Surgery for Idiopathic Epiretinal Membrane and Macular Edema
jeudi 7 avril 2011 par Didier ducournau
We argue that the most important factor responsible for the visual symptoms in the disease is not the presence of a membrane (which would block the photons from reaching the photoreceptors) but the
consequent edema. Furthermore, we emphasize the importance of vitreous fibers remaining attached to the ILM ; iERM is not a structure composed of newly secreted collagen that is seeded by non-neuroretinal cells but remnants of cortical vitreous mixed with astrocytic gliosis which is the body’s healing attempt to the injury the retina sustained.
We propose the systematic removal of the ILM in eyes with ME - whether this is present in conjunction with astrocytic gliosis (iERM) or is a stand-alone condition such as in some eyes with diabetes or vascular occlusion. We will explain the “double-edged sword” significance of ILM removal : achieving true completeness of vitreous removal as well as stimulating Muller cell gliosis, which is a more effective repair mechanism for the treatment of, as well as a prophylaxis against, ME.
Finally, we describe our surgical technique in the area of the vitreoretinal interface : the posterior cortical vitreous, the iERM, and the ILM, highlighting the fact that « mini-invasive surgery » is not limited to gauge size and the lack of scleral/conjunctival suturing but also applies to the extent of intraocular manipulations.
Idiopathic Epiretinal Membrane
iERMs represent perhaps the number one indication today for the vitreoretinal surgeon. Nevertheless, there is still a lot of confusion, misinterpretation, and misunderstanding regarding this condition. What exactly is an iERM, how does it develop, why does the vision decrease, is that the primary variable to indicate surgery, how does ILM removal work etc. ?
These are just some of the questions we are trying to answer.
|Confusion regarding the terminology of pathologies at the vitreoretinal interface|
Since the early 1980s, excellent pathological descriptions have been available of “idiopathic epiretinal membranes”. It was reported that the “simple” membrane was composed of “abundant normal vitreous collagen” and glial cells and some hyalocytes  Despite this straightforward finding, confusion later surfaced.(A big part of the confusion is related to the fact that the membrane that is examined by the pathologist is only what the surgeon supplied ; the pathologist is unable to see the entire globe, just as much as the surgeon is unable to see the ultrastructure of the membrane itself. Major improvements can be expected to our understanding of the pathophysiology if we could examine enucleated eyes containing ERMs).
1. “Idiopathic” : “Of unknown cause” (Medterms.com). In the literature, however, ERMs of known causes (retinal detachment, inflammation, diabetes etc.) have been mixed with truly idiopathic membranes ,  This confusion is important because non-idiopathic membranes have a different composition as they contain numerous cell types that do not originate in the neuroretina : macrophages, fibrocytes, myofibroblasts, RPE cells etc , .
2. “Epi” : “On the top of…” (Medterms.com). With this understanding of “epiretinal”, the pathological process occurs upon (above) the retina without involving the retina itself. This notion was reinforced by confusing secondary ERMs (often named "puckers") with truly idiopathic ones (e.g., ERM following RD). The true understanding of the term means : On top of the ILM without its involvement.
3. “Membrane” : “A pliable, sheet-like, usually fibrous tissue that covers, lines, or connects plant and animal organs or cells” (thefreedictionary.com). It reinforces the notion that a “membrane” is a newly-grown, stand-alone entity, independent of the surrounding tissues.
4. The collagen component of the membrane : It is a commonly accepted belief that iERM formation occurs after PVD. In eyes with iERM, PVD rates of 92% have been reported . Consequently, it can be thought that no collagen fiber is left on the retina. In reality, even in eyes with PVD, type II (vitreous) collagen fibers are left behind . This is what our group described first in 1992 using SEM (Fig 1).
5. The macroscopic architecture of the posterior vitreous. Spontaneous and complete separation of the posterior vitreous from the retina (Fig 2a) is much less common than previously thought, even in eyes with high myopia, where it has been described as occurring “early” . Not only is PVD rarely as “absolute” as previously assumed (see above : vitreous fibers left on the retinal surface in eyes with PVD), but what appears pre- and even intraoperatively as PVD is often not true PVD. It is either a split in the posterior cortical vitreous (vitreoschisis) , maintaining a layer of vitreous on the retina (Fig 2b), or a fragmentation of the posterior cortical vitreous, leaving “plaques” of vitreous on the retinal surface (Fig 2c). This frequent configuration of plaques left on the retinal surface explains the very frequent stage B PVR observed in retinal detachment. (video1)
6. The final issue to clarify is whether the removed specimen (iERM, PH) contains fragments of the ILM. Indeed, up to 75% of ERM specimens have been reported to also contain unintentionally ILM . Even surgical PVD can result in the tearing and partial removal of the ILM. This can explain the encouraging visual improvement reported by surgeons who claim not have felt the need to intentionally remove the ILM.
|Clinical description of iERMs|
Idiopathic epiretinal membrane is a disease composed of an epiretinal structure that is recognized by the ophthalmologist by the following signs (table 1) :
- Table 1. The clinical characteristics of different ERMs (The numbers correspond to those in Table 2).
- Change in the retinal reflection. This is a ubiquitous sign. The reflection is not changed in proportion to the functional symptoms.
- ME. The extracellular accumulation of fluid - in the outer layers of the retina - may be impossible to notice without a high-resolution OCT. The amount of fluid is not proportional to the functional symptoms.
Additional, not-always-present signs include :
- Vascular distortion. The contracting membrane alters the normal course of the retinal blood vessels. There is moderate correlation between the distortion and the visual symptoms.
