8-28a2-4691-baf5-44ff7d258039","added_by":"auto","created_at":"2024-04-22 01:56:22","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1731674,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFunctional and pathway enrichment analysis of diagnostic markers was based on GSEA.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) GO analysis of FTO based on GSEA. (B) KEGG analysis of FTO based on GSEA. (C) GO analysis of CBLL1 based on GSEA. (D) KEGG of CBLL1 based on GSEA. (E) GO analysis of IGFBP3 based on GSEA. (F) KEGG of IGFBP3 based on GSEA. (G) GO analysis of YTHDF2 based on GSEA. (H) KEGG analysis of YTHDF2 based on GSEA.\u003c/p\u003e","description":"","filename":"figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/5711228d5eb9c7ab0994a902.png"},{"id":55057723,"identity":"7e614971-1531-411d-b5e0-2373f51df565","added_by":"auto","created_at":"2024-04-22 01:56:22","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1421007,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eImmune cell infiltration analysis in GSE56815.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Analysis of differences in immune cells infiltration between OP patients and normal controls in GSE56815dataset. (B) A contrast of immune cell penetration between normal controls and OP patients was visualized using a heat map. (C) Diagnostic markers were investigated for their correlation with immune cell infiltration in a correlation analysis.\u003c/p\u003e","description":"","filename":"figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/d73f64a3bedd9100060f2bae.png"},{"id":55057729,"identity":"96d236b5-c143-49dd-bc9b-cc4281e59292","added_by":"auto","created_at":"2024-04-22 01:56:23","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":755791,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEstablishment of two m6A subtypes of OP based on diagnostic markers.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) The consensus matrix heatmap was generated when k equaled 2(B) Consensus CDF when k = 2–9. (C) The area under CDF curve experienced relative alterations.(D) Box plots of differential expression of diagnostic markers in two m6A subtypes of OP. (E) The differential expression of diagnostic markers for OP in two m6A subtypes is depicted by the heat map. (F) PCA analysis of two m6A isoforms. (G) Differential analysis of m6Ascore between two m6A subtypes.\u003c/p\u003e","description":"","filename":"figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/5dddcf7b0092c54617ea01d0.png"},{"id":55057727,"identity":"24eaeeaf-943c-4402-b7e5-0e978fa8b116","added_by":"auto","created_at":"2024-04-22 01:56:22","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":593924,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifferential expression analysis and enrichment analysis of two m6A subtypes.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Heat map demonstrating DEGS between two m6A subtypes. (B) GO functional enrichment analysis of DEGs.(C) Transcription factor targets enrichment analysis of DEGs.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/3e833835c89c06ab3e988171.png"},{"id":70258156,"identity":"53c67c1d-5f12-470f-870c-e3715d14195b","added_by":"auto","created_at":"2024-11-30 17:46:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6709439,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4153446/v1/74b12e5e-e44f-43a5-81f3-cbf3ec3dc5ee.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Identification of m6A-associated diagnostic biomarkers and subtypes for osteoporosis's disease diagnosis and risk prediction based on machine learning","fulltext":[{"header":"Introdution","content":"\u003cp\u003eOsteoporosis (OP) is a prevalent degenerative disease associated with aging, symbolic of diminished bone density that leads to an elevated risk of fractures. A highly heritable trait, bone mineral density (BMD), represents the utmost clinically significant risk factor for fracture diagnosis in OP [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Each year, a total of 8.9\u0026nbsp;million osteoporotic fractures are incurred globally [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Among postmenopausal women, osteoporotic fractures are suffered by around half [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. The health of elderly individuals is adversely impacted by osteoporotic fractures [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. After hip fractures, a significant correlation exists between OP and notable morbidity and mortality rates [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].As a result of the accelerated aging of the global population, an escalation of OP rates has been observed, posing a considerable challenge to public healthcare [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Based on its etiology, OP can be classified into two categories: primary OP, which comprises of postmenopausal, senile, and juvenile OP, and secondary OP, which stems mainly from prolonged drug usage [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. At an early stage, the diagnosis of osteoporosis remains a challenge despite improvements in therapeutic strategies, and the prognosis is poor [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Hence, there is an imperative necessity to enhance our comprehension of the pathogenesis of OP, with the aim of identifying innovative biomarkers and formulating novel therapeutic mechanisms for both the prevention and treatment of this disease.