characteristics of the corneal endothelium and pseudoexfoliation syndrome in patients with senile...

Original Article

Characteristics of the corneal endothelium andpseudoexfoliation syndrome in patients withsenile cataract_ 449..455

Lorena Quiroga MD,1 Van C Lansingh MD,1,2 Margarita Samudio PhD,3 Fernando Y Peña MD1 andMarissa J Carter PhD4

1Fundación Visión; 2Fundacion Hugo D. Nano, Buenos Aires, Argentina; 3National University of Asuncion, Asuncion, Paraguay; and4Strategic Solutions, Inc., Cody, Wyoming, USA

ABSTRACT

Background: The characteristics of the corneal endot-helium in patients with/without pseudoexfoliationsyndrome (PEX) with senile cataract in Paraguay areunknown.

Methods: Endothelial density, mean coefficient ofvariation (%) in cell size and mean cell hexagonality(%) were measured using automated specularmicroscopy. Multivariate general linear model analy-sis was used to determine the effect of age, genderand PEX on endothelial characteristics. Analysisof variance tests analysed the effect of age onendothelial variables without regard to PEX status.Categories of <2000 and �2000 cells/mm2 werecompared in the presence/absence of PEX using ageas covariate to calculate odds ratios for cornealdecompensation.

Results: Out of 468 eligible patients, 51 wereexcluded. Sixty-one patients had PEX. Unadjustedmean endothelial cell density was 2451 cells/mm2;cell size coefficient of variation was 34.3; and themean percentage of hexagonal cells was 57.7.Adjusting for age, only the mean difference betweenthe groups with and without PEX for endothelial celldensity was significant (PEX = 2315, no PEX = 2482,P = 0.002). Of the total study population, at-risk

endothelial cell densities were found in 46 eyes(11%); and 13 (28%) of those were found to havePEX. When PEX was present, the calculated oddsratio for corneal decompensation following surgerywas 1.90 after adjustment for age.

Conclusions: Endothelial cell density data were con-sistent with published literature. Because Paraguayhas a high prevalent of PEX, it is suggested thatspecular microscopy screening be carried out for allpatients scheduled for intraocular surgery wherefeasible.

Key words: cataract surgery, endothelial characteri-stic, Paraguay, pseudoexfoliation syndrome, specularmicroscopy.

INTRODUCTION

The transparency of the cornea relies on the stromaremaining relatively dehydrated, for which purposethe metabolic operation of the endothelium isessential. The endothelium consists of a monolayerof hexagonal cells of limited regenerative power ina very vulnerable anatomical location.1 The meandensity of endothelial cells considered normalfor adults is approximately 2500 cells/mm2, withcorneal oedema and decompensation occurringwhen it falls below 500 cells/mm2.2

� Correspondence: Dr Marissa J Carter, Strategic Solutions, Inc., 1143 Salsbury Avenue, Cody, WY 82414, USA. Email: mcarter@

strategic-solutions-inc.com

Received 25 August 2009; accepted 8 March 2010.

Author declarations: Fundación Visión funded the study; LQ and FYP are full-time employees of Fundación Visión; and VCL is a consultant

to Fundación Visión.

Clinical and Experimental Ophthalmology 2010; 38: 449–455 doi: 10.1111/j.1442-9071.2010.02313.x

© 2010 The AuthorsJournal compilation © 2010 Royal Australian and New Zealand College of Ophthalmologists

During cataract surgery, factors such as cornealdistortion, contact of lens fragments or the intraocu-lar lens with the endothelium, or the release of freeradicals can provoke loss of endothelial tissue. Insome cases, especially if the function of the endothe-lium has been compromised before surgery, the losscan generate irreversible corneal oedema.2

Even aside from outside factors such as trauma orsurgery, or pathological factors such as glaucoma, itis known that cell density and the morphologicalcharacteristics of the corneal endothelium are depen-dent on age. It has been estimated that the endothe-lial cell density diminishes at a rate of 0.5% annuallybeginning at birth, a rate that can increase to 2%after intraocular surgery.3,4 The physiological loss ofendothelial cells with age can also accelerate inresponse to factors, such as pseudoexfoliation syn-drome (PEX), which places endothelial function andtransparency of the cornea at risk following intrao-cular surgery.

