transepithelial riboflavin/ultraviolet. a corneal cross-linking in keratoconus: morphologic studies...

Transepithelial Riboflavin/Ultraviolet. A CornealCross-linking in Keratoconus: Morphologic Studies

on Human Corneas

RITA MENCUCCI, IACOPO PALADINI, ERICA SARCHIELLI, ELEONORA FAVUZZA,GABRIELLA BARBARA VANNELLI, AND MIRCA MARINI

� PURPOSE: To evaluate histologic and molecularchanges in human keratoconic corneas after the proce-dure of transepithelial collagen cross-linking (CXL),without the removal of corneal epithelium.� DESIGN: Experimental laboratory investigation.� METHODS: Thirty corneal buttons were examined, 18of which were from patients affected by severe keratoco-nus and submitted to penetrating keratoplasty (PK).Among these, 8 were analyzed without any treatment, 4were treated with transepithelial CXL 2 hours beforePK, and 6were treatedwith transepithelial CXL3monthsbefore PK. Twelve normal corneal buttons from healthydonors were used as controls. The corneal buttons werethen evaluated by hematoxylin-eosin staining and byimmunostaining with markers of epithelial junctionproteins (ß-catenin and connexin 43), of stromal kerato-cytes (CD34), of apoptosis (terminal deoxynucleotidyltransferase dUTP nick end labeling [TUNEL] assay),and of collagen type I fibers.� RESULTS: The analysis of epithelial markers showeda clear defective expression in keratoconic corneas beforeand soon after the transepithelial CXL treatment, return-ing to normal in corneas analyzed 3 months after transe-pithelial CXL. The analysis of stroma componentsindicated a loss of keratocytes in the upper stroma ofkeratoconic corneas and a trend toward a normal situa-tion 3 months after transepithelial CXL; similarly,collagen fibers appeared disorganized in keratoconus,while their pattern appears to be close to normal 3monthsafter treatment.� CONCLUSIONS: Histologic and immunohistochemicalfindings on human keratoconic corneas showed the pres-ence of biochemical and morphologic alterations in theepithelium and the upper stroma that are significantlyimproved 3 months after transepithelial CXL. However,further studies are necessary to assess to what extentthese results correlate with measurable biomechanical

Accepted for publication Jun 17, 2013.From the Department of Surgery and Translational Medicine, Eye

Clinic (R.M., I.P., E.F.), and Department of Experimental and ClinicalMedicine, Section of Anatomy and Histology (E.S., G.B.V., M.M.),University of Florence, Florence, Italy.

Inquiries to Rita Mencucci, Department of Surgery and TranslationalMedicine, Eye Clinic, Largo Brambilla 3, 50134 - Florence, Italy; e-mail:[email protected]

874 � 2013 BY ELSEVIER INC.

effects. (Am J Ophthalmol 2013;156:874–884.� 2013 by Elsevier Inc. All rights reserved.)

KERATOCONUS IS A NONINFLAMMATORY DYSTROPHY

of the cornea, usually bilateral, characterized byaxial thinning, fragmentation of the epithelial

basement membrane, breaks and scarring at the levelof the Bowman membrane, keratocyte alteration, and,ultimately, stromal scarring.1 Typically, the conditionstarts at puberty, progressing in approximately 20% of thecases to such an extent that surgical procedures becomenecessary.2,3

A new method has been developed for the treatment ofprogressive keratoconus: corneal collagen cross-linking(CXL), which increases the stiffness of the cornea usingultraviolet A (UVA) and riboflavin.4,5 Studies on rabbitand porcine eyes have shown after CXL a 70% increasein corneal rigidity in treated vs untreated corneas, andsimilar studies on human corneas showed a 328.9%increase in stiffness.6–9 Keratocyte apoptosis has beensuggested to be important for replacement of defectivecells and the formation of better-structured collagen fibers,as confirmed by in vivo confocal microscopy studies.10

Recently, a modified method called transepithelial cross-linking has been introduced as a promising therapy forkeratoconus patients to stiffen the cornea without epithe-lial debridement, using a new riboflavin solution able topenetrate and cross the epithelial barrier. However,the real efficacy of this treatment is still under debate,although some clinical studies have shown encouragingresults.11–14 Mazzotta and associates reported in vivoconfocal microscopy studies in which a limited apoptoticeffect was visible as compared to the classic epi-off CXLprocedure,13 whereas Filippello and associates demon-strated in a clinical study a significant arrest of keratoconusprogression evidenced after transepithelial CXL, witha statistically significant improvement in visual acuity andtopographic parameters.12 However, histologic evaluationsof transepithelial CXL results in human corneal architec-ture are necessary to ascertain the effects of treatment.The purpose of the current study was to investigate the

morphologic findings in human corneas treated with trans-epithelial CXL before penetrating keratoplasty (PK), andto compare these findings with normal corneas and

0002-9394/$36.00http://dx.doi.org/10.1016/j.ajo.2013.06.025

ALL RIGHTS RESERVED.

Delta:1_given name

Delta:1_surname

Delta:1_given name

Delta:1_surname

Delta:1_given name

Delta:1_surname

mailto:[email protected]

http://dx.doi.org/10.1016/j.ajo.2013.06.025

untreated keratoconic corneas. More specifically, weanalyzed the effects of transepithelial CXL on epithelialcells, keratocytes, and collagen fibers.

