Photochemical Keratodesmos for Repair of Lamellar Corneal Incisions

Abstract

purpose. To determine the efficacy of photochemical keratodesmos (PKD) for closing surgical incisions in the cornea of enucleated rabbit eyes compared with that achieved using sutures and self-sealing incisions.

methods. A 3.5-mm incision, at an angle parallel to the iris, was made in the cornea of enucleated New Zealand White rabbit eyes. The intraocular pressure required to cause leakage (IOP_L) from the untreated incision was then recorded. Photochemical keratodesmos treatment was then performed by application of a dye, Rose Bengal (RB), in saline solution to the surfaces of the incision wound, followed by laser irradiation at 514 nm from an argon ion laser. Immediately after treatment, the IOP_L was measured. Both dose and laser irradiance dependencies were studied in five or more eyes for each condition and appropriate control eyes. The IOP_Ls were compared with those obtained using conventional interrupted 10-0 nylon sutures. Other dyes were tested in a similar fashion.

results. The IOP_L of 300 mm Hg was obtained using a fluence of 1270 J/cm² with an irradiance of 1.27 W/cm² (laser exposure time, 16 minutes 40 seconds). No sealing was observed using dye or light alone where control pressures of approximately 30 mm Hg were found. At higher dose (1524 J/cm²) and irradiance (3.82 W/cm²; 6 minutes 35 seconds), PKD was less effective, which may be attributable to thermal effects. PKD produced IOP_Ls similar to those in closure by sutures. Other dyes such as riboflavin-5-phosphate and N-hydroxy-pyridine thione also produced efficient bonding after PKD. Nonphotochemically active dyes did not produce significant increases in the IOP_L at which leakage occurred.

conclusions. The increase in IOP_L after PKD treatment, comparable with that with sutures, in enucleated rabbit eyes demonstrates the feasibility of this technique ex vivo.

The ideal technique for wound closure in the cornea would be simple and rapid and would produce a watertight seal without astigmatism or inflammation. Closure of corneal wounds is often associated with induced astigmatism, partly due to uneven suture tension. This is prevalent after penetrating keratoplasty in which numerous sutures are needed to hold the graft in place. Suturing techniques designed to evenly distribute tension across corneal grafts may still result in significant astigmatism. Although factors such as wound healing, host graft sizing, and trephination techniques also play a role in postoperative astigmatism, a method to hold the graft with equally distributed force could help reduce the postoperative astigmatism. 

Possible alternatives to sutures include adhesives, such as fibrin ^{1 2} and cyanoacrylate glue. ³ Additionally, photodynamic tissue glue, composed of a riboflavin-5-phosphate and fibrinogen mixture, is reported to close cataract incisions and attach donor cornea in corneal transplants. ^{4 5 6} Temperature-controlled tissue welding also has been attempted in the cornea. ⁷  

Photochemical keratodesmos (PKD) may offer an improved result using a relatively easy method. Unlike photothermal tissue welding, ^{8 9 10 11 12 13} PKD can produce a tissue–tissue seal without collagen denaturation or heat-induced peripheral tissue damage. PKD involves the application of a photosensitizer to the wound surfaces followed by laser irradiation to seal the wound. Strong covalent cross-links are believed to form between collagen molecules on opposing surfaces to produce a tight seal. Photosensitization of proteins has been previously reported to produce intermolecular covalent cross-links. ^{14 15 16 17 18} Light and photosensitizers also have been reported to cause collagen cross-links. ^{19 20 21} However, there are few reports of PKD. Photochemical tissue welding of dura mater has been reported, using 1,8-naphthalimides irradiated with visible light. ²²  

In this study, we have tested the efficiency of PKD for closing small keratome incisions in the cornea of enucleated rabbit eyes. Rose bengal (RB), riboflavin-5-phosphate (R-5-P), fluorescein (Fl), methylene blue (MB) and N-hydroxypyridine-2-(1H)-thione (N-HPT) have been compared as photosensitizers. 

