The Role of Bowman’s Layer in Corneal Regeneration after Phototherapeutic Keratectomy: A Prospective Study Using In Vivo Confocal Microscopy

Abstract

purpose. To examine the role of Bowman’s layer (BL) on the nature of anterior corneal regeneration after excimer laser phototherapeutic keratectomy (PTK).

methods. A cohort of 13 patients underwent PTK to remove either 7 μm of BL for treatment of primary recurrent corneal erosions (RCE; six patients) or complete BL removal (15-μm ablation) to treat RCE or poor vision secondary to map-dot-fingerprint (MDF) dystrophy (seven patients). Clinical examinations and laser-scanning in vivo confocal microscopy (IVCM) were conducted before surgery and at a mean of 4 and 8 months after surgery.

results. Total BL removal resulted in a significant decline in subbasal nerve density at 4 months (P = 0.007) that barely recovered to preoperative levels at 8 months (P = 0.055). With BL partially present, subbasal nerve density did not significantly change from preoperative levels. Superficial, wing, and basal epithelial cell density recovered to preoperative levels within 4 months after PTK, regardless of the presence of BL. Subepithelial keratocytes, however, were more densely distributed in corneas without BL relative to those with a partial BL present (P = 0.005), and increased anterior keratocyte reflectivity was noted in all eyes without BL and in no eye with a partial BL present.

conclusions. Subbasal nerve regeneration is delayed and subepithelial keratocyte density and reflectivity remain elevated up to 10 months after total BL removal by PTK. The results provide initial evidence for a possible role of BL in facilitating rapid stromal wound healing and an associated recovery of anterior corneal transparency and the restoration of epithelial innervation after epithelial trauma.

Bowman’s layer (BL) is an acellular, nonregenerating layer located between the epithelial basement membrane and the anterior corneal stroma. ¹ It is composed of randomly oriented collagen fibrils within an extracellular matrix, is approximately 8- to 14-μm thick in humans, ² and is characterized by a smooth anterior surface facing the epithelial basement membrane and an indistinct posterior surface where it merges with the less dense, but ordered, collagen lamellae of the corneal stroma proper. ¹ Unmyelinated nerve axons penetrate BL irregularly across the cornea to ultimately provide epithelial innervation. ³ Although the precise functional role of BL remains to be elucidated, it has been suggested that it may be superfluous to human corneal function, considering the absence of adverse complications in hundreds of thousands of eyes that lack BL after having undergone excimer laser photorefractive keratectomy (PRK). ⁴  

In contrast to its total removal in PRK, BL can be either removed or partially preserved in shallow-depth phototherapeutic keratectomy (PTK) for anterior corneal disorders. ^{5 6 7 8 9 10 11 12 13 14 15 16 17} PTK for the treatment of such disorders therefore provides a unique opportunity to assess the effect of the presence or absence of BL on corneal regeneration in response to anterior trauma. In addition to clinical measures of surgical outcome, a parallel morphologic investigation of PTK-treated corneas could yield insight into the role of BL in corneal recovery at the microscopic, subclinical level. 

Accordingly, we undertook this study to prospectively follow two groups of patients (13 patients in total) indicated for unilateral, shallow-depth PTK. After the procedure, one group had approximately half the thickness of BL removed, while in the other group, BL was removed fully. The treatment goal was to alleviate symptoms of recurrent corneal erosions (RCEs) or to improve vision in patients with primary RCE or map-dot-fingerprint (MDF) dystrophy. We used laser-scanning in vivo confocal microscopy (IVCM) to obtain a detailed morphologic picture of cellular and nerve regeneration in the cornea after PTK treatment. IVCM is an ideal, noninvasive tool for examining the cornea at a morphologic, cellular level that is particularly well suited for longitudinal observation ^{18 19} and corneal cell and nerve quantification. ²⁰  

Our primary goal in this study was to quantify anterior corneal morphology at the cellular level before and after PTK treatment, by using IVCM to comparatively assess corneal regeneration with and without BL. It has been observed that, although the use of PTK to treat anterior disorders of the cornea has been gaining popularity, there are surprisingly few prospective PTK studies ²¹ and there is an absence of information concerning cellular and morphologic changes occurring during the postoperative healing phase. Accordingly, a secondary purpose of the study was to provide a morphologic description of corneal regeneration after PTK. 