\u003c/p\u003e \u003cp\u003em6A, a reversible and dynamic modification of RNA, is governed by m6A WERs, commonly known as \"writers\" ,\"erasers\" ,and \"readers\"[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. The participation of m6A methylation in RNA biology in terms of mRNA splicing, translation, and decay is apparent through an expanding list of components such as m6A readers, m6A writer-complex, and erasers[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. The implication of m6A in the onset of diverse human diseases has been documented[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. It has been reported that m6A, an RNA epigenetic modification, plays a crucial role in the regulation of osteogenic differentiation[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. For example, ALKBH5 has been demonstrated to regulate osteogenic differentiation of MSCs by means of m6A demethylation activity[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. The osteogenesis of BMSCs is reported to be regulated by METTL14 through the induction of autophagy via the m6A/IGF2BPs/Beclin-1 signal axis[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. The effect of m6A methylation on osteoclasts has also been observed, with previous studies demonstrating that osteoblast exosomal Circ_0008542 promotes bone resorption induced by osteoclasts through m6A methylation[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Bone density loss can be clinically manifested as osteopenia or OP due to the dysfunction of either osteoclasts or osteoblasts, which results in perturbation[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Therefore, the m6A methylation modification is anticipated to significantly impact the pathogenesis of OP, and an investigation into the association of m6A with OP would aid in comprehending the pathophysiological mechanisms underlying its development, thereby providing clinical utility.\u003c/p\u003e \u003cp\u003eThe popularity of machine learning (ML) has experienced a rapid growth Over the last several years, as evidenced by its increasing usage[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Utilization of various machine learning algorithms was employed to Discriminate diagnostic feature biomarkers[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. In this study, transcriptome sequencing data of OP blood samples was collected from the GEO database. Machine learning algorithms were utilized to select m6A-related diagnostic markers of OP. Subsequently, a diagnostic predictive model was established based on four selected diagnostic markers. Additionally, OP was clustered, which laid the groundwork for early OP diagnosis and demonstrated practical applicability.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDatasets and data pre-processing\u003c/h2\u003e \u003cp\u003eThe RNA-seq profiles were searched in the GEO database using the keywords \"osteoporosis\" to retrieve raw gene expression profiling data. Two datasets, namely GSE56815 and GSE35956, were obtained from the search. The former dataset consisted of 40 people with low hip bone mineral density (BMD) and 40 subjects with high hip BMD. The latter dataset had 5 patients with osteoporosis (OP) and 5 normal samples, which were utilised as external validation datasets. The R package \"limma\" was utilised to do background correction and data normalisation for the aforementioned two datasets.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of m6A-related DEGs and functional annotation\u003c/h2\u003e \u003cp\u003eFrom GSEA (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org/gsea\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org/gsea\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), a total of 26 m6A-related genes were obtained, followed by the extraction of m6A-related gene expression data from the GSE56815 dataset. Utilizing the R package \"limma\", m6A-related DEGs between healthy controls and OP samples were identified with a screening criterion that included a Wilcoxon test and a p value\u0026thinsp;\u0026lt;\u0026thinsp;0.05. The functional enrichment analysis of genes is frequently conducted using the widely used bioinformatics technique known as Gene Ontology (GO) enrichment analysis. Additionally, to comprehend biological mechanisms, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis has gained significant traction among researchers. To present the results of GO and KEGG pathway analyses, the application of the GOplot program package is commonly adopted.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eScreening diagnostic markers via machine learning algorithm\u003c/h2\u003e \u003cp\u003eThe caret package-generated RF and SVM algorithm models underwent residual distribution analysis, which entailed their evaluation via the model performance function of the \"DALEX\" package in R. Employing box plots and reverse cumulative distribution plots enabled the visualization of the residual distribution. The RF analyses, on the other hand, employed the \"randomForest\" R packages.