Increased Endothelial Cell Density in the
Paracentral and Peripheral Regions
of the Human Cornea
JOSEF AMANN, MD, MPH, GLENN P. HOLLEY, BS, SANG-BUMM LEE, MD, AND
HENRY F. EDELHAUSER, PHD
● PURPOSE: To systematically investigate the central,
paracentral, and peripheral endothelial cell density
(ECD) in normal human corneas.
● DESIGN: Observational case series and experimental
study.
● METHODS: Noncontact specular microscopy was un-
dertaken to determine the ECD of the central, paracen-
tral (2.7 ⴞ 0.2 mm from center) and peripheral (4.7 ⴞ
0.2 mm from center) regions of the cornea of 48 normal
eyes. The ECDs of central and peripheral regions were
also determined with contact specular microscopy in 21
normal eyes and a group of 30 Optisol-GS eye bank
corneas were evaluated with alizarin red stain. Histologic
ECD of 13 Optisol-GS stored corneas were also deter-
mined.
● RESULTS: Paracentral and peripheral ECD measured
with the noncontact specular microscope were 5.8%
(P < .01) and 9.6% (P < .001) increased compared with
central ECD. Superior peripheral ECD was increased
compared with the other three peripheral quadrants (P <
.05) and was 15.9% higher than central ECD. Contact
specular microscopy showed an increase of 8.9% in the
peripheral ECD from the center. Alizarin red stained
corneas confirmed the specular microscopy numbers with
a 9.2% increase in the paracentral region, and a 17.2%
increase in the peripheral region. Histological cross
sections of human corneas also showed a 22.9% increase
in peripheral ECD compared with the central region.
● CONCLUSIONS: The human cornea has an increased
ECD in the paracentral and peripheral regions of cornea
compared with the central region. The superior periph-
eral region of the corneal endothelium has the largest
Accepted for publication Dec 4, 2002.
Internet Advance publication at ajo.com March 20, 2003.
From the Emory University Eye Center, Atlanta, Georgia.
This study was supported in part by NEI Grants ROI-EY00933 and
P30-EY006360, and an unrestricted grant from Research to Prevent
Blindness, New York, New York.
Inquiries to Henry F. Edelhauser, Emory University Eye Center,
1365-B Clifton Rd NE, Atlanta, GA 30322; fax: (404) 778-4143; e-mail:
ophthfe@emory.edu
584 © 2003 BY ELSEVIER INC. ALL
increase in ECD. These data on normal endothelial cell
distribution in the human cornea are especially signifi-
cant as they relate to new surgical techniques and
endothelial wound repair. (Am J Ophthalmol 2003;
135:584 –590. © 2003 by Elsevier Inc. All rights
reserved.)
S
PECULAR BIOMICROSCOPIC REFLECTION OF THE COR-
neal endothelium was introduced by Vogt1 in 1920.
Since the 1970s, several contact and noncontact
clinical specular microscopes have enabled in vivo pho-
tography of the human corneal endothelium. Qualitative
and quantitative analyses of endothelial photographs have
provided information regarding changes in cell size (poly-
megethism), cell shape (pleomorphism), and cell density
with aging,2 contact lens wear,3–5 cataract surgery,6 corneal
preservation, and corneal transplantation.7 All of these
studies depend on the assumption that the endothelial
cells of the central cornea mirror, to some extent, the cell
populations of the cornea as a whole. However, a system-
atic study of the paracentral and peripheral endothelial cell
density (ECD) distribution in the normal human corneal
endothelium has not been undertaken. There have been a
number of reports in the literature providing ECD for
paracentral and peripheral regions of the cornea, and these
studies have shown mixed results. The reports using
contact specular microscopy8 –10 differ from the histological
studies that count cells or cell nuclei on fixed and unfixed
tissue preparations.11–15 Besides the use of various methods
to determine ECD, there is also a difference in the location
where the ECD was measured. From the published studies,
a definitive conclusion about cell densities in the different
regions of the normal human cornea has been difficult to
determine. However, Schimmelpfenning11 and Daus and
colleagues14,15 have reported a significant increase in the
peripheral ECD compared with the central ECD.
With recent developments in keratorefractive surgery,
such as the intrastromal ring, hyperopic photorefractive
keratectomy, and phakic intraocular lens implantation, it
has become important to understand the corneal endothe-
RIGHTS RESERVED. 0002-9394/03/$30.00
doi:10.1016/S0002-9394(02)02237-7
FIGURE 1. (a) Photograph of the cornea where noncontact Robo specular micrographs were taken; small white boxes show the
nine areas and the mean ⴞ SD of the paracentral and peripheral distances from center. (b) Photograph of cornea where contact
specular micrographs were taken; small white boxes show the central and peripheral regions. (c) Diagram of corneal flat mount;
small black boxes shows areas where photographs were taken and endothelial cell density analyzed. The distance from center is listed
below diagram. (d) Diagram of corneal cross section; small boxes show the central and peripheral areas where endothelial cell nuclei
were counted. Peripheral distance is listed below the Figure.

