Collagen Cross-linking with Riboflavin: What’s Really Going On?

Collagen Cross-linking with Riboflavin:  What’s Really Going On?
This year’s European Society of Cataract and Refractive Surgeons meeting (ESCRS) in Berlin once again showcased the remarkable results seen in keratoconus patients photochemically treated with riboflavin and UV-A (320 to 400nm wavelength). This technique goes by several acronyms including, CCR, CxL, CCL, and C3-R. The name, “collagen cross-linking” implies two things: firstly, that riboflavin’s mechanism of action is specifically targeted to collagen, and secondly that the stromal remodeling that occurs via this treatment is actually achieved by cross-linking collagen fibers. I’m not convinced either is entirely true. In fact, the likelihood is that there is a great deal of non-specific chemistry occurring that contributes to the effect seen in the clinical studies. It’s far more probable that singlet oxygen formation damages any nearby molecules, including collagen, just as much as any cross-linking chemistry. The corneal remodeling seen in these patients must be in response to multifaceted CCR-induced photochemical injury to the stroma.

Still, the treatment has shown simply staggering results in staunching the progression of keratoconus and in many cases reversing the progression of the disease with and without the use of Intacs. There have also been extremely encouraging data in the treatment of post-LASIK ectasia. Data from several European studies show that this effect is durable, with stable results as far as three years out. Eight trials are currently underway evaluating various aspects of the treatment around the world. Several reports have shown that the corneal topography of these patients continues to improve years after the treatment suggesting that whatever effect is occurring is not only durable, but continues to chart a favorable course.

Figure 1: Riboflavin molecular model

Riboflavin (vitamin B2) is generally regarded as safe since it is a vitamin ingested in normal diets and an omnipresent molecule in biological systems. Extensive toxicology data exist on riboflavin. Its properties as a photosensitizer have been understood for more than 30 years. It is also surmised that any residual riboflavin and any photoproducts produced during the treatment do not present any hazardous risks.

The procedure often begins with chemically stripping the corneal epithelium prior to application of the riboflavin solution. Complete debridement of the corneal epithelium is viewed as an essential component of riboflavin-UVA cross-linking therapy to permit the penetration of riboflavin into the corneal stroma. Riboflavin is added drop-wise at roughly 5-minute intervals during the procedure. The riboflavin acts both to enhance the cross-linking effect and, according to some, to protect the rest of the eye from the UV exposure. The patient’s cornea is exposed to a bank of UV emitting LEDs (wavelength = 365nm, power = 3.0 mW/cm2) for 30 minutes with repeated administration of riboflavin drops.

Figure 2: Patient undergoing CCR treatment

At the ESCRS meeting, John Kanellopoulos, MD, presented data on the use of higher fluence UV light sources (7 mW/cm² ) to reduce the exposure time to only 15 minutes. The goal was to decrease the time a patient is subjected to the treatment by changing the treatment parameters (twice the light intensity over half the time). His data showed roughly equivalent outcomes suggesting that this twist on the procedure may be practical.

As the theory goes, adjacent strands of collagen are chemically cross-linked as the riboflavin molecule swaps covalent bonds to the proteins triggered by the energy of the UV-A light, thereby mechanically stiffening the cornea. This change in the biomechanical properties of the cornea can be characterized as "corneal hysteresis" when measured by Reichert’s Ocular Response Analyzer. But is this effect specific to collagen, or are other proteins similarly affected? If so, what is the biological effect of this collateral cross-linking? The literature refers to rates of keratocyte loss. Perhaps the cells that aren’t killed mediate a remodeling resulting it the abatement of the keratoconus disease process.

So then, how deep does the riboflavin penetrate into the stroma? By some estimates using basic slit lamps, only the outer third of the stroma is treated. As most corneal surgeons with experience in this field will tell you, the endothelial layer appears to be similarly unaffected when measured by specular microscopy. The cell density appears to remain the same immediately after treatment and long after. This strongly suggests that the riboflavin and UVA treatment is “sub-lethal”. Why the use of that term?

In an apparent contradiction in theoretical mode of action, riboflavin is also being billed as a nucleic acid specific cross-linking agent.

The Mirasol system by CaridianNCT (a subsidiary of Gambro AG) is currently being developed for the photochemical decontamination of donated blood products for transfusion. In this case, the system is touted as having a specific effect on the nucleic acid of bacterial, viral, protozoan human pathogens. All the while, the proteins of the blood are suggested to be completely unaffected. This includes all the elements of the complement system, albumin, fibrinogen, etc. Non-specific chemistry has to be occurring throughout the cell altering many kinds of molecules from extracellular proteins, cell surface glycoproteins, cellular proteins, ribosomal RNA and genomic nucleic acid.

So which is it? Is riboflavin a collagen cross-linker or a nucleic acid cross-linker? You really can’t have it both ways and yet still call it a “specific” process.

The dose-response curve has not been optimized and given how rare these patients are, clinical studies to provide statistical significance in any particular experimental treatment protocol may never result in a perfected regimen. To some degree, it’s a bit like anesthesia with a brick – it works, but there really has to be a better way. I believe the mechanism of action to be one akin to a response to injury (deposition of additional collagen), which will vary depending on the individual. Though CCR is approved for use in Europe, and may represent “the standard of care” for these patients, standardizing the procedure will be key to ensuring outcomes can be compared in retrospective studies.

As is typical of any discussion regarding a new technology, this article must also end with “additional work needs to be done to understand this intriguing new treatment”. Perhaps the future will lead us down the well-trodden path of radiotherapy. Either way, thousands of patients are thankful for scientific and medical innovations like these.

