Changes of Subjective Symptoms and Tear Film Biomarkers following Femto-LASIK.

Femtosecond laser-assisted in situ keratomileusis (Femto-LASIK) represents a common treatment modality in refractive surgery and shows excellent results in terms of safety, efficacy, predictability, and long-term stability. However, patients may be affected by dry eye symptoms. The aim of this study was to identify a potential association between subjective dry eye symptoms, objective dry eye markers, and possible changes in the tear film, which could be a target for future therapy development. Therefore, clinical (dry eye) examinations (OSDI, Schirmer test, lissamine green and fluorescein staining, BUT, visual acuity) were carried out before LASIK as well as 5 and 90 days post-OP. The dry eye marker MMP-9, cytokines (IL-1ß, IL-8), and pain markers (NGF, CGRP) were quantified in tear samples with immunoassays. In addition, correlation analyses were performed. Clinical examinations revealed an upregulated OSDI score 5 days post-OP and an increased lissamine green staining score 90 days post-OP. Downregulated CGRP levels were noted 5 days post-OP, while other protein markers were not significantly altered after Femto-LASIK. Hence, Femto-LASIK surgery induced subjective symptoms like that of dry eye which could objectively rather be classified as Femto-LASIK-related discomfort. In the future, this could possibly be better detected and treated using pain markers such as CGRP.

[Long-term outcomes of PRK, LASIK and SMILE]. / PRK, LASIK, SMILE im Langzeitverlauf.

BACKGROUND: Patients and refractive surgeons have a legitimate interest in the long-term results of the most commonly used laser procedures for correction of ametropia. OBJECTIVE: To assess the safety, predictability, stability and late complications according to the current recommendations of the committee for refractive surgery (KRC). METHODS: Literature search in PubMed with the search term "long-term outcome" combined with either "PRK", "LASIK" or "SMILE". Articles about PRK and LASIK with a minimum follow-up of 10 years and SMILE with a minimum follow-up of 5 years were evaluated and duplications were eliminated. RESULTS: From a total of 440 articles that matched these search terms 15 articles were selected and analyzed. The results of PRK were described in 4 studies with a total of 666 eyes, 7 studies (566 eyes) described the results of LASIK and 5 studies (188 eyes) the results after SMILE. CONCLUSION: Long-term studies with follow-up periods of at least 10 years have shown that PRK and LASIK treatment carried out in the early days of medical excimer lasers have a very high level of safety and late complications occurred only rarely. The effectiveness of particularly high corrections declined over time. In contrast, there was no clinically significant regression within the indications recommended by the KRC. The results of SMILE were not inferior to those of PRK and LASIK. In direct long-term comparisons no procedure showed a clear superiority. Due to technological improvements in hardware and software for both preoperative diagnostics and lasers, treatment performed nowadays is presumably even safer, more efficient and also more stable over a long period of time.

Detection of postlaser vision correction ectasia with a new combined biomechanical index.

PURPOSE: To validate and evaluate the use of a new biomechanical index known as the Corvis biomechanical index-laser vision correction (CBI-LVC) as a method for separating stable post-LVC eyes from post-LVC eyes with ectasia. SETTING: 10 clinics from 9 countries. DESIGN: Retrospective, multicenter, clinical study. METHODS: The study was designed with 2 purposes: to develop the CBI-LVC, which combines dynamic corneal response (DCR) parameters provided by a high-speed dynamic Scheimpflug camera (CorVis ST; OCULUS Optikgeräte GmbH) and then to evaluate its ability to detect post-LVC ectasia. The CBI-LVC includes integrated inverse radius, applanation 1 (A1) velocity, A1 deflection amplitude, highest concavity and arclength, deformation amplitude ratio of 2 mm, and A1 arclength in millimeters. Logistic regression with Wald forward stepwise approach was used to identify the optimal combination of DCRs to create the CBI-LVC and then separate stable from LVC-induced ectasia. Eighty percentage of the database was used for training the software and 20% for validation. RESULTS: 736 eyes of 736 patients were included (685 stable LVC and 51 post-LVC ectasia). The receiver operating characteristic curve analysis showed an area under the curve of 0.991 when applying CBI-LVC in the validation dataset and 0.998 in the training dataset. A cutoff of 0.2 was able to separate stable LVC from ectasia with a sensitivity of 93.3% and a specificity of 97.8%. CONCLUSIONS: The CBI-LVC was highly sensitive and specific in distinguishing stable from ectatic post-LVC eyes. Using CBI-LVC in routine practice, along with topography and tomography, can aid the early diagnosis of post-LVC ectasia and allow intervention prior to visually compromising progression.