FIGURE 1. Set-up of transepithelial cross-linking on the eye ofa patient. The patient was asked to lie down in a recliningarmchair, and after periocular skin disinfection, a silicone ringwas placed in direct contact with the cornea and filled with 2drops of the enhanced riboflavin solution to cover the cornealapex entirely. The silicone ring was left on the cornea for30 minutes before the cornea was irradiated with ultraviolet Afor 30 minutes.

METHODS

� MATERIALS: This was an experimental laboratory inves-tigation on human corneal samples. Human cornealsamples were collected and divided into 4 study groups:(1) 12 cadaver normal corneas (donors aged 35-55 years)from Lucca Eye-Bank as untreated normal controls; (2) 8corneal buttons with keratoconus (donors 23-38 years old)removed after PK and used as untreated pathologic samples;(3) 4 corneal buttons with keratoconus (donors 23-32 yearsold) treated with transepithelial CXL 2 hours before PK(2h transepithelial CXL); (4) 6 corneal buttons with kera-toconus (donors 22-38 years old) treated with transepithe-lial CXL 3 months before PK (3m transepithelial CXL).

All patients were affected by severe keratoconus withoutstromal scarring. Patients were informed that the transepi-thelial CXL procedure could not restore their vision andavoid PK that had already been previously scheduled.Patients who had previous corneal surgery, intracornealring implants, or a history of delayed wound healing wereexcluded. The Careggi Hospital (University of Florence)Ethics Committee approved the study, and all patientsprovided written informed consent for the treatment andfor the participation in the research, after receivinga detailed description of the nature of the treatment. Prac-tice and research were in accordance with the tenets ofthe Declaration of Helsinki.

Transepithelial CXL was performed under sterile condi-tions applying an enhanced aqueous riboflavin solution,composed by riboflavin phosphate 0.1%, dextran T500,and the enhancers trometamol (tris[hydroxymethyl]amino-methane) and EDTA (ethylenediaminetetraacetic acid)sodium salt (Ricrolin TE; Sooft, Montegiorgio, Italy) onthe intact corneal epithelium for 30 minutes before irradia-tion with UVA (370 nm at 3 mW/cm2) for another30 minutes. Trometamol is a biologically inert low-toxicityamino alcohol present as buffering solution in a wide rangeof cosmetic products and as an alkalinizing agent in pharma-ceutical drugs.12 Sodium EDTA is a well-known chelator ofcalcium and magnesium ions, which are known to be impor-tant in keeping the integrity of tissues, including the epithe-lial tissue.13,15 These enhancers, by weakening epithelialintercellular junctions, help riboflavin to penetrate thecorneal stroma through an intact epithelium, therebyavoiding the need for epithelial debridement.

Three days before transepithelial CXL treatment,patients were administered single-dose eye drops of netilmi-cin 0.3%, 1 drop 4 times a day, for prophylactic purposes.The enhanced riboflavin solution was instilled 30 minutesbefore UVA exposure, with 1 drop being instilled every

VOL. 156, NO. 5 EFFECTS OF TRANSEPITHELIAL CROSS

10minutes thereafter. To reduce the risk for UVA exposureof retroirideal eye structures, miosis was induced with pilo-carpine 1.0% 30 minutes before the procedure. Twentyminutes before UVA irradiation, the cornea was anesthe-tized with single-dose anesthetic eye drops (oxybuprocainehydrochloride 0.2%), 1 drop every 5 minutes. The UVAsource (Vega;CSO, Florence, Italy)was switched on 2hoursbefore the procedure. All treatments were performed withthe power set at 2.9-3.0 mW/cm2 and a circular spot diam-eter of 8.0 mm in an outpatient setting. The patient wasasked to lie down in a reclining armchair, and the periocularskin was disinfected for 5 minutes with povidone-iodinediluted to 10.0%. To improve riboflavin penetration,a ring-shaped silicone container was designed and used.The ring container was 12.0 mm in diameter and 3.0 mmhigh with a flange 2.0 mm wide and 0.3 mm thick at itsbase. Once covered by the eyelid, the flange stabilizes thering on the cornea. The elasticity of the cylinder, which isrigid enough to resist the pressure of the eyelid, allowsminimum lid adjustments. The outer edge of the cylinderand the flange also protect the corneal limbus and its stemcells from accidental UVA irradiation. The silicone ringwas filled with 2 drops of the enhanced riboflavin solutionin direct contact with the corneal epithelium to cover thecorneal apex entirely (Figure 1). If an attempted blinkcaused part of the product to spill from the ring, more

875-LINKING ON HUMAN CORNEAS

FIGURE 2. Hematoxylin-eosin representative images of untreated control corneas, untreated keratoconus, and keratoconic corneastreated with transepithelial cross-linking 2 hours (2h) and 3 months (3m) before penetrating keratoplasty. (Top left) Control humancornea with the healthy morphology of cell layers; in particular, the stroma comprised collagen fibers highly organized interspersedwith a large number of hematoxylin-stained keratocytes. (Top middle-left) Keratoconus showing thinned corneal stroma with loss ofkeratocytes in its anterior part. (Top middle-right) A 2h transepithelial CXL cornea in which only the basal epithelial layer was wellpreserved, and a moderate reduction of keratocytes in the stroma subepithelial portion was detectable. (Top right) A 3m transepithe-lial CXL, showing a corneal morphology returned similar to control cornea. (Original magnification 43.) (Bottom) Bar graph illus-trating quantification of keratocyte number identified by counting the cell nuclei labeled with hematoxylin in at least 3 slides for eachsample and in 10 fields from each slide. Data are expressed as mean ± SD of the mean from each separate sample. CTL taken as 100%(**P < .001). CTL[ control; KER [ keratoconus; TE-CXL [ transepithelial cross-linking.

