the corneal endothelium after keratoplasty for keratoconus

RESEARCH PAPER

The corneal endothelium after keratoplasty for keratoconus

Clin Exp Optom 2013; 96: 201–207 DOI:10.1111/cxo.12022

Sven Jonuscheit*† PhDMichael J Doughty† PhDKanna Ramaesh* FRCOphth* Tennent Institute of Ophthalmology, GartnavelGeneral Hospital, Glasgow, United Kingdom† Department of Life Sciences, Glasgow CaledonianUniversity, Glasgow, United KingdomE-mail: [email protected]

Background: The aim was to assess the corneal endothelium, post-operative visual outcomeand complications following keratoplasty for keratoconus.Methods: Seventy-six corneal grafts (57 per cent penetrating keratoplasties and 43 per centdeep anterior lamellar keratoplasties) were assessed as part of routine follow-up at theOphthalmology Department at Gartnavel General Hospital in Glasgow. Routine ophthalmicassessment included visual acuity, slitlamp biomicroscopy and scanning slit confocal micro-scopy of the central cornea. Case records were reviewed and demographic, surgical andpost-operative data retrieved. Linear and logistic regression models were used to determineassociations between endothelial cell density, visual acuity and post-operative complications.Kaplan–Meier analysis was used to estimate the probability for grafts to retain more than700 endothelial cells per mm2 and to have 0.3 logMAR or better vision. Odds ratios werecalculated to assess the risk for low cell counts and poor visual outcome.Results: The mean and standard deviation of the endothelial cell count was predictablyhigher in lamellar (2033 � 643 cells per mm2) than penetrating keratoplasties (898 � 380cells per mm2) and the mean post-operative cell count for grafts without post-operativecomplications was higher (1585 � 775) than those with complications (1312 � 747 cells permm2). Mean visual acuity was 0.25 � 0.20 and 0.29 � 0.41 for lamellar and penetratingkeratoplasties, respectively. A higher number of post-operative events was associated withan increased risk for graft rejection (odds ratio 2.40, p = 0.008, multiple logistic regression)and with poorer visual outcome (odds ratio 1.38, p = 0.044). After keratoplasty, the visualoutcome and endothelial cell density were not associated, either in penetrating or deepanterior lamellar grafts (p > 0.05).Conclusions: In keratoplasties performed for keratoconus, visual outcome cannot be pre-dicted reliably by post-operative endothelial cell density but post-operative events can lead topoorer visual outcomes in both types of grafts.

Submitted: 25 June 2012Revised: 6 September 2012Accepted for publication: 3 October 2012

Key words: cornea, endothelial cell density, keratoconus, keratoplasty, visual acuity

Keratoconus is a progressive degenerativedisease of the cornea that in its advancedstages can be expected to have significanteffect on visual acuity (VA).1,2 While thedisease is generally considered to developfrom abnormalities in the anterior cornea,the posterior cornea can be affected, forexample, by substantial posterior striaedeveloping as a secondary consequence ofstretching and deformation of the anteriorcornea. Some early studies on the posteriorendothelial cell layer in advanced casesof keratoconus reported the presence ofenlarged cells (for example, when cornealhydrops had developed),3 an increased vari-ation in cell area values,4 the presence ofvacuoles or blebs3 and that the endothelialcells could be stretched and show an

orientation toward the apex of the develop-ing cone.3,4

By modern-day perspectives, early surgicalintervention in cases of keratoconus (forexample, in the form of a full-thickness orpenetrating keratoplasty, PK) is likely beforethe posterior aspect of the cornea is affected.In PK, all layers of the central region of thecornea are replaced with a donor cornealbutton so, in all likelihood, it is relativelyunimportant that the corneal endotheliumcould be affected in moderate-to-advancedstages of keratoconus. There are alternativelamellar modes of corneal transplantation,which involve only partial thickness replace-ment of the cornea. One such procedure isdeep anterior lamellar keratoplasty (DALK),in which only the affected anterior cornea is

removed and the stroma dissected close to orright to Desçemet’s membrane. This leavesthe corneal endothelial cell layer of therecipient in situ, to which a donor cornealbutton, from which the endothelium hasbeen removed, is then affixed.5,6

