In the present study, PRK and LASIK both induced increases in epithelial thickness of approximately 15% to 20% that persisted after surgery. In LASIK, the epithelial changes occurred within 1 week and remained unchanged through 3 years (
Fig. 3 ,
Table 2 ). In PRK, per-operative epithelial debridement caused an initial decrease in epithelial thickness, followed by a gradual epithelial thickening over the next 12 months. Thus, PRK and LASIK induced different initial epithelial responses to surgery, though the end point thickness was similar for the two surgical modalities. In one previous study with CMTF, no changes in epithelial thickness were found 1 year after PRK
6 ; still, other studies have reported an increase in epithelial thickness after PRK
8 11 12 and LASIK.
8 13 14 19 Case reports have demonstrated epithelial hyperplasia with an increased number of cell layers after PRK,
11 whereas the nature of the epithelial changes after LASIK are less clear. It has been suggested that epithelial hyperplasia after refractive surgery may contribute to the loss of the postoperative refractive effect.
11 12 20 21 In the present study, there was no correlation between change in epithelial thickness and change in refraction after PRK or LASIK. This lack of correlation may be attributed to the relatively few patients in each treatment group. After LASIK, however, epithelial thickness had already stabilized by 1 week, whereas the major refractive change was noted between 1 week and 1 month after surgery. This suggested that epithelial changes were not the main cause for refractive instability after LASIK. With respect to PRK, the time course of changes in epithelial thickness and refraction was similar, but, as noted, no correlation between the two parameters could be identified.
PRK and LASIK both induced stromal regrowth during the first year after surgery
(Fig. 4) ; however, wound repair after PRK gave rise to significantly more stromal tissue deposition than did LASIK
(Table 2) , and the increase in stromal thickness correlated with the postoperative loss of refractive effect
(Fig. 7) . PRK has previously been reported to induce a more aggressive wound-healing response than LASIK and to entail more myopic regression and more haze development.
4 5 Studies have also demonstrated significant amounts of stromal tissue deposition after PRK.
6 8 In contrast, stromal changes after LASIK remain controversial, with one study indicating a minor (insignificant) increase over time,
19 one reporting stability,
8 and one even indicating a decrease in total corneal and stromal thickness.
13 However, the present randomized study is the first to allow a direct comparison of corneal sublayer thickness between LASIK and PRK in human eyes and shows that LASIK causes less stromal tissue deposition than PRK for identical myopic corrections, supporting well-established clinical observations of differences in wound repair (haze development and myopic regression).
4 5 22 Mechanisms leading to more aggressive wound repair after PRK remain unclear. However, in previous studies of rabbit eyes, we demonstrated that the integrity of the epithelial-stromal barrier at the basement membrane level appeared to be of major importance for the gravity of the subsequent stromal wound repair.
23 24 Cell culture studies have supported this observation,
25 as have clinical observations of other surgical approaches that destroy the epithelial-stromal barrier. Those approaches include laser subepithelial keratomileusis (LASEK), in which the epithelial sheet is supposedly kept intact but in which haze and myopic regression also may be seen.
26
Interestingly, the stromal regrowth that was observed after LASIK in the present study was found to be localized entirely to the residual stromal bed, whereas flap thickness remained constant through the 3-year follow-up
(Fig. 5) . This contrasts with our previous observations in rabbit eyes in which LASIK caused stromal regrowth in both flap and residual stroma.
27 The present observation is important in patients who are considered for LASIK retreatment because the thickness of the residual stromal bed is a major safety parameter for the amount of refractive correction that can be applied.
28 29 In contrast to PRK, changes in stromal thickness after LASIK could not be correlated with changes in postoperative refraction, possibly because of the relatively few patients in both treatment groups. Still, it should be noted that determination of changes in central thickness alone does not allow comprehensive evaluation of the relationship between stromal tissue deposition and refraction.
Both PRK and LASIK showed very good correlation between initial changes in stromal thickness and expected photoablation depth. More stromal tissue was removed by PRK than by LASIK
(Table 2) , even though the nomogram used for the excimer laser photoablation was identical for the two procedures. The reason for the observed difference in the photoablation depth remains unclear, but may be due to differences in stromal hydration at the time of surgery,
30 either because of a longer stromal exposure during epithelial debridement in PRK or because of local variation in stromal hydration with depth.
31 In accordance with the observed difference in the amount of ablated stromal tissue, spheroequivalent refraction by 1 week averaged −0.23 D for PRK and −0.76 D for LASIK. On average, 1 D refractive change required stromal ablation of approximately 12 μm during LASIK and 13 μm during PRK, in accordance with the expected value for a 6 mm ablation.
32 Interestingly, this observation suggested that the very large difference between PRK and LASIK in central epithelial thickness by 1 week (
Fig. 3 ;
Table 2 ) did not have any major impact on the refractive result. Once again, this indicates that evaluating thickness changes in only the center of the cornea may be insufficient to estimate postoperative refractive changes.
33
In conclusion, the present study of patients with myopia randomly assigned to PRK or LASIK firmly demonstrates that the two surgical procedures induce different changes in corneal sublayer thickness. Epithelial and stromal wound repair occurs after both PRK and LASIK, but the time course is longer and the amount of tissue deposition is greater after PRK. Still, after 1 year, corneas treated with PRK or LASIK appear to be stable and to have undergone no further significant changes in corneal sublayer thickness. Initial changes in stromal thickness (after PRK) and total corneal thickness (after PRK or LASIK) appear to contribute to postoperative refractive regression. In contrast, the increase in epithelial thickness seems to have no refractive impact. In the present study, all thickness measurements were obtained only in the center of the cornea. To better evaluate the relation between wound repair and postoperative refractive changes, topographic variations in corneal sublayer thickness over time should be determined. Unfortunately, this is not realistic with in vivo confocal microscopy, but it is hoped that future technical development will allow such investigations.