Materials and Methods

Young, albino rabbit eyes were received on ice (Pel-Freez, Rogers, AR) approximately 17 to 24 hours after death and enucleation. The eyes were kept on ice and used the same day. The eye to be studied was mounted on a plastic-covered polystyrene block and fixed in position by needles inserted through the extraocular muscles into the polystyrene. The eye was then placed under a dissecting microscope allowing visualization of the treated area during the entire procedure. A 27-gauge needle was inserted parallel to the iris, 2 mm anterior to the limbus into clear cornea, and positioned above the lens in the anterior chamber. The needle was connected to both a blood pressure transducer (Harvard Apparatus, Holliston, MA) and a mini-infuser (Bard 400; Harvard) through a T coupler. The pressure transducer consists of a transducer element that is hardwired to an amplifier box and uses a semidisposable dome with an integral silicone rubber membrane. Pressure inside the dome is transmitted through the membrane to a plastic button, the motion of which is translated to a voltage. The voltage generated by the transducer–amplifier combination is proportional to the lower limit of intraocular pressure (IOP). Signals from the transducer amplifier were recorded on a computer (Macintosh G3 Powerbook; Apple, Cupertino, CA, equipped with a PCMICA[ Daqcard-1200] data acquisition card; National Instruments, Austin, TX). Data acquisition was controlled using programs written in commercial software (LabView 4; National Instruments). The voltage from the transducer and amplifier was converted to pressure by calibrating with a standing manometer. 

Experiments on individual eyes were initiated by increasing the IOP to 30 to 40 mm Hg, using water infusion at a rate of 1 ml/min. An incision was made in the cornea, 1 mm anterior to the limbus (j) and parallel to the iris, using a 3.5-mm angled keratome (Becton Dickinson, Lincoln Park, NJ). For each eye, the IOP required to produce fluid leakage from the incision (IOP_L) was recorded before and after PKD treatment. The photosensitizer, dissolved in phosphate buffer solution (PBS [pH 7.2]; Gibco, Grand Island, NY) was applied to the walls of the incision using a gastight 50-μl syringe (Hamilton, Reno, NV) with a 27-gauge needle. Confocal fluorescence spectroscopy confirmed the location of RB on the incision walls and indicated that the photosensitizer penetrated approximately 100 μm laterally into the wall of the incision. 

The photosensitizers, their absorption maxima, and their absorption coefficients at the laser wavelength were: RB, 550 nm, 33,000 dm³/mol per centimeter at 514 nm; Fl, 490 nm, 88,300 dm³/mol per centimeter at 488 nm; MB, 664 nm, 15,600 dm³/mol per centimeter at 661 nm; R-5-P, 445 nm, 4330 dm³/mol per centimeter at 488 nm; and N-hydroxypyridine-2-(1H)-thione (N-HPT), 314 nm, 2110 dm³/mol per centimeter at 351 nm. The photosensitizers were used as received, with the exception of N-HPT, which was recrystallized twice from aqueous ethanol before use. The concentrations of the photosensitizers were adjusted so that all the solutions had an absorbance of approximately 1.0 in a path length of 200 μm at the laser irradiation wavelength, with the exception of N-HPT for which the absorption was lower by approximately a factor of 10. 

Irradiations used a continuous-wave (CW) argon-ion laser (Innova 100; Coherent, Palo Alto, CA) at 488 nm (for Fl and R-5-P), 514.5 nm (for RB), or 351 nm (for N-HPT). An argon-ion pumped-dye laser (CR-599; Coherent) with 4-dicyanomethylene-2-methyl-6-(p-dimethylaminostyryl)-4H-pyran dye (Exciton, Dayton, OH) was used for irradiation at 661 nm (for MB). Laser light was coupled into a 1-mm diameter quartz fiber, and a 1- cm diameter spot on the tissue was created by using a combination of 1- and 2-in. focal length, S1-UV–grade fused silica, biconvex lenses (Esco Products, Oak Ridge, NJ), mounted in a cage assembly (SM1 series; ThorLabs, Newton, NJ). The 1-cm diameter circular spot was sufficient to cover the entire incision, and the optics were adjusted so that the laser light was incident on the cornea at an angle approximately 45^o to the plane of the incision. Dose–response curves were obtained by varying the duration of the irradiation at a constant irradiance. In separate experiments, the effects of laser irradiance were investigated by comparison of the same delivered dose using different irradiances. The doses used ranged from 124 to 1524 J/cm², and the irradiances used were 0.64, 1.27, 2.55, and 3.86 W/cm². The laser exposure time varied from 33 seconds for the lowest dose using the highest irradiance to 26 minutes, 27 seconds for the highest dose using the lowest irradiance. 

The IOP_L was recorded immediately after treatment. Infusion was started (1 ml/min), and the IOP increased until a maximum was reached, followed by a sharp decrease, corresponding to the opening of the incision and leakage of fluid from the anterior chamber. A typical trace showing the changes in IOP with infusion time is shown in Figure 1 . Five to 10 rabbit eyes were tested for each condition of dose and irradiance. 

