Cornea
Corneal Endothelial Toxicity of Air and SF6
Hubert Landry,1,2 Anahid Aminian,1,2 Louis Hoffart,1,2,3 Ossama Nada,1 Thouria Bensaoula,1
Stéphanie Proulx,4 Patrick Carrier,4 Lucie Germain,4 and Isabelle Brunette1,2
PURPOSE. The authors conducted in vivo assessment of corneal
endothelial toxicity of air and SF6 in the feline model. This
research was motivated by the increased use of air in anterior
segment surgery in human subjects.
METHODS. This was a prospective masked study. The eyes of 16
healthy adult cats were randomly assigned for the injection of
0.7 mL air into the anterior chamber of one eye and SF6 in the
contralateral eye. Daily examination included slit lamp photo-
graphs, pachymetry, and tonometry. Specular microscopy was
performed before, 7 days after, and 10 days after injection. The
animals were euthanatized, and the corneas were processed
for alizarin red-trypan blue staining and for light and electron
microscopy.
RESULTS. SF6 remained in the anterior chamber significantly
longer than air. Both groups showed postinjection inflamma-
tion, which on average was maximal at day 2 and more severe
with SF6. No difference in IOP was observed between the two
groups. Specular microscopy showed significant endothelial
cell loss in the SF6 group (mean postinjection cell loss, 132 ⫾
50 cells/mm2) but not in the group injected with air. Alizarin
red staining revealed significant regional differences in cell
density only in the SF6 group and more pronounced endothe-
lial cell loss in the superior area.
CONCLUSIONS. These results indicate that both air and SF6 in-
jected into the anterior chamber of the eye can induce intra-
ocular reaction in the feline model and that SF6 is more toxic
than air in terms of endothelial cell loss and anterior chamber
inflammation. (Invest Ophthalmol Vis Sci. 2011;52:2279 –2286)
DOI:10.1167/iovs.10-6187
1
From the Maisonneuve-Rosemont Hospital Research Center,
Montreal, QC, Canada; 2Department of Ophthalmology, University
of Montreal, Montreal, QC, Canada; 3Département d’ophtalmologie,
Université de la Méditerranée, Marseille, France; and 4Laboratoire
d’Organogénèse Expérimentale, Centre de recherche FRSQ du Cen-
tre hospitalier affilié universitaire de Québec, and Départements de
chirurgie et d’ophtalmologie, Université Laval, Québec, QC, Can-
ada.
Supported by the Fonds de la recherche en ophtalmologie de
l’Université de Montréal (FROUM), the FRSQ Research in Vision Net-
work (IB, AA), and the Canadian Institutes of Health Research (CIHR
IB, LG). LH held a postdoctoral scholarship from the Société Française
d’Ophtalmologie, Paris, France. ON held a postdoctoral scholarship
from the Egyptian Ministry of Higher Education, Egypt. IB holds the
Charles-Albert Poissant Research Chair in Corneal Transplantation,
University of Montreal, Montreal, Canada. LG holds the CIHR Canadian
Research Chair in Stem Cells and Tissue Engineering.
Submitted for publication July 10, 2010; revised August 28 and
October 11, 2010; accepted October 14, 2010.
Disclosure: H. Landry, None; A. Aminian, None; L. Hoffart,
None; O. Nada, None; T. Bensaoula, None; S. Proulx, None; P.
Carrier, None; L. Germain, None; I. Brunette, None
Corresponding author: Isabelle Brunette, Department of Ophthalmol-
ogy, Maisonneuve-Rosemont Hospital, 5415, boulevard de L’Assomption,
Montreal, QC, H1T 2M4 Canada; i.brunett@videotron.ca.
Investigative Ophthalmology & Visual Science, April 2011, Vol. 52, No. 5
Copyright 2011 The Association for Research in Vision and Ophthalmology, Inc.
Downloaded from iovs.arvojournals.org on 10/10/2023
I njection of air into the anterior chamber of the eye has
become standard practice in anterior segment surgery.
The air bubble is used to exert direct pressure on an intra-
ocular structure. Because air is less dense than aqueous
humor, the bubble remains upward. The patient is asked to
maintain the eye position that will optimize contact be-
tween the air bubble and the tissue layer to be stabilized. In
Descemet stripping automated endothelial keratoplasty
(DSAEK),1 air is used to promote donor tissue adherence to
the recipient bed, thus eliminating the need for sutures. The
graft is centered over the posterior surface of the cornea,
and the patient is asked to lie in a supine position and to
look straight at the ceiling for 1 hour after surgery to dis-
tribute pressure from the air bubble over the graft surface.
Similarly, the standard treatment for graft dislocation after
DSAEK consists in air reinjection with head positioning.2 Air
can also be used to tamponade a detached Descemet’s mem-
brane after trauma or complicated cataract surgery.3 For
superior detachment, the patient is asked to keep his head in
the upright position. Air has also been used recently to
shorten the period of corneal edema when hydrops devel-
ops in keratoconus eyes.4 Sulfur hexafluoride (SF6) has been
proposed as an alternative to air because it remains in the
anterior chamber for a longer period.5,6 Although now rou-
tinely used in several anterior segment surgeries, the safety
of air and SF6 for the corneal endothelium still must be
demonstrated. The purpose of this study was to compare the
in vivo corneal endothelial toxicity of air and SF6 in the
feline model.
METHODS
All experiments were conducted in accordance with the ARVO
Statement for the Use of Animals in Ophthalmic and Vision Research
and with the tenets of the Declaration of Helsinki. This research
protocol was approved by the Maisonneuve-Rosemont Hospital
Committee for Animal Protection (Montreal, QC, Canada).
Population
Sixteen healthy adult cats (mean weight, 5.06 ⫾ 0.44 kg; range,
2.2– 8.6 kg) were obtained from a certified supplier. One eye of each
animal received an intracameral injection of 20% (nonexpansive mix)7
sulfur hexafluoride (SF6) (Labtician Ophthalmics Inc., Oakville, ON,
Canada), and the other eye received a similar volume of filtered air. The
choice of the eye was randomized, and all observers were masked for
all steps (injection, daily examination, and endothelial cell morphomet-
ric analysis).
Surgical Protocol and Medication
Surgery was performed under general anesthesia. Each animal was
premedicated with an intramuscular injection of premixed
acepromazine (0.05 mg/kg; Sanofi Aventis, Laval, QC, Canada),
glycopyrrolate (0.01 mg/kg; Axcan Pharma, Mont Saint Hilaire, QC,
Canada), and buprenorphine (0.01 mg/kg; Schering-Plough, Kirk-
land, QC, Canada) and was intubated. Anesthesia was induced and
2279
 2280 Landry et al.
maintained by inhalation of isoflurane 2% (Baxter, Mississauga, ON,
Canada). Atracurium (0.25 mg/kg, followed by 0.1 mg/kg every 20
to 30 minutes as needed; Sandoz, Boucherville, QC, Canada) was
used to induce paralysis of the extraocular muscle. Full pupil dila-
tion was obtained with topical administration of tropicamide 1%
(Alcon, Mississauga, ON, Canada), phenylephrine 2.5% (Alcon), and
cyclopentolate 1% (Alcon). Two 30-gauge needles were introduced
through the limbus into the anterior chamber, and 0.7 mL air or SF6
was injected while the aqueous was tapped from the anterior
