Continued Endothelial Cell Loss

Ten Years after Lens

Implantation

William M. Bourne, MD,I Leif R. Nelson, BA,I David O. Hodge, MS 2

Purpose: To investigate the effects of cataract extraction and lens implantation on

the central corneal endothelium 10 years after surgery.
Methods: The authors conducted a prospective study of 253 consecutive eyes
that underwent cataract extraction with or without lens implantation by one surgeon
from 1976 to 1982. Three types of lens implant were used during this period. The
protocol included ophthalmic examinations and specular microscopy on all eyes pre-
operatively, and 2 months and 1, 3, 5, and 10 years postoperatively.
Results: The 10-year analysis was conducted on 67 (26%) of the 253 total eyes.
The remaining patients died (86 eyes [34%]), were unable to return 10 years later (93
eyes [37%]), or had secondary implants (5 eyes [2%]) or penetrating keratoplasty (2
eyes [1%]). There were no statistically significant differences among the median 10-
year endothelial cell losses of 36% in 17 control cataract extractions without lens im-
plantation (15 extracapsular and 2 intracapsular), 40% in 15 medallion iris suture implants,
32% in 28 transiridectomy clip implants, and 32% in 7 posterior chamber implants. The
median exponential rate of chronic cell loss from 1 to 10 years after surgery was 2.5%
per year, which did not differ significantly among the three implant groups or between
the implants (2.4% per year) and controls (2.7% per year). The chronic cell loss rate
was significantly higher (7.2% per year) in six eyes with cornea guttata, which was the
only preoperative endothelial morphologic feature that was significantly associated with
the chronic cell loss rate.
Conclusions: Ten years after cataract extraction, eyes continued to lose endothelial
cells from the central cornea at a rate of 2.5% per year, 2.5 to 8.0 times the rate in
healthy unoperated eyes. The rate was not affected significantly by the presence of the
three types of lens implants that the authors used. Postoperative eyes with cornea
guttata continued to lose cells at more than twice this rate. Preoperative specular mi-
croscopy did not provide additional information helpful in predicting postoperative en-
dothelial status or outcome. Ophthalmology 1994;101:1014-1023

We know from published investigations that the number

Originally received: October 12, 1993.
Revision accepted: January 31, 1994.
I Department of Ophthalmology, Mayo Clinic and Mayo Foundation,
Rochester.
2 Section of Biostatistics, Mayo Clinic and Mayo Foundation, Rochester.
Presented at the American Academy of Ophthalmology Annual Meeting,
Chicago, November 1993.
Supported in part by research grants EY 02037 and EY 08039 from the
National Institutes of Health, Bethesda, Maryland, an unrestricted grant
from Research to Prevent Blindness, Inc, New York, New York, and
the Mayo Foundation, Rochester, Minnesota.
Reprints requests to William M. Bourne, MD, Mayo Clinic, 200 First
St SW, Rochester, MN 55905.
of central corneal endothelial cells after intraocular lens
implantation decreases at rates greater than those of
healthy unoperated corneas for at least 5 years. 1,2 It is not
known if this accelerated cell loss continues or if the cor-
neas eventually stabilize with the number of central en-
dothelial cells decreasing at a minimal rate that is no
greater than that in unoperated eyes. We conducted a
prospective trial of lens implantation at the Mayo Clinic
from 1976 to 19823 ; the 5-year postoperative results were
reported in 1989. 1 To investigate the endothelial changes
that occur over a longer period, we studied the lO-year
postoperative results from this trial in the cohort with
operations performed by a single surgeon.

