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

RESEARCH

Corneal endothelial changes in type 2 diabetes mellitus relative

to diabetic retinopathy

Clin Exp Optom 2019 DOI:10.1111/cxo.12971

Irfan Durukan MD
Ophthalmology Department, Era Goz Hospital, Ankara,
Turkey
E-mail: dr.irfandurukan@gmail.com
Submitted: 16 September 2018
Revised: 23 April 2019
Accepted for publication: 16 August 2019
Background: To evaluate the morphological features of corneal endothelial cells and their
relationship with the stage of retinopathy in diabetes mellitus (DM).
Methods: Patients with type 2 DM and age-matched controls were included in this prospec-
tive, cross-sectional study. Patients with diabetic retinopathy (DR) were divided into no-DR,
non-proliferative DR and proliferative DR based on the fundus findings. Endothelial mea-
surements were obtained using specular microscopy (Topcon SP-3000P, Japan). Endothelial
cell density, average cell area, co-efficient variation of cell area and percentage of hexagonal
cells were evaluated. Central corneal thickness (CCT) was measured using an ultrasound
pachymeter. Endothelial cell parameters and CCTs of DM and control groups were com-
pared and subgroup analysis of the patients with DM was performed.
Results: One hundred and twenty patients (mean age 59.5  8.1 years) were in the DM
group, 112 patients (mean age 57.3  7.2 years) were in the control group. Both the endo-
thelial cell density and the hexagonal cell rate were lower, while the CCT was higher in the
DM group. There were no significant differences between these two groups in terms of
average cell area and co-efficient variation of cell area. Significant differences were detected
between endothelial cell density values of subgroups, and endothelial cell density decreased
as the stage of the DR increased. The average cell area, co-efficient variation of cell area
and CCT measurements were similar across subgroups. Hexagonality values were signifi-
cantly different between subgroups, with the lowest ratio of hexagonal cells in the prolifera-
tive DR group.
Conclusion: Additional precautions should be taken to reduce the risk of endothelial
decompensation prior to intraocular surgery, especially in patients with proliferative DR.

Key words: cornea, diabetes, endothelium, retinopathy

Diabetes mellitus (DM) affects almost all
ocular structures of the eye. Diabetic reti-
nopathy (DR) is one of the leading causes of
blindness worldwide.1 Additionally, morpho-
logical changes such as a decrease in cor-
neal endothelial cell count, pleomorphism
and polymegathism are more frequent in
diabetic patients.2,3 The morphological and
functional integrity of the endothelial layer
is very important for the maintenance of
corneal transparency. Morphological prop-
erties of the endothelium can be assessed
with high reliability and accuracy by using a
specular microscope.4
Patients with DM tend to develop cataracts
more frequently and earlier than the non-
diabetic population, and the frequency of cat-
aracts increases with the duration of DM.5 At
the present time, one of the important cau-
ses of bullous keratopathy is endothelial
damage after cataract surgery.6 Therefore,
having information about the number and
morphology of endothelial cells before cata-
ract surgery is important in an effort to take
the necessary precautions to reduce the risk
of endothelial failure, particularly in patients
with DM.
Presence and stage of DR can be easily
detected by dilated fundus examination in all
eye clinics. Specular microscopy is required
for morphological assessment of corneal
endothelial cells. However, this device is
available at a limited number of clinics in
undeveloped and developing countries. The
aim of the present study was to determine
whether there is a relationship between the
retinopathy stage and the morphological fea-
tures of corneal endothelial cells in patients
with DM, and whether an assessment can be
made about the possibility of endothelial
failure after cataract surgery based on reti-
nopathy stage.
Methods
Participants
This prospective and cross-sectional study
was conducted in accordance with the Decla-
ration of Helsinki upon the approval of the
Ankara Numune Education and Research
Hospital Ethics Committee. All participants
were recruited and examined by the same
clinician (ID) in Era Eye Hospital. Written
informed consent was obtained from all par-
ticipants prior to enrolment. The patients
with type 2 DM (DM group) and their age-
matched controls were included and only the
right eyes of the subjects were analysed. The
presence of type 2 DM was confirmed by the

