Journal of Refractive Surgery

Original Article 

Influence of the Reference Surface Shape for Discriminating Between Normal Corneas, Subclinical Keratoconus, and Keratoconus

David Smadja, MD; Marcony R. Santhiago, MD; Glauco R. Mello, MD; Ronald R. Krueger, MD, MSE; Joseph Colin, MD†; David Touboul, MD

Abstract

PURPOSE:

To compare the discriminating ability of corneal elevation generated by a dual Scheimpflug analyzer calculated with different reference surfaces for distinguishing normal corneas from those with keratoconus and subclinical keratoconus.

METHODS:

A total of 391 eyes of 208 patients were prospectively enrolled in the study and divided into three groups: 167 eyes of 113 patients with keratoconus, 47 contralateral topographically normal eyes of patients with clinically evident keratoconus in the fellow eye, and 177 eyes of 95 refractive surgery candidates with normal corneas. All eyes were measured with a dual Scheimpflug analyzer (GALILEI Analyzer; Ziemer Ophthalmic Systems AG, Port, Switzerland). Maximum elevation values were recorded within the central 5-mm diameter in both anterior and posterior elevation maps. Discriminating ability of corneal elevation measurements obtained by best-fit toric and aspheric (BFTA) and best-fit sphere (BFS) reference surfaces were compared by receiver operator characteristic (ROC) curves.

RESULTS:

ROC curve analysis showed that corneal elevation measured by BFTA had a significantly better ability than with BFS for distinguishing normal corneas from those with keratoconus and forme fruste keratoconus (P = .01). Posterior elevation measured by BFTA had a significantly higher predictive accuracy for forme fruste keratoconus than anterior elevation with an area under ROC curves of 0.88 and 0.80, respectively (P = .01). The sensitivity and specificity achieved with the maximum posterior elevation for detecting keratoconus and forme fruste keratoconus were 99% and 99% for keratoconus and 82% and 80% for forme fruste keratoconus with the cut-off value at 16 and 13 μm, respectively.

CONCLUSIONS:

The ability to discriminate between normal cornea and forme fruste keratoconus with elevation parameters was significantly improved by using BFTA instead of BFS reference surface.

[J Refract Surg. 2013;29(4):274–281.]

From the University Center Hospital of Bordeaux, Anterior Segment and Refractive Surgery Unit, Bordeaux, France (DS, JC, DT); Cole Eye Institute, Cleveland Clinic Foundation, Refractive Surgery Department, Cleveland, Ohio (DS, MRS, RRK); the Department of Ophthalmology, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil (MRS); and the Department of Ophthalmology, Federal University Of Paraná, Curitiba, Brazil (GRM).

†: Deceased.

Dr. Krueger is a board member and consultant for Alcon Laboratories, Inc. The remaining authors have no financial or proprietary interest in the materials presented herein.

AUTHOR CONTRIBUTIONS

Study concept and design (JC, DS); data collection (DS); analysis and interpretation of data (RRK, GRM, DS, MRS, DT); drafting of the manuscript (DS); critical revision of the manuscript (JC, RRK, GRM, DS, MRS, DT); statistical expertise (DS)

Correspondence: David Smadja, MD, Bordeaux University Hospital, Hopital Pellegrin, Place Amelie Raba Leon, Bordeaux 33076, France. E-mail: davidsmadj@hotmail.fr

Received: September 19, 2012
Accepted: February 19, 2013

Abstract

PURPOSE:

To compare the discriminating ability of corneal elevation generated by a dual Scheimpflug analyzer calculated with different reference surfaces for distinguishing normal corneas from those with keratoconus and subclinical keratoconus.

METHODS:

A total of 391 eyes of 208 patients were prospectively enrolled in the study and divided into three groups: 167 eyes of 113 patients with keratoconus, 47 contralateral topographically normal eyes of patients with clinically evident keratoconus in the fellow eye, and 177 eyes of 95 refractive surgery candidates with normal corneas. All eyes were measured with a dual Scheimpflug analyzer (GALILEI Analyzer; Ziemer Ophthalmic Systems AG, Port, Switzerland). Maximum elevation values were recorded within the central 5-mm diameter in both anterior and posterior elevation maps. Discriminating ability of corneal elevation measurements obtained by best-fit toric and aspheric (BFTA) and best-fit sphere (BFS) reference surfaces were compared by receiver operator characteristic (ROC) curves.

RESULTS:

ROC curve analysis showed that corneal elevation measured by BFTA had a significantly better ability than with BFS for distinguishing normal corneas from those with keratoconus and forme fruste keratoconus (P = .01). Posterior elevation measured by BFTA had a significantly higher predictive accuracy for forme fruste keratoconus than anterior elevation with an area under ROC curves of 0.88 and 0.80, respectively (P = .01). The sensitivity and specificity achieved with the maximum posterior elevation for detecting keratoconus and forme fruste keratoconus were 99% and 99% for keratoconus and 82% and 80% for forme fruste keratoconus with the cut-off value at 16 and 13 μm, respectively.

