RESIDUAL CELLULAR PROLIFERATION

ON THE INTERNAL LIMITING MEMBRANE

IN MACULAR PUCKER SURGERY
ARND GANDORFER, MD, FEBO, CHRISTOS HARITOGLOU, MD, RENATE SCHELER, MTA,
RICARDA SCHUMANN, MD, FEI ZHAO, MS, ANSELM KAMPIK, MD, FEBO

Purpose: To provide pathology data on the completeness of epiretinal membrane (ERM)

removal with and without internal limiting membrane (ILM) peeling.
Methods: Twenty-two patients with idiopathic ERM formation underwent vitrectomy
with ERM removal and subsequent staining of the vitreomacular interface with brilliant blue.
If the ILM was still present after ERM removal, it was peeled off. Both ERM and ILM
specimens were harvested in different containers and prepared for flat-mount phase-
contrast and interference microscopy, immunocytochemistry, and transmission electron
microscopy.
Results: In 14 patients (64%), the ILM was still present at the macula after ERM removal.
On average, 20% (range, 2–51%) of the total cell count was left behind at the ILM if the ERM
was removed only. There were mainly glial cells on the ILM, and few hyalocytes. In nine
eyes, the cells were forming cell clusters. In 8 patients (36%), both ERM and ILM were
removed together. Electron microscopy showed cellular proliferation directly attached to
the ILM in these eyes, whereas in the sequentially peeled group, there was collagen
interposed between the ERM and the ILM. Surgical ERM removal resulted in splitting of the
vitreous cortex in these eyes, leaving the ILM with residual cells behind.
Conclusion: Simple ERM removal results in sufficient separation of fibrocellular tissue in
one third of cases, only. In 2 of 3 patients with idiopathic ERM, the vitreous cortex splits
when the ERM is removed, leaving an average of 20% of the total cell count behind on the
ILM. As these cells are capable of proliferation and causing ERM recurrence, staining of the
ILM with subsequent removal seems beneficial in macular pucker surgery.
RETINA 32:477–485, 2012

E piretinal membrane (ERM) removal has become

a standard technique in vitreoretinal surgery, with
good visual outcome in many cases.1,2 However, visual
recovery is rarely complete, and the recurrence rate of
idiopathic ERM has been reported to be up to 21%.3
Both ERM recurrence and incomplete visual recovery
after surgery are thought to be caused by an
incomplete removal of the ERM. Therefore, additional
peeling of the internal limiting membrane (ILM) of the
retina has been proposed in patients with ERM
formation.3–6
From the Department of Ophthalmology, Ludwig-Maximilians-
University, Munich, Germany.
The authors have no financial interest or conflicts of interest.
Reprint requests: Arnd Gandorfer, MD, FEBO, Department of
Ophthalmology, Ludwig-Maximilians-University, Mathilden-
strasse 8, 80336 Munich, Germany; e-mail: arnd.gandorfer@
med.uni-muenchen.de
Previous ultrastructural investigation of ERM speci-
mens showed that fragments of the ILM were removed
unintendedly in about half of cases.7 Recently, residual
ILM was visualized intraoperatively with Brilliant Blue
G staining.8 In this study, 51% of eyes showed that the
ILM mostly or partially remained after ERM removal,
whereas in 49% the ILM was mostly removed together
with the ERM. Flat-mount immunocytochemistry
revealed remnant cells on residual ILM.8
We wanted to quantify the completeness of epi-
macular cell removal in patients with ERM formation
undergoing vitrectomy. Therefore, we investigated
ERM and ILM specimens removed sequentially from
the macula in 22 consecutive patients with idiopathic
ERM formation. We performed flat-mount prepara-
tion, immunocytochemistry and electron microscopy
to evaluate the degree of ILM removal during ERM
peeling and to clarify the proportion of cells left
behind when the ILM is not additionally approached.

