Kozobolis Vassilios, Sideroudi Haris, Giarmoukakis Athanassios, Gkika Maria, Labiris Georgios
Eye Institute of Thrace, Democritus University - Greece; and Department of Ophthalmology, University Hospital of Alexandroupolis - Greece.
Eur J Ophthalmol. 2012 Nov-Dec;22(6):920-30. doi: 10.5301/ejo.5000184. Epub 2012 Jul 3.
To evaluate the sensitivity and specificity of corneal biomechanical metrics, anterior segment data, and a combination model in differentiating forme fruste keratoconus (FFK) from normal corneas.
A total of 50 FFK eyes were identified by calculation of the KISA index and recruited FFK group. Results were compared with 50 normal eyes (NG group) randomly selected from 50 patients. The following parameters were evaluated for their diagnostic capacity by evaluation of their receiver operating characteristic curves (ROC): corneal hysteresis (CH), corneal resistance factor (CRF), corneal astigmatism (Cyl), anterior chamber depth (ACD), corneal volume (CV) at 3 mm (CV3) and at 5 mm (CV5), maximum posterior elevation value (PEL), central corneal thickness (CCT), thinnest corneal thickness (TCT) and its coordinates (TCTx, TCTy ), the ratio TCT/CCT, pachymetric progression indexes (PPImin, PPIavg, and PPImax), and Ambrósio's relational thickness (ARTmin, ARTavg, and ARTmax). Logistic regression was attempted for identification of a combined diagnostic model.
Significant differences were detected in all studied parameters except the Cyl, ACD, TCTx, and CV. Among individual parameters, the highest predictive accuracy was for ARTavg (area under the curve [AUC] 95.4%, sensitivity 90%, specificity 88.9%) and TCT (AUC 95.3%, sensitivity 90.9%, specificity 89%). Sufficient predictive accuracy (AUC 99.4%, sensitivity 98.8%, specificity 94.6%) was identified in a diagnostic model that combined the CRF, ARTavg, and PEL parameters.
None of the individual parameters provide sufficient diagnostic capacity in FFK. However, diagnostic models that combine biomechanical and tomographic data seem to provide high accuracy in differentiating FFK from normal corneas.
评估角膜生物力学指标、眼前节数据及联合模型在鉴别亚临床圆锥角膜（FFK）与正常角膜方面的敏感性和特异性。
通过计算KISA指数确定50只FFK眼并纳入FFK组。结果与从50例患者中随机选取的50只正常眼（NG组）进行比较。通过评估受试者操作特征曲线（ROC）来评价以下参数的诊断能力：角膜滞后（CH）、角膜阻力因子（CRF）、角膜散光（Cyl）、前房深度（ACD）、3mm处角膜体积（CV3）和5mm处角膜体积（CV5）、最大后表面高度值（PEL）、中央角膜厚度（CCT）、最薄角膜厚度（TCT）及其坐标（TCTx、TCTy）、TCT/CCT比值、厚度进展指数（PPImin、PPIavg和PPImax）以及安布罗西奥相关厚度（ARTmin、ARTavg和ARTmax）。尝试采用逻辑回归来确定联合诊断模型。
除Cyl、ACD、TCTx和CV外，所有研究参数均检测到显著差异。在各个参数中，预测准确性最高的是ARTavg（曲线下面积[AUC]95.4%，敏感性90%，特异性88.9%）和TCT（AUC 95.3%，敏感性90.9%，特异性89%）。在一个联合CRF、ARTavg和PEL参数的诊断模型中发现了足够的预测准确性（AUC 99.4%，敏感性98.8%，特异性94.6%）。
单个参数均不能为FFK提供足够的诊断能力。然而，结合生物力学和断层扫描数据的诊断模型似乎在区分FFK与正常角膜方面具有较高的准确性。