Have a question or comment on this article? Use the "Comment" link above to leave your thoughts, and the author will respond.

* * * UPDATE * * *

Shortly after submitting this piece, I received a personal communication from Dr. Gregor Wollensak. He has agreed to let me post the information he shared in our correspondence. Being one of the foremost pioneers of this technology, Dr. Wollensak is perfectly placed to offer advice on the technique based on his many years of experience. I am indebted to Dr. Wollensak for his generosity in taking the time to explain some of the key points of the photochemistry and the clinical outcomes of this procedure. Hopefully, you too will find his comments “illuminating”.

From: Gregor Wollensak
Sent: Sunday, October 19, 2008 4:00 AM
To: Ryan Alfonso
Subject: crosslinking

Dear Ryan,

I also attended this year´s 2008 ESCRS meeting in Berlin and read with interest your recent article entitled "Collagen Cross-linking with Riboflavin: What´s really going on?" on OphthalmologyWeb. I introduced this new technique for the treatment of keratoconus in 2003 (Wollensak et al. AJO 2003;135:620-627) and I think I can clarify some of the issues and add some corrections.

1. The UVA wavelength used in the cornea is 370nm.

2. The acronym used is mostly CCL or CXL. C3-R refers to a modification of the technique by Boxer-Wachler from California without removal of the epithelium.

3. The biomechanical effect by the riboflavin/UVA treatment is induced by predominantly intrafibrillar photo-crosslinking of the collagen fibrils leading to covalent dityrosine bonds (Kato et al. Photochem Photobiol 1994; 59:343-349), an increased collagen fiber diameter (Wollenak et al. Cornea 2004; 23:503-507) and a collagen polymer product bigger than 1000 kDa (Wollensak et al. Cornea 2008;27:353-356). The biomechanical effect is present immediately after the irradiation leading to an increase of the biomechanical rigidity of the cornea by about 300% ((Wollensak et al. JCRS 2003; 29:1780-1785) stabilizing the cornea at once.

4. It seems that the regression of corneal ectasia continues to some extent in about 50% of the patients in the years after crosslinking (Raiskup-Wolf et al. JCRS 2008; 34:796-801). This is however mostly only a reduction by 2-3 diopters and might be due to the wound healing response including the presence of contractile myofibriblasts (Wollensak et al. Cornea 2007; 26:600-605).

5. The removal of the epithelium is essential for the homogeneous diffusion of riboflavin into the corneal stroma. Without epithelial removal (like in C3-R) the biomechanical effect is less than 50% of the standard crosslinking procedure.

6. The attempts by some groups (for example Kanellopoulos) to reduce the exposure time using fractionated or "flash" irradiation etc. are dangerous I think as long as there are no data from animal experiments regarding its toxicity and biomechanical effeciency. In addition, I think these variations are not really needed because the present standard treatment is safe, comfortable and well accepted by the patients. Besides, it already is the result of experiments aimed at optimizing the procedure including dose-response curves (Speoerl et al. Exp Eye Res 1998; 66: 97-103)

7. The increase in biomechanical rigidity of the cornea after CXL cannot be monitored by the Ocular Response Analyzer because Corneal Hystersis is not correlated with Young´s modulus and the device only measures the biomechanical properties in the sagittal direction using an air-puff system and not in the tangential direction like in biomechanical stress-strain measurements.

8. The riboflavin penetrates throughout the corneal stroma. However, a strong crosslinking effect is only present in the anterior 250µm of the cornea (Wollensak et al. JCRS 2007; 33:516-521) because in the deeper stroma the UVA-irradiance gets too low due to its riboflavin- enhanced absorption in the anterior layers. Similarly, also the cytotoxic effect on keratocytes is present only in the anterior 300µm of the stroma (Wollensak et al. Cornea 2004; 23: 43-49).

9. Using the standard irradiance of 3 mW/cm² and a minimal stromal thickness of 400 µm the endothelium is completely safe because its cytotoxicy threshold of 0.36 mW/cm² is not reached (Wollensak et al. JCRS 2003; 29: 1786-1790).

10. Riboflavin is only used as a photosensitzer increasing the UVA absorption. The UVA induxes oxygen radicals which in turn induce covalent crosslinking of all kinds of proteins such as collagen, fibrinogen etc. In addition, UVA can induce DNA and RNA lesions, an effect which is used for the desinfection of water etc. and also for the sterilization of apheresis blood products killing efficiently viruses, bacteria and other pathogens. These cytotoxic DNA lesions are also the reason for the cytotoxic effect of the new treatment on corneal keratocytes.

11. The riboflavin/UVA system is an efficient collagen crosslinking method just like formaldehyde for example. The main and almost exclusive protein of the cornea is collagen type I so that in the cornea the effect is focused on collagen. In addition, there is a DNA-toxic effect leading to keratocyte apoptosis. In the blood with all its various ingredients it is a completely different story.

12. I agree that the crosslinking treatment should be standardized that is it should be applied as described originally in 2003 (Wollensak et al. AJO 2003;135:620-627) : epithelial removal, irradiation (of 3mW/cm²) for 30 minutes, postoperative antibiotic ointment, no steroid, no contact lens. Using this standard treament scheme no major problems have been encountered so far confirming two well-known proverbs: "Keep it simple" and "Never change a winning team"!

With friendly regards,
Wollensak (Berlin)

  • <<
  • >>