© 2019 Optometry Australia Clinical and Experimental Optometry 2019

1
Corneal morphometry in diabetes Durukan
endocrinology department. The presence
and stages of DR in patients with DM were
investigated using fundus photography, fun-
dus fluorescein angiography and/or optical
coherence tomography by the same clinician.
The Early Treatment of Diabetic Retinopathy
Study criteria were used to describe various
stages of DR.7 For detailed analysis of
patients with DR, patients with DM were fur-
ther divided into three groups as no-DR, non-
proliferative DR and proliferative DR based
on the fundus findings by the same clinician
(ID). Also, the duration of DM and the most
recent HbA1c values were recorded. All of the
control subjects were healthy without any
known systemic disease and had applied to
the ophthalmology clinic for a routine ocular
examination and/or presbyopic complaints.
The exclusion criteria of the study were: a
history of ocular surgery, retinal photocoagu-
lation, or trauma, glaucoma, uveitis, hyper-
opia or myopia more than 3.00 D, and
astigmatism more than 1.00 D. Corneal dis-
eases such as keratoconus, Fuchs endothelial
dystrophy, corneal opacities, dry eye, contact
lens wearers, patients with chronic topical
ophthalmic medications and other systemic
diseases except DM were also excluded.
Techniques
All subjects underwent a comprehensive
ophthalmic examination including best-
corrected visual acuity assessment using the
Snellen chart, intraocular pressure measure-
ments with a pneumotonometer, slitlamp
biomicroscopy, and fundus examination.
Corneal endothelial cell measurements were
performed by the same masked technician.
Figure 1. Specular microscopic findings of A: a patient with diabetes mellitus and B: a control subject are seen
Clinical and Experimental Optometry 2019
2
Three digital photographs from the central
cornea were obtained using the same non-
contact specular microscope (Topcon SP-
3000P; Topcon Corp., Tokyo, Japan) to evalu-
ate the corneal endothelium. Subjects were
asked to look at the central fixation target
and the auto-alignment function was used.
All corneal endothelial cells that were clearly
visible on the picture were manually marked.
At least 100 cells per measurement were
included in each analysis. The centre method
which is a common technique incorporated
into the specular microscope was used.
Endothelial cell density, average cell area, co-
efficient variation of cell area, and percentage
of hexagonal cells were calculated by the
software of the specular microscope
(Figure 1A–B). The co-efficient variation of cell
area in cell size (standard deviation divided
by the mean cell area) was used as an index
of the extent of variation in the cell area
(polymegethism), and the percentage of hex-
agonal cells in the analysed area was used as
an index of variation in cell shape (pleomor-
phism). After the completion of ophthalmic
examinations and specular measurements,
the endothelial morphological parameters of
DM and control groups were compared. Sub-
sequently, subgroup analysis of the patients
with DM was undertaken to assess the effect
of DR status on endothelial morphological
parameters.
Central corneal thickness (CCT) was mea-
sured using an ultrasound pachymeter
(US 4000; Nidek, Tokyo, Japan). During the
measurements, the subjects fixated on a
distant target and the pachymeter probe
was placed perpendicularly and centrally to
the cornea. Three consecutive measure-
ments were taken for each participant and
the average values were selected for data
analysis.
Statistical analysis
An a priori power analysis using the PASS
11 calculation (Power and Sample Size, ver-
sion 11; NCSS Statistical Software) showed
that approximately 84 patients with DM and
84 healthy controls should be enrolled to
reach a power of 80 per cent in the study.
The number of participants achievable dur-
ing the study period was 120 patients with
DM and 112 controls. Accordingly, the
power of the study was 89.5 per cent. After
that, a second power analysis was per-
formed for the patient subgroups with
various stages of DR and it was revealed
that at least 28 patients for each subgroup
should be enrolled to reach a power equal
to at least 80 per cent. Descriptive data
were presented as mean  standard devia-
tions, frequency distributions and percent-
ages. The chi-squared test was used to
analyse the categorical variables. Normal
distribution of the variables was checked
by visual (histogram and probability graphs)
and analytical methods (Kolmogorov–Smir-
nov/Shapiro–Wilk tests). Equality of vari-
ances was tested by the Levene test. An
independent sample t-test was used for nor-
mally distributed data, and the Mann–
Whitney U-test was used for non-normally
distributed data to compare the DM and
control groups. Additionally, one-way analy-
sis of variance, Welch analysis of variance,
and Kruskal–Wallis tests were used to
© 2019 Optometry Australia