Materials and Methods

Patients

A prospective cohort study was conducted in 13 eyes of 13 patients who underwent shallow-depth excimer laser PTK to treat RCEs of various etiologies (12 patients, 6 with MDF dystrophy) or poor vision alone arising from MDF dystrophy (1 patient). The PTKs were performed in consecutive cases by one of two surgeons (PF, JG) at the Linköping University Hospital from January to June 2008. Only eyes without prior surgical treatment were included in the study. All patients gave informed consent to participate, and the study adhered to ethical principles for research involving human subjects as stated in the Declaration of Helsinki. 

Preoperative Evaluation

Patient charts were reviewed before surgery to document symptoms and previous nonsurgical treatments. All eyes with RCE had failed to respond to one or more conventional methods of treatment including topical medication with hyperosmotic agents or lubricants, therapeutic contact lenses, or patching. Preoperative examination on the day of surgery included best spectacle-corrected visual acuity (BSCVA), slit lamp photography, and laser scanning in vivo confocal microscopy (IVCM; HRT3-RCM, Heidelberg Engineering, Heidelberg, Germany). A single operator (NL) performed all IVCM examinations; details of the IVCM procedure have been described elsewhere. ²² In this study, IVCM was used to scan the corneal apical region to obtain a series of 400 × 400-μm cross-sectional images through the full corneal thickness for subsequent nerve and cell quantification. In addition, the epithelium and anterior stroma were scanned laterally over several millimeters of central cornea to document any signs of pathologic morphology, with particular attention given to the epithelium, Bowman’s layer, and the anterior stroma. 

Surgical Technique

After application of local anesthesia (tetracaine 1.0%; Novartis Ophthalmics, Täby, Sweden), the corneal epithelium was removed mechanically with a no. 57 Beaver blade. All patients were then treated with a 193-nm ArF excimer laser system (Technolas 217; Bausch & Lomb, Tampa, FL) with a fluence of 120 mJ/cm² and a repetition rate of 50 Hz. The ablation depth was set to 15 μm for complete removal of BL in MDF cases and 7 μm for removal of half the thickness of BL in other, nondystrophic cases. The treatment zone was a 7.0-mm diameter area of the central cornea in all eyes. 

Postoperative Care and Follow-up

After surgery, patients received cinchocaine hydrochloride 0.5% (Cincain; Ipex Medical, Solna, Sweden) and chloramphenicol 1% (Chloromycetin; Pfizer, Sollentuna, Sweden) eye ointments before installation of an eye patch. During the first postoperative day, cinchocaine ointment was applied every 4 hours and one tablet of dextropropoxyphene (Dexofen 50 mg; AstraZeneca, Södertälje, Sweden) was taken three times. In addition, chloramphenicol ointment was prescribed four times daily for the first five postoperative days. Follow-up examinations for this study occurred at 4 and 8 months after surgery, and included assessment of symptoms, BSCVA, slit-lamp observation, and IVCM examination. 

IVCM Image Analysis

From each IVCM examination, six en-face confocal images from the central cornea were selected by a single observer for quantitative analysis. The first image represented the corneal surface or superficial epithelium, which was the most anterior cell layer that could be visualized. The second image contained epithelial wing cells at a depth of 20 to 25 μm below the surface, the third image contained the most posterior epithelial cells or basal epithelium, and the fourth image contained nerve fiber bundles within the subbasal nerve plexus (subbasal nerves), located immediately posterior to the basal epithelium and running anterior to BL. The fifth and sixth images contained the most anterior layer of corneal keratocytes (without BL visible in the image), and stromal keratocytes 10 to 15 μm posterior to the most anterior keratocyte layer, respectively. These keratocytes, present within the anterior 5% of the corneal stroma, are hereafter referred to as subepithelial keratocytes and were selected for analysis on the basis of their close apposition to the ablated region. BL itself was identified in IVCM images on the basis of its acellularity, slightly increased reflectivity relative to the anterior stromal extracellular matrix, and the visibility of nerve fiber bundles (in a Schwann cell sheath) comprising the underlying subepithelial nerve plexus. ^{3 23} Acellularity was not used as the sole criteria for detecting BL, since cell-like structures resembling keratocyte nuclei were sometimes observed within BL in patients with MDF. Cellular invasion of BL has been similarly noted in other corneal diseases. ⁴  