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment and verification of the diagnostic model\u003c/h2\u003e \u003cp\u003eThe determination of risk coefficients for all diagnostic markers was achieved through the utilization of logistic regression.The Nomogram predicting the likelihood of OP was created via the \"rms\" toolkit in R. Statistical significance was established for distinctions in prospective diagnostic genes between OP and controls with a threshold of p\u0026thinsp;\u0026lt;\u0026thinsp;0.05. The predictive efficacy of probable diagnostic genes in the validation set was evaluated using ROC analysis from the \"pROC\" package.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eGSEA of diagnostic markers\u003c/h2\u003e \u003cp\u003eTo explore the potential roles of diagnostic markers in OP, GO and KEGG based on gene set enrichment analysis (GSEA) were conducted. Enriched functional categories and pathways were identified by utilizing the R package \"clusterprofiler\". The reference gene sets, c5.go.symbols.gmt and c2.cp.kegg.symbols.gmt were selected and a cut-off criteria of p adjusted value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was applied.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eEvaluation and correlation analysis of immune cell infiltrating\u003c/h2\u003e \u003cp\u003eTo investigate the degree of immune cell infiltration based on the expression profiling of 28 immunity-relevant signatures and gain a better understanding of infiltrating immune cells between OP and control groups, single-sample gene set enrichment analysis (ssGSEA) was carried out. The correlation between diagnostic markers and immune infiltrating cells was examined and visualized using the \"limma\", \"reshape2\", \"tidyverse\", and \"ggplot2\" packages.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eSubclusters analysis\u003c/h2\u003e \u003cp\u003eTo identify the molecular subtypes of OP, the OP patients were partitioned into distinct subclusters with the aid of the R package \"ConsensusClusterPlus\". The DEGs of the two subtypes were subjected to GO enrichment analysis and Transcription factor targets enrichment analysis by utilizing the online analytical tool Metascape (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://metascape.org/gp/index.html\u003c/span\u003e\u003cspan address=\"https://metascape.org/gp/index.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eData processing and analysis were conducted using R software (version 4.2.3). The Wilcoxon test was used to assess the disparities between independent variables and non-normally distributed variables. Spearman correlation analysis was performed to compute the correlation coefficients among various genes, the correlation between genes and immune cells, and the relationships between modules and traits. A p-value less than 0.05 was deemed to be statistically significant.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eA flow chart illustrating the study is presented\u003c/h2\u003e \u003cp\u003eOur analysis flow chart is shown in the figure above\u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eScreening and functional analysis of m6A-related DEGs\u003c/h2\u003e \u003cp\u003eIn GSE56815, four m6A-related genes, which exhibited differential expression between OP patients and controls, were detected\u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA,B\u003cb\u003e).\u003c/b\u003e GO functional enrichment analysis showed that the terms of biological function(BP) of m6A-related DEGs were mainly enriched in regulation of mRNA stability and RNA stability. The major enriched terms in the cellular component (CC) were nuclear speck and cytoplasmic stress granule. For molecular function (MF), terms of protein phosphatase activator activity and insulin\u0026thinsp;\u0026minus;\u0026thinsp;like growth factor binding were enriched\u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC\u003cb\u003e).\u003c/b\u003e Additionally, KEGG paythway enrichment analysis highly indicated that m6A-related DEGs were mainly enriched in terms of p53 signaling pathway and growth hormone synthesis, secretion and action \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of diagnostic markers for OP by machine learning algorithm\u003c/h2\u003e \u003cp\u003eThe residual boxplots and reverse cumulative distribution plots were analyzed to compare the performance of RF and SVM in screening diagnostic markers of OP. It was observed that the residual values of the models built using RF were found to be smaller than those obtained from SVM. These results indicate that the RF is better suited for screening diagnostic markers of OP \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA,B\u003cb\u003e).\u003c/b\u003e Meanwhile, the AUC value of RF was shown to be 1 according to the ROC curve, while the AUC value of SVM was only 0.819, indicating that the diagnostic efficiency of RF was higher \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC\u003cb\u003e).