From the perspective of anterior lamellarkeratoplasty as a common surgical proce-dure for the management of advancedkeratoconus, it seems useful to reconsiderthe status and role of the corneal endothe-lium. If the recipient endothelium is consid-ered to be essentially intact following asuccessful DALK, it could be hypothesisedthat this might result in patients havinga ‘healthier’ cornea post-operatively andeven perhaps an improved visual outcome.Corneal endothelial cell density (ECD) is

C L I N I C A L A N D E X P E R I M E N T A L

OPTOMETRY

© 2013 The Authors Clinical and Experimental Optometry 96.2 March 2013

Clinical and Experimental Optometry © 2013 Optometrists Association Australia 201

a commonly reported indicator of theoutcome for corneal grafts,7–12 along withVA.2,13–16 Both of these outcome variableshave been considered within the context ofgraft rejection.13,15,17,18

In this paper, we report on ongoingstudies and analyses that consider thecorneal endothelium following keratoplastyfor keratoconus, carried out at a singlesurgical centre in Glasgow, Scotland, withrespect to the visual outcome and thenumber of post-operative complications (oradverse events). Part of the rationale forundertaking the studies was to review thesecomplications and to assess whether anincreased number of adverse events couldbe linked to a loss of endothelial cells andpoorer visual outcome in those keratoplast-ies that are generally thought to have anunaffected, healthy endothelium prior tosurgery for keratoconus. A consecutive seriesof individuals that had been treated with PKand DALK was analysed.

METHODS

All subjects were patients attending forroutine follow-up of their corneal graft atthe Tennent Institute of Ophthalmology atGartnavel General Hospital in Glasgow. Thestudy adhered to the tenets of the Declara-tion of Helsinki. Inclusion criteria werea history of either PK or DALK for kerato-conus. Medical case records were reviewedand demographic and surgical informationextracted, including age, date of surgery,type of graft, history of post-operative com-plications and rejection episodes. Immuno-logic rejection was defined by clinical signs,including substantial corneal oedema, sub-epithelial infiltrates, anterior chamber reac-tion and presence of rejection lines. Surgicalprocedures were carried out by different sur-geons using local protocols. For DALK, the‘big bubble’ technique was used.19 This sur-gical procedure involves partial replacementof the corneal tissue with donor material,using the injection of air into the stroma toseparate the corneal layers. Anterior cornealtissue (epithelium and stroma) is removed,whereas Desçemet’s membrane and theendothelium remain in place. The post-surgical treatment usually consisted ofa course of topical chloramphenicol 0.5per cent eye drops and topical corticoster-oids (dexamethasone 0.1 per cent) thatwere gradually tapered over several months.Rejection episodes were treated intensivelywith frequent topical corticosteroids

(dexamethasone 0.1 per cent) every one totwo hours.

For this analysis, only clear grafts wereconsidered. Clear grafts were defined asthose that were functioning at the time ofassessment, that is, grafts that had no sub-stantial oedema and therefore, were suitablefor scanning-slit confocal microscopy. At thefollow-up visit, that was the basis for theseanalyses, all subjects underwent an ophthal-mic assessment according to local protocols,including VA, slitlamp biomicroscopy andintraocular pressure (IOP) by Goldmanntonometry. VA was obtained using a pinholeand standard Snellen charts. Snellen frac-tions were converted to logMAR valuesfor statistical analysis. Scanning-slit in vivoconfocal microscopy was carried out (Con-foscan 4, Nidek Technologies, Albignasego,Italy) following topical anaesthesia (MinimsProxymetacaine 0.5 per cent, Bausch &Lomb, Kingston upon Thames, UnitedKingdom) and application of a couplinggel (Viscotears, Novartis, Frimley, UnitedKingdom). Images were obtained from thecentral graft and stored on a computer. Theclearest image frame showing the cornealendothelium was selected for analysis, theendothelial cell areas were measured byplanimetry20 and the ECD was calculatedfrom the average area value. Initial notationof a post-operative event was countedand coded ‘1’ if it had occurred and ‘0’if it had not occurred. Events were countedonly once as first recorded in the casenotes. The main outcome variables werepost-operative ECD, VA and the number ofcomplications.

