DECISIVE FACTORS IN THE CLINICAL RESPONSE

The speed of disease progression in keratoconus can vary greatly. Some of the risk factors linked to changes in corneal biomechanics and topography include early age, ^, thyroid hormone variations, ^, eye-rubbing habits, pregnancy, ^, and the use of medications such as estrogen regulators. Moreover, it is hypothesized that such factors may also alter the ability of CXL to stabilize keratoconus over time. It is believed, however, that two elements would be most relevant in this regard: the age of the patient at the time of performing the CXL ^, and the collagen turnover time.

Cross-linking is considered a natural process that occurs with age not only in the cornea but also in other ocular structures, such as the lens. In the cornea, the diameter of the corneal fibrils increases over time because of age-dependent glycosylation. This means that the cornea becomes progressively stiffer, broadly decreasing the likelihood of keratoconus progression as people grow older. A further relevant point that may be related to the long-term stability of CXL is the CXL-induced delay in corneal collagen turnover. Collagen turnover usually declines with age. Estimates of collagen turnover time vary from 3 to 7 years, but this process could be delayed by at least 12 to 24 months by CXL due to CXL-induced changes such as increased resistance of corneal collagen and matrix components to enzymatic degradation.

EMERGING PROTOCOLS

Since the introduction and adoption of the standard-of-care Dresden protocol, a number of other approaches to cross-linking the cornea have emerged. The Dresden protocol requires 30 minutes of UV-A irradiation, and the first modifications made were to try and make the total procedure faster—in other words, accelerated CXL. In theory, this is supported by the Bunsen-Roscoe law of reciprocity, which states that an equivalent photochemical effect could be achieved with any combination of intensity and illumination time, if one maintains the total fluence. This would suggest that 30 minutes of 3 mW/cm ² illumination for 30 minutes would give an equivalent effect to 30 mW/cm ² for 10 minutes. However, the Bunsen-Roscoe law fails to hold true, as decreasing effects on corneal strength are seen as illumination intensity increases and intensity time decreases, and our group was the first to show that the photochemical cross-linking reaction also requires another molecule that has a restricted availability within the corneal stroma: oxygen.

To better understand the effect of accelerated CXL on the cornea, the biomechanical effect of cross-linking under different combinations of irradiation times and UV intensity (with the same total fluence) was studied. The stiffening effect was significantly lower when the procedure was accelerated, with 9 mW/cm ² energy being delivered for 10 minutes resulting in a significantly lower stiffening effect than that achieved with the Dresden protocol, and this decrease in effect was even more pronounced when 18 mW/cm ² of UV-A energy was applied for 5 minutes. Similarly, more distinct changes in gene transcription were observed with Dresden protocol cross-linking compared with accelerated CXL protocols. This tendency for accelerated CXL to produce a weaker effect than the Dresden protocol has also been observed clinically, with mixed results being observed when using faster CXL protocols.

The dependence of the CXL photochemical reaction on oxygen may also explain some of the more unsatisfactory results of newer cross-linking protocols under development or already in clinical use, such as CXL with epithelium-preserving techniques (“epi-on” CXL). ^{, ,} The epithelium is removed in the Dresden protocol because it forms an effective barrier for riboflavin penetration into the collagen-rich corneal stroma where the cross-linking reaction needs to take place. Even if an epi-on technique managed to provide adequate riboflavin saturation of the stroma, leaving the epithelium in place results in not only a ~20% attenuation of UV energy that reaches the stroma, but the epithelium also limits the diffusion of oxygen into the stroma. This is in part because the corneal epithelium has a high oxygen requirement, consuming approximately10 times more oxygen than the stroma. For this reason, even if transepithelial techniques could improve the diffusion of riboflavin into the stroma, they cannot influence the speed of oxygen diffusion. These factors may explain why numerous transepithelial CXL protocols evaluated in the past have not been able to halt keratoconus progression as effectively as the Dresden protocol.

ROOM FOR IMPROVEMENT

The Dresden protocol was developed with an abundance of caution. UV-A energy can damage corneal endothelial cells, so the protocol was designed with wide safety margins to avoid this; hence the UV-irradiation settings and the requirement for a minimum corneal thickness of 400  µm. However, as more is known about the safety profile of CXL, as well as what is involved in the photochemical reaction, some of those parameters are being altered to improve the efficacy of the accelerated and epithelial-sparing cross-linking protocols. Although the studies examining this are in their infancy, and we still lack an ideal clinical metric for examining the true efficacy of CXL (we currently rely on a number of relatively suboptimal measurements such as refraction, corneal topography, and a number of air puff tonometry corneal deflection-related measurements), we are beginning to see some promising initial results.