\u003c/b\u003e The 4 DEGs were initially filtered using a random forest algorithm. Subsequently, the optimal number of variables in a binary tree node (mtry) was determined, and the model's recurrent random forest classification was performed on all possible values of 1\u0026ndash;4 variables. The model's average error rate was assessed to determine the optimal variable number parameter, which was determined to be 3 while keeping the out-of-band error and number of variables to a minimum. Furthermore, By employing 500 trees as the model's parameters, a correlation between the number of decision trees and the model's error has been established \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD\u003cb\u003e).\u003c/b\u003e In order to ensure the reliability of our findings, we employed the Gini coefficient approach to assess the significance of each variable throughout the model building process, taking into account both accuracy and mean square error. Diagnostic markers for OP were identified by selecting four genes (YTHDF2, IGFBP3, FTO, and CBLL1) with an importance greater than 2\u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eEstablishment of a diagnostic markers-based nomogram for OP\u003c/h2\u003e \u003cp\u003eIn order to enhance the efficiency of biomarkers in diagnosis, a novel diagnostic risk score was formulated by performing multivariate logistic regression analysis to determine the corresponding coefficient for each gene, which was then multiplied by its respective gene expression. The resulting diagnostic score is as follows: Z= (-3.514* expression level of CBLL1) + (-2.247 * expression level of YTHDF2) + (3.880 * expression level of IGFBP3) + (5.483 * expression level of FTO).Optimal diagnostic risk model\u0026thinsp;=\u0026thinsp;1/(1\u0026thinsp;+\u0026thinsp;e^-Z). A diagnostic tool for osteoporosis was developed using diagnostic markers, with the construction of a nomogram. Each diagnostic marker was assigned a corresponding score in the nomogram, and the total score was computed by summation of all scores. The total score corresponded to varying risks of OP \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF\u003cb\u003e)\u003c/b\u003e. Evaluating the accuracy of nomograms for estimating the occurrence of OP using calibration curves showed that the model had high precision performance \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eG\u003cb\u003e).\u003c/b\u003e Nomogram could be beneficial clinically to patients diagnosed with OP, as demonstrated by decision curve analysis and clinical impact curve (CIC) analysis \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eH, I\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eExternal independent dataset validate differential expression and diagnostic efficiency of diagnostic markers\u003c/h2\u003e \u003cp\u003eDifferential expression analysis of diagnostic markers was conducted in the GSE35956 dataset, revealing three genes (YTHDF2, FTO, and CBLL1) that showed a consistent trend of expression in the osteoporotic group as compared to GSE56815 \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA-D\u003cb\u003e).\u003c/b\u003e The ROC curve displayed a high diagnostic yield for OP using these markers, with AUC values of 0.72 for CBLL1, 0.82 for YTHDF2, 0.7 for IGFBP3, and 0.68 for FTO \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eE\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eGSEA of diagnostic markers\u003c/h2\u003e \u003cp\u003eIn the disease group FTO and IGFBP3 were highly expressed, whereas CBLL1 and YTHDF2 were lowly expressed. In OP, GO functions of the FTO high expression group are enriched in inner mitochondrial membrane protein complex \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA\u003cb\u003e).\u003c/b\u003e KEGG pathways of FTO high expression group are mainly enriched in proteasome and proteasome \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB\u003cb\u003e).\u003c/b\u003e GO functions of the IGFBP3 high expression group are enriched in ATP dependent activity acting on dna \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC\u003cb\u003e).\u003c/b\u003e KEGG pathways of IGFBP3 high expression group are mainly enriched in mismatch repair and mismatch repair \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD\u003cb\u003e).\u003c/b\u003e GO functions of the CBLL1 low expression group are enriched in external encapsulating structure and G-protein coupled receptor activity \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE\u003cb\u003e).\u003c/b\u003e KEGG pathways of cbll1 low expression group are mainly enriched in neuroactive ligand receptor interaction and primary immunodeficiency \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF\u003cb\u003e).\u003c/b\u003e GO functions of the YTHDF2 low expression group are enriched in gated channel activity and signaling receptor regulator activity (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eG). KEGG pathways of YTHDF2 low expression group are mainly enriched in focal adhesion and focal adhesion \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eH\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eImmunologic infiltration analysis of OP\u003c/h2\u003e \u003cp\u003eThrough differential analysis of immune cells, high infiltration of CD56 dim natural killer cells was observed in osteoporotic patients \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA, B\u003cb\u003e).