Statistical analysis was carried out usingthe STATA SE software package (version11.2, STATA Corporation, College Station,TX, USA). Descriptive statistics were gener-ated and differences assessed using appro-priate parametric and non-parametricmethods. Univariate simple and multiplelinear regressions were used to determinepredictors for the variables VA and ECD.The coefficient of determination R2 was cal-culated. Logistic regression models weregenerated to determine predicting factorsfor the following dichotomous dependentvariables: corneal graft rejection (1 = yes;0 = no); VA of worse than 0.3 (1 = yes; 0 = no)and endothelial cell count of 700 cells permm2 or less (1 = yes; 0 = no). Odds ratios(OR) including the 95 per cent confidenceintervals were generated. Subject age, grafttype, time since surgery, post-operativeECD and number of complication events

were included as independent variables. Forregression analyses where ‘ECD of 700 cellsper mm2 or less’ was the dependent dichoto-mous variable, continuous ECD was omittedfrom the independent variable list. Kaplan–Meier analysis was used to estimate the prob-ability of grafts to remain rejection-freeover time, to retain an ECD of greater than700 cells per mm2 and to achieve VA of 0.3logMAR or better. Linear regression analy-ses were undertaken using a Spearman rho,with a statistical trend being consideredsignificant at p < 0.05.

RESULTS

Patient group characteristicsAssessments were made of the post-operativeoutcome of corneal transplantation for kera-toconus on 76 eyes of 59 patients (mean ageof 37 � 12 years), who had been diagnosedwith moderate-to-severe keratoconus andwho had subsequently received a kerato-plasty. Just over half (57 per cent) of thesurgical procedures in this cohort wereperformed using PK and the rest were per-formed using DALK. The mean donorbutton size was 8.16 � 0.37 mm, the recipi-ent bed was generally 0.25 mm smaller; sizeinformation was missing for 12 grafts. Themean IOP at the follow-up visit was 14 �

3 mmHg.

Overall outcome measuresFor the two surgical procedures, a substan-tial difference in corneal ECD was expected.Initially, following both modes of surgicalintervention, the endothelial cells appearedslightly to substantially enlarged. Figures 1Aand 1B show endothelia following DALKand PK at six months after surgery. At longerpost-operative times this difference becameless substantial. Figures 2A and 2B illustrateendothelia following DALK and PK a fewyears after surgery. The endothelial cellswere considerably enlarged following bothsurgical modalities but still retained a celldensity of 1,000 cells per mm2 (Figure 2).

Post-operative eventsPost-operative events were frequently notedfor the cohort studied. At least one eventwas noted for 71 per cent and multipleevents were noted for 46 per cent of indi-viduals included in the analyses. Most ofthese events were self-limiting or resolvedfollowing medical intervention. Some events

Corneal endothelium in keratoconus Jonuscheit, Doughty and Ramaesh

Clinical and Experimental Optometry 96.2 March 2013 © 2013 The Authors

202 Clinical and Experimental Optometry © 2013 Optometrists Association Australia

required minor surgical intervention, suchas suture removal (for broken sutures) orastigmatic keratotomy (for persistent highastigmatism). A list of post-operative com-plications is provided in Table 1. In brief,commonly encountered events were suture-

related and included loose, broken orburied sutures (38 per cent of grafts).Twenty-three grafts were noted to haveepithelial defects. Relatively common werealso Desçemet’s membrane folds (11 cases),corticosteroid-induced increase in IOP (10

cases) and high persistent astigmatism (ninecases).

Sixteen grafts in this cohort experiencedat least one episode of rejection and typi-cally, this affected patients who had receiveda penetrating graft (15 cases). Analysis ofcases of PK using univariate logistic re-gression (with rejection as the dependentdichotomous variable; ‘yes = 1’ and ‘no = 0’)indicated that a greater cumulative numberof post-operative events was associated withan increased probability of immunologicrejection (OR 2.02, p = 0.008; 95 per cent CI:1.20 to 3.38). Using multiple logistic regres-sion and controlling for subject age, timesince surgery and ECD as additional inde-pendent variables provided a higher oddsratio of 2.40 (p = 0.008) for the dependent(outcome) variable ‘Complication count’than in simple logistic regression. For the15 individuals who were treated with PKthat developed rejection, the probability toremain rejection-free was 95 per cent withinyear one after surgery. At the last observa-tion, at 34 years (or 411 months) post-operatively these patients had a probabilityof 37 per cent to have remained rejection-free (Figure 3).

