Dan Z. Reinstein, MD, MA(Cantab), FRCSC, DABO



What differentiates the cornea from materials such as a contact lens? The cornea is an elastic collagen lamellar structure, the curvature of which is maintained firmly and constant by IOP. It is not difficult to imagine that differential thinning of the stroma can lead to differential bowing of the corneal layer producing central bulging.1 Additionally, the lamellar packing can be altered by the creation of a flap and tissue removal-the peripheral corneal lamellae adjacent to the keratectomized layer are no longer held tense and can relax, thus potentially causing a pull on the central cornea and causing central flattening.2 Finally, the epithelial thickness profile determines the majority of the final refractive power of the cornea, so changes in the profile of the epithelial layer can also cause changes in refraction.3

"Regression is reversion to an earlier condition or state." To illustrate this, imagine a myopic eye, undergoing LASIK for -4 D, that was found to be plano on postoperative day 1, -0.25 to 0.25x180 at 1 month, -0.75 at 3 months, 6 months, and 1 year. In this case, there was an initial correction of -4 D, which "regressed" to -0.75 D where it stabilized. Why did this case regress? Is it because the epithelium thickened in the center? Is it because the flap was too thick, and there was bowing of the central cornea forward?

For the purposes of this article, regression is defined narrowly to the observation of a shift in refraction postoperatively that tends to reverse the intended effect. The author is, therefore, excluding the situation where the refraction does not stabilize and continues to change due to plastic deformity of the cornea-a process known as keratectasia.4, 5 Differences among regression, primary undercorrection, and ectasia will now be illustrated.

If an eye was treated by LASIK for -8 D and on day 1, through to 1 week, 3 months, 6 months, and 1 year, the eye was found to be stable at -1 D, this would be defined as a primary undercorrection, rather than regression. If this same eye had been -0.75 D at postoperative day 1, -1.50 D at 1 month, -1.75 D at 3 months, but then stable through to 1 year, then the eye would have been said to have had a primary undercorrection of -0.75 D followed by regression from -0.75 to -1.75 D. It is important to differentiate between the causes of undercorrection because they point to the possible etiology and how to avoid keratectasia.

Primary undercorrection has many etiologies, but these can be divided into corneal and non-corneal causes. Non-corneal causes will include inaccurate preoperative refraction, inadequate laser energy delivery (eg, excess bed hydration, room humidity, inappropriate laser energy calibration, laser head energy instability). Corneal causes essentially include cases where the biomechanics of the cornea change due to the keratectomy and a stable but unpredicted curvature change is obtained. For example, this can happen if the residual stromal thickness was much less than 250 microns but not thin enough to cause long-term destabilization (ectasia). Examples of this will be shown.

For this tutorial, the author groups together both primary undercorrection for corneal causes (mostly biomechanical) and regression (mostly epithelial). The information and data that the author based this tutorial on contribute to ongoing studies to analyze the accuracy of LASIK with respect to epithelial and biomechanical changes. In this study, the author isolates and measures the effect of epithelial and biomechanical changes on the attempted corneal power change.

Fifty-two eyes that underwent routine LASIK between 1998 and 1999 with both the Moria LSK-One (Moria, Antony, France) and Hansatome (Bausch & Lomb, St Louis, Mo) microkeratomes, and with either the Nidek EC5000 (Nidek, Japan) or the B&L 217C (Bausch & Lomb) excimer laser6 were studied. Myopia from -1.00 to -10.25 D was included. Patients were scanned by 3D Artemis VHF digital ultrasound (Ultralink LLC,
St Petersburg, Fla) to obtain the thickness profile and optical power of the epithelium and stroma separately before and after LASIK. The author measured front and back surface curvature of the cornea using the Orbscan II (B&L, St. Louis, Mo) before and at least 3 months after LASIK. Epithelial thickness (ET) and residual stromal bed thickness (RST) 3D maps were produced from the Artemis data. The Orbscan determined the anterior and posterior corneal best-fit spheres (BFS).