Epi-Bowman keratectomy versus alcohol-assisted photorefractive keratectomy: wound healing and complications.

PURPOSE: In Epi-Bowman Keratectomy™ (EBK), a new dynamic multi-blade single-use device (Epi-Clear™, Orca Surgical, Israel) is utilized to remove the epithelium by sweeping movements across the corneal surface. Epithelial cells are discarded. Alcohol or other chemical agents are not utilized. We wanted to compare clinical results of Epi-Clear photorefractive kertectomy (PRK) to alcohol-assisted PRK. STUDY DESIGN: Retrospective, comparative study. METHODS: Consecutive case series: Adult patients seeking laser vision correction of myopia or myopic astigmatism without ocular diseases or prior surgery were included. The Epi-Clear PRK group comprised 50 consecutive eyes of 27 patients and the PRK group 50 eyes of 25 patients. RESULTS: No intraoperative complications occurred. Epi-Clear PRK: At day 1, day 4, and 3 months, uncorrected distance visual acuity (UDVA) was 0.41/0.47/0.93 (decimal scale); epithelial defect diameter was 4.7/0.2/0 (mm); pain level was 3.8/0.3/0 (visual analogue scale, 0-10), respectively. At 3 months, efficacy index was 0.86 and spherical aberrations were unchanged. Three eyes (6%) showed (presumably sterile) infiltrates. PRK: At day 1, day 4, and 3 months, UDVA was 0.56/0.46/1.15; epithelial defect diameter was 6.3/0.2/0 (mm); pain level was 5.0/0.3/0, respectively. At 3 months, efficacy index was 1.1 and spherical aberrations were unchanged. CONCLUSION: The new method of epithelial debridement with Epi-Clear before laser ablation seems to offer fast epithelial removal without nicking Bowman's layer (as observed with the laser microscope). However, we found significantly inferior results of Epi-Clear PRK compared to alcohol-assisted PRK. Furthermore, after Epi-Clear PRK corneal infiltrates as a new type of postoperative complication were observed.

Results of thin-cap small-incision lenticule extraction.

PURPOSE: To evaluate the influence of a thin cap in small-incision lenticule extraction (SMILE) for the correction of myopia or myopic astigmatism. SETTING: Tertiary care private practice. METHODS: A chart review of 102 eyes of 51 patients was performed. The effect of 120 µm vs 100 µm cap thickness on postoperative spherical equivalent refraction (SEQ), cylinder, corrected and uncorrected visual acuities, and ease of lenticule separation was assessed in a contralateral manner, whereas all other parameters were identical between eyes (including optical zone, minimum lenticule thickness, incision size, and energy and spot settings) using paired t test. RESULTS: At 3 months postoperatively, SMILE with 120 µm cap thickness was undercorrected in SEQ relative to SMILE with 100 µm cap thickness in a cohort of 102 eyes of 51 consecutive patients. The difference of 0.06 ± 0.39 diopter (or 0.7% ± 5.7%) did not reach statistical significance. Postoperative cylinder was not statistically different in both groups. Visual acuity was similar in both groups. Ease of lenticule separation was identical in both groups. Suction time was shorter with a 100 µm cap (P < .005). Postoperative central residual stromal thickness was 20 ± 15 µm thicker with a 100 µm cap (P < .0001). Adverse events were comparable. CONCLUSIONS: Postoperative refraction, visual acuity, ease of lenticule separation, and incidence of adverse events were not significantly affected by cap thickness. Surgeons might safely use 100 µm instead of 120 µm caps without nomogram adjustment. Alternatively, after suction loss during the lenticule cut, a second docking with a programmed cap thickness of 100 µm (and a larger optical zone) might be a rescue technique enabling surgeons to still perform the intended SMILE procedure.

Repeatability and reproducibility of manifest refraction.