CONCLUSIONS:

The ability to discriminate between normal cornea and forme fruste keratoconus with elevation parameters was significantly improved by using BFTA instead of BFS reference surface.

[J Refract Surg. 2013;29(4):274–281.]

From the University Center Hospital of Bordeaux, Anterior Segment and Refractive Surgery Unit, Bordeaux, France (DS, JC, DT); Cole Eye Institute, Cleveland Clinic Foundation, Refractive Surgery Department, Cleveland, Ohio (DS, MRS, RRK); the Department of Ophthalmology, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil (MRS); and the Department of Ophthalmology, Federal University Of Paraná, Curitiba, Brazil (GRM).

†: Deceased.

Dr. Krueger is a board member and consultant for Alcon Laboratories, Inc. The remaining authors have no financial or proprietary interest in the materials presented herein.

AUTHOR CONTRIBUTIONS

Study concept and design (JC, DS); data collection (DS); analysis and interpretation of data (RRK, GRM, DS, MRS, DT); drafting of the manuscript (DS); critical revision of the manuscript (JC, RRK, GRM, DS, MRS, DT); statistical expertise (DS)

Correspondence: David Smadja, MD, Bordeaux University Hospital, Hopital Pellegrin, Place Amelie Raba Leon, Bordeaux 33076, France. E-mail: davidsmadj@hotmail.fr

Received: September 19, 2012
Accepted: February 19, 2013

Identifying ectasia-susceptible corneas remains the major concern of preoperative refractive surgery screening. Although its prevalence is low, with estimates between 0.04%1 and 0.6%,2 post-LASIK ectasia is a severe complication that compromises the visual prognosis and could lead to corneal transplant.3 It has been long debated whether the earliest detectable sign of keratoconus would be a focal steepening identified with Placido corneal topography4,5 or a bulging of the posterior surface detected by tomography.6–8 Several recent studies that used a tomography imaging system have reported the sensitivity of posterior corneal elevation as a key variable for differentiating early keratoconus from normal corneas.7–10

Whereas Placido disk technology is able to accurately determine only the elevation of the anterior surface, tomography imaging systems allow for a direct acquisition of the corneal reliefs by measuring the spatial coordinates of several points located in both anterior and posterior surfaces.11 Height data are presented relative to reference shapes so that the clinician does not analyze the actual elevation data but data after subtracting out the reference shape. This method has been used to magnify the differences and allow for qualitative maps that will highlight clinically significant areas.12 Elevation display highly depends on reference surface shape, diameter, alignment, and fitting zone. The reference shape from which the corneal surface height is measured is often chosen as a sphere without positioning constraint (float mode) and is known as the best-fit sphere (BFS). However, considering variable corneal toricity and asphericity, a reference surface that is both toric and aspherical would fit better to the real corneal shape and therefore might help to enhance local changes and underlying abnormalities more sensitively.

The purpose of the current study was to assess the influence of the reference surface shape on the ability to detect subclinical keratoconus and to discriminate between normal corneas and those with subclinical keratoconus and keratoconus.

Patients and Methods

This prospective, comparative study was conducted at the University Hospital of Bordeaux, France, in the National Reference Center for Keratoconus and at the Cole Eye Institute of the Cleveland Clinic Foundation in Cleveland, Ohio. The study was approved by an Institutional Review Board and conducted in accordance with the tenets of the Declaration of Helsinki.

Patients

A total of 391 eyes of 208 patients were prospectively enrolled in the study. Corneas were classified into three groups based on eye conditions. Groups were defined as follows.

Normal eyes (177 eyes of 95 patients) were enrolled among suitable candidates undergoing a screening examination for refractive surgery and among the general population undergoing a routine ophthalmological examination. All patients had discontinued daily-wear soft contact lens use at least 1 week before evaluation. Eyes were considered normal when no clinical signs of keratoconus and no suggestive topographic or tomographic patterns of suspect keratoconus were found, such as asymmetric bowtie with skewed radial axes, focal or inferior steepening, central keratometry greater than 47.0 diopters (D), or corneas thinner than 500 μm. Exclusion criteria for this group were previous ocular surgery, ocular pathology, familial history of keratoconus, and contact lens wearing in the past week.

The subclinical keratoconus group comprised 47 eyes of 47 patients with forme fruste keratoconus, defined as the contralateral eyes of clinically evident keratoconus in the fellow eye (n = 47). These eyes had no clinical signs of keratoconus and a normal topographical aspect with no asymmetric bowtie and no focal or inferior steepening pattern. This condition is also known in the literature as “subclinical keratoconus” because it has already been reported that approximately 50% of clinically normal fellow eyes of patients with unilateral keratoconus progressed to keratoconus within 16 years with a greater risk during the first 6 years of onset.13

Eyes with keratoconus (167 eyes of 113 patients) were enrolled among patients referred to the National Reference Center for Keratoconus for a regular control visit for moderate to advanced keratoconus. Diagnosis of clinical keratoconus was previously defined and includes a combination of findings characteristic of keratoconus14,15: corneal topography with asymmetric bowtie pattern or localized steepening, irregular cornea determined by distortion of the retinoscopic or ophthalmoscopic red reflex, and at least one of the following slit-lamp findings: stromal thinning, Fleischer ring greater than 2 mm arc, Vogt striae, and corneal scarring consistent with keratoconus. Eyes that wore contact lenses and eyes that had undergone a specific treatment for keratoconus, such as collagen cross-linking, intracorneal rings, or keratoplasty, and marginal pellucid degeneration were excluded from the study.