\u003c/b\u003e Furthermore, correlation analysis revealed that immune cell infiltration is associated with three diagnostic markers, namely YTHDF2, IGFBP3, and FTO, indicating their involvement in the regulation of immune cell infiltration in OP \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eConstruction of two m6A subtypes of OP based on diagnostic markers\u003c/h2\u003e \u003cp\u003eBy utilizing the consensus clustering approach, OP was clustered built on the expression profiling of four m6A-related diagnostic markers. Two distinct subtypes (named A and B) were identified \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA,B,C\u003cb\u003e).\u003c/b\u003e Differential expression analysis demonstrated significant differences in three diagnostic markers between the two subtypes \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eD,E\u003cb\u003e).\u003c/b\u003e Remarkable differences were also observed between the subtypes using principal component analysis (PCA) \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eF\u003cb\u003e).\u003c/b\u003e Additionally, a significant disparity was discovered in the M6A score between the two subtypes \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eG\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eDifferential genes and functional enrichment analysis of two subtypes\u003c/h2\u003e \u003cp\u003eA heat map was generated to display the differential gene expression between two subtypes\u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA\u003cb\u003e).\u003c/b\u003e The results of the GO analysis of DEGs were presented with a focus on the regulation of hematopoietic stem cell differentiation, anatomical structure size regulation, and response to lithium ion \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB\u003cb\u003e).\u003c/b\u003e The transcription factor targets enrichment analysis revealed that these genes were predominantly regulated by TATA 01 and NRSF 01 transcription factors \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eC\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eDue to inadequate early diagnostic methods and tools, OP patients have been increasingly susceptible to developing pathological fractures resulting in the need for surgical intervention and internal fixation. These interventions incur substantial physical, emotional, and financial burdens on the patients[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. In this study, differential expression analysis was initially conducted on m6A-related genes in GSE56815 dataset, resulting in identification of four statistically significant genes. Subsequently, RF and SVM residual values were compared to create a diagnostic model, which demonstrated that the RF algorithm yielded higher diagnostic ability. Following this, four diagnostic markers (YTHDF2, IGFBP3, FTO, and CBLL1) were selected via the RF algorithm. Ultimately, diagnostic risk assessment model for OP patients was successfully developed using the aforementioned diagnostic markers and validated in the independent dataset GSE35956, which showed that the model exhibited favorable diagnostic efficacy.\u003c/p\u003e \u003cp\u003eThe four diagnostic markers related to m6A have not been described in OP so far.. The recognition of m6A sites on target mRNAs and the regulation of their fate are performed by YTHDF2, which functions as an m6A 'reader' in this process [\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. The dynamic m6A modification, which affects the translation status and lifespan of mRNA, has been reported to be recognized by YTHDF2 [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. In this study, a decreased expression level of YTHDF2 was observed among patients with osteoporosis, which led to a hypothesis that the reduced YTHDF2 expression might have contributed to destabilization of mRNA, thereby leading to abnormal cell function and accelerated development of osteoporosis. IGFBP3 is a member of the IGFBP family, which possesses high affinity binding ability to specific peptide growth factors (such as IGF-I and -II), resulting in the regulation of their biological activity, including metabolic activity, cellular differentiation, and proliferation, by preventing the activation of the IGF-I receptor[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe modulation of IGFBP3 mainly involves inducing apoptosis and/or inhibiting cellular proliferation, which has been thoroughly investigated. The effects are mediated through diverse effectors, including the transforming growth factor (TGF)-βV receptor, tumor necrosis factor (TNF)-α receptor, retinoid X receptor (RXR)-α, and nuclear factor-κB (NF-κB) cascades, as well as by IGF-independent attenuation of IGF-I's interaction with its receptor[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. In the GSE56815 dataset, IGFBP3 exhibited significantly high expression levels in OP samples, which suggests that IGFBP3 may have involvement in apoptosis in patients with osteoporosis. The existing function of this gene supports this idea. However, the external validation set GSE35956 showed a tendency for under-expression in disease groups. This could be due to sampling error as a result of the small sample size of the GSE35956 dataset, necessitating further validation in osteoporotic