Corneal endothelial morphology and centralthickness in patients with type II diabetes mellitus

Allan Storr-Paulsen,1 Amardeep Singh,1 Helene Jeppesen,1 Jens C. Norregaard1 and Jesper Thulesen2

1Department of Ophthalmology, Frederiksberg University Hospital, Frederiksberg, Denmark2Roskilde Eye Clinic, Roskilde, Denmark

ABSTRACT.

Purpose: To investigate corneal endothelial cell density and morphology in type II

diabetic and non-diabetic patients and to relate potential differences to the glycae-

mic status.

Methods: A prospective clinical study including 107 patients with type II diabetes

and 128 non-diabetic patients. Sample size was based on a power calculation

(power = 0.90; p = 0.05). The diabetic patients had on average more than four

HbA1c tests performed (mean 4.1; range 2–14) with intervals of at least 3 months

as a reflection of the long-term glycaemic status. The controls had no diabetes con-

firmed by two causal blood tests. The endothelial cell density, the variation in

endothelial cell size (CV), the percentage of hexagonal cells, and the central cor-

neal thickness (CCT) were recorded.

Results: Type II diabetic subjects did not differ from the non-diabetic control sub-

jects with regards to endothelial cell density, hexagonality or variation in CV, but

showed a significant increase in CCT (538 versus 546 lm, p < 0.05). In the dia-

betic group, lower cell counts were associated with higher HbA1c values

(p < 0.05). The HbA1c did not, however, have any impact on the CCT.

Conclusion: Type II diabetes has no impact on corneal cell density or morphology

in subjects with good glycaemic status. However, higher HbA1c was associated

with lower endothelial cell density. CCT was significantly increased in the diabetic

group.

Key words: central corneal thickness – corneal endothelial cells – diabetes mellitus type II –

glycaemic status – morphology

Acta Ophthalmol. 2014: 92: 158–160ª 2013 The Authors

Acta Ophthalmologica ª 2013 Acta Ophthalmologica Scandinavica Foundation. Published by Blackwell Publishing

Ltd.

doi: 10.1111/aos.12064

Introduction

Worldwide, the incidence of type IIdiabetes mellitus is increasing, reach-ing epidemic proportions in develop-ing countries. The disease entity ischaracterized by hyperglycaemia andthe development of micro- and macro-vascular disorders, leading to func-tional and morphological disorders inseveral organs. Ocular manifestations

include anterior ischaemic neuropathy,glaucoma, cataract, retinal vein andarterial occlusions and retinopa-thy ⁄maculopathy (Stanga et al. 1999;Jeganathan et al. 2008). The develop-ment of many of the diabetic compli-cations is related to the duration ofthe disease and the degree of meta-bolic dysregulation.

Several studies have indicatedchanges in human corneal endothelial

cell morphology in patients with typeII diabetes (Itoi et al. 1989; Ros-zkowska et al. 1999; Inoue et al.2002a). Hypothetically, these phenom-ena could be caused by chronicmetabolic changes at the cellularlevel that primarily affect the singlelayer of the coherent endothelialcells (Morikubo et al. 2004). These lar-gely hexagonal cells have practicallyno proliferative activity. They areresponsible for maintaining the hydra-tion of the stroma by actively remov-ing water, thus playing a pivotal rolein maintaining the transparency of thecornea.

The present study was undertakento investigate potential differences incorneal endothelial structure and cen-tral corneal thickness (CCT) betweena cohort of well-regulated type II dia-betic subjects and non-diabetic sub-jects.

Materials and Methods

Subjects were recruited amongpatients attending our outpatientclinic. The diagnosis of type II diabe-tes was based on the medical history,and all subjects were on oral or paren-teral antidiabetic medication. Allpatients were Caucasians. Serum gly-cosylated haemoglobin HbA1c wasobtained in all diabetic patients toevaluate their glycaemic status. Atleast two samples of HbA1c weretaken with a 3-month interval toextrapolate the mean glycaemic con-trol for a period of at least half ayear. The diabetic patients had no ormild diabetic retinopathy. None had

Acta Ophthalmologica 2014

158

proliferative retinopathy or a historyof previous photocoagulation. Non-diabetic patients from the same timeperiod were used as controls, and twocasual blood glucose tests were takenin accordance with the recommenda-tions of the American Diabetes Asso-ciation (ADA) to disclose undetecteddiabetes (American Diabetes Associa-tion 2007). Criteria of exclusion wereany corneal pathology, pseudoexfolia-tion and glaucoma, history of oculartrauma or intraocular surgery, intra-ocular inflammation, contact lenswear, age younger than 40 years, andendothelial cell count <1500 cell ⁄mm2.

