Plentiful evidence supports the utility of CXL for slowing or stopping progression of KC [103, 115–127]. However, much remains to be learned about the relationship of delivery technique to the safety, efficacy, and both short and longer term stability of CXL. Riboflavin concentration, corneal thickness, intensity and duration of UV light exposure (both total exposure time and pulsed versus nonpulsed light), as well as the presence or absence of epithelium are all variables that must be considered when performing CXL.

Non-Invasive, Epi-On CXL, if done correctly, offers the benefits of CXL (slowing/stopping progression of KC, improvement of corneal shape, improved vision) with very minimal risk of vision loss. Nawaz et al [128] evaluated Epi-On CXL compared to Epi-Off CXL and found similar CDVA and topographic changes between the two groups. Two of the patients (10 %) in the Epi-Off CXL group, however, developed persistent stromal haze. Not surprisingly, Epi-On CXL group patients reported superior comfort post-procedure [128]. In addition to causing significant postoperative pain [128, 129], removal of the corneal epithelium is also associated with a longer recovery time before resuming contact lens usage and activities of daily living. Following Epi-On CXL, patients typically return to contact lenses within a week as compared to 1 month after Epi-Off CXL. While there can be discomfort during the first 24 h after Epi-On CXL, the majority of patients are fully functional and return to baseline levels of visual function after 1–2 days postoperatively. Additionally, as noted above, Epi-Off CXL increases the risk of complications, including pain, corneal haze, melt, infection, and endothelial decompensation [130–134]. As such, there is great interest and enthusiasm amongst both patients and physicians for an Epi-On CXL approach.

Differences in riboflavin instillation techniques, including riboflavin loading time and technique, concentration of riboflavin, and parameters related to the light source may result in varied efficacy from either Epi-On or Epi-Off CXL. More precise ways to measure the concentration of riboflavin in the cornea prior to and during UV light application may be necessary to titrate the predictability of the location and magnitude of corneal strengthening for either CXL approach.

There are several unique technical considerations that are important when performing Epi-On CXL. Because intact epithelium can slow penetration of riboflavin into the corneal stroma, adequate stromal saturation of riboflavin through intact epithelium requires different techniques than with Epi-Off CXL. Loading times for Epi-On CXL can vary, depending on the concentration, pH and osmolarity of the riboflavin formulation and the presence or absence of excipient substances to enhance epithelial permeability, and may range from 15 min to 3 h. The CXLUSA Study Group (comprised of 17 centers nationwide across the USA) uses a patent pending proprietary riboflavin formulation and delivery system. The riboflavin formulation and non-iontophoretic transepitleial stromal loading system have been optimized for rapid, consistent, proprietary, patent pending penetration of corneal epithelium to achieve homogeneous transepithelial stromal loading. The unique proprietary riboflavin formulation has been shown in independent laboratory animal and human clinical trials to be unique in consistently achieving transepithelial stromal riboflavin loading [the 11th Intl CXL Congress, USA 2015; and Stulting’s Binkhorst Lecture, ASCRS 2016]. The average loading time using proprietary riboflavin 0.5 % with the proprietary loading system takes 15–20 min. Prior to these developments, transepithelial, consistent, homogeneous and reliable stromal loading using isotonic 0.1 % riboflavin, both with and without without numerous loading delivery systems used by investigators in the CXLUSA study took on average 40 min to 2 h. Other variables affecting stromal loading may be corneal thickness, steepness of the cone and any haze or scarring. An incomplete load due to a steep cone may require peripheral or sectorial loading. Other techniques, which have been described in the literature, enhancing penetration through intact epithelium include the use of topical anesthetics, iontophoresis [135], and benzalkonium chloride-EDTA (BAC-EDTA) riboflavin-UVA Epi-On [136]. The use of permeability enhancers may have an additive effect on epithelial riboflavin permeability and thereby increase riboflavin penetration into the corneal stroma [137, 138]. Some of these agents such as BAK-EDTA have been shown to act chemically as singlet oxygen quenchers thus potentially interfering with CXL [139].

