To determine the accuracy of total corneal astigmatism measurements with a Scheimpflug imager and a color light-emitting diode corneal topographer, and to compare the accuracy of total corneal astigmatism measurements with the accuracy of measurements that are based only on the anterior corneal surface.
Prospective validity assessment.
This study was conducted at the Rotterdam Ophthalmic Institute, Rotterdam, Netherlands. The study population consisted of 91 eyes of 91 patients with monofocal, non-toric intraocular lenses (IOLs). Refractive astigmatism was measured with the ARK-530A autorefractor (Nidek, Gamagori, Japan). Anterior and total corneal astigmatism were measured with the Pentacam HR (Oculus, Wetzlar, Germany) and the Cassini (i-Optics, The Hague, Netherlands). Under the assumption that refractive astigmatism must equal total corneal astigmatism in these patients, accuracy of the corneal astigmatism measurements was defined as the vectorial difference with the refractive astigmatism, with lower vector differences denoting higher accuracy.
The median refractive astigmatic magnitude was 0.84 diopter (D). The mean difference vector lengths were 0.61 D, 0.58 D, 0.49 D, and 0.45 D for Pentacam anterior, Cassini anterior, Pentacam total, and Cassini total corneal astigmatism, respectively. The mean difference vector length decreased by 0.12 and 0.13 D for Pentacam and Cassini, respectively, if the total instead of anterior corneal astigmatism was measured. These decreases were statistically significant ( P < .001).
With Pentacam as well as with Cassini, the accuracy of total corneal astigmatism measurements was higher than that of anterior corneal astigmatism measurements. Measuring total instead of anterior corneal astigmatism may therefore decrease the residual astigmatism in toric IOL implantation.
Toric intraocular lenses (IOLs) have become an increasingly popular means of correcting corneal astigmatism during cataract surgery. A high number of patients achieve good uncorrected distance visual acuity with this type of IOL. Accurate measurement of corneal astigmatism is essential for achieving good results. Traditionally, only the anterior surface of the cornea has been taken into account in these measurements. It has recently been shown, however, that ignoring astigmatism of the posterior corneal surface is the most important cause of residual refractive astigmatism after cataract surgery. With the advent of Scheimpflug imaging in ophthalmology for imaging of the anterior chamber, the contribution of the posterior corneal surface to total corneal astigmatism has gained a lot of interest. The posterior corneal surface acts as a negative lens and has a mean astigmatism of approximately 0.3 diopter (D), with the steepest meridian oriented vertically (“with-the-rule”) in most cases. If only the anterior corneal surface is taken into account in toric IOL selection, this causes corneal astigmatism to be overestimated when the anterior steep meridian is oriented vertically (“with-the-rule”) and underestimated when the anterior steep meridian is oriented horizontally (“against-the-rule”). Koch and associates proposed a nomogram to account for posterior corneal astigmatism in such cases. Because this nomogram assumes an average configuration of the posterior corneal surface, it will only be valid in eyes in which this assumption is met. Therefore, a better alternative might be to implement a device that reliably measures total corneal astigmatism, such as a Scheimpflug imager, in the practice of toric IOL implantation.
Recently, another device capable of measuring total corneal astigmatism has been released. The Cassini (i-Optics, The Hague, Netherlands) uses specular reflections of 679 colored light-emitting diodes (LEDs) to construct topographic maps of the anterior corneal surface, and the specular reflections of 7 additional infrared LEDs to measure the curvatures of the posterior corneal surface. Using ray tracing, it can then calculate total corneal astigmatism. Although anterior keratometry with the Cassini has been described in the literature, we are not aware of any reports evaluating its total corneal astigmatism function.