The PEX is an age-related condition characte-rized by the deposit of anomalous filamentousmaterial upon some eye structures.5 The presence ofthese filaments alerts the ophthalmologist beforeintraocular surgery that a higher than normal riskexists for surgical complications, such as zonulardehiscence, posterior capsule rupture or loss of thevitreous.6

It has been demonstrated by specular or electronmicroscopy that morphological and quantitativechanges occur in patients with PEX, leading to thehypothesis that a specific pathology of the endo-thelium exists, which increases the vulnerability ofendothelial cells to intraocular surgery.1 The currentstudy proposes to describe the characteristics of thecorneal endothelium in patients with and withoutPEX in a population over 50 years of age with senilecataract.

METHODS

Patient selection

Fundación Visión is an ophthalmology clinicthat treats approximately 60 000 primarily lowincome patients and performs 2000 cataract surger-ies per year in Asuncion, Paraguay and the sur-rounding region. The selection of patients for thisstudy was performed based on a diagnosis of senilecataract with indications for surgery, without regardto the presence or absence of systemic illnesses.The study was approved by the Fundación VisiónEthics Committee, and was performed in accor-dance with the ethical standards laid down inthe 1964 Declaration of Helsinki. All patientssigned a consent form before their inclusion inthe study.

Inclusion and exclusion criteria

Male and female patients were eligible for the studyif they were age 50 or older and were diagnosed withsenile cataract with indications for cataract surgery,and on whom semiautomated specular microscopyhad been performed. These patients presented con-secutively at Fundación Visión between January2005 and April 2006.

Excluded from the study were patients diagnosedas having congenital, metabolic or traumatic cataract;a history of eye trauma or prior intraocular surgery;a history or sequelae of uveitis; a history of chronicuse of contact lenses; a biomicroscopic examina-tion showing oedema, guttata, keratic precipitates,corneal dystrophies, extensive leukomas, ocularhypertension or glaucoma; and diabetes.

Patients presenting with an endothelial celldensity varying by more than 10% in three measure-ments were excluded. For the remaining patients,two of the three measurements that were closesttogether were chosen.

Examinations

A complete ophthalmological exam was performed,which included biomicroscopic assessment of theanterior segment with and without dilation, mea-surement of intraocular pressure, and fundoscopyusing a slit-lamp and hand-held 90-dioptre lens, orbinocular indirect ophthalmoscope. Patients wereconsidered to demonstrate pseudoexfoliation whenbiomicroscopic examination of the anterior segmentunder dilation showed the characteristic grayish-white exfoliative material on the anterior capsule ofthe lens or pupillary margin in the eye to be operatedon for cataract.

Specular microscopy

Only one eye from each patient was included in thestudy, with the first eye to be operated on chosenin cases of bilateral cataract. As a result, the numberof eyes studied was equal to the number of patientswho participated. Specular microscopy was per-formed on each of these eyes by a single person usinga semiautomatic non-contact Konan Noncon RoboPachy SP-9000 specular microscope (Konan MedicalInc., Hyogo, Japan).

Before beginning the study, the intratest vari-ability of the specular microscope was determined.Variability was calculated by performing 10 mea-surements on a single patient’s eye using the sameobserver.

The method consisted of taking three digitalphotographs of the central cornea (within a radius of3 mm from the centre). Digital images were analysed

450 Quiroga et al.


only if the resolution of the image was clear, withthe cell border being distinguishable (area of endot-helium imaged was 400 ¥ 220 mm). If the imageswere not clear, these cases were excluded. To reduceerrors due to intratest variability of the equipment/operator, the two measurements of endothelialparameters nearest in value to each other were used.The mean, standard deviation (SD) and intratestvariability for each eye parameter tested were thendetermined. (Intratest variability was the relationbetween the SD and the mean multiplied by 100,with variability of 10% or less being consideredacceptable. Those patients showing greater variationwere excluded.)