477
478 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES  2012  VOLUME 32  NUMBER 3
Methods
We harvested ERM and ILM specimens in 22
consecutive patients with idiopathic ERM formation
who underwent vitrectomy with membrane removal
during April and July 2010. The study was approved
by the Institutional Review Board and performed in
accordance with the ethical standards of the 1989
Declaration of Helsinki. Only those patients whose
epiretinal specimens were removed from the same
macular area, that is, the ERM was removed first, and
the ILM if still present was removed in a second step
from the macula, were included.
Surgery consisted of a standard 23-gauge pars plana
vitrectomy with induction of posterior vitreous de-
tachment (PVD) and removal of the vitreous gel. The
ERM was grasped with an end-gripping forceps and
removed from the macula. Special attention was paid
to remove the membrane in 1 fragment from the
macular area. Brilliant Blue G (0.2 mL, brilliant peel;
Geuder, Heidelberg, Germany) was injected onto the
vitreomacular interface, immediately washed out, and
the ILM was removed from the macula. The retinal
periphery was checked for breaks, and balanced salt
solution was left in the eye at the end of the procedure.
Both ERM and ILM specimens were harvested in
different containers and prepared for flat-mount pre-
paration, phase-contrast and interference microscopy,
immunocytochemistry, and transmission electron
microscopy. The specimens were placed in parafor-
maldehyde and processed as described previously.9,10
Under a stereomicroscope (Leica MS 5; Leica,
Wetzlar, Germany), the specimens were manipulated
by using glass pipettes and unfolded to show the
maximum area of their surface. Antifading mounting
medium DAPI (4’,6-diamidino-2-phenylindol; Diano-
va AKS-38448; Dianova, Hamburg, Germany) and
a cover slide were added.
We performed phase-contrast microscopy using a
modified microscope (Leica DM 2500; Leica). Phase-
contrast microscopy is a contrast-enhancing optical
technique that can be used to produce high-contrast
images of transparent specimens. It uses an optical
mechanism to translate minute variations in phase into
corresponding changes in amplitude, which can be
visualized as differences in image contrast.
Interference microscopy was additionally applied
to detect changes in surface height, such as cellular
proliferations when compared with bare ILM. A
modified microscope was used, where the entering
light is split into two beams that pass through the
specimen and are recombined in the image plane where
the interference effects make the transparent object
details become visible as intensity differences. Cell
nuclei were stained by 4’,6-diamidino-2-phenylindol
and counted using Image J software (National Institutes
of Health, Bethesda, MD). We used a digital camera
(ProgRes CF; Jenoptik, Jena, Germany) to image the
vitreoretinal interface at magnifications between 350
and 3400. Each specimen was documented by using
a standardized protocol for all imaging techniques.
Statistical analysis was performed by applying the
Mann–Whitney U-test, comparing the cell count and
cell density of ERM and ILM specimens between the
sequential and complete peeling group. After photo-
documentation, the specimens were divided and pre-
pared for transmission electron microscopy and
immunocytochemistry.
For immunocytochemistry, antibodies were used for
glial cells (anti-glial fibrillary acidic protein), hyalo-
cytes (anti-CD45, -CD90, -CD163), retinal pigment
epithelial cells (anti-cellular retinaldehyde-binding
protein [CRALB]), and expression of alpha-smooth
muscle actin, according to protocols described pre-
viously.9,10 Anti-Ki67 was applied to detect pro-
liferating cells.
For transmission electron microscopy, postfixation
was performed in osmium tetroxide 2% (Dalton
fixative), and then dehydration in graded concen-
trations of ethanol and embedding in Epon 812
followed. Semithin sections of 400 nm were stained
with an aqueous mixture of 1% toluidine blue and 2%
sodium borax. Ultrathin sections of 60 nm were
contrasted with uranyl acetate and lead citrate.
Analysis and imaging was performed using a Zeiss
EM 9 S-2 electron microscope (Zeiss, Jena, Germany).
Results
Twenty-two patients (22 eyes) with idiopathic ERM
formation were enrolled in this study. There were 7
women and 15 men. The age at time of surgery ranged
between 48 years and 79 years. Complete PVD was
present in 13 patients. In nine patients, there were
remaining adhesions of the vitreous to the macula
confirmed intraoperatively, resembling the clinical
characteristics of vitreomacular traction syndrome.
All patients underwent uneventful vitrectomy with
ERM and ILM removal without any complication.
The ERM appeared as a sheetlike whitish structure,
which could be easily removed from the macula in 14
eyes (Table 1). In these eyes, additional blue staining
of the vitreomacular interface revealed that the ILM
was only removed in small areas (nine eyes) or was
completely left in place (five eyes). Peeling of the ILM
was performed then, and both the specimens (ERM
 PATHOLOGY OF ILM PEELING IN ERM SURGERY  GANDORFER ET AL 479

