Use of Nanoindentation in Determination of Regional Biomechanical Properties of Rabbit Cornea After UVA Cross-Linking

Abstract

Purpose: To evaluate the regional effects of different corneal cross-linking (CXL) protocols on corneal biomechanical properties.

Methods: The study involved both eyes of 50 rabbits, and the left eyes were randomized to the five intervention groups, which included the standard CXL group (SCXL), which was exposed to 3-mW/cm² irradiation, and three accelerated CXL groups (ACXL1–3), which were exposed to ultraviolet-A at irradiations of 9 mW/cm², 18 mW/cm², and 30 mW/cm², respectively, but with the same total dose (5.4 J/cm²). A control (CO) group was not exposed to ultraviolet-A. No surgery was done on the contralateral eyes. The corneas of each group were evaluated by the effective elastic modulus (E_eff) and the hydraulic conductivity (K) within a 7.5-mm radius using nanoindentation measurements.

Results: Compared with the CO group, E_eff (in regions with radii of 0–1.5 mm, 1.5–3.0 mm, and 3.0–4.5 mm) significantly increased by 309%, 276%, and 226%, respectively, with SCXL; by 222%, 209%, and 173%, respectively, with ACXL1; by 111%, 109%, and 94%, respectively, with ACXL2; and by 59%, 41%, and 37%, respectively, with ACXL3 (all P < 0.05). K was also significantly reduced by 84%, 81%, and 78%, respectively, with SCXL; by 75%, 74%, and 70%, respectively, with ACXL1; by 64%, 62%, and 61%, respectively, with ACXL2; and by 33%, 36%, and 32%, respectively, with ACXL3 (all P < 0.05). For the other regions(with radii between 4.5 and 7.5 mm), the SCXL and ACXL1 groups (but not the ACXL2 and ACXL3 groups) still showed significant changes in E_eff and K.

Conclusions: CXL had a significant effect on corneal biomechanics in both standard and accelerated procedures that may go beyond the irradiated area. The effect of CXL in stiffening the tissue and reducing permeability consistently decreased with reducing the irradiance duration.

The cornea is a significant part of the eye's refractive system, providing approximately 70% of the refractive power.¹ The corneal stroma accounts for approximately 85% of the entire corneal thickness.² Collagen fibrils interweave into a reticular structure in the stroma, and a liquid unevenly fills the gaps in the collagen fibril grid, forming a porous biological tissue with a fluid content of approximately 80%,³ affecting corneal biomechanics and helping maintain corneal morphology and transparency. In corneal ectatic diseases such as keratoconus (KC), in which progressive steepening of the cornea occurs, the normal parallel organization of collagen fibrils is disrupted, resulting in progressive myopia, irregular astigmatism, and significant effects on the vision and quality of life of patients.⁴ Corneal cross-linking (CXL) effectively increases the mechanical stiffness of collagen fibrils and their ability to resist collagenase lysis, thereby increasing tissue stiffness and halting or slowing the progression of corneal distortion.⁵ 

Over time, CXL has gradually evolved from the early classic cross-linking with long treatment time to accelerated cross-linking protocols with shorter treatment time.⁶ However, the regional changes of corneal biomechanical properties after CXL with different operative protocols are understudied, and the effective treatment range and the degree of local stiffening have not been fully evaluated. To address this need, biomechanical measurements such as inflation tests⁷ and uniaxial stretch experiments⁸ have been used to evaluate the cornea as a whole, and Brillouin microscopy⁹ and optical coherence elastography¹⁰ have been useful in quantifying the changes in elastic modulus at different stromal depths after CXL. Nevertheless, little attention has been given to the spatial characterization of corneal stiffness and to the area outside the irradiation range. In this study, nanoindentation was used to address these gaps. Additionally, due to the high fluid content and porous structure of the cornea, fluid-dependent viscoelasticity (also known as poroelastic viscoelasticity,¹¹ permeability,¹² or hydraulic conductivity¹²) plays a key role in influencing the biomechanical response of the tissue. This parameter was therefore included in our study while utilizing the fluorescence recovery after photobleaching (FRAP) technique^13,14 and permeameter with diffusion chamber.¹⁵ 