Dual-Scheimpflug Analyzer System and Procedure

Measurements were performed with the GALILEI dual Scheimpflug analyzer system (software version 5.2.1; Ziemer Ophthalmic Systems, Port, Switzerland) according to the manufacturer’s guidelines. The device was first brought into focus (Placido rings into sharp focus) and aligned with the patient’s visual axis (central fixation light). Then patients were asked to blink just before the measurement. Only measurements that satisfied the minimum quality required by the system were included in this study.

The GALILEI Analyzer is a rotating Scheimpflug tomography-based device combining dual-channel Scheimpflug cameras and a Placido disk. The system acquires between 15 and 60 Scheimpflug images per scan and two Placido top view images at 90° apart, as the cameras rotate around the central axis. Placido and Scheimpflug data are acquired simultaneously, and then a motion correction algorithm is applied to the combined dataset. This correction compensates for the patient’s eye motion during scanning by a tracker that locates and tracks a patch on the iris, matching its location on every scan.

The combination of technologies, Placido disk and Scheimpflug imaging, has been designed to enhance the accuracy of the anterior corneal curvature calculation. Height data from the Scheimpflug images and slope data, converted into height data from the Placido disk, are merged to provide a surface fitted to the anterior corneal data, whereas posterior corneal surface data are measured using edge detection in images provided by the dual Scheimpflug system.

Analyzed Parameters and Description

All patients had both eyes imaged with the GALILEI Analyzer (although not all eyes were included in the study) and a detailed preoperative ophthalmic evaluation including uncorrected visual acuity, best spectacle-corrected visual acuity using Early Treatment of Diabetic Retinopathy Study charts, manifest refraction, slit-lamp evaluation, applanation tonometry, and fundus examination.

Elevation data were measured with two different reference surfaces over an 8-mm calculation zone: the BFS and the best-fit toric and aspherical surface (BFTA). For the BFS, the float mode, where the reference surface fits any direction so that the average distance between the corneal surface and the BFS is optimized to be as small as possible, was chosen. The BFTA model has been previously described with the following characteristics.16 The model is generated using the two orthogonal apical radii of maximum and minimum curvature of the cornea, thus reproducing the corneal astigmatism. Along each meridian, the curvature flattens with an elliptical progression from the center to the periphery according to the mean corneal eccentricity that is measured. Therefore, the fits of the asphero-toric model will be closer to the cornea from which it is computed (Figure 1). Values were collected on elevation maps of the anterior and posterior surfaces at the maximum value above the reference surface by manually guiding the cursor within the central 5-mm diameter zone. These parameters were called maximum anterior corneal elevation (MAE) with BFS or BFTA and maximum posterior corneal elevation (MPE) with BFS or BFTA. Elevations maps were all displayed with 5-μm color-coded scales.

Comparison of the best-fit sphere (BFS) and the best fit toric and aspheric surface (BFTA) when displaying elevation data for a normal asphero-toric cornea. Representation of a cross-section at the steepest axis of the cornea. Top (from left to right): Axial curvature map of a normal astigmatic cornea; the anterior elevation map calculated with the BFS shows a typical ridge pattern due to the corneal astigmatism; Representation of the BFS model in a cross-section passing through the steepest axis. Bottom (left to right): Same axial curvature map; the anterior elevation measured with the BFTA has minimized the effect of the astigmatism, so that there is no ridge pattern visible. Representation of the BFTA model in a cross section passing through the steepest axis.

Figure 1. Comparison of the best-fit sphere (BFS) and the best fit toric and aspheric surface (BFTA) when displaying elevation data for a normal asphero-toric cornea. Representation of a cross-section at the steepest axis of the cornea. Top (from left to right): Axial curvature map of a normal astigmatic cornea; the anterior elevation map calculated with the BFS shows a typical ridge pattern due to the corneal astigmatism; Representation of the BFS model in a cross-section passing through the steepest axis. Bottom (left to right): Same axial curvature map; the anterior elevation measured with the BFTA has minimized the effect of the astigmatism, so that there is no ridge pattern visible. Representation of the BFTA model in a cross section passing through the steepest axis.

Statistical Analysis

Comparisons of corneal elevation values were made between normal and forme fruste keratoconus, normal and keratoconus, and forme fruste keratoconus and keratoconus. Receiver operating characteristics (ROC) curves were used as described17 to assess the discriminating ability and to determine the optimal corneal elevation cut-off values obtained by different reference surfaces. Comparisons of area under the ROC curve values were made by the DeLong method as previously described18 to test differences between reference surface shapes and between anterior and posterior surface. In principle, P values less than .05 were considered statistically significant. To correct for multiples testing, P values were adjusted according to the Bonferroni–Holm procedure. All calculations were performed with STATA/SE (StataCorp 2005, version 9.0, College Station, TX).

