We are often asked which CXL treatment methods we prefer, most frequently the question is whether we perform “epi-on” or “epi-off” (Dresden protocol) cross-linking.
As technology has advanced, our answer has evolved from “epi-off” in most cases, to “epi-on” in most cases. Here’s why.
The first step of classical cross-linking requires the removal of the outermost layer of the cornea, the corneal epithelium, so that riboflavin (vitamin B₂) can penetrate the cornea correctly. The success rate of this “epi(thelium)-off” protocol is extremely high, well over 90%. A relative disadvantage is that the cornea has to heal again. In the first few days there are inconveniences such as a burning and biting sensation in the eyes. Of course you also have to make sure that no infection occurs. However, with professional surgery, patient education and meticulous follow-up, this risk is extremely low. We have not seen a single case of post-CXL corneal infection in 10 years.
The idea of epi-on cross-linking came up because it does have theoretical advantages: less postoperative burning sensation of the eyes and a reduced risk of infection. The reason why ELZA did not adopt epi-on CXL in the past was that the epi-on success rate was around 60–70% – well below the 90% success rate of epi-off CXL. In 2013, our research group discovered that in addition to vitamin B₂ and UV light, the presence of oxygen is another essential factor for successful cross-linking. However, using this knowledge, ELZA’s surgeons and researchers worked to develop better formulations of riboflavin that penetrate the epithelium, and smarter cross-linking UV irradiation protocols that have helped deliver epi-on CXL success rates that approach those of traditional epi-off CXL protocols.
Our choice: epi-on
We reserve epi-off CXL for only the most aggressive cases of corneal ectasias where the absolute maximum cross-linking effect is necessary. For all other cases, the reduced pain and risk of post-operative infection are in our view, acceptable trade-offs for a marginal difference in cross-linking efficacy.