The nanoindentation used in this study is a method of measuring mechanical properties by pressing probes of a certain shape and size into materials.¹⁶ Although traditional nanoindentation techniques were developed for stiff materials,¹⁷ nanoindentation testing of hydrogels,¹⁸ soft tissues,¹⁹ and cells²⁰ has subsequently attracted substantial attention because of the high spatial resolution nanoindentation allows with local testing of mechanical properties of soft matter that is not possible using macroscale techniques. Over the past decade, despite the challenges encountered with substantial inelastic and associated creep deformation of soft biomaterials, nanoindentation technology, especially utilizing spherical tipped indenters, along with the classic Hertz model has gradually begun to be used for reliable measurement of soft matter.²¹ However, the measurement approach still needs to meet certain objective preconditions, such as contact strain (contact radius divided by indenter tip radius) < 0.4 for the samples tested.²² In corneal measurement, nanoindentation technology has been applied in elasticity measurement,²³ creep testing,²³ porous media analysis,¹² and multilayer structure measurement,³ and it has become a widely accepted technique for material characterization. This study sought to assess corneal regional biomechanical properties after ultraviolet-A (UVA) cross-linking through the use of nanoindentation technology, leading to measurements of the effective elastic modulus (E_eff) and hydraulic conductivity (K) across the corneal surface. 

Materials and Methods

Experimental Animals

Fifty New Zealand White rabbits (3 months old; weight, 2.0–3.0 kg) were included in this study. The left eyes were randomly divided into five equal groups: standard CXL (SCXL; 3 mW/cm² for 30 minutes), accelerated CXL 1 (ACXL1; 9 mW/cm² for 10 minutes), ACXL2 (18 mW/cm² for 5 minutes), ACXL3 (30 mW/cm² for 3 minutes), and a control (CO) group, which was unirradiated but underwent epithelium removal and riboflavin instillation) (Table 1). The contralateral eyes did not undergo any surgical procedure. All animals were obtained from the Animal Breeding Unit of Wenzhou Medical University and observed for 2 weeks before the experimental study commenced. All animals were treated in agreement with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, subject to the approval of the Animal Care and Ethics Committee of the Eye Hospital of Wenzhou Medical University. 

Table 1.

 

Corneal Cross-Linking Settings Adopted in the Different Specimen Groups

 

Corneal Thickness Measurement

Metal rings with radii of 3 mm and 6 mm were stained with dye and gently placed on the cornea to mark the thickness measurement positions. The central, paracentral, and peripheral corneal thicknesses (at 0, 3, and 6 mm away from pupil center, respectively) were measured in vivo with an ultrasound pachymeter (PachPen; Accutome, Malvern, PA, USA) along two meridional lines (nasal horizontal and temporal horizontal). Each measurement was taken three times preoperatively (pre) and 1-month after CXL (pos1m), and the average values were recorded. 

Preparation and CXL Procedure

Before CXL, the rabbits were premedicated with a subcutaneous injection (SU-MIANXIN, 0.2 mL/kg; Veterinary Institute at University of Munitions, Changchun, China). General anesthesia was administered with an intramuscular injection (pentobarbital sodium, 30 mg/kg; Merck KGaA, Darmstadt, Germany). Additional topical anesthetic (Alcaine Eye Drops; Alcon Laboratories, Ft. Worth, TX, USA) was instilled into the left eyes, and a wire eyelid speculum was positioned in the same eyes. Prior to UVA (370-nm) irradiation, the central 9 mm of the corneal epithelium was carefully removed using a hockey knife, and the corneas were saturated with 0.1% riboflavin drops (Peschke Meditrade GmbH, Huenenberg, Switzerland) at 3-minute intervals over a total period of 30 minutes. Following this step, the CXL procedure was conducted while performing the protocols explained in Table 1 using the CL-01 CXL system (SiHaiTong Co., Suzhou, China). The protocols involved subjecting the 9-mm-diameter central corneal area to irradiation (D_i) (Figs. 1a, 1b) with different intensities and exposure times but with a constant total energy dose of 5.4 J/cm². The cornea was kept moist with a balanced salt solution²⁴ (one drop every 2 minutes) during the entire irradiation procedure. Immediately after CXL and three times a day for 1 week, the left eyes received tobramycin ophthalmic ointment (TOBREX; Alcon Laboratories) and deproteinized calf blood extract eye gel (Xingqi; Shenyang Xingqi Pharmaceutical Co., Ltd., Shenyang, China) to ensure complete re-epithelialization. 