chamber by passive filling of the second needle. The same proce-
dure was performed in both eyes; the only difference was the nature
of the injected gas. Wounds were checked for leaks. Betamethasone
acetate and phosphate (3 mg in 0.5 mL; Sandoz, Boucherville, QC,
Canada), tobramycin (10 mg in 0.25 mL; Alcon), and cefazolin (50
mg in 0.25 mL; Novopharm, Toronto, ON, Canada) were injected in
the inferior fornix, and 1 drop of atropine 1% (Alcon) was instilled.
Prednisone (5 mg/d; Apotex, Weston, ON, Canada) was given orally
on a daily basis.
Postinjection Follow-up
All animals were examined on a daily basis for 10 days after injec-
tion. Slit lamp assessment (Haag-Streit, Bern, Switzerland) was al-
lowed to document the decrease in air and SF6 bubble size and the
signs of inflammation. The size of the bubble in the anterior cham-
ber was estimated by measuring the ratio between the diameter of
the bubble and that of the corneal diameter in the vertical meridian.
Aqueous flare and cell responses were graded according to the
criteria of Schlaegel et al.8 Intraocular pressure (IOP; Tonovet,
TV01; Tiolat Oy, Helsinki, Finland) and central corneal thickness
(CCT; Ultrasound Pachymeter SP 3000; Tomey, Nagoya, Japan) were
measured on a daily basis. Superior corneal thickness (SCT) and
inferior corneal thickness (ICT) measurements and central noncon-
tact specular microscopy (Cellchek XL specular microscope; Konan
Medical USA, Torrance, CA) were performed before surgery and 7
and 10 days after the injection. Specular microscopy photographs
were taken in triplicate. Eleven animals were euthanatized (pento-
barbital sodium 2 mL/4.5 kg IV; Sandoz) on day 10, three animals
were euthanatized at 3 weeks and the two animals were euthana-
tized at 8 weeks. Both eyes were enucleated and examined.
Tissue Preparation
Corneoscleral buttons were dissected and cut in three. A thin,
vertical, central, 3-mm-wide band was fixed in glutaraldehyde 2.5%
for transmission electron microscopy, and the nasal cornea was
fixed in 10% formaldehyde for scanning electron microscopy. Spec-
imens for electron microscopy were then processed as described in
Proulx et al.9
The temporal cornea was used for vital staining. Two radial, 4-mm,
full-thickness, peripheral relaxing incisions were made to allow flat
mounting of the specimen. The endothelium was stained with trypan
blue 0.25% and alizarin red S 0.2% (Sigma-Aldrich, Oakville, ON, Can-
ada).10 Three standardized photographs were obtained from each of
the superior, central, and inferior endothelial areas (objective plan apo;
1.5⫻; SteREO Discovery V12; Carl Zeiss Canada, Toronto, ON, Can-
ada).
Endothelial Cell Morphometric Analysis
Endothelial cell densities and morphometric analyses of specular
microscopy and stereo microscopy images were performed (KSS-
409SP software version 2.10; Center Method) available on the Ko-
nan (Irvine, CA) noncontact specular microscope system. A mini-
mum of 100 cells were counted in each studied area. The studied
parameters included endothelial cell density (cells/mm2) and aver-
age endothelial cell area (␮m2). The coefficient of variation (CV; SD
of cell area/mean cell area) was used as a measure of polymegeth-
ism, and the percentage of hexagonal cells was used as an index of
Downloaded from iovs.arvojournals.org on 10/10/2023
IOVS, April 2011, Vol. 52, No. 5
pleomorphism.11,12 Bilateral endothelial morphometric analyses
were performed.
Analyses of the central corneal endothelium as a function of time
(before and 7 and 10 days after injection) were based on in vivo
specular microscopy. Because only central measurements can be
taken with specular microscopy in the living animal, characteriza-
tion of the geographic distribution of the endothelial cell damage
across the corneal surface (superior, central, and inferior cornea)
was made using vital staining of postmortem tissues. Alizarin red
and trypan blue endothelial stainings were performed on all post-
mortem corneas. Enucleation was performed at 10 days in 11
animals, 3 weeks in 3 animals, and 8 weeks in 2 animals. Enucleation
at day 10 was preferred to detect the early distribution pattern of
endothelial damage before cell migration and uniformization of the
endothelial mosaic.13–15 The decision to postpone enucleation in
five animals was based on the presence of a fibrin membrane
attached to the corneal endothelium, either bilateral (four animals)
or unilateral (in the SF6 eye of one animal). These membranes might
have masked the endothelial mosaic and prevented postmortem
morphometric analysis using a posterior approach. The five pairs of
corneas with a longer follow-up were first compared with the 11
pairs with a 10-day follow-up. Because cell damage distribution was
the same, all corneas were analyzed together.
Statistical Analysis
We studied the effect of two factors (time and type of treatment) on
the evolution of the measured parameters (e.g., bubble size, anterior
chamber cells, flare). The two treatment modalities assessed were air
and SF6. The effect of time was measured from day 0 to day 10 after
injection. ANOVA with two repeated factors was used on full data sets.
In case of an interaction between the type of treatment and time, each
of these two factors was analyzed separately with repeated-measures
ANOVA or paired t-tests while fixing the other factor. Mean and SEM
are reported. Analyses were performed with SPSS (Chicago, IL) soft-
ware, version 15.0, and P ⬍ 0.05 was considered statistically signifi-
cant.
RESULTS
Bubble Size
Bubble size was studied as a function of time after injection and
type of treatment. A significant statistical interaction was found
between time and type of treatment (P ⬍ 0.001), and both
parameters were found to affect bubble size. On the day of
injection (day 0), the bubble size was the same in both groups
(air and SF6), but it decreased significantly more quickly in the
group injected with air (Figs. 1A–F, 2A). Complete clearance of
the bubble in all eyes took 9 days with air (0.50% ⫾ 0.35% of
the anterior chamber volume at day 8 and no air at day 9) and
⬎10 days with SF6 (2.19% ⫾ 1.29% of the anterior chamber
volume at day 10).
Anterior Chamber Inflammation
A significant statistical interaction was found between time
and type of treatment for each of the two parameters of
anterior chamber inflammation (cells and flare; P ⬍ 0.001).
Air and SF6 induced inflammation in both groups (Figs. 2B,
2C). This inflammation was significant from day 1 (P ⬍
0.001 for cells and flare), reached a maximum on day 2, and
decreased progressively. The amounts of cells and flare were
greater and decreased significantly more slowly with SF6
than with air. Inflammation did not subside totally by day 10
(mean difference in cell grade between presurgery and day
10: air, 1.19 ⫾ 0.23, P ⫽ 0.006; SF6, 1.69 ⫾ 0.25, P ⬍ 0.001;
mean difference in flare: air, 1.50 ⫾ 0.30, P ⫽ 0.010; SF6,
2.25 ⫾ 0.35, P ⫽ 0.001). In some cases, inflammation led to
 IOVS, April 2011, Vol. 52, No. 5 Air and SF6 Corneal Endothelial Toxicity 2281