Control experiments included: irradiation with no photosensitizer application, photosensitizer application only and no photosensitizer or laser irradiation. In the experiments using no photosensitizer, PBS was applied to the incision walls, using the same method as described for the photosensitizers. In control experiments with no laser irradiation, the eye was allowed to stand for the same period as the laser-treated samples. 

Results

Photochemical Keratodesmos with RB

Treatment of incisions with RB (1.5 mM) and 514-nm laser light resulted in an increase in posttreatment IOP_L. Control experiments demonstrated that a significant increase (P < 0.005) in the IOP_L after PKD treatment occurred only when both RB and laser irradiation were applied and not after either alone (Fig. 2) . The mean IOP_L of incisions treated with RB and 514-nm laser light was greater than 300 ± 48 mm Hg, whereas laser irradiation alone or photosensitizer alone produced no significant increase between the pre- and posttreatment IOP_Ls. 

Dose response curves for IOP_L are shown in Figure 3 for doses delivered at irradiances of 1.27, 2.55, and 3.82 W/cm². A clear dose–response relationship was observed at the lowest irradiance (1.27 W/cm²) for doses between 508 and 1270 J/cm² (Fig. 3A) . No significant increase in the IOP_L was observed for doses below 508 J/cm² at any irradiance. PKD was most efficient at 1270 J/cm² delivered at an irradiance of 1.27 W/cm², All doses delivered at the two lower irradiances (1.27 and 2.55 W/cm²) gave IOP_L greater than 100 mm Hg. Treatment using irradiances of 2.55 and 3.82 W/cm² produced no obvious dose–response pattern. In general, for a selected dose the IOP_L was lower at higher irradiances. For example, at 1270 J/cm² the mean IOP_Ls are 274, 150, and 130 mm Hg for the irradiances 1.27, 2.55, and 3.86 W/cm², respectively. 

In addition to reduced IOP_L, thermal damage was consistently observed at doses of 762 to 1524 J/cm² at the highest irradiance (3.82 W/cm²) and occasionally at 2.55 W/cm². Tissue shrinkage and deformation around the wound site were taken as signs of thermal damage. Significant bleaching and color change of the photosensitizer were observed after irradiation. 

A seal was considered a failure if the IOP_L after treatment was in the range of 30 to 40 mm Hg, the same range of pressures measured for pretreatment IOP_Ls. A mean failure rate of 40% was found for the PKD treatment of incisions with RB and light doses of 508 to 1524 J/cm² for all irradiances. Failure of a treatment did not correlate with light dose or irradiance. However, high failure rates correlated with obvious deterioration of the enucleated eyes by the time of the experiment, as was evident from globes that were not firm and corneas that appeared milky. In addition, nonuniform application of the dye as determined by observation through the dissecting microscope, also appeared to correlate with treatment failure. 

Investigation of Other Photosensitizers for PKD

RB was chosen for these studies, because it photosensitizes cross-linking of soluble collagen as detected by sodium dodecyl sulfate (SDS)–gel electrophoresis. RB is known to undergo reductive and oxidative electron transfer reactions as well as singlet oxygen production. ^{23 24} With this in mind, additional dyes were evaluated that were also known to produce radicals after photoexcitation. R-5-P was selected for evaluation, because flavins undergo photoelectron transfer processes, in addition to singlet oxygen production. Riboflavin has been shown to photosensitize the formation of cross-linked collagen molecules ²⁵ and to increase the stiffness of the cornea through the induction of collagen cross-links. ²⁰ Our previous studies have shown that R-5-P and 355-nm light efficiently cross-link soluble collagen. ²⁶ The application of 11 mM R-5-P and irradiation using 488 nm light, at the same irradiances used for RB, and doses of 762 and 1016 J/cm², significantly increased IOP_L after PKD treatment (P < 0.05; Fig. 4 ). The IOP_Ls observed using R-5-P were of a magnitude similar to those for RB. However, the IOP_Ls observed for each dye at the same irradiance and dose were not comparable. Although the treatment produced significant increases in IOP_L, no simple pattern between the two dyes was observed. 

Another photosensitizer, N-HPT, produces hydroxyl radicals and other reactive species after UV irradiation, ^{27 28 29 30} and studies have shown it to be an efficient agent for cross-linking of soluble collagen. ²⁶ A 4.5-mM solution of N-HPT was applied to the walls of the incision and irradiated using 351 nm light (0.64 W/cm²) at doses ranging from 127 to 508 J/cm². No significant increase in the IOP_Ls was observed at the lowest dose. However, mean IOP_L values of 60 ± 23 and 126 ± 40 mm Hg were produced when using the doses of 254 and 508 J/cm², respectively—lower doses than used for the other photosensitizers. 