samples. FTO, a well-known m6A demethylase, is recognized to have a critical role in m6A methylation[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. FTO has been identified as the first demethylase of N6-methyladenosine (m6A) in RNA, which can oxidatively convert m6A to adenosine. This suggests that FTO may participate in lipid metabolism through m6A-mediated epigenetic regulation[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. It has been suggested by earlier research that lipogenesis is modulated by m6A levels via FTO regulation[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. The reciprocal differentiation of BMSCs gives rise to both osteogenic and adipogenic cells, indicating a closely interconnected relationship exists between the two processes[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. It can be speculated that the indirect or direct impact of FTO on osteogenesis may contribute to the development of OP. Additionally, CBLL1, an E3 ubiquitin ligase, has been repeatedly suggested as a possible component of the m6A methyltransferase complex, although its specific function has yet to be definitively determined [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. In summary, the importance of m6A-related genes as important contributors to the development of OP has been highlighted; with particular emphasis on the significance of the expression levels of YTHDF2, IGFBP3, FTO, and CBLL1.\u003c/p\u003e \u003cp\u003eTwo subtypes of OP were constructed based on m6A diagnostic markers, whereby type A had a higher m6A score in comparison to type B. The methylation pattern of m6A could be fully analyzed using the m6A score [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. The m6A landscape of OP could be reflected by our clustering study, potentially facilitating the early diagnosis and intervention of the disease.\u003c/p\u003e \u003cp\u003eIt should be noted that some limitations existed in our study, although machine learning algorithms identified m6A diagnostic markers associated with OP and their diagnostic efficacy was confirmed in external datasets, further experimental validation through RT-qPCR, western blotting, and immunohistochemical analysis is necessary to comprehensively understand the role of these diagnostic markers and their potential regulatory mechanisms associated with OP.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eUsing machine learning methods, this study identified 4 m6A-related diagnostic markers significantly associated with OP. Additionally, a m6A-related genetic diagnostic prediction model for OP was developed, providing a necessary foundation for the exploration of molecular mechanisms underlying OP biomarkers and therapeutic target selection. However, larger biological studies are required to validate the diagnostic ability of these genes for OP before clinical application.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eShengzhong Luo and Jiahao Liu: Methodology, Visualization, Investigation, Writing - Original Draft, Writing \u0026ndash; review \u0026amp; editing. Xinxin Miao: Methodology, Writing - Original Draft. Jianjian Deng: Methodology, Writing-Review and Editing. Tianlong Wu: Writing - Review and Editing. Xigao Cheng: conceptualization, writing - review and editing, supervision. All authors have read and agreed to the published version of the manuscript. All authors reviewed and approved the final version of the manuscript\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe present study was supported by the National Natural Science Foundation of China (grant no. 82060403), the Thousand Talents Program of Jiangxi Province, China (grant no. JXSQ2019201026), and\u0026nbsp;the Natural Science Foundation of Jiangxi Province of China (grant no.20224BAB216032).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data were available in the GEO database (https://www. ncbi.nlm.nih.gov /geo/query/acc.cgi?acc=GSE56815 and https://www.ncbi.nlm.nih.gov /geo/query/acc.cgi? acc=GSE35956 ). All the data analyzed and displayed in the present manuscript are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of interest\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no conflicts of interest with the contents of this article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors express their appreciation for the data provided by participants in GEO, including patients and researchers.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMorris JA, Kemp JP, Youlten SE, Laurent L, Logan JG, Chai RC, Vulpescu NA, Forgetta V, Kleinman A, Mohanty ST, Sergio CM, Quinn J, Nguyen-Yamamoto L, et al. An atlas of genetic influences on osteoporosis in humans and mice. Nat Genet. United States; 2019; 51: 258\u0026ndash;66. \u003c/li\u003e\n\u003cli\u003eZhang W, Zhou X, Hou W, Chen E, Ye C, Chen M, Lu Q, Yu X, Li W. Reversing the imbalance in bone homeostasis via sustained release of SIRT-1 agonist to promote bone healing under osteoporotic condition. Bioact Mater. China; 2023; 19: 429\u0026ndash;43. \u003c/li\u003e\n\u003cli\u003eHansen KE, Johnson RE, Chambers KR, Johnson MG, Lemon CC, Vo TNT, Marvdashti S. Treatment of Vitamin D Insufficiency in Postmenopausal Women: A Randomized Clinical Trial. JAMA Intern Med. United States; 