FIGURE 1. Slit lamp and vital staining assessment. (A–F) In vivo slit lamp images of air (first column; right eye) and SF6 (second column; left eye)
bubbles 1, 5, and 10 days (top, center, and bottom rows, respectively) after injection. The air bubble occupied 30% (A), 5% (B), and 0% (C) and
the SF6 bubble occupied 50% (D), 35% (E), and 0% (F) of the anterior chamber 1, 5, and 10 days after injection, respectively. The fibrin
membrane floating at day 5 in the anterior chamber of the eye injected with SF6 (white arrow) had retracted toward the angle superiorly
at day 10. (G–L) Ex vivo stereo microscopy images of the alizarin red and trypan blue-stained corneal endothelium of the same animal 10
days after the injection of air in the right eye (third column) and SF6 in the left eye (fourth column). The superior (first row), central (second
row), and inferior (third row) corneal endothelial areas are shown (G, 2632 cells/mm2; H, 2688 cells/mm2; I, 2558 cells/mm2; J, 1527
cells/mm2; K, 2342 cells/mm2; L, 2717 cells/mm2). Scale bars: 100 ␮m (G–L).

the formation of a loose fibrinous membrane in the anterior
chamber, which tended to wrap the bubble and attach
either to the iris surface or to the cornea.
Intraocular Pressure
No statistical interaction was found between time and type
of treatment. Intraocular pressure was affected by time after
injection (P ⬍ 0.001) but not by type of treatment (air vs.
SF6). In both groups, a temporary and nonsignificant in-
crease in IOP was seen early after injection (mean increase,
11.4 ⫾ 3.2 mm Hg; P ⫽ 0.255; Fig. 2D). IOP then decreased
below preoperative levels (this decrease was significant
from day 3) and remained low until the end of the 10-day
study period.
Corneal Thickness
Analysis of corneal thickness was limited by the high propor-
tion of missing data because ultrasound pachymetry could not
be performed in the presence of a retrocorneal bubble in direct
contact with the cornea. Corneal thickness was measurable in
all eyes before surgery. In eyes injected with air, time after
injection and position on the cornea (SCT, CCT, ICT) had no
significant effect on the ability to measure corneal thickness. In
eyes injected with SF6, the percentage of available measure-
ments was significantly affected by time and position on the
cornea (on day 10, the percentages of available data were still
inferior to normal: SCT, 25%; CCT, 56%; ICT, 94%; SCT, ⬍ICT;
P ⬍ 0.001).
Central Corneal Thickness. As a consequence, analyses
on complete CCT data sets (air and SF6 from day 0 to day 10)
could only be performed on four animals (Fig. 2E). No inter-
action was found between time and type of treatment. CCT
was affected by type of treatment (central corneas were
thicker with SF6 than with air; P ⫽ 0.019) but not by time of
treatment. Corneas exposed to SF6 remained thicker until the
end of the study period.
Inferior and Superior Corneal Thickness. Analyses on
complete ICT data sets (air and SF6 from days 0, 7, and 10)
could be performed on 13 animals. No significant interaction
was found between time and type of treatment. The ICT was
not affected by type of treatment, but it was affected by time
after injection (P ⫽ 0.002). Paired comparisons revealed a
significant postinjection increase in ICT (ICT: preinjection,
684 ⫾ 18 ␮m; day 7, 732 ⫾ 20 ␮m [P ⫽ 0.007]; day 10, 724 ⫾
21 ␮m [P ⫽ 0.046]). Statistical analyses could not be per-
formed on SCT because of the high number of incomplete data
sets.
Specular Microscopy
Endothelial Cell Counts. A significant interaction was
found between time and type of treatment (P ⫽ 0.025). Al-
though cell counts were similar in both groups before injec-
tion, they were significantly lower after injection in the SF6
group (mean air-SF6 difference before injection: ⫺35 ⫾ 37
cells/mm2, P ⫽ 0.357; after injection: 128 ⫾ 41 cells/mm2, P ⫽
0.008). Postinjection values consisted in the mean of day 7 and
day 10 values. A mean cell loss of 132 ⫾ 50 cells/mm2 was
observed in the SF6 group (P ⫽ 0.021; Fig. 3A).
Average Cell Area. Similarly, a significant interaction was
found between time and type of treatment (P ⫽ 0.034). Al-