Although Scheimpflug devices have been used in research to investigate the properties of the posterior corneal surface (see above), their accuracy in measuring total corneal astigmatism has not yet been assessed. Therefore, the purpose of this study was to determine the accuracy of total corneal astigmatism measurements with Scheimpflug imaging (Pentacam HR, Oculus, Wetzlar, Germany) and with color LED corneal topography (Cassini). In addition, we compared the accuracy of total corneal astigmatism measurements with the accuracy of measurements that are based only on the anterior corneal surface. In order to do this, we included patients in whom a non-toric IOL was implanted. In such patients, the only optical component in the eye giving rise to astigmatism is the cornea. Therefore, the refractive astigmatism must be equal to the total corneal astigmatism in these eyes. Accuracy of corneal measurements can then be assessed by comparing anterior and total corneal astigmatism to the refractive astigmatism of the cornea.
This prospective validity assessment was conducted at the Rotterdam Ophthalmic Institute, Rotterdam, Netherlands. Ninety-one eyes of 91 patients who underwent cataract surgery with implantation of a monofocal, non-toric IOL were enrolled. If both eyes of a patient were eligible for inclusion, the eye that was operated most recently was selected. Exclusion criteria were corneal diseases (eg, keratoconus, corneal scarring, pterygium), pseudoexfoliation syndrome, previous corneal or refractive surgery, contact lens wear, any significant comorbid ocular disease, or inability to fixate on the fixation target of the measurement devices. In addition, patients were excluded if complications occurred during the surgery that could influence the outcome variables, such as a capsular tear or zonular rupture causing tilt or decentration of the IOL. The study adhered to the tenets of the Declaration of Helsinki. Approval for the study was prospectively obtained from the institutional review board of the Erasmus Medical Center in Rotterdam, and informed consent was received from all subjects.
A 2.2 mm self-sealing clear corneal incision or scleral tunnel incision was made with a standard dual-bevel slit knife (INTREPID 2.2; Alcon, Fort Worth, Texas, USA). Incisions were either placed superiorly (at 100 degrees) or at the steepest corneal meridian. After routine cataract extraction by phacoemulsification and bimanual cortex removal, a hydrophobic acrylic single-piece IOL (Acrysof SA60AT; Alcon) was inserted in the capsular bag using a standard Monarch III injector (Alcon). This IOL has a 6.0-mm-diameter monofocal non-toric optic, an overall length of 13.0 mm, and no haptic angulation.
All measurements were performed approximately 4 weeks postoperatively. Refractive astigmatism was measured with the ARK-530A autorefractor (Nidek, Gamagori, Japan). Corneal astigmatism was measured with 2 devices. The Pentacam HR (software version 6.07r12) is a Scheimpflug imaging system that generates tomographic images of the anterior chamber and uses edge detection to find the curvatures of the anterior and posterior corneal surfaces. It offers several options to calculate total corneal astigmatism, including True Net Power, Equivalent K-readings, and Total Refractive Power. Of these, we used the Total Refractive Power option in the 3.0-mm zone, because this option uses ray tracing to combine the contributions of both corneal surfaces and therefore most realistically reconstructs total corneal power. Moreover, this option most closely resembles the total corneal astigmatism calculation method of the Cassini. The Cassini (software version 2.2.0 beta) analyzes specular reflections of 679 colored LEDs on the anterior corneal surface and reflections of 7 infrared LEDs on the posterior surface to calculate the curvatures of both corneal surfaces at the 3.0-mm central zone. Total corneal astigmatism is calculated using ray tracing.
For each measurement on each device, the subjects were asked to fixate on the fixation target of the instrument and blink before the measurement. We used the internal quality check of each device to decide whether or not the measurement was acceptable. For the Pentacam, measurements were accepted if the quality check indicated that the measurement was “OK.” For the Cassini, measurements were accepted if at least 6 of the 7 infrared reflections were captured, as indicated by the posterior surface quality indicator. If not acceptable, the measurement was repeated until it was acceptable or until no acceptable measurements could be obtained after several repeated attempts.
Initial Data Processing
Refractive astigmatism was converted to the corneal plane using the method described by Holladay and associates. To calculate anterior corneal surface astigmatism for the measurements with the Pentacam and the Cassini, anterior radii of curvature were converted to meridional power using a keratometric index of 1.3375. Total corneal astigmatism was calculated by the Pentacam and Cassini using ray tracing, in which Snell’s law is applied to calculate refraction of a large number of light “rays” when incident on the anterior and, subsequently, the posterior corneal surface. For the Pentacam, these values can be found under “Total Refractive Power” (TRP) (we used the values calculated at the 3.0-mm zone), whereas for the Cassini they can be found under “Total Corneal Astigmatism” (TCA).
