Beijing Institute of Ophthalmology, Beijing Tongren Eye Center, Beijing Tongren Hospital, Capital Medical University, Beijing Ophthalmology & Visual Sciences Key Laboratory, Beijing, China.
Beijing Advanced Innovation Center for Big Data-Based Precision Medicine, Beihang University & Capital Medical University, Beijing, China.
Semin Ophthalmol. 2021 Nov 17;36(8):671-678. doi: 10.1080/08820538.2021.1896752. Epub 2021 Mar 18.
: To explore the feasibility of corneal morphological and biomechanical parameters for keratoconus and forme fruste keratoconus diagnosis.: This case-control study included a total of 517 eyes from 408 keratoconus patients (KC group), 83 eyes from 83 forme fruste keratoconus patients (FFKC group), and 158 eyes from 158 patients with normal corneas (NL group). All subjects underwent routine ophthalmologic examinations. Pentacam and Corneal Visualization Scheimpflug Technology (Corvis ST) were used to obtain corneal morphological and biomechanical parameters. Differences between groups were compared using receiver operating characteristic (ROC) curve analysis.: ROC analysis showed that all corneal morphological parameters and most biomechanical parameters distinguished KC from NL, with an area under the curve (AUC) greater than 0.80, of which Belin-Ambrósio enhanced ectasia total deviation index (BAD-D) and tomographic and biomechanical index (TBI) were most efficient. The AUC for distinguishing KC from NL of the BAD-D was 0.989 and the TBI was 0.993, which were not statistically significant (DeLong et al., = .232). The BAD-D cut-off point of 1.595 provided 95.9% sensitivity for distinguishing KC from NL with 100% specificity. The TBI cut-off point of 0.515 provided 96.7% sensitivity for distinguishing KC from NL with 100% specificity. The ability of other parameters to distinguish KC from NL was lower than that of BAD and TBI. Except for central astigmatism from the anterior corneal surface (AstigF), the AUC that distinguished FFKC from NL was 0.862. The AstigF cut-off point of 4.65 provided 73.5% sensitivity for distinguishing FFKC from NL with 99.3% specificity. Other parameters distinguished FFKC from NL with low efficiency. Among them, the AUC for distinguishing FFKC from NL of the TBI was 0.722, whose cut-off point of 0.273 provided 55.4% sensitivity for distinguishing KC from NL with 79.7% specificity.: BAD-D and TBI have the highest efficiency, sensitivity, and specificity for distinguishing KC from NL. Except for AstigF, other corneal morphological and biomechanical parameters have a relatively low ability to distinguish FFKC from NL.
探讨角膜形态学和生物力学参数在圆锥角膜和亚临床圆锥角膜诊断中的可行性。
本病例对照研究共纳入 408 例圆锥角膜患者（KC 组）的 517 只眼、83 例亚临床圆锥角膜患者（FFKC 组）的 83 只眼和 158 例正常角膜患者（NL 组）的 158 只眼。所有受试者均行常规眼科检查。使用 Pentacam 和角膜可视化 Scheimpflug 技术（Corvis ST）获取角膜形态学和生物力学参数。采用受试者工作特征（ROC）曲线分析比较组间差异。
ROC 分析显示，所有角膜形态学参数和大多数生物力学参数均能区分 KC 与 NL，曲线下面积（AUC）大于 0.80，其中 Belin-Ambrósio 增强的角膜扩张总偏差指数（BAD-D）和断层和生物力学指数（TBI）效率最高。BAD-D 区分 KC 与 NL 的 AUC 为 0.989，TBI 为 0.993，差异无统计学意义（DeLong 等，=0.232）。BAD-D 的截断值为 1.595 时，对区分 KC 与 NL 的敏感性为 95.9%，特异性为 100%。TBI 的截断值为 0.515 时，对区分 KC 与 NL 的敏感性为 96.7%，特异性为 100%。其他参数区分 KC 与 NL 的能力低于 BAD 和 TBI。除前角膜表面中央散光（AstigF）外，FFKC 与 NL 的 AUC 为 0.862。AstigF 的截断值为 4.65 时，对区分 FFKC 与 NL 的敏感性为 73.5%，特异性为 99.3%。其他参数区分 FFKC 与 NL 的效率较低。其中，TBI 区分 FFKC 与 NL 的 AUC 为 0.722，截断值为 0.273 时，对区分 KC 与 NL 的敏感性为 55.4%，特异性为 79.7%。
BAD-D 和 TBI 对区分 KC 与 NL 的效率、敏感性和特异性最高。除 AstigF 外，其他角膜形态学和生物力学参数区分 FFKC 与 NL 的能力较低。
