Cross-linking round table, part 1: Controversies, techniques and treatment decisions

At the 2013 European Society of Cataract and Refractive Surgeons meeting, Ocular Surgery News convened a round table of international experts to discuss the current state of corneal cross-linking. The first part of that wide-ranging discussion, moderated by Roy S. Rubinfeld, MD, is featured in this issue of OSN.

Roy S. Rubinfeld, MD: At this moment, there is a lot of controversy in cross-linking. We are at a tipping point in terms of epithelium-on vs. epithelium-off methods. Multicenter U.S. clinical trials are ongoing to study both methods, and there has been a great deal of research and development as well as presentations that are significantly affecting clinical practice. We know much more now than we did just 1 to 2 years ago.

Originally epi-off cross-linking was developed in the 1990s because the riboflavin formulations, with high-molecular-weight dextran, would not penetrate the epithelium to get into the stroma well. At the time, it was an easy mental jump to think of PRK and remove the epithelium. However, there are newer proprietary riboflavin formulations and methods for performing highly effective epi-on cross-linking wherein the riboflavin loads the stroma well, UV light does penetrate, and cross-linking is achieved. Studies are showing best corrected visual acuity and maximum keratometry outcomes to be similar between methods.

Any treatment that is completely effective is suspicious. We have identified now 31 eyes of about 2,200 eyes in our epi-on study that we believe are true treatment failures; that is 0.01%. In Michael Mrochen and colleagues’ Journal of Cataract and Refractive Surgery article “Complications and failure rates after corneal cross-linking,” the rate was about 5%.

William B. Trattler, MD: To the panel: What do you think your personal progression rate is with epi-off cross-linking? In Dr. Mrochen’s paper, it was 5%. Is that about what you are observing now with epi-off? Do you ever see progression?

Arthur B. Cummings, FRCS: Using mean keratometry and maximum keratometry readings as the two parameters, I would say about 5% do not respond.

George O. Waring IV, MD: It is a good idea to separate out the complications from the failure rates because presumably the complication rate may be higher with epi-off but the progression rate might be lower. You have to separate those two variables when you look at this percentage.

Technique-dependent

Trattler: The success rate vs. failure rate depends on the epi-on technique. In the CXLUSA clinical trial, we do a more robust epi-on technique, and our rate of progression appears to be in the same range as that for epi-off, but I have not done enough epi-off cases personally to know for certain.

Rubinfeld: To use a phrase I have heard Dr. Trattler say, “Not all epi-on cross-linking is the same.” I have observed procedures wherein, to standardize the treatment, a well is placed on the eye, a riboflavin solution is placed on the eye, and exactly 30 minutes later treatment starts, but no one has looked at the cornea to confirm the stroma is loaded with riboflavin.

Trattler: What you are describing has been published in a couple of papers that compared epi-on vs. epi-off and reported that epi-on was not as effective. The technique that they used for epi-on was exactly as you described: a timer technique in which the cornea is never evaluated to make sure there is sufficient riboflavin in the cornea for an effective cross-linking procedure. And that is why those papers showed inadequate effect with epi-on; the issue was their technique.

Aleksandar Stojanovic, MD: In combination with laser, I now do phototherapeutic keratectomy first for slight topography-guided optical regularization of the corneal surface and then epi-off cross-linking, but I have done epi-on, too, for a couple of years. With epi-on, I wait quite a long time, maybe 45 to 50 minutes, to get saturation and to identify flare. I look for homogenous riboflavin staining of the stroma and anterior chamber flare — signs that the riboflavin has penetrated. I always wait quite a long time, so that is a disadvantage. After saturation I wash the epithelium, and then I use UV.

Rubinfeld: If you have too much riboflavin on the surface, how much UV goes through?

Trattler: In our experience with epi-on for the CXLUSA study group, we considered that the epithelium might be blocking some UV light, so we switched the riboflavin concentration and the UV power. We went from applying 3 mW for 30 minutes to applying 4 mW for 30 minutes and saw a nice jump in efficacy.

Jorge L. Alió, MD, PhD: I use the Dresden protocol or the one from Avedro. I never reduce the time.

Paolo Vinciguerra, MD: Twenty or 25 minutes could be enough. We get a demarcation line almost at the same depth, but flattening was much less with epi-on. That is our experience using 3 mW and 9 mW.

Rubinfeld: What metric should we use now to determine when to start UV light treatment?

Vinciguerra: First, always check the pachymetry because sometimes if you use the standard solution with dextran, you have a strong reduction in the pachymetry, and then you apply the UV light and have corneal burn and a deep opacity. So always check that you are within the Dresden protocol.

Rubinfeld: Looking at the cornea at the slit lamp, can you use the presence of a homogeneous green in the stroma to be a metric for loading the solution?

Vinciguerra: We published a paper in the Journal of Refractive Surgery that compares use of optical coherence tomography and electrophoresis to measure not only depth but also concentration against standard epi-off procedures. That is another metric that could be used.