A total of 234 images were selected for analysis (13 patients, three examinations/patient, and six images/examination). All images were randomized before analysis to mask observers to the patient, depth of ablation, and time of examination. ImageJ image-processing software ²⁴ was used to superimpose a 200 × 100-μm (width × height) rectangle onto each image of superficial, wing, and basal epithelial cells to highlight a region within the image containing cells with the most well-defined borders. For images of subepithelial keratocytes, a similar procedure was used, but with a 250 × 250-μm square placed in the geometric center of the image. Within each selected region (rectangle or square), a band-pass filter was applied to attenuate high- and low-frequency noise (periodic variations below 2 pixels and above 80 pixels, respectively). In this manner, variations in background intensity were suppressed while cell or nuclear features were sharpened to maximize the ability of an observer to discriminate individual cells or nuclei (Fig. 1) . Images with subbasal nerves were not preprocessed in any way before analysis. 

The cells and nerves within 100% of the images were quantified by two independent observers (NL, JG) on separate occasions. The cells were counted manually by each observer with the assistance of point-and-click cell marking and counting software. ²⁴ Only cells with clearly distinguishable borders were counted. Cells crossing the top and right edges of the rectangular region were counted, while those crossing the bottom and left edges were excluded. In addition, for stromal images, only in-focus, bright objects presumed to represent keratocyte nuclei were counted. In cases of presumed fibroblast activity where multiple keratocyte nuclei appeared to aggregate into larger bright regions, an observer counted as many keratocyte nuclei as were presumed to exist within the region, based on region features and the size of surrounding distinct nuclei, similar to the method described by Patel et al. ²⁵ We traced the subbasal nerves by a semiautomated method using nerve tracing software, ²⁶ as described in detail elsewhere. ²⁷ The total number of cells or total nerve length (in pixels) per image was recorded by both observers separately in commercial spreadsheet software (Excel 2003; Microsoft Corp, Redmond, WA). 

Cell density (cells/mm²) and nerve density (μm/mm²) were calculated for each image, and the images were unmasked to enable grouping according to cell type. Interobserver repeatability was determined by computing the 95% limits of agreement for each cell type according to the method of Bland and Altman. ²⁸ Before further statistical analysis, the mean cell or nerve density determined by the two observers was taken as the cell or nerve density for that image. ²⁵ In addition, the mean density of keratocyte nuclei from the two stromal images per cornea was taken to yield a single value of subepithelial keratocyte density per cornea. 

From additional confocal scans obtained across a 3- to 4-mm central region of each cornea, the presence of several morphologic features was determined: Bowman’s layer (presence of at least two confocal images with BL characteristics posterior to the basal epithelium), a population of dendritic (Langerhans-type) cells at the level of the subbasal nerve plexus, ²⁹ epithelial microcysts or dots, ³⁰ characteristic irregularities and folds in the epithelial basement membrane (map or fingerprint features), ³⁰ and increased reflectivity of subepithelial keratocytes (relative to keratocyte reflectivity in deeper anterior stromal layers). 

Statistical Analysis

Cell and nerve densities were grouped according to time and PTK ablation depth. Time-dependent tests were performed on all study patients as a single group, and within two subgroups of patients (full and partial BL removal). For time-dependent comparisons, repeated-measures ANOVA was used. One-way, repeated-measures ANOVA was used when normality and equal variance criteria for densities were met. Significant differences were isolated post hoc with the Student-Newman-Keuls method. For normally distributed data with unequal variances, we opted to use multiple paired t-tests rather than the nonparametric Friedman test, because of to the greater sensitivity of the paired t-test in the case of normally distributed data. In addition, the two patient subgroups were compared at a fixed time by using an independent t-test or the Mann-Whitney rank sum test. In all cases, the Kolmogorov-Smirnov procedure was used to test normality while equal variances were determined with the Levene median test. All statistical tests were two-tailed, and a P < 0.05 was considered statistically significant. Statistical testing was performed with commercial software (SigmaStat 3.5; Systat Software Inc., Chicago, IL). 