enhanced riboflavin solution was instilled until riboflavinwas homogeneously dispersed over the corneal apex andits epithelial layer. The silicone ring filled with enhancedriboflavin solutionwas left in direct contact with the corneafor 30 minutes before the cornea was irradiated with UVAfor 30 minutes.13 During the irradiation, a homogeneouslevel of the enhanced riboflavin solution was constantlymaintained by adding 1 drop every 3-5 minutes. In contrastto the initial permeation phase, during the irradiationphase, riboflavin formed only a thin layer over the epithe-lium without building up. Thus, UVA rays were notprevented from penetrating the stroma, shielded by thecorneal epithelium. A simple adhesive sterile plaster waspositioned on the outer edge to collect any drops of theenhanced riboflavin solution leaking from the corneal ringduring the procedure. At the end of the procedure, the sili-cone ring was removed and all residue of the enhanced ribo-flavin solutionwas rinsed away with a sterile physiologic saltsolution. Immediately postoperatively, the treated eye wasmedicated with 1 drop of single-dose norfloxacin. The

876 AMERICAN JOURNAL OF

patient was given a prescription for single-dose netilmicin,1 drop 3 times a day for a week; and sodium hyaluronate0.15%, 1 drop 3 times a day for 20 days.PK of CXL-treated corneas was performed 2 hours or

3 months after the transepithelial CXL procedure. For all3 keratoconus groups and for normal corneas, we analyzedcentral corneal buttons of 7 mm diameter. Immediatelyafter removal, the corneal buttons were cut in small piecesthat were fixed in formalin solution, routinely processed,embedded in paraffin, and used for immunohistochemistryand morphologic detail examination (hematoxylin-eosinstaining).

� IMMUNOHISTOCHEMICAL STUDIES: Immunohisto-chemical studies were performed on tissue sections, aspreviously described.16 Briefly, the slides were stained bythe indirect immunoperoxidase technique using as primaryantibodies the mouse monoclonal antibodies anti-CD34(1:100 dilution; Dakopatts, Carpinteria, California, USA),anti-collagen type I (1:500 dilution; Sigma-Aldrich,

NOVEMBER 2013OPHTHALMOLOGY

TABLE. Analysis of Epithelial (ß-Catenin, Connexin-43) and Stromal (CD34, Collagen I) Markers, Keratocyte Number, and TUNEL-Positive Cells in Untreated Control Corneas, Untreated Keratoconus, and Keratoconic Corneas Treated With Transepithelial

Cross-linking 2 Hours and 3 Months Before Penetrating Keratoplasty

Specimens

Epithelium Stroma

ß-Catenin CX43 Keratocytes CD34 TUNEL Collagen I

Control 100 6 15.54 100 6 10.94 100 6 5.9 100 6 28.02 100 6 65 100 6 12.56

Keratoconus 70.17 6 9.47a 63.57 6 6.07a 66.78 6 8.5a 46.5 6 12.88a 400 6 90a 64.67 6 5.1a

2h TE-CXL 58.46 6 9.45a 43.61 6 4.88a 67.8 6 8.9a 42.26 6 9.19a 650 6 120a 66.72 6 4.18a

3m TE-CXL 100.23 6 19.74 87.72 6 17.55b 97.5 6 6.4 98.13 6 13.65 300 6 70a,c 89.8 6 18

2h¼ 2 hours before penetrating keratoplasty; 3m¼ 3months before penetrating keratoplasty; CX43¼ connexin-43; TE-CXL¼ transepithelial

cross-linking; TUNEL ¼ terminal deoxynucleotidyl transferase dUTP nick end labeling.aP < .001 vs control.bP < .01 vs control.cP < .05 vs keratoconus.

Data are expressed as percent 6 SD of the control value, taken as 100%.

St. Louis, Missouri, USA), anti-ß-catenin (1:100 dilution;Santa Cruz Biotechnology, Santa Cruz, California, USA),and anti-connexin-43 (1:100 dilution; Sigma-Aldrich).The diameter of the collagen fibers in the anterior stromawas measured by a computerized image analyzer program(ImageJ 1.38x, W.S. Rasband, U. S. National Institutes ofHealth, Bethesda, Maryland, USA) on 15 fields for slide.16

Computer-assisted quantification of CD34, ß-catenin, andconnexin-43 (Cx43) was performed using Adobe Photoshop6.0 software (Adobe System Incorporated, San Jose,California, USA).16

� TERMINAL DEOXYNUCLEOTIDYL TRANSFERASE DUTPNICK END LABELING ASSAY: Apoptotic cells were identi-fied using Fragel DNA Fragmentation Detection Kit,Fluorescent – TdT Enzyme (Calbiochem, San Diego,California, USA). Paraffin-embedded tissues were deparaf-finized and subjected to the terminal deoxynucleotidyltransferase dUTP nick end labeling (TUNEL) assayaccording to the manufacturer’s protocol, as previouslydescribed.17 The numbers of apoptotic cells were countedin 10 separate fields of 4 slides.

� STATISTICAL ANALYSIS: The results were expressed asmean 6 SD. Statistical analysis was performed by theMann-Whitney test or by analysis of variance (ANOVA)with post hoc test (Bonferroni correction), as appropriate.P < .05 was considered significant. All statistical analyseswere carried out using SPSS software version 15.0 (SPSSInc, Chicago, Illinois, USA).