The keys to adequate riboflavin loading prior to UV light application include homogeneity, or how even or uneven (patchy) the loading is, concentration (or density) of the loading. If patchy or incomplete loading is seen (Figs. 4.2 and 4.3, areas of poor loading seen by arrows), additional loading should be performed. An example of a homogeneous, adequately concentrated corneal stromal loading is seen in Fig. 4.4. Additionally, time should be allotted to allow clearance of riboflavin from the epithelium before proceeding to UV in order to minimize UV absorption by the epithelium (generally about 5 min is sufficient). Figure 4.5 illustrates adequate stromal loading with absence of riboflavin in the epithelium as evidenced by lack of autofluorescence in the presence of cobalt blue light. Given that the CXL treatment is targeted to stromal tissue, proceeding with UVA application is best performed when the corneal epithelium is clear, non-disrupted and free of visible riboflavin.

As with any CXL approach, UV light can be applied continuously or “fractionated,” whereby the light is “pulsed” on and off. The pulsed-UV application is thought to allow oxygen levels in the cornea to recover, which should allow for more effective CXL [140, 141]. The irradiance (typically 3–4 mW) and diameter (9–12 mm) of the light are additional variables that likely affect outcome.

The CXLUSA Study Group is evaluating both Epi-On and Epi-Off CXL in nonrandomized cohorts. Eligible patients in the study must have one of the following conditions: forme fruste keratoconus (FFKC), post-LASIK ectasia, pellucid marginal degeneration (PMD), or fluctuating vision after radial keratotomy. Trattler and Rubinfeld (ISRS Refractive Surgery Subspecialty Day, November 13, 2015) evaluated UCVA, CDVA, and Kmax at 6 months, 1 year, and 2 years in 381 eyes with average preoperative Kmax of 63.5 D and average thinnest corneal thickness (TCT) of 408 μm with Epi-On CXL using a proprietary system. At 6 months (n = 206), 1 year (n = 153), and 2 years (n = 62), average improvement in UCVA was 1.71, 2.26, and 2.65 lines, respectively. CDVA improved 0.68, 0.79, and 1.18 lines, at 6 months, 1 year, and 2 years, respectively. There was noted to be 0.75 D, 1.18 D, and 1.23 D of flattening of Kmax at 6 months, 1 year, and 2 years, respectively.

Following initial enthusiastic reports of efficacy with Epi-On CXL, long-term follow-up showed loss of effect in 19 % [142] and 23 % [143] of eyes between year 1 and year 2 postoperatively. Both of these studies utilized a commercially available riboflavin formulation containing tromethamine and EDTA to promote absorption, continuous UVA exposure at 3 mW/cm2, and continued application of riboflavin during UVA exposure. It is also noteworthy that these protocols did not include verification of riboflavin saturation by slit lamp examination before light exposure.

The CXLUSA Epi-ON protocol differs from that utilized in the studies reported by Caporossi’s and Soeters’ groups in several ways. First of all, the CXLUSA protocol requires slit lamp verification of adequate corneal stromal riboflavin saturation before light exposure, with an extended treatment time if saturation is not adequate. Second, the patent pending CXLUSA riboflavin formulation is completely different from the commercially available one that has been reported to lose effect long-term. Third, patented sponges are utilized to enhance riboflavin absorption under the CXL-USA protocol. Finally, the UVA light utilized by CXLUSA is more intense (4 mW/cm² vs 3 mW/cm²) and pulsed to allow deeper oxygen penetration into the corneal stroma.

Interim 2-year analysis of Epi-On CXLUSA outcomes was recently reported to produce a 1-1 ½ Snellen line improvement in CDVA that is stable from year 1 to year 2 postoperatively, as shown in Figure 4.6
[Stulting’s Binkhorst Lecture, ASCRS 2016].

Corneal hysteresis (CH) is another potential measure of corneal biomechanics that can be assessed after Epi-On CXL. De Bernardo et al [144] showed stabilization of patients with progressive KC at 6 months after Epi-On CXL with minimal change in CH, while Lombardo et al [145] showed a biomechanical strengthening effect on donor globe eyes with Epi-On CXL. Torricelli et al [136] showed that benzalkonium chloride-EDTA (BAC-EDTA) riboflavin-UVA Epi-On CXL statistically increased biomechanical corneal stiffening compared to the standard Epi-Off CXL treatment in a rabbit model.