1014
 Bourne et al . Endothelial Cell Loss 10 Years after Lens Implant
Methods
Two hundred fifty-three eyes of 206 patients were entered
into the trial by one surgeon (WMB) from 1976 to 1982.
Informed consent was obtained from all patients. We ex-
cluded patients with uveitis or proliferative diabetic reti-
nopathy, or those who had had previous intraocular surgery.
Those younger than 65 years of age and those with glaucoma,
nonproliferative diabetic retinopathy, or cornea guttata were
excluded initially, but included later. Cornea guttata was
graded according to four stages: 1+ (a small number of
sparsely scattered guttae) to 4+ (confluent guttae). After
routine observation during the first postoperative week, all
patients were asked to return for examinations 2 months
and 1, 3, 5, and 10 years after cataract extraction. We asked
all living patients who were unable to return for the 1O-year
examination to complete a survey inquiring about visual
acuity and any further ocular procedures.
The surgical techniques have been described previ-
ously".4 We administered preoperative digital ocular
massage for 10 minutes to most eyes. For intracapsular
operations, we performed cryoextractions; a-chymotryp-
sin was instilled in the posteriQr chamber in one fourth
of the eyes. For extracapsular procedures, we used an an-
teriorcapsulotomy, nuclear expression, and cortical as-
piration. Six eyes received a posterior capsulotomy at sur-
gery. A commercial balanced salt solution was used for
intraocular irrigation; it was supplemented with glucose
(5.6 mM), sodium bicarbonate (18 mM), and epinephrine
(I :5,000,000) in one half of the eyes. To dilate or constrict
the pupil during the procedure, the surgeon injected into
the anterior chamber acetylcholine (I %) in 17 eyes, car-
bachol (0.0 1%) in 33 eyes, and epinephrine (l: lOO,OOO)
in 21 eyes.
Of the 253 total eyes that underwent cataract extrac-
tion, 76 did not receive an implant and are considered
controls. We implanted three types of lens implants se-
quentially over the course of the trial. Of the 177 eyes
receiving implants, the first 54 received a medallion iris
suture lens with intracapsular cataract extraction and a
10-0 polypropylene iris suture. The next 90 eyes received
a transiridectomy clip lens with extracapsular extraction.
This lens is a two-loop, capsular-fixated implant with a
prepupillary lenticular portion. The final 32 eyes received
a posterior chamber lens with extracapsular extraction;
the implant had two J-Ioops that were placed within the
capsular bag. One additional patient received an angle-
supported anterior chamber lens after extracapsular ex-
traction and anterior vitrectomy. Intraocular viscoelastic
materials were not used.
Topical corticosteroids were administered in low doses
for the first 3 to 6 postoperative weeks. We performed
central corneal endothelial cell photography and pachy-
metry with a specular microscope and applanation to-
nometry preoperatively and 2 months and I, 3, 5, and 10
years postoperatively. In some of the eyes with cornea
guttata, it was necessary to take photographs of the para-
central cornea to obtain cellular images that could be an-
alyzed. We also recorded any complications, reoperations,
or contact lens wear.
Individual endothelial cell sizes were measured from
the projected images of the photographic negatives mag-
nified approximately 500 times. We either used a com-
puter-assisted image-processing system5 to analyze trac-
ings of all visible cell outlines on three photographs or
measured the areas of at least 50 individual cells with an
electronic planimeter. The two methods of analysis had
comparable accuracy and reproducibility.6 Mean cell area
was converted to cell density (cells/mm2), and cell loss
was expressed as a percentage of the preoperative cell
density. As a measure of polymegethism, we calculated
the coefficient of variation of cell area (standard deviation/
mean) for each examination.
We derived an exponential model rather than a linear
one for the decrease in endothelial cell density over time
more than I year after cataract extraction. We chose the
exponential model as being more compatible with biologic
changes, which are usually first-order processes. Only the
data from the 1., 3-, 5-, and 1O-year examinations were
used in case the endothelial cell population had not sta-
bilized within 2 months of the operation. 7,8 Using the
least-squares method, we computed a regression line for
each eye by applying the natural logarithm of cell density
versus time in years. The slope of this line was used as a
measure of the annual rate of change of cell density for
that eye. The rates derived are thus the instantaneous an-
nual rates of chronic postoperative endothelial cell loss.
Three eyes (1 implant, 2 controls) had measurements for
only the I O-year examination, and therefore were excluded
from this analysis.
We used both parametric and nonparametric statistical
analyses, based on the distribution of the data under con-
sideration. Group differences for continuous variables
were tested using the two-sample Student's t test for nor-
mally distributed data and the Wilcoxon rank-sum test
for non-normally distributed data. Differences among
more than two groups were tested using analysis of vari-
ance for normal data, and the Kruskal-Wallis test for non-
normal data. Differences for categorical variables were
tested using the chi-square test for independence. Corre-
lations between continuous variables were obtained using
Pearson's correlation coefficient for normal data and
Spearman's rank correlation coefficient for non-normal
data. A two-tailed probability of 0.05 or less was consid-
ered statistically significant.
For several comparisons in which no statistically sig-
nificant difference was found, we calculated the power of
the study to detect a true difference. For those comparisons
involving data that were not distributed normally, the
power estimates represent approximations made under
the assumption of a normal distribution.
Results
Of the 253 eyes entered in the trial, 1O-year postoperative
examinations were performed on 67 eyes of 57 patients.
Eighty-six eyes were in 69 patients who were known to
have died within 10 years after cataract extraction, and
93 eyes were in 83 patients who were unable to complete
1015
 Ophthalmology Volume 101, Number 6, June 1994

the final examination. The remaining seven eyes returned
but were excluded from the analysis: 5 control eyes that
had undergone secondary lens implantation and 2 control
eyes that had undergone subsequent penetrating kerato-
plasty (both had 3+ cornea guttata preoperatively). There
were no statistically significant differences between the
186 eyes that did not return at ten years and the 67 eyes
with 1O-year follow-up, both control and implant groups,
for any preoperative variables (Tables 1 and 2). The en-
dothelial cell loss noted 2 months postoperatively, how-
ever, was significantly larger in the patients returning at
10 years (Table 2). Of the 67 eyes with 1O-year data, 17
were controls without lens implantation (2 intracapsular
and 15 extracapsular extractions), 15 had medallion iris
suture implants, 28 had transiridectomy clip implants,
and 7 had posterior chamber implants. All of the lens
implantations except the first two medallion iris suture
implants were performed under air. 5
We attempted to locate and send survey letters to the
83 patients (93 eyes) who were unable to complete the
10-year examination. Responses were obtained from 22
patients (25 eyes, 19 implants, 6 controls). All except two
patients (1 implant and 1 control) could read newsprint;
only one eye, a control, was reported as having had a
secondary ocular procedure (a posterior capsulotomy)
since the cataract extraction.
Preoperative data for the 67 eyes with 1O-year follow-
up are given in Tables 3 and 4. The preoperative endo-
thelial cell densities ranged from 1594 to 3581 cells/mm2.
The controls differed from the implant group in that they
were younger and had a higher percentage with cornea
guttata and with diabetes mellitus.
The endothelial cell densities and cell losses for the 10-
year postoperative period are presented in Table 5 and
Figure 1. The ranges of actual examination times for each
of the postoperative intervals are 41 to 97 days (mean, 58
days) for the 2-month examination, 186 to 538 days
(mean, 379 days)for the I-year examination, 734 to 1415
days (mean, 1104 days) for the 3-year examination, 1462
to 2288 days (mean, 1841 days) for the 5-year examina-
tion, and 3303 to 4432 days (mean, 3728 days) for the
1O-year examination. The mean cell densities were de-
creased significantly from the previous interval at the 2-
month (paired Student's t test, Sd = 470 cells/mm2, P <
0.001), 3-year (Sd = 448 cells/mm2, P < 0.001), and 10-
year (Sd = 417 cells/mm2, P < 0.001) examinations. There
was a significant correlation between the preoperative and
1O-year postoperative cell densities (r = 0.61, P < 0.001).
There was no significant difference between implant and
control groups at any interval. When the eyes with cornea
guttata were removed from both groups, there was still
no significant difference in endothelial cell loss at 10 years
between the control group and the implant group.
Table 6 and Figure 2 present the endothelial cell den-
sities and cell losses in the three implant groups. A sig-
nificant difference in cell density among the three groups
was present at all intervals except 2 months postopera-
tively, the eyes with medallion iris suture implants having
lower densities. The average ages of the patients in the
three implant groups were not significantly different