Part of the statistical analyses concerned the question of dependence between observations. Because for most patients the data included both eyes, it seemed plausible that there would be intraclass correlation within the patient, so the whole database cannot be considered as independent observations. Therefore, we initially used the mixed model rather than analysis of variance for comparing groups because the mixed model takes account of the dependence among observations. However, results of this analysis indicated that there was no dependence, so the independence assumption was valid.

Results

Baseline demographic characteristics of the patients by groups are summarized in Table 1.

Demographic Characteristics of the Patients by Groups

Table 1: Demographic Characteristics of the Patients by Groups

Mean Corneal Elevation

Elevation values with BFTA were significantly lower than with BFS in the normal group, whereas it was the opposite in the forme fruste keratoconus and keratoconus groups. Anterior and posterior corneal elevation with the two reference surfaces was significantly higher in the keratoconus group than in the two other groups. Anterior and posterior elevation (MAE and MPE) with BFTA in the forme fruste keratoconus group was significantly higher than in the normal group, whereas elevation values with BFS were not statistically different between these two groups. The mean values of corneal elevation and clinical data in each group are summarized in Table 2.

Means and Intergroup Comparison of Anterior and Posterior Corneal Parameters

Table 2: Means and Intergroup Comparison of Anterior and Posterior Corneal Parameters

Discriminating Ability and ROC Curves

Cut-off values and discriminating ability of the analyzed corneal elevation parameters are summarized in Tables 3 to 5.

Optimized Cut-off Values and Corresponding Pairs of Sensitivity/Specificity for the Four Elevation Parametersa,b

Table 3: Optimized Cut-off Values and Corresponding Pairs of Sensitivity/Specificity for the Four Elevation Parameters,

Comparison of Discriminating Ability Between the Elevation Parameters for Distinguishing Between Normal and Keratoconus

Table 4: Comparison of Discriminating Ability Between the Elevation Parameters for Distinguishing Between Normal and Keratoconus

Comparison of Discriminating Ability Between the Elevation Parameters for Distinguishing Between Normal and Form Fruste Keratoconus

Table 5: Comparison of Discriminating Ability Between the Elevation Parameters for Distinguishing Between Normal and Form Fruste Keratoconus

Comparison Between Reference Surface Shapes

ROC curves showed an overall high predictive accuracy of both anterior and posterior corneal elevation for distinguishing between normal and keratoconus with an area under the ROC curve ranging from 0.973 to 0.997. However, the discriminating ability with BFTA achieved a significantly better performance than with BFS (Table 3). ROC curves of corneal elevation with different reference surfaces for discriminating between normal and keratoconus are showed in Figure 2.

Comparison of receiver operator characteristic curves of anterior (MAE) and posterior (MPE) corneal elevation with different reference surfaces (best-fit sphere [BFS] and the best-fit toric and aspheric surface [BFTA]) for the discrimination between normal corneas and keratoconus.

Figure 2. Comparison of receiver operator characteristic curves of anterior (MAE) and posterior (MPE) corneal elevation with different reference surfaces (best-fit sphere [BFS] and the best-fit toric and aspheric surface [BFTA]) for the discrimination between normal corneas and keratoconus.

The ability to discriminate between normal and forme fruste keratoconus with elevation parameters also significantly improved by using BFTA instead of the BFS reference surface. The area under the ROC curve increased from 0.63 in BFS to 0.80 in BFTA (P = .01) with the MAE, and from 0.59 to 0.88 (P = .01) with the MPE. ROC curves of corneal elevation with different reference surfaces for discriminating between normal and forme fruste keratoconus are represented in Figure 3. However, the sole use of corneal elevation parameters for discriminating between normal and forme fruste keratoconus was significantly weaker than for discriminating between normal and keratoconus (Table 3).

Comparison of receiver operator characteristic curves of anterior (MAE) and posterior (MPE) corneal elevation with different reference surfaces (best-fit sphere [BFS] and best-fit toric and aspheric surface [BFTA]) for the discrimination between normal corneas and forme fruste keratoconus.

Figure 3. Comparison of receiver operator characteristic curves of anterior (MAE) and posterior (MPE) corneal elevation with different reference surfaces (best-fit sphere [BFS] and best-fit toric and aspheric surface [BFTA]) for the discrimination between normal corneas and forme fruste keratoconus.

Comparison Between Anterior and Posterior Corneal Elevation

When comparing MAE and MPE values with the same reference surface, there was no difference for differentiating between normal and keratoconus. However, for discriminating between normal and forme fruste keratoconus, posterior elevation was significantly better than anterior elevation (P = .02) (Table 5) when using the BFTA. The sensitivity and specificity achieved with MPE were 82% and 80% by setting the cut-off value at 13 μm compared to 51% and 92% achieved with the MAE when setting the cut-off value at 8 μm. However, no difference was found between MAE and MPE when using the BFS (P = .49).