Due to its atraumatic nature, Epi-On CXL is well tolerated by all age groups and varying degrees of disease and may a better option in specific type of cases. Young patients, including children as young as 9 years old, tolerate the procedure well with fast recovery. Young patients, however, need to be monitored carefully for progression, given risk of progression, and repeat Epi-On CXL or Epi-Off CXL treatments may need to be considered earlier in this population. Older patients (greater than age 35) as well as corneas with steep average keratometry (Kavg) can also benefit, as they may be slower to reepithelialize with Epi-Off CXL compared to other patients without such attributes. A study done by Koller et al [49] showed that preoperative Kmax greater than 58 D increased risk for continued progression after Epi-Off CXL, so these eyes specifically may be at higher risk for complications given slower reepithelialization. For this reason, in addition to the other types of cases listed here, these steep eyes (Kmax > 58) may be good candidates for an initial Epi-On CXL treatment. Additionally, eyes of patients of age greater than 35 undergoing Epi-Off CXL were found to have a clinically significant increased risk for complications after surgery (eyes losing 2 or more Snellen lines) [49]. Moreover, eyes that have undergone prior keratorefractive surgery including LASIK, RK, AK, or ISCR may be able to proceed more safely with Epi-On CXL. Including all the standard risks with Epi-Off CXL (mentioned previously), there may be increased risk for additional complications in these eyes such as LASIK flap-related complications including DLK, incision gaping post-RK or AK, or risk of melt over an ISCR.

Repeat CXL treatments may be necessary for stabilization of corneal ectasia in some cases, and variables such as loading time, UV light application, and strength may need to be varied depending on the severity of the condition. To date, there is no published results of repeat Epi-On treatments, but a secondary treatment could be considered as an option for eyes where added effect may be necessary, and in our experience do not increase the risk of posttreatment complications. Given the safety of Epi-On CXL, repeat treatments may become more mainstream if the technique gains popularity.

Despite the promise of Epi-On CXL, the literature supporting its use for treatment of corneal ectasia remains limited. There is no debate as to the superiority of Epi-On CXL from a safety perspective, but the data supporting efficacy and stability comparable to standard Epi-Off CXL are, to date, sparse. A study by Kocak et al [146] showed in a comparison study of Epi-On CXL to Epi-Off CXL that Epi-On CXL did not effectively halt progression of KC while Epi-Off CXL not only halted progression but improved corneal shape parameters [146]. Other studies also report KC parameter instability [142–149], particularly in the pediatric population, where additional CXL was often required [142]. It is the opinion of these authors that many of the published studies yielding negative or mixed outcomes likely relate to inadequate riboflavin loading of corneal stroma and/or application of UV light without allowing riboflavin to clear from the corneal epithelium first. In fact, clinical experience has led these authors and many other investigators from the CXL-USA study group to conclude that the therapeutic index of Epi-On CXL is favorable enough to recommend it as first line treatment of keratectasia, followed by either repeat Epi-On or Epi-Off CXL if further progression is documented. Nevertheless, given the paucity of well-controlled long-term studies examining the stability of visual acuity, corneal shape, and biomechanics after Epi-On CXL, future research will be of critical importance to elicit the optimal clinical indications and treatment parameters.

One very promising future application of Epi-On CXL is the ability to administer selective treatment based on corneal topography and biomechanics to achieve controlled corneal shape modifications that can predictably improve corneal strength and optics simultaneously. It is proposed that the biomechanical change with ectatic corneas is focal in nature, rather than a uniform generalized weakening, and that the focal reduction in elastic modulus precipitates a cycle of biomechanical decompensation that is driven by asymmetry in these biomechanical properties [150]. Seven et al. [151] showed clinically significant reductions in astigmatism are possible with patterned CXL, and the magnitude of the effect was dependent on patient-specific geometry, effective stiffening pattern, and treatment orientation. Other novel approaches to measuring biomechanical changes from CXL will also be useful in our understanding of the effects of CXL or ISCR. Sinha et al [152] describe a unique module which uses topography to determine a patient-specific finite element model of an ectatic eye assuming hyperelastic properties. In this model [129], human eyes were studied and noted to have an increase in corneal elastic module after CXL, which could be correlated to changes in topography findings. This study emphasized the importance of patient-specific approaches including a possible topography-guided approach to CXL. Kanellopoulos and Asimellis [153] discuss a clinical novel application of topographically-customized CXL (both Epi-On and Epi-Off CXL) to attempt to achieve hyperopic refractive changes [153]. Such approaches are likely to succeed only once the significant variables of epithelial healing and resultant corneal stromal remodeling are no longer part of postoperative healing.

