KEY CONCEPTS
- •
Corneal cross-linking (CXL) is a photopolymerization process using riboflavin as initiator or photo mediator.
- •
In the cornea, oxygen and energy (ultraviolet light) are consumed during this process, riboflavin is degraded, free radicals (reactive oxygen species) are formed, and cross-links are formed as new covalent bonds within the extracellular matrix.
- •
This chapter highlights the whole process in detail and discusses the experimental and clinical studies related to CXL.
Basic Parameters
Corneal cross-linking (CXL) is a photopolymerization process using riboflavin as initiator or photomediator. Photopolymerization initiators are used in many fields to generate photocurable composites. Within these composites, polymerization is triggered by an irradiation interacting with the photomediator, for example, with ultraviolet (UV) light. This leads to altered physical properties of the composites such as solubility, viscosity, and elasticity. In particular, the phenomenon in which a liquid state changes into a solid state is most useful and is applied to surface-treating techniques in fields including painting, printing inks, dental materials, and lithography, among others.
In the cornea and in particular in the extracellular matrix, the composites are represented by collagen and proteoglycans. Using riboflavin as initiator, which is stimulated by near UV light with a wavelength of approximately 365 nm, polymerization does not occur, because the third adjunct, oxygen, is missing. Although the detailed steps of the process are not fully understood yet, we can summarize by stating that (1) oxygen and energy (UV light) are consumed during the process, (2) riboflavin is degraded, (3) free radicals (reactive oxygen species) are formed, and (4) cross-links are formed as new covalent bonds within the extracellular matrix. (For more details, see the paper of Hayes et al. )
The first ingredient of CXL is riboflavin.
In a deepithelialized cornea, the intrastromal gradient of riboflavin was measured by two-photon microscopy after application of 0.1% riboflavin solutions in either 20% dextran or 1.1% hydroxypropyl methylcellulose (HPMC) for 10 or 30 minutes ( Fig. 26.1 ). The results appear similar and are in accordance with the mathematical diffusion equation indicating a one-directional passive diffusion process. As riboflavin experiences only minor photodegradation, the effective loss of riboflavin happens mainly by diffusion across the endothelium into the aqueous. The riboflavin diffuses not through endothelial cells but passes through intercellular spaces. This is one reason why, during the irradiation only every 5 minutes, riboflavin drops have to be applied. The riboflavin film acts as protection for more posterior structures (additional UV absorption) and prevents corneal dehydration. It is not clear whether all riboflavin molecules act as reaction partners for CXL or whether a fraction of riboflavin molecules build dimers that would not act as photomediators. This would mean that the “active” riboflavin concentration is significantly smaller than 0.1%.
The second component of CXL is UV light of 365 nm in wavelength, which was selected based on the absorption spectrum of riboflavin ( Fig. 26.2 ). The second absorption maximum is at 440 nm, which, however, was avoided owing to the potential hazard of blue light for the retina. UV light is absorbed by riboflavin according to Lambert-Beer’s law (exponential decay). However, as the concentration of riboflavin decreases with depth in the cornea, the gradient of the light intensity flattens toward the endothelium ( Fig. 26.3 , blue line ). In total, approximately 90% of the UV light is absorbed in the riboflavin-soaked cornea. In principle, it is one photon that lifts the riboflavin molecule into an excited singlet or triplet state and, therefore, only the number of photons counts for the efficiency regarding CXL. Holding the radiant exposure constant at 5.4 J/cm 2 , the number of photons hitting the cornea is identical in cases of 3 mW/cm 2 for 30 minutes, 9 mW/cm 2 for 10 minutes, or 30 mW/cm 2 for 3 minutes, and all these combinations should provide identical results. However, for many biological systems, this rule (Bunsen-Roscoe law) does not hold, and in particular, for CXL it has been shown to be invalid. ,
The third ingredient is oxygen, which participates in radical formation and in the chemical bonding process. The gradient of oxygen in the deepithelialized corneal stroma without cross-linking decreases only minimally: at the Bowman’s layer level, it equals atmospheric pressure (21% of volume) decreasing to 17% at 100 microns and to 14% at 300 microns. Again, the oxygen diffuses passively into the cornea, and the gradient is flatter compared with the riboflavin gradient because of the much smaller size of the O 2 molecule and its faster diffusion (see Fig. 26.3 , black line ).
