Accelerated Corneal Collagen Cross-Linking Using Topography-Guided UV-A Energy Emission: Preliminary Clinical and Morphological Outcomes

Purpose. To assess the clinical and morphological outcomes of topography-guided accelerated corneal cross-linking. Design. Retrospective case series. Methods. 21 eyes of 20 patients with progressive keratoconus were enrolled. All patients underwent accelerated cross-linking using an ultraviolet-A (UVA) exposure with an energy release varying from 7.2 J/cm2 up to 15 J/cm2, according to the topographic corneal curvature. Uncorrected (UDVA) and corrected (CDVA) distance visual acuity, topography, in vivo confocal microscopy (IVCM), and anterior segment optic coherence tomography (AS-OCT) were evaluated preoperatively and at the 1, 3, 6, and 12 months postoperatively. Results. 12 months after surgery UDVA and CDVA did not significantly vary from preoperative values. The average topographic astigmatism decreased from −4.61 ± 0.74 diopters (D) to −3.20 ± 0.81 D and coma aberration improved from 0.95 ± 0.03 μm to 0.88 ± 0.04 μm after surgery. AS-OCT and IVCM documented differential effects on the treated areas using different energies doses. The depths of demarcation line and keratocyte apoptosis were assessed. Conclusions. Preliminary results show correspondence between the energy dose applied and the microstructural stromal changes induced by the cross-linking at various depths in different areas of treated cornea. One year after surgery a significant reduction in the topographic astigmatism and comatic aberration was detected. None of the patients developed significant complications.


Introduction
Conventional riboflavin UVA corneal cross-linking (CXL) represents an evolving therapy for the conservative treatment of progressive keratoconus (KC) [1,2] and secondary corneal ectasia [3,4] due to its capacity to increase the corneal biomechanical resistance and the intrinsic anticollagenase activity. The physiochemical basis of conventional CXL lies in the photodynamic types I-II reactions induced by the interaction between riboflavin molecules, absorbed in corneal tissue, and UVA rays delivered at 3 mW/cm 2 for 30 minutes (5.4 J/cm 2 energy dose), which releases reactive oxygen species able to mediate cross-link formation between and within collagen fibers and within proteoglycan core proteins in the interfibrillary space [5,6].
Conventional CXL with epithelium removal (epitheliumoff) represents an evidence based and scientifically wellsupported treatment, with documented long-term efficacy in stabilizing progressive keratoconus and secondary ectasia as reported in a series of nonrandomized and randomized clinical trials [7,8]. Since conventional CXL procedure requires long treatment time (1 hour approximately), [9] accelerated cross-linking (ACXL) treatment protocols have been proposed with the purpose of shortening treatment time, improving patient's comfort and reducing hospital waiting lists.
Recent studies have shed light on the chain of chemical events occurring during the photochemical activation of riboflavin with ultraviolet light, emphasizing the importance of corneal oxygenation during treatment. With pulsed 2 Journal of Ophthalmology fractionation of ultraviolet-A (UVA) radiation, cross-linking efficiency may be improved by allowing rediffusion of oxygen during UVA light exposure pauses.
Topography-guided ACXL was first proposed as a potential approach to improve optical predictability of CXL and maximizing corneal regularization in a patient-specific computational modeling study of keratoconus progression and differential responses to CXL [10]. In simulations comparing broad-zone CXL treatments to focal, cone-localized treatment, much greater reductions in cone curvature and higher order aberrations (HOA) were observed with cone-localized patterns for a variety of patient tomographies [10]. Given that corneal ectasia is driven by focal rather than generalized weakness [11], focal stiffening of the cone region may promote a more favorable material property redistribution with compensatory steepening of surrounding areas, thereby enhancing topographic normalization [10].
Here we present the 1-year functional and morphological results of the first topography-guided ACXL study performed in Italy.

Materials and Methods
The study was conducted at the Siena International Cross-Linking Center and at the Ophthalmic Unit of the Arcispedale Santa Maria Nuova of Reggio Emilia. The high-irradiance corneal collagen ACXL with topography-guided UVA energy release treatment protocol was approved by the Institutional Review Board. All patients gave informed consent and the study was conducted according to the ethical principles for medical research stated in the Helsinki Declaration as renewed in 2013.

