American Journal of Ophthalmology
Volume 141, Issue 5 , Pages 799-809.e1, May 2006

Corneal Keratocyte Deficits After Photorefractive Keratectomy and Laser In Situ Keratomileusis

  • Jay C. Erie, MD

      Affiliations

    • Department of Ophthalmology, Mayo Clinic College of Medicine, Rochester, Minnesota, USA.
    • Corresponding Author InformationInquiries to Jay C. Erie, MD, Mayo Clinic, 200 First Street SW, Rochester, MN 55905
    • ,
  • Sanjay V. Patel, MD

      Affiliations

    • Department of Ophthalmology, Mayo Clinic College of Medicine, Rochester, Minnesota, USA.
    • ,
  • Jay W. McLaren, PhD

      Affiliations

    • Department of Ophthalmology, Mayo Clinic College of Medicine, Rochester, Minnesota, USA.
    • ,
  • David O. Hodge, MS

      Affiliations

    • Division of Biostatistics, Mayo Clinic College of Medicine, Rochester, Minnesota, USA.
    • ,
  • William M. Bourne, MD

      Affiliations

    • Department of Ophthalmology, Mayo Clinic College of Medicine, Rochester, Minnesota, USA.

Accepted 5 December 2005. published online 13 February 2006.

Article Outline

Purpose

To measure changes in keratocyte density up to five years after photorefractive keratectomy (PRK) and laser in situ keratomileusis (LASIK).

Design

Prospective, nonrandomized clinical trial.

Methods

Eighteen eyes of 12 patients received PRK to correct a mean refractive error of −3.73 ± 1.30 diopters, and 17 eyes of 11 patients received LASIK to correct a mean refractive error of −6.56 ± 2.44 diopters. Corneas were examined by using confocal microscopy before and six months, one year, two years, three years, and five years after the procedures. Keratocyte densities were determined in five stromal layers in PRK patients and in six stromal layers in LASIK patients. Differences between preoperative and postoperative cell densities were compared by using paired t tests with Bonferroni correction for five comparisons.

Results

After PRK, keratocyte density in the anterior stroma decreased by 40%, 42%, 45%, and 47% at six months, two years, three years, and five years, respectively (P < .001). At five years, keratocyte density decreased by 20% to 24% in the posterior stroma (P < .05). After LASIK, keratocyte density in the stromal flap decreased by 22% at six months (P < .02) and 37% at five years (P < .001). Keratocyte density in the anterior retroablation zone decreased by 18% (P < .001) at one year and 42% (P < .001) at five years. At five years, keratocyte density decreased by 19% to 22% (P < .05) in the posterior stroma.

Conclusions

Keratocyte density decreases for at least five years in the anterior stroma after PRK and in the stromal flap and the retroablation zone after LASIK.

 

Development of the excimer laser by Trokel and associates1 has resulted in a marked increase in refractive surgery in the past decade. Photorefractive keratectomy (PRK) sculpts the corneal surface by using an excimer laser to remove a layer of the anterior stroma. In laser in situ keratomileusis (LASIK), an anterior corneal flap is cut and an excimer laser removes a layer of the middle stroma. The epithelium and the anterior stroma are preserved during LASIK, and this is thought to modify the corneal wound-healing response compared with PRK.2

The corneal stroma is populated by keratocytes, whose nuclei are visible in confocal microscopy. Keratocytes remodel structural proteins to maintain homeostasis, mediate wound repair, migrate in response to injury, and die, through apoptosis, in response to wounding.3, 4, 5, 6 Studies of human corneas after PRK and LASIK by histologic methods and confocal microscopy demonstrate a period of active wound healing augmented by activated keratocytes (≤6 months after surgery), followed by a long period of corneal remodeling associated with quiescent keratocytes.7, 8, 9, 10 Keratocyte density estimated by confocal microscopy and light microscopy decreases in the first three years after PRK and LASIK.9, 10, 11, 12, 13, 14 It is not known if this gradual loss of keratocytes continues beyond three years after laser refractive surgery or if keratocyte density recovers. We also do not know the consequences of keratocyte loss to the stroma, or if there is a minimum number of keratocytes needed to maintain the health of the cornea.

The clinical confocal microscope provides a means of repeated noninvasive examination of corneal keratocyte nuclei. Keratocyte density estimated by using confocal microscopy is consistent with density estimated by using light microscopy, DNA distribution, or vital dyes.9, 10, 15, 16, 17 The purpose of this report is to extend our previous observations11, 12, 13, 14 of keratocyte density after PRK and LASIK to five years.

Back to Article Outline

Methods 

We studied 18 eyes of 12 patients (three men, nine women) who received PRK and 17 eyes of 11 patients (one man, 10 women) who received LASIK at the Mayo Clinic, Rochester, Minnesota, between July 1998 and January 1999. PRK patients were 40 ± 7 years old (range 22 to 53 years) and had a mean preoperative refractive error of −3.73 ± 1.30 diopters (range −1.25 to −5.75 diopters). LASIK patients were 32 ± 9 years old (range 22 to 50 years) and had a mean preoperative refractive error of −6.56 ± 2.44 diopters (range −2.00 to −11.00 diopters). All surgical procedures were bilateral; six eyes of six PRK subjects and five eyes of five LASIK subjects had a reoperation for residual myopia between three months and four years after surgery and were excluded from analysis. All patients had a complete ophthalmologic examination before surgery to ensure a normal cornea and anterior segment. Patients with previous ocular surgery, glaucoma, topical ocular medications, or diabetes were excluded. After surgery, none of the patients had a reoperation or wore contact lenses beyond five days postoperatively. One LASIK patient failed to return for the five-year examination. Each patient gave informed consent to participate after the nature and possible consequences of the study had been explained. The study was approved by the Institutional Review Board of Mayo Clinic and followed the Declaration of Helsinki for research involving human subjects.