The situation changes dramatically once UV light initiates the CXL process (see Fig. 26.3 , black interrupted line ). Whereas the riboflavin gradient and the UV decay remain practically unchanged, the oxygen concentration decreases within 20 seconds and to less than 3% at 100-microns depth at an irradiance of 3 mW/cm 2 and to 1% at 9 mW/cm 2 . With irradiances higher than 9 mW/cm 2 , oxygen falls beyond the measurement sensitivity of the sensor (1%). When the depletion process is faster and extinguishes oxygen more completely, the irradiance is higher. With 3 mW/cm 2 even at 200-micron depth, the equilibrium oxygen concentration (during CXL) is at 1.5%, which means that there is still some oxygen to support radical formation. This result is clearly not achieved when using 18 mW/cm 2 or more. Compared with riboflavin and UV light, the oxygen gradient is now by far the steepest (see Fig. 26.3 ) and determines, therefore, the depth of CXL. As a consequence, oxygen represents a kind of bottleneck parameter that limits the efficiency of CXL, and this may be the reason the Bunsen-Roscoe law is not valid for irradiances higher than 9 mW/cm 2 .
A new technical approach employing an oxygen concentration of >95% over the cornea tries to counteract this bottleneck and, indeed, under such a hyperoxygenation even with 9 mW/cm 2 , there is a measurable oxygen concentration of 1.2% at a depth of 300 microns.
Experimental Studies
For electron microscopy of the cornea, aldehyde fixatives are used, and the result is a rigid and clear piece of tissue looking like glass. Glutaraldehyde and formaldehyde belong to the group of cross-linking fixatives and were used in the initial cross-linking experiments in Dresden in the 1990s. The stiffening effect was spectacular; however, the side effects were also significant. Another approach was the use of sugar solutions, which are known to produce advanced glycation end products (AEGs) in biologic tissues. These cross-links are also responsible for complications in diabetes mellitus. The Boston group tried another photomediator approach (rose bengal and green light), which delivered significant stiffening of the cornea. However, in animal experiments side effects such as prolonged epithelial healing occurred.
After 5 years of experimental work in Dresden in the late 1990s, it was clear that the technique of riboflavin (0.1%) in dextran solution and UV light was most promising. Riboflavin is a natural vitamin and is not toxic. In a saturated cornea, at least 90% of the UV light is absorbed, so the residual UV irradiance is at least a factor of 10 below the damage thresholds of iris, lens, and retina. In the pre-LED era, we had to use a mercury vapor lamp and band filters delivering 3 mW/cm 2 at the cornea level. In rabbit experiments, we learned that a radiant exposure of 5.4 J/cm 2 killed keratocytes approximately 300 microns deep and the endothelial cells showed neither apoptosis nor necrosis. , Based on theoretical calculations of the riboflavin concentration at the level of the endothelium (again: diffusion equation), the UV damage threshold of the endothelium was determined, which corresponded to the irradiance at corneal depth at approximately 330 microns. Including safety margins, we concluded that the cornea prior to CXL should have a minimal thickness of at least 400 microns. Recently, we discovered that the solution of the diffusion equation was overly pessimistic and that the measured riboflavin concentration at the endothelium level is at least a factor of two smaller, which implies a revision of the 400-micron rule.
Wollensak who joined the group only in the late 90-ties discovered that crosslinked cornea is more resistant against digesting enzymes16, which plays a role during corneal melting. Native pig corneal buttons were digested in trypsin solution within 2 days, whereas in cross-linked corneas, the process took 5 days and more.
Clinical Studies
The first clinical application of CXL was anti-melting treatment in four eyes with therapy-refractory corneal melting. Three of the four corneas healed, and in only one eye was a keratoplasty à chaud necessary. As many of these melting processes create ulcers localized in the periphery of the cornea, this application has become the domain of customized CXL, and we mostly use ring segment irradiation patterns ( Fig. 26.4 ). During this treatment, the 400-micron rule is clearly violated but sacrificing endothelium of a small limited area in this circumstance is considered reasonable.