DM group (n = 120)
Age (years) 59.5  8.1
Gender (male: female) 67:53
HbA1c (%) 9.2  1.5
IOP (mmHg) 17.1  1.6
ECD (cell/mm ) 2
2,295  311
AVG 411.6  52.1
CV 36.0  6.2
HEX (%) 49.1  10.2
CCT (μm) 544.2  38.0
AVG: average cell area, CCT: central corneal thickness, CV: co-efficient of variation of
cell area, ECD: endothelial cell density, HbA1c: glycosylated haemoglobin, HEX: per-
centage of hexagonal cells, IOP: intraocular pressure.
Bold values indicate p < 0.05.
† in the endothelial layer, making corneas of
Chi-squared test.
‡
Independent samples t-test.
Table 1. Demographic data, endothelial cell characteristics and central corneal thick-
ness measurements in the diabetes mellitus (DM) and control groups
determine if there were any significant dif-
ferences between the three subgroups of
DM. Post-hoc tests for pairwise comparisons
were also performed. A probability level of
p < 0.05 was considered statistically
significant.
Results
A total of 232 patients were included in the
study: 120 patients (67 male, 53 female,
mean age 59.5  8.1 years) were in the DM
group, 112 patients (60 male, 52 female,
mean age 57.3  7.2 years) were in the con-
trol group. Mean duration of the DM was
11.5  6.2 years, and the mean level of
HbA1c was 9.2  1.5 per cent in the DM
group. The demographic data, corneal endo-
thelial cell characteristics, and CCT values of
patients with DM and healthy subjects are
shown in Table 1. There were no differences
between the groups in terms of age and
gender (p > 0.05), but the average HbA1c
level was significantly higher in the DM
group (p < 0.001). Both the endothelial cell
density and the percentage of hexagonal
cells were significantly lower (p = 0.003 and
p = 0.04), while the CCT was significantly
higher in the DM group (p = 0.005). There
were no significant differences between
these two groups in terms of average cell
area and co-efficient variation of cell
area (p > 0.05).
The clinical data, endothelial cell charac-
teristics, and CCT measurements of the
© 2019 Optometry Australia
Control group (n = 112) p age was a significant factor (p = 0.003).
57.3  7.2 0.310‡
60:52 0.100†
5.0  0.3 < 0.001‡
17.4  2.0 0.501‡
2,501  302 0.003‡
408.2  48.8 0.603‡
34.9  4.0 0.344‡
55.2  10.0 0.040‡
521.5  31.5 0.005‡
patients with different stages of DM are
shown in Table 2. Of the total 120 patients
with DM, 40 patients did not have any sign
of DR (no-DR), 51 patients had non-
proliferative DR and the remaining 29 had
proliferative DR. While there was no age dif-
ference between no-DR and non-
proliferative DR groups, the mean age of
proliferative DR group was significantly
higher compared to the other two groups.
There were significant differences among
the three groups in terms of the average
HbA1c level and the duration of DM
(p < 0.05). In the proliferative DR group,
both the average HbA1c level and the dura-
tion of DM were the highest. In the no-DR
group, both the average HbA1c level and the
duration of DM were the lowest. A signifi-
cant difference was detected between endo-
thelial cell density values of subgroups
(p = 0.02), and endothelial cell density
decreased with increasing stage of the
DR. The average cell area, co-efficient varia-
tion of cell area and C values were similar
across subgroups (p > 0.05). Hexagonality
values were significantly different between
subgroups (p = 0.04), with the lowest ratio
of hexagonal cells in proliferative DR group.
The factors, including age, gender, dura-
tion of DM and HbA1c values, which can
affect the corneal endothelial cell character-
istics were evaluated by univariate and mul-
tivariate analyses. On univariate analysis,
significant factors affecting endothelial cell
characteristics were age (p = 0.001) and
duration of DM (p = 0.02); however, multiple
Corneal morphometry in diabetes Durukan
linear regression analysis revealed that only
Table 3 showed the endothelial cell charac-
teristics and CCT measurements of the
patients with different stages of DM.
Discussion
Diabetes mellitus affects all structural layers
of the cornea. Diabetic keratopathy findings
such as recurrent epithelial erosion, delayed
wound healing, and neurotrophic ulcers are
seen in a significant proportion of patients.
Structural and functional changes also occur
diabetic patients more vulnerable to intraoc-
ular surgery. Researchers have tried to
explain the effects of DM on endothelial cells
with many pathophysiological mechanisms
such as osmotic damage due to excessive
sorbitol accumulation, oxidative damage due
to glycation end products accumulation, and
damage to the F-actin fibrils, which are
important in maintenance of the hexagonal
shape. The increase in corneal thickness may
be due to osmotic swelling in the stroma or
a decrease in endothelial function.8
In the present study, it was detected that
the number of endothelial cells and the
ratio of hexagonal cells (206 cells/mm2 and
6.1 per cent, respectively) were lower and
the cornea was thicker (23 μm) in the DM
group compared to the age-matched con-
trol group. When studies evaluating the
effects of DM on the corneal endothelium
are examined, there were significant differ-
ences between the results. These differ-
ences may be caused by differences in
ethnicity, types of DM, duration of DM,
glycaemic status and HbA1c levels between
study groups. In the study of Modis et al.,9
there was significant endothelial cell loss in
type 1 DM compared to the control group,
but there was no significant difference
between type 2 DM and control groups. In
another study, significant endothelial cell
loss was detected in both types of DM, but
cell loss was found to be higher in type
1 DM than in type 2 DM.10 In the study by
Storr-Paulsen et al.,11 endothelial cell den-
sity and morphology were not affected in
patients with type 2 DM with good
glycaemic control, whereas endothelial cell
density was found to be low in patients
with high HbA1c. Although glycaemic con-
trol was good, the corneas of diabetic
patients were found to be thicker than the
control group.
Clinical and Experimental Optometry 2019
3
Corneal morphometry in diabetes Durukan

