Enhanced Riboflavin Stromal Delivery Using Microchannel-Assisted Iontophoresis for Corneal Crosslinking

Abstract

Purpose: The purpose of this study was to determine if iontophoresis of riboflavin (Rf) combined with femtosecond generated epithelial microchannels (MCs) could shorten the time required for Rf stromal delivery and subsequent crosslinking (CXL).

Methods: Eighty-four ex vivo rabbit eyes were divided into three groups: group 1 = iontophoresis alone; group 2 = iontophoresis with MC; and group 3 = MC alone. Iontophoresis was performed using a small electric current with varying current and time. MCs were generated using a 1030 nm femtosecond (FS) laser to machine the corneal epithelial surface in a 6 mm diameter region. Ex vivo eyes were treated with topical iso-osmolar Rf solution, and stromal Rf concentration was determined spectrophotometrically by eluting Rf from trephined stromal buttons. In vivo, six rabbits underwent iontophoresis or iontophoresis + MC followed by ultraviolet-A crosslinking (UVA CXL) and imaged for collagen autofluorescence (CAF) signal to determine CXL efficacy.

Results: Ex vivo iontophoresis achieved a threefold increase in stromal Rf concentration when combined with MC for 5 to 10 minutes (equivalent to MC alone for 30 minutes compared to iontophoresis alone). In vivo, iontophoresis + MC resulted in a significantly higher (4-fold) CAF intensity than iontophoresis alone.

Conclusions: MCs and low current iontophoresis produce a significantly higher stromal Rf concentration than iontophoresis alone and equivalent to MC alone at greatly reduced Rf treatment time. Additionally, the combined treatment results in a four-fold increase in CAF intensity over iontophoresis alone.

Translational Relevance: A combined treatment of iontophoresis and MC significantly enhances stromal Rf concentration resulting in increased CXL while significantly reducing procedure time.

Introduction

Keratoconus (KC) and ectasia following corneal refractive procedures, such as keratorefractive surgery (LASIK or surface ablation) or Refractive Lenticule Extraction (RLE), are characterized by progressive thinning, mechanical weakening of the stroma, and corneal bulging. These conditions can result in visual impairment due to irregular astigmatism and reduced spectacle-corrected distance visual acuity.^1,2 They may also heighten sensitivity to light and cause glare. The etiopathogenesis of KC is complex, and is no longer considered a noninflammatory disease, as elevated levels of proinflammatory cytokines have been found in the tears and epithelium of patients with KC.^3,4 Both genetic and environmental factors seem to contribute to the development of KC, with eye rubbing and a recently identified nocturnal ocular compression playing a role in the pathogenesis.^5,6 

Although mild KC can be treated with glasses and contact lenses, more advanced stages may require surgery.¹ For instance, synthetic polymethylmethacrylate (PMMA) intrastromal corneal ring segments (ICRS) are often used to reshape the cornea, reduce irregular astigmatism, and delay the need for more invasive procedures, like corneal transplantation. However, their ability to halt KC progression remains debated.^7,8 On the other hand, corneal collagen crosslinking (CXL), which involves soaking the cornea in the photosensitizer, riboflavin (Rf), and exposing it to ultraviolet-A (UVA) irradiation, has been shown to stop or even reverse KC progression.⁹ In very advanced cases of KC, penetrating keratoplasty or deep anterior lamellar keratoplasty (DALK) may be required.¹ However, less invasive anterior keratoplasty techniques, such as Bowman’s layer transplantation, corneal allogenic intrastromal ring segments (CAIRS), or intrastromal lenticules, have shown promising results.^10–13 With regard to the most common procedure currently used to slow or halt KC progression, corneal collagen CXL mediated by UVA photoactivation of the Rf and the generation of oxygen free radicals has been shown to induce covalent crosslinks within and between the collagen fibrils and proteoglycan core proteins, resulting in mechanical stiffening and flattening of the cornea.^14,15 In addition to the stiffening effect (up to 300%), CXL has been shown to enhance blue collagen autofluorescence (CAF).^16,17 However, as photoactivation of Rf and the generation of oxygen free radicals is cytotoxic and harmful to living tissue, specific parameters for CXL are used to minimize exposure damage to deeper ocular structures, such as the corneal endothelium, by limiting the power, treatment time, and concentration of Rf.^17–20 