PURPOSE: To evaluate the intraexaminer repeatability and the interobserver reproducibility of manifest refraction. SETTING: Tertiary referral center. DESIGN: Retrospective study. METHODS: Patients attending at least 2 preoperative refractions before undergoing subsequent refractive surgery were included. All manifest refractions were performed by 1 of 4 experienced optometrists using an automated phoropter according to a standard protocol. The first manifest refraction was performed after obtaining automated refraction and measuring the spectacles of the patient. The second refraction was typically refined from the first also considering wavefront refraction and tomography/topography. RESULTS: The latest 2 manifest refractions of 1000 eyes obtained at 2 separate visits showed a mean pairwise absolute difference of 0.16 ± 0.19 diopter (D) (range 0 to 1.38 D) in spherical equivalent (SE). This SD was better than 0.25 D (the minimum measurement increment of refraction itself). The 95% limit of agreement (LoA) was within 0.50 D for sphere, cylinder, and SE. The SD of the astigmatism axis was approximately 10 degrees, and the 95% LoA was within 22 degrees (the difference in axis decreasing significantly with the measured cylinder magnitude). The SD for corrected distance visual acuity (CDVA) was half a Snellen line and the 95% LoA was within 1.5 lines (with increasing deviation with worse vision). There were no clinically meaningful differences in reproducibility (2 optometrists) compared with repeatability (same optometrist) in sphere, axis, and CDVA. CONCLUSIONS: Reproducibility was 0.16 D irrespective whether refractions were performed by 1 or 2 different optometrists. Obtaining multiple refractions preoperatively might increase the predictability of surgery and decrease the enhancement rate.

Thermal Load During Corneal Excimer Laser Ablation: Impact of Different Ablation Parameters.

PURPOSE: To characterize the impact of different ablation parameters on the change in peak corneal temperature during corneal excimer laser ablation. METHODS: Forty-two ablations were performed (21 on polymethylmethacrylate [PMMA] and 21 on porcine eyes) simulating photorefractive keratectomy (PRK), transepithelial PRK, and laser in situ keratomileusis (LASIK) treatments. Each ablation was recorded using infrared thermography. The change in peak corneal temperature was analyzed with respect to varying ablation parameters. RESULTS: Excellent correlation between the temperature rise in PMMA and porcine eyes was found. The change in peak corneal temperature was significantly higher in myopic than in hyperopic corrections, ranging from 6 °C for hyperopic corrections to 16 °C for a -18.00 diopters (D) correction. The change decreased with larger optical zones. Of all analyzed metrics, the refractive correction together with the maximum slope of the treatment showed the best coefficient of determination. Transepithelial ablations led to a higher change in peak corneal temperature than pure stromal ablations. Isothermal lines as a function of the refractive correction and optical zone diameter were calculated with a hypersurface model. If keeping the change in peak corneal temperature below 8 °C is the limit, only 3.00 to 4.00 D can be treated (6.5 mm or larger optical zone); for 10 °C, a maximum of 7.00 D (7 mm or larger optical zone) can be treated and for 12 °C, 10.00 D can be treated (6 mm or larger optical zone). CONCLUSIONS: The effect of different parameters on change in peak corneal temperature during corneal laser ablation could be quantitatively evaluated. Cooling the cornea may aid in avoiding thermal denaturation. This is particularly important for transepithelial ablations, which further increase the thermal load due to the extra ablation of the epithelium. [J Refract Surg. 2020;36(10):667-676.].

Small incision lenticule extraction for the correction of high myopia.

PURPOSE: Excimer laser-based refractive procedures can have less predictable results when used for correcting high myopia than when used for moderate myopia. Small incision lenticule extraction might overcome this weakness. However, small incision lenticule extraction is only Food and Drug Administration approved for use in myopic eyes up to -8 D with astigmatism of -3 D or less. We report outcomes of small incision lenticule extraction in highly and moderately myopic eyes and compare these to modern laser-assisted in situ keratomileusis. METHODS: Retrospective, observational consecutive case series. Inclusion criteria: attempted myopic spherical correction ⩾-8 or-3 to -7.75 D with astigmatism ⩽-3 D, and corrected distance visual acuity of 1.0 (decimal scale) or better. RESULTS: A total of 62 highly myopic and 407 moderately myopic eyes were included. At 3 months postoperatively, the highly myopic eyes had a mean spherical equivalent refraction of -0.28 ± 0.41 D (range: -1.13 to +0.75 D). Mean uncorrected distance visual acuity was 1.0. Mean efficacy index was 0.84. Mean safety index was 1.03. Uncorrected distance visual acuity same or better than corrected distance visual acuity: 61%. Astigmatism was ⩽0.5 D in 90% and ⩽1 D in 100%. The results in the moderately myopic eyes were comparable. CONCLUSION: We found equally good visual and refractive outcomes after small incision lenticule extraction for the correction of high and of moderate myopia combined with an astigmatic correction of up to 3 D, respectively.

