Introduction

Riboflavin UV A corneal collagencross-linking (CXL) is a new conser-vative approach for progressive kera-toconus introduced by Wollensak,Spoerl et al. in 2003 (Wollensak et al.2003a,b,c) that can delay progressionof corneal ectasia, reducing or elimi-nating the need for keratoplasty (Wol-lensak et al. 2003a,b,c; Caporossiet al. 2006; Wollensak 2006). There isalso experimental evidence of possiblebenefits of this method in post-lasikectasia (Hafezi et al. 2007) because ofincreased corneal biomechanical resis-tance and reduced collagenase activity(Spoerl et al. 2004). The techniqueinvolves photo-polymerisation of stro-mal collagen fibres induced by thecombined action of a photosensitisingsubstance (riboflavin or vitamin B2)and ultraviolet A light, increasing cor-neal stiffening (Wollensak 2006) andcorneal collagen resistance to enzy-matic digestion (Spoerl et al. 2004;Wollensak & Redl 2008). Accordingto Wollensak, Spoerl et al.(Wollensaket al. 2003a,b,c), Caporossi and Mazz-otta (Caporossi et al. 2006) introducedthe cross-linking procedure in Italy in2004, assessing the corneal micro-structural modifications induced bythis method directly in vivo in humansat the Department of Ophthalmologyof Siena University (Mazzotta et al.2006, 2007a,b, 2008) by HRT II laserscanning corneal confocal microscopy

Morphological and functionalcorrelations in riboflavin UVA corneal collagen cross-linkingfor keratoconus

Cosimo Mazzotta,1 Tomaso Caporossi,2 Rosario Denaro,1

Cristina Bovone,1 Caterina Sparano,1 Anna Paradiso,1

Stefano Baiocchi1 and Aldo Caporossi1

1Department of Ophthalmology, Siena University, Siena, Italy2Department of Ophthalmology, Rome Catholic University, Rome, Italy

ABSTRACT.

Purpose: To investigate the correlations between corneal structural modificationsassessed by in vivo corneal confocal microscopy with visual function [uncorrectedvisual acuity (UCVA), best spectacle-corrected visual acuity (BSCVA)] and mor-phological data (corneal topography, pachymetry, elevation analysis) after ribofla-vin UV A corneal collagen cross-linking (CXL) for the stabilization of progressivekeratoconus.

Methods: Forty-four eyes with progressive keratoconus were enrolled in the SienaEye Cross Study (prospective nonrandomized phase II open trial). All eyes underwentRiboflavin UV A CXL. Preoperative and postoperative evaluation comprised: UCVA,BSCVA, optical pachymetry (Visante OCT, Zeiss, Germany), corneal topography(CSO, Florence, Italy) and tomography (Orbscan IIz; B&L, Rochester, NY, USA)and in vivo confocal microscopy (Heidelberg Retina Tomograph II; Rostock, Heidel-berg Gmbh, Germany). Examinations were performed preoperatively 6 months andone day before treatment and at 1, 3, 6 and 12 months of follow-up.

Results: In vivo corneal confocal microscopy showed time-dependent postoperativeepithelial and stromal modifications after cross-linking. Epithelial thinning associatedwith stromal oedema and keratocytes apoptosis explained initial tendency towardsslightly reduced VA and more glare one month postoperatively in 70% of eyes. Fur-thermore, a statistically not significant early worsening of topographic mean K valueswas observed. Orbscan II analysis significantly underestimated pachymetric valuesafter treatment. Pachymetric underestimation was rectified by high-resolution opticalpachymetry provided by the Visante OCT system. After the third post-CXL month,epithelial thickening, disappearance of oedema and new collagen compaction recordedby in vivo corneal confocal microscopy explained the improvements in visual perfor-mance during the follow-up. Changes in stromal reflectivity and collagen compactionobserved by in vivo confocal microscopy were associated with corneal flattening andreduction in anterior elevation values recorded by differential topographic analysis.

Conclusion: Corneal structural changes assessed by in vivo corneal confocalmicroscopy demonstrated significant correlations with visual function (UCVA andBSCVA) and morphological (corneal topography, pachymetry, elevation analysis)findings recorded after riboflavin-UV A-induced CXL.