In PRK, a 6.3-mm-diameter circular area of epithelium was removed by using the laser-scrape technique. The corneal surface was ablated by using a VISX Star 2 excimer laser (VISX, Santa Ana, California, USA) to a mean planned depth of 46 ± 18 μm (± SD, range, 13 to 90 μm; Figure 1). Postoperatively, patients wore a bandage soft contact lens (SofLens 66, Bausch & Lomb Inc, Rochester, New York, USA) until the cornea epithelialized (two to five days). Topical medications were ofloxacin 0.3% four times daily until epithelialization was complete and fluorometholone 0.1% four times daily with a taper over eight to 12 weeks.

  • View full-size image.
  • FIGURE 1. 

    Keratocyte density distribution throughout the normal full-thickness stroma.17 Photorefractive keratectomy removed the keratocyte-dense anterior stroma during surface ablation (light gray area). Laser in situ keratomileusis (LASIK) removed the less-keratocyte-dense middle stroma (dark gray area).

In LASIK, a Hansatome microkeratome (Chiron Vision Corp, Claremont, California, USA) was used to create a flap with a superior hinge and a planned thickness of 180 μm. The measured mean flap thickness was 160 ± 28 μm.18 The midstroma was ablated by using a VISX Star 2 excimer laser to a mean planned depth of 63 ± 26 μm (Figure 1). Postoperatively, topical medications were ofloxacin 0.3% four times daily for five days and fluorometholone 0.1% four times daily with taper over three weeks.

Corneas were examined by using a tandem scanning confocal microscope (Tandem Scanning Corporation, Reston, Virginia, USA) before and at six months and one, two, three, and five years after PRK and LASIK. The method of examination has been previously described.17 Briefly, hydroxypropyl methylcellulose 2.5% optical coupling medium was placed on the tip of the objective lens and the lens was advanced until the solution contacted the cornea. The objective lens was aligned with the center of the cornea by centering the light and dark rings of the epithelial image. After alignment, the focal plane was scanned through the cornea at approximately 72 μm per second from anterior to posterior. Digital images were captured with the video camera in an automatic-gain mode (gain was set by the camera to optimize image brightness) and were stored on a computer workstation (Indy, Silicon Graphics Inc, Mountain View, California, USA) at 30 frames per second. We also recorded images with the camera gain fixed manually to estimate the brightness of scattered light from the cornea. Brightness of these images was adjusted for brightness of a fluorescent glass model cornea. Each image represented a coronal section of 475 μm × 350 μm (horizontal × vertical), was separated from the adjacent image by an average of 2.4 μm, and had a depth of field of 11.9 μm.19 Four to eight scans were recorded at each visit. All scans were within the central 4 mm of the cornea, although not in the identical region each time. No scans were obtained in the peripheral cornea, specifically near the LASIK flap hinge. Typically, keratocyte density was assessed from one scan from each eye with the least transverse movement and no anteroposterior movement.

Changes in confocal image contrast19 and cell brightness that could affect estimates of cell density were evaluated over the five-year study period. Cell brightness was calculated as the mean intensity of individual cells in scans recorded with the gain of the video camera fixed. Image contrast was calculated as the difference between the cell brightness and brightness of the background region around the cell divided by the brightness of the background region. Contrast was determined from the same scans used to assess cell density, with the video gain automatically controlled by the camera.

In PRK, five layers were considered (Figure 2) as described previously: (1) 0% to 10% (anterior), (2) 11% to 33%, (3) 34% to 66%, (4) 67% to 90%, and (5) 91% to 100% (posterior) depth.12, 14 In the pre-PRK and post-PRK cornea, the boundaries of the stromal layers were determined relative to the depth of the most anterior keratocytes. Stromal thickness was the distance between the first focused image of the most anterior keratocytes and the last focused image of the posterior keratocytes, but without visible endothelial cells. In the pre-PRK cornea, the thickness of Bowman’s layer was the distance between the first focused image of subbasal nerves and the most anterior keratocytes.

  • View full-size image.
  • FIGURE 2. 

    Schematic representation of the five stromal layers studied in the pre-PRK (photorefractive keratectomy) and post-PRK cornea. Stromal layers in the pre-PRK cornea are compared with corresponding layers in the post-PRK cornea.