RESULTS

� MORPHOLOGIC DETAILS: Morphologic studies withhematoxylin-eosin staining showed a thinning of thecorneal thickness and a reduction of 33.2% in keratocytes


number in all keratoconus specimens (Figure 2, Topmiddle-left) with respect to control (Figure 2, Top leftand Bottom; Table). In 2h transepithelial CXL corneas(Figure 2, Top middle-right), the outer layers of theepithelium were lost, likely because of the effect of theenhancers in the transepithelial riboflavin formulation,but the deeper layers were well preserved; the stroma stillshowed a reduction of keratocytes (Figure 2, Bottom;Table) (reduction of 32.2%). In 3m transepithelial CXLcorneas (Figure 2, Top right) a normal multilayeredepithelial structure was again detected, and keratocytedensity increased (increase of 30.7% vs keratoconus and29.7% vs 2h transepithelial CXL), even though still lowerthan in control healthy corneas (Figure 2, Bottom; Table)(reduction of 2.5%).

� IMMUNOHISTOCHEMISTRY: Epithelium. ß-Catenins aremultifocal proteins involved in cellular junctions andpivotal members of signal transduction pathways.18

Immunohistochemical analysis of ß-catenin expression inthe epithelium showed a uniform labeling in untreatedcontrol corneas (Figure 3, Top left), whereas inkeratoconic corneas only few epithelial cells showedimmunoreactivity for ß-catenin (Figure 3, Top middle-left). In 2h transepithelial CXL corneas, ß-cateninimmunostaining in the preserved epithelial layers wasweak and scattered (Figure 3, Top middle-right). Finally,in 3m transepithelial CXL corneas high ß-cateninexpression was detected in the whole epithelium(Figure 3, Top right), similar to untreated controlcorneas. Quantitative analysis (Figure 3, Bottom; Table)indicated a significant relative decrease of theimmunostaining in keratoconus compared with controlcorneas (P < .001) and 3m transepithelial CXL (P <.001). Such staining for ß-catenin decreased even furtherin 2h transepithelial CXL corneas (P < .001).Connexins are a polygenic family of transmembrane

proteins that form gap junction channels. These gap


FIGURE 3. Immunodetection of ß-catenin in untreated control corneas, untreated keratoconus, and keratoconic corneas treatedwith transepithelial cross-linking 2 hours (2h) and 3 months (3m) before penetrating keratoplasty. (Top left) Control cornea, inwhich the immunoreactivity for b-catenin was detected in the full thickness of the epithelium. (Top middle-left) Keratoconic cornea,in which only few b-catenin-stained epithelial cells are detectable. (Top middle-right) A 2h transepithelial CXL cornea, showingweak and scattered b-catenin immunolabeling in the remaining epithelial layers. (Top right) A 3m transepithelial CXL cornea, inwhich corneal epithelial cells returned to express b-catenin with a pattern similar to control. (Original magnification 203.) (Bottom)Bar graph illustrating the densitometric quantification of b-catenin staining evaluated from 3 slides for each sample and 15 fields foreach slide using Adobe Photoshop 6.0 software. Data are expressed as percent increase ± SD over the control value taken as 100%(**P < .001 vs control). CTL [ control; KER [ keratoconus; TE-CXL[ transepithelial cross-linking; O.D.[ optical density.

junctions permit the direct cell-to-cell exchange of ions,secondary messengers, water, electrical impulses, metabo-lites, and nutrients, and thus mediate intercellular func-tional communication among corneal epithelial cells.19,20

Epithelial cells of control corneas (Figure 4, Top left)strongly expressed Cx43 throughout the epithelium. Kera-toconic corneas (Figure 4, Top middle-left) exhibiteda decreased and scattered immunolabeling pattern, and in2h transepithelial CXL corneas (Figure 3, Top middle-right) Cx43 immunolabeling was observed only sporadi-cally in the epithelium. In 3m transepithelial CXL corneas(Figure 4, Top right), Cx43 expression was characterized byan immunostaining pattern similar to control corneas.Quantitative analysis (Figure 4, Bottom; Table) indicatedthat the lowest amount of Cx43 expression was presentin 2h transepithelial CXL (P < .001); 3m transepithelialCXL showed an immunopositivity stronger than keratoco-nus (P < .001), though still lower with respect to controlcorneas (P < .01).

Stroma. Keratocytes, the main stromal cells, play animportant role in the preservation of corneal transparency


and mechanical stability through the synthesis and main-tenance of the collagen component.16,21 Keratocytedistribution in normal and keratoconic corneas has beenevaluated by CD34 staining. CD34 is a glycosylated type Itransmembrane protein recently used as a specific markerof keratocytes in human corneas.16,22 In control corneas(Figure 5, Upper row left and Lower row left), keratocyteswere found throughout the corneal stroma, whereas inkeratoconus (Figure 5, Upper row middle-left and Lowerrow middle-left) a patchy loss of CD34 immunoreactivitywas observed mainly in the anterior part of the cornea(decrease of 53.5%). In 2h transepithelial CXL corneas(Figure 5, Upper row middle-right and Lower row middle-right), a decrease of 57.74% in CD34 cell positivitywas present in the anterior stroma, while in 3mtransepithelial CXL corneas (Figure 5, Upper row rightand Lower row right) the CD34 positivity was regularlydetected in keratocytes of the all corneal stroma showinga distribution similar to control. Quantitative analysisof CD34 staining (Figure 5, Bottom; Table) showedno significant difference between 3m transepithelialCXL and control, while a significant increase of the