No-DR (n = 40) Non-proliferative DR (n = 51) Proliferative DR (n = 29) p

Age (years) 57.3  7.8 57.6  7.1 63.0  6.0 0.020†, 0.661‡, 0.032§, 0.035¶
HbA1c (%) 7.1  1.1 8.1  1.8 9.0  1.9 0.011†, 0.010‡, 0.022§, 0.020¶
Duration of DM 6.5  3.2 9.7  5.0 12.2  6.3 < 0.001†, < 0.001‡, < 0.001§, < 0.001¶
ECD (cell/mm2) 2,862  271 2,683  245 2,390  253 0.020†, 0.028‡, 0.001§, 0.010¶
AVG 402.1  51.0 412.0  55.2 399.9  49.6 0.201†
CV 33.8  4.2 34.6  5.2 36.0  4.9 0.125†
HEX (%) 53.4  11.0 51.7  8.1 46.0  9.1 0.030†, 0.218‡, 0.011§, 0.023¶
CCT (μm) 532.2  31.6 530.7  36.4 522.2  36.0 0.243†
AVG: average cell area, CCT: central corneal thickness, CV: co-efficient of variation of cell area, DM: diabetes mellitus, DR: diabetic reti-
nopathy, ECD: endothelial cell density, HEX: percentage of hexagonal cells.
Bold values indicate p < 0.05.
†
Significance in analysis of variance (comparison among three groups).
‡
Significance between no-DR and non-proliferative DR groups (pairwise comparison).
§
Significance between no-DR and proliferative DR groups (pairwise comparison).
¶
Significance between non-proliferative DR and proliferative DR groups (pairwise comparison).