Karolinne Rocha, MD, PhD: We published a paper in the Journal of Refractive Surgery last year showing significant irregularity, variability and alterations in regional epithelial thickness in keratoconus using spectral-domain OCT. Significant epithelial thinning is observed over areas in which the anterior stromal curvature is steep and the surface elevated, while thickening of the corneal epithelium is seen over areas of stromal thinning, especially in severe cases. When we talk about riboflavin and light distribution, is it possible to compensate for these areas of epithelial thickening and irregularity?

Michael Mrochen, PhD: Riboflavin, especially if you use it in high concentrations, absorbs that strong light. If you substantially reduce the concentration of riboflavin in the epithelium, then that affects the concentration in the stroma.

Energy distribution

Mrochen: There is a complete misunderstanding in the ophthalmic field about this 5.4 J/cm². If you look at the original data, the peak fluence we had was about 7.6 J/cm². In the periphery, we are actually going out to about 7 J/cm². Starting with 3 mW for 30 minutes is maybe too low. By increasing the numbers, you are in the range that is used in most devices. We started off with the original device in which we had something of a Gaussian profile, but we did not achieve enough effect in the periphery. Sheehan et al showed that about 30% more energy is needed in the periphery, 3 mm from the center, to have the same cross-linking effect. We applied an optical system that delivers much more light in the periphery, about 30% more. In 1-year data presented in Cambridge by Prof. Theo Seiler, 90% of patients had flattening of more than 1 D. In about 60% of cases, there was more than 2 D of flattening.

There is a lot of discussion about time. And there is a lot of discussion about whether we should leave on the epithelium. To optimize the outcome, in my opinion, we have to work on the energy distribution applied to the cornea to achieve the appropriate outcome because, at the end of the day, it is the combination and not just the riboflavin. If you just put in riboflavin, nothing happens.

There is a complete misunderstanding about the energy distribution of the light on the cornea. The light distribution, the energy distribution, over time is the driver.

Developing a decision tree

Waring: Take a step back and look at it from a more global perspective: epi-on vs. epi-off. It is like a risk-benefit analysis. Even if epi-off is slightly more efficacious, it may still be OK to do epi-on in most cases if there is progression because you have the added safety profile and you can presumably re-treat more easily if needed. Potentially, there is a whole treatment algorithm that we could come up with for certain cases that may do better with epi-off, but the vast majority may do better with an optimized epi-on procedure. That may be the future of cross-linking.

Mrochen: I would envision a decision tree. Rather than saying, “This is the right way. This is the standard protocol,” we develop a decision tree for surgeons based on evidence-based criteria so we can offer patients a solution.

Waring: It is a myopic view to think that you are just going to treat one way. A decision tree could reflect more than just a biomechanical outcome, but more of a truly refractive outcome as well.

Mrochen: Whatever we are achieving with cross-linking today is not really improving the vision of the patient. If someone comes in with bad vision, you might fix the problem and it does not get any worse, but that does not have a refractive solution.

Rubinfeld: As a standalone procedure, this is generally the case.

Trattler: We have found improvement in vision with epi-on cross-linking. About 50% of patients improve one line or more, and patients often report improvement in quality vision.

Thin vs. steep

Trattler: Question to the panel: About beam profiles, we have some patients with keratoconus and central thinning and other patients with pellucid marginal degeneration who have peripheral thinning. When using a 9-mm beam, is it best to treat the central cornea with UV light for every patient? Or should we alter the placement of the beam based on the thinnest portion of the cornea?

Mrochen: We advise surgeons to focus or center the beam on the thinnest part of the cornea to make sure they are not harming the endothelium. If you have a de-centered pellucid situation, we advise that surgeons protect the limbus area and de-center the beam. We basically have a profile around 8 mm or 9 mm, and we de-center the whole thing, so the upper part would not be cross-linked.

Waring: There is some really interesting finite element modeling data that suggest that if you do not treat the whole cornea, you may get improved shape. Presumably, there is a gradient of disease and the whole cornea is not equally ectatic. You have your super-threshold point where there is a thinnest point and the weakest part. We want to flatten one part but relatively steepen the other. If you cross-link the entire cornea, you may hold that back. You may achieve more effect by not cross-linking the entire cornea. That is what the model suggests and is something we should look at.

Rocha: The steepest corneal curvature may or may not correlate to the thinnest point in keratoconus. Finite element modeling data of the cornea and in vivo spectral-domain imaging studies suggest that the “stress” in corneal ectasia does not correlate to the steepest area but more likely to the thinnest point or the “weakest point.”

Mrochen: That is the reason why our beam profile has less of the center and goes more on the periphery.

Waring: The steepest area is not the thinnest area because of the epithelial hyperplasia.

Trattler: So we want to be on the thinnest spot, not the steepest.

Rubinfeld: If fibers are stretched thin beyond a certain point, whether or not you cross-link them, you are not going to stop the progression of disease, correct?

