CORNEAL ENDOTHELIAL CELL LOSS NINE YEARS AFTER EXCIMER LASER KERATOREFRACTIVE SURGERY

Abstract

Purpose

To determine the long-term changes in the corneal endothelium after LASIK and PRK.

Methods

Twenty-nine eyes (16 patients) received myopic LASIK or PRK, with intended correction to emmetropia. Central endothelial photographs were taken before and 9 years after surgery, and were analyzed by the same masked investigator after appropriate calibration for magnification. Comparisons were made by using generalized estimating equation models to account for any correlation between fellow eyes of the same subject. The annual exponential rate of cell loss was compared to cell loss over a ten-year period in 42 normal (unoperated) corneas of 42 subjects.

Results

Endothelial cell density at 9 years after LASIK and PRK decreased by 5.3% from preoperative (P<0.001, n=29), whereas coefficient of variation of cell area and percentage of hexagonal cells did not change (P=0.24 and P=0.19, respectively, n=29). The annual rate of cell loss after refractive surgery (0.6 ± 0.8 %) was not different from that in normal corneas (0.6 ± 0.5 %, P=0.88; minimum detectable difference = 0.5%, α=0.05, β=0.20).

Conclusions

LASIK and PRK had no long-term effect on the corneal endothelium. Corneas that have received LASIK or PRK can be considered as donors for posterior lamellar keratoplasty procedures.

INTRODUCTION

Photoablation of the corneal stroma by excimer lasers is a method of correcting refractive errors. Laser in situ keratomileusis (LASIK) requires the creation of an anterior corneal flap and photoablation of the mid-stroma, whereas photorefractive keratectomy (PRK) involves epithelial removal and anterior stromal photoablation. The long-term effects of photoablation on epithelial and stromal thickness,¹ keratocyte density² and corneal nerve regeneration³ are known, but the long-term effect on the corneal endothelium has not been reported. There have been conflicting reports of the effect of photoablation on the corneal endothelium in the short-term with most studies finding no effect, ^4-11 although a minority of studies have indicated endothelial cell loss higher than age-related physiologic cell loss,^12-14 possibly because of mechanical trauma from shockwaves, local oxidative changes, or thermal effects.^15-18

In this study, we examined changes in the corneal endothelium nine years after excimer laser keratorefractive surgery, and made comparisons to changes in the normal (unoperated) corneal endothelium over a ten-year period. In addition, we explored the relationships between endothelial cell loss and ablated and residual bed thicknesses.

METHODS

Subjects

Twenty-nine eyes of 16 patients that had endothelial photographs taken before myopic keratorefractive surgery at Mayo Clinic between July 1998 and January 1999 were re-examined with endothelial photography nine years later. Twenty eyes of 10 patients received LASIK, and 9 eyes of 6 patients received PRK. Exclusion criteria included diabetes mellitus or other significant systemic disorders, glaucoma or ocular hypertension (≥22 mm Hg), use of any ocular medications, use of any systemic medications known to have adverse effects in the cornea, and any ocular surgery, with the exception of refractive surgery enhancement procedures, in the period between examinations. A LASIK enhancement procedure was performed in 8 eyes of 5 patients for under-correction, and a PRK enhancement procedure was performed in one eye.

For LASIK patients, age at surgery was 34 ± 8 years (mean ± standard deviation; range, 23-47 years) and mean preoperative spheroequivalent refractive error was −6.2 ± 1.4 D (range, −4.0 to −9.25 D). Eight of the 10 patients were current contact lens wearers whereas the other 2 patients had worn contact lenses in the past. For PRK patients, age at surgery was 39 ± 6 years (range, 31-44 years) and mean preoperative spheroequivalent refractive error was −3.5 ± 1.7 D (range, −1.25 to −5.75 D). Five of the 6 patients were current contact lens wearers and the other patient had worn contact lenses in the past.

This study adhered to the tenets of the Declaration of Helsinki, and was approved by the Mayo Clinic Institutional Review Board. Informed consent was obtained from all subjects after explanation of the nature and possible consequences of the study.

Refractive surgery procedures

For LASIK, the flap was created by using a mechanical microkeratome (Hansatome, Bausch & Lomb, Rochester, NY) with a flap diameter of 8.5 mm or 9.5 mm and an intended thickness of 180 μm. For PRK, the epithelium was removed by using the laser-scrape technique. For LASIK and PRK, the stroma was ablated with a VISX Star excimer laser (VISX, Santa Ana, CA). Emmetropia was attempted in all cases by using an ablation zone that ranged from 6.0 × 6.0 mm for spherical corrections to 4.5 × 6.0 mm for some astigmatic corrections.