Figure 1.

 

UVA cross-linking and corneal specimen preparation. (a) Irradiation diameter was set at 9.0 mm for CXL. (b) Cross-sectional view of the cornea. (c) Fixation of the corneal strip. (d) Division of corneal measurement area.

Specimen Preparation

One month after CXL, a corneal ring with a diameter of 9 mm was placed in the central pupil to mark the corneal irradiation area in vivo, and the corneal diameter (D_c, nasal–temporal) was measured at the same time. The rabbits were then over-anesthetized to death. Both eyes were immediately removed and kept in PBS until they were used. The corneal epithelium was completely scraped gently using a corneal epithelial scraper, and the remaining corneal tissue, along with a 1- to 2-mm scleral band, was separated with ophthalmic scissors. Considering that the cornea has a surface curvature and is difficult to be fixed as a whole, each cornea was cut with a blade into a 5-mm-wide strip containing the nasal–temporal corneal rim. The corneal strip was then glued to the base of a Petri dish, and the endothelial surface was spread and gently affixed to avoid ruffling of the corneal anterior surface (Fig. 1c). A Vernier caliper was used to measure the nasal–temporal limbus-to-limbus length of the strip (L_c) and the length of the irradiation area (L_i) using the marked positions of the metal ring (Fig. 1c). A storage medium of PBS was then injected into the Petri dish to maintain hydration during the nanoindentation measurements. 

Nanoindentation Measurements

Nanoindentation measurements were performed using a Piuma instrument (Optics11 Life, Amsterdam, The Netherlands). A glass indenter with a tip radius of 261 to 270 µm was used, and the maximum indentation depth was set at 30 µm. The indents were performed along the specimen center line (Fig. 1d) with a constant spacing of 500 µm. The indents were separated into five groups based on their locations; these indents were within distances of 0 to 1.5 mm, 1.5 to 3.0 mm, 3.0 to 4.5 mm, 4.5 to 6.0 mm, and 6.0 to 7.5 mm from the specimen center, respectively (Fig. 1d). All measurements for each eye were completed within 1 hour. 

With the cornea considered a porous elastic (poroelastic) material, two properties were determined from the nanoindentation tests: the elastic modulus and the permeability.¹² In the analysis, the balance of forces between the indenter and indented structure is represented by  

\begin{eqnarray} F = {F_e} + {E_p}\quad \end{eqnarray}

(1)

 

where F is the applied force, F_e is the elastic resistance given by the Hertz equation,²⁵ and F_p is the resistance provided by permeability and is derived from Darcy's law.¹² Therefore, Equation (1) takes the form  

\begin{eqnarray} F = (4/3){E_{eff}}{R^{1/2}}{h^{3/2}} + \pi (1/K){(2Rh)^{3/2}}\frac{{{\rm{d}}h}}{{{\rm{d}}t}}\quad \end{eqnarray}

(2)

 

where E_eff is the effective elastic modulus, h is the indentation depth, R is the radius of the indenter tip, K is the hydraulic conductivity, and dh/dt is the indenter displacement rate. The load-displacement responses at different indentation rates are shown in Figure 2. Based on Equation (2), an estimate of K may be had by loading at two different indentation rates, dh₁/dt and dh₂/dt (Fig. 2):  

\begin{eqnarray} K = \frac{{\pi {{(2Rh)}^{3/2}}({\rm{d}}{h_1}/{\rm{d}}t - {\rm{d}}{h_2}/{\rm{d}}t)}}{{{F_1} - {F_2}}}\quad \end{eqnarray}

(3)

knowing K, one can use Equation (2) to determine E_eff. The average values of K and E_eff during loading within an indentation depth between 8 and 22 µm were obtained from the analysis. 

Figure 2.

 

Schematic diagram of the load–displacement response at two loading rates based on the analysis developed for an elastic permeable material.

Statistical Analysis

Quantitative data are presented as the mean ± SD, and all statistical analyses were performed using PASW Statistics 25.0 (IBM, Chicago, IL, USA). Comparisons of the dimension changes in the corneas (D_c and L_c) and irradiated areas (D_i and L_i) were performed using paired t-tests. Comparisons of results obtained for the different specimen groups were performed using analysis of variance (ANOVA) and the Games–Howell post hoc test. Spearman correlations examined relationships among different parameters. P < 0.05 was indicative of statistical significance. 