FIGURE 4. Immunodetection of connexin-43 (Cx43) in untreated control corneas, untreated keratoconus, and keratoconic corneastreated with transepithelial cross-linking 2 hours (2h) and 3months (3m) before penetrating keratoplasty. (Top left) Cx43was highlyexpressed in the epithelium of a control cornea. (Top middle-left). A modified and scattered expression pattern was observed in thekeratoconic cornea. (Top middle-right) In 2h transepithelial CXL corneas, Cx43 immunolabeling was rarely detected in the epithe-lium, whereas (Top right) in 3m transepithelial CXL corneas, epithelium showed an immunostaining pattern similar to control. (Orig-inal magnification 203.) (Bottom) Bar graph illustrating the densitometric quantification of Cx43 staining evaluated from 3 slides foreach sample and 15 fields for each slide. Data are expressed as percent increase ± SD over the control value taken as 100% (**P <.001; *P< .01 vs control). CTL[ control; KER[ keratoconus; TE-CXL[ transepithelial cross-linking; O.D.[ optical density.

immunoreactivity was evident when compared withkeratoconus (P < .001) and 2h transepithelial CXLcorneas (P < .001).

Stromal keratocyte apoptosis by the TUNEL assay isshown in Figure 3. This technique was developed asa method to identify individual apoptotic cells by labelingthe ends of the peculiar DNA fragments with the poly-merase terminal deoxynucleotidyl transferase (TdT).23

Apoptotic fluorescent nuclei in TUNEL-stained sectionswere virtually absent in controls (Figure 6, Top left) andpresent in the anterior stroma of keratoconic (Figure 6,Top middle-left) and treated corneas (Figure 6, Topmiddle-right and Top right). More specifically, in bothgroups of transepithelial CXL–treated corneas theapoptotic process appeared to be delimited in a stromalthickness of about 120 mm (Figure 6, Top middle-rightand Top right). The percentage of apoptotic cells in 2htransepithelial CXL corneas was significantly increasedwith respect to all other samples (550%, P < .001 vscontrol; 250%, P < .01 vs keratoconus; 350%, P < .001vs 3m transepithelial CXL). In 3m transepithelial CXLsamples (Figure 6, Top right) the amount of apoptosis,although higher with respect to control (P < .001), was


significantly decreased in comparison to keratoconus(P < .05) (Table).Immunohistochemical analysis of the extracellular

matrix (ECM) of the corneal stroma with anti-collagentype I antibody is shown in Figure 4. This analysis allowedassessment of the diameter and orientation of collagen typeI fibers. Control (Figure 7, Top left) and 3m transepithelialCXL (Figure 7, Top right) corneas showed a tightly packedand highly interwoven distribution of collagen type I fibersin the anterior stroma, while in keratoconus (Figure 7,Top middle-left) and 2h transepithelial CXL (Figure 7,Top middle-right) samples, collagen fiber arrangementwas clearly modified, showing a reduction of the inter-weaving pattern. Relevant changes in the amount collagentype I staining were revealed by quantitative analysis(Table). A statistically significant increase of the immuno-positivity in 3m transepithelial CXL corneas comparedwith untreated keratoconus (P< .001) and 2h transepithe-lial CXL (P < .001) was observed (Figure 7, Bottom left).The pattern and the amount of staining of 3m transepithelialCXL returned to similar to healthy control corneas, withno significant difference in the collagen fiber diameter(Figure 7, Bottom right).


FIGURE 5. CD34 expression by immunolocalization in untreated control corneas, untreated keratoconus, and keratoconic corneastreated with transepithelial cross-linking 2 hours (2h) and 3months (3m) before penetrating keratoplasty. (Upper row left and Lowerrow left) Control cornea showing intense and diffuse expression throughout the corneal stroma. (Upper row middle-left and Lowerrow middle-left) Keratoconic cornea showing poor and patchy CD34 immunoreactivity, localized mainly in the anterior stroma.(Upper row middle-right and Lower row middle-right) A 2h transepithelial CXL cornea in which the faint CD34 immunostainingremains similar to the keratoconic cornea. (Upper row right and Lower row right) A 3m transepithelial CXL cornea in which stainingintensity and pattern distribution of CD34 appears like in the normal cornea. (Hematoxylin counterstained; upper panels, originalmagnification 103; lower panels, original magnification 203.) (Bottom) Bar graph illustrating the densitometric quantification ofCD34 staining evaluated from 3 slides for each sample and 15 fields for each slide. Data are expressed as percent increase ± SDover the control value taken as 100% (**P < .001 vs control). CTL [ control; KER [ keratoconus; TE-CXL [ transepithelialcross-linking; O.D. [ optical density.

Dataobtainedby semiquantitative evaluationof epithelialand stromal corneal markers are summarized in the Table.

