Introduction

A positive linear correlation existsbetween stromal water content andthe thickness of the cornea (Ytteborg& Dohlman 1965). Thickness regula-tion therefore implies regulation ofstromal hydration, determined by thebarrier functions of the epitheliumand endothelium. Interference witheither of these boundary layers is fol-lowed by oedema and an increase inthickness. Because of the mechanicalstructure of the corneal fibres, abreakdown of the epithelial barrierresults in only a modest thicknessincrease: values > 0.6 mm are rare.

By contrast, endothelial decompensa-tion is followed by considerable stro-mal swelling, with thickness values of0.8–1.0 mm, which, because of theinterwoven fibril texture of the humancorneal stroma, is the maximum swell-ing possible.

With increasing thickness, the ten-dency to absorb fluid, the imbibitionpressure, is reduced from 60 mmHgat normal hydration (Hedbys & Do-hlman 1963; Olsen & Sperling 1987)to virtually zero at 1 mm thickness.When the epithelium is healthy andintact, and almost water-imperme-able, fluid to increase the stromalvolume comes from the aqueous

humour. The net inflow representsthe difference between passive pres-sure-driven inflow and active meta-bolic pumping out:

Inflow¼P=R� pump flow

Inflow¼ðimbibition pressureþ IOPÞ=R� pump flow

where P = pressure, R = resistance,and IOP = intraocular pressure.Inflow stops, which means that thick-ness stabilizes over time, when inflowequals outflow.

Passive inflow over the endothe-lium can be reduced by diminishingstromal imbibition pressure. Bycross-linking, the stromal swellingcapacity and swelling pressure maybe reduced. The cross-linking canbe accomplished by free oxygen radi-cals produced by ultraviolet (UV)radiation of riboflavin (vitamin B2).This treatment modality was intro-duced by Wollensak et al. (2003a),mainly with the purpose of strength-ening the stromal fibril texture inthe longitudinal direction, and hasbeen used in the treatment of kera-toconus.

During some preliminary animalexperiments (Ehlers et al. 2003), theidea of cross-linking stromal fibres inan attempt to reduce corneal oedemaby reducing swelling tendency andthickness, and to increase opticalclarity by reducing fibril scattering,emerged. This paper presents thepreliminary clinical results of stromalcross-linking, and suggests that thistreatment principle does in factwork.

Riboflavin-ultraviolet lightinduced cross-linking inendothelial decompensation

Niels Ehlers and Jesper Hjortdal

Department of Ophthalmology, Arhus University Hospital, Arhus C, Denmark

ABSTRACT.

Purpose: To evaluate the potential of collagen cross-linking in the treatment

of corneal oedema caused by endothelial decompensation.

Methods: Riboflavin-ultraviolet (UV) treatment induces cross-linking and

reduces stromal swelling. Eleven patients with corneal oedema were treated.

The technique comprised: epithelial abrasion; instillation of 0.1% riboflavin in

saline, and 5.4 J ⁄ cm2 illumination with 365 nm UV-A light over approxi-

mately 30 mins (3 mW ⁄ cm2).

Results: A reduction in corneal thickness was observed in 10 patients. The

majority also experienced improvement in vision. The effect occurred over

weeks and lasted for months.

Conclusions: The study shows a potential application of collagen cross-linking

in the management of patients with corneal oedema. Experimental and addi-

tional clinical studies are necessary in order to define the precise indications

for this type of treatment.

Key words: cornea – cross-linking – endothelium – riboflavin – thickness – UV light

Acta Ophthalmol. 2008: 86: 549–551ª 2008 The Authors

Journal compilation ª 2008 Acta Ophthalmol

doi: 10.1111/j.1600-0420.2007.01085.x

Acta Ophthalmologica 2008

549

Materials and Methods

The treatment with riboflavin andsubsequent UV radiation essentiallyfollowed the technique described byWollensak et al. (2003b).

The treatment was administeredunder topical anaesthesia with dropsof 0.4% oxybuprocaine and 1% lido-caine gel with the patient in the supineposition. The patient was asked tolook at a red fixation light in theoperating microscope. The epitheliumwas mechanically abraded (which wasvery easy in cases of incipient or man-ifest oedema). The cornea was thenexposed to 0.1% riboflavin solution in0.9% saline for 15–30 mins. The cor-nea was treated by establishing asmall, open reservoir with an 8-mmcorneal marker lightly pressed againstthe cornea. Some patients requiredanother drop of anaesthetic duringthe treatment. When the staining per-iod ended, the eye was flushed withsaline to remove any riboflavin on theconjunctiva and limbal cornea.