Key words: confocal microscopy – cross-linking – elevation analysis – functional results – morp-

hological results – pachymetry

Acta Ophthalmol. 2012: 90: 259–265ª 2010 The Authors

Journal compilation ª 2010 Acta Ophthalmol

doi: 10.1111/j.1755-3768.2010.01890.x

Acta Ophthalmologica 2012

259

(Rostock Cornea Module; HeidelbergEngineering, Heidelberg Gmbh, Ger-many) (Mazzotta et al. 2006, 2007a,b,2008; Kymionis et al. 2009).

The current study (a prospectivenonrandomized phase II open trial)relies on comparative analysis of early(<3 months) and late micro-structuralcorneal modifications recorded byin vivo confocal microscopy, func-tional outcomes [uncorrected visualacuity (UCVA), best spectacle-cor-rected visual acuity (BSCVA)] andmorphological findings (Orbscan IIztomography, Visante OCT opticalpachymetry and corneal topography)in 44 patients undergoing riboflavinUV A CXL. The aim was to analysewhether corneal confocal structuralchanges assessed by confocal micros-copy can explain certain functionaland morphological outcomes recordedin the first year after treatment.

Methods

After Siena University Ethical Com-mittee approval in September 2004,micro-morphological examination of44 cross-linked human corneas wasperformed at the Department ofOpthalmology of Siena University byin vivo HRT II confocal laser scanningmicroscopy (Rostock Cornea Module;Heidelberg Engineering). Patientsunderwent clinical and instrumentalexamination of UCVA, BSCVA mea-sured in Snellen decimal equivalents,corneal topography (CSO Eye Top,Florence, Italy), corneal tomography(Orbscan IIz; B&L, Rochester, NY,USA), optical pachymetry (VisanteOCT; Zeiss, Jena, Germany and Orb-scan IIz; Baush & Lomb, Rochester,NY, USA). Statistical analysis wasconducted by U-Test (Mann–Whitney)for nonparametric data (UCVA andBSCVA) and by paired t-test for para-metric data (mean K power, cornealthickness, elevation analysis and sym-metry index). Control examinationswere performed preoperatively (6months and one day before treatment)and at 1, 3, 6 and 12 months postoper-atively. Inclusion criteria were: agebetween 10 and 40 years, keratoconusprogression documented clinically,UCVA and BSCVA worsening in thelast 3–6 months, increased topographi-cal mean K readings (>0.5 diop-tres ± pachymetry reduction in thethinnest point ‡10 lm), mean K value

under 55 dioptres, optical cornealpachymetry ‡ 400 lm in the thinnestpoint, absence of corneal opacities andVogt striae at slit lamp examination.

Riboflavin UV A-induced CXL wasperformed as follows: pilocarpin 1%drops 30 min preoperatively, topicalanaesthesia with lidocaine 4% drops15 min before and once after epithelialremoval, corneal mechanical (bluntmetal spatula) epithelial scraping ofan area 9 mm in diameter, preirradia-tion corneal soaking for 15 min inriboflavin 0.1% + dextrane 20% sol-ution (Ricrolin�; Sooft, Montegiorgio(AP), Italy), exposure to solid stateUV A illuminator (Caporossi, Baioc-chi, Mazzotta Vega X-Linker; CSO)(Wollensak et al. 2003a,b,c; Mazzottaet al. 2006, 2007a,b, 2008) for 30 min,irradiating an area 8 mm in diameter(energy delivered: 3 mW ⁄ cm2), antibi-otic medication with ofloxacin drops,flurbiprofen drops and lacrimal substi-tutes 4 times a day, therapeutic softcontact lens bandage for 4 days. Pre-servative-free topical steroids (fluor-metholone 0.2% drops) were alsoused in the postoperative period for4–6 weeks for haze prevention (Mazz-otta et al. 2006, 2007a,b, 2008).

Visual Function Results

Mean UCVA was slightly lower thanpreoperatively (mean 0.33 Snellenlines, ±0.12) one month after treat-ment (mean 0.31 Snellen lines, ±0.14)(p = 0.783), while an improvement

was recorded at 3 months (mean 0.45Snellen lines, ±0.17) (p = 0.084),6 months (mean 0.49 Snellen lines,±0.13) (p = 0.0031) and 1 year offollow-up (mean 0.51 Snellen lines,±0.11) (p = 0.000119).