The cut-off values of anterior and posterior elevation for discriminating normal from keratoconus and forme fruste keratoconus with both reference surfaces are shown in Table 3.

Discussion

In addition to advancements in anterior segment imaging, several corneal indices based on elevation,8 thickness profile,19,20 or wavefront21 have been reported to improve the sensitivity of subclinical keratoconus detection. However, although differences in corneal elevation between normal corneas, mild keratoconus, and frank keratoconus have been extensively studied, few reports have addressed the influence of the reference surface shape on the detection of the mildest forms of keratoconus and its differentiation from normal.

In the current study, we demonstrated that the use of the BFTA surface for calculating elevation is better than the BFS for discriminating between normal and keratoconus and between normal and forme fruste keratoconus. Similarly, Kovács et al.22 reported that the use of a toric and ellipsoid reference surface (BFTE) with the Pentacam system (Oculus Optikgeräte GmbH, Wetzlar, Germany) enabled better discrimination between normal and keratoconus (P = .05) with an area under the ROC curve of 0.99 with the BFTE and 0.96 with the BFS.

When analyzing elevation data, it is essential to be aware of the influence of corneal toricity and asphericity on elevation calculation, and therefore on elevation map patterns. Using a multiple regression analysis, Kovács et al.22 demonstrated that corneal cylinder highly influenced elevation calculations when spherical reference surface (or BFS) was used as the reference surface as opposed to ellipso-toric fitting surface.

More recently, Gatinel et al.23 further highlighted the importance of toricity and asphericity when analyzing elevation data. The authors modeled the cornea as a biconic surface to analyze the effects of corneal toricity and asphericity on elevation map patterns generated relative to spherical surface (BFS). They demonstrated that the complete ridge pattern seen in elevation maps was induced by an increase in corneal toricity, whereas the incomplete ridge and island pattern reflected an increase in prolateness.

An ellipse surface is defined by its apical radius (on the vertex of the conic), which can be expressed in terms of a circle with the same degree of curvature and its asphericity, and represents the variation in curvature from the apex toward the periphery. Therefore, the ellipso-toric or asphero-toric model (BFTA) is generated by incorporating the difference in curvature between the two principal meridians (corneal cylinder) and the mean corneal asphericity of the measured cornea.16 With that method, it effectively approximates to the average corneal shape by minimizing the difference between the fitting surface and the cornea that is analyzed. This surface geometry has been considered a good model of the average topography of a normal asphero-toric cornea.16,23,24 However, in a real eye, it is assumed that this asphero-toric model is a simplified model, and in some cases of asymmetric and irregular corneal surface, the cornea differs from the perfectly smooth asphero-toric surface.16

Smolek and Klyce25 showed that a 4th-order Zernike polynomial would be reliable for modeling a normal cornea, whereas it would be inappropriate for modeling an abnormal corneal surface because of significant loss of fine-detail shape information. A 5th-order Zernike polynomial would be required for accurately fit corneas expressing more than 0.5 D of astigmatism and 7th and 12th order Zernike polynomials for reliably fitting suspected keratoconic and mild keratoconic corneas, respectively. Consequently, these findings suggest that the mildest forms of keratoconus may have substantial higher elevation irregularities that may not be highlighted with a simple spherical surface such as the BFS. In contrast, as described above, the aspherical and toric reference surface would lie closer to the natural corneal shape and neutralize, in some ways, the ridge pattern commonly seen in elevation maps due to the effect of corneal toricity. This way, it might help revealing the fine abnormalities that deviate from a regular asphero-toric surface and would be otherwise hidden by the ridge pattern seen in elevation maps calculated relative to a spherical surface (Figure 4). This difference between the BFS and BFTA displays becomes particularly relevant when tracking subtle abnormalities in elevation maps for detecting subclinical keratoconus.

Comparison of the best-fit sphere (BFS) and the best-fit toric and aspheric surface (BFTA) when displaying elevation data for a case of forme fruste keratoconus. Representation of a cross-section passing through the underlying conus area: (A) Normal topography aspect of a forme fruste keratoconus; (B) aspect of normal posterior elevation surface with the BFS display with the ridge pattern due to the astigmatism; (C) clear posterior elevation asymmetry revealed with BFTA fitting method with maximum posterior elevation value of + 16 μm.

Figure 4. Comparison of the best-fit sphere (BFS) and the best-fit toric and aspheric surface (BFTA) when displaying elevation data for a case of forme fruste keratoconus. Representation of a cross-section passing through the underlying conus area: (A) Normal topography aspect of a forme fruste keratoconus; (B) aspect of normal posterior elevation surface with the BFS display with the ridge pattern due to the astigmatism; (C) clear posterior elevation asymmetry revealed with BFTA fitting method with maximum posterior elevation value of + 16 μm.