DISCUSSION

THIS STUDY DOCUMENTS THE MORPHOLOGIC CHANGES

occurring in human corneas after transepithelial CXL.In the classic protocol for the CXL procedure, epithelial

cells have to be removed mechanically from the cornealsurface in order to allow the saturation of the cornealstroma by riboflavin, and finally the denuded corneas are


exposed to UVA irradiation.24 In this treatment protocol,the epithelial debridement can cause severe pain and visualimpairment during the first days after treatment, untilcomplete regrowth of the epithelium occurs.12 Recently,a modified protocol without removal of the epithelium(transepithelial CXL) has been introduced, to avoid theseearly postoperative complications. The maintenance ofthe epithelium layer preserves corneal morphology, makesthe procedure more comfortable for the subject, and allowsthe treatment of thinner corneas and/or problematicpatients.14 Corneal epithelium is a multicellular layerthat undergoes continuous renewal by cells differentiatingfrom the basal layer. It acts as a barrier to prevent the entry


FIGURE 6. Assessment of terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)-positive cells in untreatedcontrol corneas, untreated keratoconus, and keratoconic corneas treated with transepithelial cross-linking 2 hours (2h) and 3 months(3m) before penetrating keratoplasty. Apoptotic cells appear with pale blue/whitish nuclei because of the co-localization of DAPI(blue nuclear staining) and green signal of TUNEL. (Top left) Fluorescence microscopy analysis showed normal DNA staining inthe control cornea, (Top middle-left) while DNA fragmentation was visible in the anterior stroma of both keratoconus and (Topmiddle-right and Top right) 2 hours and 3 months transepithelial CXL corneas, in which the apoptotic process was confined toa stromal thickness of about 120 mm. (DAPI counterstained; original magnification 103). (Bottom) Bar graph illustrating quanti-fication of TUNEL-positive cells. TUNEL-positive cells have been identified by counting the number of apoptotic cells in at least 3slides for each sample and in 10 fields from each slide. Data are expressed as mean ± SD of the mean from each separate sample. CTLtaken as 100% (**P< .001 vs CTL; #P< .05 vs KER). CTL[ control; KER[ keratoconus; TE-CXL[ transepithelial cross-linking.

of external elements (eg, microorganisms, chemical orphysical agents), and contributes to corneal transparencyand homeostasis.20,25 Corneal epithelial cells areconnected to each other by structures, such as tightjunctions, adherens junctions, and gap junctions, that areinvolved in epithelial function.20 Adherens junctionscontain cadherin- and catenin-type proteins. In the cate-nin family, ß-catenin is a multifocal protein involved inepithelial architecture through the formation of cell surfacecomplexes with E-cadherin, and a pivotal member ofa signal transduction pathway regulating epithelial differ-entiation and proliferation.18,26,27 Gap junctionscontribute to the regulation of corneal epithelial cellfunction such as growth, differentiation, adhesion,migration, and barrier formation.20,28 These junctionscontain connexins, a family of integral membraneproteins that are expressed in a cell type–specific fashion.Cx43 is expressed in the suprabasal and basal layers of


the corneal epithelium and in the anterior stroma.19,20

Various observations suggest that Cx43 plays animportant role in regulating corneal epithelial cellgrowth and makes a crucial contribution to epithelialand stromal integrity.19 The data here reported showthat ß-catenin and Cx43 are highly expressed in theepithelia of human control healthy corneas, while inkeratoconic corneas there is a decreased expression anda different distribution pattern. These modificationsmay indicate a possible underestimated role of the epithe-lium in keratoconus development.29,30 In fact, previousstudies have shown that Cx43 mutations are responsiblefor some human genetic disorders associated with variousophthalmologic findings (ie, microcornea, cataract,glaucoma) and, in particular, a reduced expression ofCx43 seems to be potentially involved in thekeratoconus pathophysiology.19,31 In 2h transepithelialCXL corneas there is a further decrease of both Cx43


FIGURE 7. Collagen type I immunolocalization in untreated control corneas, untreated keratoconus, and keratoconic corneastreated with transepithelial cross-linking 2 hours (2h) and 3 months (3m) before penetrating keratoplasty. (Top left) Control corneashowing highly interwoven collagen type I fibers in the anterior stroma. (Top middle-left) Keratoconic cornea in which an altereddistribution pattern of collagen type I with decreased interweaving is visible. (Top middle-right) A 2h transepithelial CXL corneain which the collagen type I pattern remains as in untreated keratoconic cornea. (Top right) A 3m transepithelial CXL cornea,showing an interweaving pattern of collagen type I fibers similar to the healthy control cornea. (Original magnification 103.) (Bottomleft) Bar graph illustrating the densitometric quantification of collagen type I staining evaluated from 3 slides for each sample and15 fields for each slide. Data are expressed as percent increase ± SD over the control value taken as 100% (**P< .001 vs control).(Bottom right) Collagen fiber diameter evaluation. Measurement of collagen fiber diameter using ImageJ program in 3 slides for eachsample and 15 fields for each slide. Diameters are expressed inmm±SD.No statistically significant difference was detected in collagenfiber diameter among the different corneal samples. CTL[ control; KER[ keratoconus; TE-CXL[ transepithelial cross-linking;O.D. [ optical density.

and ß-catenin expression, confined to the residual basalepithelium, likely attributable to the disassemblement ofthe epithelial layers by the penetration enhancers (Tris/EDTA) treatment. Such disassemblement may result inthe decreased barrier function of the epithelium andthe increased penetration of riboflavin in the underlyingstroma. Most interestingly, 3 months after transepithelialCXL, the newly reformed epithelium contains a normalamount and a normal pattern of Cx43 and ß-cateninexpression. These findings are in line with the recentlypublished clinical confocal microscopy analysis of humancorneas after transepithelial CXL.14

