EyeWorld is the official news magazine of the American Society of Cataract & Refractive Surgery.
Issue link: https://digital.eyeworld.org/i/274531
E W RESIDENTS 88 February 2014 by Dean McGee Eye Institute residents: Evan Allan, MD, Derek Bitner, MD, Yasser Elshatory, MD, Jim Hoffmann, MD, Blake Isernhagen, MD, Jared Jackson, MD, Jonathan Perkins, MD, Tyler Sorensen, MD, Evan Sussenbach, MD, Jeremy Tan, MD, Jonathan Walgama, MD, Andy Wilson, MD; attendings: Alex Cohen, MD, R. Michael Siatkowski, MD Review of "Validation of metrics for the detection of subclinical keratoconus in a protocols. This particular article ad- dresses one possible metric to aid in the diagnosis of keratoconus at an earlier, subclinical stage. Study summary This retrospective cross-sectional study was performed to determine if the metric previously created by the authors could identify patients with subclinical keratoconus with a clini- cally acceptable level of accuracy. The study evaluated 32 clinically normal eyes with topographically e vident keratoconus in the con- tralateral eye. Patients were diag- nosed with keratoconus based on an asymmetric bow-tie appearance on corneal topography with or without skewed axes. The fellow eyes (with- out clinical or topographic evidence of keratoconus) were presumed to have subclinical keratoconus if they had less than 1.4 D of paracentral I-S dioptric difference. Other inclusion criteria included no signs or symp- toms of keratoconus, no previous contact lens wear for four weeks (rigid) or two weeks (soft), and no history of ocular problems or surger- ies. These patients were chosen be- cause the authors thought that they most likely represented eyes that would go on to develop clinically definite keratoconus. The control group was obtained by retrospective convenience sam- pling of 245 eyes of 245 patients who had undergone hyperopic or myopic LASIK at a university refrac- tive clinic at Goethe University in Frankfurt am Main, Germany. The authors retrospectively reviewed the records of the control group for any subtle signs of keratectasia. Absence of keratectasia over the 12-month period post-LASIK was defined by the absence of increasing corneal curvature or astigmatism, inferior steepening, or increasing refractive astigmatism of more than 0.5 D compared to the one-month postop- erative refraction. Another inclusion criterion was the absence of de- creased corrected distance vision at 12 months postoperatively. All post- operative topographic maps were re- viewed by a single expert (JB) to rule out iatrogenic keratectasia, and the observer was masked to correspon- ding preoperative topographic maps to avoid classification bias. The authors used the Orbscan IIz (Bausch + Lomb, Rochester, N.Y.) to acquire corneal topography. The i nput features used to construct the discriminant function coefficients obtained included Zernike coeffi- cients of anterior corneal surface (DA), Zernike coefficients of poste- rior corneal surface (DP), Zernike coefficients of anterior and posterior corneal surface (DAP), pachymetry metrics (DT), and Zernike coeffi- c ients of anterior and posterior corneal surface and pachymetry metrics (DAPT). Strength of discrimi- nation was determined using re- ceiver operating characteristic (ROC) curve analysis. Anterior and poste- rior corneal surface aberrations were calculated using Visual Optics Lab-Pro and Matlab software. Wilcoxon-Mann-Whitney U tests (non-normal distributions) and Student t tests (normal distributions) were used to look for intergroup dif- ferences between keratoconus fellow eyes and normal eyes across kerato- metric indices, total higher order aberrations, root mean square val- ues, coma RMS, spherical aberration RMS, residual HOA RMS, 1st to 7th order RMS, and individual Zernike coefficients. Discriminant function coefficients were obtained from both a previous study as well as the cur- rent study population. Receiver op- erating characteristic curves were plotted and the area under the ROC curve was calculated using Matlab. Sensitivity, specificity, and accuracy were also calculated from the ROC curve data. Only parameters that correctly classified keratoconus fel- low eyes versus normal eyes at a rate of 80% or higher were identified in the study. The anterior C(1,–1) and C(3,1) coefficients had the highest area under the ROC curve (both 0.87). The anterior 5th order RMS was the RMS value with the maximum area under the ROC curve (0.90). The dis- criminant function with input from anterior and posterior Zernike coeffi- cients (DAP) and DAP including pachymetry data (DAPT) performed best (area under curve 0.864 and 0.857, respectively). DAPT reached Validation of metrics for the detection of subclinical keratoconus in a new patient collective Jens Bühren, MD, Thomas Schäffeler, MD, Thomas Kohnen, MD, PhD, FEBO J Cataract Refract Surg (Feb) 2014; 40: 259-268 Purpose: To validate the discriminative ability of wavefront- and pachymetry-based corneal topographic metrics to detect subclinical keratoconus in a new patient collective. Setting: Department of Ophthalmology, Goethe-University, Frankfurt am Main, Germany. Design: Retrospective cross-sectional study. Methods: Normal fellow eyes with early keratoconus and preoperative eyes with an uneventful follow-up without signs of iatrogenic keratectasia 12 months after laser in situ keratoconus were included. Zernike coeffi- cients from the anterior and posterior surfaces and corneal thickness spatial profiles and corresponding discriminant functions were assessed for their usefulness to discriminate between eyes with subclinical kerato- conus and normal eyes using receiver-operating-characteristic (ROC) curve analysis. Discriminant functions were obtained from a previous study and constructed de novo from the present collective. Results: The anterior C(1,−1) and C(3,−1) coefficients had the highest area under the ROC curve (both 0.87). The anterior 5th-order root mean square (RMS) was the RMS value with the maximum area under the ROC curve (0.90). The discriminant function with input from anterior and posterior Zernike coefficients (DAP) and DAP including pachymetry data (DAPT) performed best (area under ROC curve 0.864 and 0.857, respec- tively). Applying cutoff values from a previous study resulted in a minimal drop in accuracy (0.0% to 1.3%). The construction of discriminant func- tions from the present dataset resulted in a gain in accuracy of between 3.5% and 9.6%, with DAPT reaching the maximum area under the ROC curve of 0.956. Conclusion: Validation in a new and larger patient collective proved the usefulness of metrics based on corneal wavefront and pachymetry for the detection of subclinical keratoconus. Financial disclosure: No author has a financial or proprietary interest in any material or method mentioned. Background D iagnosing disease at earlier stages leads to more prompt initiation of treatment and potential reduction in morbidity, but diagnosis at an early stage may be elusive for many con- ditions. One such disease is kerato- conus. The gold standard for diagnosis remains corneal topogra- p hy, and most diagnoses occur after the disease manifests clinically. Various attempts are being made to develop accurate metrics for the di- agnosis of subclinical disease, which could aid in decision-making in different clinical scenarios such as proceeding with refractive surgery or early intervention with collagen crosslinking. This could impact cost of care and public health screening R. Michael Siatkowski, MD, residency program director, Dean McGee Eye Institute T his month I asked the Dean McGee residents to review this paper on diagnosing subclinical keratoconus for EyeWorld. –David F. Chang, MD, chief medical editor EyeWorld journal club 84-90 Residents_EW February 2014-DL2_Layout 1 1/30/14 11:13 AM Page 88