Downloaded from iovs.arvojournals.org on 10/10/2023

 2282 Landry et al. IOVS, April 2011, Vol. 52, No. 5

AIR
A Bubble size SF6
B Anterior chamber cells
100 4
90
80
3

AC cells grade
% of AC volume

70
60
50 2
40
30
1
20
10
0 0
1 2 3 4 5 6 7 8 9 10 Pre-I 1 2 3 4 5 6 7 8 9 10
Injection
Time (days) Time (days)

C Anterior chamber flare D Intraocular pressure

4 45

40
3
AC flare grade

35

IOP (mm Hg)

30
2
25

1 20

15
0 10
Pre-I 1 2 3 4 5 6 7 8 9 10 Pre-I I 1 2 3 4 5 6 7 8 9 10

Time (days) Time (days)

E Central corneal thickness

780
760
740
720
CCT (µm)

700
680
660
640
620
600
580
Pre-I 1 2 3 4 5 6 7 8 9 10
Time (days)

FIGURE 2. Slit lamp parameters, IOP, and central pachymetry values from day 0 to day 10. (A) Percentage of the anterior chamber occupied by
the bubble. (B) Anterior chamber cells. (C) Anterior chamber flare. (D) Intraocular pressure. (E) Central corneal thickness. Slit lamp measurements
were obtained daily on all eyes (n ⫽ 16 animals with complete data sets for air and SF6 from day 0 to day 10), whereas central pachymetry complete
data sets could be obtained in only four animals (ultrasound pachymetry is not possible in the presence of a retrocorneal bubble). Error bars indicate
SEM.