Table 2. Demographic, clinical, and endothelial cell characteristics and central corneal thickness measurements of the patients
with different stages of DM

In our study, while cell density was differ-
ent between the DM and control groups as
well as DM subgroups, cell area and co-
efficient of variation (polymegathism) were
similar. Similar findings were obtained from
the study by Islam et al.12 and Sudhir et al.2
However, as is known, the reduction in the
number of corneal endothelial cells is
restored by expansion of neighbouring cells.
The lack of results from our study to con-
firm this expectation may be related to the
small number of patients or less reliable
endothelial cell area measurement than
endothelial cell density measurement.13
In the studies by Lee et al.14 and Briggs
et al.,15 it was detected that changes in corneal
thicknesses and endothelial cells are associ-
ated with the duration of DM and patients
having DM for more than 10 years were found
to have more prominent changes. In contrast
to these studies, in another study, although a
decrease in endothelial density and a deterio-
ration in morphology were detected in
patients with type 2 DM, there were no corre-
lations between quantitative and structural
changes, and HbA1c levels and duration of
DM.16,17 However, it is beneficial to remem-
ber that there were significant differences
between diabetic patients and healthy sub-
jects in terms of corneal thickness, endothe-
lial cell count, and morphology in the
majority of studies, while there was no signif-
icant difference in some of the studies.2,11,17
Another finding of the present study was
that there are correlations between endo-
thelial changes and DR, and endothelial
changes become more prominent as the
stage of DR increases. In the literature, the
results of the studies on this subject vary. In
the study by El-Agamy et al.,17 patients were
grouped as no-DR, non-proliferative DR and
proliferative DR, but there were no

No-DR (n = 40) Non-proliferative DR (n = 51) Proliferative DR (n = 29) p

ECD (cell/ mm2) 2,862  271 2,683  245 2,488  250 0.025†, 0.028‡, 0.001§, 0.013¶
AVG 402.1  51.0 412.0  55.2 400.8  48.8 0.245†
CV 33.8  4.2 34.6  5.2 35.8  4.8 0.133†
HEX (%) 53.4  11.0 51.7  8.1 47.2  9.1 0.032†, 0.218‡, 0.015§, 0.024¶
CCT (μm) 532.2  31.6 530.7  36.4 523.1  35.8 0.258†
AVG: average cell area, CCT: central corneal thickness, CV: co-efficient of variation of cell area, DR: diabetic retinopathy, ECD: endothe-
lial cell density, HEX: percentage of hexagonal cells.
Bold values indicate p < 0.05.
†
Significance in analysis of variance (comparison among three groups).
‡
Significance between no-DR and non-proliferative DR groups (pairwise comparison).
§
Significance between no-DR and proliferative DR groups (pairwise comparison).
¶
Significance between non-proliferative DR and proliferative DR groups (pairwise comparison).

Table 3. Endothelial cell characteristics and central corneal thickness measurements of the patients with different stages of dia-
betes mellitus after adjusted for age