Method for Isolating Epithelial and Biomechanical Changes

The curvature of Bowman's surface was calculated from the anterior BFS and the epithelial thickness profile from the Artemis. Gradient optics and lens formulae were used to calculate total corneal power from anterior, Bowman's, and posterior corneal interfaces. Back surface curvature change was defined as a bowing factor. The corneal power change (CPC) was calculated in this manner by comparing the preoperative and postoperative data. To isolate the effects of epithelial and biomechanical changes, the postoperative data were split into the following three permutations:

Epithelial factors: the bowing factor was removed by subtracting the back surface curvature change from each of the anterior, Bowman's, and posterior corneal surfaces.
Bowing factors: the epithelial factor was removed by substituting the postoperative epithelium for the preoperative epithelium.
Remove epithelial and bowing factors: the bowing factor was removed by substituting the postoperative back surface for the preoperative back surface and the epithelial factor was removed by substituting the postoperative epithelium for the preoperative epithelium.
Linear regression and paired t-tests were used to determine the epithelial and/or bowing contributions to the final refraction by correlating the attempted correction to the CPC.

General Observations

The author found that for the cohort of eyes, the minimum RST was 262 microns. Below an RST of 300 microns, postoperative back-surface curvature change correlated strongly with RST (R2=0.5). Attempted vs. achieved subjective refractive change was highly correlated (R2=0.95) with a slope of 0.92. CPC measurement by the calculation method was validated by a high correlation between change in clinical refraction and calculated CPC (R2=0.67, slope=0.90).

Isolating Epithelial and Bowing Factors

As described earlier, attempted correction was correlated with calculated CPC subtracting epithelial factors, bowing factors, and both epithelial and bowing factors. Removing epithelial changes gave a significant change in CPC and increased the slope to 0.94 (R2=0.66), whereas removing bowing gave a significant change in CPC and increased the slope to 0.99 (R2=0.46). Removing both epithelial and bowing factors resulted in a significant difference in CPC, with a correlation slope of 1.03 (R2=0.4, (all P<.01). This led to the conclusion that there were significant biomechanical and epithelial effects occurring, and that corneal elastic bowing and epithelium changes could practically account for the inaccuracy of LASIK.

In the magnitude analysis, mechanical changes accounted for a 15% decrease in intended reduction in central corneal power (P<.001). Epithelial changes accounted for a 5% decrease in intended reduction in central corneal power (P<.001). RST was correlated to the mechanical shifts calculated (R2=0.32). Ablation depth was highly correlated to the mechanical shifts observed (R2=0.89). Thickness of the stromal component of the flap significantly correlated with the spherical equivalent postoperative error (P<.05). Strong significant nonlinear correlations were found between the level of myopia treated and the epithelial (P<.001) or biomechanical (P=.011) power shift measured.

The Epithelium in Regression

The mean central epithelial thickness before surgery was 51 microns (range: 47-62 microns), whereas three months after LASIK, the central epithelium had thickened to an average of 61 microns (range: 44-75 microns). Central thickening amounts to a relative increase in curvature and, therefore, a regression in the myopic refractive effect of LASIK flattening.

In a raw comparison between the error in the postoperative spherical equivalent (all patients had an intended postoperative refractive error of zero) and the epithelial power shift as measured by Artemis scanning, the author found a statistically significant correlation. In other words, the postoperative refractive error could be (at least partly) accounted for by shifts in the power of the epithelium. The fact that these shifts were correlated alone is a testament to the significance of the epithelium in regression, considering the number of biomechanical factors that were also in play. A simple linear regression demonstrated a correlation in which for every diopter of postoperative spherical equivalent error, the epithelium could be accounting for about 25% of the error.

Plotting the level of myopia treated against the amount of epithelial thickening in the center and at the 3-, 4-, 5-, 6-, and 7-mm zones, a strong correlation between amount of epithelial thickening and the level of myopia treated was found. Two things stood out from this analysis. First, more thickening in the central cornea than the peripheral was found. Second, the epithelial thickening response was steep and linear for lower myopia, but appeared to level off as the level of myopia increased, which implies that the epithelium has the ability to reverse central flattening, but only to a certain extent.

