{"title":"抗dsdna IgG糖基化与treatment-naïve系统性红斑狼疮患者器官受累相关。","authors":"Zhuochao Zhou, Yuhong Liu, Xiaotong Gu, Haowen Zhang, Panan Zhang, Yue Sun, Honglei Liu, Xiaobing Cheng, Yutong Su, Hui Shi, Qiongyi Hu, Huihui Chi, Jianfen Meng, Jinchao Jia, Tingting Liu, Mengyan Wang, Cui Lu, Yunping Cai, Yijun You, Dehao Zhu, Shifang Ren, Jialin Teng, Jingyi Wu, Chengde Yang, Junna Ye","doi":"10.1136/lupus-2025-001665","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>This study aimed to leverage machine learning algorithms to explore the relationship between anti-double-stranded DNA (anti-dsDNA) immunoglobulin G (IgG) glycosylation and the degree of organ involvement in patients with SLE.</p><p><strong>Methods and analysis: </strong>We enrolled 86 consecutive treatment-naïve patients with SLE positive for anti-dsDNA antibodies from the Department of Rheumatology and Immunology at Ruijin Hospital, Shanghai, between 2017 and 2019. We quantified and classified the degree of organ involvement in patients with SLE and analysed each glycoform and a combination of glycoforms of purified anti-dsDNA IgG. A random forest classifier and artificial neural network were applied to evaluate the correlation between the levels of glycoform pairs and the degree of organ involvement.</p><p><strong>Results: </strong>Pearson's correlation analysis revealed a strong connection between the involved and uninvolved organs in patients with SLE. Random forest analysis showed that the combination of IgG1Gal and IgG3/4Bis had the highest accuracy (0.7692) and area under the curve (0.8187). In terms of predicting the degree of involvement using an artificial neural network, IgG3/4Bis and IgG1Gal showed the lowest mean squared error (0.0244).</p><p><strong>Conclusions: </strong>Our study showed the effectiveness of combining glycoforms to classify and predict the degree of SLE organ involvement. Different glycoforms were correlated with the involvement degree to various extents, and the combination of anti-dsDNA IgG3/4Bis and IgG1Gal exhibited the best correlation with organ involvement.</p>","PeriodicalId":18126,"journal":{"name":"Lupus Science & Medicine","volume":"12 2","pages":""},"PeriodicalIF":3.5000,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12414148/pdf/","citationCount":"0","resultStr":"{\"title\":\"Glycosylation of anti-dsDNA IgG correlates with organ involvement in treatment-naïve patients with systemic lupus erythematosus.\",\"authors\":\"Zhuochao Zhou, Yuhong Liu, Xiaotong Gu, Haowen Zhang, Panan Zhang, Yue Sun, Honglei Liu, Xiaobing Cheng, Yutong Su, Hui Shi, Qiongyi Hu, Huihui Chi, Jianfen Meng, Jinchao Jia, Tingting Liu, Mengyan Wang, Cui Lu, Yunping Cai, Yijun You, Dehao Zhu, Shifang Ren, Jialin Teng, Jingyi Wu, Chengde Yang, Junna Ye\",\"doi\":\"10.1136/lupus-2025-001665\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Objective: </strong>This study aimed to leverage machine learning algorithms to explore the relationship between anti-double-stranded DNA (anti-dsDNA) immunoglobulin G (IgG) glycosylation and the degree of organ involvement in patients with SLE.</p><p><strong>Methods and analysis: </strong>We enrolled 86 consecutive treatment-naïve patients with SLE positive for anti-dsDNA antibodies from the Department of Rheumatology and Immunology at Ruijin Hospital, Shanghai, between 2017 and 2019. We quantified and classified the degree of organ involvement in patients with SLE and analysed each glycoform and a combination of glycoforms of purified anti-dsDNA IgG. A random forest classifier and artificial neural network were applied to evaluate the correlation between the levels of glycoform pairs and the degree of organ involvement.</p><p><strong>Results: </strong>Pearson's correlation analysis revealed a strong connection between the involved and uninvolved organs in patients with SLE. Random forest analysis showed that the combination of IgG1Gal and IgG3/4Bis had the highest accuracy (0.7692) and area under the curve (0.8187). In terms of predicting the degree of involvement using an artificial neural network, IgG3/4Bis and IgG1Gal showed the lowest mean squared error (0.0244).</p><p><strong>Conclusions: </strong>Our study showed the effectiveness of combining glycoforms to classify and predict the degree of SLE organ involvement. Different glycoforms were correlated with the involvement degree to various extents, and the combination of anti-dsDNA IgG3/4Bis and IgG1Gal exhibited the best correlation with organ involvement.</p>\",\"PeriodicalId\":18126,\"journal\":{\"name\":\"Lupus Science & Medicine\",\"volume\":\"12 2\",\"pages\":\"\"},\"PeriodicalIF\":3.5000,\"publicationDate\":\"2025-09-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12414148/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Lupus Science & Medicine\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1136/lupus-2025-001665\",\"RegionNum\":2,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"RHEUMATOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Lupus Science & Medicine","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1136/lupus-2025-001665","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"RHEUMATOLOGY","Score":null,"Total":0}
Glycosylation of anti-dsDNA IgG correlates with organ involvement in treatment-naïve patients with systemic lupus erythematosus.
Objective: This study aimed to leverage machine learning algorithms to explore the relationship between anti-double-stranded DNA (anti-dsDNA) immunoglobulin G (IgG) glycosylation and the degree of organ involvement in patients with SLE.
Methods and analysis: We enrolled 86 consecutive treatment-naïve patients with SLE positive for anti-dsDNA antibodies from the Department of Rheumatology and Immunology at Ruijin Hospital, Shanghai, between 2017 and 2019. We quantified and classified the degree of organ involvement in patients with SLE and analysed each glycoform and a combination of glycoforms of purified anti-dsDNA IgG. A random forest classifier and artificial neural network were applied to evaluate the correlation between the levels of glycoform pairs and the degree of organ involvement.
Results: Pearson's correlation analysis revealed a strong connection between the involved and uninvolved organs in patients with SLE. Random forest analysis showed that the combination of IgG1Gal and IgG3/4Bis had the highest accuracy (0.7692) and area under the curve (0.8187). In terms of predicting the degree of involvement using an artificial neural network, IgG3/4Bis and IgG1Gal showed the lowest mean squared error (0.0244).
Conclusions: Our study showed the effectiveness of combining glycoforms to classify and predict the degree of SLE organ involvement. Different glycoforms were correlated with the involvement degree to various extents, and the combination of anti-dsDNA IgG3/4Bis and IgG1Gal exhibited the best correlation with organ involvement.
期刊介绍:
Lupus Science & Medicine is a global, peer reviewed, open access online journal that provides a central point for publication of basic, clinical, translational, and epidemiological studies of all aspects of lupus and related diseases. It is the first lupus-specific open access journal in the world and was developed in response to the need for a barrier-free forum for publication of groundbreaking studies in lupus. The journal publishes research on lupus from fields including, but not limited to: rheumatology, dermatology, nephrology, immunology, pediatrics, cardiology, hepatology, pulmonology, obstetrics and gynecology, and psychiatry.