Table 1. Preoperative Categorical Variables in Eyes with a Ten-year Follow-up

Compared with Those without a Ten-year Follow-up
10-yr Follow-up No 10-yr Follow-up
(n = 67) (n = 186)
Variable No. (%) No. (%) P*
Males 26 (39) 74 (40) 0.89
Implants 50 (75) 127 (68) 0.33
Lens style
Iris suture medallion 15 (22) 39 (21) 0.81
Posterior chamber 7 (11) 25 (13) 0.53
T ransiridectomy clip 28 (42) 62 (33) 0.21
Anterior chamber 0(0) 1 (1) 0.54
Controls 17 (25) 59 (32) 0.33
Cornea guttata
Absent 56 (84) 167 (90) 0.18
1+ 5 (8) 9 (5)
2+ 4 (6) 3 (2)
3+ 2 (3) 6 (3)
4+ 0(0) 1 (1)
Diabetes mellitus 5 (8) 9 (5) 0.42
Glaucoma 0 3 (2) 0.30
Vitreous loss at surgery 4 (6) 6 (3) 0.32
(anterior vitrectomy)

• Chi-square test.

1016
 Bourne et al . Endothelial Cell Loss 10 Years after Lens Implant

Table 2. Continuous Variables in Eyes with a Ten-year Follow-up Compared

with Those without a Ten-year Follow-up
10-yr Follow-up No 10-yr Follow-up
(n = 67) (n = 186)
Preoperative Variable Mean ± SD Median Mean ± SD Median p.
Age (yrs) 70 ± 8 70 72±13 74 0.10
Endothelial cell density (cells/mm2) 2703 ± 428 2737 2675 ± 540 2697 0.71
Coefficient of variation of cell area 0.30 ± 0.06 0.30 0.30 ± 0.12 0.29 0.78
Corneal thickness (mm) 0.55 ± 0.04 0.55 0.55 ± 0.04 0.55 0.80
Intraocular pressure (mmHg) 15 ± 3 16 15 ± 3 15 0.80
2-mo postoperative endothelial cell loss (%)t 17 ± 18 16 10 ± 21 9 0O. O8t

SD = standard deviation .
• Two-sample Student's t test.
t n = 66 and 180.
t Wilcoxon rank·sum test.

(P = 0.61). There were no significant differences in cell There were 38 reoperations required by 32 of the 67
loss among the three implant groups at any postoperative eyes over their IO-year postoperative courses. We per-
interval. formed posterior capsulotomies, either surgical discissions
The annual rates of chronic postoperative endothelial or Y AG laser capsulotomies, in 2S (SO%) of the SO eyes
cell loss for the three implant types are included in Table with extracapsular extractions, in S (33%) of 15 controls,
7. The rates did not appear to decrease with time (Figure and in 20 (57%) of 35 implants. There was no statistically
1); the annual cell loss from one to five years was not significant difference in chronic endothelial cell loss be-
significantly different from that from five to 10 years (P tween the eyes with discissions or capsulotomies and those
= 0.86). From one to 10 years postoperatively, the cell without, whether in all eyes (P = 0.41), controls (P =
density decreased 2.S% (median) in all 64 eyes-2.4% per 0.52), or implants (P = 0.82). There were three scleral
year in the implant group and 2.7% per year in the control buckles, two panretinal photocoagulations, and one each
group. The difference in rates between the two groups was of the following procedures: wound repair, peripheral ir-
not significant (P = 0.S2). There was also no difference idectomy, argon laser trabeculoplasty, trephine filtering
among the three types of implants (P = 0.60). The rate procedure, suture fixation of implant, anterior chamber
of chronic cell loss was significantly correlated with the reformation after scleral buckle, retinal cryopexy, and pars
2~month cell loss (p = 0.32, P = 0.01). plana vitrectomy.
We attempted to identify preoperative factors that were We recorded 73 postoperative complications (48 im-
predictive of chronic endothelial cell loss. There was no plants and 25 controls) in 42 eyes (31 implants and II
significant correlation between preoperative endothelial controls) of 37 patients over their IO-year postoperative
cell density, coefficient of variation of cell area, corneal courses. Opacification of the posterior capsule developed
thickness, intraocular pressure, age, or sex with the 10-
year cell loss or annual rate of chronic cell loss. Specifically,
the preoperative coefficient of variation of cell area was Table 3. Preoperative Data: Categorical Variables
not correlated significantly with cell density (Spearman's
p = 0.09, P = 0.49) or cell loss (p = O.IS, P = 0.24) 10 Implants Controls
years postoperatively or with the rate of chronic postop- (n = 50) (n = 17)
erative cell loss (p = O.OS, P = 0.72). The only preoperative Variable No. (%) No. (%) p.
factor listed in Table I that was associated with an increase
in the chronic cell loss was the presence of cornea guttata. Males 20 (40) 6 (35) 0.73
For this analysis we used only the six eyes with 2+ or Cornea guttata
more cornea guttata; these eyes would be considered by Absent 45 (90) 11 (65) 0.02
1+ 3 (6) 2 (12)
most examiners to have a progressive endotheliopathy
2+ 1 (2) 3 (18)
(i.e., early Fuchs dystrophy). Although there was no sig- 3+ 1 (2) 1 (6)
nificant difference in the 2-month cell loss, the six eyes
Diabetes mellitus 1 (2) 4 (24) 0.01
with cornea guttata had significantly greater cell losses at
10 years (median, SI % versus 31 %; P = 0.03) and greater Vitreous loss at surgery
(anterior vitrectomy) 2 (4) 2 (12) 0.24
rates of chronic endothelial cell loss (median, 7.2% versus
2.1 % per year; P = 0.003, Wilcoxon rank-sum test) than • Chi-square test.
the 61 eyes without guttae.