In LASIK, six layers of stroma were considered (Figure 3) as described previously: (1) anterior half of the stromal flap, (2) posterior half of the stromal flap, (3) anterior half of the 100-μμ-thick retroablation zone, (4) posterior half of the retroablation zone, (5) posterior 66% to 90% of the pre-LASIK stroma, and (6) posterior 91% to 100% of the pre-LASIK stroma.11, 13 Small bright objects in the anterior stroma of all post-LASIK scans were used to identify the flap interface in postoperative corneas.9, 10, 11, 12, 13 The thickness of the stromal flap (distance from the most anterior keratocytes to the flap interface) and the stromal bed (distance from the flap interface to the endothelium), determined at the one-month post-LASIK scan, was used to delimit the corresponding anterior and posterior stromal layers in the pre-LASIK cornea. This left a gap between the flap and the bed in the preoperative cornea, a gap that represented the tissue destined for ablation. Cell density in this gap was not assessed. This allowed us to compare the same tissue layers in the pre-LASIK and post-LASIK stroma (Figure 3). After LASIK, the posterior 66% to 90% and 91% to 100% of stroma were determined from the pre-LASIK stromal thickness (distance between the most anterior keratocytes and the endothelium).

  • View full-size image.
  • FIGURE 3. 

    Schematic representation of the six stromal layers studied in the pre-LASIK (laser in situ keratomileusis) and post-LASIK cornea. The thickness of stroma equivalent to the ablation depth (that is, ablation zone) in the pre-LASIK cornea was omitted from analysis to allow comparison of the same tissue layers preoperatively and postoperatively. RAZ = retroablation zone; ANT = anterior; POST = posterior.

Two images with no motion artifact were selected from each stromal layer for assessment of cell density, for a total of 10 images in PRK and 12 images in LASIK per eye per visit. In the most anterior stromal layer, one of the two images was always the most anterior image containing keratoctyes. The selected confocal images before PRK and LASIK and at all visits after PRK and LASIK were randomly presented to an observer who was masked to the patient, time after surgery, and stromal layer. Keratocyte nuclei (cells) were identified as bright objects in a predefined area (0.109 mm2) of each selected image to calculate keratocyte density (cells/mm3) by using a custom computer program (Figure 4).17 The mean cell density in each layer after PRK and LASIK was compared with the mean cell density in the corresponding layer before PRK and LASIK (FIGURE 2, FIGURE 3). The full-thickness density was estimated by dividing the number of cells in a full-thickness stromal column with a cross-sectional area of 1 mm2 by the stromal thickness. The number of cells in the column was estimated from the mean density in each analyzed frame weighted by the distance between frames.17 Cell density estimates are approximately 30% higher in the current study than in our previous studies11, 12, 13, 14 because image depth of field was recalculated to be 11.9 μm in the current study rather than 16 μm, which was used in previous studies.19

  • View full-size image.
  • FIGURE 4. 

    Method to determine keratocyte density. Keratocyte nuclei (Top) were manually counted to determine keratocyte density (cells/mm3) in two images from each stromal layer, as described by Patel and associates.17 By convention, nuclei overlapping the edges of the bounding box were counted on only two sides (Bottom).

We examined intraobserver variability of cell density assessment by repeating estimates of cell density in a subset of subjects after a two-year interval (between three and five years after surgery). Seven subjects were randomly selected from the PRK group, and nine were selected from the LASIK group. After the five-year examination, three-year cell density was reassessed by a masked observer in the same frames as were originally used.

Mean cell density in 212 normal untreated corneas measured at different times during the five years of the study was examined for systematic changes in our ability to assess cell density. Mean cell density of two frames between 10% and 33% of the stromal depth was graphed by date of examination and compared with densities in the posterior flap of corneas treated with LASIK. All untreated corneas were normal and were assessed by one of three investigators. For purposes of this comparison, all densities were adjusted for an age of 45 years, by assuming a normal rate of decrease of 96 cells/mm3 per year of age.17

Keratocyte densities at all postoperative visits were compared with keratocyte densities before surgery by using a paired t test. Significances of differences were Bonferroni-adjusted for five comparisons. The rate of cell loss expressed as a percent per year was assumed to be equal to 100 × ek × 1 year, where k is the slope of the line fitted by linear regression to the natural log of the five mean densities versus time between six months and five years. The percent loss per year was calculated in each stromal layer for each subject and was compared with zero by using a one-sample t test. Tests were checked for the effects of potential correlations between measurements from both eyes of the same subject by using generalized estimating equation (GEE) models.20 The results of the GEE models were similar to results of the standard tests, and only the results of the standard tests are presented. The significance of the differences between cell density assessed from the same frames at three and five years was determined by using a paired t test. All statistics were calculated by using SAS software (SAS Institute Inc, Cary, North Carolina, USA). P < .05 was considered statistically significant.

Back to Article Outline

Results 

Keratocyte densities before and after PRK are shown in Table 1. Keratocyte density in the post-PRK anterior stroma was reduced by 40% at six months (26,049 ± 3396 cells/mm3) when compared with density in the pre-PRK anterior stroma (43,313 ± 7925 cells/mm3, P < .001, Table 1). Between six months and five years, keratocyte density in this most anterior stroma continued to decrease at a rate of 3.4% per year (P = .02, Table 2). At five years, keratocyte density in the anterior 10% stroma was reduced by 47% (23,006 ± 6140 cells/mm3) when compared with preoperative (P < .001, Table 1, Figure 5).