All patients had a slit lamp and aretinal evaluation performed, andbest-corrected visual acuity and intra-ocular pressure (IOP) were measured.Central endothelial cell density(cells ⁄mm2), variation in size of endo-thelial cells (CV), the percentage ofhexagonal cells, and CCT were analy-sed using a non-contact specularmicroscope (SP 2000P; Topcon,Tokyo, Japan) with the Image-Net�imaging system (version 2.1; Topcon).At each visit, three photographs ofeach cornea were taken and analysedindependently by three ophthalmolo-gists unaware of the diabetic status. Asemi-automated technique was used inwhich the automatic cell outlines ofeach photograph were reviewed andthen corrected manually (Cheung &Cho 2000). All three ophthalmologistswere skilled in reading endothelial cellphotographs, and preliminary testswere performed to test for interob-server variation in evaluation of theendothelial cell parameters. An inte-robserver variation below 5% ofrepeated evaluations of the same pho-tograph demonstrated the precision ofour technique. The mean of nine cor-rected readings was calculated andused to determine the mean value foreach visit.

The study protocol was approvedby the local ethical committee andreported to the Danish Data Protec-tion Agency and to the ClinicalTri-als.gov Protocol Registration Systemin USA (Identifier: NCT01084850).Informed consent was obtained fromparticipating patients.

Statistical analysis

Tests for differences among groups fordemographic and clinical characteris-

tics were performed using the chi-square test for categorical variablesand t-test for continuous variables.Mean visual acuity was calculatedbased on LogMAR transformation.Visual acuity is presented as Snellendecimal fractions. Multivariate regres-sion analyses were performed toidentify variables associated withendothelial cell density and CCT. Age,gender, IOP, casual blood glucose andHbA1c were included as independentvariables. Ninety-five percent confi-dence intervals (CI) were calculated.Sample size was based on a power cal-culation (power 0.90; p = 0.05). Prob-ability values <5% were consideredstatistically significant.

Results

Two hundred and thirty-five Cauca-sian patients were enrolled in thestudy. One hundred and sevenpatients had type II diabetes mellitus,and 128 were non-diabetic subjects.Demographic and clinical status isgiven in Table 1. No statistically sig-nificant differences in gender, IOP,and visual acuity were observedbetween the diabetic and the non-dia-betic subjects. A small but statisticallysignificant difference in mean age wasobserved between the two groups(72.1 versus 75.6 years, p < 0.05).Mean HbA1c level was 7.3% (95%CI ±0.2%; range 5.5–10.7) in thediabetic group, indicating a goodcompliance with the antidiabetic treat-

ment. HbA1c was in average mea-sured four times (mean 4.1; range 2–14) during the study, which reflectsthe glucose level for a period of about1 year. In the non-diabetic subjects,the mean casual blood glucose levelwas 5.7 mm (95% CI ±0.1; range4.2–7.8) compared to 9.0 mm (95% CI±0.7; range 2.9–23.1) in the diabeticgroup. In the non-diabetic group, glu-cose tolerance tests were performedaccording to the recommendations ofADA. Four subjects with non-recog-nized type II diabetes were excludedfrom the control group and from thestudy after positive glucose tolerancetests.

The endothelial cell density was 2578(95% CI±77) cells ⁄mm2 in the diabeticpatients and 2605 cells ⁄mm2 (95% CI±66) in normal subjects (p = 0.60)(Table 2). Hexagonality and CV werealso similar in the two groups (Table 2).The CCT was 546 lm (95% CI ±7) inthe diabetic group compared to 538 lm(95% CI ±5) in the normal subjects.The difference in CCT was statisticallysignificant (p < 0.05).