Using the mouse, the centre of each cell wasmarked, beginning with a cell from the centralregion and continuing with the cells adjacent to itand then others adjacent to those already marked ina counterclockwise direction until a minimum of 60cells were marked. Once the cells had been marked,the team calculated the endothelial parameters andgenerated the results.

Parameters studied

Of the parameters generated by the equipment(automatically calculated by the software), onlythree (endothelial cell density, coefficient of variation[CV] in cell size and percentage of hexagonality)were used in this study. The endothelial cell density,or endothelial density, is the number of cells presentin a 1-mm2 area, and is calculated by the specularmicroscope. Variation in cell size is the estimateddegree of polymegathism (variation of endothelialcell size) and was calculated as SD/X where SD is thestandard deviation and X is the mean cell size.

The percentage of hexagonal cells is the percent-age of cells that have six edges, and is calculated bythe specular microscope. This also permits determi-nation of the degree of pleomorphism (variation incell shape) by deduction, as the greater the degree ofhexagonality present the less pleomorphism there is,and vice versa.

Statistical analysis

Data from the case history and specular micro-scopy were recorded on an Excel v8.0 spreadsheet(Microsoft, Redmond, WA, USA) and were analysedusing SPSS v.16 (SPSS, Inc., Chicago, IL, USA). Therelationships between gender and age and preva-lence of PEX were examined separately by the Fish-er’s exact test and Kendall’s tau-c, respectively (agewas categorized into four groups: 50–59 years; 60–69years; 70–79 years; and �80 years). Multivariategeneral linear model (GLM) analysis was used to

determine the effect of age, gender and PEX onendothelial characteristics simultaneously (endo-thelial cell density, cell size CV and percentage ofhexagonal cells). Separate one-way analysis of vari-ance (ANOVA) tests analysed the effect of age on thethree endothelial variables without regard to PEXstatus. A univariate GLM analysis was also con-ducted to study the effect of pertinent variables onendothelial cell density.

To identify the corneas most susceptible to decom-pensation after intraocular surgery, corneas wereconsidered to be at risk when the endothelial celldensity was less than 2000 cells/mm2, as proposed inthe literature.4 Categories of <2000 and �2000 cells/mm2 were compared in the presence or absence ofpseudoexfoliation using the Fisher’s exact test (one-sided) and the relationship between age and corneasat risk was examined using the gamma statistic. Alogistic regression was performed to calculate oddsratios (ORs).

A statistically significant result was consideredwhen the probability was <0.05.

RESULTS

Out of 468 eligible patients, 51 (10.9%) wereexcluded from the study with the following reasons(exclusion criteria): 36 (uveitis, glaucoma, cornealanomaly, age <50 years); 4 (full medical charts notavailable); and 11 (>10% variation in endothelialparameters).

The study included 417 eyes of 417 patients whowere diagnosed with cataract. The mean � SD age ofthe patients was 70.7 � 9.36 years (range: 50–94years); and 239 (57.3%) were female (Table 1). Of allthese patients, 61 (14.6%) presented with PEX, andof those, 50.8% were men, indicating that genderhad no effect on PEX prevalence (Fisher’s exact test,P = 0.207). The prevalence of PEX with regard to age,

Table 1. Demographic and clinical characteristics of the studypopulation

Characteristic Value

GenderFemale n (%) 239 (57.3)Male n (%) 178 (42.7)

Age (years)Range 50–94Mean � standard deviation 70.7 � 9.36

Distribution of ages (total = 417)50–59, n (%) 57 (13.7)60–69, n (%) 125 (30.0)70–79, n (%) 160 (38.4)�80, n (%) 75 (18.0)

Frequency of pseudoexfoliation, n (%) 61 (14.6)

Pseudoexfoliation syndrome and cataract 451


however, showed that maximum prevalence wasreached beginning with the 70- to 79-year age group(Table 2) (P < 0.001).

The intratest variability results of the specularmicroscope were an endothelial cell density of 2.5%,CV of cells of 10.5% and percentage of hexagonalcells of 10.4%.