Radiation was performed with com-mercially available equipment (IROCUV-X; IROC A6, Zurich, Switzer-land), according to the manufacturer’sinstructions. The radiation intensitywas checked before each treatmentand the lamp placed above the corneaat the optical focal distance. Thewavelength was 365 nm (UV-A), withan effect of 3 mW ⁄ cm2. The treatmenttime was intended to be 30 mins(5.4 J ⁄ cm2), but this was shortened insome cases when the patient feltuncomfortable. The immediate after-care comprised voltaren drops admin-istered a maximum of four times perday and chloramphenicol ointment toprevent infection until the epitheliumhad healed.

Postoperative follow-up in all casesexceeded 3 months and included slit-lamp biomicroscopy, optical non-con-tact pachymetry (Haag-Streit, Bern,Switzerland) and determination ofvisual acuity (VA).

Patients

This preliminary clinical study com-prised 11 patients. Diagnoses includedFuchs’ dystrophy (n = 4), secondarybullous keratopathy (n = 2), oedema-tous rejected grafts (n = 3), cornealoedema resulting from longstandingsecondary glaucoma (n = 1) and

endothelial damage possibly caused bymechanical damage after a complicatedforceps delivery (n = 1). Four of thepatients were treated twice (patients 4–7). Table 1 shows patient details.

In all cases the alternative to UVcross-linking would have been graft-ing. The patients accepted the treat-ment in the hope of avoiding orpostponing transplantation and thetime-consuming aftercare it involves.

Results

All patients indicated some pain inthe first few postoperative days untilepithelial healing was complete.

Effect of cross-linking on central corneal

thickness

As Table 1 shows, 10 of 11 casesshowed a redction in central cornealthickness (CCT), the exception beingcase 6, which showed very advancedsecondary bullous keratopathy aftercataract operation. Examination ofthe central corneal button after apenetrating graft revealed almost com-plete loss of endothelium. Fourpatients (4, 5, 6 and 7) were treatedtwice and showed a further decreasein thickness after retreatment. Severalpatients demonstrated increased thick-ness in the first postoperative daysbefore the epithelium had healed andthen a decrease in thickness thatoccurred over several weeks. Thissame pattern of events was observed

after the retreatments. Fig. 1 illus-trates the clinical response.

Effect on visual acuity

In some patients (1, 2 and 10), VAimproved (Table 1). In other cases thiseffect was limited as the dystrophic-degenerative changes in the very pos-terior aspect of the cornea remainedopaque and reduced the total cornealtransparency, despite an evidentincrease in the clarity of the stroma.

The effect over time

Usually the reduced stromal thicknesswas observed after weeks, but in somecases a result was seen only after alonger period. Other patients showeda longer term increase in thickness,suggesting that the obtained effectmay not be permanent.

Complications of treatment

No corneal side-effects were noticed,in particular no corneal infections, noretarded epithelial healing, no intraoc-ular inflammation, and no progressionin lenticular opacification. However,the observation period was too shortto allow any conclusions about possi-ble complications.

Corneal thickness

In all cases but one, corneal oedemadecreased. The graphs show an imme-diate increase in CCT, no doubt

Table 1. Patients treated by ultraviolet cross-linking for stromal oedema.

Diagnosis

Preoperative

CCT

Visual

acuity

Postoperative

CCT

Visual

acuity

1. 68-year-old woman Fuchs’ dystrophy 0.68 0.3 0.60 0.5

2. 54-year-old woman Fuchs’ dystrophy 0.58 0.6 0.51 0.8

3. 79-year-old woman Fuchs’ dystrophy 0.66 0.2 0.53 0.2

4. 69-year-old woman* Fuchs’ dystrophy 0.65 0.7 0.61

0.61 0.58

5. 80-year-old woman* Bullous keratopathy 0.73 0.75

0.75 0.51 0.05

6. 76-year-old man* Bullous keratopathy 0.9 0.05 0.78 0.05

0.78 0.90

7. 52-year-old man* Graft 0.80 0.01 0.66

0.66 0.57 0.05

8. 86-year-old woman Graft 0.76 0.02 0.67 0.05

9. 55-year-old woman Graft 0.69 0.05 0.58 0.05

10. 60-year-old man Glaucoma 0.595 0.1 0.50 0.32

11. 58-year-old woman Trauma 0.659 0.1 0.560 0.25

* Cases 4 to 7 were treated twice.