lial cell density in the paracentral and peripheral corneal
endothelial regions. As the clinical use of these new
surgical techniques is ongoing, the possibility of endothe-
lial cell damage during and after surgery is of concern. The
purpose of this study was to systematically investigate the
central, paracentral, and peripheral endothelial cell den-
sity in normal human corneas using a noncontact and a
contact specular microscope and to compare the results
with histological studies on human donor corneas.
METHODS
THE STUDY WAS PERFORMED AFTER INSTITUTIONAL RE-
view Board approval from the human investigation com-
mittee at Emory University School of Medicine. Informed
consent was obtained from all subjects participating in the
study. All images and photographs of the endothelial cells
were analyzed by a masked reader to prevent bias.
The study group consisted of 24 patients with no history
of eye disease or trauma. Of the 24 patients, 14 patients
were female and 10 patients were male. The mean ⫾ SD
age was 29.3 ⫾ 7.1 years with a range of 19 to 50 years.
Ten of the investigated patients had a history of contact
lens wear. Endothelial images were obtained by using the
Konan Noncon Robo SP-8000 noncontact specular micro-
scope (Konan, Fairlawn, New Jersey, USA). We recorded
one central, four paracentral, and four peripheral images of
the corneal endothelium in the four different quadrants
(Figure 1a). The ECD for the paracentral and peripheral
area of the corneas were calculated by taking the average
of the four ECDs illustrated in Figure 1a. The mean ⫾ SD

