Age-Related Thinning of Bowman's Layer in the Human Cornea In Vivo

Abstract

Purpose.: To determine the thickness of Bowman's layer (BL) in vivo in a healthy population and to determine its variation with age.

Methods.: Eighty-two subjects aged 15 to 88 years with clear, healthy corneas were examined bilaterally with laser scanning in vivo confocal microscopy (IVCM). Bowman's layer thickness was determined from IVCM images of anterior and posterior BL boundaries. For a given eye, BL thickness was averaged across four central locations by two independent observers. In addition, central corneal thickness was measured by time-domain optical coherence tomography.

Results.: A significant negative correlation of BL thickness with age was found in right eyes (Pearson r = −0.579, P < 0.0001) and in left eyes (r = −0.558, P < 0.0001). Linear regression analysis yielded a decline in BL thickness of 0.06 μm per year. In 41 older subjects (mean age, 64.4 years), BL thickness was significantly thinner (mean ± SD, 8.6 ± 1.7 μm in right eyes) than that in 41 younger subjects (mean age, 31.6 years) (mean ± SD, 10.7 ± 1.6 μm in right eyes) (P < 0.001). No correlation of corneal thickness with age or of BL thickness with corneal thickness was observed. Strong intereye correlations in BL thickness (r = 0.771, P < 0.0001) and corneal thickness (r = 0.969, P < 0.001) were found.

Conclusions.: Bowman's layer thins with age in the normal cornea, losing one-third of its thickness between the ages of 20 and 80 years. In vivo measurement of BL thickness by IVCM could aid in clinical assessment and planned treatments of the anterior cornea.

Introduction

Bowman's layer (BL) is a normally acellular layer located immediately posterior to the corneal epithelium. It is structurally composed of collagen fibrils, smaller in diameter and randomly oriented compared with stromal collagen, in which fibrils are larger and directionally oriented in parallel lamellae. ^1,2 The functional role of BL is not completely known, but it is believed to serve as a barrier that protects the corneal stroma from traumatic injury. ^3–5 In this manner, BL aids in the maintenance of corneal transparency and epithelial innervation in cases of injury, disease, or surgical intervention. ⁵  

The protective nature of BL may be related to its structural strength and thickness. Knowledge of BL thickness could enable an assessment of its protective ability, and this parameter is also of importance when planning various types of laser surgery. ^5–7 Much discrepancy, however, exists in published values of the mean BL thickness in humans, with values ranging from 8 to 19 μm depending on the type of measurement (in vivo or ex vivo), the tissue preparation method, and the measurement technique. ^2,7–13  

Recently, we developed an accurate method to measure BL in vivo based on laser scanning in vivo confocal microscopy (IVCM). ¹⁴ By this method, three-dimensional laser scanning confocal depth scans of BL are obtained in live subjects, and an image analysis procedure is used to determine the upper and lower boundaries of BL. In contrast to other in vivo techniques in which reflectivity of epithelial and stromal interfaces is used to infer BL thickness, our method directly examines morphologic features of BL to determine its boundaries. 

To date, no studies have examined BL thickness in a large, healthy population for several reasons, including the difficulty in obtaining healthy donor cornea tissue from a wide age range, the unknown effects of tissue fixation and shrinkage of BL, and the difficulty in accurately measuring this thin layer in vivo with lower-resolution techniques. In this study, we used laser scanning IVCM to accurately measure BL thickness in the central cornea in a large, healthy population with a wide age range. 

Methods

Recruitment and Clinical Examination of Subjects

After obtaining approval from the regional ethical review board in Linköping, Sweden, and with signed informed consent, 115 healthy volunteers were recruited for this study. The study adhered to the tenets of the Declaration of Helsinki. Volunteers were recruited from the population of persons accompanying patients to the Department of Clinical and Experimental Medicine–Ophthalmology at Linköping University. Each volunteer underwent full ophthalmic examination in both eyes, including slitlamp biomicroscopy, intraocular pressure, refraction, best spectacle-corrected visual acuity (BSCVA), IVCM, and anterior segment optical coherence tomography (OCT). Individuals with ocular pathology or diabetes mellitus, a history of contact lens wear, prior corneal surgery, or a current eye medication regimen were excluded. No further selection criteria were initially used; however, as the study progressed, individuals were selected based on age to achieve a target final cohort consisting of four age groups (15–30, 31–45, 46–60, and ≥61 years) of 20 individuals each. 