2015; 175: 1612\u0026ndash;21. \u003c/li\u003e\n\u003cli\u003eJin F, Li J, Zhang Y-B, Liu X, Cai M, Liu M, Li M, Ma C, Yue R, Zhu Y, Lai R, Wang Z, Ji X, et al. A functional motif of long noncoding RNA Nron against osteoporosis. Nat Commun. England; 2021; 12: 3319. \u003c/li\u003e\n\u003cli\u003eGowen M, Stroup GB, Dodds RA, James IE, Votta BJ, Smith BR, Bhatnagar PK, Lago AM, Callahan JF, DelMar EG, Miller MA, Nemeth EF, Fox J. Antagonizing the parathyroid calcium receptor stimulates parathyroid hormone secretion and bone formation in osteopenic rats. J Clin Invest. United States; 2000; 105: 1595\u0026ndash;604. \u003c/li\u003e\n\u003cli\u003eWang Y, Deng P, Liu Y, Wu Y, Chen Y, Guo Y, Zhang S, Zheng X, Zhou L, Liu W, Li Q, Lin W, Qi X, et al. Alpha-ketoglutarate ameliorates age-related osteoporosis via regulating histone methylations. Nat Commun. England; 2020; 11: 5596. \u003c/li\u003e\n\u003cli\u003eDi Iorgi N, Rosol M, Mittelman SD, Gilsanz V. Reciprocal relation between marrow adiposity and the amount of bone in the axial and appendicular skeleton of young adults. J Clin Endocrinol Metab. United States; 2008; 93: 2281\u0026ndash;6. \u003c/li\u003e\n\u003cli\u003eYao X, Liu M, Jin F, Zhu Z. Comprehensive Analysis of Differentially Expressed Circular RNAs in Patients with Senile Osteoporotic Vertebral Compression Fracture. Biomed Res Int. United States; 2020; 2020: 4951251. \u003c/li\u003e\n\u003cli\u003eLi T, Hu P-S, Zuo Z, Lin J-F, Li X, Wu Q-N, Chen Z-H, Zeng Z-L, Wang F, Zheng J, Chen D, Li B, Kang T-B, et al. METTL3 facilitates tumor progression via an m(6)A-IGF2BP2-dependent mechanism in colorectal carcinoma. Mol Cancer. England; 2019; 18: 112. \u003c/li\u003e\n\u003cli\u003eYu J, Chen M, Huang H, Zhu J, Song H, Zhu J, Park J, Ji S-J. Dynamic m6A modification regulates local translation of mRNA in axons. Nucleic Acids Res. England; 2018; 46: 1412\u0026ndash;23. \u003c/li\u003e\n\u003cli\u003eYi D, Wang R, Shi X, Xu L, Yilihamu Y, Sang J. METTL14 promotes the migration and invasion of breast cancer cells by modulating N6‑methyladenosine and hsa‑miR‑146a‑5p expression. Oncol Rep. Greece; 2020; 43: 1375\u0026ndash;86. \u003c/li\u003e\n\u003cli\u003eLi D, Cai L, Meng R, Feng Z, Xu Q. METTL3 Modulates Osteoclast Differentiation and Function by Controlling RNA Stability and Nuclear Export. Int J Mol Sci. Switzerland; 2020; 21. \u003c/li\u003e\n\u003cli\u003eLi Z, Wang P, Li J, Xie Z, Cen S, Li M, Liu W, Ye G, Zheng G, Ma M, Wang S, Yu W, Wu Y, et al. The N(6)-methyladenosine demethylase ALKBH5 negatively regulates the osteogenic differentiation of mesenchymal stem cells through PRMT6. Cell Death Dis. England; 2021; 12: 578. \u003c/li\u003e\n\u003cli\u003eHe M, Lei H, He X, Liu Y, Wang A, Ren Z, Liu X, Yan G, Wang W, Wang Y, Li G, Wang T, Pu J, et al. METTL14 Regulates Osteogenesis of Bone Marrow Mesenchymal Stem Cells via Inducing Autophagy Through m6A/IGF2BPs/Beclin-1 Signal Axis. Stem Cells Transl Med. England; 2022; 11: 987\u0026ndash;1001. \u003c/li\u003e\n\u003cli\u003eWang W, Qiao S-C, Wu X-B, Sun B, Yang J-G, Li X, Zhang X, Qian S-J, Gu Y-X, Lai H-C. Circ_0008542 in osteoblast exosomes promotes osteoclast-induced bone resorption through m6A methylation. Cell Death Dis. England; 2021; 12: 628. \u003c/li\u003e\n\u003cli\u003eMansky KC. Aging, human immunodeficiency virus, and bone health. Clin Interv Aging. New Zealand; 2010; 5: 285\u0026ndash;92. \u003c/li\u003e\n\u003cli\u003eYang W, Fidelis TT, Sun W-H. Machine Learning in Catalysis, From Proposal to Practicing. ACS Omega. United States; 2020; 5: 83\u0026ndash;8. \u003c/li\u003e\n\u003cli\u003eZhang H, Hu J, Zhu J, Li Q, Fang L. Machine learning-based metabolism-related genes signature and immune infiltration landscape in diabetic nephropathy. Front Endocrinol (Lausanne). Switzerland; 2022; 13: 1026938. \u003c/li\u003e\n\u003cli\u003eWang X, Pei Z, Hao T, Ariben J, Li S, He W, Kong X, Chang J, Zhao Z, Zhang B. Prognostic analysis and validation of diagnostic marker genes in patients with osteoporosis. Front Immunol. Switzerland; 2022; 13: 987937. \u003c/li\u003e\n\u003cli\u003eNguyen LH, Robinton DA, Seligson MT, Wu L, Li L, Rakheja D, Comerford SA, Ramezani S, Sun X, Parikh MS, Yang EH, Powers JT, Shinoda G, et al. Lin28b is sufficient to drive liver cancer and necessary for its maintenance in murine models. Cancer Cell. United States; 2014; 26: 248\u0026ndash;61. \u003c/li\u003e\n\u003cli\u003eWang X, He C. Reading RNA methylation codes through methyl-specific binding proteins. RNA Biol. United States; 2014; 11: 669\u0026ndash;72. \u003c/li\u003e\n\u003cli\u003eHuang H, Weng H, Sun W, Qin X, Shi H, Wu H, Zhao BS, Mesquita A, Liu C, Yuan CL, Hu Y-C, H\u0026uuml;ttelmaier S, Skibbe JR, et al. Recognition of RNA N(6)-methyladenosine by IGF2BP proteins enhances mRNA stability and translation. Nat Cell Biol. England; 2018; 20: 285\u0026ndash;95. \u003c/li\u003e\n\u003cli\u003eWang X, Zhao BS, Roundtree IA, Lu Z, Han D, Ma H, Weng X, Chen K, Shi H, He C. N(6)-methyladenosine Modulates Messenger RNA Translation Efficiency. Cell. United States; 2015; 161: 1388\u0026ndash;99. \u003c/li\u003e\n\u003cli\u003eShi H, Wei J, He C. Where, When, and How: Context-Dependent Functions of RNA Methylation Writers, Readers, and Erasers. Mol Cell. United