Endothelial Cell Analysis

Before surgery, the central corneal endothelium was photographed with a non-contact specular microscope (Konan Non-Robo SP8000, Konan Medical Inc., Hyogo, Japan). With the patient's head stabilized by using the chin and forehead rests, the observer aligned the microscope with the center of the cornea and used the automatic function to capture a focused image of the central endothelium. Images were stored to digital media. Images consisted of 640 x 480 pixels (horizontal × vertical), which corresponded to a field size of 284 μm × 364 μm (0.103 mm²) based on calibration measurements of a micrometer slide.

At nine years after surgery, the endothelium of the same corneas was photographed by using a contact in vivo confocal microscope (ConfoScan 4, Nidek Technologies, Greensboro, NC). With the patient's head stabilized, the observer aligned the objective lens, coated with optical coupling medium (GenTeal Gel, Novartis Pharmaceuticals Corp., East Hanover, NJ), with the center of the cornea and digital images of the central endothelium were recorded. Images consisted of 768 × 576 pixels and were rescaled to 640 × 480 pixels for analysis; image field size was 434 μm × 320 μm (0.139 mm²) based on calibration measurements of the same micrometer slide used to calibrate the specular microscope.

Digital images were transferred to an image analysis system (KSS-400, Konan Medical USA, Torrance, CA) and the center of each endothelial cell was digitized by one observer who was masked to the identity of the images. 50-100 cells were counted in each endothelial photograph depending on image quality. The analysis program calculated the mean cell area and its reciprocal, endothelial cell density (ECD), the coefficient of variation (standard deviation/mean) of cell area (CV), and the percentage of hexagonal cells (HEX).

Measurement of ablated and residual bed thicknesses

A Tandem Scanning Confocal Microscope (Tandem Scanning, Reston, VA) was used to examine corneas in vivo before and at 1 month after surgery as described in detail previously.^1,19 All confocal scans were manually reviewed and scans with the least lateral ocular movement, and with no anteroposterior movement of the cornea relative to the objective were selected for analysis. An intensity profile of backscattered light was generated from the confocal images of the selected scan.^20,21 Peaks in the light intensity profiles of preoperative and postoperative corneas corresponded to the superficial epithelium, the endothelium, the subbasal nerve plexus, and the most anterior keratocytes.^1,19,20 The video image corresponding to each intensity peak was displayed. Profiles generated from corneas after LASIK also showed a peak corresponding to the lamellar interface, and this was typically confirmed by the presence of interface debris in the corresponding video image.¹⁹

The thickness of the ablated tissue was calculated as the difference between preoperative and postoperative stromal thickness. Before PRK, stromal thickness was defined as the distance from the subepithelial plexus to the endothelium (and therefore included Bowman's layer and Descemet's membrane); after PRK, and before and after LASIK, stromal thickness was defined as the distance from the most anterior keratocyte to the endothelium (after PRK, Bowman's layer was not present, and after LASIK, Bowman's layer remained unchanged compared to before LASIK). Residual bed thickness was defined as the distance between the most anterior keratocyte and the endothelium after PRK and as the distance between the lamellar interface and the endothelium after LASIK.

Data Analysis

Corneal ECD, CV and HEX were compared before and nine years after surgery. Comparisons were made by using generalized estimating equation models to adjust for any correlation between fellow eyes of the same patient.²² P< 0.05 was considered statistically significant.

We calculated the annual rate of corneal endothelial cell loss by assuming cell loss occurred as a first-order exponential process according to the following relationship:

$${\mathrm{ECD}}_9={{\mathrm{ECD}}_0}^{\mathrm{e}-kt}$$

where ECD₀ = endothelial cell density before surgery, ECD₉ = endothelial cell density at 9 years after surgery, k = exponential rate constant, and t = time between examinations in years. We compared the annual exponential rate of cell loss after keratorefractive surgery to cell loss in 42 normal, unoperated corneas of 42 subjects who were examined twice over a 10-year period, as previously reported.²³ At the first examination, the normal subjects’ age was 50 ± 17 years (mean ± standard deviation, range, 20 to 74 years); over the 10 years between repeat examinations, none of the normal subjects wore contact lenses or developed diabetes, and none had any ocular surgery.²³ In the previous paper,²³ because the exponential rate constants k were small, they approximated and were used as the annual exponential rates of cell loss (this approximation gives an error of 0.001 percent per year when the rate of decline is 0.5 percent per year).

Correlations between the percentage of endothelial cell loss at 9 years after refractive surgery (compared to preoperative) and the preoperative spheroequivalent refractive error, measured ablation, and residual bed thicknesses at 1 month after surgery were assessed by using the Pearson correlation coefficient if the data were distributed normally and by the Spearman test if they were not; significances were completed by using generalized estimating equation models.