Clinical outcomes of small incision lenticule extraction versus advanced surface ablation in low myopia.

PURPOSE: To compare the visual and refractive outcomes of small incision lenticule extraction and advanced surface ablation for low myopia or myopic astigmatism. METHODS: Retrospective, observational case series of our first 50 consecutive small incision lenticule extraction patients compared to refraction-matched 50 advanced surface ablation treatments with attempted spherical equivalent correction ⩽-3.5 D, astigmatism ⩽-1.5 D, and corrected distance visual acuity of 1.0 (decimal scale) or better. Only one eye per patient was included. RESULTS: Small incision lenticule extraction: mean attempted spherical equivalent correction was -2.80 ± 0.63 D. Uncorrected distance visual acuity was 0.85 and 1.0 at days 1 and 5, respectively. At 3 months, mean spherical equivalent refraction was 0.02 ± 0.32 D (range: -0.5 to +0.75 D), mean cylinder was -0.24 ± 0.21 D (range: 0 to -0.75 D), mean uncorrected distance visual acuity was 1.27, mean efficacy index was 0.96, and mean safety index was 1.05. Uncorrected distance visual acuity was same or better than corrected distance visual acuity in 96%, astigmatism ⩽0.5 D in 98% and ⩽1 D in 100% of eyes, respectively. Advanced surface ablation: mean attempted spherical equivalent correction was -2.75 ± 0.5 D. Uncorrected distance visual acuity was 0.72 and 0.61 at days 1 and 5, respectively. At 3 months, mean spherical equivalent refraction was 0.22 ± 0.32 D, mean cylinder was -0.27 ± 0.27 D, mean uncorrected distance visual acuity was 1.21, mean efficacy index was 1.03, and mean safety index was 1.08. CONCLUSION: Small incision lenticule extraction for low myopia was found to be safe and effective with outcomes at 3 months similar to those obtained with advanced surface ablation while offering a quicker visual recovery.

Varifocal Versus Monofocal LASIK in Presbyopic Hyperopic Eyes.

PURPOSE: To compare varifocal (SupraCor; Technolas Perfect Vision GmbH, Munich, Germany) to monofocal (Zyoptix TissueSaving; Bausch & Lomb, Rochester, NY) LASIK in patients with hyperopic presbyopia. METHODS: In this prospective, non-randomized, comparative case series, consecutive patients with hyperopia, presbyopia, and emmetropia as target refraction were bilaterally treated with varifocal (8 patients) or monofocal (7 patients) LASIK. The study was designed for 35 patients, but was terminated early after interim analysis. Outcomes (preoperative and 1 day, 1 week, 1 month, and 3 months postoperative) were: monocular and binocular uncorrected near visual acuity (UNVA), distance-corrected near visual acuity (DCNVA), uncorrected distance visual acuity (UDVA), corrected distance visual acuity (CDVA), low-contrast UDVA and CDVA, efficacy, and safety. RESULTS: Preoperative data were similar in both groups. Monocular and binocular UNVA were not significantly different between both groups at any follow-up visit. At 3 months, mean monocular UNVA was 0.40 logMAR in both groups. Monocular DCNVA and binocular CDVA were not significantly different between groups. Monocular mean CDVA was 0.00 ± 0.06 logMAR after varifocal LASIK and -0.06 ± 0.04 logMAR after monofocal LASIK. The efficacy index was 0.9 after vari-focal LASIK and 0.88 after monofocal LASIK (not significant). The safety index was 1.08 after varifocal LASIK and 1.125 after monofocal LASIK (not significant). CONCLUSIONS: With emmetropia as target refraction, varifocal ablations yielded no additional benefit compared to monofocal ablations in hyperopic presbyopic LASIK. The authors speculate that epithelial remodeling masks the impact of a varifocal ablation pattern and that a myopic postoperative refraction (modified monovision) may be necessary to further improve near and intermediate vision. These results demonstrate the value of a control group in studies evaluating presbyopia corrections. [J Refract Surg. 2019;35(7):459-466.].