Since the initial clinical reports on the efficacy of CXL for progressive KC and postoperative corneal ectasia [118, 120], tremendous international interest and research has surrounded this topic. To date there have been numerous proof-of-concept, basic science, and clinical reports from the Standard Protocol (commonly referred to as the Dresden Protocol) [118, 154–161].

The Standard Protocol has clear efficacy and an acceptable safety profile. However, there are drawbacks to this protocol that have stimulated interest in modifications to either improve upon safety or efficacy, or to reduce treatment times, and ultimately to avoid epithelial removal for optimal treatments. Among these drawbacks, treatment time is a significant clinical issue. The Standard Protocol requires 30 min or more of riboflavin soak time followed by 30 min of treatment. If this treatment time could be reduced it could benefit clinical practice and patient tolerance of the procedure. This is especially important in younger patient population, who stand to gain the most from CXL.

The Bunson-Roscoe law of reciprocity, discussed below, implies that alterations to the Standard Protocol should yield equivalent results as long as the same total dose or irradiance is delivered. However, in biologic systems other factors play a role and may alter this otherwise fundamental principle.

In order to better understand the current state of CXL protocols, an examination of the literature must be undertaken to establish proof of concept, anatomic features of Standard and Accelerated protocols, and comparative clinical outcomes.

CXL with riboflavin (a vitamin B2 molecule) and ultraviolet-A may nowadays be considered as an established option [183] for the management of progressive KC [184]. This is supported by more than 10 years of experience following the introduction of the technique by the Dresden Protocol [118, 185]. The procedure increases corneal resistance and inhibits progression of the ectatic disorder [186], which is applicable not only in KC, but also in the treatment of PMD [187] and induced post-LASIK ectasia [120].

CXL, although initially employed to arrest keratectasia, has recently been increasingly evaluated in regard to its optimal refractive outcome that may significantly affect the quality of everyday life. Several adjuvant treatments may combine with CXL to offer a far wider reach of options. TG-PRK, TE-PTK, ISCR implantation, and PIOL implantation are many of the refractive options that may be combined with CXL.

Topo-guided excimer ablation combined with CXL treatment has been among such options [4, 188]. A pioneering report presented significant clinical improvement of a keratoconic patient who underwent TG-PRK 1 year after CXL [4]. Variations in technique have revolved around procedure timing and sequencing, recommended maximum ablation depth, and the use of Mitomycin-C. We have shown that same-session partial TG-PRK followed by CXL is more effective than sequential CXL with delayed (6 months or more) PRK in achieving visual rehabilitation in keratoconic eyes [7]. Several other studies confirmed the safety and/or efficacy of the simultaneous TG-PRK followed by CXL in patients with KC and post-LASIK ectasia; long-term stability of this combined procedure has also been demonstrated [8, 9, 13–15].

Our team in Athens has contributed many of the evolutionary steps of the initially introduced CXL technique:

Specifically, we have introduced the concept of accelerated, high-fluence CXL in post-LASIK ectasia [189], as well as the utilization of prophylactic CXL in routine LASIK [190], and in situ femtosecond laser-assisted treatment of corneal ectasia [184], in attempting corneal deturgescence [191] in bullous keratopathy [192], and as a prophylactic intervention adjuvant to Boston keratoprosthesis surgery [193].

Seiler and colleagues first introduced CXL to halt the progression of KC. That was approximately 15 years ago, after many years of intensive laboratory research, to demonstrate safety and efficacy [118, 154, 155, 158, 159]. Though not yet FDA approved in the United States, several studies from around the world have demonstrated encouraging results, not only in halting the progression of KC but in many instances there has also been an improvement in UDVA [125, 186, 228]. The original technique researched and described by Seiler and co-workers is an epithelium-off (Epi-Off) technique also known as the “Dresden technique” [118].