1017
 Ophthalmology Volume 101, Number 6, June 1994

Table 4. Preoperative Data: Continuous Variables

Implants (n = 50) Controls (n = 17)
Variable Mean ± SD Median Mean ± SD Median p'
Age (yrs) 72 ±6 73 62 ±8 62 <0.01
Intraocular pressure (mmHg) 16 ±3 16 15 ±2 16 0.42
Corneal thickness (mm) 0.55 ± 0.04 0.54 0.56 ± 0.04 0.56 0.26
Endothelial cell density (cells/mm2) 2710 ± 419 2743 2684 ± 466 2666 0.84
Coefficient of variation of cell area 0.29 ± 0.06 0.28 0.32 ± 0.06 0.33 0.06t

SD = standard deviation .
• Two-sample Student's t test.
t Wilcoxon rank-sum test.

in 25 eyes and hyphemas occurred in 8. The hyphemas
all cleared without treatment or sequelae. The remaining
complications and number of eyes for each were: seven
with macular edema, six with glaucoma, five with post-
operative ocular hypertension requiring short-term treat-
ment, five with vitreocorneal contact, four with failure of
implant to achieve capsular fixation, three with retinal
detachment, two with implant dislocation, two with cor-
neal contact by implant, two with proliferative diabetic
retinopathy, one with wound leak, one with pupillary
block, one with varicella zoster keratitis, and one with
retinal tear. The 10-0 polypropylene iris suture was intact
in all 15 eyes with the medallion iris suture implant 10
years after surgery; however, the suture subsequently has
broken in two of the eyes.
Chronic open-angle glaucoma developed from 3 to 10
years postoperatively in six eyes (3 implants, 3 controls)
of five patients. Although these eyes had a greater annual
rate of chronic postoperative endothelial cell loss (median,
5.0% versus 2.2%; P = 0.18) and a greater lO-year cell
loss (median, 51 % versus 32%; P = 0.09) than the eyes
without glaucoma, the differences were not statistically
significant. One of the eyes underwent argon laser tra-
beculoplasty and a trephine filtering procedure.
The five eyes of patients with diabetes (Table 1) were
not significantly different from the 62 eyes of those without
diabetes, with respect to corneal thickness, endothelial
cell density, or the rate of chronic endothelial cell loss.
There was also no significant difference in 10-year cell
loss or annual rate of late endothelial cell loss between
the eight control eyes that wore contact lenses and the
nine control eyes that did not.
Discussion
This report illustrates the difficulties in studying the long-
term (lO-year) results of cataract extraction, a procedure
in which the average age of the patient was 69.5 years.
Approximately one third of patients died within 10 years
of the operation, and another third was unable to complete
the lO-year evaluation. We attempted to survey the pa-

Table 5. Endothelial Cell Density and Cell Loss·

Implants (n = 50) Controls (n = 17)
Cell Density Cell Density
(cells/mm 2) Cell Loss (%j (cells/mm 2) Cell Loss (%j
Postoperative
Time No. Mean ± SD Median Mean ±SD Median No. Mean ± SD Median Mean ± SD Median
Preoperative 50 2710 ± 419 2743 17 2684 ± 466 2666
2 mos 49 2221 ± 588 2179 19 ± 18 17 17 2344 ± 427 2206 12 ± 17 14
1 yr 48 2181 ± 601 2247 20 ± 17 18 15 2343 ± 457 2458 12 ± 20 18
3 yrs 49 2004 ± 672 1988 27 ± 19 25 15 2236±61O 2257 17 ± 15 20
5 yrs 47 1926 ± 610 1984 29 ± 18 24 15 2123 ± 705 2033 22 ± 18 19
10 yrs 50 1718 ± 548 1734 37 ± 17 32 17 1743 ± 659 1885 36 ± 19 36
10 yrs, no cornea
guttata 45 1734 ± 553 1736 36 ± 17 32 11 1991 ± 639 2116 32 ± 18 26

SD = standard deviation .
• The mean cell densities at the 2-month, 3-year, and IO-year examinations were Significantly decreased from the previous interval (P < 0.001, two-
tailed paired Student's t test). There were no statistically significant differences between the implant and control groups at any interval in either
endothelial cell density (Student's t test) or cell loss (Wilcoxon rank-sum test).