Mean BSCVA was lower than pre-operatively (mean 0.58 Snellen lines,±0.09) at one month (mean 0.53Snellen lines, ±0.12) (p = 0.815),while an improvement was recordedat 3 months (mean 0.65 Snellen lines,±0.1) (p = 0.334), 6 months (mean0.69 Snellen lines, ±0.08) (p =0.00311) and 1 year post-CXL treat-ment (mean 0.75 Snellen lines,±0.075) (p = 0.00023), Fig. 1.

Morphological Results

In vivo Scanning Laser Confocal Analysis

Epithelium: One month after cross-linking, corneal epithelium measuredin the apex region of keratoconus wasvery thin (10–20 lm), Fig. 2A–A1.After 3 months, the thickness incre-ased (30–40 lm) to slightly under thenormal value (50 lm), Fig. 2B–B1.Normal epithelium thickness, resem-bling preoperative pachymetric data(mean 50 ± 10 lm), was detectedbetween 3 and 6 months after thecross-linking procedure (40–50 lm),Fig. 2C–C1. Micro-morphological an-alysis of epithelium showed time-dependent postoperative stratificationand compensatory function, especiallyin the apex region of keratoconus,

1.00

0.90

0.80

0.70

0.60

0.50

Sn

elle

n li

nes

0.40

0.30

0.20

0.10

0.00Preop

0.58

0.33 0.31

0.53

0.65

0.450.49

0.69 0.71

0.51

1 3Months

UCVA–BSCVA

6 12

UCVA

BSCVA

Fig. 1. Functional data (UCVA-BSCVA) in the first year after riboflavin UV A corneal colla-

gen cross-linking. At 1 month was recorded a statistically not significant slightly reduction in

visual acuity. Improvement in UCVA and BSCVA became statistically significant after the sixth

postoperative month. BSCVA, best spectacle-corrected visual acuity; UCVA, uncorrected visual

acuity.

Acta Ophthalmologica 2012

260

where epithelium thickness is gen-erally thinner (Reinstein et al. 2009).We did not observe statistically signifi-cant differences in epithelial thicknessbetween 6 months and 1 year post-CXL treatment. Epithelial thickeningand its progressive stratification wereestablished through a transition bet-

ween basal epithelial cells and sub-epithelial nerve plexus.

Stromal Healing: a major finding ofin vivo confocal microscopy between 3and 6 months after treatment wasincreased density (hyper-reflectivity) ofthe extracellular matrix combined withthe presence of keratocyte nuclei,

Fig. 3B, compared with the preopera-tive scan, Fig. 3A. This was detectedconstantly at the longer follow-up.A major finding was evidence of colla-gen compaction by new structuredfibres in the anterior-mid stroma aftercross-linking, expressed by late hyper-density of the extracelluar matrix,Fig. 4A, with activated (hyper-reflec-tive) keratocyte nuclei and elongatedcell processes, Fig. 4B. Polymericneedle-shaped hyper-reflective micro-bands were revealed by in vivo confo-cal examination, Fig. 4C. This aspectwas constantly observed at the longerfollow-up.

Endothelium: preoperative meanendothelial cell density was 2451 cells ⁄mm2 ± 130.444 (range 2092–3016cells ⁄mm2). A statistically insignificantreduction in endothelial cells countwith respect to physiological reductionwas observed after treatment, namelya mean of 2% per year. Endothelial

(A) (B) (C)

(A1) (B1) (C1)

Fig. 2. Micro-morphological analysis of corneal epithelium performed by HRT II in vivo confocal microscopy. One month after cross-linking, epi-

thelium thickness in the apex region was very thin, A–A1. After 3 months, the thickness increased but did not reach the normal value, B–B1.

Normalization of epithelium thickness in the apex region was detected between 3 and 6 months postcross-linking, C–C1.

(A) (B)

Fig. 3. In vivo HRT II corneal confocal microscopy of corneal stroma performed 3 months

after treatment (B) showed increased density (hyper-reflectivity) of the extracellular matrix ver-

sus preoperative scan (A), combined with the presence of keratocytes nuclei.

(A) (B) (C)

Fig. 4. In vivo HRT II corneal confocal microscopy of corneal stroma. Cross-linking-induced collagen ‘compaction’ was revealed by postoperative

hyper-density of extracelluar matrix (A); activated (hyper-reflective) keratocytes nuclei with elongated cell processes (B); newly formed, polymeric,

needle-shaped, hyper-reflective collagen micro-bands (C).