RESULTS

For all eyes, ECD was 5.3% lower at 9 years after refractive surgery compared to preoperative (P<0.001, n=29), whereas no differences were found for CV or HEX (Table 1). In eyes that received LASIK, ECD decreased by 6.3% (P<0.001, Table 1), and in eyes that received PRK, CV and HEX improved over the 9-year period (P=0.009 and P=0.01, respectively, Table 1). At 9 years after LASIK, ECD in eyes that received an enhancement procedure did not differ from ECD in eyes that did not receive an enhancement procedure (Table 2).

Table 1.

Endothelial cell density and morphology before and after keratorefractive surgery

	Endothelial Cell Parameters (mean ± standard deviation)
	ECD (cells/mm2)	CV	HEX (%)
All eyes (n=29)
Preoperative	2,837 ± 336	0.34 ± 0.05	54 ± 8
9 years	2,685 ± 351	0.33 ± 0.03	56 ± 5
P	<0.001	0.24	0.19
MDD	-	0.03	5.7
LASIK eyes (n=20)
Preoperative	2,925 ± 303	0.33 ± 0.05	55 ± 9
9 years	2,741 ± 308	0.33 ± 0.03	56 ± 5
P	<0.001	0.98	0.77
MDD	-	0.04	7.4
PRK eyes (n=9)
Preoperative	2,641 ± 340	0.36 ± 0.03	51 ± 5
9 years	2,559 ± 423	0.32 ± 0.04	58 ± 5
P	0.45	0.009	0.01
MDD	230	-	-

ECD, endothelial cell density; CV, coefficient of variation of cell area; HEX, percentage of hexagonal cells.

Comparisons between preoperative and 9 years were made by using generalized estimating equation models to account for any correlation between fellow eyes of the same patient. MDD, minimum detectable difference for non-significant comparisons (α=0.05, β=0.20).

Table 2.

Endothelial parameters at 9 years after LASIK in eyes with or without an enhancement procedure.

	Endothelial Cell Parameters (mean ± standard deviation)
	ECD (cells/mm2)	CV	HEX (%)
No enhancement (n=12)	2,670 ± 340	0.33 ± 0.03	57 ± 6
Enhancement (n=8)	2,847 ± 234	0.34 ± 0.04	53 ± 2
P	0.89	0.53	0.04
MDD	410	0.04	-

ECD, endothelial cell density; CV, coefficient of variation of cell area; HEX, percentage of hexagonal cells.

Comparisons were made by using generalized estimating equation models to account for any correlation between fellow eyes of the same patient. MDD, minimum detectable difference for non-significant comparisons (α=0.05, β=0.20).

For all eyes after refractive surgery, the annual rate of endothelial cell loss was 0.6 ± 0.8 % (mean ± standard deviation, n=29; range-1.1 to 2.2). This did not differ significantly from the annual rate of cell loss of 42 normal (unoperated) corneas of 42 adults, which was 0.6 ± 0.5 % (P=0.88, minimum detectable difference= 0.5 %, α=0.05, β=0.20). CV at the first examination of the normal corneas (0.26 ± 0.05, n=42) was lower than the corneas at 9 years after refractive surgery (0.33 ± 0.03, n=29; P<0.001), and HEX at the first examination of the normal corneas (67 ± 8 %, n=42) was higher than the corneas at 9 years after refractive surgery (56 ± 5 %, n=29; P<0.001).

Measured ablation depth was 55 ± 35 μm (n=29) and residual bed thickness was 329 ± 55 μm (n=29) (Table 3). There was no correlation between the percentage of endothelial cell loss from preoperative and preoperative spheroequivalent refractive error (r= 0.01, P=0.96, n=29), measured ablation (r= 0.17, P=0.39, n=29) or residual bed thickness (r= −0.25, P=0.20, n=29).

Table 3.

Measured ablation and residual bed thicknesses after LASIK and PRK measured by confocal microscopy in vivo.

	Thickness, urn (mean ± standard deviation)
	Measured Ablation	Residual Bed
All eyes (n=29)	55 ± 35	329 ± 55
LASIK eyes (n=20)	66 ± 34	302 ± 31
PRK eyes (n=9)	31 ± 19	389 ± 49