though the average cell area was similar in both groups before
injection, cells were significantly larger after injection in the
SF6 group (mean air-SF6 difference: preinjection, 4.77 ⫾ 6.12
␮m2, P ⫽ 0.449; postinjection, ⫺24.85 ⫾ 8.72 ␮m2, P ⫽ 0.014;
Fig. 3B). SF6 induced a significant increase in cell area, but air
did not (mean increase: SF6, 25.52 ⫾ 10.01 ␮m2, P ⫽ 0.024;
air, ⫺4.10 ⫾ 5.56 ␮m2, P ⫽ 0.473).
Coefficient of Variation of Cell Area. No interaction
was found between time and type of treatment. The CV
increased after injection in both treatment groups (preinjec-
tion, 24.05 ⫾ 0.68; postinjection, 28.58 ⫾ 1.11; P ⬍ 0.001;
Fig. 3C). The CV was also globally higher in the SF6 group
(air, 25.86 ⫾ 0.67; SF6, 28.03 ⫾ 1.14; P ⫽ 0.007). However,
because this difference was also present before injection, no
conclusion could be drawn regarding the comparative effect
of air and SF6 on the CV.
Hexagonality. Once again, no interaction was found
between time and type of treatment. Hexagonality was af-
fected only by time; both treatments induced a similar de-
crease in the percentage of hexagonal cells (before injec-
tion, 67.01% ⫾ 1.06%; after injection, 61.87% ⫾ 1.43%; P ⫽
0.009; Fig. 3D).
Alizarin Red and Trypan Blue Staining
Morphometric analyses demonstrated that signs of endothelial
cell damage and instability were highly dependent on the
position on the cornea (superior, central, or inferior), with a
greater damage in the superior position. Figures 1G to 1L
illustrate the typical aspect of the endothelium after injection
of air in one eye (third column) and SF6 in the contralateral eye
(fourth column).
Endothelial Cell Density. A significant interaction was
found between position and type of treatment (P ⫽ 0.045).
Although position on the cornea did not affect endothelial
cell density in eyes injected with air, endothelial cell loss
was significantly greater in the superior cornea of eyes
injected with SF6 (mean difference in cell density with SF6:
superior-center ⫽ ⫺615 ⫾ 186 cells/mm2, P ⫽ 0.016; supe-
rior-inferior ⫽ ⫺1224 ⫾ 290 cells/mm2, P ⫽ 0.003; center-
inferior ⫽ ⫺609 ⫾ 189 cells/mm2, P ⫽ 0.018; Fig. 4A). In
the superior position, however, the overall difference be-
tween the two groups only tended to be significant (mean
air-SF6 difference: 619 ⫾ 303 cells/mm2, P ⫽ 0.060).
Average Endothelial Cell Area. No interaction was found
between position and type of treatment, and neither position
nor type of treatment was found to affect the average cell area
(Fig. 4B).
Coefficient of Variation of Cell Area. No interaction was
found between position and type of treatment. A significant
position effect was observed (P ⫽ 0.011), but no differences
were detected between air and SF6 groups (Fig. 4C).

Downloaded from iovs.arvojournals.org on 10/10/2023

 IOVS, April 2011, Vol. 52, No. 5 Air and SF6 Corneal Endothelial Toxicity 2283

A Specular microscopy: Cell density AIR scattered on the posterior surface of the cornea were occasion-
2600 SF6
ally found in eyes exposed to either air or SF6 (Fig. 5E).
2550
Cell density (cells/mm2)

2500
Transmission Electron Microscopy
2450

2400 In corneas exposed to SF6, it was found that the endothelial

2350 cells were thinner and more elongated than in the corneas
2300 exposed to air. Some endothelial cells appeared to be detached
2250 from Descemet’s membrane and showed nonspecific protein
2200
Post-Injection
accumulation in the subendothelial cell space. A definite distur-
Pre-Injection bance in the structure of the mitochondrial cristae was observed
and was a sign of cell stress. The inflammatory membrane seen
B Specular microscopy: Average cell surface
460
over the endothelium included multiple cells, among which
monocytes, immature fibroblasts, and epithelioid histiocytes (ker-
Average cell surface (µm2)

450

440
atic precipitates) could be identified. Some similar but less ad-
vanced and more subtle changes were observed in the endothelial
430
cells of corneas exposed to air. It could be seen that these cells
420
were more viable. An example of the transmission electron mi-
410

400

390 A CD_AIR
Post-Injection Alizarin staining: Cell density CD_SF6
Pre-Injection 3200
3000

Cell density (cells/mm2)

C Specular microscopy: Coefficient of variation 2800
0,35
2600
2400
Coefficient of variation

0,3 2200
2000
1800
0,25
1600
1400
0,2 1200
Superior Central Inferior

0,15
Post-Injection B Alizarin staining: Average cell surface
Pre-Injection 410
Average cell surface (µm2)

390
D Specular microscopy: Percentage of hexogonal cells
80 370
350
70
Hexogonal cell (%)

330
310
60
290
50 270
250
40 Superior Central Inferior

30
Post-Injection C Alizarin staining: Coefficient of variation
Pre-Injection 0,37

FIGURE 3. Preinjection and postinjection endothelial cell morphomet-

Coefficient of variation

0,32
ric analysis of specular microscopy photos (n ⫽ 14 complete data sets).
The mean of day 7 and day 10 values are reported for postinjection 0,27

values. (A) Cell density. (B) Average cell area. (C) Coefficient of
0,22
variation of cell area. (D) Percentage of hexagonal cells. Error bars
indicate SEM.
0,17

0,12
Hexagonality. No interaction was found between position Superior Central Inferior
and type of treatment. The percentage of hexagonality was
highly dependent on corneal position (P ⬍ 0.001). All paired D Alizarin staining: Percentage of hexagonal cells
differences were statistically significant (superior-center ⫽ 80

⫺12.56% ⫾ 3.78%, P ⫽ 0.015; superior-inferior ⫽ ⫺20.99% ⫾

Hexogonal cell (%)

70
5.37%, P ⫽ 0.005; center-inferior ⫽ ⫺8.43% ⫾ 2.81%, P ⫽
0.029). No differences were detected between air and SF6 60

groups (Fig. 4D). 50

Scanning Electron Microscopy 40

Scanning electron microscopy was performed on four repre- 30

sentative pairs of eyes. It confirmed the observations made by Superior Central Inferior

specular microscopy and vital staining, namely a greater endo-
thelial cell damage in eyes exposed to SF6 than in those
exposed to air, and a greater damage in the superior cornea,
consisting of cell membrane disruption, missing cells, and
increased endothelial cell area (Fig. 5A–D). Inflammatory cells
FIGURE 4. Postmortem endothelial cell morphometric analysis after
vital staining. Measurements were taken in three corneal positions
(superior, central, and inferior) (n ⫽ 15 complete data sets). (A) Cell
density. (B) Average endothelial cell area. (C) Coefficient of variation in
cell area. (D) Percentage of hexagonal cells. Error bars indicate SEM.