Surgical Technique
Topography-guided ACXL procedures were carried out under sterile operating conditions and topical anesthesia with the application of 4% lidocaine and 0.2% oxybuprocaine hydrochloride anesthetic drops. Topical pilocarpine 2% was administered 10 minutes before treatment. After application of a lid speculum, the corneal epithelium was removed by a blunt metal spatula in the central 9 mm area. Corneal stroma was soaked for 10 minutes with a riboflavin 0.1% Hydroxyl-Propyl Methyl-Cellulose (HPMC) dextranfree solution (VibeX Rapid Avedro Inc., Waltham, MA, USA).
Topography-guided ACXL were carried out with the KXL II6 UVA illuminator (Avedro Inc., Waltham, MA, USA) using a 30 mW/cm 2 UV-A power with pulsed-light emission (1 second on/1 second off). Treatments were individually planned by using a dedicated software (Avedro Mosaic System version 1.0, Avedro Inc., Waltham, MA, USA), according to the preoperative topography data. The 30 mW/cm 2 customized, topography based, ACXL treatments consisted in a differentiated energy dose release according to the different corneal curvatures showed by each keratoconus. The entry level energy dose of 7. The thinnest point and the area of major posterior elevation were included within the highest dose treatment zone. The irradiation patterns shapes included arc, circular, oval, and combined patterns according to keratoconus tomography and shape. The irradiation pattern was aligned by using a direct real time visualization of the cornea, maintaining a perfect centration by the eye-tracking system provided by the machine.
After UV-A irradiation, the cornea was washed with sterile balanced salt solution (BSS), medicated with antibiotics (moxifloxacin), cyclopentolate eye drops, and dressed with a therapeutic soft contact lens that was removed after four days. Inclusion criteria and treatment protocol are listed in Table 1.
Statistical analysis was performed using the Wilcoxon test. All analyses were performed using the IBM SPSS Statistics version 16.0 (Armonk, USA). A p value of ≤0.05 was considered statistically significant.

Results
Twenty-one eyes of 20 patients, 16 males and 4 females, mean age 28.9 ± 5.8 years (range 22-34 years) with progressive keratoconus, were included in the study. One patient Arc patterns for "peripheral cones" (apex distance ≥3 mm from the pupillary center); circular patterns for "central cones"

Anterior Segment OCT Analysis
Corneal OCT scan (Figure 4(a)) showed a double demarcation line according to the different energy doses delivered in the corneal tissue and different exposure times. Treatment irradiation patterns combined a peripheral single arc illumination (7.2 J) followed by a central circular irradiation pattern (10 J), Figure 4  demarcation line, measured from epithelial surface, was 150 ± 18 m SD in the flatter corneal area and 300 ± 37 m SD

IVCM Outcomes
Different demarcation lines were also documented with IVCM at 150 m ± 28 m SD depth in the flattest areas (48 D and under), irradiated at 7.2 J/cm 2 (Figures 6(a),