Small-incision lenticule extraction for the correction of myopic astigmatism.

PURPOSE: To report the outcomes of small-incision lenticule extraction (SMILE) in myopic eyes with astigmatism of 1.0 to 3.0 diopters (D). SETTING: Center for Refractive Surgery, Eye Department, St. Francis Hospital, Muenster, Germany. DESIGN: Retrospective case series. METHODS: Inclusion criteria were myopia of -0.5 D or more, astigmatism between -1.0 D and 3.0 D, and decimal corrected distance visual acuity (CDVA) of 1.0 or better. RESULTS: The study comprised 206 eyes. Preoperatively, the mean attempted spherical equivalent (SE) correction was -5.45 ± 2.37 D (range -1.25 to -11.38 D) and the mean cylinder was -1.52 ± 0.57 D (range -1.0 to -3.0 D). Three months postoperatively, the mean SE refraction was -0.07 ± 0.38 D (range -1.25 to +1.00 D) and the mean cylinder was -0.32 ± 0.29 D (range 0.0 to -1.5 D). The mean decimal uncorrected distance visual acuity (UDVA) was 1.13, the mean efficacy index was 0.87, and the mean safety index was 1.00. The UDVA was the same as or better than the CDVA in 67% of cases. The angle of error was ±5 degrees in 67% and ±15 degrees in 94% of cases, and the residual astigmatism was 0.5 D or less in 88% and 1.0 D or less in all patients. CONCLUSION: Small-incision lenticule extraction for the correction of myopic astigmatism was safe and effective, with outcomes comparable to those reported for laser in situ keratomileusis using modern eye-tracking systems.

Flap Melting Over Corneal Inlay for Hyperopic Correction.

PURPOSE: To report two cases of flap melting over corneal inlays for hyperopic correction several years after implantation. METHODS: The corneal inlay (+6.00 diopters [D]; PermaVision intracorneal lens; Anamed Inc., Lake Forest, CA) was implanted underneath a LASIK flap in two patients. RESULTS: Visual acuity and slit-lamp findings of both patients were stable for several years. At 9 years postoperatively, the first patient presented with a sudden loss of corrected distance visual acuity (CDVA) due to partial flap melting over the inlay while the surrounding anterior cornea showed opacification. The inlay was explanted immediately. Three years after explantation, CDVA returned to 1.0. Similarly, the second patient presented 11 years postoperatively with loss of vision, opacification, and partial flap melting. The inlay was explanted. Two years later, CDVA recovered. CONCLUSIONS: Potentially severe biocompatibility issues of corneal inlays may occur even after several uneventful years. [J Refract Surg. 2018;34(11):775-778.].

Atypical endophthalmitis after intraocular collamer lens implantation.

Unilateral endophthalmitis occurred after bilateral immediately sequential intraocular collamer lens (Visian Implantable Collamer Lens) implantation for the correction of myopic astigmatism in a 49-year-old woman. The surgeries and initial recovery were unremarkable. However, after 3 days, the patient noticed increasing blurring of vision in the left eye and presented to our clinic. The anterior segment of this eye was quiet, while the vitreous cavity was densely infiltrated. Immediate explantation of the corrective lens and vitrectomy combined with intravitreal antibiotic injection were performed. Staphylococcus epidermidis was detected in the vitreous tap and treated with intravenous antibiotics, antibiotic eyedrops, and parabulbar injections of dexamethasone. The corrected distance visual acuity recovered to 20/20 over a 6-month period. However, a cataract developed thereafter.

Corneal Ectasia After LASIK Combined With Prophylactic Corneal Cross-linking.

PURPOSE: To report a case of unilateral corneal ectasia following LASIK surgery combined with prophylactic corneal cross-linking (CXL) in a young patient. METHODS: Case report. RESULTS: Preoperative topography was unremarkable in both eyes with a minimum corneal thickness of 554 µm in the right eye and 546 µm in the left eye. Preoperative corrected distance visual acuity (CDVA) was 1.0 (20/20 Snellen) in both eyes with a refraction of +1.25 -2.75 × 10 in the right eye and +0.50 -2.00 × 163 in the left eye. LASIK combined with CXL was uneventful. After 12 months, postoperative topography was unremarkable with an uncorrected distance visual acuity (UDVA) of 1.0 in both eyes. Two years after surgery, the patient presented with a loss of vision (UDVA 0.25) and an inferior steepening on topography in the left eye. Standard CXL was performed to arrest further progression. CONCLUSIONS: This report illustrates that the currently used prophylactic CXL protocol in combination with LASIK may not be effectively preventing corneal ectasia in every case. [J Refract Surg. 2017;33(1):50-52.].