Measurement of Central Corneal Thickness

Using anterior segment OCT (Visante; Carl Zeiss Meditec, Jena, Germany), all participants underwent a global high-resolution pachymetry scan bilaterally before IVCM examination. Central corneal thickness was obtained as the mean value in the central 2-mm diameter zone as determined by the OCT software. 

IVCM of BL

An in vivo confocal microscope (HRT3-RCM; Heidelberg Engineering, Heidelberg, Germany) was used to conduct all examinations; details of the IVCM procedure have been described elsewhere. ^14,15 Subjects were examined bilaterally, and up to 10 confocal scans were obtained for each examined eye. Scans were obtained in both volume and section scan modes. In the volume scan mode, depth is automatically scanned by an internal motor to obtain 40 images spanning a thickness of 80 μm. The volume scan mode was initiated when a central corneal wing cell layer was visible in the real-time image display, and the scan automatically terminated in the anterior stroma. The section scan mode was used to acquire scans with 100 images at eight frames per second, while the focal depth was varied manually to cross BL two to three times during the scan. Manual depth adjustment was facilitated by the use of a joystick-driven motor control. After each volume or section scan, the lateral positioning of the microscope objective was adjusted (in a random direction but keeping within the nonoblique central zone) to image a different central location, and scanning was repeated. Typically, several central corneal locations were scanned for each eye to sample a larger region of the central cornea. Because the volume scans resulted in more closely spaced images axially (separated by approximately 2 μm in depth), most of the scans obtained were volume scans. 

Measurement of BL Thickness

The IVCM scans were used to determine BL thickness based on a previously reported method ¹⁴ but with the following adjustments. First, images from volume scans were used where possible, and image sequences with obvious motion artifacts (lateral and/or axial) were excluded. Also, obviously tilted (oblique) images and images with extensive artifacts due to mechanical pressure on the cornea were excluded. Two observers (JG and NL) selected the final scans for further analysis. Scans were ordered in a randomized manner, with observers masked to the patient age. Next, the selected scans were exported as sets of individual images without image depth or patient-identifying information. For each sequence of images, only the morphologic features in a small subregion of the image frame were considered to further avoid the influence of image obliqueness and pressure-induced artifacts. The same subregion of the image was observed in successive image frames to identify the features defining the anterior and posterior limits of BL. The anterior limit of BL was considered the most anterior epithelial layer with haze (often with a slight haze and subbasal nerves visible). The posterior limit of BL was considered the most anterior layer of stroma with haze where indistinct keratocytes were visible. The image numbers corresponding to the anterior and posterior BL limits were recorded, and only later was the image depth unmasked. Bowman's layer thickness was determined independently by two trained observers (JG and NL). For each eye, a mean BL thickness was calculated from measurements of thickness in approximately four different regions in the central cornea. 

Quantitative Analysis and Statistical Analysis

The Bland-Altman method ¹⁶ was used to compare differences in BL thickness measurements between the two independent observers. Interobserver differences are reported as the 95% limits of agreement. 

Interobserver correlation of BL thickness and correlation of central corneal thickness or BL thickness with eye and age were determined by the Pearson product moment correlation test. Difference in BL thickness across age groups was tested with one-way ANOVA and the Student's t-test. Differences in corneal thickness or BL thickness in male and female subjects were tested with the Student's t-test. Statistical and regression analyses were performed using the software package SigmaStat 3.5 for Windows (Systat Software Inc., Chicago, IL), and two-tailed P < 0.05 was considered significant. 

Results

Subject Characteristics

Of 115 examined volunteers, 82 subjects (164 eyes) were included in the study. Patients were excluded because of the presence of corneal haze, localized corneal scarring, endothelial pigment deposits, diabetes mellitus, corneal dystrophy, dry eye, prior corneal trauma or surgery, or IVCM images without sufficient quality for analysis. Characteristics of the subjects included in the study are summarized in Table 1. The mean subject age was 48 years (age range, 15–88 years) at the time of examination. Roughly equal numbers of participants were examined in each age category. 

Table 1. 