Multivariate analysis revealed a sig-nificant association between thinnerCCT and higher age (p < 0.05). In thecontrol group, cell counts were notassociated with any of the independentvariables. In the diabetic group, lowercell counts were associated with higherHbA1c values (p < 0.05) after control-ling for age. CV and the percentage ofhexagonal cells were not associatedwith any of the examined variables.

Table 1. Demographics and clinical status.

Diabetics,

n = 107

Controls,

n = 128 p-values

Age in years; mean (SD) 72.1 (11.0) 75.6 (8.9) p < 0.05

Gender (male ⁄ female) 51 ⁄ 56 49 ⁄ 79 p = 0.18

Casual blood glucose (mm); mean (95% CI) 9.0 (±0.2) 5.7 (±0.1) p < 0.05

HbA1c in %; mean (95% CI) 7.3 (±0.2) –

Number of HbA1c samples taken

with 3-month interval; mean (range)

4.1 (2–14) –

SD, standard deviation; 95%CI, 95% confidence intervals.

Table 2. Corneal endothelial cell count, cell size variation coefficient, percentage of hexagonal

cells, and central corneal thickness (CCT) in diabetic and non-diabetic subjects (mean values).

Diabetics Controls p-values

Cells ⁄mm2 (95% CI) 2578 (±77) 2605 (±66) p = 0.60

Hexagonal cells, % (95% CI) 58.0 (±1.4) 58.5 (±1.1) p = 0.67

Cell size variation coefficient (95% CI) 31.9 (±1.0) 31.9 (±0.9) p = 0.91

CCT, lm (95% CI) 546 (±7) 538 (±5) p < 0.05

95% CI, 95% confidence intervals.

Acta Ophthalmologica 2014

159

Discussion

Central corneal endothelial morphology

Evaluation of the endothelial cell mor-phology not only reflects the quantita-tive corneal properties, but alsoprovides a qualitative description of thefunctional status in terms of variation incell area and cell shape. A reduction ofcells that is not detectable by cell densitymeasurements alone may be detected byquantization of the CV, the percentageof hexagonal cells, and the CCT.

In the present study, similar celldensity, hexagonality, and CV wereobserved in the two groups; but meanCCT was higher in the diabetic group.Although statically significant, thisdifference in CCT might be of no clin-ical relevance as no corneal oedemawas observed in any of the patients.In the diabetes group, HbA1c and celldensity were inversely correlated.

Several articles have paid attentionto the corneal morphology in type IIdiabetics and normal subjects, but theresults are inconsistent. Thus, themorphology of the corneal endothe-lium in conditions with diabetes mell-itus remains to be fully elucidated.

Studies showing differences inmorphology between diabetics and normalsubjects

Su and co-workers examined the rela-tionship of diabetes and hyperglyca-emia with CCT in 3239 eyes,including 748 diabetics, and demon-strated mean CCT 6.5 lm thicker indiabetics than in persons without dia-betes (p < 0.001) (Su et al. 2008).Moreover, they found a correlationbetween corneal thickness and thehigher HbA1c. These results are inaccordance with the present studyshowing that poor diabetic compliancewas associated with lower cell counts.

Roszkowska et al. reported a signifi-cant decrease in cell density in type IIdiabetic patients compared to normalsubjects. Unfortunately, the number oftype II diabetic patients is not speci-fied, the metabolic control was lessstrict (HbA1c < 9.5%), and the agerange of the diabetics was restricted to40–60 years (Roszkowska et al. 1999).Inoue et al. investigated the corneal

morphology of 99 type II diabeticsand 97 non-diabetics and found a sig-nificant decrease in cell density and anincrease in CV among the diabeticpatients (Inoue et al. 2002a). However,it is not mentioned in the articlewhether the endothelial photographswere evaluated manually or automati-cally. This is important, because auto-mated cell analysis significantlyunderestimates average cell size andhexagonality and overestimates endo-thelial cell density and CV (Cheung &Cho 2000). Lee et al. evaluated the dif-ferences in corneal morphology in 200patients with insulin-dependent diabe-tes compared to 100 normal controlsubjects. The study showed that thediabetics had thicker corneas, less celldensity and hexagonality, and moreirregular cell size (higher CV) (Leeet al. 2006). However, the study doesnot differentiate between patients withtype I and II diabetes, stating only thephrase ‘insulin-dependent diabetics’.

