Distribution of Anterior and Posterior Corneal Astigmatism in Eyes With Keratoconus MOHAMMAD NADERAN, MOHAMMAD TAHER RAJABI, AND PARVIZ ZARRINBAKHSH PURPOSE: To investigate the magnitude, with-the-rule ERATOCONUS (KC) IS A PROGRESSIVE, USUALLY (WTR) or against-the-rule (ATR) orientation, and vec- bilateral ectatic corneal disorder, characterized by 1,2 tor components (Jackson astigmatic vectors [J0 and J45] K corneal thinning and protrusion. KC starts at and blurring strength) of the anterior and posterior puberty and progresses to the third or fourth decade of corneal astigmatism (ACA and PCA) in patients with life, causing myopia and astigmatism, which results in keratoconus (KC) in a retrospective study, and to try to severe vision distortion and sometimes even blindness.1 find suitable cutoff points for ACA and PCA in an Astigmatism is a refractive error that is mostly caused by attempt to discriminate KC from normal corneas. toricity of the anterior corneal surface leading to visually DESIGN: Retrospective age- and sex-matched case- significant optical aberration. Both the anterior and poste- control study. rior corneal surfaces contribute to the total corneal METHODS: Using the Pentacam images, the aforemen- astigmatism. Recently, the direct and quantitative mea- tioned parameters were compared between 1273 patients surement of the posterior corneal measurements in a with KC and 1035 normal participants. clinical setting has been possible with new imaging tech- RESULTS: The mean magnitude of the ACA and PCA nologies such as slit-scanning, Scheimpflug, or optical was 4.49 ± 2.16 diopter (D) and 0.90 ± 0.43 D, respec- coherence devices.3,4 tively. The dominant astigmatism orientation of the Assessment of the corneal astigmatism plays an impor- ACA was ATR in KC patients and WTR in normal par- tant role in vision correction procedures such as rigid gas- ticipants (P < .001), while for the PCA it was WTR in permeable lens prescription or intraocular lens (IOL) im- KC patients and ATR in normal participants (P < .001). plantation in KC patients. Ho and associates3 reported There was a significant agreement between the axis orien- that neglecting the posterior corneal astigmatism may tations of ACA and PCA in KC patients (k [ 0.077, result in significant deviation in the estimation of the P < .001), but not in the normal group (P [ .626). corneal astigmatism. There are several studies evaluating 5 ACA and PCA magnitude, M, J0,J45, and blur signifi- anterior corneal astigmatism in KC patients ; however, a cantly increased by increasing KC severity. There was a few studies evaluated the magnitude and orientation of trend for increasing anterior ATR and posterior WTR, the posterior corneal astigmatism in patients with KC.6 and decreasing oblique astigmatism on both corneal sur- Moreover, several studies tried to differentiate between faces by increasing the KC severity according to the KC and normal corneas by means of corneal astigmatism Amsler-Krumeich classification. A cutoff value of 1.8 and suggested various cutoff points with different sensi- D for ACA had 90.2% sensitivity and specificity, and tivity and specificity.7–12 But the results are inconclusive. that of 0.4 D for PCA had 89.5% sensitivity and In the current study, we aimed to investigate the magni- 85.0% specificity for discriminating KC from normal cor- tude, orientation, and vector components of the anterior neas. and posterior corneal astigmatism in patients with KC in CONCLUSION: Our findings can help clinicians in the comparison with normal corneas and according to different diagnosis of KC and lens manufacturers in designing suit- KC severity stages. Furthermore, we tried to find a suitable able contact or intraocular lenses. (Am J Ophthalmol cutoff point for the anterior and posterior corneal astigma- 2016;167:79–87. Ó 2016 Elsevier Inc. All rights tism that could discriminate between KC and normal cor- reserved.) neas with the highest sensitivity, specificity, and accuracy. METHODS Supplemental Material available at AJO.com. Accepted for publication Mar 31, 2016. A RETROSPECTIVE AGE- AND SEX-MATCHED CASE-CONTROL From the Eye Research Center, Farabi Eye Hospital, Tehran University of Medical Sciences (M.N., M.T.R.); Ophthalmic Research Center, study was conducted in Zarrinbakhsh Eye Clinic, Tehran, Shahid Beheshti University of Medical Sciences (M.N.); and Iran, reviewing the documents of the patients diagnosed Zarrinbakhsh Eye Clinic (P.Z.), Tehran, Iran. with KC who attended the clinic from 2010 to 2015. Inquiries to Mohammad Naderan, Eye Research Center, Farabi Eye Hospital, Qazvin Square, Tehran, Iran 1336616351; e-mail: Moh@ This study was in accordance with the tenets of the Decla- Naderan.com ration of Helsinki and the Institutional Review Board 0002-9394/$36.00 Ó 2016 ELSEVIER INC.ALL RIGHTS RESERVED. 