CCT = central corneal thickness.

Acta Ophthalmologica 2008

550

related to the epithelial abrasion. Overlonger periods (weeks to months) theCCT was reduced.

Visual acuity

Only one patient (patient 9) experi-enced visual deterioration. The otherpatients all experienced some improve-ment in vision, in particular when theposterior boundary (Descemet’s mem-brane and endothelium) showed littlealteration.

In case 10 the treatment was givenprior to a planned phacoemulsifi-cation in an attempt not to elicita decompensation. Three monthsafter cataract surgery, CCT thicknesswas 0.53 mm, VA 0.32 and IOP12 mmHg.

Conclusions

The cross-linking induced by ribo-flavin-UV irradiation seems to leadto stromal deswelling and increasedtransparency. This effect can be

attributed to decreased stromal imbi-bition pressure and hence reducedtransendothelial inflow.

The small series presented herewould seem sufficient to suggest away of treating corneal oedema thatattempts to avoid or postpone cornealgrafting, and possibly offers someprotection to the cornea during andafter cataract surgery. Endothelialdamage by the irradiation is a poten-tial complication of this treatment.Ultraviolet-B radiation, with wave-lengths of around 280 nm, is absorbedin aromatic aminoacids of the cellnuclei (Olsen & Ringvold 1982;Ringvold et al. 1982; Riley et al. 1987;Hamil 2005). The effect of UV-A,with a wavelength of 365 nm, hasbeen less studied. If the photochemicaldamage to cells caused by UV-A isdose-related, a fractionation may bepreferable.

Detailed experimental studies willbe needed before this treatment canfind a place in the management ofcorneal oedema.

ReferencesEhlers N, Hjortdal JØ, Funding M & Vorum

H (2003): Modification of properties of

corneal stroma by molecular cross-linking.

[Abstract.] European Vision and Eye

Research, Annual Meeting, Alicante,

Spain, October 2003.

Hamil MB (2005): Mechanical injury. In:

Krachmer JH, Mannis MJ & Holland EJ

(eds.) Cornea, 2nd edn. Philadelphia: Else-

vier-Mosby 1245–1261.

Hedbys BO & Dohlman CH (1963): A new

method for the determination of the swell-

ing pressure of the corneal stroma in vitro.

Exp Eye Res 2: 122–129.

Olsen EG & Ringvold A (1982): Human cor-

nea endothelium and ultraviolet radiation.

Acta Ophthalmol 60: 54–56.

Olsen T & Sperling S (1987): The swelling

pressure of the human corneal stroma as

determined by a new method. Exp Eye Res

44: 481–490.

Riley MV, Peters SS & Schwartz CA (1987):

The effect of UV-B irradiation on the corneal

endothelium. Curr Eye Res 6: 1021–1033.

Ringvold A, Davanger M & Olsen EG

(1982): Changes of the cornea endothelium

after ultraviolet irradiation. Acta Ophthal-

mol 60: 41–53.

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

Riboflavin ⁄ ultraviolet-A-induced collagen

cross-linking for the treatment of kera-

toconus. Am J Ophthalmol 135: 620–627.

Wollensak G, Spoerl E, Seiler T & Wilsch M

(2003b): Endothelial cell damage after com-

bined riboflavin-UVA treatment. J Cata-

ract Refract Surg 29: 1786–1790.

Ytteborg J & Dohlman CH (1965): Corneal

oedema and intraocular pressure. Arch

Ophthalmol 74: 477–484.

Received on April 27th, 2007.

Accepted on September 3rd, 2007.

Correspondence:

Niels Ehlers

Department of Ophthalmology

Arhus University Hospital

8000 Arhus C

Denmark

Tel: + 45 8949 3222

Fax: + 45 8612 1653

Email: ehler@as.aaa.dk

Days0 50 100 150 200 250

Cor

neal

thic

knes

s (m

m)

0.55

0.60

0.65

0.70

0.75

0.80

0.85

UVUV

Fig. 1. Corneal thickness decrease after cross-linking induced by riboflavin-ultraviolet in a

52-year-old man (case 7) grafted 3 years previously for a perforated corneal ulcer. A rejection

was treated by steroids, with only temporary effect. The graft had been oedematous for almost

a year before treatment. The patient was treated twice, as indicated by the arrows.

Acta Ophthalmologica 2008

551