Acta Ophthalmologica 2012

261

cell density and morphology wereunaltered at corneal confocal micro-scopic examination 1 year post-CXLtreatment.

Optical Pachymetry

Orbscan IIz� (Rochester, NY, USA):in all eyes, a statistically significantreduction in the corneal thinnest pointwith respect to preoperative evalua-tion was detected in the first12 months, mean )130 ± 24.9 lm()115 ⁄ )160 lm) (p =0.0016) at1 month, )116 lm ± 22.4 ()92 ⁄)144lm) (p = 0.0033) at 3 months,)86 lm ± 16.1 ()46 ⁄ )122 lm) (p =0.0063) at 6 months and )52 lm ±24.4 (+10 ⁄ )84 lm) (p = 0.4092) at1 year follow-up, Fig. 5A.

Visante OCT� (Zeiss, Jena, Ger-many): differential optical pachymetryperformed after cross-linking showeda statistically not significant reductionin corneal thinnest point, comparedwith preoperative analysis. Differ-

ential OCT pachymetry performedafter 1, 3, 6 and 12 months showedmean thinnest points of )6 lm ± 8.1()18 ⁄+2 lm) (p = 0.069), +10 lm± 5.7 (+2 ⁄+ 16 lm) (p = 0.912),+13 lm ± 6.4 (+3 ⁄+31 lm) (p =0.943) and + 14.5 lm ± 6.0 (+8 ⁄+28 lm) (p = 0.962), respectively,Fig. 5B.

Corneal Topography

Differential corneal topography withtangential algorithm showed variabledata after treatment. In the first post-operative month, we observed anincrease in central steepening withapparent worsening of mean K value(mean 51.6 D ± 3.2) (p = 0 971) com-pared with preoperative data (mean51.4 D ± 2.8). Differential tangentialmaps performed at 3 months showedinitial flattening in the inferior-tempo-ral quadrant with compensatorysteepening (red bound) in the supe-rior-nasal quadrant Fig. 6A, with a

mean K value reduction of50.3 D ± 2.5 (p = 0.0065). Betweenthe 6th and 12th months, corneal flat-tening involved the remaining inferior-central cornea and steepening (redspot) Fig. 6A, involved the superiorand superior-temporal central corneawith a mean K value improvementof 50.2 D ± 2.6 (p = 0.00412) at6 months and 50.1 D ± 2.6 (p =0.00337) at 12 months, Fig. 6B.

Elevation Analysis

Subtractive analysis provided by ele-vation maps of the anterior cornealsurface showed a reduction in the areaof keratoconus between preoperativeand the first postoperative month,preserving the best fit sphere (BFS)ray without significant elevation dif-ferences between preoperative andpostoperative examinations by Orb-scan IIz tomography. Reduction ofanterior elevation became statisticallysignificant after the third month,Fig. 7A. Differential analysis per-formed using BFS showed a constantand statistically significant increase inBFS compared with preoperative data(mean 7.12 mm, range 6.41 ⁄ 7.58 mm),+0.23 mm ± 0.14 (range )0.06 ⁄ 0.44mm) at 1 month, +0.26 mm ± 0.17(range )0.03 ⁄ 0.52 mm) at 3 months,+0.31 mm ± 0.15 (range 0.03 ⁄ 0.58mm) at 6 months, +0.33 mm ± 0.17(range 0 ⁄ 0.57 mm) at 12 months.

Anterior elevation analysis using afixed (preoperative) fitting sphereshowed statistically not significant

(A) (B)

Fig. 5. Difference pachymetry by Orbscan IIz in the first postoperative months gave a statisti-

cally significant underestimation of corneal thickness (A). Orbscan pachymetric underestimation

was rectified by Visante-OCT high-resolution difference pachymetry that gave statistically not

significant data (B).

(A) (B)

Fig. 6. Differential tangential corneal topography after cross-linking performed by CSO Eye Top Topographer in the first postoperative month

showed a central steepening (A) with statistically not significant worsening of mean K value (B). Constant corneal flattening in the inferior-tempo-

ral quadrant with compensatory steepening (red bound, push up effect) in the superior-nasal quadrant (A) with parallel reduction in mean K

values were recorded after 3, 6 and 12 months (B).