1018
 Bourne et al Endothelial Cell Loss 10 Years after Lens Implant
3,000
Implants (n = 50) IN
2,800
Controls (n =17)
Gi~
u"
_E 2,600
.!!! E
.c
-CD
Ou
=
GiCil 2,400
0""'
""'NN...... " """
............
"O~
2,200 +1 +1 +1 +1 +1
5i.~ ...... N""'oo""
e: UI
",e: 2,000 ---------------------------------------------------_. 'l""""41"""'4NNrt"')

CD CD
~"O
1,800

1,600 +----r------,c------r-----r---~
o 2 6 8 10

Time after cataract extraction (yr)

Figure 1. Changes in mean endothelial cell density over time in 67 eyes
observed for 10 years after cataract extraction. The mean cell densities
at the 2-month and 3- and 10-year examinations were decreased signifi-
cantly ftom the previous interval (P < 0.001, paired Student's t test).
There was no statistically significant difference between eyes with and
without lens implants.

tients who did not return, but most did not respond. Of ~
the 74 eyes that were followed 10 years postoperatively, ,.-J a
~
7 had a secondary procedure (secondary implant or pen- U
V)
+1
etrating keratoplasty), which made the IO-year exami- +1 +1 +1 +1 +1
O\OON"""","
nation of the corneal endothelium invalid. Thus, the study .....-I ....... N N ("f"')

is based on the remaining 67 eyes, which represent only

26% of the 253 eyes that were entered in the trial. This
proportion, however, is not low for a clinical study, es-
pecially a 10-year postoperative one. Martin et al 9 were
able to achieve examinations in only 29% of patients (17
of 59) 5 years after cataract surgery and 7% (4 of 59) 7
years after surgery. Werblin lO was able to examine 37%
of eyes (34 of 93) 5 years after phacoemulsification.
One caveat concerning this study is that we cannot be
certain that differences did not exist between the 67 eyes ::::::
Q)
that returned 10 years postoperatively and the 186 eyes U
that did not. The group that returned did seem represen-
tative of the entire cohort preoperatively (Tables I and
2). We excluded patients with glaucoma and cornea gut-
tata from the early period of the study, so fewer patients
in this trial may have these conditions than those under- ooNN"","oo
........ N N N ........
going cataract extraction in the general population. Thus, +1 +1 +1 +1 +1
our results may represent a "best-case" scenario in which "'" !'- 00 !'- ""'"
N N "'" "'" ""'"
the continuing cell loss that we measured is actually less
than that in the average postoperative eye. On the other
hand, the patients who returned at 10 years had signifi-
cantly more operative (2-month) cell loss than those who
did not (Table 2). Because operative cell loss was correlated
significantly with the chronic postoperative loss rate, the
chronic rate that we measured might be artificially high.
We have no way of knowing if either of these possibilities
is operative; they would, however, tend to offset one an-
other.
A second caveat concerning this study is that the con-
trol group is not truly a valid one. Even in 1976, we be-
lieved that the decision of having a lens implant needed
to be made by the well-informed patient; we could not
assign patients randomly to receive or not receive a lens
implant with their cataract extractions to achieve a valid