In this study, we showed that posterior elevation had significantly better ability than anterior elevation for differentiating forme fruste keratoconus from normal corneas. This finding supports the hypothesis that posterior elevation is more sensitive than anterior elevation for identifying ectasia-susceptible corneas.8,26 However, it does not further clarify the potential differences or synergies for anterior curvature (topography) and posterior elevation (tonography). A leading current hypothesis is that keratoconic disease may be first detectable at the posterior surface, and modifications of the posterior surface are of particular interest in the mildest form of the disease in the sense that corneal epithelium has been shown to have the potential of smoothing corneal topographic irregularities27,28 and masking the presence of an underlying cone on the anterior surface of mild keratoconus.29

In the current study, the optimal posterior elevation cut-off value with BFTA was set at 13 and 16 μm for detecting keratoconus and forme fruste keratoconus, with 99% sensitivity and specificity for keratoconus and 82% sensitivity and 80% specificity for forme fruste keratoconus. De Sanctis et al. found a higher posterior elevation cut-off value of 29 μm using a fixed 9-mm diameter BFS with the Pentacam and achieved 68% sensitivity and 90.8% specificity for detecting subclinical keratoconus.9 Uçakhan et al. reported a posterior elevation cut-off value for detecting subclinical keratoconus of 20.5 μm using a similar setting and achieved 81.8% sensitivity and 66% specificity.10

The discrepancies in results, cut-off values, and predictive accuracy are likely due to four major points. First, although the GALILEI Analyzer is also a Scheimpflug-based imaging system, it has been demonstrated to lead to significantly lower values than Pentacam in both anterior and posterior elevation maps, which may in part explain our lower cut-off values.30 Second, our lower values are also explained by the different fitting method used in our study. Kovács et al. have shown that the use of an ellipso-toric reference surface influenced the magnitude and led to lower elevation value.22 Third, we included forme fruste keratoconus in our subclinical keratoconus group, which represents the earliest stage of the ectatic disease with a normal topographic aspect, whereas the studies by de Sanctis et al.9 and Uçakhan et al.10 included more advanced cases of keratoconus already expressing topographic manifestations such as asymmetric bowtie pattern. Finally, in the above-cited studies, the authors used a reference sphere set at 9-mm diameter in contrast to our study where both reference surfaces were set at 8 mm, and it is well known that a larger diameter of the reference surface results in greater elevation values.31

Therefore, by achieving comparable and even more sensitive performance than previous reports for detecting the mildest stage of the disease, the BFTA fitting method seems to be a useful and relevant approach for improving the detection of subclinical keratoconus. In addition, this fitting method has shown to better discriminate between normal and subclinical keratoconus than the classically used BFS. However, the sensitivity achieved for detecting forme fruste keratoconus by using this sole parameter remains insufficient for a screening program with 18% false-negative responses and should be considered in combination with other relevant parameters for identifying subclinical keratoconus.