Corneal endothelium in relationto post-operative eventsThe mean ECD for all 76 grafts was 1,391 �

760 cells per mm2 with a substantial rangefrom a high of 2,941 to the lowest value of452 cells per mm2. As expected, the meanendothelial cell densities were significantlydifferent between the two types of grafts(p < 0.001) with a mean of 2,033 � 643 and898 � 380 cells per mm2 for DALK and PK,respectively.

The ECD can be expected to vary withincreasing post-operative time and the timesince surgery was significantly associatedwith ECD (R2 = 0.31, p < 0.001, univariatelinear regression). One objective of thisanalysis was to evaluate if the decline in ECDin relation to the time after surgery couldbe predicted, based on the occurrence ofpost-operative complications. This analysis isshown in Figure 4, comparing the ECD ofcorneal grafts, for which one or more eventshad been noted (Figure 4A) and thosewith no recorded post-operative events(Figure 4B). The mean endothelial cell den-sities were 1,585 � 775 for grafts withoutcomplications and 1,312 � 747 cells per mm2

for those grafts that had one or more com-plications; this difference was not statisticallysignificant (p = 0.14, Wilcoxon rank-sum

Figure 1. Post-operative endothelial confocal images six months after deep anteriorlamellar keratoplasty (A) and penetrating keratoplasty (B)

Figure 2. Post-operative endothelial confocal images five years after deep anteriorlamellar keratoplasty (A) and penetrating keratoplasty (B)




test). When the type of graft was considered,the occurrence of post-operative eventsappears to have less impact on the graftendothelia following DALK as compared toPK (Table 2). For example, even in lamellargrafts with events, the mean ECD at the timeof the assessment was still 2,049 cells permm2 compared to penetrating grafts (withevents) with a mean ECD of just 843 cellsper mm2.

No significant relationship could be iden-tified between the number of post-operativeevents and ECD, despite a statistically signi-ficant overall difference in ECD betweenkeratoplasties with and without events (p <0.001). Multiple regression analysis con-firmed the type of graft as the main predictorfor ECD (p < 0.001).

Eleven grafts (10 PK, one DALK) had anECD of 700 cells per mm2 or less, with a meancell count of just 574 � 73 cells per mm2

(range 452 to 672). Kaplan–Meier analysisshowed that during the first post-operativeyear, the probability for grafts to retain 700or more endothelial cells per mm2 was nearly100 per cent for both penetrating and lamel-lar grafts, but for the PK this probabilitydeclined (as expected) over time with aprobability of these grafts to retain at least700 cells per mm2 being about 34 per cent at34 years after surgery (Figure 5).

Visual outcome in relation to ECDand post-operative eventsThe mean VA for all grafts was 0.28 � 0.33.Seventy-eight per cent of grafts achieved a

0 100 200 300 400Time (months)

0.00

0.25

0.50

0.75

1.00

Pro

babi

lity

Figure 3. Kaplan–Meier plot illustrating the probability of apenetrating keratoplasty remaining rejection-free over time

0 100 200 300 400 0 100 200 300 400

3,000

2,000

1,000

0

EC

D

One or more complications No complications

Time since surgery (months)

Figure 4. Scatter plots showing endothelial cell density (ECD) asa function of post-operative time for grafts with and withoutpost-operative events

Type of complication Frequency all grafts(n = 76)

Frequency in PK(n = 43)

Frequency in DALK(n = 33)

Broken or loose suture 29 16 13Epithelial defects 23 14 9Rejection 16 15 1Desçemet’s folds 11 6 5Steroid-induced increase inIOP

10 5 5

Persistent astigmatism 9 6 3Sub-epithelial infiltrates 7 6 1Endothelial pigment 6 6 0Graft neovascularisation 5 5 0Graft oedema 4 1 3Incomplete information 4 4 0DM perforation 3 n/a 3Stromal haze 3 2 1Iritis 2 2 0Stromal scarring 2 2 0Trauma 2 2 0Anterior chamber activity 1 1 0Bullous keratopathy 1 1 0Herpes keratitis 1 1 0Repeat graft 1 1 0