A major concern about UVA CXL is the method of stromal Rf application. Because the corneal epithelium acts as a barrier to prevent the diffusion of solutes into the stroma, UVA CXL generally requires complete epithelial debridement to ensure sufficient stromal Rf penetration (the Dresden protocol) and effective CXL. This procedure is painful for the patient, requires several days to recover, and increases the risk of microbial keratitis.^{16,19,21–24} To reduce these side effects, several alternative approaches have been used, including full-thickness partial mechanical de-epithelization,^3,25,26 multifocal epithelial disruption using a specially designed metallic instrument²⁷ or a surgical sponge, and partial thickness epithelial removal using an excimer laser.²⁸ In addition, to avoid any ocular surface disruption, transepithelial UVA CXL (TE-UVA CXL) procedures have been proposed that achieve stromal Rf penetration without removing the corneal epithelium. Current methods include the addition of chemical excipients, such as benzalkonium chloride (BAK) or ethylenediaminetetraacetic acid (EDTA), to loosen epithelial tight junctions. Additions of glutathione, vitamins A, C, and E, polyethylene glycol, gentamicin, tetracaine, and corneal iontophoresis during Rf application have also been proposed.^29–36 However, CXL as a result of TE-UVA CXL has been shown to be far less effective than traditional UVA CXL.^29,37–41 This is likely due in part to the reduced Rf penetration into the corneal stroma. The long-term efficacy of this technique compared to the standard epi-off Dresden protocol also has not been determined. 

Iontophoresis, a novel method of facilitating TE-UVA CXL, is a noninvasive method where a small electric current is applied to a surface to increase penetration of an ionized substance into the tissue. This method's benefit lays in its ability to provide high drug tissue concentration (50–100 × higher concentration compared to passive permeation),^42,43 while minimizing systemic drug exposure. Iontophoresis is used in various fields of medicine including ophthalmology and has been used for transcorneal dexamethasone application post-surgery and methylprednisolone treatment post-penetrating keratoplasty (PKPs) to minimize graft rejection.^44–47 Recently, this procedure has been advocated for use in TE-CXL by enhancing intrastromal Rf diffusion while retaining an intact epithelium.^29,38–41 

The first clinical application of ocular iontophoresis was performed by Wirtz in 1908 to treat serpiginous ulcer of the cornea with 0.5% zinc sulfate using a current of 2 mA for 1 minute.⁴⁸ Since then, dyes, antiviral and antibacterial agents, steroids, and genes have been delivered to the eye using this method. Iontophoresis has been lauded for its simple approach requiring only a current generator, an eye cup, and the drug solution as well as providing a noninvasive delivery method that allows the drug to overcome natural permeability barriers. Clinical trials have demonstrated that iontophoresis-assisted CXL is a valid method for treating KC progression but has a far lower efficacy compared to standard CXL.^48–50 

In standard iontophoresis, the Rf is applied with a negatively charged electrode and the circuit completed with an electrode inserted into the vitreous of the eye ex vivo, or the nape or forehead in vivo. The electrode then conducts the current through the corneal tissue. Rf is a great candidate for corneal iontophoresis due to its water solubility, negative charge at physiological pH, and low molecular weight allowing for enhanced penetration. Other variables that influence penetration are the substance's concentration, size, and the application time. Studies have shown that safe and optimal transepithelial iontophoresis of Rf can be achieved using a 1 mA current applied over a duration of 5 minutes that can provide sufficient intrastromal Rf concentration for CXL stromal collagen.^16,30,38 Although these values are clinically used for iontophoresis-assisted CXL and claim to reduce the Rf penetration time from 30 minutes to 5 to 10 minutes,^{18,38,49,51,52} the achieved Rf stromal penetration has been reported to be 2-fold less than the conventional Dresden protocol, as shown by Cassagne et al. and Mastropasqua et al. in human donor corneas.^39,53 We have previously shown that femtosecond (FS) laser based micro-machining of the corneal epithelium to form epithelial microchannels (MCs; 2 mm in diameter and 25 µm deep into the corneal epithelium spaced 50–100 µm apart) can greatly improve passive Rf stromal penetration while also avoiding cellular damage that commonly occurs from the addition of BAK/EDTA.³³ Even so, this procedure still requires 30 minutes of Rf application to achieve maximum stromal delivery.⁵⁴ The purpose of this study was to determine if iontophoresis when combined with MC for stromal Rf delivery could both shorten the treatment time while also improving Rf stromal delivery by iontophoresis, resulting in more effective CXL. 