Table 1. Morphometric Characteristics of ERM and ILM Specimens

ERM Specimen ILM Specimen
Number of Complete Area in Cell Cells per ILM Area in Cell Cells per Cell
Patients Age/Sex PVD mm2 Number mm2 Fragments mm2 Number mm2 Cluster
1 73/m + 15.06 7.140 47.410 + 11.95 934 7.816 +
2 74/m + 9.52 3.577 37.574 + 11.22 1.683 150 +
3 48/m 2 5.83 6.062 103.979 + 9.48 500 5.274 +
4 72/m + 3.25 653 20.092 + 4.00 186 4.650 2
5 65/m + 1.28 876 69.523 2 3.02 516 17.086 +
6 62/f 2 0.57 332 58.246 2 5.89 176 2.988 2
7 70/m 2 23.21 9.316 40.138 2 16.88 276 1.635 2
8 69/m 2 8.38 2.546 30.382 + 12.15 622 5.119 +
9 74/f 2 3.56 801 225 + 2.79 858 30.753 +
10 70/f + 11.38 3.037 26.687 2 4.31 482 11.183 +
11 69/f + 7.41 1.785 23.992 + 2.72 70 2.574 +
12 70/m + 10.81 4.144 38.335 + 15.11 1.719 11.377 2
13 69/m + 2.50 936 37.440 2 3.74 104 2.781 +
14 75/m + 20.63 7.563 36.660 + 2.14 163 7.617 2
ERM and ILM removed together
15 70/f + 37.45 13.772 36.774 + 2 2 2 2
16 74/f 2 23.82 4.274 17.943 + 2 2 2 2
17 64/m + 2.45 103 4.204 + 2 2 2 2
18 59/m 2 3.99 1.207 30.251 + 2 2 2 2
19 67/f + 6.02 311 5.166 + 2 2 2 2
20 77/m + 11.31 2.987 26.410 + 2 2 2 2
21 78/m 2 7.45 737 9.893 + 2 2 2 2
22 79/m 2 10.49 2.661 25.367 + 2 2 2 2
F, female; m, male.

and ILM) were collected in separate containers. This
was the ‘‘sequential peeling group.’’
In eight eyes, the ERM was more adherent to
the macula. After ERM removal, blue staining of the
vitreomacular interface showed that the ILM had
already been removed from the macular area together
with the ERM. Additional ILM removal was performed
to enlarge the area of ILM removal peripherally to the
area of the former ERM, aiming at one and a half disk
diameter of ILM removal surrounding the fovea. This
was the ‘‘complete peeling group.’’
Sequential Peeling Group
The ERM specimen consisted of a thick membrane
with numerous cells in light microscopy (Figure 1).
Phase-contrast and interference microscopy revealed
a multilayered membrane with massive cellular pro-
liferation. The cell count varied between 332 and 9,316
cells (average, 3,483; standard deviation [SD], 2,955).
The cell density was between 200 cells per mm2 and
1,039 cells per mm2 (average, 424; SD, 224). In nine
specimens, areas of the ILM could be identified in
phase-contrast and interference microscopy. These
ILM fragments, which were unintendedly removed
together with the ERM, were characterized by
a wrinkled fragment of the ILM with clear-cut
boundaries. Internal limiting membrane fragments
were mostly located at the border of the ERM specimen
and covered only a small area of the entire specimen.
The ILM specimen removed sequentially from the
area of the former ERM showed a translucent
membrane with a varying number of cells adherent
to it. The cell count was between 70 and 1,719 cells
(average, 592; SD, 541). The cell density was between
16 cells per mm2 and 307 cells per mm2 (average, 89;
SD, 79). There were significantly less cells on the ILM
than within the ERM specimen. Mostly single cells
were situated on the ILM. However, there were
clusters of cells in nine specimens. The average
percentage of cells left behind on the ILM after ERM
removal was 20% (SD, 15) compared with the total
cell count in both ERM and ILM, with a range
between 2% and 51%.
Complete Peeling Group
In 8 eyes, the ERM was removed together with the
ILM (Figure 2). These specimens were characterized by
a cellular proliferation on the ILM, containing between
103 and 13,772 cells (average, 3,256; SD, 4,491). The
cell density was between 42 cells per mm2 and 367 cells
per mm2. The cellular proliferation covered most parts
of the ILM, varying between 2.45 mm2 and 37.45 mm2
480 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES  2012  VOLUME 32  NUMBER 3

Fig. 1. Sequential peeling.