DALK: deep anterior lamellar keratoplasty, PK: penetrating keratoplasty, IOP: intra-ocular pressure,DM: Desçemet’s membrane

Table 1. Frequency of post-operative complications




VA of 0.3 or better (6/12). The mean VAin this group was 0.15 � 0.12. On average,DALK (0.25 � 0.20) appeared to provide abetter visual outcome than PK (0.29 � 0.41)but the difference was not statistically signifi-cant (p = 0.797, Wilcoxon rank sum test).The grafts with a very low ECD (less than 700cells per mm2) were still functional basedon assessments of the mean logMAR VA,which was 0.25 � 0.30 on average (range -0.1to 0.8).

The sub-group analyses suggest that therewas very little difference in post-operative VA(Table 3). While two grafts had a rather poor

visual status at the time of the assessment,no association could be detected betweenVA and ECD (Spearman’s rho = -0.0217,p = 0.852). Analysing cases treated with pen-etrating and lamellar keratoplasty separatelyprovided similar results (Figure 6A; rho =-0.468, p = 0.766 and Figure 6B; rho = -0.038,p = 0.833). In addition, removing the twooutliers did not result in a significant asso-ciation between ECD and VA (rho = -0.021,p = 0.861).

Univariate linear regression suggestedthat a greater number of post-operativeevents was weakly associated with lower VA

(R2 = 0.10, p = 0.005). At the univariate level,VA was not predicted by graft type, timesince surgery or ECD (p > 0.05). Multipleanalysis with graft type, time since surgeryand ECD as additional independent vari-ables confirmed this finding; the numberof post-operative events remained the onlysignificant predictor of VA (p = 0.004).With univariate logistic regression, a similartrend was noted for a greater total numberof post-operative events to be associatedwith a higher risk for having VA of poorerthan 0.3 and this association was just statis-tically significant (OR 1.34, p = 0.042, 95per cent CI: 1.01 to 1.79). The trend wasconfirmed when multiple logistic regressionwas used, in which the time since surgery,ECD and type of graft were taken intoaccount (OR 1.38, p = 0.044, 95 per cent CI:1.01 to 1.89).

Kaplan–Meier analysis showed that withinthe first year, the two types of grafts had aprobability of 95 (PK) and 93 per cent(DALK) to have VA of 0.3 or better. Forpatients who had received a lamellar kerato-plasty, this probability was 49 per cent fiveyears after the procedure (last observation).Individuals who had received penetratinggrafts had about a 43 per cent chance ofhaving VA of 0.3 or better 34 years aftersurgery (Figure 7).

DISCUSSION

This study confirms that visual outcome ofkeratoplasty for keratoconus cannot be reli-ably predicted by ECD. The data and analy-ses reported here are part of an ongoingassessment of cases with advanced kerato-conus undergoing keratoplasty and specifi-cally, graft outcomes at a single centre inGlasgow.

Corneal grafts for keratoconus have agood long-term prognosis with a recentstudy reporting 80 per cent of penetratinggrafts still being clear, with nearly half ofthe eyes examined having VA of 0.3 (6/12)or better at (on average) 27 years aftersurgery.21 For grafts carried out in the UK, ithas been reported recently that patientsreceiving penetrating grafts were more likelyto achieve better vision after surgery thanrecipients of lamellar grafts,2 so there isclearly an ongoing need for further analyses,if DALK is to be considered as a superiorapproach to PK.

The analyses of the present cohortindicate that a greater number of post-operative events in a single corneal graft

Number ECD (cells per mm2)

Mean � SD Median Minimum Maximum

All grafts 76 1,391 � 760 1,083 452 2,941All grafts (no events) 22 1,585 � 775 1,411 563 2,941All grafts (with events) 54 1,312 � 747 1,029 452 2,762DALK 33 2,033 � 643 2,041 514 2,825DALK (no events) 12 2,005 � 588 1,933 1,049 2,825DALK (with events) 21 2,049 � 686 2,105 514 2,762PK (all grafts) 43 898 � 380 851 452 2,941PK (no events) 10 1,081 � 678 901 563 2,941PK (with events) 33 843 � 216 851 452 1,250