Results

Clinical

The cohort consisted of six women and seven men aged 48 ± 14 years (mean ± SD) at the time of surgery. Twelve patients had symptoms of recurrent corneal erosions, while one patient with MDF dystrophy underwent surgery for poor vision. Of the total patient group, 54% (seven eyes) had central corneal features consistent with MDF dystrophy, as determined by preoperative slit lamp observation (Fig. 2)and in vivo confocal microscopy examination. The remaining 46% (six eyes) of patients had recurrent corneal erosions of nondystrophic origin (four cases of corneal trauma and two idiopathic cases). All patients were examined before surgery and at a mean of 4 (range, 3–5) and 8 (range, 6–10) months after surgery. Preoperative and postoperative patient characteristics are given in Table 1 . PTKs were completed without intraoperative complications. At the final follow-up visit, mean BSCVA was 1.0 (20/20 Snellen equivalent; range, 1.0–1.2), without reported recurrence of erosive events in any surgically treated eye. The development of a persistent subepithelial haze during the postoperative period was a minor complication observed with the slit lamp in three patients with MDF and in one patient with RCE, with visual acuity appearing to be unaffected in all cases. 

Cell Quantification and Analysis

The 95% limits of agreement for repeatability of interobserver cell and nerve density determination were superficial epithelium (±45%), epithelial wing cells (±14%), basal epithelium (±24%), subbasal nerves (±34%), and anterior stromal keratocytes (±31%). 

Within the cohort, the density of subbasal nerves was significantly reduced (P = 0.002) at 4 months, relative to the preoperative level. This reduction was followed by a significant increase from 4 to 8 months (P = 0.009) to establish a subbasal nerve density not significantly different (P = 0.14) from the preoperative level (Fig. 3) . Superficial, wing, and basal epithelial cell density and subepithelial keratocyte density after PTK did not significantly differ from preoperative levels (Table 2) . 

When subgrouped by ablation depth, patients without BL exhibited recovery of epithelial cells and subepithelial keratocytes to preoperative density levels within 4 months. Subbasal nerve density, however, was significantly reduced at 4 months (P = 0.009) and remained reduced at 8 months relative to preoperative levels, although the reduction at 8 months failed to reach statistical significance (P = 0.055). By contrast, patients with a partial BL exhibited a less marked reduction in subbasal nerve density at 4 months followed by a more complete recovery at 8 months, without significant change from preoperative levels throughout the postoperative period. In addition, epithelial wing cell density in patients with a partial BL increased significantly (P = 0.03) in the postoperative period. 

Finally, comparison of patients with total BL removal with patients with a partial BL a fixed time yielded no significant differences in cell density before or after PTK, with the exception of subepithelial keratocytes, which had a significantly higher density after total BL removal compared with partial BL removal at 8 months (P = 0.005). 

IVCM Microstructural Findings

Qualitative analysis of corneal morphology by IVCM revealed the presence of BL in all patients before surgery. After surgery, BL was present in six of the six patients who underwent PTK ablation to a depth of 7 μm and in none of the seven patients who had a 15-μm ablation (Fig. 4) . A high density of dendritic cells was observed in the central cornea of two patients with RCE and two with MDF before surgery. Of these patients, only one patient with RCE did not exhibit a dense dendritic cell population after surgery. In addition, in one patient with RCE in whom dendritic cells were not observed before surgery, a population of central dendritic cells was present after surgery (Fig. 5) . Epithelial microcysts or dots were present in six of seven MDF eyes and in three of seven RCE eyes before surgery. At the latest follow-up, no dots were found in any eye. Central corneal map regions were observed in all MDF eyes before surgery, whereas only one eye had observable fingerprint lines. After surgery, no fingerprint lines were found while map regions persisted in two MDF eyes (Fig. 6) . At the latest follow-up, subepithelial keratocytes exhibited increased reflectivity in seven of seven patients without BL and in none of six patients with a partial BL (Fig. 6) . 