References

  1. Randleman JB, Russell B, Ward MA, Thompson KP, Stulting RD. Risk factors and prognosis for corneal ectasia after LASIK. Ophthalmology. 2003;110:267–275 doi:10.1016/S0161-6420(02)01727-X [CrossRef] .
  2. Pallikaris IG, Kymionis GD, Astyrakakis NI. Corneal ectasia induced by laser in situ keratomileusis. J Cataract Refract Surg. 2001;27:1796–1802 doi:10.1016/S0886-3350(01)01090-2 [CrossRef] .
  3. Woodward M, Randleman JB, Russell B, Lynn MJ, Ward MA, Stulting RD. Visual rehabilitation and outcomes for ectasia after corneal refractive surgery. J Cataract Refract Surg. 2008;34:383–388 doi:10.1016/j.jcrs.2007.10.025 [CrossRef] .
  4. Klyce SD. Chasing the suspect: keratoconus. Br J Ophthalmol. 2009;93:845–848 doi:10.1136/bjo.2008.147371 [CrossRef] .
  5. Rabinowitz YS, McDonnell P. Computer-assisted corneal topography in keratoconus. Refract Corneal Surg. 1989;5:400–408.
  6. Rao SN, Raviv T, Majmudar PA, Epstein RJ. Role of Orbscan II in screening keratoconus suspects before refractive corneal surgery. Ophthalmology. 2002;6420:1642–1646 doi:10.1016/S0161-6420(02)01121-1 [CrossRef] .
  7. Schlegel Z, Hoang-Xuan T, Gatinel D. Comparison of and correlation between anterior and posterior corneal elevation maps in normal eyes and keratoconus-suspect eyes. J Cataract Refract Surg. 2007;34:789–795 doi:10.1016/j.jcrs.2007.12.036 [CrossRef] .
  8. Saad A, Gatinel D. Topographic and tomographic properties of forme fruste keratoconus corneas. Invest Ophthalmol Vis Sci. 2010;51:5546–5555 doi:10.1167/iovs.10-5369 [CrossRef] .
  9. de Sanctis U, Loiacono C, Richiardi L, Turco D, Mutani B, Grignoto FM. Sensitivity and specificity of posterior corneal elevation measured by Pentacam in discriminating keratoconus/subclinical keratoconus. Ophthalmology. 2008;115:1534–1539 doi:10.1016/j.ophtha.2008.02.020 [CrossRef] .
  10. Uçakhan ÖÖ, Cetinkor V, Özkan M, Kanpolat A. Evaluation of Scheimpflug imaging parameters in subclinical keratoconus, keratoconus, and normal eyes. J Cataract Refract Surg. 2011;37:1116–1124 doi:10.1016/j.jcrs.2010.12.049 [CrossRef] .
  11. Ambrósio R, Belin MW. Imaging of the cornea: topography vs tomography. J Refract Surg. 2010;26:847–849 doi:10.3928/1081597X-20101006-01 [CrossRef] .
  12. Belin MW, Khachikian SS. An introduction to understanding elevation-based topography: how elevation data are displayed–a review. Clin Experiment Ophthalmol. 2008;37:14–29 doi:10.1111/j.1442-9071.2008.01821.x [CrossRef] .
  13. Li X, Rabinowitz YS, Rasheed K, Yang H. Longitudinal study of the normal eyes in unilateral keratoconus patients. Ophthalmology. 2004;111:440–446 doi:10.1016/j.ophtha.2003.06.020 [CrossRef] .
  14. Rabinowitz Y. Keratoconus. Surv Ophthalmol. 1998;42:297–319 doi:10.1016/S0039-6257(97)00119-7 [CrossRef] .
  15. Zadnik K, Barr JT, Edrington TB, et al. Baseline findings in the Collaborative Longitudinal Evaluation of Keratoconus (CLEK) Study. Invest Ophthalmol Vis Sci. 1998;39:2537–2546.
  16. Calossi A. Corneal asphericity and spherical aberrations. J Refract Surg. 2007;23:505–514.
  17. Hanley J, McNeil B. The meaning and use of the area under receiver operating characteristic (ROC) curve. Radiology. 1982;143:26–36.
  18. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating curves: a non parametric approach. Biometrics. 1988;44:837–845 doi:10.2307/2531595 [CrossRef] .
  19. Ambrósio R, Alonso RS, Luz A, Coco Velarde LG. Corneal-thickness spatial profile and corneal-volume distribution: tomographic indices to detect keratoconus. J Cartaract Refract Surg. 2006;32:1851–1859 doi:10.1016/j.jcrs.2006.06.025 [CrossRef] .
  20. Ambrósio R, Caiado AL, Guerra FP, et al. Novel pachymetric parameters based on corneal tomography for diagnosing keratoconus. J Refract Surg. 2011;27:753–758 doi:10.3928/1081597X-20110721-01 [CrossRef] .
  21. Bühren J, Kook D, Yoon G, Kohnen T. Detection of subclinical keratoconus by using corneal anterior and posterior surface aberrations and thickness spatial profiles. Invest Ophthalmol Vis Sci. 2010;51:3424–3432 doi:10.1167/iovs.09-4960 [CrossRef] .
  22. Kovács I, Miháltz K, Ecsedy M, Németh J, Nagy ZZ. The role of reference body selection in calculating posterior corneal elevation and prediction of keratoconus using rotating Scheimpflug camera. Acta Ophthalmologica. 2011;89:251–256 doi:10.1111/j.1755-3768.2010.02053.x [CrossRef] .
  23. Gatinel D, Malet J, Hoang-Xuan T. Corneal elevation topography: best fit sphere, elevation distance, asphericity, toricity, and clinical implications. Cornea. 2011;30:508–515.
  24. Navarro R, González L, Hernández JL. Optics of the average normal cornea from general and canonical representations of its surface topography. J Opt Soc Am A Opt Image Sci Vis. 2006;23:219–232 doi:10.1364/JOSAA.23.000219 [CrossRef] .
  25. Smolek MK, Klyce SD. Goodness-of-prediction of Zernike polynomial fitting to corneal surfaces. J Cataract Refract Surg. 2005;31:2350–2355 doi:10.1016/j.jcrs.2005.05.025 [CrossRef] .
  26. Ambrósio R, Dawson DG, Salomão M, Guerra FP, Caiado AL, Belin MW. Corneal ectasia after LASIK despite low preoperative risk: tomographic and biomechanical findings in the unoperated, stable, fellow eye. J Refract Surg. 2010;26:906–911 doi:10.3928/1081597X-20100428-02 [CrossRef] .
  27. Gatinel D, Racine L, Hoang-Xuan T. Contribution of the corneal epithelium to anterior corneal topography in patients having myopic photorefractive keratectomy. J Cataract Refractive Surg. 2007;33:1860–1865 doi:10.1016/j.jcrs.2007.06.041 [CrossRef] .
  28. Rocha KM, Perez-Straziota CE, Stulting RD, Randleman JB. Spectral-domain OCT analysis of regional epithelial thickness profiles in keratoconus, postoperative corneal ectasia, and normal eyes. J Refract Surg. 2013;29:173–179 doi:10.3928/1081597X-20130129-08 [CrossRef] .
  29. Reinstein DZ, Gobbe M, Archer TJ, Silverman RH, Coleman DJ. Epithelial, stromal, and total corneal thickness in keratoconus: three-dimensional display with Artemis very-high frequency digital ultrasound. J Refract Surg. 2010;26:259–272 doi:10.3928/1081597X-20100218-01 [CrossRef] .
  30. Salouti R, Nowroozzadeh MH, Zamani M, Fard AH, Niknam S. Comparison of anterior and posterior elevation map measurements between 2 Scheimpflug imaging systems. J Cataract Refract Surg. 2009;35:856–862 doi:10.1016/j.jcrs.2009.01.008 [CrossRef] .
  31. Khachikian SS, Belin MW. Posterior elevation in keratoconus. Ophthalmology. 2009;116:816–817 doi:10.1016/j.ophtha.2009.01.009 [CrossRef] .