Methods

Ex Vivo Tissue Preparation

Eighty-four ex vivo New Zealand albino rabbit eyes were shipped overnight (Pel-Freez, Rogers, AK, USA) and prepared for experimentation as previously published.^32,55,56 The eyes were first rinsed in phenol free, low glucose, Dulbecco's Modified Eagle Medium (DMEM; Sigma-Aldrich, St. Louis, MO, USA) and placed in 12-well tissue culture plates containing sufficient media to cover the sclera up to the limbus. The eyes were then placed in a humidified, 37°C, 5% CO2 incubator for 2 hours to restore endothelial pump function and allow the corneal thickness to return to normal. The corneas were then stained with Lissamine Green (10 mg/mL in PBS; Sigma-Aldrich, St. Louis, MO, USA) to verify an intact, damage free epithelium. The eyes with Lissamine green staining, indicating epithelial damage, were discarded. Eyes were then imbibed with an Rf solution with one of three methods: iontophoresis alone (group 1), iontophoresis with MC (group 2), or MC alone (group 3). Table 1 lists detailed group sizes, Rf solution makeup, and treatment information. 

Table 1.

 

Ex Vivo Stromal Riboflavin Delivery and Concentration

 

Microchannels

Epithelial MCs were generated using a 1030 nm, 1 kilohertz (kHz), amplified FS laser (One Five Origami, NKT Photonics, Denmark) with a 5 µJ pulse energy adjusted using a polarizing beam cube and half wave plate, as reported previously.³³ Using custom Labview software, the FS laser beam was focused to precisely cut to a 25 µm deep, 2 µm diameter channel, with 50 µm spacing over the central 6 mm diameter area of the cornea. This procedure results in approximately 10% of the corneal surface having MCs. As previously published, this approach produces sufficient stromal Rf penetration for corneal CXL without damage to the corneal epithelium.^32,55,56 

Iontophoresis

The basic setup for iontophoresis is shown in Figure 1 and involves the use of a constant current source (see Fig. 1A) from OPIA Technologies (Iontofor-CXL) and two electrodes. The anode (see Fig. 1B, return electrode) was a sterile 18-gauge needle inserted into the vitreous chamber (ex vivo) or the rabbit's neck (in vivo), whereas the cathode (see Fig. 1C, main electrode) was a stainless steel mesh enclosed in an 8 mm diameter circular cup with a surrounding 1 mm wide annular suction ring that attaches the device to the cornea secured by a vacuum syringe (see Fig. 1D) for the duration of the iontophoresis treatment. The cathode was placed 8 mm away from the cornea to allow air bubbles to escape from the side. The mesh reservoir in the cathode was filled with a topical Rf solution (1.0% riboflavin-5-phosphate in PBS; Sigma-Aldrich, St. Louis, MO, USA) with optimized osmolarity for penetration (300 milliosmolal [mOsm]/L), which also acts as contact between the cornea and electrode. When treatment begins, the current generator (see Fig. 1A) outputs a constant direct current (0.5 or 1 milliampere [mA]) for a set period of time (5 or 10 minutes). Voltage fluctuations during treatment were monitored using a digital multimeter to maintain a 3 to 5 volt (V) range. Rf was suctioned out with the extra syringe and new solution was applied for each eye. 

Figure 1.

 

Iontophoresis Rf delivery. Iontophoresis riboflavin delivery system modified for use in ex vivo eyes. The constant current source (A) is connected to the positive electrode (B), a 22-gauge needle is inserted into the vitreous chamber, and the negative electrode (C) containing the riboflavin solution is submerging the stainless-steel grid. The annular suction ring is secured by the vacuum syringe (D) to steady the device during the procedure.

Experimental Design

The 84 ex vivo rabbit eyes were divided into 3 treatment groups, as shown in Table 1. The first two groups (group 1 = iontophoresis alone, 31 eyes) and (group 2 = iontophoresis with MC, 48 eyes) received Rf iontophoresis using three different settings of: 1 mA for 10 minutes (see Fig. 1A; group 1: n = 17 and group 2: n = 17), 1 mA for 5 minutes (see Fig. 1B; group 1: n = 5 and group 2: n = 16), and 0.5 mA for 10 minutes (see Fig. 1C; group 1: n = 9 and group 2: n = 15). These variations were chosen to determine the treatment which maximized stromal Rf penetration while minimizing current and treatment time. The last group (group 3: n = 5) received MCs alone and was used as a control. 