Downloaded from iovs.arvojournals.org on 10/10/2023

 2284 Landry et al.
croscopy aspect of the endothelium of a pair of corneas exposed
to air and SF6 is shown in Figure 6.
DISCUSSION
Our results demonstrate that in the living feline model, both
air and SF6 injected into the anterior chamber of the eye can
FIGURE 6. Transmission electron microscopy. Pair of corneas of the
same animal exposed to SF6 in one eye (A) and air in the other eye (B).
In the cornea exposed to SF6, a nonspecific material lay in the suben-
dothelial cell space. An inflammatory membrane, in which multiple
cells such as monocytes, immature fibroblasts, and epithelioid histio-
cytes (keratic precipitates) were identified, covered the endothelium.
No inflammatory membrane was found over the endothelial cells of
cornea exposed to air. Scale bar: 2 ␮m (A, B).
Downloaded from iovs.arvojournals.org on 10/10/2023
IOVS, April 2011, Vol. 52, No. 5
FIGURE 5. Scanning electron mi-
croscopy. Two patterns of endothe-
lial damage observed in the superior
cornea of animals exposed to SF6 in
one eye (left) and air in the other eye
(right). Each row illustrates a pair of
mate corneas. Damage was either
similarly minimal (A, B) or signifi-
cantly more important (C, D) in the
eye injected with SF6 than in the
mate eye injected with air. Endothe-
lial cell damage included cell mem-
brane disruption and missing cells.
Patches of inflammatory cells scat-
tered on the posterior corneal sur-
face were occasionally seen (E).
Scale bars: 50 ␮m (A–D); 10 ␮m (E).
induce an intraocular reaction. SF6 was more toxic than air
in terms of anterior chamber inflammation (cell, flare, and
corneal edema) and corneal endothelial cell losses, and the
endothelial damage was significantly greater in the superior
cornea. Air did not significantly affect central endothelial
cell counts, and it had no effect on cell damage distribution.
Both air and SF6 induced pleomorphism and polymegeth-
ism, two signs of reversible endothelial instability, which
herein were primarily marked in the superior cornea.
To the best of our knowledge, no previous report on the
corneal endothelial toxicity of intracameral gases16 –21 com-
pared air and SF6 toxicity in a living animal model with
corneal endothelial wound healing characteristics similar to
those of human subjects. The rabbit corneal endothelium
model, which was used most often,16 –20 is known to differ
significantly from the human model. Rabbit endothelial cells
actively regenerate after wounding, showing extensive cel-
lular division at the margin of the wound, migration, elon-
gation, coalescence, endothelial multilayering, and replace-
ment of old degenerated endothelial cells with new
endothelial cells.13,16,22,23 Endothelial repair in the rabbit is
rapid and occurs in hours or days, depending on the severity
of the damage.13,23 Contrary to the rabbit, endothelium
repair in the human takes place without mitosis.24 –26 Cells
enlarge, spread, and migrate to cover the deficit left by the
loss of neighboring cells, eventually restoring a more stable
pattern but never returning to a normal size, and the cell
count remains low.27 The nonregenerative properties of the
corneal endothelium of the cat were shown to be similar to
those of the human corneal endothelium.13–15,22
 IOVS, April 2011, Vol. 52, No. 5
Several older reports16 –21 using varying methodologies,
mostly descriptive, occasionally quantitative, and usually in
the rabbit model, all suggest some degree of corneal endo-
thelial toxicity (corneal edema, aqueous flare and fibrin, and
endothelial cell loss) for both air and expansive gases. Stud-
ies from Olson et al.21 and Foulks et al.17 report results in
the cat model in accordance with our findings, showing
signs of intraocular toxicity to air and SF6, respectively.
As mentioned by Green et al.18 and as suspected herein,
longevity of the gas in the anterior chamber seems to play a
key role in the severity of its deleterious effect. In the
present study, the endothelial cell damage was located pri-
marily in the superior cornea, where the gas bubble re-
mained for the longest period. This finding parallels results
from Doi et al.28 showing increased retinal toxicity (abnor-
mal glutamate distribution and thinning of the outer plexi-
form layer) of intravitreal injection of air, SF6, and C3F8 in
the superior retina, where gases were in continuous con-
tact.
Endothelial toxicity may be explained by several mecha-
nisms. Interference of the bubble with the aqueous humor
nutrients17,21 would most likely be limited to cases in which
the bubble fills the entire anterior chamber, with smaller mo-
bile bubbles allowing enough exchange between the endothe-
lium and the aqueous humor.
A mechanical interaction, resulting either from surface