Results

Table 2 shows the corneal diameters (D_c), lengths of corneal strips (L_c), and lengths of strips within the irradiation area (L_i) of rabbit eyes in each intervention group; the results show no statistically significant differences among the five groups in all three parameters (all P > 0.05). During the preparation of corneal strips in vitro, the diameter of the cornea increased by 2.30 ± 0.54 mm from D_c to L_c (P < 0.001), and the diameter of the irradiated area increased by 0.24 ± 0.12 mm from D_i (9 mm) to L_i (P = 0.001). 

Table 2.

 

Corneal Dimensions and Thickness Measurements in All Intervention Groups

 

Corneal Thickness at Different Positions

Corneal thickness measurements taken before the CXL treatment showed no significant differences among all of the intervention groups (all P > 0.05) (Table 2). The central corneal thickness after CXL decreased significantly compared to the preoperative values for SCXL and ACXL1 groups (P < 0.001 and P = 0.010, respectively). The significant decrease was found only in the paracentral regions of the SCXL groups (P < 0.001 and P = 0.002, respectively). In contrast, the peripheral thickness in all groups did not reduce significantly (all P > 0.05). 

Elastic Modulus at Different Indentation Rates

Each measurement point was measured twice at two loading speeds of 60 µm/min and 600 µm/min; in both cases, the time interval between successive measurements was 10 minutes to ensure tissue recovery. The Hertz equation was used to fit and obtain elastic modulus values at different indentation rates, denoted as E₆₀ and E₆₀₀. The mean values of E₆₀ (20.11 ± 15.78 kPa), E₆₀₀ (26.80 ± 23.26 kPa), and E_eff (19.31 ± 15.09 kPa) for all five groups are shown in Figure 3, where E₆₀₀ differs from both E_eff (P < 0.001) and E₆₀ (P < 0.001). No significant difference was observed between E_eff and E₆₀ (P = 0.395). 

Figure 3.

 

Elastic modulus (kPa) at different indentation rates. The box indicates the lower and upper quartiles, and the central line marks the median; whiskers are 0.5 times the interquartile range. **Difference against P < 0.001.

Regional Changes in the Effective Elastic Modulus

The E_eff values of the different intervention groups are shown in Table 3 and Figure 4. Compared with the CO group, the E_eff of all regions was significantly higher in the SCXL, ACXL1, and ACXL2 groups (all P < 0.05) but not in some regions of the ACXL3 groups (P = 0.092 for region 4.5–6.0 mm; P = 0.326 for region 6.0–7.5 mm). No significant difference was detected between the 0- to 1.5-mm region and the 1.5- to 3.0-mm region for all groups (P = 0.409 for SCXL group; P = 0.695 for ACXL1 group; P = 0.946 for ACXL2 group; P = 0.150 for ACXL3 group; P = 0.926 for CO group). No significant difference was detected between the 3.0- to 4.5-mm region and the 4.5- to 6.0-mm region for the ACXL2 group (P = 0.085), ACXL3 group (P = 0.164), and CO group (P = 0.792). For the ACXL3 group and CO group, no significant difference was detected in the 1.5- to 3.0-mm region (P = 0.175) and the 3.0- to 4.5-mm region (P = 0.322). 

Table 3.

 

Regional Effective Elastic Modulus (E_eff) for Different Groups

 

Figure 4.

 

Regional effective elastic modulus (E_eff) for different groups. The box indicates the lower and upper quartiles, and the central line marks the median; whiskers are 0.5 times the interquartile range. *Significant difference from the CO group (P < 0.05). #Difference had no statistical significance (P ≥ 0.05).

Regional Changes of Hydraulic Conductivity

The hydraulic conductivity values of different intervention groups are shown in Table 4 and Figure 5. Compared with the CO group, the hydraulic conductivity for all regions was significantly lower in the SCXL, ACXL1, and ACXL2 groups but not the ACXL3 group (P = 0.332 for the 4.5–6.0-mm region; P = 0.423 for the 6.0–7.5-mm region). For different regions of each group, no significant difference was detected between the 0- to 1.5-mm region and the 1.5- to 3.0-mm region for all groups (P = 0.979 for SCXL group; P = 0.830 for ACXL1 group; P = 0.946 for ACXL2 group; P = 0.359 for ACXL3 group; P = 0.997 for CO group). No significant difference was detected between the 1.5- to 3.0-mm region and the 3.0- to 4.5-mm region for SCXL (P = 0.170), ACXL1 (P = 0.118), and ACXL2 (P = 0.220) groups. Furthermore, no significant difference was detected between the 3.0- to 4.5-mm region and 4.5- to 6.0-mm region for the ACXL1 (P = 0.161), ACXL2 (P = 0.083), ACXL3 (P = 0.773), and CO (P = 1.000) groups. 