MB is a frequently used dye in ophthalmic surgery that has been reported to photosensitize collagen cross-links in rat tail tendon. ¹⁹ Our previous studies showed that MB and 355 nm light did not produce efficient cross-linking of soluble collagen, ²⁶ and MB was therefore used as a control in these ex vivo studies. MB (3 mM) applied to the walls of the incision and irradiated with 0.64 W/cm² of 661 nm light at doses of 508, 762, and 1016 J/cm² did not increase the posttreatment IOP_L. However, it was observed that MB did not stain the corneal tissue efficiently, explaining its low efficiency in PKD. 

Laser-activated tissue welding has been studied in a variety of tissues. ^{8 31 32 33 34 35 36 37 38} In tissue welding, the laser radiation is used to heat the tissue to temperatures at which collagen denatures and, on cooling, the collagen molecules intertwine to form a weld. Additionally, dye-enhanced thermal welding has been investigated. ^{8 39} In this method, the dye selectively absorbs the laser energy and then releases heat to the desired area, reducing peripheral tissue damage. These methods, however, are not appropriate for the cornea because of the potential reduction in visual acuity that would result from the corneal deformation produced by thermal tissue damage. When performing PKD on the cornea, heating must be avoided. 

We evaluated the possibility that nonphotochemical processes contribute to wound closure by comparing PKD produced by RB with that produced by Fl, a dye with a similar structure but one that is not expected to induce protein cross-links. RB and Fl are both xanthene dyes. However, RB is halogenated (four iodines and four chlorines), and the presence of these heavy atoms causes RB to be photochemically active. ⁴⁰ Fl has a high quantum yield of fluorescence ⁴¹ and lower quantum yield of triplet state formation than RB ⁴⁰ and will, therefore, produce a lower proportion of active species with the potential to produce collagen cross-links. A solution of 0.6 mM Fl was applied and irradiated using 488-nm laser light at the same range of irradiances used for RB and at doses from 508 J/cm² to 1016 J/cm² (Fig. 5) . No increase in IOP_L was observed for the incisions treated with the two lowest doses using the two lowest irradiances studied. However, at the highest dose for all irradiances, an increase in IOP_Ls was observed with pressures ranging from 63 ± 30 to 89 ± 42 mm Hg, although this is much less efficient than RB (compare Figs. 3 and 5 ). These results suggest that PKD is indeed produced by photochemical processes. The IOP_L of 116 ± 40 mm Hg obtained using a dose of 762 J/cm² at 3.82 W/cm² (laser exposure time of 3 minutes, 10 seconds) is considerably higher than any other observed using Fl. The sealing observed at the highest irradiance (3.82 W/cm²) and dose (762 J/cm²) suggests that some other effect is operating, such as a thermal mechanism under these conditions. 

Photochemical Keratodesmos Versus Sutures

The IOP_L after PKD treatment was compared with that obtained using sutures. Two interrupted radial sutures of black monofilament 10-0 nylon (Ethilon suture; Ethicon, Piscataway, NJ) were used to close the keratome incision. The sutures were placed in a radial fashion at approximately 90% corneal depth. Preliminary experiments produced IOP_Ls of approximately 230 mm Hg. This pressure is similar for the incisions closed with PKD treatment. However, it was observed that the IOP was maintained even when there was leakage around the sutures. The leaks surrounding the sutures were reversible, whereas with PKD the damage was irreversible after the opening of the incision. 

Discussion

This study demonstrates the feasibility of using PKD treatment to close small incisions made in the cornea of rabbit eyes ex vivo. The results show that PKD produces a significant increase in the immediate IOP_L of enucleated rabbit eyes after treatment of 3.5-mm corneal incisions. 

The dose–response pattern observed for the PKD treatment, using RB as a photosensitizer, is not simple. Reduced IOP_L and tissue shrinkage were observed consistently at the highest irradiance of 3.82 W/cm² and occasionally at 2.55 W/cm² for doses between 762 and 1524 J/cm², which suggests contributions from both photochemical and photothermal processes. The ideal conditions to produce a clinically relevant IOP_L with PKD are those that balance the shortest treatment time with the highest dose; the limitation is the thermal effects produced using high irradiances. In the cornea, photothermal effects may produce collagen contraction resulting in distortion of the patient’s vision. Therefore, higher irradiances that would allow a shorter treatment time are limited by thermal effects. 