 

Demographic Characteristics of Healthy Volunteer Subjects

Variable	Total Group	Age Group, y
		15–30	31–45	46–60	≥61
Subjects, n	82	20	20	18	24
Sex, n
Female	45	10	8	15	12
Male	37	10	12	3	12
Age, mean, y	48.0*	25.1	37.4	54.4	71.2

*  The mean ages were 48.9 years for female subjects and 46.9 years for male subjects.

Of 164 eyes examined, 67 right eyes were emmetropic (spherical equivalent refraction greater than −0.5 diopter [D] but less than +1.5 D), 8 were hyperopic (+1.5 D or greater), and 7 were myopic (−0.5 D or less). Sixty-eight left eyes were emmetropic, seven were hyperopic, and seven were myopic. Fifty-three of 82 subjects had a bilateral uncorrected visual acuity of 1.0 (20/20 Snellen equivalent). The mean ± SD decimal BSCVAs were 1.0 ± 0.3 lines in right eyes and 1.0 ± 0.3 lines in left eyes. The mean ± SD intraocular pressures were 15.4 ± 2.8 mm Hg in right eyes and 15.5 ± 3.0 mm Hg in left eyes. 

Central Corneal Thickness

The mean ± SD central corneal thicknesses by OCT were 535.4 ± 31.3 μm in right eyes and 533.9 ± 31.2 μm in left eyes. Thickness in right eyes and left eyes was highly correlated (Pearson r = 0.969, P < 0.001). No significant correlation of central corneal thickness was found with age in right eyes (r = −0.038, P = 0.74) or left eyes (r = 0.020, P = 0.86) (Fig. 1); however, the SD increased with increasing age (Table 2). No difference in corneal thickness between female and male subjects was found for either eye. 

Figure 1. 

 

Variation in central corneal thickness with age determined by OCT. The linear regression line is indicated. No correlation with age was observed in either eye; however, thicknesses in right and left corneas among subjects were highly correlated. The SD of corneal thickness increased with age.

Table 2. 

 

Central Corneal Mean (SD) Thickness in Micrometers by OCT, Grouped by Age and Eye

Eye	Age Group, y
	15–30	31–45	46–60	≥61
Right	534 ± 27	542 ± 30	537 ± 34	532 ± 35
Left	531 ± 28	539 ± 27	538 ± 33	530 ± 36

Interobserver Differences in BL Thickness

For each subject and eye, IVCM image scans were analyzed to select those that did not contain significant motion-induced artifacts (in axial or lateral directions). For all eyes, two to five image scans contained suitably stable images for determination of BL thickness at different central corneal locations. Data from volume scans were used to determine BL thickness; however, in some cases in which suitable volume scans were not available, section scans were used. In 82 subjects, BL thickness was determined at a mean of 3.9 different central corneal locations in the right eye and at a mean of 3.7 in the left eye. For each eye, a single mean central BL thickness was determined by averaging values from the different locations. The mean value for each eye was compared between observers (Fig. 2). The values between observers were highly correlated (P < 0.001 for both eyes); however, inclusion of the slope equals one line in Figure 2 indicated a tendency of observer 1 to record slightly thicker values than observer 2 in cases in which BL thickness exceeded 10 μm. 

Figure 2. 

 

Bowman's layer thickness measurements between two observers were highly correlated (P < 0.001). The dashed line indicates the slope equals one line of perfect positive correlation.

Bland-Altman analysis indicated a mean difference of 0.05 μm in BL thickness between observers in right eyes and 0.11 μm in left eyes. The 95% limits of agreement were ±3.4 μm for right eyes and ±3.0 μm for left eyes (Fig. 3). 

Figure 3. 

 

Bland-Altman plots of interobserver agreement in BL thickness measurement. Each circle represents the eye of one subject. The solid horizontal line indicates the mean interobserver difference, while dashed lines indicate the lower and upper boundaries for the 95% limits of agreement.

Age Dependence of BL Thickness

Bowman's layer thicknesses for each eye were averaged across observers to yield a single value for central BL thickness. The resulting BL thickness was significantly negatively correlated with subject age in right eyes (r = −0.579, P < 0.0001) and in left eyes (r = −0.558, P < 0.0001) (Fig. 4). In addition, BL thickness in right eyes and left eyes of the same subject was highly correlated (r = 0.771, P < 0.0001). Regression analysis yielded the following formulas for BL thickness:  

 

 

 

which corresponds to a decline in thickness of approximately 0.06 μm per year.  