Table 4.

 

Regional Hydraulic Conductivity (K) for Different Groups

 

Figure 5.

 

Regional hydraulic conductivity for different groups. The box indicates the lower and upper quartiles, and the central line marks the median; whiskers are 0.5 times the interquartile range. *Significant difference from the CO group (P < 0.05). #Difference had no statistical significance (P ≥ 0.05).

Spearman’s correlation was used to determine the relationship between the effective elastic modulus (E_eff) and the hydraulic conductivity (K). For all measurement points, the results showed a negative correlation (r_s = –0.745, P < 0.001). A simple power law relationship was found using a nonlinear model (red line) as shown in Figure 6. Further, different regions were analyzed for each group, and the results are shown in Table 5, which shows negative correlations between E_eff and K. 

Figure 6.

 

The relationship between the effective elastic modulus (E_eff) and hydraulic conductivity (K) for all of the nanoindentation tests undertaken.

Table 5.

 

Correlation of Effective Elastic Modulus (E_eff) With Hydraulic Conductivity (K) in Different Regions for All of the Nano-Indentation Tests Undertaken in Intervention Groups

 

Additionally, there were no statistical differences in all parameters (effective elastic modulus, hydraulic conductivity, and corneal thickness) in each region among the six non–cross-linked groups, including one control eye group and five contralateral eye groups (all P > 0.05) (Table 6). 

Table 6.

 

Corneal Regional Biomechanical Parameters and Thickness Measurements in CO Group and All of the Contralateral Eye Groups

 

Discussion

CXL is a minimally invasive surgical technique that stabilizes the progression of corneal ectasia and postpones the need for lamellar or penetrating keratoplasty.²⁶ Comprehending the change of corneal biomechanical behavior is critical to assessing the effectiveness of CXL. Corneal solid components (e.g., collagen fibrils, proteoglycans) primarily determine the elasticity of the cornea as measured by the elastic modulus.¹² On the other hand, characterization of the corneal mobile liquid components (e.g., interstitial fluid) can be expressed by the permeability or hydraulic conductivity of the tissue,²⁷ which also relates to corneal nutrition transport and transparency.^12,28 The present study found a highly radial dependence of tissue elasticity and permeability after CXL treatment. In terms of corneal elasticity, the strengthening was highest in the central cornea and decreased toward the periphery (Fig. 4, Table 3). In terms of corneal permeability, an opposite trend was observed (Fig. 5, Table 4). This uneven radial effect of CXL may be related to the more tightly arranged collagen fibrils in the central region.^29,31 

CXL enhances stiffness to the tissue by producing additional covalent bonds within and between collagen fibrils. In this study, the increase in stiffness gradually reduced with reducing the UV-A power duration. This trend is compatible with earlier studies in which ACXL led to reduced effectiveness.^7,30 Oxygen is necessary to drive CXL process within the corneal stroma, which is impaired in hypoxia.³¹ The smaller effect of accelerated CXL has been attributed to the higher consumption and shortage of oxygen in the stroma.^32,33 Kamaev et al.³⁴ noted that oxygen consumption occurs within seconds during UVA irradiation and that, following cessation of irradiation, oxygen levels can take several minutes to be restored. 