The mean endothelial cell density was 2451 �390 cells/mm2, the mean CV of the cell size 34.3 �5.39% and the mean percentage of hexagonal cells57.7 � 8.17% (Table 3). Endothelial cell densitydecreased in regard to age with the following meandifferences between age groups significant: 50–59and 70–79 years, 191 (P = 0.003); 50–59 and �80years, 247 (P = 0.001); 60–69 and 70–79 years, 147(P = 0.005); and 60–69 and �80 years, 204 (P =0.003). There was no obvious relationship betweenage and the other two variables; and ANOVA resultswere not significant in either instance.

In the multivariate analysis (Appendix I), adjust-ing for age, only the mean difference between thePEX and non-PEX groups for endothelial cell densitywas significant (PEX = 2315, no PEX = 2482, P =0.002; Table 4; differences between sex were non-significant). In the univariate analysis, only age and

the interaction between sex and PEX were signi-ficant (model R2 = 0.082). After adjusting for age,whereas non-PEX women had lower mean endo-thelial cell densities compared with non-PEX men(women: 2442, men: 2522), PEX men had lowerendothelial cell densities compared with PEXwomen (women: 2368, men: 2262). We studiedthe relationship between age and endothelial celldensity to determine if non-linear models wouldprovide a better fit but determined that power andcubic equations only increased R2 by 0.02, indicatinglittle difference.

Of the total study population, at-risk endothelialcell densities were found in 46 eyes (11%); and ofthose, 13 (28%) were found to have PEX (Table 5)(Fisher’s exact test, one-sided, P = 0.008). The fre-quency of corneas with at-risk endothelial cell den-sities increased with age as follows: 4% in thoseaged 60–69 years; 16.3% in those aged 70–79 years;and 20% in those aged 80 years or more (P < 0.001).At-risk endothelial cell densities (<2000 cells/mm2)were found in 21.3% of patients with PEX, but only9.3% of patients without PEX (Table 5). The logisticregression results indicated that the OR for risk ofdecompensation with regard to age was 0.92 (95%confidence interval: 0.889–0.960). The OR for risk ofdecompensation with regard to PEX after adjustmentfor age was 1.90, but this was not significant (95%confidence interval: 0.911–3.980).

DISCUSSION

This study describes the endothelial characteris-tics of a Paraguayan population with senile cataract,and demonstrates that reduced endothelial celldensities are found in patients with PEX. The

Table 2. Prevalence of PEX by age (%)

Age (years) PEX No PEX Total

50–59 1 (2) 56 (98) 5760–69 11 (9) 114 (91) 12570–79 33 (21) 127 (79) 160>80 16 (21) 59 (79) 75Total 61 356 417

PEX, pseudoexfoliation syndrome.

Table 3. Characteristics of the corneal endothelium in the study population broken down by age (N = 417)

Characteristic Mean � SD Age � SD (years)

50–59 60–69 70–79 �80

ECD (cells/mm2) 2451 � 390.1 2582 � 321 2539 � 334 2391 � 413 2335 � 423CS CV (%) 34.3 � 5.39 34.7 � 5.97 33.6 � 4.78 34.1 � 5.31 35.6 � 5.90HC (%) 57.7 � 8.17 56.9 � 9.81 57.4 � 7.39 57.9 � 8.15 58.4 � 8.15

CS, cell size; CV, coefficient of variation; ECD, endothelial cell density; HC, hexagonal cells; SD, standard deviation.

Table 4. Endothelial characteristics by gender in patients with and without pseudoexfoliation syndrome (PEX) after adjusting for age

Characteristic PEX No PEX P value Male Female P value

ECD (cells/mm2) 2315 (49.13) 2482 (20.36) 0.002 2392 (37.25) 2405 (37.01) 0.802CS CV (%) 33.9 (0.71) 34.4 (0.29) 0.499 34.3 (0.54) 34.0 (0.53) 0.676HC (%) 56.1 (1.06) 57.9 (0.44) 0.123 56.6 (0.81) 57.4 (0.80) 0.458

Figures represent mean (standard error of the mean). CS, cell size; CV, coefficient of variation; ECD, endothelial cell density; HC,hexagonal cells.