Acta Ophthalmologica 2012

262

reduction in elevation by a mean of)3 lm ± 9 (range +11 ⁄ )22 lm)(p = 0.144) at 1 month that increasedsignificantly by a mean of )15 lm ± 7(range 0 ⁄ )24 lm) (p = 0.00116), )21lm ± 6 (range 0 ⁄ )32 lm) (p =0.000811) and )24 lm ± 5 (range)4 ⁄)33 lm) (p = 0.000784) at 3, 6and 12 months, respectively, Fig. 7A.

A finding constantly recorded in alltreated eyes during follow-up wasreduction of the difference betweensuperior and inferior corneal meridi-ans (flattest ⁄ steeper) expressed by thepreoperative and postoperative com-parative values of the topographicalsymmetry index (SI). SI increased inthe first postoperative month by amean of 5.72 D ± 1.04 (p = 0.919)compared with preoperative meanvalue (5.66 D ± 0.98). After apparentnonsignificant worsening in the first

month, the SI improved significantlyby a mean of 4.47 D ± 0.79 (p =0.0714), 4.17 D ± 0.72 (p = 0.043)and 4.09 D ± 0.74 (p = 0.038) at 3,6 and 12 months, respectively, Fig. 7B.

Discussion

In vivo confocal data recorded afterRiboflavin UV A CXL for progres-sive keratoconus demonstrated inter-esting correlations between functional(UCVA, BSCVA) and morphological(corneal pachymetry, topography, ele-vation analysis) findings. Micro-mor-phological analysis allowed to assessqualitative modifications in treatedkeratoconic eyes over time withaccurate spatio-temporal definition ofcorneal repair processes (Mazzottaet al. 2007a,b, 2008; Kymionis et al.2009).

In the first 4–6 weeks, in 70% oftreated eyes, functional data (UCVAand BSCVA) appeared negativelyinfluenced by transient corneal oedema(glare) expressed by stromal hypo-reflectivity, epithelial thinning andkeratocyte apoptosis in the anterior-mid stroma detected by confocalexamination. Generally, epithelialre-stratification was complete afterthree months but in the apex region,the stratification process was slowerprobably because of a local reducedcellular adhesion influenced by blink-ing process and not homogeneous tearfluid distribution. Contact lens wearingmay also give a contribute generatinga constant micro-trauma, but a pri-mary (dystrophic) origin in failure ofintercellular adhesion mechanismsshould not be excluded as cause of epi-thelial growth retardation in the apexregion. In the 4–6 weeks post-CXL,there was a tendency (with statisticallynot significant values) towards slightlyreduced visual acuity (VA) and moreglare one month postoperatively in xof 44 eyes. Visual improvement gener-ally started 4–6 weeks after treatmentbecoming statistically significant afterthe sixth month. At the same follow-up indeed, in vivo confocal analysisdemonstrated progressive epithelialstratification and thickening, oedemareduction, keratocytes repopulationand heterogeneous changes in stro-mal density (reflectivity). Only in a10–15% of patients, a visual discom-fort (glare) may last over 6 weeks,generally because of a persistentcorneal oedema (bio microscopically

(A) (B)

Fig. 7. Difference anterior elevation analysis performed by Orbscan IIz tomography using a

fixed (preoperative) fitting sphere showed a statistically not significant reduction in anterior ele-

vation 1 month after cross-linking (A). Reduction in anterior elevation became statistically sig-

nificant at 3, 6 and 12 months, (A). Trend of Symmetry Index (SI) values after cross-linking

recorded by CSO Eye Top Topographer showed a statistically not significant reduction in the

first postoperative month (B). SI values improved significantly at 3, 6 and 12 months, (B).

Fig. 8. Summary of confocal findings, functional outcomes (UCVA and BSCVA) and morphological data (pachymetric, topographic, elevation

analysis) in the first year after riboflavin UV A corneal collagen cross-linking. Structural changes detected by corneal confocal microscopy during

follow-up explained some correlations between functional and morphological data recorded during early and late follow-up. In the early follow-

up, UCVA and BSCVA appeared negatively influenced by transient corneal oedema (glare), epithelial thinning and keratocyte apoptosis in the

anterior-mid stroma. Thinner epithelium explained also the statistically not significant worsening of mean K value recorded in the first month. In

the late follow-up, normalization of epithelial thickness combined with resolution of corneal oedema and increased density of the extracellular

matrix with collagen ‘compaction’ explained parallel improvements of functional and morphological data. BSCVA, best spectacle-corrected visual

acuity; UCVA, uncorrected visual acuity.