ECD: endothelial cell density, SD: standard deviation, DALK: deep anterior lamellar keratoplasty,PK: penetrating keratoplasty

Table 2. Endothelial cell density with respect to post-operative events

Number VA

Mean � SD Median Minimum Maximum

All grafts 76 0.28 � 0.33 0.20 -0.10 1.80All grafts (no events) 22 0.20 � 0.18 0.20 -0.10 0.60All grafts (with events) 54 0.31 � 0.37 0.20 -0.10 1.80DALK 33 0.25 � 0.20 0.20 -0.10 0.80DALK (no events) 12 0.18 � 0.15 0.20 -0.10 0.50DALK (with events) 21 0.30 � 0.20 0.20 0.00 0.80PK (all grafts) 43 0.29 � 0.41 0.30 -0.10 1.80PK (no events) 10 0.22 � 0.22 0.20 -0.10 0.60PK (with events) 33 0.32 � 0.45 0.20 -0.10 1.80

VA: visual acuity, SD: standard deviation, DALK: deep anterior lamellar keratoplasty, PK: penetratingkeratoplasty

Table 3. VA logMAR with respect to post-operative events in grafts




can be associated with a greater risk forcorneal graft rejection. The occurrence ofcomplications observed during the initialtwo years after PK is associated with graftfunction at 10 years22 and while this studyincluded various indications for PK, theanalyses indicated that half of the graftsthat had complications within the firsttwo years had failed after 10 years. Anotherrecent study concluded that post-operativerisk factors play a greater role in long-termgraft survival than pre-operative factors.23

Only clear corneal grafts were analysed inthe present study to assess whether com-monly used indicators of graft outcome(VA and ECD) could be shown to differin relation to post-operative events. Theoverall finding is novel and underpinsthe importance of continuous and regularmonitoring and counselling of cornealgraft patients, especially those with PK.The results further extend the outcome ofprevious studies that reported the occur-rence of one or more complications is asso-ciated with an increased risk of graft failure,whereas specific types of complicationswere not related to risk for graft failure.24

Some of the complications noted inthe present study could be considered lesssevere than others, such as Desçement’sfolds and endothelial pigmentation.Excluding these from the overall complica-tions did not affect substantially the resultsof the analyses. For example, the p-values ofunivariate regression analysis of the effectof the number of complications on ECDare very similar (that is, p = 0.358, whenfolds and pigment are included and p =0.301, without folds and pigment). Simi-lar minimal changes to the results wereobserved on the effect of the complicationcount on VA (that is, p = 0.005 versus p =0.004), indicating that excluding theseminor complications does not affect theoverall outcome.

A number of previous cohort studies9,12,21

that examined between 13 and 61 eyes havereported functional grafts with endothelialcell densities below a suggested lower limitof 700 cells per mm2 and it is this limitthat was selected in the present ongoinganalyses. The mean ECD in the other studiesranged from 509 to 894 cells per mm2, whichis lower than the overall ECD in the presentcohort (of 1,391 cells per mm2). A range ofmodels has been proposed to describe thedecline, including a bi-exponential modelwith a rapid loss of cells during the initialpost-operative period and a slower decline

0 100 200 300 400Time (months)

0.00

0.25

0.50

0.75

1.00PKDALK

Pro

babi

lity

Figure 5. Kaplan–Meier plot illustrating the probability of grafts to retain an endothelialcell density of 700 cells per mm2 or greater over timePK: penetrating keratoplasty, DALK: deep anterior lamellar keratoplasty

0 1,000 2,000 3,000 0 1,000 2,000 3,000

DALKPK

ECD

0

0.5

1

1.5

2

VA

logM

AR

Figure 6. Scatter plot showing VA in relation to post-operative endothelial cell density(ECD) for penetrating keratoplasty (A) and deep anterior lamellar keratoplasty (B)

0 100 200 300 400Time (months)

0.00

0.25

0.50

0.75

1.00

PKDALK

Pro

babi

lity

Figure 7. Kaplan–Meier plot illustrating the probability of grafts achieving visual acutity(VA) of 0.3 or better over timePK: penetrating keratoplasty, DALK: deep anterior lamellar keratoplasty




thereafter.25 The results from the presentcohort (Figure 4) broadly support thiseffect, especially for penetrating grafts.