452 Quiroga et al.


mean endothelial cell density of our populationwas 2451 cells/mm2. However, because our studyincluded only patients with cataract, the resultscannot be extrapolated to the general population.

There are reports in the literature stating thatendothelial cell densities are lower in patients withcataract than in those without cataract.7 Inoue et al.reported an mean endothelial cell density in Japan of2543 cells/mm2 in patients aged 50–89 years.8 In con-trast, Praveen et al. reported an mean endothelial celldensity of 2242 cells/mm2 among a similar popula-tion in India.7 Snelligen et al. however, studied 1235eyes with cataract, and described an mean endothe-lial cell density of 2720 cells/mm2 in cataract patientsaged 40–75 years from three different ethnic groupsin Asia (southern India, Bangladesh and Nepal).9

The greater difference between the results of thisstudy and our own may be partly due to the higherage range (50–94 years) of patients in our study.

The overall frequency of PEX in the populationwith cataract in our study was 14.6%. Because ourresearch dealt only with patients from a single insti-tution and included only patients age 50 or olderwho were indicated for cataract surgery, we cannotaddress the question of the prevalence of PEX inParaguay. Nonetheless, the data are useful in thatthey document how frequently one surgical facilityencounters PEX and the implied risk for the country.

In 2007, a retrospective study was carried out atthe Hospital de Clinicas (university-teaching hospi-tal) in Asuncion that included 872 eyes of patientsoperated on for cataract and described a frequencyof PEX of 18.8%,10 slightly higher than that foundin our own study. Similar studies performed inother countries report widely varied rates of preva-lence. For example, in Poland, a European country,a prevalence of 8.2%11 was reported, whereas inEthiopia, in Africa, the figure found was 39.3%.12

The differences may be explained not only by ethnicfactors, but also by methodological differences, suchas different age ranges, inclusion of patients withglaucoma, type of facility, the population studied orother factors. In our study, the prevalence of PEXincreased with age, reaching a plateau in the 70- to79-year age group.

Various articles exist describing changes in histo-pathological endothelial characteristics,6,13 but fewuse specular microscopy to analyse the possibledifferences between patients with or without PEX.

We found that the mean age-adjusted endothelialcell density was lower in patients with PEX com-pared with those without PEX (2315 and 2482 cells/mm2, respectively), a difference that was statisticallysignificant. Knorr et al. studying 4432 eyes withoutprior intraocular surgery or ocular hypertension,reported an mean endothelial cell density of2302 cells/mm2, whereas 182 eyes with PEX had avalue of 1812 cells/mm2, and 59 eyes with PEX andglaucoma presented with 1482 cells/mm2.14 Inoue, ina similar study, reported finding a mean endothelialcell density of 2632 cells/mm2 in 30 eyes withoutPEX, and 2336 cells/mm2 in 21 eyes with PEX.15 Inour study, the differences in endothelial cell densitybetween eyes with and without PEX was 167 cells/mm2, which is less that the 300–500 cells/mm2

reported in the literature.1,12,15,16 This difference maybe due to the cited studies including patients withglaucoma in the group with PEX, as high intraocularpressure is an added risk factor for loss of endothelialcells.

At-risk endothelial cell densities (less than2000 cells/mm2) were found in 11% (n = 46) of theeyes in our population, similar to the 12% reportedin Japan by Ishikawa.17 Of these eyes, 89% belongedto patients over age 70, and 29% presentedpseudoexfoliation. The literature documents theeffect on the corneal endothelium of aging, whichcreates a reduction in the endothelial cell density andpercentage of hexagonal cells, as well as an increasein variability of cell size.4,8,18,19 In our study, althoughANOVA results demonstrated that endothelial celldensities significantly diminished with greater age,no significant association was found between ageand variability of cell size or percentage of hexagonalcells. The intratest variability of the equipment used(which was greater for these parameters than for theendothelial cell density), and the smaller size ofthe sample may explain the difference. After adjust-ment for age, the effect of PEX after intraocularsurgery does increase the risk of corneal decompen-sation (OR 1.90), but due to the small sample sizethis result was not significant.