Clinical and Experimental Optometry 2019 © 2019 Optometry Australia

4

correlations between the status of DR, endo-
thelial parameters and CCT. Similar to the
present study, in the study by Shenoy
et al.,18 correlations were detected between
changes in endothelial parameters and the
status of DR, and these changes were more
prominent in patients with proliferative
DR. The correlation between DR and corneal
endothelial cell loss may be due to common
pathophysiological mechanisms such as
accumulation of advanced glycation end
products and increased oxidative stress.
One of the disadvantages of the present
study is that fasting blood glucose levels
were not measured in the control group.
However, the patients in the control group
may have DM even though they are not
aware of it. Also, the mean age was found
to be higher in the proliferative DR sub-
group compared to the other two sub-
groups. Multivariate analysis showed that
age had a significant effect on endothelial
cell parameters in this study. However, the
results obtained from the comparison of
age-adjusted endothelial cell parameters
were similar to the unadjusted comparison
results between DM subgroups. Nonexis-
tence of age correction in evaluation of the
endothelial parameters of this group is one
of the drawbacks of this study. The fact that
we did not consider factors that can influ-
ence endothelial parameters such as cor-
neal diameter and smoking may be counted
as another disadvantage of the present
study.
© 2019 Optometry Australia
In conclusion, in the present study, increase
in corneal thickness, reduction in endothelial
cell density, and distortion of morphology
were detected in patients with type
2 DM. Therefore, additional precautions
such as use of dispersive ophthalmic visco-
elastic device, low phaco parameters, and
the phaco chop technique should be con-
sidered to reduce the risk of endothelial
decompensation prior to intraocular sur-
gery. Consequently, we suggest that endo-
thelial evaluation should be a part of the
routine inspection prior to intraocular sur-
gery in diabetic patients. Another important
finding of this study was the correlations
between endothelial changes and the stage
of DR. If specular microscopy is not available,
the stage of DR may be used to evaluate the
risk of endothelial failure after intraocular
surgery.
REFERENCES
1. Vieira-Potter VJ, Karamichos D, Lee DJ. Ocular compli-
cations of diabetes and therapeutic approaches.
Biomed Res Int 2016; 2016: 1–14.
2. Sudhir RR, Raman R, Sharma T. Changes in the cor-
neal endothelial cell density and morphology in
patients with type 2 diabetes mellitus: a population-
based study, Sankara Nethralaya diabetic retinopathy
and molecular genetics study (SN-DREAMS, report 23).
Cornea 2012; 31: 1119–1122.
3. Choo M, Prakash K, Samsudin A et al. Corneal
changes in type II diabetes mellitus in Malaysia. Int J
Ophthalmol 2010; 3: 234–236.
4. Mc Carey BE, Edelhauser HF, Lynn MJ. Review of corneal
endothelial specular microscopy for FDA clinical trials of
refractive procedures, surgical devices, and new intraoc-
ular drugs and solutions. Cornea 2008; 27: 1–16.
5. Becker C, Schneider C, Aballéa S et al. Cataract in
patients with diabetes mellitus-incidence rates
Corneal morphometry in diabetes Durukan
in the UK and risk factors. Eye (Lond) 2018; 32:
1028–1035.
6. Veldman PB, Terry MA, Straiko MD. Evolving indica-
tions for Descemet’s stripping automated endothelial
keratoplasty. Curr Opin Ophthalmol 2014; 25:
306–311.
7. Early Treatment Diabetic Retinopathy Study Design
and Baseline Characteristics. ETDRS report number 7.
Ophthalmology 1991; 98: 741–756.
8. Lutty GA. Effects of diabetes on the eye. Invest
Ophthalmol Vis Sci 2013; 54: 81–87.
9. Modis L Jr, Szalai E, Kertesz K et al. Evaluation of the
corneal endothelium in patients with diabetes
mellitus type I and II. Histol Histopathol 2010; 25:
1531–1537.
10. Roszkowska AM, Tringali CG, Colosi P et al. Corneal
endothelium evaluation in type I and type II diabetes
mellitus. Ophthalmologica 1999; 213: 258–261.
11. Storr-Paulsen A, Singh A, Jeppesen H et al. Corneal
endothelial morphology and central thickness in
patients with type II diabetes mellitus. Acta
Ophthalmol 2014; 92: 158–160.
12. Islam QU, Mehboob MA, Amin ZA. Comprasion of cor-
neal morphological characteristics between diabetic
and non diabetic population. Pak J Med Sci 2017; 33:
1307–1311.
13. Doughty MJ, Müller A, Zaman ML. Assessment of the
reliability of human corneal endothelial cell-density
estimates using a noncontact specular microscope.
Cornea 2000; 19: 148–158.
14. Lee JS, Oum BS, Choi HY et al. Differences in corneal
thickness and corneal endothelium related to dura-
tion in diabetes. Eye (Lond) 2006; 20: 315–318.
15. Briggs S, Osuagwu UL, AlHarthi EM. Manifestations of
type 2 diabetes in corneal endothelial cell density, cor-
neal thickness and intraocular pressure. J Biomed Res
2016; 30: 46–51.
16. Galgauskas S, Laurinavičiu  tė G, Norvydaitė D et al.
Changes in choroidal thickness and corneal parame-
ters in diabetic eyes. Eur J Ophthalmol 2016; 26:
163–167.
17. El-Agamy A, Alsubaie S. Corneal endothelium and cen-
tral corneal thickness changes in type 2 diabetes
mellitus. Clin Ophthalmol 2017; 11: 481–486.
18. Shenoy R, Khandekar R, Bialasiewicz A et al. Corneal
endothelium in patients with diabetes mellitus: a his-
torical cohort study. Eur J Ophthalmol 2009; 19:
369–375.
Clinical and Experimental Optometry 2019