Measuring Stromal Riboflavin Concentration

Measurements of stromal Rf concentration were performed as previously published.³³ Briefly, excess Rf was wiped with cotton gauze and the epithelium removed with a Tooke knife. Using a 6 mm trephine, a central corneal button was cut out and then placed in 1 mL of PBS solution overnight to allow the Rf to elute from the button. This process was done quickly (3 minutes from cutting to depositing in PBS) to avoid loss of Rf into the anterior chamber. The weight of the solutions were measured before and after addition of the button to account for any variance in sample volumes. 

After 24 hours, the Rf concentration of the eluate was measured with a spectrophotometer (Spectramax Gemini XPS microplate reader; Molecular Devices, San Jose, CA, USA) with 380 nm excitation and 540 nm emission. Serial dilutions of the Rf solution were made and used to create a standard curve for the calculation of Rf concentration (ug/mL) which was used to quantify each sample's measured intensity values. 

Live Rabbit Model

A total of 6 New Zealand albino rabbits (4–5 months old) were used in this study and euthanized 24 hours after treatment. All animals were treated according to the ARVO statement on the use of animals in vision research, and experiments were approved by the IACUC of the University of California - Irvine (AUP 23-016). Only the right eye of each of the rabbits was treated in our experiment. 

Surgery

Rabbits were split into 2 groups of 3 each and treated with a current of 1 mA for a duration of 5 minutes. These values were chosen as they have been used extensively and are the current clinical standard for iontophoresis-assisted CXL.^{39,40,42,45,47,49,52,53,57} The first group underwent iontophoresis alone, whereas the second group received iontophoresis with MCs. Prior to treatment, all animals were anesthetized with a subcutaneous injection of 30 to 50 mg/kg of ketamine hydrochloride (Zoetis, Parsippany-Troy Hills, NJ, USA) and 5 to 10 mg/kg of xylazine (Akorn, Lake Forest, IL, USA). Each rabbit was set in a rabbit restrainer with a speculum positioned on the eye. Each cornea had a baseline line anterior segment scan taken with an optical coherence tomography (OCT; DRI OCT Triton plus; Topcon, Japan). 

After Rf treatment, UVA CXL was performed using the previously described protocol.⁵⁶ Rabbits then received a drop of 0.5% tetracaine hydrochloride (Alcon, Ft. Worth, TX, USA) to prevent pain. For the post-treatment, each rabbit received 0.1 to 1 mg/kg of Revertidine (Modern Veterinary Therapeutics, Miami, FL, USA) to reverse the sedative and 0.1 mL of buprenorphine hydrochloride (0.3 mg/mL; Reckitt Benckiser Healthcare, UK) for analgesia. Additionally, each rabbit was given a drop of 0.3% gentamicin (Allergan, Irvine, CA, USA) in the treated eye 3 times in the next 24 hours after treatment to prevent infection. 

Euthanization

After 24 hours, another OCT image was taken of each eye and then the treated rabbit corneas were stained with Lissamine green to determine the extent of epithelial damage and the rabbit was euthanized with an intravenous injection Euthasol (1 mL; Virbac AH, Fort Worth, TX, USA) into the marginal ear vein. Immediately after death, the corneas were fixed in situ under 20 millimeters of mercury (mm Hg) pressure via perfusion of 2% PFA (Mallinckrodit Baker, Inc., Phillipsburg, NJ, USA) in PBS at 4°C to maintain the in vivo collagen structure. They were then removed and fixed overnight. 

CAF Imaging

To verify and quantify the degree of CXL using blue CAF, corneas from in vivo experiments were embedded in 10% low melting point agarose (Lonza, Rockland, ME, USA) and sectioned along the superior-inferior meridian into 200 micron thick sections using a vibratome (Campden Instruments, Loughborough, England, UK). The samples were then imaged for blue CAF using a Leica Stellaris 8 multiphoton confocal microscope with a Chameleon FS laser (Coherent, Santa Clara, CA, USA) as the excitation source. The samples were excited at 760 nm and the CAF signal was collected between 400 and 450 nm using a 20 × objective. Images of the central cornea were exported to an image analysis software program (Metamorph, Molecular Devices, Sunnyvale, CA, USA) and three 100 × 100 pixel (100 pixels = 25 µm) areas in the central anterior CXL region and posterior stroma (background region) were identified in 3 sections for each cornea. The first 100 × 100 region was placed in the center of the CAF region in the anterior stroma with the other two 100 × 100 regions placed adjacently. This process was repeated in the posterior of the stroma to determine the background. The average intensity within both regions were subtracted to identify the average CAF value in the anterior of the cornea (CAF_ANT). 