Discussion

Our PTK treatment strategy of complete BL removal in patients with MDF and partial preservation of BL otherwise was clinically successful. No recurrence of symptoms was reported, and all patients had excellent vision up to 10 months after surgery. Subbasal nerves, which were completely removed during surgery, only partially regenerated after 4 months, whereas substantial recovery occurred after 8 months. A delay in nerve regeneration was observed in patients without BL relative to patients with a partial BL. In cases of total BL removal, the subepithelial nerve fiber bundles directly posterior to BL (comprising the subepithelial nerve plexus, which subsequently penetrate BL to give rise to the nerves of the subbasal nerve plexus) ^{3 31} may have been damaged or completely removed along with the removal of BL. No nerves from the subepithelial nerve plexus were observed in IVCM images after total BL removal, although the increased anterior stromal reflectivity in these corneas may have masked their presence. By contrast, in eyes with a partial BL present, nerves of the subepithelial plexus were observed posterior to BL in IVCM images. From these observations, it appears that in the absence of both BL and the underlying subepithelial plexus, subbasal nerves are slow to recover as they would presumably regenerate from peripheral nerves outside the treatment zone or from deeper stromal nerves. On the other hand, with a partial BL present, preservation of the underlying subepithelial nerve plexus would promote faster regeneration of central subbasal nerves. It is postulated that in cases of epithelial trauma, BL may therefore serve as a protective barrier to the underlying subepithelial nerve plexus—the presence of which provides for quick recovery of the subbasal nerves essential for epithelial innervation and restoration of the protective corneal aversion reflex. ³²  

In our cohort, the density of cells within the various epithelial layers recovered to preoperative levels within the first few postoperative months. Apart from map regions present within the treatment zone in two patients with MDF, the regenerated epithelium after PTK appeared morphologically normal under confocal microscopic examination. Moreover at any given time, epithelial, nerve, and subepithelial keratocyte densities were apparently independent of the presence of BL, with the only exception being a significant increase in subepithelial keratocyte density 8 months after surgery in corneas without BL. This increased density manifested in the confocal images as an aggregation of keratocytes into regions of increased scatter, interpreted as wound-healing fibroblast activity. ^{33 34} Trauma from the deeper ablation (reaching the anterior stromal surface) was likely to result in prolonged stromal activation and wound healing. Qualitatively, by IVCM we observed a persistent stromal activation in all cases where BL was removed and an absence of activation where BL was present. Clinically, a mild subepithelial haze was present on slit lamp examination. A similar mild or transient postoperative haze has been reported in some patients after shallow-depth PTK, ^{11 13 35} and our morphologic findings suggest that this haze may be more prevalent at the subclinical, morphologic level, particularly in patients without BL. Similar morphologic findings by IVCM have been reported in the context of PRK, where subsequent normalization of keratocyte density and reflectivity was associated with restoration of corneal transparency and disappearance of haze at a minimum of 12 months after the procedure. ³³ It has been suggested that only wounds that traverse BL into the stroma result in prolonged myofibroblast transformation. ³⁴ Our present observations of subepithelial keratocyte activity therefore suggest a further possible function of BL in accelerating the process of anterior stromal wound healing and the restoration of corneal transparency after epithelial trauma. 

It is interesting to note that the possible functions of BL outlined herein are consistent with observations of a sensory deficit and postoperative haze in PRK-treated eyes devoid of BL. ^{33 34 36} Although an apparent lack of significant complications has been suggested in the hundreds of thousands of PRK-treated eyes devoid of BL, ⁴ we suggest that future epithelial trauma in these eyes may result in prolonged stromal wound healing and recovery of transparency and corneal sensation, compared with eyes with BL present. 