TABLE 1. Keratocyte Density (Mean ± SD, Cells/mm3) Before and After Photorefractive Keratectomy (PRK)
Stromal Layer (% depth)Time After PRK
Pre-PRK6 moP⁎1 yP⁎2 yP⁎3 yP⁎5 yP⁎
0%–10% (anterior)43,313±792526,049±3396<.00126,319±5258<.00125,143±5712<.00123,848±5975<.00123,006±6140<.001
11%–33%25,121±281427,042±4777ns24,592±3511ns23,921±6672ns25,370±4391ns23,870±4984ns
34%–66%24,387±486423,632±5054ns22,240±5459ns23,416±4922ns23,222±3817ns18,826±5204.001
67%–90%24,473±336723,875±3050ns22,748±2892ns22,216±4092ns23,285±3292ns20,178±5333.02
91%–100% (posterior)23,524±500324,775±4234ns22,564±4065ns23,049±4362ns22,337±3894ns17,935±6668.04
Full-thickness26,220±289724,756±3302ns23,301±3564.00123,776±35960.00423,459±3394.00921,017±4534<.001

ns = not significant (P > .05).

⁎ Significance of differences compared with Pre-PRK by using a paired t test (Bonferroni-adjusted for five comparisons).

TABLE 2. Change in Keratocyte Density Between Six Months and Five Years After Photorefractive Keratectomy (PRK)
Stromal Layer%/year⁎P†
0%–10% (anterior)−3.4±5.4.02
11%–33%−1.6±4.9.17
34%–66%−4.4±3.3<.001
67%–90%−3.5±4.8.007
91%–100% (posterior)−7.5±8.4.002
Full-thickness−3.2±2.8<.001

⁎ Percent change in keratocyte density per year (mean ± SD).

† One-sample t test.

  • View full-size image.
  • FIGURE 5. 

    Keratocyte density before and after photorefractive keratectomy (PRK). In the anterior 10% of the stroma, keratocyte density was diminished at all post-PRK visits relative to density before PRK. Cell density in most remaining stromal layers was not decreased until five years after PRK. *P < .001 and †P < .05, compared with densities before PRK. ANT = anterior.

Keratocyte density in the middle and posterior stroma did not change after PRK at each visit for up to three years (Table 1, Figure 5). At five years, a decrease in keratocyte density was first measured in the middle 34% to 66% (P = .001), posterior 67% to 90% (P = .02), and posterior 91% to 100% (P = .04) layers (Table 1, Figure 5).

Keratocyte densities before and after LASIK are shown in Table 3. By six months after LASIK, keratocyte density was decreased by 23% (P = .02), 19% (P < .001), and 15% (P = .004) in the anterior stromal flap, posterior stromal flap, and anterior retroablation zone, respectively, when compared with the density in the same layers before LASIK (Table 3). Between six months and five years, keratocyte density continued to decrease in these layers (Table 4), and by five years, densities were reduced by 32%, 42%, and 42% of pre-LASIK densities, respectively (P < .001, Table 3, Figure 6). Keratocyte density decreased from pre-LASIK densities in the posterior retroablation zone (P = .02), the posterior 66% to 90% layer (P < .001), and the posterior 91% to 100% layer (P = .04) at five years, but not earlier (Table 3, Figure 6).

TABLE 3. Keratocyte Density (Mean ± SD, Cells/mm3) Before and After Laser In Situ Keratomileusis (LASIK)
Stromal LayerTime After LASIK
Pre-LASIK6 mo†P⁎1 y†P⁎2 y†P⁎3 y†P⁎5 y‡P⁎
Anterior flap47,004±689641,063±4877.0237,864±7910.00237,156±7071.00534,620±5306<.00131,976±4071<.001
Posterior flap34,277±546027,604±4571<.00124,999±4300<.00125,822±5141<.00121,892±3238<.00119,715±4378<.001
Anterior RAZ28,793±402324,474±3811.00423,560±3981.00321,389±3929<.00121,229±4712.00216,559±2676<.001
Posterior RAZ27,599±478726,073±2907ns26,279±2680ns27,452±2973ns25,913±3335ns22,944±3285.02
66%–90% (posterior)29,752±427527,467±3823ns27,559±3760ns27,878±3720ns26,668±3586ns23,430±3359<.001
91%–100% (posterior)28,770±545727,444±4694ns27,879±4232ns25,091±3673ns27,490±3205ns23,454±3463.04
Full-thickness31,108±498428,337±2863ns27,533±2757.0427,491±2693.0526,320±1973.00522,982±1829<.001

ns = not significant (P > .05); RAZ = retroablation zone.

⁎ Significance of differences compared with pre-LASIK by using a paired t test (Bonferroni-adjusted for five comparisons).

† 17 eyes of 11 patients.

‡ 16 eyes of 10 patients.

TABLE 4. Change in Keratocyte Density Between Six Months and Five Years After Laser In Situ Keratomileusis (LASIK)
Stromal Layer%/year⁎P†
Anterior flap−4.3±3.1<.001
Posterior flap−7.2±4.3<.001
Anterior RAZ (0–50 μm)−8.4±3.7<.001
Posterior RAZ (51–100 μm)−2.6±4.1.02
Posterior 66%–90%−3.5±3.4<.001
Posterior 91%–100%−3.1±2.2<.001
Full-thickness−4.2±2.2<.001

RAZ = retroablation zone.

⁎ Percent change in keratocyte density per year (mean ± SD).