VOL. 135, NO. 5 INCREASED ENDOTHELIAL CELL DENSITY IN HUMAN CORNEA 585
distance of the four paracentral regions from the center was
2.7 ⫾ 0.2 mm and for the peripheral regions from the
center was 4.7 ⫾ 0.2 mm. The distance from center was
determined each time by printing out the image location
box and measuring how far the image box was from the
center using an internal calibration scale. Images of endo-
thelial cells were recorded by using the HAI ISR/CL 3.33
image management software (HAI Labs, Arlington, Mas-
sachusetts, USA). The cells were analyzed with the Konan
computer-assisted analysis program using the center (dot)
method. This method is semiautomatic and the operator
has to hand-digitize the center of each cell in a contiguous
group of cells (150 –200 cells). The computer algorithm
calculates the endothelial cell density, mean cell area,
coefficient of variation of cell area (polymegethism), and
percentage of hexagonal cells (pleomorphism). For the
central and paracentral images the patient has to look at a
fixation light in the specular microscope. To get peripheral
images, the patient is instructed to look well beyond the
fixation light at a prepositioned black circle fixed on the
eyepiece of the Robo SP 8000 at the opposite side of where
the endothelial specular photograph is taken.
We examined the endothelium by using the Keeler-
Konan SP-1 contact specular microscope. Topical anes-
thetic drops of 0.5% proparacaine were instilled in both
eyes. Specular photographs were taken of the central and
the peripheral corneal endothelium of 21 eyes from 11
subjects (mean age, 30.9 ⫾ 4.1 years, range, 21–50). For
the peripheral photographs, the patient was asked to look
at a fixation light temporal to the opposite side that the
specular photographs were taken. The contact objective
lens (after wetting with balanced salt solution [BSS]) was
placed on the temporal limbus of the examined eye and
then moved toward the center of the cornea until the first
endothelial cells were observed. At that point, a photo-
graph of the peripheral endothelium was taken. Based on
the position of the contact objective lens, we estimated
that the photographs were taken 4.2 to 5.0 mm from the
center of the cornea. The best negatives of the four
photographs were printed and enlarged to ⫻400 magnifi-
cation. The magnification of the photographs was cali-
brated using a micrometer scale. Individual cells on the
endothelial photographs were subjected to a computer-
assisted morphometry of the area and shape as described
previously.16 Contiguous endothelial cells (100 –150) were
traced and digitized by touching cell apices (corner
method) with a Hewlett Packard 9111A graphics tablet
pen (Hewlett-Packard, Brookfield, Wisconsin, USA). The
entered cell coordinates were analyzed by endothelial
analysis software as previously described.17
A total of 30 human corneas from donors ranging in age
from 14 to 72 years were obtained from the Georgia Eye
Bank, Atlanta. It was difficult to age match the corneas
obtained from the Eye Bank to the corneas in the clinical
specular microscopy study because typically the younger
corneas are used for keratoplasty and are not sent to
586 AMERICAN JOURNAL
research. Two groups of 15 corneas were investigated. In
group 1 (mean donor age, 40.6 years; range, 14 –72) a
corneal button excised with a 10 mm trephine was pre-
pared and stained with alizarin red. This enabled us to
evaluate the ECD in the paracentral region. In group 2
(mean donor age, 55.3 years; range, 17–71), excised
corneas with a scleral rim were stained with alizarin red,
and provided a corneal flat mount with complete periph-
eral endothelial cells. All specimens were flattened on a
slide by radial incisions after staining. The supravital dye of
alizarin red was dissolved in isotonic saline (0.2 g/100 ml)
and applied to the cornea with the scleral rim for 5 to 7
minutes using the method described by Taylor and Hunt.18
The corneas were then rinsed with phosphate-buffered
saline and photomicrographs were obtained through a
Zeiss light microscope at a magnification of ⫻100. In group
1, two central and four paracentral (2.4 to 3.3 mm from
center) photomicrographs were taken. In group 2, two
central and four peripheral (4.4 to 5.3 mm from the center)
photomicrographs were taken. The pictures were printed
to a final magnification of ⫻290. Contiguous endothelial
cells (100 –150) were traced and digitized by touching cell
apices (corner method) with a Hewlett Packard 9111A
graphics tablet pen (Hewlett-Packard, Brookfield, Wiscon-
sin, USA). The analysis of the data was performed with the
same software as for the contact specular microscopy
photographs.17
Optisol GS stored human corneas (n ⫽ 13; mean donor
age, 45.5 years; range, 17–72) from the Georgia Eye Bank
were fixed in a 4% buffered formaldehyde solution, rou-
tinely processed, embedded in paraffin, sectioned (7 ␮m),
and stained with hematoxylin-eosin for histologic evalua-
tion. The endothelial cell counts were determined by
counting the nuclei of central and peripheral cells across
the diameter of a ⫻40 objective field using a standard light
microscope. The mean of five different cross-sectional
endothelial cell counts in the center and in the periphery
were calculated. Endothelial cell density was determined
from a previously described nomogram,19 which correlates
the number of endothelial cells per high power field to cell
density. The histological ECDs were determined in a
masked fashion by Hans E. Grossniklaus, MD, ophthalmic
pathologist.
Statistical analysis of the data was performed using an
analysis of variance (ANOVA) and unpaired t test. Find-
ings with an error probability less than 0.05 were consid-
ered to be statistically significant.
RESULTS
THE RESULTS OF THE NONCONTACT SPECULAR MICROS-
copy are summarized in Table 1. We found a mean ⫾ SD
of 2,730 ⫾ 224 cells/mm2 in the center of the cornea,
2,887 ⫾ 213 cells/mm2 in the paracentral region (2.7 ⫾
0.2 mm from the center), and 2,993 ⫾ 229 cells/mm2 in
OF OPHTHALMOLOGY MAY 2003
 TABLE 1. Central, Paracentral, and Peripheral Endothelial Cell Density*

No. of Increase in ECD Hexagonal Coefficient of

Corneal Region Corneas ECD Cells/mm2 From Center (%) Cells (%) Variation

Central 48 2,730 ⫾ 224†‡ — 63 ⫾ 6 0.32 ⫾ 0.04

Paracentral 48 2,887 ⫾ 213† 5.80 65 ⫾ 4 0.31 ⫾ 0.03
Peripheral 48 2,993 ⫾ 229‡ 9.60 65 ⫾ 5 0.31 ⫾ 0.03

ECD ⫽ endothelial cell density.

*Percent hexagonal cells and coefficient of variation (mean ⫾ SD) measured with Robo noncontact specular microscope (age 29.4 ⫾ 7.1
years).
†
P ⬍ .01.
‡
P ⬍ .001.