Some previous studies have shown changes of corneal elastic modulus after CXL. Nohava et al.²³ used nanoindentation to analyze the regional elastic modulus of human corneal stroma. The results showed that the elastic modulus of the central (radius 0–1.0 mm), paracentral (radius 1.0–2.5 mm), and peripheral (radius 2.5–4.0 mm) regions increased by 108%, 79%, and 63%, respectively, after SCXL; the corresponding values were 89.9 ± 42.4 kPa, 65.0 ± 17.9 kPa, and 37.7 ± 20.4 kPa, respectively. In this study, the effective elastic modulus (E_eff) values for the 0- to 1.5-mm region, 1.5- to 3.0-mm region, and 3.0- to 4.5-mm region in the SCXL group were 47.65 ± 20.58 kPa, 41.12 ± 19.87 kPa, and 31.17 ± 16.23 kPa, respectively, which increased by 309%, 276%, and 226%, respectively, compared with the CO group (Table 3). As can be seen from Figure 3, the higher the indentation rate, the greater the penetration resistance and thus the greater the elastic modulus. In this study, the effect of indentation rate penetration resistance was excluded, and instead the effective elastic modulus (E_eff), independent of penetration rate, of the rabbit cornea was obtained, and it was not statistically different from the elastic modulus E₆₀ (Fig. 3) for E_eff at a speed of 60 µm/min.²³ 

The main reasons for the difference from the previous results are as follows. First, the rabbit cornea does not contain the outer Bowman's layer, which is stiffer than the stroma,³⁵ and its collagen fibrils are much smaller than those of the human cornea.³⁶ Second, the elastic modulus of the normal rabbit cornea is smaller than that of the human cornea.³⁷ As a result, although the elastic modulus of rabbit cornea is smaller than that of human cornea, the proportion of increase after cross-linking is higher than in the human cornea. Third, although both studies used the nanoindentation technique, the actual measurement parameters were set differently, such as the size of the spherical indenter. With increasing indenter radius, the contact diameter increases for the same depth of penetration. Also, the volume of the region beneath the indenter that is compressed scales directly with the indenter contact radius; that is, the bigger the indenter radius and contact diameter, the greater the depth of the cornea that is experiencing compressive deformation. As the elastic modulus of the cornea decreases with depth, especially after cross-linking, this gradient structure of the cornea becomes more obvious.³⁸ Therefore, measurement of the human cornea with a bigger indenter (500 µm) also incorporates the elastic modulus of the deeper stroma, resulting in a much smaller proportion of the increase of the elastic modulus after cross-linking than that of the rabbit. 

The reported progression rate of KC after CXL is up to 23%,³⁹ and the reason may be that CXL is commonly focused on the central cornea for irradiation. KC can theoretically occur anywhere in the cornea, and it has been reported that it is mostly found in localized lesions in the inferior temporal part of the cornea.⁴⁰ With cone apexes 1 to 2 mm off the corneal center and with cone borders up to 3 mm from the center,⁴¹ the eccentricity values are even higher in pellucid marginal degeneration (PMD).⁴² Because the position of greatest need for strengthening is not located in the center of the cornea, where CXL is most effective, this may result in an overestimation of the expected outcome of the procedure. In such cases, peripheral compensatory irradiation or customized irradiation centered on the cone apex could achieve better results. 

The strengthening range of corneal stiffness was larger than the irradiation range set in this study; that is, the strengthening effect was still evident in the non–cross-linked region. As can be seen from Figure 4, E_eff increased by 133%, 124%, and 62% after SCXL, ACXL1, and ACXL2, respectively, in the 4.5- to 6.0-mm region (all P < 0.001). However, this region contains a small fraction of the irradiated area (0.24 ± 0.12 mm) (Table 2), so the increase in E_eff cannot be used to accurately determine whether the non–cross-linked region has strengthened. It is worth noting that the E_eff still increased by 40%, 39%, and 32% after SCXL, ACXL1, and ACXL2, respectively, in the 6.0- to 7.5-mm region (P = 0.002, P = 0.002, and P = 0.016, respectively). Webb et al.⁴³ performed blue-light CXL (447 nm) on porcine eyes and shaded the outside of the irradiated region to quantify the changes of corneal stiffness using Brillouin microscopy; they confirmed that the strengthening effect occurs beyond the set range. The factors that influence the biomechanical properties of a non–cross-linked region are various and may include diffusion of oxygen radicals⁴⁴ and scattering of UVA in the stroma,⁴⁵ as well as reflection of light through the cornea. Also, the device and conjunctival capsule (Fig. 1a) can lead to extensive strengthening of the non–cross-linked region. The strengthening of the non–cross-linked region, especially the limbus, may have unintended adverse consequences on the cornea. Limbal epithelial stem cells are considered to be the only cells responsible for retaining the corneal epithelium in a steady state, and their integrity is a key determinant for maintaining a clear, avascular cornea.⁴⁶ Previous studies have reported cases of delayed corneal epithelial healing after CXL,^47,48 and the reason has not yet been explored, although it may be due to the cytotoxic effects of UVA on epithelial stem cells at the corneal limbus⁴⁹ or the increase of corneal stromal stiffness that affects the maintenance and differentiation of epithelial stem cells.⁵⁰ Jeyalatha et al.⁵¹ suggested that the use of a polymethylmethacrylate covering could prevent UVA damage in the epithelial stem cells in the limbus. Otherwise, recent studies have developed customized CXL treatments that change the irradiation energy and irradiation site according to the specific morphology of the cornea^52,53 and show that the irradiation border of eccentric CXL is closer to the limbus than conventional CXL, so more attention should be paid to the influence of CXL on the limbus. 