tension29 or from direct trauma by the bubble itself, could
be a possible source of endothelial cell damage. Hong et
al.30 used corneas mounted on an artificial anterior chamber
filled at 40% with an air bubble and rotated 180° for 50 times
to simulate the movement of an eye filled with air after
DSAEK. The proportion of viable endothelial cells docu-
mented by trypan blue and alizarin red staining was signifi-
cantly lower after air bubble trauma (79.8% ⫾ 0.04%) than
was the proportion of control fellow corneas (89.9% ⫾
0.02%; P ⫽ 0.03; n ⫽ 12 pairs). In the present study, the
surgical stress was the same for all eyes and cannot be held
responsible for differences between groups.
Endothelial cell injury could also be secondary to inflam-
mation, which in our study was more severe with SF6 than
with air. An inflammatory reaction in cat and rabbit eyes
after the intracameral injection of air, SF6, or C3F8 has been
reported by others.17,19 The inflammatory response ob-
served herein in cat eyes was more marked than that rou-
tinely observed in human eyes, such as after DSAEK surgery.
Changes in the antioxidant system of the aqueous humor
have also been reported after intravitreal injection of SF6, as
measured by the increased activity of catalase and superoxide
dismutase and increased malondialdehyde concentrations.31
These authors concluded that the increased occurrence of
active oxygen species in the aqueous humor leads to insuffi-
ciency of the antioxidant system and intensification of the
peroxidation processes, reflected by increased malondialde-
hyde concentration.
It should be mentioned that the numerous reports on the
beneficial effect of air injected into the anterior chamber to
prevent its collapse during intracapsular or extracapsular cat-
aract extraction32–39 were of no help in the present study. In
these older reports, the considerable mechanical protection
offered by the short-term (few minutes only) filling of the
anterior chamber with air would have masked any eventual
sign of air endothelial toxicity.
In conclusion, both air and SF6 injected into the anterior
chamber of the eye were shown to induce intraocular reac-
tion in the living feline model. SF6 was more toxic than air
in terms of endothelial cell loss and anterior chamber in-
flammation. Based on the results obtained herein, the lack of
endothelial toxicity of air and SF6 in the human subject
Downloaded from iovs.arvojournals.org on 10/10/2023
Air and SF6 Corneal Endothelial Toxicity 2285
should not be considered as granted. We therefore recom-
mend that these gases be used with caution in the anterior
chambers of human subjects. Their use should be at the
minimal dose necessary to obtain the desired tamponade
effect and for no longer than required, and air should be
favored over SF6.
Acknowledgments
The authors thank Michel Asselin, John Douglas Cameron, Michel
Carrier, Miguel Chagnon, Marie-Eve Choronzey, André-François Cou-
ture, Marie-Josée Guyon, Julie Dubeau, Angèle Halley, Serge Rosolen,
Denis Sherknies, and Marian-Ananda Zaharia for their technical assis-
tance and professional advice.
References
1. Gorovoy MS. Descemet-stripping automated endothelial kerato-
plasty. Cornea. 2006;25(8):886 – 889.
2. Price FW Jr, Price MO. Descemet’s stripping with endothelial kera-
toplasty in 200 eyes: early challenges and techniques to enhance
donor adherence. J Cataract Refract Surg. 2006;32(3):411– 418.
3. Mahmood MA, Teichmann KD, Tomey KF, al-Rashed D. Detach-
ment of Descemet’s membrane. J Cataract Refract Surg. 1998;
24(6):827– 833.
4. Miyata K, Tsuji H, Tanabe T, Mimura Y, Amano S, Oshika T.
Intracameral air injection for acute hydrops in keratoconus. Am J
Ophthalmol. 2002;133(6):750 –752.
5. Mearza AA, Qureshi MA, Rostron CK. Experience and 12-month
results of Descemet-stripping endothelial keratoplasty (DSEK) with
a small-incision technique. Cornea. 2007;26(3):279 –283.
6. Zusman NB, Waring GO 3rd, Najarian LV, Wilson LA. Sulfur
hexafluoride gas in the repair of intractable Descemet’s membrane
detachment. Am J Ophthalmol. 1987;104(6):660 – 662.
7. Abrams GW, Edelhauser HF, Aaberg TM, Hamilton LH. Dynamics
of intravitreal sulfur hexafluoride gas. Invest Ophthalmol. 1974;
13(11):863– 868.
8. Schlaegel TF Jr, O’Connor GR. Current aspects of uveitis: general
considerations. Int Ophthalmol Clin. 1977;17(3):1– 42.
9. Proulx S, Audet C, Uwamaliya J, et al. Tissue engineering of feline
corneal endothelium using a devitalized human cornea as carrier.
Tissue Eng Part A. 2009;15(7):1709 –1718.
10. 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(12):815– 819.
11. Bergmanson JP. Histopathological analysis of corneal endothelial
polymegethism. Cornea. 1992;11(2):133–142.