DISCUSSION

The annual rate of endothelial cell loss after LASIK and PRK did not differ from the annual rate of age-related endothelial cell loss in normal, unoperated corneas. Even though our study was small, we had sufficient statistical power to detect a 0.5% difference in the rate of cell loss between these two groups of eyes. Our results support the findings of numerous short-term studies which found no significant endothelial cell loss after LASIK^5,7,8,10 and PRK.^4,6,9,11 There have been few studies that have examined endothelial cell loss at five or more years after excimer photoablative surgery; Kato et al. noted 1.2% cell loss at 5 years after LASIK and indicated this was within the physiologic age-related cell loss.²⁴ We found no relationship between endothelial cell loss and either the thickness of ablated tissue or the residual bed thickness, indicating that the deeper stromal ablations with LASIK compared to PRK did not affect the endothelium. A safe residual bed thickness was respected in all cases, and this probably protects the endothelium as demonstrated by animal studies in which endothelial cell loss only occurred when the corneal stroma was photoablated within 40 microns of Descemet's membrane.^16,17

We did not find changes in endothelial cell morphology when we combined eyes that received LASIK or PRK, but we did notice an improvement in the CV and HEX in the small number of eyes that received PRK. While the improvement in morphology after PRK could be attributed to cessation of contact lens wear, we would have expected a similar improvement in the eyes that received LASIK, because most of these eyes had also worn contact lenses prior to surgery. Contact lens wear is known to induce morphologic changes in the corneal endothelium without affecting cell density, ^25-28 but it is not known how quickly, if at all, the morphologic changes reverse after cessation of lens wear.^29,30 Improvement in endothelial cell morphology has been described after LASIK,^5,10 although Collins et al. were unable to associate this with the cessation of contact lens wear in a multivariate analysis.⁵

The importance of the findings in our study relates to the use of corneas that have received LASIK or PRK as donor tissue. When keratorefractive surgery started to increase in popularity, the Cornea Donor Study was initiated in the United States to address the concern that the supply of donor tissue would diminish because donor corneas would be unusable for penetrating keratoplasty after refractive surgery.³¹ While the latter was clearly a concern for penetrating keratoplasty, in recent years, posterior lamellar keratoplasty techniques (such as Descemet-stripping with endothelial keratoplasty) have become the preferred method for treating corneal endothelial dysfunction.^32-34 Because posterior lamellar keratoplasty does not involve transplantation of the anterior cornea, eye banks in the United States do accept donor corneas that have had excimer laser keratorefractive surgery for posterior lamellar keratoplasty. Our finding of no difference in endothelial cell loss after keratorefractive surgery compared to normal suggests that corneas after keratorefractive surgery should be suitable for posterior lamellar keratoplasty. The increased polymegethesim and pleomorphism that we found in corneas after refractive surgery compared to normal unoperated corneas could result from a small sample size, or might represent increased stress to the endothelium³⁵ related to prior contact lens wear. Nevertheless, potential corneal donors are not excluded by eye banks because of increased polymegethism or pleomorphism, or because of a history of contact lens wear. Although none of the LASIK eyes in our study had flaps created with a femtosecond laser, femtosecond laser preparation of donor tissue for posterior lamellar keratoplasty has not been associated with endothelial cell loss³⁶; nevertheless, long-term data of endothelial cell loss after femtosecond laser-assisted LASIK are warranted to ensure that the combination of femtosecond and excimer laser energy are not detrimental to the endothelium. Whether eye banks should accept corneas that have had previous PRK with adjuvant mitomycin-C for donation is unknown; some studies have indicated no effect of PRK with mitomycin-C on the endothelium^37-39 whereas other studies have found detrimental effects, including increased endothelial cell loss.^40-42 However, determining which potential donor corneas have had PRK is a challenge for eye banks,^43-45 and determining which corneas also received adjuvant mitomycin-C will be more difficult.

The major limitation of our study was the small number of eyes available for analysis. Nevertheless, we had adequate statistical power to be confident that cell loss after refractive surgery is no higher than physiologic age-related cell loss. Our enhancement data were limited by sample size and statistical power, and no conclusions can be drawn from our data. Although we used specular microscopy to photograph the corneal endothelium before surgery and confocal microscopy after surgery, we previously reported that careful calibration of both instruments for image magnification resulted in interchangeable data.⁴⁶ With both specular and confocal microscopy, only a very small proportion of the corneal endothelium can be visualized, introducing variation of repeated cell density measurements,⁴⁷ and limiting the analysis to one region of the cornea (central cornea in this study). Isager et al. found that magnification of specular microscopes decreased <1% (for most cases of excimer refractive surgery) with decreased corneal thickness, which could result in an overestimation of postoperative ECD⁴⁸; we did not adjust our data for corneal thickness, but if we had, the cell loss at 9 years compared to preoperative would have been approximately 0.5% more than indicated, which would not have altered any of our conclusions.

In summary, corneal endothelial cell loss at 9 years after LASIK and PRK (without mitomycin-C) did not differ from age-related cell loss found in normal corneas. Eye banks and surgeons can consider donor corneas that have had LASIK or PRK for posterior lamellar keratoplasty.