The clinical outcome in our cohort suggests a re-examination of shallow-depth treatment strategy in patients with MDF presenting with recurrent erosions. Nonzero rates of recurrence have been reported in patients with MDF who undergo PTK ablation to a depth of 5 to 8 μm ^{7 8 10 14 16 35} and in patients with RCE (with and without dystrophy) who undergo ablation as shallow as 5 μm. ^{13 37} On the other hand, in 15 patients with MDF who received PTK ablation to a mean depth of 10 μm, Pogorelov et al. ¹¹ reported no recurrence of erosion after a mean follow-up of 4.8 years, while Örndahl and Fagerholm ¹² reported one instance of recurrence in 10 patients who underwent PTK ablation to a mean depth of 11 μm, after a mean follow-up period of 24 months. In patients with MDF, both in this prospective study and in an earlier series of 10 patients treated in our clinic for recurrent erosions with a mean ablation depth of 16 μm, no recurrences were reported, with a mean follow-up time of 25 months (range, 13–47 months) in the earlier series (Germundsson JR, Fagerholm PP, manuscript in preparation). These studies indicate a possible association between post-PTK recurrence in patients with MDF and the partial presence of BL. Improved outcome after repeat procedures of 5- to 7-μm depth PTK ^{8 14 16 37} (which would result in total BL removal) provides support for such an association. We surmise that in some cases of MDF where BL may be affected, partial ablation of BL may not be sufficient to prevent recurrence of the underlying disease. 

We noted in this study that in patients with clinical symptoms of RCE, reliance on slit lamp findings alone can sometimes make accurate differential diagnosis of idiopathic RCE from MDF dystrophy difficult. In a retrospective study of 22 patients with MDF dystrophy, Labbé et al. ³⁰ reported three (13.6%) patients with a normal corneal structure on slit lamp examination who were diagnosed with MDF dystrophy only after IVCM examination. In our cohort, one presumed RCE patient was similarly rediagnosed with MDF dystrophy after map, dot, and fingerprint features were observed with IVCM. The preoperative examination in this case enabled us to modify the PTK treatment strategy for complete BL removal. In this patient, a preoperative visual acuity of 0.4 (20/50 Snellen equivalent) subsequently improved to 1.0 (20/20) after 6 months without recurrence of erosion. We suggest that IVCM may be a useful screening tool before PTK treatment in patients with presumed RCE. In addition, IVCM may also be useful in determining preoperative BL thickness before PTK. Given the interindividual variability in BL thickness, ³⁸ it is conceivable that in some patients total BL removal may not be achieved with an ablation depth of 15 μm and conversely in some patients with a thin BL, 7-μm ablation may result in total BL removal. 

Dendritic cells were observed in a few patients in this study, both before and after PTK, and exhibited a morphology consistent with both mature (cells bearing dendritic processes) and immature (cell bodies only) forms. ^{29 39} High densities of dendritic cells residing in the central cornea are thought to accompany irritation, ⁴⁰ wound healing, ²⁹ or inflammation, ^{29 39 40} all of which may have been present to various degrees in the corneas examined in this study. The significance of our findings of dendritic cells is unclear. Given the ongoing discussion surrounding the interpretation of findings of dendritic cells in the central cornea, ³⁹ a more detailed study of their relationship to RCE, MDF dystrophy, and the PTK treatment is warranted. 

One limitation of this study was the small size of the central corneal region used for quantitative analysis. Because many patients with MDF did not have map, dot, or fingerprint features present at the central apical region, the cell densities (for wing and basal epithelial cells) would not represent the true density averaged over a larger corneal region. The true density would be lower due to an apparent acellularity in regions with microcysts or basement membrane folds. Accordingly, although epithelial cell densities as reported in this study did not generally change after PTK treatment, an absence of map, dot, or fingerprint features after treatment would probably have resulted in a marked increase in epithelial cell density, if a larger corneal area were to be considered. 

