Department of Ophthalmology, Philipps-University Marburg, Marburg, Germany.
Institute of Medical Bioinformatics and Biostatistics, Philipps-University Marburg, Marburg, Germany.
PLoS One. 2022 Jul 25;17(7):e0270120. doi: 10.1371/journal.pone.0270120. eCollection 2022.
To evaluate a multivariable model predicting the individual probability of successful intravitreal ocriplasmin (IVO) treatment in eyes with vitreomacular traction (VMT).
Data from three prospective, multicenter IVO studies (OASIS, ORBIT, and INJECT) were pooled. Patients were included if they were treated for a symptomatic VMT without a full-thickness macular hole. A prediction model for VMT resolution using the factors 'age' and 'horizontal VMT diameter' was validated by receiver operating characteristic analysis and according to grouped prediction after calibration. Multivariable regression analysis was performed to check robustness and explore further improvements.
Data from 591 eyes was included. In the univariate analysis all key factors (age, gender, VMT diameter, lens status, ERM) significantly correlated to treatment success. The prediction model was robust and clinically applicable to estimate the success rate of IVO treatment (AUC of ROC: 0.70). A refinement of the model was achieved through a calibration process.
The developed multivariable model using 'horizontal VMT diameter' and 'age' is a valid tool for prediction of VMT resolution upon IVO treatment.
评估一个多变量模型，预测玻璃体黄斑牵引（VMT）患者接受玻璃体内注氧酶（IVO）治疗成功的个体概率。
汇总了三个前瞻性、多中心 IVO 研究（OASIS、ORBIT 和 INJECT）的数据。纳入标准为患有有症状的 VMT 但无全层黄斑孔的患者。使用“年龄”和“水平 VMT 直径”这两个因素的 VMT 缓解预测模型通过接受者操作特征分析和基于校准后的分组预测进行验证。进行多变量回归分析以检查稳健性并探索进一步的改进。
共纳入 591 只眼的数据。在单变量分析中，所有关键因素（年龄、性别、VMT 直径、晶状体状态、ERM）与治疗成功显著相关。该预测模型稳健且适用于临床，可估计 IVO 治疗的成功率（ROC 曲线的 AUC：0.70）。通过校准过程对模型进行了改进。
使用“水平 VMT 直径”和“年龄”的多变量模型是预测IVO 治疗后 VMT 缓解的有效工具。