Acta Ophthalmologica 2012

263

detectable), corneal sub-oedema (onlyconfocally detectable) or presence oftransient stromal haze.

Generally, beyond the third month,functional data tended to improve fur-ther because of new collagen synthesisby repopulating keratocytes andlamellar compaction, expressed byhyper-reflectivity of the extracellularmatrix, combined with newly formedcollagen fibres demonstrated by in vivoconfocal scans (Mazzotta et al. 2008).

During the follow-up, particularlyin the first 6 months after treatment,Orbscan II mean postoperative pachy-metric evaluation showed a statisti-cally significant reduction in cornealthinnest point measurement comparedwith preoperative mean values. Theunderestimation of pachymetric valuesrecorded by Orbscan II may beexplained by uneven postoperativereflectivity observed by confocal scanin the cross-linked portion of thecornea reached by the oxidative doseof UV A (Wollensak et al. 2003a,b,c,2004a,b; Mazzotta et al. 2007a,b).The apparent pachymetric worseningrecorded with the Orbscan II diagnos-tic system was completely disprovedat same follow-up by pachymetricoptical analysis performed with Visan-te OCT and confocal microscopy thatdemonstrated statistically not signi-ficant reduction in corneal thicknessafter cross-linking. The cause ofpachymetric underestimation by Orb-scan II was reasonably related tophysical limitations of the white lightoptical source (Boscia et al. 2002) thatis subject to multiple scatteringbecause of uneven changes in stromalreflectivity. Indeed, the path of lightpassing through a medium is deviatedby optical changes (in our case thecross-linked portion of cornealstroma) having a dimension greater orequal to the light’s ‘phase’, namelyhalf the shortest wavelength of thelight (Serway 1992). On the contrary,the Visante image allowed us toappreciate statistically insignificantchanges in corneal pachymetry aftercross-linking and the presence of ademarcation line between cross-linkedand noncross-linked stroma (Seiler &Hafezi 2006) at a depth similar to thepachymetric value estimated by theOrbscan measurement, confirming ourhypothesis inferred from confocalmicroscopy (Mazzotta et al. 2007a,b,2008). The Visante OCT instrument

proved to be an easy no-contact sys-tem for postoperative pachymetricevaluation, because of its long wave-length (1350 nm) and partially coher-ent light, relatively unaffected by themultiple scattering afflicting otherlight sources.

Topographic data, recorded in thefirst 4–6 postoperative weeks, showingincreased corneal central steepeningwith apparent (statistically not signifi-cant) worsening in mean K value,were related to the thinner epitheliumpresent in the early postoperative per-iod (Mazzotta et al. 2006, 2008).Starting from the third postoperativemonth, differential tangential topogra-phies showed constant corneal flatten-ing in the inferior-temporal quadrantsand compensatory steepening, a sortof topographic ‘push up effect’, in thesuperior-nasal quadrants with parallelstatistically significant reduction inmean K values. Moreover, preopera-tive and postoperative elevation analy-sis showed increased BFS raycurvature (Orbscan data) with areduction in corneal maximum eleva-tion (CSO and Orbscan II data).

A relevant finding constantlyrecorded during the follow-up was adeclining difference between superiorand inferior corneal meridians (flat-test ⁄ steeper) expressed by the preoper-ative and postoperative values of thetopographic SI (Caporossi et al.2006). The variable visual improve-ment recorded after cross-linking indifferent international studies (Wollen-sak et al. 2003a,b,c; Caporossi et al.2006; Kohlhaas 2008; Raiskup-Wolfet al. 2008; Wittig-Silva et al. 2008;Vinciguerra et al. 2009) was confirmedin our series in over 70% of treatedeyes, generally starting (with statisti-cally significant values) after thethird postoperative month, whenin vivo micro-morphological analysisdemonstrated epithelial thickening,disappearance of corneal oedema,repopulation by keratocytes andnew stromal collagen ‘compaction’.Changes recorded by confocal micros-copy in the early and late follow-upperiods (Mazzotta et al. 2007a,b,2008; Kymionis et al. 2009) supportedthe progressive improvement in meanK value, anterior elevation and SI.