A, Fundus photograph of
a 48-year-old patient with
ERM formation. B, Epi-
retinal membrane removal
resulted in dissection of the
epimacular cellular mem-
brane on a thick layer of
collagen. C, Flat-mount
preparation of the ERM
shows the ERM as it was in
situ. D, 4’,6-Diamidino-2-
phenylindol staining reveals
.6,000 cells corresponding
to a cell density of 1,040
cells per mm2. E, The ILM
was peeled consecutively.
Flat preparation shows 500
cells that were left behind at
the vitreomacular interface
after ERM removal (8% of
total cell count). F, Inter-
ference microscopy reveals
cellular protrusions and inter-
connections, forming cell clus-
ters. Magnification: (B) 3400;
(C, D, E) 350; (F) 3200.

(average, 195; SD, 121). When the ILM peripherally to
the former specimen was removed, only single cells
were seen adherent to the membrane. There were no
significant differences between the complete and
sequential peeling group in terms of cell count and
cell density.
Transmission Electron Microscopy
In the sequential peeling group, the ERM specimen
was composed of vitreous collagen and a layer of cells
on one side of the collagenous layer (Figure 3). The
cellular layer consisted of fibroblasts, myofibroblasts,
macrophage-like cells, presumably hyalocytes, and
glial cells. Single cells were also distributed within the
collagenous layer. Collagen fibrils were regularly
arranged and measured between 10 nm and 15 nm,
indicating the nature of native vitreous collagen.
Internal limiting membrane specimens removed
sequentially showed the ILM with its typical smooth
vitreal side and the undulating retinal side. Some
retinal debris was present at the retinal side, mainly
fragments of the Müller cell membrane. Retinal debris
was predominantly found within the undulations of the
ILM caused by traction. On the vitreal side, collagen
fibrils and a few cells were seen in direct contact with
the ILM.
In the complete peeling group, there was either
a thin layer of vitreous cortex measuring between
5 mm and 10 mm with cellular proliferation on
top of it or direct contact of proliferating cells to
the ILM without interposed collagen. Ultrastructural
characteristics of cellular proliferation did not vary
from ERM specimens from the sequential peeling
group.
 PATHOLOGY OF ILM PEELING IN ERM SURGERY  GANDORFER ET AL 481

Fig. 2. Complete peeling.

Both ERM and ILM were re-
moved together. A, A 74-year-
old lady with ERM formation
and incomplete PVD. 4’,6-Di-
amidino-2-phenylindol stain-
ing reveals 4,274 cells,
corresponding to a cell density
of 179 cells per mm2. B, Phase-
contrast microscopy shows
marked cellular proliferation.
C, Interference microscopy
shows cellular proliferation on
the ILM. The cells are forming
cell clusters on the ILM.
Magnification: (A) 350; (B,
C) 3200.

Immunocytochemistry the present series, 36% (8 of 22) of eyes showed

simultaneous removal of ERM and ILM, that is, the
The ERM specimens were immunoreactive mainly
ILM came off together with the ERM. In 64% (14 of
for glial cell markers and for hyalocyte markers (Table 2,
22) of eyes, however, the ILM was left in place after
Figure 4). A proportion of cells stained positive for
ERM removal, either completely (23%) or major parts
smooth muscle actin, indicating their contractile
of it (41%). The ILM contained a considerable number
properties. Some cells were positive for CRALB,
of cells that would have been left behind at the macula
a marker of both glial and retinal pigment epithelial
if the ILM was not additionally removed.
cells. A few proliferating cells stained positive with
Ultrastructural and immunocytochemical investiga-
anti-Ki67.
tion of these cells on the ILM showed mainly glial
The ILM specimens removed from the area of the
cells, few hyalocytes, and myofibroblasts. Some of
former ERM were mainly immunoreactive for glial
these cells expressed alpha–smooth muscle actin and
cell markers. Only few cells were immunoreactive for
were capable of exerting tangential traction at the
hyalocyte markers.
vitreomacular interface. In nine specimens, cell
Specimens containing the ERM and ILM (complete
clusters were seen as a sign for cellular proliferation.
peeling group) were immunoreactive for both glial and
On average, 20% of the total cell count was left behind
hyalocyte markers. There was a similar expression of
at the macula, representing a potential source of ERM
CRALB, smooth muscle actin, and Ki67 as it was seen
recurrence.
in ERM specimens from the sequential peeling group.
Internal limiting membrane peeling removes these
cells completely. Another effect of ILM peeling on the
Discussion avoidance of ERM recurrence lies in the fact that the
scaffold for cellular proliferation is removed. The ILM
There is some debate as to whether removing the cannot regenerate, and recurrent cellular proliferation
ILM during macular pucker surgery is beneficial in is only seen in areas where the ILM had not been
terms of visual outcome and avoiding recurrence. In completely removed. There is a third advantage of
482 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES  2012  VOLUME 32  NUMBER 3