1019
 Ophthalmology
2,800
--.
_ .•
----
2,600
•••
2,400 - Control(n=l?)
2,200 :
o
o ters. 12- 14 Because of the relatively small number of eyes
2,000 •
..............
'. '.
.
1,600
'. '
1,600
1,400
.... _---------------------------------
+-----..-----r----.....-----.---'=......,
o 2 4 6
Time after cataract extraction (yr)
Figure 2. Changes in mean endothelial cell density over time by implant
type. The 15 eyes with medallion iris suture lenses had significantly lower
endothelial cell densities preoperatively as well as postoperatively, so that
there was no significant difference in cell loss among the four groups.
control group. As an alternative, we used eyes that· met
the study criteria, were operated on during the study pe-
riod, and did not receive implants. Given the choice, most
patients elected to receive an implant, so the control group
is small. In the early stages of the study, we were reluctant
to place lens implants in eyes with cornea guttata or in
eyes of patients who were young or had diabetes mellitus.
Therefore, the control group was younger and contained
a significantly higher percentage of eyes with cornea gut-
tata and with diabetes (Tables 3 and 4).
A third caveat for this investigation is the possibility
that some of the significant differences that we found oc-
curred by chance alone because of the large number of
statistical tests that we performed. II This possibility is
more likely for the statistically significant comparisons
with probabilities near 0.05. Conversely, because of the
relatively small number of patients returning at 10 years,
a negative test of significance excludes only large differ-
ences. The study could detect as significant, at the 5%
level with 80% power, a difference in cell density 10 years
postoperatively of 454 cells/mm2 and a difference in cell
loss of 14%, if such differences existed. If the six eyes with
cornea guttata are omitted, the minimal detectable dif-
ference at 80% power is even greater (17% cell loss). In
Table 7. Rate of Chronic Postoperative Endothelial Cell Loss
Group
Controls
Implants
Medallion iris suture
Transiridectomy clip
Posterior chamber
SD = standard deviation .
• Wilcoxon rank-sum test.
t Kruskal-Wallis test.
1020
Volume 101, Number 6, June 1994
addition, we can only exclude with 80% confidence a dif-
Medallion (n=15)
Transiridectomyclip(n=28)
ference in rate of chronic cell loss of 2.8% per year or
Posterior chamber (n = 7) greater between the implanted eyes and controls.
We included ten bilateral cases in this study despite
the known intrasubject correlations for many parame-
remaining for analysis after 10 years, we elected to include
these ten eyes in the study because of the lack of evidence
or rationale for an intrasubject correlation for the param-
eter of concern, chronic endothelial cell loss. The corre-
lation between the two eyes of the ten bilateral subjects
8 10
was not significant for either the lO-year cell loss (p =
-0.28, P = 0.43) or the annual rate of chronic postop-
erative cell loss (p = -0.05, P = 0.90). In addition, the
mean absolute difference in chronic postoperative rate of
cell loss between the two eyes of each subject (4.3% per
year) was greater than the standard deviation of chronic
cell loss in all eyes (3.5% per year, Table 7). These findings
indicate that the two eyes of each subject are reasonably
independent with respect to chronic postoperative cell loss
and justify the use of the ten bilateral cases in this analysis.
Only one of the ten patients with bilateral vision had cor-
nea guttata; both of these eyes had 2+ cornea guttata.
We found no published studies of endothelial cell loss
10 years after cataract extraction for comparison. Five
investigations reported results of sequential endothelial
examinations before and 5 years after lens implantation.
We excluded studies of anterior chamber implants be-
cause there were none in our analysis. Liesegang et all
reported a trial that included the eyes in the current study,
and found a similar cell loss from 1 to 5 years postop-
eratively. Martin et al 9 analyzed 28 eyes 5 years after
intracapsular cataract extraction with or without Bink-
horst four-loop lens implantation, finding progressive cell
losses that appeared to be in the same range as, or some-
what higher than, our control eyes and eyes with intra-
capsular extractions and medallion iris suture implants.
Kora et al 2 presented the results of sequential endothelial
examinations on seven children; the average decrease in
cell density from 1 to 5 years after lens implantation
appeared to be somewhat greater than the 2.5% per year
that we found. Werblin lO reported endothelial exami-
nations 5 years after lens implantation in 34 eyes; the
Rate of Cell Loss
(% per year)
No. Mean ± SD Median P
15 3.8 ± 3.9 2.7 0.52"
49 2.8 ± 3.4 2.4
15 2.6 ± 4.3 1.3 0.60t
27 2.8 ± 3.1 2.4
7 2.9 ± 2.4 3.4
 Bourne et al . Endothelial Cell Loss 10 Years after Lens Implant
pooled results and variable fonow-up did not allow an
estimation of the chronic cell loss beyond I year, al-
though the mean cell loss at 5 years was less than that
in our patients. Numa et ails reported data on 15 eyes
with posterior chamber lenses that had an exponential
rate of cell loss from 1 to 5 years postoperatively of ap-
proximately 1.1 % per year.
The 2.5% chronic endothelial cell loss rate that we
found is much higher than that in healthy unoperated
eyes. In cross-sectional studies, the average annual cell
loss rate in healthy eyes appears to be approximately 0.3%
to 0.5%.16.17 In longitudinal studies, however, the annual
loss has been somewhat higher, with values of 1%,18
0.8%,10 0.6%,19 and 0.3%15 per year being reported. We
can estimate from these studies that the 2.5% annual en-
dothelial cell loss that we found in eyes 10 years after
surgery for cataract is 2.5 to 8.0 times the loss in unop-
erated eyes.
In an article often cited as evidence of the value of
preoperative endothelial morphologic analysis, Rao et al 20
reported the results in 118 eyes of 102 patients with iris
suture medallion lenses implanted at least 5 years previ-
ously. They divided the group into 40 eyes in which cor-
neal edema developed within 5 years (12 eyes required
penetrating keratoplasty for diffuse edema and 28 had an
unspecified amount of peripheral corneal edema but clear
central corneas with good vision) and 78 eyes that did
not. The group with edema had a significantly higher
mean preoperative coefficient of variation of cell area
(polymegethism). There was no difference between the
groups in endothelial cell loss, although the time between
lens implantation and endothelial photography for deter-