Normally, the cost of including specular micro-scopy as part of a routine preoperative examinationcannot justified because most PEX cases can bedetected in the slit-lamp examination. However,because of the high prevalence of PEX in Paraguay(we estimate that 25% of patients scheduled forsurgery have it; and Strohl et al.20 found it was a causeof secondary glaucoma in 27% of Paraguayanpatients with glaucoma), we suggest that it might beincluded for every patient with signs of PEX in this

Table 5. Distribution of eyes with and without PEX with at-riskcorneas for decompensation following intraocular surgery basedon endothelial cell density

Distribution Endothelial cell density(cells/mm2) (%)

<2000 >2000

PEX, n (%) 13 (21.3) 48 (78.6)No PEX, n (%) 33 (9.2) 323 (90.7)

PEX, pseudoexfoliation syndrome.



population, and perhaps others with a high preva-lence of PEX. In Paraguay, its cost as part of preop-erative care is about US$5.

It should be emphasized that, as was done in ourresearch, all the studies that have been cited alsoused specular microscopy with semiautomatedanalysis, a method that requires marking of cells tobe done by an operator. Many of them also used, aswe did, a non-contact specular microscope.7,10,14,18

Intratest variability was checked for the threeparameters used in our study; and it was foundthat the endothelial cell density suffered the leastvariation, for which reason it seems for practicalpurposes to be a useful measurement for decisionmaking. Following the example of Snelligen et al.,9

we used a minimum of 60 central cells because thephotographed area was quite small and the periph-ery not always sharp. This is a limitation in termsof precision as many studies have used 100 cellsor more. However, by taking three pictures, webelieve that we can substantially reduce the possi-bility of errors. The CV and percentage of hexago-nal cells, in contrast, varied greatly, making themmuch less trustworthy. An attempt was made toreduce variations in measurements for the study byrepeating each measurement three times and usingonly the two measurements that were closest toeach other to find a mean. The same operator fol-lowed identical instructions for all measurements;and all eyes for which endothelial cell densityresults varied more than 10% were excluded fromthe study. Inspection of the photographs and visualanalysis of the shape and arrangement of the cells,more than the numerical values calculated, continueto be very important in determining pleomor-phism and polymegathism with this type of specu-lar microscope.

Based on this study, we suggest that multicentricstudies with larger sample populations be carriedout in order to obtain results that can be extrapolatedto the population of Paraguay. In addition, studiesof the corneal endothelium in at-risk populations,such as patients with ocular hypertension, glaucomaor uveitis would be of great help in guiding oursurgical practice towards excellence.

REFERENCES

1. Wirbelauer C, Anders N, Pham DT, Wollensak J.Corneal endothelial cell changes in pseudoexfoliationsyndrome after cataract surgery. Arch Ophthalmol 1998;116: 145–9.

2. Bourne RR, Minassian DC, Dart JK, Rosen P, KaushalS, Wingate N. Effect of cataract surgery on the cornealendothelium: modern phacoemulsification comparedwith extracapsular cataract surgery. Ophthalmology2004; 111: 679–85.

3. Cheng H, Jacobs PM, McPherson K, Noble MJ.Precision of cell density estimates endothelial cell losswith age. Arch Ophthalmol 1985; 103: 1478–81.

4. Abib FC. Microscopia Especular de Córnea. Manual e Atlas.Rio de Janeiro: Revinter, 2000.

5. Schlötzer-Schrehardt U, Naumann GO. Ocular and sys-temic pseudoexfoliation syndrome. Am J Ophthalmol2006; 141: 921–37.

6. Ritch R, Schlötzer-Schrehardt U. Exfoliation syn-drome. Surv Ophthalmol 2001; 45: 265–315.

7. Praveen MR, Vasavada AR, Ghodadra B, EdelhauserHF. Corneal endothelial morphologic features incataract and clear lens an Indian population. Am JOphthalmol 2007; 144: 914–20.