Discussion
This study documented that topography-guided ACXL is safe and effective in halting keratoconus progression and improving to corneal topography at 12 months. Interestingly, regional effects on keratocyte stromal reflectivity and corneal nerves, as well as multiple stromal demarcation lines, indirectly demonstrated the effectiveness of topography-guided treatment planning according to different E doses and UV-A exposure time. Recently, accelerated cross-linking protocols have been investigated in several studies. Bunsen-Roscoe's law [12] established that photochemical reactions, including CXL, depend on the absorbed UVA energy (E) and their biological effect is proportional to the total energy dose delivered to the tissue [13,14]. According to the so-called "equal-dose" physical principle, 9 mW/cm 2 for 10 min, 18 mW/cm 2 for 5 min, 30 mW/cm 2 for 3 min, or 45 mW/cm 2 for 2 min at constant energy dose (E) of 5.4 J/cm 2 may have the same photochemical impact of conventional CXL with 3 mW/cm 2 for 30 minutes [8,9]. The clinical results of high-irradiance 30 mW/cm 2 ACXL protocol with continuous and fractionated UV-A light exposure [15,16] and 18 mW/cm 2 demonstrated keratoconus stability and endothelial safety, but less anterior corneal flattening effect compared to conventional CXL [17]. On the other hand, a recent review [18] on laser scanning in vivo confocal microscopy (IVCM) of the cornea after conventional and accelerated CXL protocol documented less intrastromal penetration using 30 mW/cm 2 UVA irradiance compared with standard 3 mW/cm 2 CXL [19,20]. The safety of conventional CXL and ACXL riboflavin UVA-induced corneal collagen cross-linking in the conservative treatment of keratoconus was evaluated and confirmed in vivo in humans by laser scanning in vivo confocal microscopy (IVCM) of the cornea [21,22]. IVCM allowed for a detailed, high magnification in vivo micromorphological analysis of corneal layers, enabling the assessment of early and late corneal changes induced by these treatments with much greater axial resolution (1 m) than traditional biomicroscopy and corneal optical coherence tomography (OCT), in both progressive keratoconus and secondary corneal ectasias [21,23].
CXL is known to be an effective mean for stabilizing keratoconus over extended follow-up periods. Even though conventional broad-beam CXL and ACXL protocols induce improvements in visual acuity and topographic and aberrometric parameters in many patients, these optical outcomes vary from case to case due to patient-specific factors and inhomogeneous responses to the intrinsic photodynamic reaction and its stiffening effects [24,25].
In order to improve patient's quality of vision, while maintaining keratoconus stability, we have investigated this novel topography-guided accelerated CXL protocol with customized energy dose release according to corneal curvatures. In this pilot study we have observed a statistically significant reduction of the mean topographic cylinder magnitude along with a decrease in coma aberration. Patients corneal topographies were characterized by the flattening of the steeper KC areas associated with a compensatory steepening on the flattest areas resulting in an improved corneal symmetry. All functional parameters (UDVA and CDVA, max , and average ) tended to improve, and we recorded a flattening of the central cone area as showed in Figure 4(d) compared with preoperative tomography (Figure 4(e)), followed by compensatory steepening of the flattest superior cornea documented in differential corneal tomography map (Figure 4(f)).
The microstructural corneal analysis performed by IVCM showed that in the topography-guided ACXL protocol using energy doses higher than conventional 5.4 J/cm 2 (from 7.2 up to 15 J/cm 2 ), keratocytes apoptosis was detected between 250 (10 J/cm 2 ) and 300 m (15 J/cm 2 ). As showed by corneal OCT scans we revealed multiple demarcation lines underlying the different energy doses and UV-A exposure times used according to the topography-guided ACXL principle, Figures  5(b) and 5(d). These preliminary observations allow us to formulate the hypothesis that CXL induced biodynamic interaction and CXL treatment volume is related not only to the UV-A power and relative exposure time, but also to energy dose delivered to the corneal tissue. In conventional 5.4 J/cm 2 energy dose CXL, we demonstrated with IVCM analysis that CXL stromal penetration (i.e., cell apoptosis) appeared to be inversely proportional to UV-A power and directly proportional to exposure time. The same basic concepts were applicable to 7.2 J/cm 2 energy dose. After high-irradiance, fractionated ACXL with 7.2 J/cm 2 our previously published IVCM data showed an increased hyperreflectivity of stromal tissue surrounding keratocytes compared to 5.4 J/cm 2 energy dose [18,20]. After topography-guided high-irradiance pulsedlight CXL, IVCM showed that by using energy doses over 7.2 J/cm 2 (10 J/cm 2 and 15 J/cm 2 ) we can achieve higher penetration (i.e., cell apoptosis) with reduced exposure time and increased UV power compared to conventional epitheliumoff CXL.
The possibility to have a topography-guided ACXL treatment capable of improving patient's quality of vision, with reduced corneal aberrations and astigmatism, by mean of a nonablative, nonincisional surgery, should be highly desirable in reducing the need of combined procedures for CXL refractive empowerment [26][27][28].
Current study addresses the potentiality of CXL customization based on corneal curvatures, differentiated energy doses, and irradiation times (at the same irradiance, which also implies a higher energy dose). Mazzotta et al. [15] and more recently Peyman et al. [29] have shown that pulsed-light CXL induces a deeper demarcation line than continuouslight CXL maintaining the same irradiance and the same energy, potentially because pulsed-light CXL has a longer overall irradiation time. In the same line, Kling et al. [30] have shown that the biomechanical effect of continuous and pulsed-light CXL (different energies and irradiances, but same overall irradiation time) is equivalent. Therefore, the deeper demarcation line in the "high energy" zone and the observed reduction of astigmatism may result from the longer irradiation time. However, considering that the depth of conventional CXL (C-CXL) with 5.4 J/cm 2 E dose and 30 minutes of total UV-A exposure time reached 300 m of depth on average [19], the topography-guided accelerated CXL reaches a comparable treatment penetration in 18 minutes instead of 30 minutes and in this contest the higher dose may be a possible explanation of the increased treatment penetration beyond the exposure time.
Topography-guided ACXL results, in our preliminary experience, reduced some degree of corneal aberrations and topographic cylinder value with improvement in patient's quality of vision that was recorded since the first postoperative month. However, the overall 1-year results of this pilot study showed no better clinical outcomes compared with literature data reported in conventional broad-beam epithelium-off CXL [1,2,7,31]. Of interest, all treated patients reported a faster improvement in their quality of vision without the typical glare disability reported in the first 1 to 3 months after conventional epithelium-off treatment.
Conventional broad-beam epithelium-off CXL and ACXL remain a safe and efficient solution to delay or halt corneal ectasia progression in progressive keratoconus, for which the primary aim is to stabilize the disease. Topography-guided ACXL may represent an adjunctive option for visual rehabilitation, both for patients with early ectasia with acceptable best spectacle corrected visual acuity and low high order aberrations and for patients with more advanced irregular or decentered cones.