Abstract：
BACKGROUND: This study aimed to explore ocular manifestations in ANCA-associated vasculitis (AAV), focusing on granulomatosis with polyangiitis (GPA), eosinophilic granulomatosis with polyangiitis (EGPA), and microscopic polyangiitis (MPA) and to examine the associations with laboratory parameters and other systemic manifestations.
METHODS: This retrospective study reviewed data from 533 AAV patients across two major Chinese medical centers from January 2016 to November 2023. Data including diagnosis, cranial manifestations of disease, ocular complications, and laboratory parameters were analyzed. Univariate and multivariable logistic regression analyses assessed associations across disease manifestations. Machine learning models were also utilized to predict the risk of retinal/eye involvement in AAV patients.
RESULTS: Among 533 patients (210 GPA, 217 MPA, 99 EGPA, and 7 unclassified AAV), ocular complications were observed in 20.64% of them, with a distribution of 36.67% in GPA, 7.37% in MPA, and 18.18% in EGPA. The most common ocular manifestations included scleritis and retro-orbital mass/dacryocystitis, which were notably prevalent in GPA patients. Retinal involvement was observed in 9.09% of EGPA cases. The machine learning models yielded that eosinophil percentage (EOS%), high-sensitivity C-reactive protein (hsCRP), and CD4 + T cell/CD8 + T cell ratio (T4/T8) can predict retinal involvement. Furthermore, the white blood cell, EOS%, APTT, IgA, hsCRP, PR3-ANCA, and T4/T8 can predict eye involvement.
CONCLUSIONS: Ocular manifestations are a prevalent complication across all forms of AAV. Predictive models developed through machine learning offer promising tools for early intervention and tailored patient care. This necessitates a multidisciplinary approach, integrating rheumatology and ophthalmology expertise for optimal patient outcomes.
METHODS: This retrospective study reviewed data from 533 AAV patients across two major Chinese medical centers from January 2016 to November 2023. Data including diagnosis, cranial manifestations of disease, ocular complications, and laboratory parameters were analyzed. Univariate and multivariable logistic regression analyses assessed associations across disease manifestations. Machine learning models were also utilized to predict the risk of retinal/eye involvement in AAV patients.
RESULTS: Among 533 patients (210 GPA, 217 MPA, 99 EGPA, and 7 unclassified AAV), ocular complications were observed in 20.64% of them, with a distribution of 36.67% in GPA, 7.37% in MPA, and 18.18% in EGPA. The most common ocular manifestations included scleritis and retro-orbital mass/dacryocystitis, which were notably prevalent in GPA patients. Retinal involvement was observed in 9.09% of EGPA cases. The machine learning models yielded that eosinophil percentage (EOS%), high-sensitivity C-reactive protein (hsCRP), and CD4 + T cell/CD8 + T cell ratio (T4/T8) can predict retinal involvement. Furthermore, the white blood cell, EOS%, APTT, IgA, hsCRP, PR3-ANCA, and T4/T8 can predict eye involvement.
CONCLUSIONS: Ocular manifestations are a prevalent complication across all forms of AAV. Predictive models developed through machine learning offer promising tools for early intervention and tailored patient care. This necessitates a multidisciplinary approach, integrating rheumatology and ophthalmology expertise for optimal patient outcomes.
摘要：
背景：本研究旨在探讨ANCA相关性血管炎（AAV）的眼部表现，专注于肉芽肿性多血管炎（GPA），嗜酸性肉芽肿性多血管炎（EGPA），和显微镜下多血管炎（MPA），并检查与实验室参数和其他全身表现的关联。
方法：这项回顾性研究回顾了2016年1月至2023年11月两个主要中国医疗中心的533例AAV患者的数据。包括诊断在内的数据，疾病的颅骨表现，眼部并发症,并对实验室参数进行了分析。单变量和多变量逻辑回归分析评估了疾病表现之间的关联。还利用机器学习模型来预测AAV患者中视网膜/眼睛受累的风险。
结果：在533例患者中（210GPA，217MPA,99EGPA，和7个未分类的AAV)，其中20.64%出现眼部并发症,GPA分布为36.67%，MPA为7.37%，和18.18%在EGPA。最常见的眼部表现包括巩膜炎和眶后肿块/泪囊炎，在GPA患者中尤其普遍。在9.09%的EGPA病例中观察到视网膜受累。机器学习模型得出嗜酸性粒细胞百分比(EOS%)，高敏C反应蛋白（hsCRP），CD4+T细胞/CD8+T细胞比值(T4/T8)可以预测视网膜受累。此外，白细胞,EOS%,APTT,IgA,hsCRP,PR3-ANCA,T4/T8可以预测眼部受累。
结论：眼部表现是所有形式的AAV的普遍并发症。通过机器学习开发的预测模型为早期干预和量身定制的患者护理提供了有前途的工具。这就需要多学科的方法，整合风湿病学和眼科专业知识，以获得最佳患者预后。
方法：这项回顾性研究回顾了2016年1月至2023年11月两个主要中国医疗中心的533例AAV患者的数据。包括诊断在内的数据，疾病的颅骨表现，眼部并发症,并对实验室参数进行了分析。单变量和多变量逻辑回归分析评估了疾病表现之间的关联。还利用机器学习模型来预测AAV患者中视网膜/眼睛受累的风险。
结果：在533例患者中（210GPA，217MPA,99EGPA，和7个未分类的AAV)，其中20.64%出现眼部并发症,GPA分布为36.67%，MPA为7.37%，和18.18%在EGPA。最常见的眼部表现包括巩膜炎和眶后肿块/泪囊炎，在GPA患者中尤其普遍。在9.09%的EGPA病例中观察到视网膜受累。机器学习模型得出嗜酸性粒细胞百分比(EOS%)，高敏C反应蛋白（hsCRP），CD4+T细胞/CD8+T细胞比值(T4/T8)可以预测视网膜受累。此外，白细胞,EOS%,APTT,IgA,hsCRP,PR3-ANCA,T4/T8可以预测眼部受累。
结论：眼部表现是所有形式的AAV的普遍并发症。通过机器学习开发的预测模型为早期干预和量身定制的患者护理提供了有前途的工具。这就需要多学科的方法，整合风湿病学和眼科专业知识，以获得最佳患者预后。