Other potential photosensitizers for PKD, chosen on the basis of suitable photochemistry, were investigated. PKD treatments using R-5-P and N-HPT produced increases in the IOP_L. Relative efficiencies of the photosensitizers were evaluated by comparing the IOP_Ls produced by optically matched solutions of the photosensitizers at the same set of irradiances and doses. However, these comparisons do not take into account considerations such as the binding efficiency of the photosensitizers, which alters the dye concentration on the incision surface. All the photosensitizers generate singlet oxygen or reactive radicals that may be toxic to cells in the cornea. Future in vivo studies are needed to determine whether this effect is relevant and, if so, to evaluate possible protective agents. 

Our results using R-5-P are comparable with those found by Khadem et al. ^{4 5 6} who used a photoactivated adhesive consisting of fibrinogen and R-5-P irradiated with argon ion laser light (488–514 nm) to close 5-mm penetrating central corneal incisions made in human cadaveric eyes. With this method of incision closure in a smaller sample size, a mean wound-bursting pressure of 154 mm Hg was found. The maximum mean IOP_L observed in our study using R-5-P was 254 mm Hg. Our results suggest that the presence of fibrinogen is not necessary to obtain a good seal. Elimination of fibrinogen from the system removes the limitations imposed by using this protein, such as the limited tensile strength and the requirement that the fibrinogen be isolated from the patient to be treated, to avoid risk of infection from donor plasma. ⁴²  

Other possible suture alternatives for use in ophthalmic surgery that have been investigated include chemical glues. ^{1 2 3} Glues are limited by the requirement that they be nontoxic, noncarcinogenic, and biodegradable. In addition, glues do not generally provide a permanent closure; they are sloughed off within weeks of application. 

Our results show that PKD treatment of small keratome incisions in rabbit cornea ex vivo produced IOP_Ls comparable with those incisions closed with sutures. The leaks associated with the sutures were reversible, but after the PKD-treated incision had been opened, the seal was completely lost. However, PKD treatment can easily and effectively be repeated on the previously treated incision. 

PKD offers many potential advantages over the methods currently used to attach corneal tissue and close incisions in a variety of surgical procedures such as penetrating keratoplasty, laser in situ keratomileusis (LASIK), and cataract surgery and in the treatment of corneal lacerations. The sutures currently used in corneal transplants can induce postoperative astigmatism, neovascularization, and rejection of the donor cornea. Furthermore, loose or broken sutures can leave a patient vulnerable to microbial keratitis. The suturing procedures used are skill intensive and are mainly performed by corneal specialists. PKD offers a simple procedure to close wounds, spot seal LASIK flaps and attach donor cornea, reducing the operating and rehabilitation time. 

 Supported by the DoD Medical Free Electron Laser program, contract number N00014-94-10927, and the Center for Innovative Minimally Invasive Therapy, Massachusetts General Hospital, Boston.

 Submitted for publication February 2, 2000; revised April 27, 2000; accepted May 31, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Typical trace of increasing IOP with infusion time for a PKD-treated eye showing IOP_L at 300 mm Hg.

Figure 2.

 

Mean IOP_L values for PKD-treated eyes (n = 5) using 514 nm light (2.55 W/cm²) and RB (1.5 mM) in PBS. Additional controls are incisions treated with RB or buffer but no laser light.

Figure 3.

 

Mean IOP_L before and after PKD using RB and 514-nm irradiation. RB (10 μl, 1.5 mM) was applied to the incision surfaces, which were then treated with the doses indicated using irradiances of (A) 1.27 W/cm², (B) 2.55 W/cm², and (C) 3.82 W/cm².

Figure 4.

 

Mean IOP_L before and after PKD using R-5-P and 488-nm irradiation. R-5-P (40 μl, 11 mM) was applied to the incision surfaces, which were then treated with the doses indicated using irradiances of (A) 1.27 W/cm², (B) 2.55 W/cm², and (C) 3.82 W/cm².

Figure 5.

 

Mean IOP_L before and after PKD using Fl and 488-nm irradiation. Fl (40 μl, 0.6 mM) was applied to the incision surfaces, which were then treated with the doses indicated using irradiances of (A) 1.27 W/cm², (B) 2.55 W/cm², and (C) 3.82 W/cm².

The authors thank Norman Michaud and Thomas Flotte for collecting the confocal images and for useful discussion, Hans-Christian Luedemann and Dominic Bua for technical help, and Béatrice M. Aveline for preparation of N-HPT.