2,907 ⫾ 306 cells/mm2 in the center of the cornea versus

TABLE 2. Endothelial Cell Density (ECD) in Peripheral
3,165 ⫾ 323 cells/mm2 in the temporal peripheral region
Regions
(4.2 to 5.0 mm from the center) of the cornea (P ⬍ .05).
Corneal Region No. of Increase in ECD
These results showed an increase of 8.9%.
n ⫽ 47 Corneas ECD Cells/mm2 From Center (%) In both groups of corneas there was an increase in
endothelial cell density in the paracentral and peripheral
Central 48 2,730 ⫾ 224 —
regions as compared with the central region. In group 1,
Superior 48 3,166 ⫾ 272† 15.9
Nasal 48 2,944 ⫾ 292† 7.8
we found a 9.2% increase in the paracentral region (2.4 to
Inferior 48 2,912 ⫾ 276† 6.7 3.3 mm from the center) with an average paracentral cell
Temporal 48 2,992 ⫾ 226† 9.6 density of 2,717 ⫾ 390 cells/mm2 vs central 2,488 ⫾ 375
cells/mm2. In group 2, the increase was 17.2% in the
*Density 4.7 ⫾ 0.2 mm from center of the human cornea peripheral region (4.4 –5.3 mm from the center) with an
(mean ⫾ SD) measured with the Robo noncontact specular average peripheral cell density of 2,992 ⫾ 511 cells/mm2 vs
microscope. 2553 ⫾ 343 cells/mm2 in the center (P ⬍ .05; Table 4).
†
P ⬍ .001 from central ECD. With the use of a histological endothelial cell nomo-
gram,19 the cell density was calculated from histologic cross
sections of Optisol-GS eye bank corneas. The endothelial
the peripheral region (4.7 ⫾ 0.2 mm from the center). The cell density for the central region was 1,982 ⫾ 205
differences between the central and paracentral, as well as cells/mm2 vs 2,436 ⫾ 351 cells/mm2 for the peripheral
between the central and peripheral regions, were statisti- region, which represented a difference of 22.9% (P ⬍ .05;
cally significant (P ⬍ .01 and P ⬍ .001, respectively). Table 5).
Paracentral and peripheral endothelial cell density were
increased 5.8% and 9.6% compared with the central
endothelial cell density. These percentages were calculated DISCUSSION
by subtracting the central ECD from the respective para-
central or peripheral ECD and dividing the difference by THIS STUDY INVESTIGATED THE PERIPHERAL CORNEAL EN-
the central ECD. All percent increases in the four methods dothelial cell density in a systematic way using four
of ECD analysis were determined in this way. The highest different methods; two histologic methods, and two in vivo
endothelial cell density was found in the superior region of specular microscopy methods (contact and noncontact).
the peripheral cornea with an average of 3,166 ⫾ 272 In this study, we found an increase in endothelial cell
cells/mm2, a 15.9% increase (P ⬍ .001) compared with the density in the paracentral and the peripheral region of the
central region (Table 2). The percentage of hexagonal cornea compared with the central region. All four methods
cells was not significantly different between the three used to determine ECD yielded similar values leading us to
different areas of the cornea (central 63% ⫾ 6%, paracen- believe that the results were accurate. Additionally, all
tral 65% ⫾ 4%, and peripheral 65% ⫾ 5%). The coeffi- observed increases in endothelial cell density, except for
cient of variation of cell area (0.32 ⫾ 0.04 for the central, the paracentral versus central measurements in the alizarin
0.31 ⫾ 0.03 for the paracentral, and 0.31 ⫾ 0.03 for the red stained corneas, were statistically significant. Because
peripheral cornea) also showed no significant difference of the differing ages in the clinical study (younger patients)
(Table 1). and eye bank donors (older patients), the absolute average
The ECDs determined by contact specular microscopy ECD may be slightly changed when comparing the ECDs
are summarized in Table 3. We found an average of by decade. This study also shows that the peripheral ECD

VOL. 135, NO. 5 INCREASED ENDOTHELIAL CELL DENSITY IN HUMAN CORNEA 587
 TABLE 3. Central and Peripheral Endothelial Cell Density*

No. of Increase in ECD Hexagonal Coefficient of

Corneal Region Corneas ECD Cells/mm2 From Center (%) Cells (%) Variation

Central 21 2,907 ⫾ 306 — 62 ⫾ 8 0.31 ⫾ 0.05

Peripheral (temporal) 21 3,165 ⫾ 323† 8.9 63 ⫾ 7 0.29 ⫾ 0.05

ECD ⫽ endothelial cell density.