Studies showing no differences in mor-phology between diabetics and normal

subjects

Several studies found no differences incorneal cell density between diabeticsand normal subjects (Larsson et al.1996; Inoue et al. 2002b; Hugod et al.2011). Most of these studies sufferfrom the inability to describe a well-defined diabetic group with goodlong-term glycaemic status and awell-defined control group withoutdiabetes confirmed by casual glucoseblood tests, according to the guide-lines of ADA (American DiabetesAssociation 2007).

The current study describes a well-defined diabetic group with good long-term glycaemic status and a well-defined control group without diabetesconfirmed by casual glucose blood tests.

Conclusion

The present study shows no differ-ences in endothelial cell density, hex-agonality or CV between a cohort ofwell-regulated diabetic subjects and acohort of (true) non-diabetic subjects.However, we found a statisticallysignificant increase in CCT in the dia-betic cohort compared to non-diabet-

ics. Multivariate analysis revealed thatpoor glycaemic control in diabeticswas associated with a decreased celldensity. Our data suggests that glycae-mic status (HbA1c) should be includedin future studies examining corneamorphology in diabetic patients.

AcknowledgementSupported by an unrestricted grantawarded by the Sofus and Olga FriisFoundation, Copenhagen, Denmark.

ReferencesAmerican Diabetes Association (2007): Diagnosis and classi-

fication of diabetes mellitus. Diabetes Care 30: S42–S47.

Cheung SW & Cho P (2000): Endothelial cells analysis

with the TOPCON specular microscope SP-2000P and

IMAGEnet system. Curr Eye Res 21: 788–798.

Hugod M, Storr-Paulsen A, Norregaard JC, Nicolini J,

Larsen AB & Thulesen J (2011): Corneal endothelial

cell changes associated with cataract surgery in patients

with type II diabetes mellitus. Cornea 30: 749–753.

Inoue K, Kato S, Inoue Y, Amano S & Oshika T (2002a):

The corneal endothelium and thickness in type II diabe-

tes mellitus. Jpn J Ophthalmol 46: 65–69.

Inoue K, Tokuda Y, Inoue Y, Amano S, Oshika T & Inoue J

(2002b): Corneal endothelial cell morphology in patients

undergoing cataract surgery. Cornea 21: 360–363.

Itoi M, Nakamura T, Mizobe K, Kodama Y, Nakagawa

N & Itoi M (1989): Specular microscopic studies of the

corneal endothelia of Japanese diabetics. Cornea 8: 2–6.

Jeganathan VS, Wang JJ & Wong TY (2008): Ocular asso-

ciations of diabetes other than diabetic retinopathy.

Diabetes Care 31: 1905–1912.

Larsson LI, Bourne WM, Pach JM & Brubaker RF

(1996): Structure and function of the corneal endothe-

lium in diabetes mellitus type I and type II. Arch Oph-

thalmol 114: 9–14.

Lee JS, Oum BS, Choi HY, Lee JE & Cho BM (2006):

Differences in corneal thickness and corneal endothe-

lium related to duration in diabetes. Eye 20: 315–318.

Morikubo S, Takamura Y, Kubo E, Tsuzuki S & Akagi Y

(2004): Corneal changes after small-incision cataract

surgery in patients with diabetes mellitus. Arch Oph-

thalmol 122: 966–969.

Roszkowska AM, Tringali CG, Colosi P, Squeri CA & Ferreri

G (1999): Corneal endothelium evaluation in type I and

type II diabetes mellitus. Ophthalmologica 213: 258–261.

Stanga PE, Boyd SR & Hamilton AM (1999): Ocular man-

ifestations of diabetes mellitus. Curr Opin Ophthalmol

10: 483–489.

Su DH, Wong TY, Wong WL et al. (2008): Diabetes,

hyperglycemia, and central corneal thickness: the

Singapore Malay Eye Study. Ophthalmology 115:

964–968.

Received on April 29th, 2012.

Accepted on November 11th, 2012.

Correspondence:

Allan Storr-Paulsen, MD

Department of Ophthalmology

Frederiksberg University Hospital

DK-2000 Frederiksberg

Denmark

Tel: +45 38 16 36 37

Fax: +45 38 16 36 39

Email: allanstorr@dadlnet.dk

Acta Ophthalmologica 2014

160