Figure 4. 

 

Decline in BL thickness with age in right eyes (top plot) and left eyes (bottom plot) of healthy subjects. The solid line indicates the best-fit linear regression line, while dashed lines indicate the 95% confidence intervals for BL thickness. Bowman's layer thickness declined significantly with increasing age (Pearson r < 0.0001 for both eyes).

When compiled according to the four age categories, a clear decline in the mean BL thickness was evident (Table 3). As age increased, the minimum and maximum BL thicknesses in the age groups generally decreased, while the variation in BL thickness within groups remained at a constant 6 to 7 μm. One-way ANOVA across age categories indicated a significant age dependence of BL thickness (P < 0.001 for both eyes); however, pairwise comparisons indicated no significant difference between the two youngest categories or the two oldest age categories independent of eye (P > 0.05). Age categories were then collapsed into two categories consisting of subjects younger than or older than (or equal to) the mean age of 48 years. The numbers of subjects in each of the two categories were equal (41 subjects). For both eyes, the Student's t-tests indicated that in older subjects (mean ± SD age, 64.4 ± 9.3 years) BL thickness was significantly thinner (mean ± SD, 8.6 ± 1.7 μm in right eyes and 8.4 ± 1.9 μm in left eyes) relative to that in younger subjects (mean ± SD age, 31.6 ± 7.6 years) (mean ± SD, 10.7 ± 1.6 μm in right eyes and 10.7 ± 1.8 μm in left eyes) (P < 0.001 for both eyes). No difference in BL thickness was found between male and female subjects for either eye. 

Table 3. 

 

Central BL Thickness Categorized by Age Group

BL Thickness, μm	Eye	Age Group, y
		15–30	31–45	46–60	≥61
Mean ± SD	Right	11.0 ± 1.7	10.5 ± 1.4	9.1 ± 1.8	8.3 ± 1.6
	Left	11.0 ± 2.2	10.5 ± 1.3	8.9 ± 2.0	8.0 ± 1.7
Range	Both	7.5–14.5	7.8–13.1	6.0–13.2	5.5–12.0

Correlation of Total Corneal Thickness and BL Thickness

No correlation was evident between central corneal thickness measured by OCT and central BL thickness measured by IVCM in either the right eyes or the left eyes. The values were r = 0.10 (P = 0.38) for right eyes and r = 0.12 (P = 0.27) for left eyes. 

Discussion

In this study, the largest to date to measure in vivo BL thickness, a strong negative correlation of BL thickness with age was found. Bowman's layer becomes thinner with age, with a loss of approximately 32% in thickness between the ages of 20 and 80 years. Stated in another way, BL is on average 20% thinner in older subjects compared with younger subjects. Although the negative correlation of BL thickness with age was highly significant, a large variation in BL thickness across individuals was noted, confirming our previous in vivo and light microscopy measurements in smaller cohorts. ¹⁴ Notably, the thinnest values for BL were approximately 50% thinner than the thickest values in the same age group. This large variation has been reported in our previous study ¹⁴ and is confirmed by other studies using slit scanning confocal microscopy ¹³ and OCT. ¹¹ The variation in BL thickness in a healthy population is much larger in relative terms than the variation observed in total corneal thickness or epithelial thickness. ^7,8,13 This large variation in BL thickness may be one reason why its age dependence has not been previously reported; a correspondingly large population of normal corneas, evenly distributed in age, is required to detect correlation when variability is large. 

Although the reason for this age-dependent thinning is not known, one hypothesis may be that BL collagen, as with stromal collagen, becomes naturally cross-linked with age. Bowman's layer is a nonregenerating layer of the cornea, ³ so its collagen is not renewed (or spatially organized) by keratocytes because BL is acellular. The gradual cross-linking of BL collagen could, therefore, result in gradual thinning of BL without a compensatory mechanism of new collagen production and deposition. An alternative hypothesis may be a gradual compacting of BL collagen with age due to collagen degradation, hydrostatic pressure, or some other mechanism. These hypotheses could be investigated by immunochemical methods and/or electron microscopy using a large sample of human donor corneas with wide age variation. It must be kept in mind, however, that tissue preservation and fixation may lead to preparation-dependent artifacts that could influence the apparent thickness. 