Accelerated protocols that deliver a higher total fluence are now achieving functional outcomes that are close to those of the Dresden protocol. Pulsed-light protocols have been developed that enable oxygen to diffuse into the stroma during the UV-A off-period (rather than being consumed in the on-period) and pilot studies of these enhanced-fluence pulsed-light iontophoresis protocols have shown—with the caveat of a currently short follow-up period—that they may be able to overcome relevant limitations of transepithelial protocols and achieve effective keratoconus stabilization rates. At present, however, more studies are needed to evaluate and validate such results in both the medium and long term.

Another approach, customized CXL, uses different energies applied unevenly in different locations on the same cornea and has displayed (again, over a short follow-up period) a stronger corneal regularization effect than standard cross-linking protocols.

Important advances have also been made in the treatment of thin corneas. ^, A new, individualized CXL modality has been introduced by our group as a way to cross-link each cornea based on the individual. The “sub400” experimental protocol takes into account the diffusion of oxygen and applies different energies according to each patient’s individual intraoperative corneal thickness. ^{, ,} This has allowed us to treat corneas below 400 µm without using methods to artificially increase the corneal thickness such as swelling with hyperosmolar riboflavin or contact lenses (which introduce another factor that can increase the variability of outcomes).

SHORT- AND LONG-TERM RESULTS

A longitudinal evaluation of the first 130 patients with keratoconus who underwent Dresden protocol CXL found that 1 year after cross-linking, the anterior keratometry (K) of the corneal apex decreased by a mean of 2.68 D in 62% of the eyes, and stability (a variation ≤0.50 D) was found in 17% of the eyes. Fig. 27.1 exemplifies typical corneal remodeling after just over a year in a patient after standard epithelium-off CXL. Similarly, the maximum K value decreased by an average of 1.46 D in 56% of the eyes, whereas 30% of eyes remained stable topographically after 1 year of follow-up. Interestingly, even with some patients being lost to follow-up, the same series showed a flattening in K apex value by an average of 4.84 D after 3 years in 78% of the 33 eyes analyzed.

Image of the anterior sagittal curvature using a Scheimpflug imaging device, showing a typical effect after performing standard epithelium-off corneal cross-linking (CXL) . The left image represents corneal curvature after CXL; the central image represents preoperative topography; and the image on the right shows differential map between both images, displaying central flattening of 2.6 D after 1.3 years.

Until 2020, only one study had been published that encompassed clinical results up to 10 years after CXL using standard epithelium-off Dresden protocol in adults. The maximum K, minimum K, and K apex values were significantly reduced 10 years after and corrected distance visual acuity (CDVA) significantly improved by an average of 0.14 logMAR. This analysis was performed in 34 eyes, and only one eye (3%) had permanent corneal scar with reduced visual acuity. Although other persistent anterior stromal opacities were found in 13 eyes (38%), these did not affect best-corrected visual acuity (BCVA). Of the entire series, two eyes (6.25%) failed the treatment and required a repeat CXL procedure because of an increase in keratometry values, at 5 and 10 years after the initial procedure.

An unpublished joint study data from the ELZA Institute (Dietikon, Switzerland), the IROC Institute (Zurich, Switzerland), and the Center for Applied Biotechnology and Molecular Medicine at the University of Zurich also showed that topographic readings tend to decrease significantly over time, and despite such changes, corrected visual acuity remained stable in a 10-year follow-up period. In this retrospective series, preliminary data show a failure rate of around 10% at 10 years after the procedure.

As with adults, standard epithelium-off CXL has also been shown to be safe and effective in providing keratoconus stability in pediatric patients. As expected, the rate of progression in pediatric patients was up to 24% in 10 years, showing that although CXL can slow down KC progression and improve functional performance, age is a decisive factor and younger patients should be followed closely.

EARLY POSTOPERATIVE

A mild haze (that does not affect vision) usually appears after CXL. For this reason, there is debate as to whether haze should be viewed as a normal finding in CXL or considered as a complication that could potentially cause loss of visual quality.