Demographic Characteristics of the Patients by Groups

Characteristics Normal (Fr/US) Forme Fruste Keratoconus (Fr/US) Keratoconus (Fr/US)
No. of eyes 177 (135 / 42) 47 (43 / 4) 167 (128 / 39)
No. of patients 95 (72 / 23) 47 (43 / 4) 113 (86 / 27)
Mean age ± SD 28.9 ± 8.6 31.8 ± 9.8 28.9 ± 10.0
Female sex (%) 105 (59.3%) 12 (25.5%) 48 (32.4%)

Means and Intergroup Comparison of Anterior and Posterior Corneal Parameters

Parameter Means ± SD (Range)
Intergroup Comparisons Mixed Model (P)a
Normal (177 Eyes/95 Patients) Forme Fruste Keratoconus (47 Eyes/47 Patients) Keratoconus (148 Eyes/102 Patients) Normal vs Forme Fruste Normal vs Keratoconus Forme Fruste vs Keratoconus
Anterior surface
  BFTA MAE (μm) 4.8 ± 1.7 (1 to 9) 8.4 ± 3.9 (1 to 19) 31.9 ± 15.8 (6 to 87) .03 < .001 < .001
  BFS MAE (μm) 5.4 ± 3.15 (0 to 15) 7.2 ± 4.3 (2 to 20) 25.2 ± 12.5 (6 to 64) .20 < .001 < .001
  Cylinder (D) 0.94 ± 0.64 (0.11 to 2.5) 1.3 ± 1 (0.3 to 5.5) 3.6 ± 2.2 (0.35 to 10.7) .31 < .001 < .001
  Eccentricity (∊2) 0.22 ± 0.12 (−0.05 to 0.5) 0.3 ± 0.3 (−0.6 to 1.1) 0.92 ± 0.8 (−1.5 to 2.81) .30 < .001 < .001
Posterior surface
  BFTA MPE (μm) 8.6 ± 2.8 (4 to 17) 16.9 ± 6.9 (4 to 39) 57.8 ± 28.4 (10 to 135) .004 < .001 < .001
  BFS MPE (μm) 13.1 ± 5.2 (3 to 30) 15.4 ± 6.5 (3 to 34) 46 ± 20.6 (6 to 109) .20 < .001 < .001
  Cylinder (D) −0.3 ± 0.11 (−0.7 to −0.1) −0.3 ± 0.16 (−0.7 to −0.1) −0.7 ± 0.3 (−1.7 to −0.1) .99 < .001 < .001
  Eccentricity (∊2) 0.19 ± 0.19 (−0.2 to 0.8) 0.32 ± 0.3 (−0.4 to 1.09) 1.3 ± 0.96 (−1.3 to 3.6) .14 < .001 < .001

Optimized Cut-off Values and Corresponding Pairs of Sensitivity/Specificity for the Four Elevation Parametersa,b

Parameter Sensitivity (%)
Specificity (%)
Cut-off Values (μm)
Normal vs Form Fruste Keratoconus Normal vs Keratoconus Normal vs Form Fruste Keratoconus Normal vs Keratoconus Normal vs Form Fruste Keratoconus Normal vs Keratoconus
MAE BFS 62% 92% 63% 91% 6 11
MAE BFTA 66% 98% 86% 99% 7 9
MPE BFS 51% 93% 55% 95% 14 21
MPE BFTA 82% 99% 80% 99% 13 16

Comparison of Discriminating Ability Between the Elevation Parameters for Distinguishing Between Normal and Keratoconus

Parameters AUROC 95% CI MAE BFS Normal vs Keratoconus (P)
MAE BFTA MPE BFS MPE BFTA
MAE BFS 0.98 0.96 to 0.99
MAE BFTA 0.99 0.99 to 1 .01
MPE BFS 0.97 0.95 to 0.99 .25 .008
MPE BFTA 0.99 0.99 to 1 .01 .05 .01

Comparison of Discriminating Ability Between the Elevation Parameters for Distinguishing Between Normal and Form Fruste Keratoconus

Parameters AUROC 95% CI MAE BFS Normal vs Form Fruste Keratoconus (P)
MAE BFTA MPE BFS MPE BFTA
MAE BFS 0.63 0.54 to 0.72
MAE BFTA 0.80 0.72 to 0.88 .01
MPE BFS 0.59 0.50 to 0.69 .49 .01
MPE BFTA 0.88 0.82 to 0.95 .01 .03 .01

10.3928/1081597X-20130318-07

Sign up to receive

Journal E-contents