*Percent hexagonal cells and coefficient of variation (mean ⫾ SD) measured with Keeler-Konan contact specular microscope (age 30.9 ⫾
4.1 years).
†
P ⬍ .05.

TABLE 4. Central and Paracentral Endothelial Cell Density*

No. of Increase Cells/mm2 Hexagonal Coefficient of

Corneal Region Corneas ECD From Center (%) Cells (%) Variations

Group 1
Central 15 2,488 ⫾ 375 — 63 ⫾ 8 0.30 ⫾ 0.05
Paracentral 15 2,717 ⫾ 390 9.2 59 ⫾ 11 0.33 ⫾ 0.06
Group 2
Central 15 2,553 ⫾ 343 — 62 ⫾ 7 0.32 ⫾ 0.06
Peripheral 15 2,992 ⫾ 511† 17.2 55 ⫾ 10 0.35 ⫾ 0.07

ECD ⫽ endothelial cell density.

*Percent hexagonal cells and coefficient of variation (mean ⫾ SD) measured on alizarin red stained corneas (group 1, a 10 mm corneal
button, age 40.6 ⫾ 36 years; group 2, whole cornea with 1 to 2 mm scleral rim, age 55.3 ⫾ 16.1 years).
†
P ⬍ .05.

8000 Specular microscope. When the internal calibration

TABLE 5. Central and Peripheral Endothelial Cell Density
is used (which is not an X and Y calibration, but a single
(ECD)*
X or Y calibration) the cell density result is generally 6%
No. of Increase in ECD less than the result given using the contact specular
Central Region Corneas ECD† Cells/mm2 From Center (%) microscope (corners method). Ohno and colleagues20 re-
ported on the 6% difference and that only by bringing in
Central 13 1,982 ⫾ 205 —
Peripheral 13 2,436 ⫾ 351‡ 22.90
an external calibration with appropriate software would
the result for the contact and the Robo noncontact
ECD ⫽ endothelial cell density. microscopes be essentially the same. The present study was
*Mean ⫾ SD determined from histological cross-sections (age primarily interested in the difference between the ECDs in
45.5 ⫾ 19.2 years). the three regions (the percent difference) as measured with
†
ECD determined from nomogram in Williams and cowork- the instrument in use. The basis of this report is a
ers.19
systematic evaluation of ECD using four distinct methods.
‡
P ⬍ .05.
It is important to note that the two methodologies (Robo
noncontact vs Keeler-Konan contact specular microscope)
yielded a similar percentage difference when comparing
in the older alizarin red stained and histological corneas central to peripheral ECDs. The peripheral ECD was 8.9%
were age dependent and there was a greater difference in and 9.6% increased compared with central for the contact
cell number between the central and peripheral regions and noncontact methods, respectively.
compared with the younger specular microscopy corneas. In previous studies using histological methods to inves-
The central and peripheral cell densities in the Keeler- tigate the cell density distribution,11,12,14,18 the central cell
Konan contact group were approximately 6% higher than density was found to be between 10% and 15% decreased
in the Robo noncontact group (6.4% for central and 5.7% compared with the periphery. However, Binder and asso-
peripheral) with similarly aged patients. This difference is ciates13 using scanning electron microscopy on cadaver
attributable to the internal calibration that is used in the eyes found that the central corneal endothelial cell density
center dot method analysis provided with the Robo SP- was 10% higher than in the periphery. This variation may