In conclusion, in vivo confocal anal-ysis after cross-linking demonstratedsignificant correlations between func-tional (UCVA and BSCVA) and mor-

phological (topographic, pachymetric,elevation analysis) data. A summaryof the data is reported in Fig. 8. Gen-erally, in the first 4–6 weeks afterCXL, corneal confocal microscopyshowed a low stromal reflectivitybecause of a predominant cornealoedema and keratocytes loss. Thesefindings with initial thin epitheliumexplain tendency towards slightlyreduced VA and more glare onemonth postoperatively in x of 44 eyes.After 4–6 weeks post-CXL, stromaloedema gradually disappears with aparallel increase of stromal reflectivityand keratocytes repopulation. Hetero-geneous stromal hyper reflectivity rep-resents an important indirect sign ofcross-linked tissue. This confocalaspect is different from haze(expressed by very high confocalreflectivity and positivity at slit lampexamination). A reliable quantificationof reflectivity, universally acceptable,is difficult because of the extreme vari-ability of the results; however, a paral-lelism between confocal aspects andslit lamp examination should beattempted for differential diagnosisbetween post-CXL stromal hyper den-sity and haze. in vivo corneal confocalanalysis performed after treatmentenabled us to link functional out-comes to different morphological find-ings recorded during early and latepostoperative follow-up, demonstrat-ing that epithelial and stromal colla-gen modifications significantlyinfluenced topographic, pachymetricand visual outcomes after cross-link-ing with a reliable spatio-temporalcorrelation. In many cases, the vari-ability of functional results recordedafter CXL seems to be significantlyrelated to casual and unpredictabledistribution or rearrangement ofcross-linked and newly structured cor-neal collagen. Riboflavin UV A cor-neal collagen indeed should not beconsidered a refractive procedure buta stabilizing treatment with frequentpositive refractive implications relatedto uneven collagen compaction.

Acknowledgement

The authors had no financial supportand have no financial conflict ofinterest or anything to disclose. Thestudy (prospective nonrandomizedphase II open trial, Head: AldoCaporossi, MD, FRCS; Scientific

Acta Ophthalmologica 2012

264

Coordinator: Cosimo Mazzotta, MD,PhD) was approved in September2004 by the Siena University MedicalHospital Institutional Review Boardand by local Ethical Committee ofthe University of Siena. It was con-ducted in accordance with the ethicalstandards of the 1964 Declaration ofHelsinki as renewed in 2000. Allpatients gave their specific writteninformed consent prior to inclusionin the study.

ReferencesBoscia F, La Tegola MG, Alessio G & Sbor-

gia C (2002): Accuracy of Orbscan optical

pachymetry in corneas with haze. J Cata-

ract Refract Surg 28: 253–258.

Caporossi A, Baiocchi S, Mazzotta C, Tra-

versi C & Caporossi T (2006): Parasurgical

therapy for keratoconus by riboflavin-ultra-

violet type A induced cross-linking of cor-

neal collagen: preliminary refractive results

in an Italian study. J Cataract Refract Surg

32: 837–845.

Hafezi F, Kanellopoulos J, Wiltfang R &

Seiler T (2007): Corneal collagen crosslink-

ing with riboflavin and ultraviolet A to

treat induced keratectasia after laser in situ

keratomileusis. J Cataract Refract Surg 33:

2035–2040.

Kohlhaas M (2008): Collagen crosslinking

with riboflavin and UVA-light in keratoc-

onus. Ophthalmologe 105: 785–793.

Kymionis GD, Diakonis VF, Kalyvianaki M,

Portaliou D, Siganos C, Kozobolis VP &

Pallikaris AI (2009): One-year follow-up of

corneal confocal microscopy after corneal

cross-linking in patients with post laser in

situ keratosmileusis ectasia and keratoc-

onus. Am J Ophthalmol 147: 774–778.

Mazzotta C, Traversi C, Baiocchi S, Sergio

P, Caporossi T & Caporossi A (2006):

Conservative treatment of keratoconus by

riboflavin-UVA-induced cross-linking of

corneal collagen: qualitative investigation.

Eur J Ophthalmol 16: 530–535.