To investigate further, the author divided the study eyes into three groups: low, moderate, and high myopia, and determined the thickening profile for each group separately.

The shift in power due to the epithelium will be related to the difference between central and peripheral thickening. For example, if the epithelium were to thicken evenly by 3 microns there would be no power shift (no change in curvature). For the low myopic group, there was considerably more thickening in the center than the periphery-8 microns versus 4 microns, and as myopia increased, the difference between the central and peripheral thickening diminished-in other words, the epithelium appears to be causing regression to a greater extent in lower myopia than higher myopia, despite the fact that there is less absolute thickening in lower myopia.

Thus, the shift in power due to epithelial profile changes was more significant for lower myopia than for higher myopia. Based on the central epithelial thickening reaching a maximum level beyond which increasing myopic ablation depth will not result in further central epithelial thickening while the peripheral epithelium can still thicken for higher myopic ablations as the peripheral ablation depth increases, the author has postulated a hypothesis to explain this shift.


Important nonlinear biomechanical and epithelial effects have been observed and characterized. Biomechanical changes appear well correlated to the residual stromal thickness, which is a function of the total amount of keratectomy, largely determined by the initial corneal thickness and flap thickness. Therefore, it follows that an accurate knowledge of the residual stromal thickness can be important when deciding to perform further enhancement surgery. Knowledge of the mechanical state of the cornea can be obtained by 3D residual stromal thickness mapping as provided by the Artemis. It is important to differentiate regression due to elastic and stable bowing of the cornea (more pronounced for thin residual stromal thickness) from regression due to epithelial changes (thick residual stroma). If the residual stromal thickness is low and responsible for mechanical changes, further tissue removal could, at best, produce an inaccurate result, but may unfortunately risk converting a stable elastic state into an unstable plastic corneal ectasia. This is particularly true of the newer wavefront-guided repair profiles, which aim to reduce spherical aberration and enlarge the optical zone, as they are very tissue intensive. In the final analysis, true customized ablation may require epithelial and biomechanical predictive modeling for achieving low aberration vision.


Seitz B, Torres F, Langenbucher A, Behrens A, Suarez E. Posterior corneal curvature changes after myopic laser in situ keratomileusis. Ophthalmology. 2001; 108:666-672; discussion 673.
Roberts C. The cornea is not a piece of plastic. J Refract Surg. 2000; 16:407-413.
Srivannaboon S, Reinstein DZ, Sutton HFS, Silverman RH, Coleman DJ. Effect of epithelial changes on refractive outcome in LASIK. Invest Ophthalmol Vis Sci. 1999; 40:S896.
Reinstein DZ, Srivannaboon S, Sutton HFS, Silverman RH, Shaikh A, Coleman DJ. Risk of Ectasia in LASIK: revised safety criteria. Invest Ophthalmol Vis Sci. 1999; 40:S403.
Barraquer JI. Queratomileusis y queratofakia.
Bogota: Instituto Barraquer de America; 1980.
Reinstein DZ, Srivannaboon S, Silverman RH, Coleman DJ. The accuracy of routine LASIK; isolation of biomechanical and epithelial factors. Invest Ophthalmol Vis Sci. 2000; 41:S318.

Question asked from a reader:

Dr. Dan Reinstein writes;

Quote: In the final analysis, true customized ablation may require epithelial and biomechanical predictive modeling for achieving low aberration vision.


Dr. Reinstein, as you know it is very rare for virgin eyes to have high levels of higher order aberration, central flattening, epithelial hyperplasia or excessive bowing. The changes you describe here are caused by an elective, medically unnecessary surgery. Now that you have identified these problems and realize that you can't prevent them, why on earth would you perform even one more surgery on a pair of healthy virgin eyes?