The primary safety considerations with CXL focus on the corneal endothelium and stem cells present in the limbus, lens, and retina. Originally, one of the major concerns was potential endothelial damage caused by UV irradiation: 400 μm of corneal minimum thickness was required to protect eye structures, as a result of riboflavin shielding. A threshold of 0.35 mW/cm² of corneal endothelium UV irradiation would lead to cell death by apoptosis, putting corneal homeostasis and transparency at stake. Although this limit might be overestimated according to new evaluations, most of the few reported cases of endothelial dysfunction after CXL arose because surgeons failed to respect the protocol to measure corneal thickness immediately before UV irradiation. In both pediatric and adult patients, long-term follow-up studies have not reported endothelial cell changes or failures. ^,

Richoz et al. investigated whether the amount of UV irradiation delivered in CXL can affect corneal limbal stem cells. They showed that irradiating with double the standard fluence (10.8 J/cm ² ) on the limbus, altered neither the regenerative capacity of the limbal epithelial cells nor the expression pattern of the putative stem cell marker p63, suggesting that CXL can be used safely, even in eccentric irradiation situations as in PMD.

Another relevant point concerns epithelial cell removal. Although currently highly effective, the CXL treatment protocol is invasive and may be painful, since the most effective treatment occurs when the epithelium is removed. Photoactivated chromophore for keratitis corneal cross-linking (PACK-CXL) has been used successfully for the treatment of keratitis and has the ability to reduce the microbial load, depending on the total fluence used, to a lesser or greater degree. Therefore at the end of each CXL procedure, the irradiated corneal surface has a lower microbial load, which would potentially protect the surface from infections acquired at this point in time. However, the persistence of epithelial defects poses a postoperative risk of contracting a corneal infection in cases where the open surface is not properly handled. ^,

Other unexpected responses to CXL have also been reported. As a result of cellular immunity to staphylococcal antigens, peripheral sterile infiltrates have been reported as complications after CXL. ^, Finally, herpes reactivation could be triggered by UV light even in individuals with no apparent history of clinical herpes virus ocular infections.

LATE POSTOPERATIVE

There are distinctive forms of haze after CXL. As mentioned earlier, a transient mild haze that does not affect visual quality is observed in most patients. However, haze can also become permanent. In these cases, this is usually associated with corneal flattening.

In a retrospective assessment of 127 patients, approximately 9% developed clinically significant persistent haze at 12 months of follow-up. Advanced keratoconus with decreased corneal thickness, higher keratometry values, or a reticular pattern of stromal microstriae (as seen by in vivo confocal microscopy) are considered risk factors for the development of late and permanent haze. ^{, ,} Interestingly, eyes with haze associated with massive corneal remodeling may not have visual acuity negatively affected. On the contrary, visual acuity can improve, even in eyes with moderate haze and flattening of up to almost 10 D after CXL. Corneal flattening in highly aberrated corneas preoperatively may be potentially more beneficial than the formation of the haze itself, and despite the haze, some patients display improvements in their BCVA.

In a minority of cases, corneal flattening can occur in the early years after CXL, and this usually stabilizes after 1 to 3 years. Nevertheless, cases of continuous flattening have recently been reported—even after up to 12 years after CXL. Interestingly, such cases had no stromal opacity other than transient stromal haze, hence this process differs from the standard long-term behavior of corneas after CXL.

Ectasia progression after treatment is considered to be a CXL failure and could be viewed as a long-term potential complication; however, because of the phenomenon of corneal collagen turnover one could also interpret this as a spontaneous process that occurs once keratocytes have been renewed after many years have passed. A failure rate of 7.6% was reported after 12 months in a study that prospectively evaluated 117 eyes. This study identified that high preoperative maximum K was a significant risk factor for failure. For context, long-term follow-up studies available report a failure rate between 6.25% and 10% after 10 years of CXL using the standard epithelium-off protocol. ^,

Questions can be raised as to when to evaluate whether re–cross-linking would be necessary and which protocol to use, especially owing not only to long-term reactivation of progression, but also to immediate treatment failures occurring as early as 6 months after the procedure. ^, Although there is no consensus, 6 months after CXL seems to be an adequate time to reconsider a new procedure, in the rare event of a treatment failure. Moreover, repeating CXL very early after the first procedure is unlikely to increase corneal stiffness any further, as has been shown in in vivo studies. Finally, it is interesting to mention that there is an absence of complete agreement between corneal imaging devices, especially in highly aberrated keratoconic corneas. There remains the lack of an ideal metric for comparing and monitoring patients using topographers and tomographers available on the market today, ^, although technologies like optical coherence tomography-based devices may help better assess the success in the future.