The avascular and solute diffusion of the cornea becomes an important mechanism for cellular transport of nutrients and wastes,⁵⁴ which depends on the transport of substances by corneal endothelial cells and the structure of the corneal stromal collagen network.¹⁴ Combined with the results of this study and our recent work,⁷ CXL may change the spatial structure of corneal stroma collagen fibrils and reduce corneal permeability, thereby increasing the resistance of corneal stroma to the diffusion of certain solutes. Previous studies also showed a trend of decreased transport after CXL, with direct measurements of fluorescein diffusion through the corneal stroma showing that solute permeability significantly decreased following cross-linking in rabbit and porcine eyes.^15,55 Furthermore, the regional variation of K (Fig. 5, Table 4) indicates that the effect of CXL on solute transport depends on the lateral position from the corneal center point. The reduction of corneal permeability was higher in the central region than in the peripheral region; that is, like elastic modulus, the greatest effect of CXL on corneal permeability occurred in the central region. It is worth noting that the non–cross-linked region was also affected to some extent, resulting in a slight decrease in permeability. This confirmation has important clinical implications and provides a more detailed understanding of the influence of CXL on cornea transport behavior. For example, drug delivery to certain regions of the cornea would be incorrectly overestimated if a uniform decrease in permeability was present. Again, the present study preliminarily presented a power function relationship between E_eff and K (Fig. 6, Table 5)—that is, the stiffer the region, the less permeable it becomes. This aspect suggests that surgeons should consider the effect of CXL on drug transport when selecting surgical options for patients who require long-term drug use, especially for high-energy, low-irradiance CXL. Apart from that, with further investigations of the quantitative relationship between these factors, K has the potential to become a novel and important in vivo biomechanical parameter¹³ with great diagnostic value for corneal biomechanical weakening diseases in the future. 

The differences in biomechanical parameters among treatment groups were primarily caused by the variations adopted in the CXL protocols, which were able to be demonstrated by the comparison between the non–cross-linked groups (Table 6), but there are still some limitations in the current study. Because of the difficulty in obtaining a sufficient number of human donor corneas, rabbit corneas were used instead, as they have been shown to be suitable alternatives in previous experiments.^7,30 Also, depth dependence and radial variations of corneal biomechanical properties after CXL could not be further explored and will be investigated in future studies. Finally, the regional biomechanical changes after CXL in this study were not confirmed by corresponding histological analysis of collagen fibrils. The results of future studies should address these limitations. 

In conclusion, this study confirms the uneven effect of CXL on corneal biomechanical properties. This effect is most evident in the corneal center, and the effectiveness decreases with distance away from the center. In the case of severely eccentric KC, peripheral compensatory irradiation or customized CXL centered on the cone apex may be necessary to ensure maximum impact on corneal biomechanics and ensure stability of the cornea. 

Acknowledgments

Supported by grants from the Zhejiang Provincial Natural Science Foundation of China (LQ20A020008, LY22H180005), the National Natural Science Foundation of China (82271049, 31771020, 82001924), and the Medical Science and Technology Program of Zhejiang Province (WKJ-ZJ-1829). 

Disclosure: X. Zheng, None; Y. Xin, None; C. Wang, None; Y. Fan, None; P. Yang, None; L. Li, None; D. Yin, None; E. Zhang, None; Y. Hong, None; H. Bao, None; J. Wang, None; F. Bao, None; W. Zhang, None; S. Chen, None; A. Elsheikh, None; M. Swain, None