Cummings: Mrochen and Seiler published a paper 3 or 4 years ago suggesting that any cornea steeper than 58 D would probably not do well. I do not agree entirely. We have cross-linked corneas up to 65 D, sometimes even 70 D, that flatten by 3 D or 4 D, but in some, the cornea is so stretched and the fibers that you are trying to cross-link are so damaged that it is like trying to cross-link soup. There is no material there. It is all so destroyed that there is nothing you can really cross-link. We do not have hard and fast evidence, but I think for cross-linking to work, there needs to be collagen fibers that have some sort of structural integrity that can be strengthened. Otherwise, there will not be much effect.

Documenting progression

Rocha: Progression documentation prior to cross-linking treatment is important when we are comparing clinical trials. For example, I was involved in the U.S. Food and Drug Administration corneal collagen cross-linking clinical trial at Emory University for 1 year, and we needed to prove progression to enroll patients. This is a discussion among European and U.S. surgeons: “Do we need to prove progression to treat the patients?” For some studies, this is not a requirement. We definitely had eyes that were progressing, and then they were treated with epi-off cross-linking. We observed a stop in progression and even topographic improvement. I believe keratoconus patients should be treated upon the diagnosis despite significant clinical and/or topographic progression, especially in young patients. Cross-linking is a safe procedure if you follow the safety standards and guidelines, including the correct riboflavin concentration, UV light exposure, residual stromal bed and endothelial cell count.

Stojanovic: One point that has been made clear in the discussion whether to treat early disease is this: If you have a patient younger than 20 years with signs of keratoconus, then, go, treat.

Rubinfeld: This comes back to the idea of epi-on, epi-off, safety of the procedure and the development of an algorithm, as Dr. Waring suggested.

Trattler: I feel very comfortable treating early. We have a safe procedure in epi-on in which the risk is extremely low for all patients, young and old. Not only will epi-on stop progression, but often you will see improvement in corneal shape. I expect that over time we will see more people jump on the epi-on bandwagon.

Mrochen: In Europe, all of the systems are approved for progressive keratoconus. So if you do a treatment that is not shown to be progressive for the patient, you end up with off-label use.

Waring: Do you believe that progression needs to be documented before you treat?

Alió: I believe in treating case by case. With good vision and a stable condition in terms of evolution, I do not indicate immediately cross-linking as a dogmatic indication. We have a tool, but when should we use it? We use it when the patient is at risk for losing vision. Epi-off is badly tolerated in children, and infections frequently can happen. This is a tool that has a risk-benefit that you have to balance.

Trattler: I have been doing epi-on exclusively for 3.5 years. The procedure itself is extremely safe. The downside is close to zero for patients, and I feel very comfortable recommending and treating right away. Even if the patient ends up seeing about the same on the Snellen chart, that is OK. There are still typically improvements in corneal shape and in quality of vision.

Patient’s perspective

Cummings: The other point I want to make is this: At the end of the day, it is the patient’s opinion that matters, so that has to be a part of our decision tree.

Typically, patients get some sort of psychological reassurance from the positive effects of cross-linking. That helps them a huge amount. They start rubbing their eyes less. They start chilling out a bit. They can wear lenses more readily, and so that makes them happier.

Trattler: If you ask a patient a month after cross-linking and he has received a new scleral contact lens, he may say that he sees better, but was it the cross-linking or just the fact that he got fitted with a new contact lens? That is what makes questionnaires such as the National Eye Institute Visual Function Questionnaire 25 so hard to decipher because cross-linking may not be the only reason the patient is seeing better. He has new contact lenses. If the physician has optimized the ocular surface with punctal plugs or topical cyclosporine, then his dry eye may be a little better as well — all factors that can help the patient see better.

One of the biggest innovations for our keratoconus patients has been the introduction of the scleral contact lens, which vaults over the center of the cornea but does not touch the cornea. Patients are more comfortable, and they achieve high-quality vision. Comparing rigid gas permeable lenses to scleral lenses, I do have one concern about rigid gas permeables. One of the biggest risk factors for progression of keratoconus is eye rubbing, which is essentially compression of the cornea. I wonder whether the compression of the cornea with each blink with a rigid gas permeable contact lens in any way affects the stability of the cornea.

Waring: There may be an orthokeratologic component to that as well, which we observed in some of our original studies. When we got the patients back into a rigid lens, not a scleral lens, early after treatment, we found that the lens may act almost like a cast as the cornea remolds. Even though it may not have a lasting effect, there may be some temporary effect, so that seemed to get a slightly improved benefit from an orthokeratologic standpoint. And, certainly, there is the idea that you get a more normalized cornea so patients are more likely to be able to wear their contact lens in the first place. In our experience, we had the best luck with hybrid lenses.

Rubinfeld: If we are going to say that someone has had a cross-linking treatment failure, then that patient has subjective and objective vision loss?

Cummings: That makes sense. It is very subjective, and it depends on your contact lens expert. We have a fantastic contact lens person who tells us who is doing well and who is not.

Rubinfeld: Of seven cases in my clinic that were failures, all were associated with eye rubbing.

Look for a discussion of epi-on vs. epi-off in part 2 of this round table in an upcoming issue of OSN.