From early26 to more recent27 studies onthe endothelium of corneal grafts, cor-neal transparency has been considered tobe dependent upon endothelial function,although the mechanism by which this isachieved still remains unclear.28,29 During PKespecially, the endothelium can be compro-mised by surgical trauma associated withcorneal button preparation, handling andsuturing. The same does not apply to thedeep anterior lamellar mode of transplanta-tion, yet several of the grafts assessed in thepresent study showed much lower thanexpected endothelial cell densities. Varioussub-group analyses have been undertaken toassess possible reasons. For example, a smallnumber of grafts experienced an episodeof ocular hypertension associated withintensive corticosteroid treatment of post-operative events, including possible onset ofrejection. Two of these patients had receiveda lamellar graft, where the endothelial celldensities were just 514 and 1,045 cells permm2 at 11 and 27 months post-operatively.A similar ECD could be found followingPK, for which a steroid-induced increase inIOP had been recorded. When evaluatedat follow-up, all of these 10 eyes (five casesof DALK and five cases of PK) had IOPwithin normal limits (DALK 14 � 3 mmHg,PK 15 � 3 mmHg). The mean ECD waslower (1,206 � 703 cells per mm2) than thatof the overall cohort. The same trend wasnoted when analysing the data with respectto graft type (ECD for lamellar grafts withraised IOP was 1,591 � 796 cells per mm2

and for penetrating grafts was 820 � 325cells per mm2). These post-operative eventsrequire further investigation on a suffi-ciently large cohort that will allow for analy-ses as to whether or not there is a statisticallysignificant effect (with controlled post-operative time).

The analyses and the overall outcome ofthe present study are limited by the relativelysmall number of patients evaluated and bythe imbalance in the post-operative assess-ment period. The study is considered usefuland valid to determine the extent by whichgraft outcome (as functional vision) can bepredicted by the status of the endothelialcells.

In summary, the analyses of ECD in rela-tion to VA indicate poor inter-dependencebetween the two variables. It can be con-cluded that corneal transparency is not obvi-

ously dependent on endothelial status, atleast as assessed by ECD. In this study, it wasnot possible to predict VA from ECD, norcould it be shown that a decline in ECDpreceded a reduction in VA. Furtherstudies on these inter-relationships would beuseful, including more detailed analyses ofendothelial morphology.

ACKNOWLEDGEMENTSThis study was supported by an ophthalmol-ogy research grant from The Royal Collegeof Surgeons, Edinburgh. The authors thankDr Claudia Geue, University of Glasgow, forassistance with the statistical analysis.

The authors have no proprietary interestin any of the instrumentation or softwareused in these studies and no competinginterests.

REFERENCES1. Rabinowitz YS. Keratoconus. Surv Ophthalmol 1998;

42: 297–319.2. Jones MN, Armitage WJ, Ayliffe W, Larkin DF,

Kaye SB. Penetrating and deep anterior lamellarkeratoplasty for keratoconus: a comparison of graftoutcomes in the United Kingdom. Invest Ophthal-mol Vis Sci 2009; 50: 5625–5629.

3. Laing RA, Sandstrom MM, Berrospi AR, LeibowitzHM. The human corneal endothelium in kerato-conus: A specular microscopic study. Arch Ophthal-mol 1979; 97: 1867–1869.

4. Matsuda M, Suda T, Manabe R. Quantitative analy-sis of endothelial mosaic pattern changes inanterior keratoconus. Am J Ophthalmol 1984; 98:43–49.

5. Luengo-Gimeno F, Tan DT, Mehta JS. Evolution ofdeep anterior lamellar keratoplasty (DALK). OculSurf 2011; 9: 98–110.

6. Reinhart WJ, Musch DC, Jacobs DS, Lee WB,Kaufman SC, Shtein RM. Deep anterior lamellarkeratoplasty as an alternative to penetrating kerato-plasty a report by the American Academy of Oph-thalmology. Ophthalmology 2011; 118: 209–218.

7. Abbott RL, Fine M, Guillet E. Long-term changes incorneal endothelium following penetrating kerato-plasty. A specular microscopic study. Ophthalmology1983; 90: 676–685.

