ABSTRACT
Corneal-collagen crosslinking (CXL) is on the verge of achieving US Food and Drug Administration approval for treatment of progressive keratoconus (KCN) therefore, being able to accurately assess progression and the effect of CXL is a fundamental requirement for clinical management. Currently, the corneal tomography images from the Pentacam HR device (Oculus; Weltzar, Germany) are heavily relied on in the evaluation of KCN progression. Estimates of parameters such as corneal curvature and corneal thickness are analyzed serially over time to assess change. However, within each parameter, there is a degree of variability. In the normal population, this variability has been well characterized. This study was designed to provide the same characterization of repeatability and reproducibility in eyes with mild to moderate KCN.
Patients with mild to moderate KCN (cases were classified using the topographical KCN classification, which is based on corneal front shape, provided within the Pentacam device, patients were grade “1” to “2–3”), were recruited from a specialized anterior segment unit at the Jules-Gonin Eye Hospital in Lausanne, Switzerland. The study was approved by the Cantonal Committee for Ethics and Human Research, Vaud, Lausanne, and adhered to the tenets of the Declaration of Helsinki. Tomography measurements were obtained using the Pentacam HR (V 1.20r02). Three repeated measurements in the 25-picture mode by 2 independent observers, first 3 images taken by KH, then 3 images by IG. Scans were taken in automatic release mode. Only measurements with a quality factor of “OK” or when >95% of the data were validated by the system were used for analysis.
The precision of the Pentacam HR in terms of repeatability (Sr) and reproducibility (SR) was assessed per the recommendations from the British Standards Institute and the International Organization for Standardization. Repeatability limits (r) equals the within-subject SD for repeated measures with same observer, which is derived by a 1-way analysis of variance (ANOVA). The repeatability limit (r) is reported as Sr *1.96* √2, which gives the likely limits within which 95% of measurements should occur. We calculated Sr and r for observer 1. The reproducibility (SR) and reproducibility limits (R) were calculated similarly, using a 1-way ANOVA from the measurement results from observer 1 compared with observer 2. In addition, a comparison was made using the single image, the mean of the first 2 images, and the mean of all 3 images from both observers. Keratoconus pairs of eyes were asymmetric, patient number and eye were parameters included in the ANOVA analysis. Analysis was performed with R Version 2.15.1.
Images from 32 eyes (16 right, 16 left) of 20 patients (6 females, 14 males) were taken. The mean age of patients was 31 years (range, 18–47; median, 29; SD, ±8.6). Overall, 192 images (2 observers taking 3 images on 32 eyes) underwent statistical evaluation. Baseline mean and SD for thinnest corneal thickness, maximal corneal curvature (Kmax), mean corneal curvature Km anterior, astigmatism, anterior chamber (AC) depth, and corneal volume at 7 mm were 482.1±36.8 μm, 52.3±3.7 diopters (D), 46.0±2.2 D, −3.25±1.6 D, 3.3±0.3 mm, and 23.5±1.6 mm3, respectively.
Anterior and posterior corneal curvature (K values) measurements were found to be repeatable (Table 1); for curvature measures the worst repeatability limits were for Kmax with a value of 1.97 D. Summary data in terms of KCN power deviation, AC depth, AC volume, and AC angle estimates were moderately repeatable, but were all greater than normative values. Front surface elevation maps were more repeatable than back surface elevation maps. Pachymetry estimates had good repeatability limits for pupil center, corneal apex, and thinnest corneal thickness, with most measures inside normal limits.