A relatively high interobserver variability was reported in the quantification of a few of the corneal parameters in this study. Borders of superficial epithelial cells were often difficult to detect because of a high specular reflection, variable reflectivity and desquamation of epithelial cells, ²² and a general reduction in image contrast of this outermost corneal layer as visualized with the laser scanning in vivo confocal microscope. In addition, basal epithelial cell borders were sometimes indistinct, and subepithelial keratocytes were sometimes difficult to quantify in the presence of high background reflectivity (in the wound-healing phase). The best agreement was for epithelial wing cells due to their dark cytoplasm with highly reflective cell borders. Considering our analysis of 100% of images by both observers and given the nature of the corneas, the limits of interobserver agreement are reasonable in comparison to other studies employing a similar method of analysis. ^{41 42}  

In conclusion, while the absence of BL does not appear to impede recovery of stratified epithelial morphology after shallow-depth PTK, two potential roles of BL after epithelial trauma in the human cornea were identified. BL may present a physical barrier that protects the subepithelial nerve plexus and thereby hastens epithelial innervation and sensory recovery, and may also serve as a barrier that prevents direct traumatic contact with the corneal stroma and thus accelerates stromal wound healing and the associated restoration of anterior corneal transparency at the morphologic level. Further studies with a larger patient population could provide additional evidence to support these putative roles. 

 Supported by grants from the County Council of Östergötland (PF) and the Carmen and Bertil Regnérs Foundation (NL).

 Submitted for publication March 30, 2009; revised April 16, 2009; accepted June 30, 2009.

 Disclosure: N. Lagali, None; J. Germundsson, None; P. Fagerholm, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

The appearance of preprocessed IVCM images before cell counting. Band-pass filtering within the region of interest enhanced the ability to visualize borders of distinct cells. (A) The 200 × 100-μm region of interest in the basal epithelial cell layer. (B) The 250 × 250-μm region of interest in the subepithelial stroma.

Figure 2.

 

Preoperative slit lamp photographs of corneas from (A) a 57-year-old male patient with MDF dystrophy, with geographic map regions clearly visible (arrow), and (B) a 33-year-old female with recurrent corneal erosions, with dotlike epithelial deposits (arrow).

Table 1.

 

Clinical Characteristics of the Study Cohort

Patient No.	Age	Sex	Condition	Eye	Ablation Depth (μm)	BSCVA			Final Follow-up (mo)	Postoperative Complications*
														Preop	Final
1	33	F	RCE	OS	7	1.0	1.2	9	Subepithelial haze
2	39	F	RCE	OD	7	1.0	1.0	8
3	30	M	RCE	OD	7	1.0	1.0	7
4	63	M	RCE	OS	7	0.9	1.0	8
5	52	M	RCE	OD	7	1.0	1.0	9
6	50	F	RCE	OD	7	0.5	1.0	8
7	29	F	MDF	OS	15	1.0	1.0	7	Subepithelial haze
8	60	F	MDF	OS	15	0.2	1.0	10
9	57	M	MDF	OS	15	1.0	1.0	6
10	65	M	MDF	OD	15	0.7	1.0	7
11	50	M	MDF	OS	15	1.0	1.0	7
12	31	F	MDF	OD	15	0.4	1.0	6	Subepithelial haze
13	65	M	MDF	OD	15	1.0	1.0	6	Subepithelial haze

*  Subepithelial haze indicates a diffuse hazy appearance or discrete, punctate, or fleck opacities.

Figure 3.

 

Subbasal nerve regeneration after PTK in a 29-year-old female with MDF dystrophy. (A) Preoperative subbasal nerve plexus. (B) Four months after PTK, presumed sprouting subbasal nerves (white arrows) and a regenerating subbasal nerve (arrowhead) were present at the basal epithelium. (C) Seven months after PTK, presumed sprouts (white arrow) and mildly tortuous regenerated subbasal nerves (arrowheads) were visible at the basal epithelium. Activity in the subepithelial keratocyte layer is observed in the same image frame (black arrows) indicating the absence of BL. Image size, 400 × 400 μm.

Table 2.