Fig. 3. Transmission elec-

tron microscopy of ERM and
ILM specimens. (A) and (B)
are from the sequential
peeling group. A, ERM
specimen shown in Figure 1.
A continuous cellular pro-
liferation is present on a col-
lagenous layer. B, Higher
magnification reveals a mul-
tilayered membrane com-
posed of cells and collagen.
Note that no ILM is present.
(C) and (D) belong to the
complete peeling group. C,
The ILM was removed to-
gether with a thin collage-
nous layer and cellular
proliferation on top of it. A
single cell is present close to
the ILM. Same specimen as
shown in Figure 2. D, A
multilayered cellular mem-
brane in direct contact with
the ILM. Note that there is no
collagen between the cells
and the ILM. Magnification:
(A) 31,000; (B) 31,600; (C)
31,800; (D) 33,000.

ILM peeling in ERM surgery. Cellular proliferation proliferation or a consequence of contraction and
has also been described underneath the ILM, which wrinkling of the ILM caused by the fibrocellular tissue
means on its retinal side, and this is a common electron on the vitreal side of the membrane.
microscopic finding in ILM specimens removed from All three arguments are in favor of ILM removal in
macular pucker eyes.11 It has been speculated before macular pucker surgery, although final conclusions
whether these cells are a possible source of epiretinal regarding the visual benefit cannot be drawn from
a clinicopathologic correlation. In clinical terms, there
are two retrospective case series dealing with ILM
Table 2. Immunocytochemistry of ERM and ILM removal in ERM surgery. In both reports, recurrence
Specimens
was limited to the group without additional ILM
Sequential peeling, and reached 18% and 21%, respectively,
Peeling supporting the concept that the ILM is necessary for
Complete Peeling
Antibody ERM ILM ERM + ILM ERM recurrence.3,4
GFAP + + +
Sebag12–14 proposed the hypothesis of anomalous
CRALB + + + PVD, in which vitreoretinal separation does not occur
CD45 + (2) + cleanly between the vitreous cortex and the ILM, but
CD90 + 2 + rather splits the vitreous cortex, leaving cortical
CD163 + (2) + remnants adherent at the ILM, a phenomenon called
SMA + 2 +
Ki67 + (2) +
‘‘vitreoschisis.’’ Ultrastructural support of this concept
is provided by the present study.
‘‘+’’ indicates positive, ‘‘(2)’’ indicates sparse, and ‘‘2’’ We observed splitting of the vitreous cortex during
indicates negative.
GFAP, glial fibrillary acidic protein; SMA, smooth muscle actin; ERM removal in 64% of eyes (Figure 5). In ultra-
CRALB, cellular retinaldehyde-binding protein. structural terms, the fibrocellular membrane was
 PATHOLOGY OF ILM PEELING IN ERM SURGERY  GANDORFER ET AL
Fig. 4. Immunocytochemistry of ERM and ILM specimens. A, ERM
specimen from the sequential peeling group. Green signal indicates
hyalocytes (anti-CD163). Magenta indicates glial cells expressing in-
termediate-type filament protein (anti–glial fibrillary acidic protein).
Red signal shows glial cells or retinal pigment epithelial cells (anti-
cellular retinaldehyde-binding protein). B, Combined ERM/ILM
specimen. Orange signal indicates anti-smooth muscle actin expression.
Magenta indicates hyalocytes (anti-CD45). Cell nuclei are stained by
4’,6-diamidino-2-phenylindol. Magnification: 3200.
separated from the macula at the level of the vitreous
cortex, which partially remained behind at the ILM. In
36% of eyes, separation of the ERM occurred at the
level of the retinal side of the ILM, obviously caused
by the firm adhesion of the ERM to the ILM. These
specimens showed either direct attachment of cells at
the vitreal side of the ILM without intermingled
collagen or a thin collagenous layer, which might have
been present in some areas only. In other areas of the
ERM, direct interaction of cells and the ILM may have
caused a stronger adhesion than in areas of interposed
collagen, and both ERM and ILM were removed
simultaneously. Our findings support the hypothesis
that the vitreous cortex splits into lamellae during