The accuracy of the measurements that took into account only the anterior corneal surface, as well as of those that took into account both corneal surfaces, was calculated as the difference of these measurements with the refractive astigmatism. The smaller the difference, the better the accuracy. The refractive astigmatism was considered the “gold standard” because it consists of the astigmatism of all refractive surfaces of the eye combined. Since a non-toric IOL was inserted into the eye, the only refractive surfaces that can be astigmatic are the anterior and posterior corneal surface. Thus, refractive astigmatism must equal the astigmatism that results from both corneal surfaces. To calculate the accuracy, all astigmatism measurements were decomposed into orthogonal X and Y components (ie, a vector) according to the method described by Holladay and associates. Accuracy was inversely related to the length of the vector denoting the difference between the “gold standard” vector (refractive astigmatism) and the vector representing the measurement of interest. This will be called the “difference vector length.”
Statistical analysis was performed using Excel 2010 (Microsoft, Redmond, Washington, USA) and SPSS version 21 software (IBM, Armonk, New York, USA). Normality of the distribution of all outcome variables was checked by inspection of histograms and using the Kolmogorov-Smirnov test. All variables except astigmatic magnitude were normally distributed. Thus, difference vector lengths were presented as mean ± standard deviation. The statistical significance of the change in difference vector length owing to the use of total instead of anterior corneal astigmatism measurements was calculated using the 1-sample t test. P values < .05 were considered statistically significant.
Sample Size Calculation
The smallest change in difference vector length that might be clinically relevant was defined to be 0.125 D. Based on pilot data of 24 patients, the standard deviation of this change was estimated to be 0.24 D. Setting α to 0.05 and power (1-β) to 0.9, the minimum required sample size was calculated to be 39 eyes.
Mean patient age was 69.4 ± 8.4 years (range 45.9–90.0). Forty-one patients (45%) were men, and 50 eyes (55%) were right eyes. The mean preoperative axial length was 23.92 ± 1.39 mm (range, 20.89–27.63), and the mean preoperative corneal power was 43.64 ± 1.48 D (range, 40.30–47.44). Measurements of acceptable quality could be obtained for the Pentacam and Cassini in 78 (86%) and 60 eyes (66%), respectively. An overview of the astigmatism magnitudes as measured with the different devices is given in Table 1 .
|Device||N||Anterior Surface |
Median [IQR] (Range) (D)
|Posterior Surface |
Median [IQR] (Range) (D)
|Total Cornea |
Median [IQR] (Range) (D)
|Autorefractor||91||–||–||0.84 [0.49–1.25] (0.12–3.51)|
|Pentacam||78||0.75 [0.48–1.16] (0.06–2.52)||0.40 [0.28–0.51] (0.06–1.14)||0.95 [0.50–1.30] (0.20–2.60)|
|Cassini||60||0.91 [0.52–1.34] (0.12–2.30)||0.35 [0.23–0.52] (0.06–1.17)||0.90 [0.55–1.29] (0.23–2.40)|
The difference vector lengths for the various techniques are shown in Table 2 . Shorter difference vectors indicate that the measurements are closer to the refractive astigmatism and therefore more accurate. It should be noted that when only the anterior corneal surface is taken into account, the mean difference vector length is approximately 0.6 D for both devices. Also, for Pentacam and Cassini, the mean difference vector length becomes shorter when both corneal surfaces are taken into account. Figure 1 shows for each technique the percentage of patients within a certain accuracy. Higher proportions of eyes fall within limits of 0.25, 0.50, 0.75, and 1.00 D when both corneal surfaces are measured. Table 3 shows the difference in astigmatic magnitude and axis with the refractive astigmatism for the various techniques. The magnitude difference decreases up to 0.11 D and the axis difference decreases up to 6.4 degrees when both corneal surfaces are taken into account instead of only the anterior surface.