 

Cell or Nerve Density in the Cohort

PTK Ablation Depth/Cell Layer	Time of Examination					ANOVA P *
		Preoperative	4 mo	8 mo
All (n = 13)
A	1,163 ± 272	1,115 ± 268	1,113 ± 311	0.90
B	5,729 ± 1,018	6,219 ± 799	6,063 ± 654	0.21
C	9,538 ± 995	9,273 ± 1,480	9,517 ± 1,756	0.84
D	15,669 ± 9,483	5,447 ± 3,541	10,553 ± 6,467	0.002/0.009/0.14
E	735 ± 249	815 ± 240	725 ± 229	0.54
15 μm (n = 7)
A	1,243 ± 306	1,118 ± 251	1,189 ± 353	0.79
B	6,000 ± 851	5,893 ± 381	5,757 ± 642	0.69
C	9,717 ± 1,223	9,125 ± 564	9,207 ± 1,808	0.57
D	16,180 ± 10,660, †	3,918 ± 2,411, †	9,280 ± 4,897	0.009, ‡
E	852 ± 245	801 ± 190	858 ± 188, §	0.88
7 μm (n = 6)
A	1,071 ± 215	1,113 ± 311	1,025 ± 256	0.88
B	5,413 ± 1,181, † ∥	6,600 ± 1,017, †	6,421 ± 498∥	0.03
C	9,329 ± 694	9,446 ± 2,192	9,879 ± 1784	0.80
D	15,072 ± 8,870	7,231 ± 4,004	12,038 ± 8,166	0.07/0.13/0.63
E	630 ± 223	733 ± 169	535 ± 128, §	0.22

 Data are the mean ± SD. Cell layers are defined as follows: A, superficial epithelial cells (cells/mm²); B, epithelial wing cells (cells/mm²); C, basal epithelial cells (cells/mm²); D, subbasal nerves (μm/mm²); and E, subepithelial keratocytes (cells/mm²). Statistically significant (P < 0.05) results are indicated in bold.

*  One-way, repeated-measures ANOVA was used where normality and equal variance criteria were met. For normally distributed data with unequal variance, a paired t-test was used, with probabilities reported for the pair-wise comparisons: preoperative/4 mo, 4 mo/8 mo, and preoperative/8 mo, respectively.

†  Values with the same symbol are significantly different.

‡  Preoperative/4 mo significance is P = 0.007 and preoperative/8 mo significance is P = 0.055 (Student-Newman-Keuls method).

§  Values differed significantly from each other (P = 0.005; independent t-test).

Figure 4.

 

Corneal morphology in a 31-year-old woman with MDF dystrophy. (A) The pre-PTK BL exhibited increased reflectivity in patches, with narrow folds evident (black arrows) and unidentified cell or nuclear inclusions present within BL (white arrows). (B) Seven months after PTK, basal epithelial cells (b), keratocyte nuclei (white arrows), and a regenerated subbasal nerve (black arrow) were present at the same corneal depth, indicating the absence of BL. Image size, 400 × 400 μm; depth of images from the corneal surface, (A) 43 and (B) 46 μm.

Figure 5.

 

Basal epithelial layer in the central cornea of a 33-year-old female patient with RCE (A, B) and in a 39-year-old female patient with RCE (C, D). (A) Before surgery, dendritic cell bodies, some with dendrites (arrowhead) occupied the central cornea (>100 cells per field). (B) Nine months after surgery, a population of dendritic cell bodies remained (>50 per field). (C) Before surgery, the basal epithelium was free of dendritic cell bodies. (D) Eight months after surgery, dendritic cell bodies were present (25 per field). Note the presence of a partial BL with the posterior surface defined by the existence of subepithelial nerve fiber bundles (arrows). Image size, 400 × 400 μm; depth of images from the corneal surface: (A) 40, (B) 42, (C) 52, and (D) 52 μm.

Figure 6.

 

Geographic map regions in the central cornea of a 65-year-old male patient with MDF (A, B) and the most anterior stromal layer in the central cornea of a 50-year-old male patient with MDF (C, D). (A) Preoperative map region with presumed activated keratocytes present (arrows). (B) Four months after surgery, a map region is present at the level of the basal epithelium. (C) Before surgery, distinct keratocyte nuclei were visible just posterior to BL. (D) Seven months after surgery, patches of increased stromal reflectivity were observed just posterior to the basal epithelium. Image size 400 × 400 μm; depth of images from the corneal surface: (A) 69, (B) 45, (C) 52, and (D) 39 μm.