Changes in the corneal stroma are the most relevant forkeratoconus pathology and treatment. It has already beenshown that a small amount of apoptotic keratocytes arepresent in keratoconic corneas and that massive keratocyteapoptosis could be considered an initiator response of theCXL procedure.16 In this study we confirm these previousfindings, showing a decreased amount of keratocytes inpathologic corneas and a dramatic increase of theirapoptosis 2 hours after transepithelial CXL. This indicatesthat the transepithelial CXL procedure affects the cornealstroma in a way that is very similar to the traditional


procedure with epithelium removal; however, this effectwas limited to about 100-120 mm stromal depth in transe-pithelial CXL corneas, where a demarcation line wasdetectable by both CD34 immunopositivity and TUNELanalysis. Most interestingly, after 3 months from transepi-thelial CXL treatment we observed a decreased amountof keratocyte apoptosis, still higher than in normal or kera-toconic corneas, but lower than soon after treatment (2htransepithelial CXL); however, the density and the distri-bution of regenerated keratocytes throughout the wholecorneal stroma recurred to similar to normal corneas.Collagen organization influences corneal shape and

biomechanical properties, playing an important role indetermining visual acuity. Anterior collagen fibers showa high degree of branching, progressively decreasing inthe mid and the posterior stroma.32 Previous studies haveshown an alteration in the anterior lamellar organizationof keratoconus compared with normal corneas, consistingof reduced fiber interweaving.33 In this study, we alsoobserved a reduced anterior fiber interweaving and a reduc-tion of collagen protein expression in keratoconus and2h transepithelial CXL samples. In 3m transepithelialCXL the organization of collagen distribution and fiber


diameter returned to similar to control corneas. However,in contrast to the standard CXL procedure, we did notsee a detectable increase in collagen fiber diameter, thussuggesting that transepithelial CXL, working in the moresuperficial part of the stroma, can only affect the patternof collagen distribution without modifying fiber diameter.

In conclusion, data here presented support a previouslyunderestimated involvement of the epithelium in kerato-conus and show the efficacy, at least from a biochemicaland histologic perspective, of the transepithelial CXLprocedure. This treatment is able to reestablish conditions


similar to normal corneas both at the epithelial (Cx43 andß-catenin expression) and upper stromal level, restoring analmost-normal keratocyte density and distribution andregular collagen fiber appearance. These results are inagreement with, and give a biological support to, thepublished clinical data showing the midterm efficacy ofthe transepithelial CXL procedure.12,14 Further researchis already under way to investigate to what extent themodifications induced by the transepithelial CXLprocedure also correlate with the biomechanical andviscoelastic properties of the cornea.

ALL AUTHORSHAVE COMPLETED AND SUBMITTED THE ICMJE FORM FOR DISCLOSUREOF POTENTIAL CONFLICTS OF INTEREST.The authors indicate no financial disclosures. No funding/support was received. Contributions of authors: design of the study (R.M., I.P., E.S., E.F., M.M.,G.B.V.); conduct of the study (R.M., I.P., E.S., E.F., M.M., G.B.V.); collection, management, analysis, and interpretation of data (R.M., I.P., E.S., E.F.,M.M., G.B.V.); and preparation, review, or approval of manuscript (R.M., I.P., E.S., E.F., M.M., G.B.V.).

REFERENCES

1. Hayes S, Boote C, Tuft SJ, Quantock AJ, Meek KM. A studyof corneal thickness, shape and collagen organisation in kera-toconus using videokeratography and X-ray scattering tech-niques. Exp Eye Res 2007;84(3):423–434.

2. Tuft SJ, Moodaley LC, Gregory WM, Davison CR,Buckley RJ. Prognostic factors for the progression of keratoco-nus. Ophthalmology 1994;101(3):439–447.

3. Kennedy RH, Bourne WM, Dyer JA. A 48-year clinical andepidemiologic study of keratoconus. Am J Ophthalmol 1986;101(3):267–273.

4. Wollensak G, Spoerl E, Seiler T. Riboflavin/ultraviolet-a-induced collagen crosslinking for the treatment of keratoco-nus. Am J Ophthalmol 2003;135(5):620–627.

5. Wollensak G. Crosslinking treatment of progressive kerato-conus: new hope. Curr Opin Ophthalmol 2006;17(4):356–360.

6. Wollensak G, Spoerl E, Seiler T. Stress-strain measurementsof human and porcine corneas after riboflavin ultraviolet-A-induced cross-linking. J Cataract Refract Surg 2003;29(9):1780–1785.

7. Spoerl E,Wollensak G, Seiler T. Increased resistance of cross-linked cornea against enzymatic digestion.Curr Eye Res 2004;29(1):35–40.

8. Spoerl E, Wollensak G, Ditterd DD. Thermomechanicalbehavior of collagen-cross-linked porcine cornea.Ophthalmo-

logica 2004;218(2):136–140.9. Kohlhaas M, Spoerl E, Schilde T, Unger G, Wittig C,

Pillunat LE. Biomechanical evidence of the distribution ofcross-links in corneas treated with riboflavin and ultravioletA light. J Cataract Refract Surg 2006;32(2):279–283.

10. Mazzotta C, Traversi C, Baiocchi S, et al. Corneal healingafter riboflavin ultraviolet-A collagen crosslinking deter-mined by confocal laser scanning microscopy in vivo: earlyand late modifications. Am J Ophthalmol 2008;146(4):527–533.