mination of cell loss varied for each eye and was not re-
ported. Although our study and that of Rao et al are not
directly comparable, our results do not seem to confirm
theirs. We did not notice clinically important peripheral
corneal edema in our patients. There was no significant
correlation between the preoperative coefficient of vari-
ation of cell area and lO-year cell loss, lO-year corneal
thickness, or chronic endothelial cell loss. Bates and
Cheng21 found no evidence of a relation of any preoper-
ative or early postoperative endothelial morphologic
characteristic to subsequent endothelial decompensation.
Numa et ailS also found no significant correlation between
preoperative coefficient of variation of cell area and sub-
sequent cell loss.
There are a number of possible causes for the elevated
chronic endothelial cell loss in eyes long after cataract
extraction. The presence of an artificial lens implant in
the eye is an obvious candidate. In the only randomized
trial oflens implantation, the Oxford Cataract Treatment
and Evaluation Team found a higher rate of cell loss in
eyes with implants than in those without in the first 4
postoperative years.22 Their control group had intracap-
sular extractions, however, whereas 15 of our 17 controls
had extracapsular extractions. The similar rates of cell
loss in the control and implanted eyes in our study make
it unlikely that lens implantation is responsible. Chronic
inflammation is another possible cause, although there is
not good evidence to support it. 23 Other possible causes
are exposure to the components of vitreous humor, which
may be detrimental to corneal endothelial cells,24 and
changes in the aqueous humor, resulting from the lack of
a crystalline lens in the eye, which may limit endothelial
cell nutrition.
What are the implications of continued endothelial cell
loss in eyes after cataract extraction? If the typical eye
loses endothelial cells at the rate of 2.5% per year after
the first year after cataract extraction when it has a cell
density of 2200 cells/mm2 (Table 5), then it will require
60 years for the cell density to decrease to 500 cells/mm2,
the level at which corneal decompensation was found to
be imminent by Bates et al. 25 This reassuring finding but-
tresses the arguments for lens implantation in young eyes
with healthy corneas. 2 If the operative cell loss is much
higher than average, however, the postoperative cell den-
sity will reach 500 cells/mm2 sooner because of the sig-
nificant correlation between operative cell loss (measured
at 2 months) and chronic postoperative cell loss rate. Also,
if the operative eye has cornea guttata, a cell density of
500 cells/mm2 will be reached much earlier because of
the increased rate of chronic cell loss with cornea guttata.
Finally, we found no preoperative endothelial factor
that was associated with the postoperative corneal status
or outcome except cornea guttata, which is most easily
observed on slit-lamp examination. These findings indi-
cate that preoperative specular microscopy is not a nec-
essary procedure for routine cataract surgery. This is con-
sistent with recently published guidelines. 26 ,27
References
1. Liesegang TJ, Bourne WM, Ilstrup DM. Prospective 5-year
postoperative study of cataract extraction and lens implan-
tation. Trans Am Ophthalmol Soc 1989;87:57-78.
2. Kora Y, Inatomi M, Fukado Y, et al. Long-term study of
children with implanted intraocular lenses. J Cataract Re-
fract Surg 1992; 18:485-8.
3. Liesegang TJ, Bourne WM, I1strup DM. Short- and long-
term endothelial cell loss associated with cataract extraction
and intraocular lens implantation. Am J Ophthalmol
1984;97:32-9.
4. Bourne WM, Waller RR, Liesegang TJ, Brubaker RF. Cor-
neal trauma in intracapsular and extracapsular cataract ex-
traction with lens implantation. Arch Ophthalmol 1981 ;99:
1375-6.
5. Bourne WM, Brubaker RF, O'Fallon WM. Use of air to
decrease endothelial cell loss during intraocular lens im-
plantation. Arch Ophthalmol 1979;97: 1473-5.
6. Bourne WM. Morphologic and functional evaluation ofthe
endothelium of transplanted human corneas. Trans Am
Ophthalmol Soc 1983;81 :403-50.
7. Galin MA, Lin LL, Fetherolf E, et al. Time analysis of cor-
neal endothelial cell density after cataract extraction. Am J
Ophthalmol 1979;88:93-6.
8. Inaba M, Matsuda M, Shiozaki Y, Kosaki H. Regional
specular microscopy of endothelial cell loss after intracap-
sular cataract extraction: a preliminary report. Acta
Ophthalmol 1985;63:232-5.
9. Martin NF, Stark WJ, Maumenee AE. Continuing corneal
endothelial loss in intracapsular surgery with and without
1021
 Ophthalmology Volume 101, Number 6, June 1994
Binkhorst four-loop lenses: a long-term specular microscopy
study. Ophthalmic Surg 1987;18:867-72.
10. Werblin TP. Long-term endothelial cell loss following
phacoemulsification: model for evaluating endothelial
damage after intraocular surgery. Refract Corneal Surg
1993;9:29-35.
11. O'Brien PC, Shampo MA. Statistical considerations for per-
forming multiple tests in a single experiment. 4. Performing
multiple statistical tests on the same data. Mayo Clin Proc
1988;63: 1043-5.
12. Ederer F. Shall we count numbers of eyes or numbers of
subjects [editorial]? Arch Ophthalmol 1973;89:1-2.
13. Rosner B. Statistical methods in ophthalmology: an ad-
justment for the intraclass correlation between eyes. Bio-
metrics 1982;38: 105-14.
14. Ray WA, O'Day OM. Statistical analysis of multi-eye data
in ophthalmic research. Invest Ophthalmol Vis Sci 1985;26:
1186-8.
15. Numa A, Nakamura J, Takashima M, Kani K. Long-term
corneal endothelial changes after intraocular lens implan-
tation. Anterior vs posterior chamber lenses. Jpn J Ophthal-
mol 1993;37:78-87.
16. Yee RW, Matsuda M, Schultz RO, Edelhauser HF. Changes
in the normal corneal endothelial cellular pattern as a func-
tion of age. Curr Eye Res 1985;4:671-8.
17. Carlson KH, Bourne WM, McLaren JW, Brubaker RF.
Variations in human corneal endothelial cell morphology
and permeability to fluorescein with age. Exp Eye Res
1988;47 :27 -41.
18. Cheng H, Jacobs PM, McPherson K, Noble MJ. Precision
of cell density estimates and endothelial cell loss with age.
Arch Ophthalmol 1985;103:1478-81.
Discussion
by
Alan Sugar, MD
Long-term clinical research is a difficult task. Long-term research
on the follow-up of surgical procedures is especially so. While
we can obtain good data on past techniques with difficulty, the
rapidly changing nature of surgical practice often makes it im-
possible to obtain late outcome data on the specific techniques
we currently use. Long-term data are helpful to the extent that
they examine techniques still current, or when they define gen-