8. Inoue K, Tokuda Y, Inoue Y, Amano S, Oshika T,Inoue J. Corneal endothelial cell morphology inpatients undergoing cataract surgery. Cornea 2002; 21:360–3.

9. Snellingen T, Rao GN, Shrestha JK, Huq F, Cheng H.Quantitative and morphological characteristics of thehuman corneal endothelium in relation to age, genderand ethnicity in cataract populations of South Asia.Cornea 2001; 20: 55–8.

10. Bogado M y col. Prevalencia de pseudo exfoliación, diámetropupilar y complicaciones en cirugía de cataratas. Asunción,Paraguay: VI Congreso Paraguayo de Oftalmología,2006.

11. Obuchowska I, Bryl-Przybylska A, Mariak Z.[Pseudoexfoliation syndrome in patients scheduledfor cataract surgery]. Klin Oczna 2006; 108: 397–400.

12. Teshome T, Regassa K. Prevalence of pseudoex-foliation in Ethiopian patients scheduled for cataractsurgery. Acta Ophthalmol Scand 2004; 82: 254–8.

13. Naumann GO, Schlötzer-Schrehardt U. Keratopathyin pseudoexfoliation syndrome as a cause of cornealendothelial decompensation: a clinicopathologicstudy. Ophthalmology 2000; 107: 1111–24.

14. Knorr HL, Junemann A, Handel A, Naumann GO.[Morphometric and qualitative changes in cornealendothelium in pseudoexfoliation syndrome]. FortschrOphthalmol 1991; 88: 786–9.

15. Inoue K, Okugawa K, Oshika T, Amano S. Mor-phological study of corneal endothelium and cornealthickness in pseudoexfoliation syndrome. Jpn JOphthalmol 2003; 47: 235–9.

16. Seitz B, Müller EE, Langenbucher A, Kus MM,Naumann GO. [Endothelial keratopathy in pseu-doexfoliation syndrome: quantitative and qualitativemorphometry using automated video image analysis].Klin Monatsbl Augenheilkd 1995; 207: 167–75.

17. Ishikawa A. Risk factors for reduced corneal endo-thelial cell density before cataract surgery. J CataractRefract Surg 2002; 28: 1982–92.

18. Rao SK, Ranjan Sen P, Fogla R, Gangadharan S,Padmanabhan P, Badrinath SS. Corneal endothelialcell density and morphology in normal Indian eyes.Cornea 2000; 19: 820–3.

19. Yunliang S, Yuqiang H, Ying-Peng L, Ming-Zhi Z,Lam DS, Rao SK. Corneal endothelial cell density and

454 Quiroga et al.


morphology in healthy Chinese eyes. Cornea 2007; 26:130–2.

20. Strohl A, Pozzi S, Wattiez R, Roesen B, Miño de KasparH, Klauss V. [Secondary glaucoma in Paraguay. Etiol-ogy and incidence]. Ophthalmologe 1999; 96: 359–63.

APPENDIX I

Details of the multivariate analysis

A full-factorial multivariate GLM showed that thethree endothelial dependent variables were signifi-cantly correlated (Bartlett’s test of sphericity, P <0.0005). Models in which interactions with age wereadded did not significantly change the R2 value; andthe full-factorial model (R2 = 0.082, adjusted R2 =0.073) was used as the base model. Box’s test of

equality of the covariance matrices was non-significant (P = 0.677) as was Levine’s test forequality of error variances for all three dependentvariables (P = 0.291, 0.372 and 0.430 for endothelialcell density, mean CV of cell size and mean percent-age of hexagonal cells, respectively).

Pillai’s trace was significant for age (P < 0.0005)and PEX (P = 0.003), but was not significant forsex (P = 0.888) or sex * PEX (P = 0.322). Tests ofbetween-subject effects showed that in regard tomain effects, only age and PEX were significant forendothelial cell density (P < 0.001 and P = 0.002,respectively); all other main effects in regard tothe dependent variables were non-significant. Inaddition, the interaction term sex * PEX was non-significant for all three dependent variables.



characteristics of the corneal endothelium and pseudoexfoliation syndrome in patients with senile...

Documents