The depth of CXL was also measured by generating three regions through the center of the cornea extending from the anterior to posterior of the stroma. We then used a line scan to calculate the mean fluorescence intensity along each line, measured the background intensity, and plotted this as a function of corneal depth – showing the change in pixel intensity from the anterior to the posterior stroma. By integrating the area under this depth-intensity curve (CAF_AUC), we estimated the total CAF between both groups. Peak pixel intensity was determined by looking at the maximum intensity value in the AUC plot for each group. CXL depth (CAF_depth) was determined by noting the distance from the beginning of the intensity signal to the point where the intensity signal equaled the background intensity. This distance was divided by the thickness of the section from the anterior to the posterior stroma and was expressed as a percent thickness of the section. 

Statistics

Statistical analysis was performed in statistical software SigmaPlot (SPSS Inc., Chicago, IL, USA) using the Tukey-Kramer method for a multiple comparison, 1-way analysis of variance (ANOVA). Groups were considered to have a statistically significant difference with a P value less than 0.05. Error bars represent the standard deviation in all the following figures. 

Results

Effects of Iontophoresis on Stromal Rf Concentration in Ex Vivo Rabbit Eyes

Table 1 and Figure 2 show the measured stromal Rf concentrations achieved in ex vivo eyes using the various treatment protocols. Corneal button weights between samples and treatment days were consistent. Iontophoresis alone (group 1, n = 31) achieved Rf stromal concentrations of 4.73 ± 0.82 ug/mL (1 mA for 10 minutes), 4.18 ± 0.42 ug/mL (1 mA for 5 minutes), and 3.65 ± 0.23 ug/mL (0.5 mA for 10 minutes). In this group, 0.5 mA treated corneas for 10 minutes achieved significantly lower Rf concentration than 1 mA for either 5 or 10 minutes. The same iontophoresis exposures combined with MCs (group 2, n = 48) achieved a 3+ fold higher (P < 0.001) Rf concentration averaging 12.64 ± 1.82 ug/mL, 10.37 ± 2.28 ug/mL, and 12.8 ± 1.38 ug/mL for the same iontophoresis protocols stated above. Iontophoresis with MCs when performed at 1 mA for 5 minutes was significantly lower than both 1 mA and 0.5 mA for 10 minutes but remained significantly higher than iontophoresis alone. Furthermore, iontophoresis with MCs for 10 minutes at 0.5 or 1 mA achieved equivalent stromal concentration of Rf as MC alone (group 3, n = 5) after 30 minutes of Rf treatment (P > 0.05). 

Figure 2.

 

Effect of Iontophoresis with and without microchannels on stromal Rf. Penetration graph of Rf stromal concentration 24 hours after procedure. The treatment groups: group 1 = iontophoresis alone, group 2 = iontophoresis + MC, and the control group (MCs for 30 minutes) are compared and show average ± SD for each variable iteration.

In Vivo Efficacy of UVA CXL

In vivo OCT images taken at baseline and post-CXL (1 day) were used to measure changes in stromal thickness using Metamorph after iontophoresis alone and iontophoresis with MCs (Figs. 3A, B, respectively). Post-treatment, the stroma was significantly increased (P = 0.0001) after iontophoresis alone, swelling 73.22% (294.67 ± 3.29 µm to 510.44 ± 90.41 µm) and 69.86% following iontophoresis + MCs (294.56 ± 5.70 µm to 500.33 ± 72.63 µm) which were not significantly different (Fig. 3C). Lissamine Green staining (Fig. 4) revealed a significantly (P = 0.0001) higher area of epithelial damage in the iontophoresis + MC group (23.87 ± 5.49 mm²) compared to the iontophoresis group (1.03 ± 0.97 mm²). Quantification of epithelial wound dimensions was performed using the open-source software QuPath by measuring the area of the Lissamine Green stain before and after the procedure. 

Figure 3.

 

OCT imaging. Representative images taken 24 hours post-treatment show significant stromal swelling in both the iontophoresis alone (A) and the iontophoresis + MC (B) groups. There was no significant difference between groups at baseline and at 24 hours (C).