11. Caporossi A, Mazzotta C, Baiocchi S, Caporossi T,Paradiso AL. Transepithelial corneal collagen crosslinkingfor keratoconus: qualitative investigation by in vivo HRT

II confocal analysis. Eur J Ophthalmol 2012;22(Suppl 7):S81–88.

12. Filippello M, Stagni E, O’Brart D. Transepithelial cornealcollagen crosslinking: bilateral study. J Cataract Refract Surg2011;38(2):283–291.

13. Mazzotta C, Caporossi T, Denaro R, et al. Morphological andfunctional correlations in riboflavin UVA corneal collagencross-linking for keratoconus. Acta Ophthalmol 2012;90(3):259–265.

14. Filippello M, Stagni E, Buccoliero D, Bonfiglio V,Avitabile T. Transepithelial cross-linking in keratoconuspatients: confocal analysis. Optom Vis Sci 2012;89:e1–7.

15. Wollensak G, Iomdina E. Biomechanical and histologicalchanges after corneal crosslinking with and without epithelialdebridement. J Cataract Refract Surg 2009;35(3):540–546.

16. Mencucci R, Marini M, Paladini I, et al. Effects of riboflavin/UVA corneal cross-linking on keratocytes and collagenfibres in human cornea. J Clin Exp Ophthalmol 2010;38(1):49–56.

17. Sarchielli E, Pacini S, Morucci G, et al. Cadmium inducesalterations in the human spinal cord morphogenesis.Biometals 2012;25(1):63–74.

18. Kase S, Osaki M, Sato I, et al. Immunolocalisation ofE-cadherin and beta-catenin in human pterygium. Br JOphthalmol 2007;91(9):1209–1212.

19. Gatzioufas Z, Charalambous P, Thanos S. Reduced expressionof the gap junction protein Connexin 43 in keratoconus.Eye (Lond) 2008;22(2):294–299.

20. Kimura K, Teranishi S, Kawamoto K, Nishida T. Protectionof human corneal epithelial cells from hypoxia-induceddisruption of barrier function by hepatocyte growth factor.Exp Eye Res 2010;90(2):337–343.

21. Hassell JR, Birk DE. The molecular basis of corneal transpar-ency. Exp Eye Res 2010;91(3):326–335.

22. Perrella G, Brusini P, Spelat R, Hossain P, Hopkinson A,Dua HS. Expression of haematopoietic stem cell markers,CD133 and CD34 on human corneal keratocytes. Br JOphthalmol 2007;91(1):94–99.

23. Heatwole VM. TUNEL assay for apoptotic cells.Methods Mol

Biol 1999;115:141–148.


http://refhub.elsevier.com/S0002-9394(13)00423-6/sref1














































































24. Thorsrud A, Nicolaissen B, Drolsum L. Corneal collagencrosslinking in vitro: Inhibited regeneration of human limbalepithelial cells after riboflavin–ultraviolet-A exposure. J

Cataract Refract Surg 2012;38(6):1072–1076.25. Lu L, Reinach PS, Kao WW. Corneal epithelial wound heal-

ing. Exp Biol Med (Maywood) 2001;226(7):653–664.26. Schlotzer-SchrehardtU,Kruse FE. Identification and character-

ization of limbal stem cells. Exp Eye Res 2005;81(3):247–264.27. Campbell M, Humphries M, Kenna P, Humphries P,

Brankin B. Altered expression and interaction of adherensjunction proteins in the developing OLM of the Rho(-/-)mouse. Exp Eye Res 2007;85(5):714–720.

28. Kimura K, Nishida T. Role of the ubiquitin-proteasomepathway in downregulation of the gap-junction proteinConnexin43 by TNF-a in human corneal fibroblasts. InvestOphthalmol Vis Sci 2010;51(4):1943–1947.


29. Rohrbach JM, Szurman P, El-Wardani M, GrubM. About thefrequency of excessive epithelial basement membrane thick-ening in keratoconus. Klin Monbl Augenheilkd 2006;223(11):889–893.

30. Maruyama Y, Wang X, Li Y, Sugar J, Yue BY. Involvementof Sp1 elements in the promoter activity of genes affectedin keratoconus. Invest Ophthalmol Vis Sci 2001;42(9):1980–1985.

31. Shurman DL, Glazewski L, Gumpert A, Zieske JD,Richard G. In vivo and in vitro expression of connexins inthe human corneal epithelium. Invest Ophthalmol Vis Sci2005;46(6):1957–1965.

32. Ruberti JW, Roy AS, Roberts CJ. Corneal biomechanics andbiomaterials. Annu Rev Biomed Eng 2011;13:269–295.

33. Meek KM, Fullwood NJ. Corneal and scleral collagens–a microscopist’s perspective. Micron 2001;32(3):261–272.


































Biosketch

Dr Rita Mencucci is currently Professor at the School of Ophthalmology at the University of Florence, Italy, and Assistant-

Professor at the Eye Clinic, University of Florence. She received her medical degree and completed her specialty training

in Ophthalmology at the University of Florence. Dr Mencucci’s special interests focus on Cornea, Anterior Segment

Surgery and Pediatric Ophthalmology. She has published more than 120 scientific papers, many of which have appeared

in major journals.

VOL. 156, NO. 5 884.e1EFFECTS OF TRANSEPITHELIAL CROSS-LINKING ON HUMAN CORNEAS

transepithelial riboflavin/ultraviolet. a corneal cross-linking in keratoconus: morphologic studies...

Documents