PVD induction, and the cleavage plane is determined
by the adhesion of the vitreous cortex to the ILM and
by the plane of ERM formation.
483
Fig. 5. Two possible dissection planes during ERM peeling. If the
ERM is located on a layer of vitreous collagen, splitting will occur in
the collagenous layer, leaving the ILM behind. If the ERM directly
grows on the ILM, the ILM will be removed together with the ERM.
Both refinement of surgical techniques using
triamcinolone for better visualization of vitreous
cortex and vital dyes to stain the ILM have brought
insight into the relationship of the vitreous and the
macula in patients with ERM formation.6,15 Doi et al16
and Yamashita et al17 observed a defect in the
posterior vitreous cortex, which developed during
PVD induction. This defect in the vitreous cortex in
patients with ERM has previously been studied by
Hikichi18 and by Kishi and Shimizu19,20 and represents
the liquefied space in front of the macula, which had
been called ‘‘bursa premacularis,’’ ‘‘premacular
vitreous pocket,’’ and ‘‘Martigiani space.’’
Yamashita et al17 have shown that 47% of eyes with
idiopathic ERM formation without preexisting PVD
show a defect in the vitreous cortex after PVD
induction, leaving the ERM on the macula. In 53%,
the vitreous cortex came off completely without
a defect, still leaving the ERM at the macula in 33%.17
In the light of our findings, these studies support
the concept of different cleavage planes during
macular pucker surgery, caused by different types of
ERM formation. If the ERM is located at the
posterior wall of the bursa premacularis and
splitting of the vitreous occurs anteriorly, there will
be a defect in the posterior vitreous cortex after PVD
induction (Figure 6). If splitting occurs posterior to
the ERM, the ERM will come off together with
some vitreous cortex, and no defect is present. If the
ERM directly grows on the ILM, splitting of the
vitreous cortex during PVD induction will occur in
front of the ERM, leaving the proliferation behind at
the macula.
484 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES  2012  VOLUME 32  NUMBER 3
Fig. 6. If an ERM is located at the posterior wall of the bursa pre-
macularis, PVD induction can occur anteriorly or posteriorly to the
ERM. In the first case, a defect will be seen in the posterior hyaloid. In
the second case, the ERM will be removed together with the vitreous
cortex, and no defect is seen.
Recently, Kifuku et al8 have shown that approxi-
mately half of patients undergoing ERM removal have
evidence of residual ILM with remnant glial fibrillary
acidic protein–positive and –negative cells. As others,
we found glial cells, hyalocytes, and myofibroblasts in
our series of idiopathic ERM.21 Noteworthy, both
ERM types (those with and without intermingled
vitreous collagen) contained hyalocytes and glial cells,
and both expressed alpha-smooth muscle actin,
a marker of epithelial–mesenchymal transdifferentia-
tion, and a molecular sign of cellular contraction. In
contrast, ILM specimens removed from the area of
a formerly peeled ERM did sparsely express hyalocyte
markers, but mainly glial cell markers. These findings
support the hypothesis that hyalocytes derived from
the mesenchymal system are located within the
vitreous cortex at some distance away from the
ILM, whereas single cells or cell clusters in direct
contact with the ILM are primarily of glial origin,
formerly named ‘‘simple ERM’’ by Foos.22,23 The
present study also supports the hypothesis discussed
earlier that especially the interaction of hyalocytes and
glial cells (and retinal pigment epithelial cells in case
of retinal breaks) triggers the formation of epiretinal
proliferation, both on a layer of vitreous cortex or
directly on the ILM.7
In essence, we have shown that ERM removal alone
does not completely separate fibrocellular tissue
cleanly from the macula, but leaves cells and collagen
behind at the ILM, and this may be a source of ERM