Anterior and posterior corneal changes after crosslinking for keratoconus.

PURPOSE: To evaluate anterior and posterior changes in corneal topography and tomography after corneal crosslinking (CXL) in eyes with progressive keratoconus. METHODS: Scheimpflug analyses (Pentacam, Oculus) of 20 eyes with keratoconus performed before and after corneal CXL were included into retrospective analysis. Mean follow-up was 2 years. Changes in topographic, tomographic, and pachymetric values were statistically analyzed applying analysis of variance. Further, the distance and direction between the anterior maximum keratometry (K(max)) and the apex as well as the distance and direction between the thinnest point in corneal thickness (TPCT) and the corneal apex before and after CXL were studied. RESULTS: Two years after CXL, a statistically significant reduction of the keratometry at the flat meridian (-0.8 D, p < 0.05), the steep meridian (-0.5 D, p < 0.05), the "index of surface variance" (-5.3, p < 0.05), and the "index of highest decentration" (-0.05, p < 0.05) could be demonstrated. While the elevation of the front surface at the apex decreased (-1.5 µm, p < 0.05), the back elevation at the apex (+2 µm, p < 0.05) increased. Although not reaching statistical significance, the maximum front and back elevation demonstrated the same trend; while maximum front elevation data remained stable (-0.3 µm, p = 0.961), maximum back elevation data increased (+6.7 µm, p = 0.122). The corneal thickness at the apex (-22.0 µm, p < 0.001) and the TPCT (-20.0 µm, p < 0.001) decreased, leading to an increase of the corneal thickness progression from the corneal apex to the periphery. The position of K(max) and TPCT remained stable. CONCLUSIONS: Corneal topography proved to be useful in the follow-up for CXL because of significant changes in the keratometry. Increasing posterior elevation values, despite a stabilized anterior corneal surface, might be a sign of ongoing ectatic changes in the posterior corneal surface.

The effect of riboflavin/UVA collagen cross-linking therapy on the structure and hydrodynamic behaviour of the ungulate and rabbit corneal stroma.

PURPOSE: To examine the effect of riboflavin/UVA corneal crosslinking on stromal ultrastructure and hydrodynamic behaviour. METHODS: One hundred and seventeen enucleated ungulate eyes (112 pig and 5 sheep) and 3 pairs of rabbit eyes, with corneal epithelium removed, were divided into four treatment groups: Group 1 (28 pig, 2 sheep and 3 rabbits) were untreated; Group 2 (24 pig) were exposed to UVA light (3.04 mW/cm(2)) for 30 minutes and Group 3 (29 pig) and Group 4 (31 pig, 3 sheep and 3 rabbits) had riboflavin eye drops applied to the corneal surface every 5 minutes for 35 minutes. Five minutes after the initial riboflavin instillation, the corneas in Group 4 experienced a 30 minute exposure to UVA light (3.04 mW/cm(2)). X-ray scattering was used to obtain measurements of collagen interfibrillar spacing, spatial order, fibril diameter, D-periodicity and intermolecular spacing throughout the whole tissue thickness and as a function of tissue depth in the treated and untreated corneas. The effect of each treatment on the hydrodynamic behaviour of the cornea (its ability to swell in saline solution) and its resistance to enzymatic digestion were assessed using in vitro laboratory techniques. RESULTS: Corneal thickness decreased significantly following riboflavin application (p<0.01) and also to a lesser extent after UVA exposure (p<0.05). With the exception of the spatial order factor, which was higher in Group 4 than Group 1 (p<0.01), all other measured collagen parameters were unaltered by cross-linking, even within the most anterior 300 microns of the cornea. The cross-linking treatment had no effect on the hydrodynamic behaviour of the cornea but did cause a significant increase in its resistance to enzymatic digestion. CONCLUSIONS: It seems likely that cross-links formed during riboflavin/UVA therapy occur predominantly at the collagen fibril surface and in the protein network surrounding the collagen.