erally applicable principles. This study, though subject to the
limitations of such work, provides both.
In the past, Dr. Bourne and his colleagues have shown that
corneal endothelial cell loss at 2 years was less after extracapsular
than intracapsular cataract extraction and that extracapsular
cataract extraction with a posterior chamber intraocular lens
(IOL) caused more cell loss than without an IOL. l Their 1984
study raised concern that the presence of an IOL, even a bag-
placed posterior chamber IOL, contributed to continuing en-
dothelial damage. A 5-year follow-up examination showed con-
tinuing cell loss in all groups, but there were no significant dif-
ferences between the groups at 5 years. 2
Bourne et al have selected their own patients from the pre-
vious prospectively assigned groups for a IO-year analysis. The
authors followed 253 eyes that had had cataract extraction from
1976 to 1982. Ofthese patients, the 76 who did not receive IOLs
were considered to be "controls," although they were selected
From the W. K. Kellogg Eye Center, University of Michigan, Ann Arbor.
1022
19. Ambrose VMG, Walters RF, Batterbury M, et al. Long-
term endothelial cell loss and breakdown of the blood-
aqueous barrier in cataract surgery. J Cataract Refract Surg
1991 ;17:622-7.
20. Rao GN, Aquavella JV, Goldberg SH, Berk SL. Pseudo-
phakic bullous keratopathy. Relationship to preoperative
corneal endothelial status. Ophthalmology 1984;91: 1135-
40.
21. Bates AK, Cheng H. Bullous keratopathy: a study of en-
dothelial cell morphology in patients undergoing cataract
surgery. Br J Ophthalmol 1988;72:409-12.
22. Oxford Cataract Treatment and Evaluation Team (OCTET).
Long-term corneal endothelial cell loss after cataract surgery.
Results of a randomized controlled trial. Arch Ophthalmol
1986;104:1170-5.
23. Tuberville A W, Wood TO. Aqueous humor protein and
complement in pseudophakic eyes. Cornea 1990;9:249-
53.
24. Fischbarg J, Stuart J. The effect of vitreous humor on fluid
transport by rabbit corneal endothelium. Invest Ophthalmol
1975;14:497-506.
25. Bates AK, Cheng H, Hiorns RW. Pseudophakic bullous
keratopathy: relationship with endothelial cell density and
use of a predictive cell loss model. A preliminary report.
Curr Eye Res 1986;5:363-6.
26. American Academy of Ophthalmology. Ophthalmic pro-
cedures assessment. Corneal endothelial photography.
Ophthalmology 1991 ;98: 1464-8.
27. Cataract Management Guideline Panel. Management of
Functional Impairment Due to Cataract in Adults. Guide-
line: Preoperative ophthalmic testing in patients with cat-
aract. Ophthalmology 1993; 100(Suppl):33S-4.
to not receive an IOL because of young age, cornea guttata, or
diabetes. By the IO-year follow-up visit, only 67 eyes (26.5%) in
57 patients were available for endothelial examination. Seventeen
were control eyes, and half of these wore contact lenses. Only
seven patients had posterior chamber IOLs. The high rate of
loss to follow-up from death or failure to return for re-exami-
nation is to be expected in patients with a mean age of almost
70 years at entry. But the small number of remaining patients
limits the power of subsequent analyses to detect small or even
moderate differences.
These authors, however, are statistically sophisticated. They
examined sources of bias between groups and between followed
and lost patients carefully and have tried to maximize the value
of the information obtained from the remaining patients. The
key point from the data is that after an initial early loss of en-
dothelial cells there was a continued cell loss at a compounded
or exponential rate of 2.5% per year. This is greater than the
0.3% to 0.8% cell loss expected in unoperated eyes. This rate
did not differ significantly between implanted and control eyes
or between IOL types within the limits ofthe small sample sizes.
The rate of continued long-term cell loss correlated to a small
to moderate degree (r = 0.32) with the cell loss at 2 months.
Because there is no longer a question of the value of pseu-
dophakia over aphakia, and the clinical success of posterior
chamber IOLs, what is the value of this information? First, it
suggests a finite "lifespan" of the endothelium in eyes with IOLs,
which on average is much longer than that necessary to maintain
 Bourne et al . Endothelial Cell Loss 10 Years after Lens Implant
the cornea during the life expectancy of most patients. Second,
it suggests that minimizing the initial cell loss will prolong that
"lifespan," a valuable gain in vulnerable eyes such as those with
cornea guttata or in those at the low end of a population distri-
bution of cell density. In fact, the ability of clinical specular
microscopy to detect early cell loss and direct and evaluate sur-
gical change to protect the endothelium was the initial con-
tribution of the work of Bourne and Kaufman 3 in the
1970s. The current cell loss seen after posterior chamber
phacoemulsification 4 and the use ofviscoelastics5 would be ex-
pected to raise the baseline on which exponential cell loss acts
and further protect the cornea.
If cell loss continues after cataract extraction and IOL inser-
tion, this would be expected to result from an effect of crystalline
lens absence, an effect of the presence of an IOL, or an accelerated
effect of aging due to the initial cell loss. Mechanical, inflam-
matory, and toxic mechanisms associated with an IOL or altered
fluid dynamics with or without an IOL, may playa role. The
lack of an increase of cell loss in pseudophakic over aphakic
eyes is at least reassuring.
The last point that Bourne et al makes is one with which I
strongly agree. The use of specular microscopy in patients having
cataract surgery is a valuable and constructive research tool, but
its use is limited in routine clinical cataract practice.
References
1. Liesegang TJ, Bourne WM, Ilstrup DM. Short- and long-
term endothelial cell loss associated with cataract extraction
and intraocular lens implantation. Am J Ophthalmol 1984;
97:32-9.
2. Liesegang TJ, Bourne WM, Ilstrup DM. Prospective 5-year
postoperative study of cataract extraction and lens implan-
tation. Trans Am Ophthalmol Soc 1989; 87:57-78.
3. Bourne WM, Kaufman HE. Endothelial damage associated
with intraocular lenses. Am J Ophthalmol1976; 81:482-5.
4. Werblin TP. Long-term endothelial cell loss following
phacoemulsification: model for evaluating endothelial
damage after intraocular surgery. Refract Corneal Surg 1993;
9:29-35.
5. Koch DD, Liu JF, Glasser DB, et al. A comparison of corneal
endothelial changes after use of Healon or Viscoat during
phacoemulsification. Am J Ophthalmol 1993; 115:188-201.
1023