In contrast to BL thickness, total corneal thickness was found to remain constant with age at the resolution available with the time-domain OCT used. The age dependence of central corneal thickness in normal, healthy corneas is well investigated in studies ^17–20 examining large groups of individuals. A slight age dependence has been reported in some studies, ¹⁸ while no age correlation was found in other studies. ^17,19,20 The values and SDs for central corneal thickness obtained in our Swedish population, however, agree well with the literature-reported values using ultrasonographic pachymetry, ^17,18,20 Scheimpflug photography, ¹⁹ slit scanning confocal microscopy, ¹³ and spectral-domain OCT.¹¹  

Although no correlation was found between BL and total corneal thickness in this study, the axial resolution of the time-domain OCT used (18 μm) ²¹ is poorer than that of laser scanning IVCM (4 μm). ²² While spectral-domain OCT has a better resolution than time-domain OCT, corneal thickness still cannot be measured with present instrumentation at the level of precision required (1–2 μm) to detect a possible correlation with BL thickness. 

Further studies are needed to corroborate our BL thickness values, which are markedly thinner than other in vivo values reported. To date, in vivo BL thickness has been measured by slit scanning confocal microscopy ^10,13 and spectral-domain OCT. ^7,11,12 In these studies, however, measurements were based on interface reflectivity of the posterior epithelium and anterior stroma; it has been noted that these reflective interfaces likely arise from subbasal epithelial nerves and anterior stromal keratocytes, respectively, and not the anterior and posterior aspects of BL. ^10,14 From the present analysis of successive image frames spaced 1 to 2 μm axially, it can be confirmed that each plane of maximum reflectivity is separated from a BL boundary by at least 2 μm, resulting in a 4-μm minimum overestimation of BL thickness by reflectivity analysis. 

One method to corroborate and potentially improve on the results reported herein would be to use laser scanning IVCM in a manual scan mode with motorized joystick-driven depth adjustment. By adjusting the frame acquisition rate to 30 frames per second and the motorized depth adjustment to the slowest speed, manually scanning through a limited depth of the anterior cornea can yield images spaced 0.5 μm axially. ²³ This approach, however, is technically challenging. 

A limitation of the method used in this study is in the subjective interpretation of images for determining BL boundaries. The method was based on earlier investigation of light and electron microscopic features of BL in the same corneas examined by IVCM ¹⁴ ; however, the interpretation of these borders is clearly observer dependent, as indicated by the Bland-Altman plots. In an attempt to overcome this limitation and improve the objectivity of the analysis, images were randomly and independently examined, image depth information was not available to observers until after boundaries were identified, and final thickness values were averaged across several scans and across both observers. 

Another limitation was that only the central cornea was imaged. A mean of four different central corneal locations was used; however, no information on BL thickness outside the central region was available. In two studies ^7,11 using spectral-domain OCT, spatially resolved BL thickness measurements were reported, and local variations in BL thickness were noted. Future studies using IVCM could image midperipheral, peripheral, and limbal locations in various sectors of the cornea to yield spatially dependent information, although the IVCM scanning procedure involves considerably longer examination times compared with fast OCT scans. 

Nevertheless, the strong age dependence of BL thickness and the intersubject variation found in the present study suggest that measurement of a patient's BL by an in vivo method could be useful in clinical assessment of the cornea. For example, treatment of epithelial basement membrane dystrophy by excimer laser ablation involves the removal of BL to promote epithelial anchoring. ⁵ In this context, it is advantageous to know the exact thickness of BL to avoid unnecessary ablation of deeper stromal tissue, which could lead to scarring and nerve damage. ⁵ Treatment of keratoconus by UV-A–riboflavin collagen cross-linking may also benefit from knowledge of BL thickness because individual variations in this layer could affect the penetration of riboflavin and/or absorption of UV energy, leading to a variation in cross-linking effect. In addition, because BL has been thought to serve as a barrier against corneal infection, ³ information about its thickness could be of potential value in assessing risk or progression of infection. 

Acknowledgments

Supported by the Swedish Research Council, Konung Gustav V and Drottning Viktorias Frimurarestiftelse, and the County Council of Östergötland (PF) and by the Cronqvist Foundation (NL). 

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