† One-sample t test.

  • View full-size image.
  • FIGURE 6. 

    Keratocyte density before and after laser in situ keratomileusis (LASIK). In the anterior and posterior stromal flap and the anterior RAZ, keratocyte density was decreased at all post-LASIK visits from density before LASIK. Cell densities in all remaining stromal layers were first decreased at five years after LASIK. *P < .005 and †P < .05, when compared with densities before LASIK. RAZ = retroablation zone; ANT = anterior; POST = posterior.

Cell densities in the posterior flap of all LASIK patients during the five years of the study and densities in a similar region of the stroma in 212 normal unoperated corneas from other studies are shown in Figure 7. Density in the posterior flap decreased at an average of 1897 cells/mm3 per year (P <. 001), whereas cell densities in normal untreated corneas decreased at 445 cells/mm3 per year (P = .004).

  • View full-size image.
  • FIGURE 7. 

    Estimated cell density in anterior 10% to 33% of stroma of 212 untreated normal corneas from other studies, and in the posterior flap of corneas after LASIK throughout this study. The solid line is the least-squares regression of cell densities in corneas treated with laser in situ keratomileusis (LASIK) (average rate of cell loss = 1897 cells/mm3 per year, r = 0.51, P <. 001), and the dashed line is the regression of cell densities in untreated corneas (average rate of cell loss = 445 cells/mm3 per year, r = 0.20, P = .004). Densities in untreated corneas were estimated by three investigators, whereas all densities in corneas treated with LASIK were estimated by one investigator.

When confocal images were randomly assessed to measure keratocyte density in the same frames by the same masked observer two years apart, the mean difference in cell density was 640 ± 4274 cells/mm3 in the PRK patients and 128 ± 2185 cells/mm3 in the LASIK patients. These differences were not significantly different from zero (paired t test, P = .70 and P = .87, respectively). The average ratio of the difference between the first and second measurement to the mean of the two measurements was 0.029 for PRK and 0.005 for LASIK. The minimum detectable difference between the first and second measurements by the same observer was 4050 cells/mm3 for PRK and 1730 cells/mm3 for LASIK (paired t test, α = .05, β = .20).

FIGURE 8, FIGURE 9 demonstrate mean brightness of cells and background, and mean contrast in images from LASIK patients throughout the study. Cell and background brightness varied by 10 to 15 intensity units but did not increase or decrease consistently throughout the study. Cell contrast consistently increased after surgery, reached a maximum at six months, and returned to presurgery levels at one year. From two to five years, mean cell contrast changed by less than 0.01.

  • View full-size image.
  • FIGURE 8. 

    Change in cell and background brightness over time. Mean brightness of cells (open symbols) and the area immediately surrounding the cells (solid symbols) in the layers immediately anterior and posterior to the interface, and the posterior 90% to 100% of stromal thickness after laser in situ keratomileusis (LASIK). Brightness was adjusted for variations in the illumination brightness and sensitivity of the video camera by using measurements of a fluorescent glass standard. Units of brightness are arbitrary and represent the digitized output of the video camera. Brightness remained steady between one and five years when keratocyte density decreased, suggesting that potential changes in confocal image brightness did not affect our ability to identify cells.

  • View full-size image.
  • FIGURE 9. 

    Change in contrast. Mean contrast between cells and background was calculated in the same frames used to determine cell density. After laser in situ keratomileusis (LASIK), contrast increased to a maximum at six months and then returned to pre-LASIK contrast by one year. Contrast changed by 0.01 or less between one and five years, suggesting that potential changes in image contrast did not affect our ability to identify cells.

Back to Article Outline

Discussion 

This prospective longitudinal clinical trial demonstrates a loss of keratocytes in the most anterior stroma after PRK, and in the stromal flap and stroma immediately posterior to the ablation interface after LASIK for up to five years. In the middle and posterior stroma, keratocyte density decreased for the first time at five years after both procedures.

In the normal cornea, keratocyte density is highest in the anterior 5% to 10% of the stroma, approximately 40% higher than cell density in the middle and posterior stroma.16, 17 In PRK, this keratocyte-rich anterior stroma is removed during photoablation. Although keratocytes are able to divide and migrate after wounding,5, 6, 21 our study shows that remaining keratocytes do not repopulate the post-PRK anterior stroma to the densities in the pre-PRK anterior stroma for at least five years. Rather than repopulate the anterior stroma, keratocytes continue to decrease between six months and five years at an accelerated rate of 3.4% per year. Early after wounding by PRK, corneal keratocytes disappear through apoptosis mediated by cytokines released from injured epithelium.7 Chronic stimulation of apoptosis could be responsible for the observed long-term losses of anterior keratocytes after PRK through continued epithelial–stromal interactions or cell–matrix interactions.3, 13

Keratocytes in the middle and posterior stromal layers (11% to 100% depth) appear to be less affected by PRK than the anterior keratocytes were. Cell density in these layers was unchanged through three years relative to density before PRK. At five years after PRK, however, keratocyte loss became significant in the middle and posterior stroma. The reason for this delayed cell loss is unclear, but persistent molecular signals to undergo apoptosis could be directed to deeper stromal keratocytes via gap junctions.22