Figure 4.

 

Lissamine Green staining. Images taken 24 hours post-treatment of rabbit eyes stained with Lissamine Green to show the extent of corneal damage. Eyes treated with iontophoresis alone (A) showed an average of 4.08% of the epithelium being damaged, whereas those treated with iontophoresis + MC (B) showed a significantly higher loss of (94.98%, P = 0.0001).

Collagen Autofluorescence

A graphical and numerical comparison of the in vivo CAF_AUC, average anterior stromal CAF accounting for background intensity (CAF_ANT), and CXL depth (CAF_depth) are shown in Figure 5 and Table 2. In group 1 (iontophoresis alone), there was a peak CAF (maximum pixel intensity in the AUC graph) value of 30 arbitrary unit (AU), total CAF area of 6512 ± 169 AU, and an average anterior stromal CAF (CAF_ANT) in the crosslinked region of 23.38 ± 1.49 AU. Group 2 (iontophoresis + MC) had respective values of 165 AU, 23151 ± 289 AU, and 96.31 ± 4.47 AU. Iontophoresis with MC (group 2, with MC, n = 3) achieved a significantly higher (4-fold) than iontophoresis alone but both groups showed a similar CXL depth (CAF_depth = 51.19 ± 4.52% vs. 45.21 ± 1.36%, respectively). 

Figure 5.

 

Collagen autofluorescence. Collagen autofluorescence from representative samples of the two experimental groups (A) iontophoresis alone and (B) iontophoresis with MC, both using 1 mA for 5 minutes. Iontophoresis achieved an average CAF value of 23.38 ± 1.49 AU with a total crosslinked area of 6512 ± 160 AU, whereas iontophoresis + MC achieved significantly higher metrics (96.31 ± 4.47 AU and 23151 ± 289 AU, respectively).

Table 2.

 

Collagen Autofluorescence and Depth of Crosslinking

 

Discussion

Although there have been several transepithelial studies to try and replicate the success of the traditional epi-off method, none have been able to achieve the same degree of CXL.^23,36,58 Although the addition of chemical excipients, such as BAK and EDTA, facilitates Rf penetration into the stroma through loosening of the epithelial tight junctions, they also increase patient pain and delay postoperative recovery similar to standard epi-off CXL.^{15,30,38–40,52} Excimer laser use prior to CXL has also been studied. In a clinical study,²⁸ Bakke et al. compared postoperative pain severity and riboflavin penetration rates between two corneal CXL methods: excimer laser superficial epithelial removal and mechanical full-thickness epithelial removal, and concluded that excimer laser superficial epithelial removal led to greater postoperative pain and a longer riboflavin application time to achieve corneal saturation. Additionally, in a recent experimental study,⁵⁹ Brekelmans et al. evaluated the efficacy of corneal CXL and chromophore penetration (Rf and Rf with 20% Dextran both at 0.1% and WST11 and WST11 with 20% Dextran both at 2.5%) after excimer laser-assisted patterned de-epithelialization in porcine eyes and showed that all chromophores induced significant CXL with similar stiffening across formulations and de-epithelialization methods. Light transmittance was lower after full de-epithelialization, but stromal chromophore penetration was comparable between the methods. The researchers concluded that excimer laser-assisted patterned de-epithelialization enabled effective CXL, but reduced chromophore concentration may impact UVA attenuation safety. 

One of the major reasons for these poor results from transepithelial attempts is insufficient Rf penetration and lack of stromal saturation during UV irradiation. Our study establishes that iontophoresis when combined with MCs, can lead to a dramatic 4-fold increased Rf concentration within 10 minutes of application compared to iontophoresis alone and equivalent to MC alone for 30 minutes. Intrastromal Rf concentration is a major factor determining CXL efficacy. To date, most iontophoresis studies have used a 0.1% hypo-osmolar Rf solution usually paired with excipients, such as BAK/EDTA/Trometamol, to enhance penetration into the stroma despite an intact epithelium. Unfortunately, this method's low stromal Rf saturation does not result in sufficient crosslink formation for the treatment to be effective, nor does it address the issue of postoperative patient pain.^14,31 With our combined iontophoresis + MC delivery system – although results are less than that achieved by complete epithelial removal – we were able to achieve a higher level of stromal Rf concentration through enhanced Rf diffusion without de-epithelialization and addition of excipients. Additionally, UVA photoactivation of Rf following iontophoresis + MC achieved an in vivo CAF level that was nearly 4 times that of iontophoresis alone while shortening the Rf application time from 30 to 5 to 10 minutes. Despite not removing the epithelium and avoiding the addition of BAK/EDTA, the iontophoresis + MC treatment resulted in the complete death of the epithelium showing that this method does not mitigate the risk of patient pain. 