12. McCarey BE, Edelhauser HF, Lynn MJ. Review of corneal endothe-
lial specular microscopy for FDA clinical trials of refractive proce-
dures, surgical devices, and new intraocular drugs and solutions.
Cornea. 2008;27(1):1–16.
13. Van Horn DL, Sendele DD, Seideman S, Buco PJ. Regenerative
capacity of the corneal endothelium in rabbit and cat. Invest
Ophthalmol Vis Sci. 1977;16(7):597– 613.
14. Ling TL, Vannas A, Holden BA. Long-term changes in corneal
endothelial morphology following wounding in the cat. Invest
Ophthalmol Vis Sci. 1988;29(9):1407–1412.
15. Huang PT, Nelson LR, Bourne WM. The morphology and function
of healing cat corneal endothelium. Invest Ophthalmol Vis Sci.
1989;30(8):1794 –1801.
16. Van Horn DL, Edelhauser HF, Aaberg TM, Pederson HJ. In vivo
effects of air and sulfur hexafluoride gas on rabbit corneal
endothelium. Invest Ophthalmol. 1972;11(12):1028 –1036.
17. Foulks GN, de Juan E, Hatchell DL, McAdoo T, Hardin J. The effect
of perfluoropropane on the cornea in rabbits and cats. Arch Oph-
thalmol. 1987;105(2):256 –259.
18. Green K, Cheeks L, Stewart DA, Norman BC. Intraocular gas
effects on corneal endothelial permeability. Lens Eye Toxic Res.
1992;9(2):85–91.
19. Lee DA, Wilson MR, Yoshizumi MO, Hall M. The ocular effects of
gases when injected into the anterior chamber of rabbit eyes. Arch
Ophthalmol. 1991;109(4):571–575.
 2286 Landry et al.
20. Leibowitz HM, Laing RA, Sandstrom M. Corneal endothelium. Arch
Ophthalmol. 1974;92(3):227–230.
21. Olson RJ. Air and the corneal endothelium: an in vivo specular
microscopy study in cats. Arch Ophthalmol. 1980;98(7):1283–
1284.
22. Bahn CF, Glassman RM, MacCallum DK, et al. Postnatal develop-
ment of corneal endothelium. Invest Ophthalmol Vis Sci. 1986;
27(1):44 –51.
23. Fukami H, Laing RA, Tsubota K, Chiba K, Oak SS. Corneal endo-
thelial changes following minor trauma. Invest Ophthalmol Vis
Sci. 1988;29(11):1677–1682.
24. Doughman DJ, Van Horn D, Rodman WP, Byrnes P, Lindstrom RL.
Human corneal endothelial layer repair during organ culture. Arch
Ophthalmol. 1976;94(10):1791–1796.
25. Kaufman HE, Capella JA, Robbins JE. The human corneal endothe-
lium. Am J Ophthalmol. 1966;61(5 pt 1):835– 841.
26. Matsuda M, Suda T, Manabe R. Serial alterations in endothelial cell
shape and pattern after intraocular surgery. Am J Ophthalmol.
1984;98(3):313–319.
27. Waring GO 3rd, Bourne WM, Edelhauser HF, Kenyon KR. The
corneal endothelium: normal and pathologic structure and func-
tion. Ophthalmology. 1982;89(6):531–590.
28. Doi M, Ning M, Semba R, Uji Y, Refojo MF. Histopathologic
abnormalities in rabbit retina after intravitreous injection of ex-
pansive gases and air. Retina. 2000;20(5):506 –513.
29. Kim EK, Cristol SM, Geroski DH, McCarey BE, Edelhauser HF.
Corneal endothelial damage by air bubbles during phacoemulsifi-
cation. Arch Ophthalmol. 1997;115(1):81– 88.
30. Hong A, Caldwell MC, Kuo AN, Afshari NA. Air bubble-associated
endothelial trauma in descemet stripping automated endothelial
keratoplasty. Am J Ophthalmol. 2009;148(2):256 –259.
Downloaded from iovs.arvojournals.org on 10/10/2023
IOVS, April 2011, Vol. 52, No. 5
31. Sztarbala T, Gos R, Kedziora J, Blaszczyk J, Sibinska E, Goralczyk
M. [Changes in the antioxidant system of the aqueous humor,
lens and erythrocytes after sulfur hexafluoride application to
the vitreous of rabbits]. Klin Oczna. 1998;100(2):73–75.
32. Norn MS. Corneal thickness after cataract extraction with air in the
anterior chamber. Acta Ophthalmol (Copenh). 1975;53(5):747–750.
33. Irvine AR, Kratz RP, O’Donnell JJ. Endothelial damage with phaco-
emulsification and intraocular lens implantation. Arch Ophthal-
mol. 1978;96(6):1023–1026.
34. Hirst LW, Snip RC, Stark WJ, Maumenee AE. Quantitative
corneal endothelial evaluation in intraocular lens implanta-
tion and cataract surgery. Am J Ophthalmol. 1977;84(6):775-
780.
35. Binkhorst CD, Nygaard P, Loones LH. Specular microscopy of the
corneal endothelium and lens implant surgery. Am J Ophthalmol.
1978;85(5 pt 1):597– 605.
36. Bourne WM, Brubaker RF, O’Fallon WM. Use of air to decrease
endothelial cell loss during intraocular lens implantation. Arch
Ophthalmol. 1979;97(8):1473–1475.
37. Sugar A, Fetherolf EC, Lin LL, Obstbaum SA, Galin MA. Endothelial
cell loss from intraocular lens insertion. Ophthalmology. 1978;
85(4):394 –399.
38. Forstot SL, Blackwell WL, Jaffe NS, Kaufman HE. The effect of
intraocular lens implantation on the corneal endothelium. Trans
Sect Ophthalmol Am Acad Ophthalmol Otolaryngol. 1977;83(2):
195–203.
39. Cheng H, Sturrock GD, Rubinstein B, Bulpitt CJ. Endothelial cell
loss and corneal thickness after intracapsular extraction and iris
clip lens implantation: a randomised controlled trial (interim
report). Br J Ophthalmol. 1977;61(12):785–790.