In earlier LASIK studies, keratocytes decreased in the stromal flap and in the stroma immediately posterior to the ablation interface as early as one month after LASIK.10, 11 Our current data demonstrate that this early keratocyte loss in the anterior and posterior stromal flap and in the anterior retroablation zone progresses to deficits of 32%, 42%, and 42% of pretreatment densities, respectively (P < .001) by five years after LASIK. Additionally, the rate of keratocyte loss appears greatest in the stroma immediately anterior and posterior to the ablation interface (−7.2% and −8.4% between six months and five years, respectively). In a recent study of 13 postmortem LASIK corneas, Dawson and associates10 demonstrated a decrease in keratocytes in the anterior stroma in the first year after LASIK but were unable to detect a decrease with increasing postoperative time. Their cross-sectional investigation, however, lacked the statistical power to detect longitudinal differences similar to ours. Previous histopathologic studies4, 7 have shown that epithelial and stromal injuries induced by the microkeratome cause keratocyte apoptosis in the stromal layers anterior and posterior to the lamellar interface. This localized cell loss has been attributed to epithelial debris with apoptosis-inducing cytokines being tracked into the interface by the microkeratome blade.7 Implanted viable and degenerating epithelial cells have been found in the LASIK interface years after surgery.9, 10 These implanted interface epithelial cells may contribute to the observed chronic keratocyte loss through continued release of apoptotic cytokines that diffuse along the interface and into the central stroma.

In addition to cell death by apoptosis, other mechanisms of keratocyte loss after PRK and LASIK are possible. First, a causal relationship between decreased keratocyte density and decreased innervation after LASIK has been hypothesized.11, 12, 13, 23 Müller and associates24 have documented direct innervation of keratocytes by stromal nerves, and a normal keratocyte population may depend on relevant cytokines and growth factors provided by a normal density of corneal nerves. Transplanted corneas, for example, have both keratocyte25 and nerve deficits.23 Consistent with this hypothesis, we recently showed that subbasal nerves are reduced by 24% at five years after LASIK.26 Although these five-year nerve densities were not significantly different from densities before LASIK, the study lacked statistical power to detect a difference less than 35%.26 What effect a reconstituted nerve density has on future keratocyte densities is unknown. Second, altered tissue stress-strain relationships in the anterior stroma after PRK and in the stromal flap after LASIK may affect keratocyte survival. Finally, the extremely high corneal ascorbate levels found in the human cornea may be affected by PRK and LASIK. Corneal epithelial ascorbic acid absorbs UV radiation and protects deeper layers of the cornea, such as stromal keratocytes, from radiation damage.27 Altered ascorbate levels could lead to accelerated keratocyte death. It is likely that the epithelium, stroma, and nerves all participate, to varying degrees, in the homeostasis of stromal keratocytes. Their eventual effect on keratocyte density after PRK and LASIK will require longer follow-up.

It is not clear what role a chronic reduction in keratocyte numbers after PRK and LASIK might play in the health of the cornea. Wilson and associates7 suggest that a high keratocyte density in the anterior stroma provides some form of protection against infection of the corneal epithelium and minimizes posterior extension of infections. Keratocyte loss after PRK and LASIK, however, does not seem to negatively affect corneal clarity or visual acuity. All of our study patients had clear corneas and a corrected visual acuity of 20/25 or better at all postoperative visits. Similarly, Rajan and associates28 have shown good clinical results for up to 12 years after PRK. Recently, Dawson and associates9 measured the image brightness from a 30-μm-thick optical section that included the 4-μm to 6-μm-thick interface scar in 13 postmortem LASIK corneas and found that the backscattered light did not differ from that in normal corneas. Although this measurement had the statistical power to detect only very large differences in backscattered light, the result is nevertheless consistent with clinical impressions that the hypocellular interface scar does not degrade vision.

Estimates of keratocyte density after PRK and LASIK by confocal microscopy have potential limitations, which we investigated as possible sources of spurious results. First, manual assessment of keratocytes is subjective and could be affected by intraobserver variability and subtle changes in the microscope and video camera during the five years of our study. Previous evaluations of keratocyte estimates in normal corneas have shown intraobserver variability to be approximately 7.8% (McLaren JW, Bourne WM. An improved automated method for estimating keratocyte density in confocal microscopy. AVRO E-abstract #1714, May 6, 2002, http://www.arvo.org, presented at Association for Research in Vision and Ophthalmology Meeting, 2002). In the current study, keratocyte density estimated on two occasions separated by two years differed by 2.9% after PRK and 0.5% after LASIK, differences that were not statistically different from zero. Systematic changes in our ability to estimate keratocyte density during the five years of this study should also have affected our estimates of keratocyte densities in normal, untreated corneas. Although we did not have concurrent controls, our estimates of mean keratocyte density in 212 normal corneas decreased by approximately 2226 cells/mm3 in five years (Figure 7). We do not know why apparent densities in normal corneas decreased, although this trend may represent subtle changes in criteria for subjective estimation of density by three investigators or changes in the confocal microscope. Cell density in the posterior flap decreased by approximately 9473 cells/mm3 during the five years after LASIK, and this rate of decline is considerably faster than the variations in normal corneas and what one would expect from subjective estimates. We cannot rule out the possibility that the small changes in cell density in the posterior stroma between three and five years resulted from subtle changes in our microscope and video system. It is unlikely that these differences were related to changes in criteria for selecting cells because individual scans from the five-year examinations were presented randomly with a sample of scans from the three-year examinations to the investigator who selected the cells, and there was no systematic shift of the repeated assessments of the three-year scans. The small but significant cell density changes in the posterior stroma must be verified by remeasuring cell density again at seven years or longer after surgery.

