Sharma and associates recently published an interesting retrospective case series of patients who manifested persistent corneal edema after collagen cross-linking (CXL) for keratoconus. The authors analyzed the clinical data of 350 patients with keratoconus who underwent CXL and identified 10 patients with postoperative corneal edema (2.9%). Corneal edema improved in 4 and resolved in 1 patient, whereas 5 patients underwent penetrating keratoplasty for visual restoration. Based on these findings, the authors underscore the potential risk for endothelial damage and consequent corneal edema after CXL, although they recognize that CXL is a procedure demonstrating a strong overall safety profile.

Considering the importance of the reported clinical data, we would like to highlight a number of aspects, which merit further attention. First, the authors applied the Dresden protocol for CXL procedure using isotonic riboflavin 0.1% drops (20% Dextrane; IROC Innocross AG, Zug, Switzerland). CLX was performed after removal of corneal epithelium and subsequent measurement of central corneal thickness to ensure that central corneal thickness remained at more than 400 μm in all cases. However, corneal thickness (CT) was measured neither at the end of riboflavin solution instillation nor during ultraviolet A irradiation, although it is well established that CT may decrease significantly during CXL in which corneal epithelium is removed because of corneal dehydration from epithelial debridement. Intra-CXL corneal thinning is transient and reversible, but it may expose the corneal endothelium to severe danger if a minimal preoperative stromal thickness of 330 μm is not respected. Therefore, careful evaluation of CT after riboflavin solution instillation, as well as repeated intratreatment CT measurements with administration of hypotonic riboflavin solution for CT values of less than 330 μm, should be part of the routine CXL protocol. This was not the case in this study.

Second, only 1 case report on endothelial damage after CXL has been published so far, whereas dozens of studies have reported the absence of irradiation damage to the endothelium in the past years. This case report also did not measure corneal thickness during surgery, similar to the cases reported here.

Third, another potential source of endothelial damage that often is underestimated is associated with the imprecise estimation of preoperative and intraoperative corneal pachymetry. The measured CT, as evaluated by portable ultrasound devices, does not always correspond to the thinnest corneal point, which is the crucial pachymetric reading determining the decision tree in a CXL procedure. In our opinion, it is of paramount importance that physicians perform preoperative and intraoperative corneal pachymetry with simultaneous evaluation of the corresponding corneal topographic data to measure the thinnest point corneal pachymetry in the area of the cone.

Finally, frequent equipment calibration also is essential for avoidance of ultraviolet light toxicity resulting from inadvertent delivery of excessive energy. In conclusion, we believe that meticulous adherence to the procedural precautions mentioned above would have prevented the endothelial damage reported here.