8. Matsuda M, Bourne WM. Long-term morphologicchanges in the endothelium of transplantedcorneas. Arch Ophthalmol 1985; 103: 1343–1346.

9. Kus MM, Seitz B, Langenbucher A, Naumann GO.Endothelium and pachymetry of clear cornealgrafts 15 to 33 years after penetrating keratoplasty.Am J Ophthalmol 1999; 127: 600–602.

10. Bourne WM. Cellular changes in transplantedhuman corneas. Cornea 2001; 20: 560–569.

11. Inoue K, Kimura C, Amano S, Oshika T, Tsuru T.Corneal endothelial cell changes twenty years afterpenetrating keratoplasty. Jpn J Ophthalmol 2002; 46:189–192.

12. Basak SK. Low endothelial cell count and clearcorneal grafts. Indian J Ophthalmol 2004; 52: 151–153.

13. Williams KA, Kelly TL, Lowe MT, Coster DJ. Theinfluence of rejection episodes in recipients of

bilateral corneal grafts. Am J Transplant 2010; 10:921–930.

14. Larkin DF, Mumford LL, Jones MN. Centre-specific variation in corneal transplant outcomes inthe United kingdom. Transplantation 2011; 91:354–359.

15. Coster DJ, Williams KA. The impact of cornealallograft rejection on the long-term outcome ofcorneal transplantation. Am J Ophthalmol 2005; 140:1112–1122.

16. Vail A, Gore SM, Bradley BA, Easty DL, Rogers CA,Armitage WJ. Conclusions of the corneal trans-plant follow up study. Collaborating Surgeons.Br J Ophthalmol 1997; 81: 631–636.

17. Tham VM, Abbott RL. Corneal graft rejection:recent updates. Int Ophthalmol Clin 2002; 42: 105–113.

18. Panda A, Vanathi M, Kumar A, Dash Y, Priya S.Corneal graft rejection. Surv Ophthalmol 2007; 52:375–396.

19. Anwar M, Teichmann KD. Big-bubble technique tobare Descemet’s membrane in anterior lamellarkeratoplasty. J Cataract Refract Surg 2002; 28: 398–403.

20. Jonuscheit S, Doughty MJ, Ramaesh K. In vivoconfocal microscopy of the corneal endothelium:comparison of three morphometry methods aftercorneal transplantation. Eye (Lond) 2011; 25: 1130–1137.

21. Jensen LB, Hjortdal J, Ehlers N. Long termfollow-up of penetrating keratoplasty for kerato-conus. Acta Ophthalmol 2010; 88: 347–351.

22. Claesson M, Armitage WJ. Ten-year follow-up ofgraft survival and visual outcome after penetratingkeratoplasty in Sweden. Cornea 2009; 28: 1124–1129.

23. Anshu A, Lim LS, Htoon HM, Tan DT. Post-operative risk factors influencing corneal graftsurvival in the Singapore Corneal TransplantStudy. Am J Ophthalmol 2011; 151: 442–448 e1.

24. Wagoner MD, Ba-Abbad R, Al-Mohaimeed M,Al-Swailem S, Zimmerman MB. Postoperative com-plications after primary adult optical penetratingkeratoplasty: prevalence and impact on graft sur-vival. Cornea 2009; 28: 385–394.

25. Armitage WJ, Dick AD, Bourne WM. Predictingendothelial cell loss and long-term corneal graftsurvival. Invest Ophthalmol Vis Sci 2003; 44: 3326–3331.

26. Irvine AR, Irvine AR, Jr. Variations in normalhuman corneal endothelium; a preliminary reportof pathologic human corneal endothelium. AmJ Ophthalmol 1953; 36: 1279–1285.

27. Edelhauser HF. The balance between cornealtransparency and edema: the Proctor Lecture.Invest Ophthalmol Vis Sci 2006; 47: 1754–1767.

28. Bonanno JA. Identity and regulation of ion trans-port mechanisms in the corneal endothelium. ProgRetin Eye Res 2003; 22: 69–94.

29. Leung BK, Bonanno JA, Radke CJ. Oxygen-deficient metabolism and corneal edema. ProgRetin Eye Res 2011; 30: 471–492.




the corneal endothelium after keratoplasty for keratoconus

Documents