Second, changes in backscattered light, cell brightness, and image contrast after PRK and LASIK could limit one’s ability to identify keratocytes in confocal images. Cell and background brightness varied somewhat after both PRK and LASIK, although they did not vary in a way that was consistent with the changes in cell density. Contrast between cells and the immediate background, which is important for identifying cell nuclei, increased to a peak at six months after LASIK, but then decreased to pre-LASIK contrast at one year. This change in contrast is consistent with the appearance and disappearance of highly reflective metabolically activated wound-response keratocytes.10, 29, 30 Image contrast remained steady between one and five years. Keratocyte density decreased during this same time. If contrast had affected our ability to estimate keratocyte density, contrast should have changed as apparent cell density changed. Therefore, we believe that the characteristics of cell images, scatter of light, cell brightness, and image contrast did not likely affect our ability to assess keratocyte density in this study. Finally, bright objects (presumed keratocytes) in the confocal images could also represent other stromal cells. However, keratocyte density estimated by using light microscopy in the normal cornea17 and in corneas after LASIK10 is consistent with density estimates using confocal microscopy.

All LASIK procedures were performed using a superior hinge, and the mean flap thickness measured by confocal microscopy one month postoperatively was 160 μm.18 It is possible that a nasal hinge or a thinner corneal flap could affect keratocytes differently. Additionally, our LASIK mean midstromal ablation depth was 17 μm greater than the PRK mean surface ablation depth. The difference in the amount of stromal tissue removed in the PRK and LASIK groups could contribute to the observed differences in keratocyte density.

Any change in stromal thickness in the postsurgical cornea could affect keratocyte density estimates. Specifically, deposition of new stromal tissue in the postsurgical cornea would increase stromal thickness, resulting in a relative decrease in keratocyte density. In our study patients, stromal thickness increased 2.2 μm (0.1% per year) in the PRK group and 5.5 μm (0.3% per year) in the LASIK group between six months and five years after surgery (Nau CB, Erie JC, Hodge DO, et al. Epithelial and stromal thickness five years after LASIK and PRK. ARVO E-abstract #4387, May 4, 2005, http://www.arvo.org, presented at Association for Research in Vision and Ophthalmology Meeting, 2005). Therefore, postsurgical stromal thickness changes had minimal effect on our keratocyte density estimates.

The best estimate of age-related keratocyte loss in the normal cornea is 0.45% per year,17 which is similar to the age-related endothelial cell loss of 0.6% per year.31 Between six months and five years after PRK and LASIK, we found an accelerated annual loss of keratocytes (3.2% and 4.2%, respectively). At present, this appears to have no detectable clinical consequences. Over decades, however, a deficiency of keratocytes, the cells that produce the collagen and proteoglycans necessary to maintain this low-turnover tissue, could affect corneal transparency or curvature.

In summary, we report long-term decreases in keratocyte density in human corneas after PRK and LASIK. Additional studies by other investigators are needed to confirm our findings. By measuring and following changes in keratocyte density in various layers of the cornea after PRK and LASIK, one may better understand the long-term biologic and clinical consequences of these surgical procedures.