There are a few notable points to discuss from our results. First, that the MC procedure by itself does not induce a complete death of epithelium – it damages a very small portion of the epithelium and the damage is hardly visible after only 3 hours of culture. The death of the epithelium is not from the manner of Rf penetration but due to the UVA irradiation post MC generation. This can be seen by the fact that the corneal epithelia remained intact after UVA irradiation post iontophoresis alone, but not when MCs were added. Not only do these results match what we have seen in our previous work,^33,60 but other groups have also seen similar results in in vivo rabbit and ex vivo porcine corneas as well as in human clinical trials. Slit lamp examinations, immunohistochemistry, topography readings, OCT imaging, and patient testimonies indicate that iontophoresis CXL with Rf solutions ranging from 0.1% (single dose) to 0.25% (double dose) without the addition of chemical excipients causes little or no change in pachymetry, endothelial cell damage, or other aberrations – suggesting maintenance of an intact epithelium after the procedure,^39,61–64 patients’ pain reports were also few and superficial. In contrast, patients and eyes subjected to iontophoresis CXL with the addition of excipients or laser ablation showed comparable epithelial and stromal damage to Dresden protocol CXL with epithelial debridement but displayed greater improvements to Kmax, best corrected visual acuity (BCVA), flattest and steepest meridian keratometry, corneal astigmatism, aberrometry, and central corneal thickness after 6, 12, and 24 months – although not at the level of standard epi-off UVA CXL.^41,65,66 We infer that iontophoresis alone without the aid of excipients or laser ablation causes little to no epithelial damage but also results in a markedly lower CXL effect. This is reflected clinically by a higher percentage of patients treated with iontophoresis CXL requiring a second CXL treatment 6 months afterward or other form of visual aid (Intacs, scleral lenses, glasses, etc.) than those that underwent iontophoresis + excipient, epi-off iontophoresis, or standard epi-off UVA CXL. 

In conclusion, iontophoresis when paired with MC allows for a much higher rate of Rf diffusion into the corneal stroma before UVA treatment that is sufficient for the formation of collagen crosslinks at a level equivalent to traditional UVA CXL. When compared to iontophoresis alone, stromal Rf penetration was tripled and CAF was four-fold greater. These values are consistent with those reported in studies where investigators report a 50% effect of iontophoresis alone when compared to epi-on UVA CXL. Further studies are possible with iontophoresis parameter variability (accelerated crosslinking) and modification of the Rf solution makeup and concentration. The greatest benefit of this iontophoresis + MC technique would be in conjunction with our nonlinear optical crosslinking (NLO CXL) treatment. This method uses a two-photon method of Rf activation as opposed to the single photon utilized by UVA CXL. With NLO, activation (and therefore CXL) only occurs within the focal volume of the laser which can be set to any depth and scanned through any pattern, allowing for crosslinking of regions below the epithelium without affecting the epithelium at all, leading to a faster and more precise CXL. 

Acknowledgments

The authors acknowledge support to the Gavin Herbert Eye Institute at the University of California, Irvine from an unrestricted grant from Research to Prevent Blindness and from NIH grant P30 EY034070. 

Supported by NIH R01EY024600. Research reported in this publication was supported by the Office of the Director, National Institutes of Health of the National Institutes of Health under Award Number S10OD028698. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 

This study was made possible in part through access to the Optical Biology Core Facility of the Developmental Biology Center, a shared resource supported by the Cancer Center Support Grant (CA-62203) and Center for Complex Biological Systems Support Grant (GM-076516) at the University of California – Irvine, Irvine, CA. 

Disclosure: R. Joshi, None; S. Bradford, M2Cor, Inc. (I), a company that may potentially benefit from the research results. The relationships with M2Cor, Inc. has been reviewed and approved by the University of California, Irvine in accordance with its conflict of interest policies; S. Luo, None; E. Farrah, None; Y. Xie, None; D.J. Brown, M2Cor, Inc. (I); T. Juhasz, M2Cor, Inc. (I); J.V. Jester, M2Cor, Inc. (I)