Mazzotta C, Balestrazzi A, Traversi C, Baioc-

chi S, Caporossi T, Tommasi C & Capo-

rossi A (2007a): Treatment of progressive

keratoconus by riboflavin-UVA-induced

cross-linking of corneal collagen: ultra-

structural analysis by Heidelberg Retinal

Tomograph II in vivo confocal microscopy

in humans. Cornea 26: 390–397.

Mazzotta C, Balestrazzi A, Baiocchi S, Tra-

versi C & Caporossi A (2007b): Stromal

haze after combined riboflavin-UVA cor-

neal collagen cross-linking in keratoconus:

in vivo confocal microscopic evaluation.

Clin Experiment Ophthalmol 35: 580–582.

Mazzotta C, Traversi C, Baiocchi S, Capo-

rossi O, Bovone C, Sparano MC, Balest-

razzi A & Caporossi A (2008): Corneal

healing after riboflavin ultraviolet-a colla-

gen cross-linking determined by confocal

laser scanning microscopy in vivo: early

and late modifications. Am J Ophthalmol

146: 527–533.

Raiskup-Wolf F, Hoyer A, Spoerl E & Pill-

unat LE (2008): Collagen crosslinking with

riboflavin and ultraviolet-A light in kera-

toconus: Long-term results. J Cataract

Refract Surg 34: 796–801.

Reinstein DZ, Archer TJ, Gobbe M, Silver-

man RH & Coleman DJ (2009): Epithelial

Thickness after Hyperopic LASIK: Three-

Dimensional Display with Artemis Very

High-Frequency Digital Ultrasound.

J Refract Surg 24: 1–10.

Seiler T & Hafezi F (2006): Corneal cross-

linking-induced stromal demarcation line.

Cornea 25: 1057–1059.

Serway RA (1992): Physics for scientist and

engineers. Philadelphia: Saunders College

Publishing, 946–948.

Spoerl E, Wollensak G & Seiler T (2004):

Increased resistance of crosslinked cornea

against enzymatic digestion. Curr Eye Res

29: 35–40.

Vinciguerra P, Albe E, Trazza S, Rosetta P,

Vinciguerra R, Seiler T & Epstein D (2009):

Refractive, topographic, tomographic, and

aberrometric analysis of keratoconic eyes

undergoing corneal cross-linking. Ophthal-

mology 116: 369–378.

Wittig-Silva C, Whiting M, Lamoureux E,

Lindsay RG, Sullivan LJ & Snibson GR

(2008): A randomized controlled trial of

corneal collagen cross-linking in progres-

sive keratoconus: preliminary results.

J Refract Surg 24: 720–725.

Wollensak G (2006): Crosslinking treatment

of progressive keratoconus: new hope. Curr

Opin Ophthalmol 17: 356–360.

Wollensak G & Redl B (2008): Gel electro-

phoretic analysis of corneal collagen after

photodynamic cross-linking treatment.

Cornea 27: 353–356.

Wollensak G, Spoerl E & Seiler Th (2003a):

Riboflavin ⁄ ultraviolet-A-induced collagen

crosslinking for the treatment of keratoc-

onus. Am J Ophthalmol 135: 620–627.

Wollensak G, Spoerl E & Seiler T (2003b):

Stress strain measurements of human and

porcine corneas after riboflavin – ultravio-

let-A induced cross-linking. J Cataract

Refract Surg 29: 1780–1785.

Wollensak G, Spoerl E, Wilsh M & Seiler T

(2003c): Endothelial cell damage after

Riboflavin – Ultraviolet – A treatment in

the rabbit. J Cataract Refract Surg 29:

1786–1790.

Wollensak G, Spoerl E, Wilsh M & Seiler T

(2004a): Keratocyte apoptosis after corneal

collagen cross-linking using ribofla-

vin ⁄UVA treatment. Cornea 23: 43–49.

Wollensak G, Spoerl E, Reber F & Seiler T

(2004b): Keratocyte cytotoxicity of ribofla-

vin ⁄UVA treatment in vitro. Eye 18: 718–

722.

Received on June 27th, 2009.

Accepted on February 11th, 2010.

Correspondence:

Cosimo Mazzotta, MD, PhD

Department of Ophthalmology

Santa Maria alle Scotte Hospital

Viale Bracci 8

53100 Siena

Italy

Tel: + 39 0577 356618

Fax: + 39 0577 356618

Email: cgmazzotta@libero.it

Acta Ophthalmologica 2012

265