79 http://dx.doi.org/10.1016/j.ajo.2016.03.051 (IRB) and ethics committee of our clinic approved the surface was between 60 and 120 degrees and as against- study. Since our study was a retrospective chart review, the-rule (ATR) when the steepest meridian was between the IRB of our clinic waived the requirement of obtaining 0 and 30 degrees or 150 and 180 degrees. Since the dioptric informed consent from the participants. power of the posterior corneal surface is negative, posterior The diagnosis of KC was based on the clinical character- corneal astigmatism was classified as WTR when the steep- istic signs, such as Fleischer ring, Vogt striae, and stromal est meridian was between 0 and 30 or 150 and 180 degrees scar, using slit-lamp examinations and also corneal topog- and as ATR when the steepest meridian was between 60 raphy evaluation using Pentacam (OCULUS Optikgerate and 120 degrees. The remaining values were classified as GmbH, Wetzlar, Germany). For the control group, a num- oblique astigmatism. ber of age- and sex-matched subjects who primarily attended The power vector method was employed to quantify the the clinic for laser in situ keratomileusis, and who did not relationship between astigmatism measurements.15 Con- fulfill the diagnostic criteria of the KC or were not KC sus- ventional script notations of manifest refractions (sphere pect based on slit-lamp examination and Pentacam imaging, [S], cylinder [C], and axis [a]) were applied to calculate po- were consecutively selected and included in the study. Based wer vector coordinates. The method uses 3 fundamental on the patients’ documents, those with a history of ocular vectors, including M ¼ S þ C/2, J0 ¼ (ÀC/2) cos2a,and trauma or surgery and any corneal or other ophthalmic dis- J45 ¼ (ÀC/2) sin2a, where S is the sphere power, C is the orders were excluded from the study. cylinder power, a is the cylinder axis, and J is the Jackson The Pentacam device uses a 475 nm monochromatic astigmatic vector. M is the spherical lens equal to the spher- blue light-emitting diode (LED) with a 180-degree rotating ical equivalent of the given refractive error. J0 value is the Scheimpflug camera. The camera rotates around the opti- cylinder power set at 90-degree and 180-degree meridians cal axes of the eye and within 2 seconds captures a total and J45 value refers to a cross-cylinder set at 45 and 135 de- of 25 images and produces a 3-dimensional model of the grees.16 The overall blurring strength is calculated through 2 2 2 1/2 anterior segment of the eye. The instrument is capable of the following formula: B ¼ (M þ J0 þ J45 ) . automatically analyzing the anterior segment, the anterior Receiver operating characteristic (ROC) curves were chamber, and the lens and performing the anterior and pos- produced to determine the diagnostic significance of terior topography of the cornea and pachymetry measure- various astigmatism measurements. The area under ROC ments. curves (AUROC) was calculated to describe the predictive Before performing the Pentacam imaging, patients were accuracy of the different measurements and the optimized asked to stop using contact lenses for at least 2 weeks. The cutoff points that could best distinguish KC from normal imaging was performed as follows. The patients were asked eyes. An AUROC between 0.90 and 1.0 represents excel- to place their chin on the chin rest of the device, press their lent discrimination, between 0.80 and 0.90 good, between forehead to the forehead strap, stare at a central target or 0.70 and 0.80 fair, between 0.60 and 0.70 poor, and be- fixation light, and, when a perfect alignment between the tween 0.50 and 0.60 very poor; <0.50 represents insuffi- patients’ eye and visual axis was obtained, they were asked cient measures. to blink, and then the imaging was performed. All measure- The best cutoff point was determined where the tests’ ments were based on the data from the annular ring that sensitivity and specificity were maximized. Sensitivity, was 3 mm in diameter around the corneal apex. In our specificity, and positive and negative predictive values clinic, we routinely assess the Scheimpflug imaging param- (PPV and NPV) were calculated for the measurements eters on the 3 mm corneal diameter. This would make our with AUROC of >_0.900 to assess the validity of cutoff results comparable with the literature.13 The Pentacam im- points for predicting KC. The sensitivity (false negative) ages were reviewed, and those with a good quality were is defined as the ability of a test to correctly identify the pa- included in the study. tients with disease. The specificity (false positive) is defined In patients with bilateral KC and in the normal group as the ability of a test to correctly identify the patients as well, only 1 eye of each participant was randomly without disease. PPV is the probability that patients with selected.