588 AMERICAN JOURNAL OF OPHTHALMOLOGY MAY 2003

be explained by the fact that different areas in the
peripheral cornea may have been evaluated. Most reported
studies do not give exact locations where the peripheral
cell density was measured. The area of highest cell density
in the periphery appears to be a small ring of endothelial
cells close to the Schwalbe line. Schimmelpfennig11 has
previously reported a 23% increase in ECD in this area.
Using noncontact and contact specular microscopy, we
also found an increased endothelial cell density in the
paracentral and peripheral areas of the cornea. In previous
investigations, the endothelial cell densities determined by
contact specular microscopy varied markedly, and some
published reports found no significant difference between
the central and the peripheral endothelial cell density.8,21
In a recent study using a contact specular microscope,
Wiffen and colleagues10 reported a significantly higher
endothelial cell density in the central region of the cornea
compared with the peripheral region of normal subjects.
The results of a study by Cheung and Cho22 show an
increased ECD in the peripheral corneal versus the central
area. However; the purpose of that study was to evaluate
the performance of the TOPCON SP-2000P and IM-
AGEnet system, and the authors did not emphasize the
increase in the peripheral distribution of the corneal
endothelial cells.
In another study, Bourne and Enoch23 demonstrated
that the magnification effect of the specular microscope
may cause the corneal ECD to vary less than 1% over the
range of corneal thickness. Although the difference in
ECD may be small, the authors state that one should
correct for the magnification effect of the specular micro-
scope. In this study, differences between the central and
the peripheral corneal endothelial cell density are much
larger than the 1% magnification effect.
Our data on ECD confirm the findings of Schim-
melpfennig11 and Daus and colleagues14,15 with both aliza-
rin red stained corneas and in vivo specular microscopy.
We were also able to confirm with specular microscopy the
Daus and associates histologic observation14 that the
highest cell counts were found in the superior region of the
peripheral human cornea. Investigating the normal distri-
bution of ECD in 75 alizarin red stained human corneas
(age range 3 to 75 years), our group has recently found a
higher cell density in the far peripheral region (next to
Schwalbe line) in all corneas regardless of age (GP Holley,
unpublished data, Association for Research in Vision and
Ophthalmology poster 2002). Daus and associates14,15
found similar results when they performed endothelial cell
mapping in 30 human donor corneas. Recent studies by
Bednarz and Engelman24 and Bednarz and associates25
have shown that there is a zone of endothelial cell
regeneration in the peripheral region of the cornea. The
authors also propose that the peripheral endothelial cells
may serve as precursor (stem) cells for the corneal endo-
thelium. Additionally, Whikehart and colleagues recently
reported significant telomerase activity in endothelial cells
VOL. 135, NO. 5 INCREASED ENDOTHELIAL CELL DENSITY
adjacent to the Schwalbe line, and lack of activity in the
central corneal endothelial cells (DR Whikehart, unpub-
lished data, Association for Research in Vision and Oph-
thalmology poster 2002). As only cancer cells, white blood
cells, and stem cells have telomerase activity these findings
suggest that corneal endothelial cells from the peripheral
area may contain progenitor (stem) cells that have the
ability to divide. The peripheral regenerative zone may
provide a slow continuous population of cells used to
maintain the central ECD. Over a lifetime, it has been
reported that there is a loss of central endothelial cells at
a rate of 0.5% to 0.6% per year,2 which is a small decline
and amounts to a loss of approximately 100 to 200 cells per
year from the central corneal endothelium. Therefore, the
endothelium adjacent to the Schwalbe line may be the
area of continual cellular supply used to maintain the
central corneal endothelial cell density at a relatively
constant level during one’s lifetime. The increased ECD in
the periphery is very important in the wound healing
response of the corneal endothelium because the periph-
eral cells provide a physiologic reserve and storage region
of endothelial cells.
In summary, this study establishes that there is an
increase in ECD in both the paracentral and peripheral