States; 2019; 74: 640\u0026ndash;50. \u003c/li\u003e\n\u003cli\u003eFirth SM, McDougall F, McLachlan AJ, Baxter RC. Impaired blockade of insulin-like growth factor I (IGF-I)-induced hypoglycemia by IGF binding protein-3 analog with reduced ternary complex-forming ability. Endocrinology. United States; 2002; 143: 1669\u0026ndash;76. \u003c/li\u003e\n\u003cli\u003eBaxter RC, Dai J. Purification and characterization of the acid-labile subunit of rat serum insulin-like growth factor binding protein complex. Endocrinology. United States; 1994; 134: 848\u0026ndash;52. \u003c/li\u003e\n\u003cli\u003eJogie-Brahim S, Feldman D, Oh Y. Unraveling insulin-like growth factor binding protein-3 actions in human disease. Endocr Rev. United States; 2009; 30: 417\u0026ndash;37. \u003c/li\u003e\n\u003cli\u003eYamada PM, Lee K-W. Perspectives in mammalian IGFBP-3 biology: local vs. systemic action. Am J Physiol Cell Physiol. United States; 2009; 296: C954-976. \u003c/li\u003e\n\u003cli\u003eYi D, Wang Q, Zhao Y, Song Y, You H, Wang J, Liu R, Shi Z, Chen X, Luo Q. Alteration of N (6) -Methyladenosine mRNA Methylation in a Rat Model of Cerebral Ischemia-Reperfusion Injury. Front Neurosci. Switzerland; 2021; 15: 605654. \u003c/li\u003e\n\u003cli\u003eSun D, Zhao T, Zhang Q, Wu M, Zhang Z. Fat mass and obesity-associated protein regulates lipogenesis via m(6) A modification in fatty acid synthase mRNA. Cell Biol Int. England; 2021; 45: 334\u0026ndash;44. \u003c/li\u003e\n\u003cli\u003eZhao X, Yang Y, Sun B-F, Shi Y, Yang X, Xiao W, Hao Y-J, Ping X-L, Chen Y-S, Wang W-J, Jin K-X, Wang X, Huang C-M, et al. FTO-dependent demethylation of N6-methyladenosine regulates mRNA splicing and is required for adipogenesis. Cell Res. England; 2014; 24: 1403\u0026ndash;19. \u003c/li\u003e\n\u003cli\u003eXu H, Yang Y-J, Yang T, Qian H-Y. Statins and stem cell modulation. Ageing Res Rev. England; 2013; 12: 1\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eCoker H, Wei G, Brockdorff N. m6A modification of non-coding RNA and the control of mammalian gene expression. Biochim Biophys Acta Gene Regul Mech. Netherlands; 2019; 1862: 310\u0026ndash;8. \u003c/li\u003e\n\u003cli\u003eJiang F, Hu Y, Liu X, Wang M, Wu C. Methylation Pattern Mediated by m(6)A Regulator and Tumor Microenvironment Invasion in Lung Adenocarcinoma. Oxid Med Cell Longev. United States; 2022; 2022: 2930310. \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"osteoporosis, m6A, diagnostic biomarkers, diagnostic predictive model, machine learning","lastPublishedDoi":"10.21203/rs.3.rs-4153446/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4153446/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOsteoporosis(OP) is a systemic skeletal dysfunction disorder that occurs in large numbers worldwide. The aim of our study was to screen potential m6A-related diagnostic markers and establish diagnostic predictive models to provide a reference framework for the diagnosis and treatment of OP.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGSE56815 and GSE35956 were retrieved from GEO. The m6A-related diagnostic markers for OP were identified through random forest algorithm screening of GSE56815, and subsequently validated for diagnostic efficacy in GSE35956. The mechanism exploration of diagnostic markers was conducted via GO and KEGG analyses based on GSEA. Immune cell infiltration estimation was performed using ssGSEA, with correlation analysis between immune cells and diagnostic markers. Consensus clustering produced two OP patient subtypes, which were compared using m6A scores.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e4 m6A-related diagnostic markers for OP have been selected by random forest. Utilizing these 4 diagnostic markers, an OP diagnostic model was successfully established, and its diagnostic value was validated using independent external data. The immune infiltration analysis of ssGSEA revealed that CD56 dim natural killer cell significantly infiltrated the OP samples, while the diagnostic markers were found to possess regulatory effects on diverse immune cells. OP samples were classified into two m6A subtypes through the concordance clustering, with type A having a higher m6A score than type B. The classification can provide more instructive assistance for the diagnosis of OP.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA diagnostic model for OP was constructed based on four m6A-related genes in the study, which provides significant references for diagnosis of OP and holds practical significance.\u003c/p\u003e","manuscriptTitle":"Identification of m6A-associated diagnostic biomarkers and subtypes for osteoporosis's disease diagnosis and risk prediction based on machine learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-22 01:56:17","doi":"10.21203/rs.3.rs-4153446/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"b568311c-ef1e-4bc4-b198-9d527bdc3e17","owner":[],"postedDate":"April 22nd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-11-30T17:38:27+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-22 01:56:17","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4153446","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4153446","identity":"rs-4153446","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}