Back to Article Outline

References 

  1. Trokel SL , Srinivasin R , Braren B . Excimer laser surgery of the cornea . Am J Ophthalmol . 1983;96:710–715
  2. Helena MC , Baerveldt F , Kim WJ , Wilson SE . Keratocyte apoptosis after corneal surgery . Invest Ophthalmol Vis Sci . 1998;39:276–283
  3. Fini EM . Keratoctye and fibroblast phenotypes in the repairing cornea . Prog Retinal Eye Res . 1999;18:529–551
  4. Wilson SE , Liu JJ , Mohan RR . Stromal-epithelial interactions in the cornea . Prog Retinal Eye Res . 1999;18:293–309
  5. Jester JV , Rodrgues MM , Villasenor RA , et al.   Keratophakia and keratomileusis (histopathologic, ultrastructural, and experimental studies) . Ophthalmology . 1984;91:793–805
  6. Zieske JD , Guimaraes SR , Hutcheon AEK . Kinetics of keratocyte proliferation in response to epithelial debridement . Exp Eye Res . 2001;72:33–39
  7. Wilson SE , Mohan RR , Mohan RR , et al.   The corneal wound healing response (cytokine-mediated interaction of the epithelium, stroma, and inflammatory cells) . Prog Retinal Eye Res . 2001;20:625–637
  8. Dawson DG , Edelhauser HF , Grosshiklaus HE . Long-term histopathologic findings in human corneal wounds after refractive surgical procedures . Am J Ophthalmol . 2005;139:168–178
  9. Dawson DG , Holley GP , Geroski DH , Waring GO , Grossniklaus HE , Edelhauser HE . Ex vivo confocal microscopy of human LASIK corneas with histologic and ultrastructural correlation . Ophthalmology . 2005;112:634–644
  10. Dawson DG , Kramer TR , Grossniklaus HE , et al.   Histologic, ultrastructural, and immunofluorescent evaluation of human laser in situ keratomileusis corneal wounds . Arch Ophthalmol . 2005;123:741–756
  11. Mitooka K , Ramirez M , Maguire LJ , et al.   Keratocyte density of central human corneas after laser in situ keratomileusis . Am J Ophthalmol . 2002;133:307–314
  12. Erie JC . Corneal wound healing after photorefractive keratectomy (a three-year confocal microscopy study) . Trans Am Ophthalmol Soc . 2003;101:293–333
  13. Erie JC , Nau CB , McLaren JW , Hodge DO , Bourne WM . Long-term keratocyte deficits in the corneal stroma after LASIK . Ophthalmology . 2004;111:1356–1360
  14. Erie JC , Patel SV , McLaren JW , et al.   Keratocyte density in the human cornea after photorefractive keratectomy . Arch Ophthalmol . 2003;121:770–776
  15. Poole CA , Brookes NH , Clover GM . Keratocyte networks visualized in the living cornea using vital dyes . J Cell Sci . 1993;106:685–692
  16. Moller-Pedersen T , Ledet T , Ehlers N . The keratocyte density of human donor corneas . Curr Eye Res . 1994;13:163–169
  17. Patel SV , McLaren JW , Hodge DO , Bourne WM . Normal human keratocyte density and corneal thickness measurement by using confocal microscopy in vivo . Invest Ophthalmol Vis Sci . 2001;42:333–339
  18. Erie JC , Patel SV , McLaren JW , et al.   Effect of myopic laser in situ keratomileusis on epithelial and stromal thickness. A confocal microscopy study . Ophthalmology . 2002;109:1447–1452
  19. McLaren JW , Nau CB , Kitmann AS , Bourne WM . Keratocyte density. Comparison of two confocal microscopes . Eye Contact Lens . 2005;31:28–33
  20. Zeger SL , Liang KY . Longitudinal data analysis for discrete and continuous outcomes . Biometrics . 1986;42:121–130
  21. Yoon YD , Waring GO , Stulting RD , et al.   Keratocyte repopulation in epikeratoplasty specimens . Cornea . 1998;17:180–184
  22. Watsky MA . Keratocyte gap junctional communication in normal and wounded rabbit corneas and human corneas . Invest Ophthalmol Vis Sci . 1995;36:2568–2576
  23. Tervo T , Vannas A , Tervo K , Holden BA . Histochemical evidence of limited reinnervation of human corneal grafts . Acta Ophthalmol . 1985;63:207–214
  24. Müller LJ , Pels L , Vrensen GFJM . Ultrastructural organization of human corneal nerves . Invest Ophthalmol Vis Sci . 1996;37:476–488
  25. Bourne WM . Cellular changes in transplanted human corneas . Cornea . 2001;20:560–569
  26. Erie JC , McLaren JW , Hodge DO , Bourne WM . Recovery of corneal subbasal nerve density after PRK and LASIK . Am J Ophthalmol . 2005;140:1059–1064
  27. Brubaker RF , Bourne WM , Bachman LA , McLaren JW . Ascorbic acid content of human corneal epithelium . Invest Ophthalmol Vis Sci . 2000;41:1681–1683
  28. Rajan MS , Jaycock P , O’Brart D , et al.   A long-term study of photorefractive keratectomy, twelve year follow-up . Ophthalmology . 2004;111:1813–1824
  29. Moller-Pedersen T , Cavanaugh HD , Petroll WM , Jester JV . Stromal wound healing explains refractive instability and haze development after refractive keratectomy (a 1-year confocal microscopic study) . Ophthalmology . 2005;107:1235–1245
  30. Jester JV , Moller-Pedersen T , Huang J , et al.   The cellular basis of corneal transparency (evidence for corneal crystallins) . J Cell Sci . 1999;112:613–622
  31. Bourne WB , Nelson LR , Hodge DO . Central corneal endothelial cell changes over a ten year period . Invest Ophthalmol Vis Sci . 1997;38:779–782
biography

Jay C. Erie, MD, is an Associate Professor of Ophthalmology in the Mayo Clinic College of Medicine. He practices in the General Ophthalmology Section. His research interests include corneal confocal microscopy.

 See accompanying Editorial on page 918.This study was supported in part by grant EY 02037 from the National Institutes of Health, Bethesda, Maryland; Research to Prevent Blindness, Inc, New York, New York; and the Mayo Foundation, Rochester, Minnesota.Data from this study were presented in part at the Annual Meeting of the American Ophthalmological Society, Sea Island, Georgia, May 24, 2005.

PII: S0002-9394(05)01299-7

doi:10.1016/j.ajo.2005.12.014

Refers to article:

  • Long-Term Corneal Keratocyte Deficits After PRK and LASIK: In Vivo Evidence of Stress-Induced Premature Cellular Senescence , 21 February 2006

    Daniel G. Dawson, Terrence P. O’Brien, Henry F. Edelhauser
    American Journal of Ophthalmology May 2006 (Vol. 141, Issue 5, Pages 918-920)

American Journal of Ophthalmology
Volume 141, Issue 5 , Pages 799-809.e1, May 2006

Article Tools

* Image Usage & Resolution