Current reports on complications after corneal cross-linking surgery are usually related to corneal infections that occur during the postoperative epithelial healing period. There are very few reports of postoperative endothelial damage, most of which are concentrated on endothelial damage after CXL treatment for keratoconus[4–7]. In our case report, this case shows that complications of reversible endothelial damage after half-femtosecond combined with rapid CXL reinforcement surgery are very rare. It is well known that UVA radiation has potential cytotoxicity, leading to the release of singlet oxygen, superoxide, and hydrogen peroxide in endothelial cells, thus causing apoptosis[8]. It has been reported that the cytotoxic level of UVA radiation to porcine keratinocytes is 4mW/square centimeter. In the evaluation of the cytotoxic effect of UVA on human endothelial cells in vitro corneal cross-linking, it has been proven that human endothelial cells are much more resistant to the cytotoxic effects of UVA than animal endothelial cells[9]. In addition, it has been proven that the absorption of riboflavin by the corneal stroma can increase the UVA absorption coefficient, limiting the irradiance of ultraviolet light through the cornea, reducing endothelial damage after corneal cross-linking[10, 11].
Studies have shown[12] that for patients with keratoconus, the probability of endothelial damage is relatively low when the corneal thickness after treatment is greater than 400mm. However, there are no reports on how much residual corneal thickness is a safe range for corneal cross-linking during corneal refractive surgery combined with corneal cross-linking. In this report, the preoperative central corneal thickness was 518µm for the right eye and 528µm for the left eye. After the laser ablation of the cornea and repositioning of the corneal flap for UVA irradiation, the thinnest corneal thicknesses were 398µm and 384µm, respectively. Considering the patient's history (excluding keratoconus and congenital corneal malnutrition and other corneal diseases), no endothelial damage was observed preoperatively, and the endothelial damage three days postoperatively was mainly due to the toxic effect of CXL surgery on the corneal endothelium. The formation of small pit-like lesions on the endothelial surface and dark areas on the endothelium may be related to the corneal thickness being lower than 400µm during UVA irradiation. Two months post-surgery, the corneal endothelial cell count returned to preoperative levels.
As is well known, endothelial cells do not have the ability to divide and proliferate, and previous reports have also shown an increase in corneal endothelial cells before and after CXL surgery, which may be due to sample errors in data collection before and after surgery. In addition, it may also be due to cell rearrangement, hiding the actual damage to endothelial cells[13]. Amanie[14] and others have shown that the endothelial cells with keratoconus significantly decreased (6.7%) three months after CXL surgery, gradually improved at six months, and decreased by 0.89% at twelve months. On the other hand, the loss of corneal endothelial cells after LASIK combined with CXL surgery is less (2.53% at three months), gradually improving and approaching preoperative baseline levels at twelve months. However, there are differences in the UVA irradiance parameters in the current literature studies, and the results of corneal endothelial damage are different and even contradictory. Cingü etal[15] found that after accelerated CXL (18mW/cm2 continuous for 5 min), endothelial cell counts and morphology changed significantly, with obvious changes observed at the first week and first month, and endothelial cell count and morphology returned to baseline values at three months. In contrast, some studies believe that accelerated CXL has little impact on ECD[16, 17], with studies using accelerated pulsed high-flux CXL (30mW/cm2, 8-minute pulse mode 1 second on/1 second off, total energy of 7.2J/cm2)[18] finding no significant change in average endothelial cell coefficient of variation after six months of follow-up, no significant change in the percentage of hexagonal cells, and no significant change in the number of endothelial cells. Another comparative study of 21 patients with single-eye standard CXL (3mW/cm2 continuous for 30 min) and contralateral accelerated CXL (7mW/cm2 continuous for 15 min) showed that rapid CXL and standard CXL had similar safety for corneal endothelium[19]. The inconsistent research data observations in the above studies are mainly due to the use of different UVA energies in these studies. So far, there is no consensus on the use of accelerated CXL, which requires more randomized clinical trials to evaluate different CXL methods and energy requirements.
So, is the decrease in postoperative corneal endothelial cell count and morphological abnormalities related to the thinning of the cornea after corneal cross-linking? In studies on standard CXL[20, 21] or accelerated CXL[22, 23], it has been reported that the thinnest corneal thickness in both groups of patients with keratoconus and post-LASIK ectasia was significantly reduced at three months postoperatively, then gradually increased, and continued to be significantly thinner at one year. The decrease in corneal thickness[24] may be related to structural changes caused by the rearrangement of corneal lamellae, changes in corneal collagen fibrils, apoptosis of corneal cells, etc., and corneal ischemia is also one of the possible reasons[25]. Although corneal thickness was significantly reduced, there was no direct statistical significance between the loss of endothelial cells and the thinnest layer of corneal thickness. This indicates that in addition to corneal thickness, there are other factors that cause changes in corneal endothelial cell count after CXL: higher UVA intensity radiation can cause nerve plexus damage, thereby disrupting the performance of the endothelial pump. The cornea is one of the organs with the highest density of peripheral nerve endings in the body. Sensory nerve fibers branch out from the long ciliary nerve, pass through the anterior elastic layer, form a subepithelial nerve plexus under the epithelium, and release neurotransmitters including acetylcholine, catecholamines, substance P, and calcitonin gene-related peptide, etc., which promote the transmission of corneal endothelial signals through the Na/K-ATPase pump and play an auxiliary role[26]. Higher UVA intensity radiation can cause damage to the subepithelial nerve plexus of the cornea, which may be the main reason for the morphological changes in corneal endothelial cells after CXL. However, such changes in corneal endothelium do not lead to complete loss or death, because when undamaged endothelial cells slide in and replace damaged endothelial cells, and as the function of the subepithelial nerve plexus of the cornea recovers, corneal endothelial cells recover. This theory is consistent with the disease course changes reported in this case (Fig. 4). We speculate that the recovery time of corneal endothelial cell function after half-femtosecond combined with CXL reinforcement surgery may be parallel to the healing response after half-femtosecond corneal flap cutting surgery. Through this case report, the current outcome of corneal endothelium is still a potential major consequence of the surgery, which deserves further long-term randomized controlled studies.
In summary, although CXL has been proven to be safe and effective, patients still face risks. There is evidence that accelerated CXL has a significant impact on corneal endothelium, with excessive UVA energy during CXL surgery being the main cause of corneal endothelial damage. Insufficient riboflavin injection may also limit the inhibition of riboflavin, thus increasing ultraviolet irradiance, cytotoxicity, and endothelial levels. To avoid endothelial damage, it is necessary to strictly implement the inclusion criteria for corneal cross-linking surgery patients in clinical work, and case selection and follow-up should be more cautious.