recurrence or incomplete visual recovery. Additional
ILM peeling is beneficial in terms of complete
removal of epi- and sub-ILM proliferation and in
eliminating the scaffold for proliferating cells. The
cleavage plane during ERM removal is determined by
the morphology of the ERM, in particular by the
presence of residual vitreous cortex. If the ERM
directly grows on the ILM, both membranes will
probably be removed together. If vitreous cortex is
intermingled between the ERM and the ILM, splitting
occurs within the collagenous layer, and the ILM with
cells and collagen left behind should be removed in
addition.
Key words: ILM, peeling, vitrectomy, ERM,
pathology.
References
1. Machemer R. [The surgical removal of epiretinal macular
membranes (macular puckers)] (author’s translation). Klin
Monbl Augenheilkd 1978;173:36–42.
2. Grewing R, Mester U. Results of surgery for epiretinal
membranes and their recurrences. Br J Ophthalmol 1996;80:
323–326.
3. Park DW, et al. Macular pucker removal with and without
internal limiting membrane peeling: pilot study. Ophthalmol-
ogy 2003;110:62–64.
4. Kwok A, Lai TY, Yuen KS. Epiretinal membrane surgery with
or without internal limiting membrane peeling. Clin Experi-
ment Ophthalmol 2005;33:379–385.
5. Sorcinelli R. Surgical management of epiretinal membrane
with indocyanine-green-assisted peeling. Ophthalmologica
2003;217:107–110.
6. Schumann RG, et al. Sequential epiretinal membrane removal
with internal limiting membrane peeling in brilliant blue G-
assisted macular surgery. Br J Ophthalmol 2010;94:1369–1372.
7. Kampik A, et al. Epiretinal and vitreous membranes.
Comparative study of 56 cases. Arch Ophthalmol 1981;99:
1445–1454.
8. Kifuku K, et al. Residual internal limiting membrane in
epiretinal membrane surgery. Br J Ophthalmol 2009;93:
1016–1019.
9. Gandorfer A, et al. Interference microscopy delineates cellular
proliferations on flat mounted internal limiting membrane
specimens. Br J Ophthalmol 2009;93:120–122.
10. Gandorfer A, et al. Pathology of the macular hole rim in flat-
mounted internal limiting membrane specimens. Retina 2009;
29:1097–1105.
11. Haritoglou C, et al. Glial cell proliferation under the internal
limiting membrane in a patient with cellophane maculopathy.
Arch Ophthalmol 2007;125:1301–1302.
12. Sebag J. Anomalous posterior vitreous detachment: a unifying
concept in vitreo-retinal disease. Graefes Arch Clin Exp
Ophthalmol 2004;242:690–698.
13. Sebag J. Vitreoschisis.Graefes Arch Clin Exp Ophthalmol
2008;246:329–332.
14. Sebag J. Vitreous: the resplendent enigma. Br J Ophthalmol
2009;93:989–991.
15. Sonoda KH, et al. Residual vitreous cortex after surgical
posterior vitreous separation visualized by intravitreous
triamcinolone acetonide. Ophthalmology 2004;111:226–230.
16. Doi N, et al. Vitreomacular adhesion and the defect in posterior
vitreous cortex visualized by triamcinolone-assisted vitrec-
tomy. Retina 2005;25:742–745.
 PATHOLOGY OF ILM PEELING IN ERM SURGERY  GANDORFER ET AL
17. Yamashita T, Uemura A, Sakamoto T. Intraoperative characteristics
of the posterior vitreous cortex in patients with epiretinal mem-
brane. Graefes Arch Clin Exp Ophthalmol 2008;246:333–337.
18. Hikichi T, et al. Relationship between premacular cortical
vitreous defects and idiopathic premacular fibrosis. Retina
1995;15:413–416.
19. Kishi S, Shimizu K. Oval defect in detached posterior hyaloid
membrane in idiopathic preretinal macular fibrosis. Am J
Ophthalmol 1994;118:451–456.
485
20. Kishi S, Shimizu K. Posterior precortical vitreous pocket. Arch
Ophthalmol 1990;108:979–982.
21. Kohno RI, et al. Possible contribution of hyalocytes to
idiopathic epiretinal membrane formation and its contraction.
Br J Ophthalmol 2009;93:1020–1026.
22. Foos RY.Vitreoretinal juncture—simple epiretinal membranes.
Graefes Arch Clin Exp Ophthalmol 1974;189:231–250.
23. Foos RY.Vitreoretinal juncture: epiretinal membranes and
vitreous. Invest Ophthalmol Vis Sci 1977;16:416–422.