corneal endothelium. With careful use of the noncontact
and contact specular microscope an increase in ECD can
easily be measured and monitored in patients after refrac-
tive and intraocular surgical procedures. Peripheral ECD
will also be of interest after phakic intraocular lens
implantation and in glaucoma patients after drainage filters
with mitomycin C and tube shunts.
ACKNOWLEDGMENTS
A special thanks to Hans E. Grossniklaus, MD, for deter-
mining the high-power endothelial cell counts.
REFERENCES
1. Vogt A. Die Sichtbarkeit des lebenden Hornhautendothels:
ein Beitrag zur Methodik der Spaltlampenmikroskopie.
Graefes Arch Clin Exp Ophthalmol 1920;101:123–132.
2. Bourne WM, Nelson LF, Hodge DO. Central corneal endo-
thelial cell changes over a ten-year period. Invest Ophthal-
mol Vis Sci 1997;38:779 –782.
3. Carlson KH, Bourne WM, Brubaker RF. Effect of long-term
contact lens wear on corneal endothelial cell morphology
and function. Invest Ophthalmol Vis Sci 1988;29:185–193.
4. MacRae SM, Matsuda M, Philips DS. The long-term effects
of polymethylmethacrylate contact lens wear on the corneal
endothelium. Ophthalmology 1994;101:365–370.
5. Setaelae K, Vasara K, Vest E, Ruusuvaara P. Effects of
long-term contact lens wear on the corneal endothelium.
Acta Ophthalmol Scand 1998;76:299 –303.
6. Schultz RO, Glasser DB, Matsuda M, Yee RW, Edelhauser
HF. Response of the corneal endothelium to cataract surgery.
Arch Ophthalmol 1986;104:1164 –1169.
7. Ing JJ, Ing HH, Nelson LR, Hodge DO, Bourne WM.
IN HUMAN CORNEA 589
 Ten-year postoperative results of penetrating keratoplasty.
Ophthalmology 1998;105:1855–1865.
8. Blackwell WL, Gravenstein N, Kaufman HE. Comparison of
central corneal endothelial cell numbers with peripheral
areas. Am J Ophthalmol 1977;84:473–476.
9. Hoffer KJ. Vertical endothelial cell disparity. Am J Ophthal-
mol 1979;87:344 –349.
10. Wiffen SJ, Hodge DO, Bourne WM. The effect of contact
lens wear on the central and peripheral corneal endothelium.
Cornea 2000;19:47–51.
11. Schimmelpfennig BH. Direct and indirect determination of
nonuniform cell density distribution in human corneal en-
dothelium. Invest Ophthalmol Vis Sci 1984;25:223–229.
12. Irvine AR, Irvine AR Jr. Variations in normal human
corneal endothelium: a preliminary report of pathologic
human corneal endothelium. Am J Ophthalmol 1953;36:
1279 –1285.
13. Binder PS, Akers P, Zavala EY. Endothelial cell density
determined by specular microscopy and scanning electron
microscopy. Ophthalmology 1979;86:1831–1847.
14. Daus W, Voelker HE, Meysen H, Bundschuh W. Vitalfaer-
bung der Hornhautendothels-erweiterte Moeglichkeitein der
Diagnostik. Fortschr Ophthalmol 1989;86:259 –264.
15. Daus W, Voelker HE, Meysen H. Clinical significance of
age-related regional differences in distribution of human
corneal endothelium. Klin Mbl Augenheilk 1990;196:449 –
455.
16. Matsuda M, Suda T, Manabe R. Quantitative analysis of
endothelial mosaic pattern changes in anterior keratoconus.
Am J Ophthalmol 1984;98:43–49.
590 AMERICAN JOURNAL
17. Schultz RO, Matsuda M, Yee RW, Edelhauser HF, Schultz
KJ. Corneal endothelial changes in type I and type II
diabetes mellitus. Am J Ophthalmol 1984;98:401–410.
18. Taylor MJ, Hunt CJ. Dual staining of corneal endothelium
with trypan blue and alizarin red S: importance of pH for the
dye-lake reaction. Br J Ophthalmol 1981;65:815–819.
19. Williams KK, Noe RL, Grossniklaus HE, Drews-Botsch C,
Edelhauser HF. Correlation of histologic corneal endothelial
cell counts with specular microscopic cell density. Arch
Ophthalmol 1992;110:1146 –1149.
20. Ohno K, Nelson LR, McLaren JW, Hodge DO, Bourne W.
Comparison of researching systems and analysis methods in
specular microscopy. Cornea 1999;18:415–423.
21. Yee RW, Matsuda M, Schultz RO, Edelhauser HF. Changes
in the normal corneal endothelial cellular pattern as a
function of age. Curr Eye Res 1985;4:671–678.
22. Cheung SW, Cho P. Endothelial cells analysis with the
TOPCON specular microscope SP-2000P and IMAGEnet
system. Curr Eye Res 2000;21:788 –798.
23. Bourne WM, Enoch JM. Some optical principles of the
clinical specular microscope. Invest Ophthalmol Vis Sci
1976;15:29 –32.
24. Bednarz J, Engelmann K. Indication of precursor cells in
adult human corneal endothelium. Invest Ophtalmol Vis Sci
2001;42(Suppl):5274.
25. Bednarz J, Rodkanaki-von Schrenck A, Engelmann K. Dif-
ferent characteristics of endothelial cells from central and
peripheral human cornea in primary culture after sub-culture.
In Vitro Cell Develop Biol Ann 1998;34:149 –153.
OF OPHTHALMOLOGY MAY 2003