- Foveal ectopia. An asymmetrically positioned membrane can drag the fovea off its normal location. There is high correlation between the distortion and the visual symptoms. This finding signifies for the surgeon that vitreous is present.
- Retinal folds. The contraction of the membrane pulls on the retina and sends radiating folds to a considerably larger area than the membrane itself. The folds can vary from the very superficial to full-thickness. This finding signifies for the surgeon that vitreous is present.
- Macular pseudohole. The membrane with vitreous component is sparing the fovea. There is no correlation between the distortion and the VA.
For the patient, the two leading symptoms are :
- Decrease in VA. While this may partly be explained by the membrane preventing photons from reaching the photoreceptors (reflection/scattering of light), the main reason is the retinal damage caused by the edema. This is proven by the lack of a proportional correlation between membrane thickness/reflectivity and VA, but there is a correlation between the duration of the edema and the VA. Membrane removal does not result in immediate restoration of the VA.
- Metamorphopsia. The retinal surface as a minimum, but often the entire retinal thickness is distorted, which manifests itself in distortion of the image itself. Metamorphopsia is a more important and characteristic symptom than VA, even if it often needs to be specifically asked for to be detected. In fact, metamorphopsia can be more disturbing for the patient than a reduction in VA ; some patients with good vision still close their affected eye to avoid disturbing the good fellow eye. Conversely, many patients whose VA remains unchanged after surgery still consider their surgery beneficial because the metamorphopsia has been eliminated.
|The pathoanatomy of the iERM|
In 1987 we analyzed specimens from 56 eyes with iERM, using type I, II, III, and IV anti-collagen antibodies, anti-vimentine antibody, anti-protein S100 antibody, anti-cytokeratine antibody, and anti-GFAP antibody. GFAP plays a dominant role in astrocytic interactions  ; its expression is essential to the preservation of tissue architecture and to the myelinization and to the integrity of the BBB .
In all the specimens only two kinds of structures were present :
- Collagen (mostly collagen type II, but sometimes type III or IV), depending on the amount of vitreous/PH (Fig 3) .
- ILM lined by GFAP antibody-positive tissue, present in all cases (Fig 4).
The GFAP-positive reaction indicated the presence of gliosis at the ILM level, indicating that the glial cells (in the retina, the astrocytes and the Mûller cells) become filled with gliofibrils of GFAP and/or the number of glial cells is increased.
We found that hyalocytes, blood cells, RPE and inflammatory cells were extremely rare in idiopathic membranes. Collagen and the ILM are acellular structures ; apart from the few cells just mentioned, the glial element appeared to be the only component in idiopathic membranes.
We attempted to make a clinico-pathologic correlation to interpret our findings. It was our attention during vitrectomy to detach the PH first and then the membrane. We encountered two different scenarios.
- 1. The two structures could be separated. (video2)
- The preretinal structure was identified by the pathologist as the normal appearance of the PH (in SEM) ; immunohistochemistry showed the typical collagen type II-positive and GFAPnegative staining (Fig 3).
- The retinal structure, composed of an acellular ILM lined by glial cells (or glial cell fragments), which is GFAP-positive and collagen type II-negative (Fig 4), except for the few remaining fibers as described earlier. Astrocyte arm expansion was obviated in SEM pictures (Fig 5).
- 2. The two structures could not be separated. (video3) In 30% of cases) the two structures were so adherent to one another that they had to be removed together ; these specimens therefore showed both GFAP-positive and collagen type II- positive staining ; no other substantial cellular contingent was present.
- Table 2 : The development of ERM
Considering these observations we concluded that (table 2) :
- a. iERM is not a proliferative process that would originate outside the vitreoretinal interface and would spread over the retinal surface.
- b. The retinal reflection we observe biomicroscopically is produced by a gliosis that develops at the ILM level and is enhanced by the still-adherent PH collagen fibers. The retinal reflection is dependent on the underlying anatomical condition :
- In absence of PVD, including vitreoschisis (numbers 3 & 4 of table2), the collagen fibers are abundant and the reflection is more pronounced, as the number of irregularly arranged fibers is higher, similar to what happens in the cornea. (video4), (video5) The astrocytic proliferation that grows through the ILM spreads beneath the posterior hyaloid. (video6)
- In case of PVD with a residual plaque of vitreous adherent to the macula (number 5 of table 2), the surface of the adherent vitreous shows much more reflection than the adjacent area (Fig 6). The vitreous plaque can be either a unique and uniform plaque, sometime limited by a sharp edge (video7) or, more frequently, a dispersed one, presenting many islets spread out around the major plaque. (video8) These different configurations can be seen whatever the distance of the PH from the retina (video9). The astrocytic proliferation covers either only the PH plaque’s area (Fig 7), (video10) or extends much more towards the temporal arcades. (video11) In this last configuration, the appearance of the extended membrane is similar to the ILM’s (and with the same coloration properties) and it spreads over the ILM. It is this appearance that can mislead the surgeon to believe that the ILM has been removed when in reality it was only the astrocytic proliferation.
- With only a small amount of vitreous fibers on the retinal surface (number 2 of table 2), the reflection is very similar to that of a regular cellophane membrane, and the condition resembles a simple retinal edema. (video12) In fact, the only difference between iERM and
ME can sometimes come down to the presence/absence of reflection in the macular area, (video13), (video14), thus explaining that these two pathologies could be regrouped in the same chapter.
We happened to have the unique opportunity to study a child’s eye, removed for anterior segment congenital abnormalities, but also containing an ERM. In HES standard staining, several cellular elements were found to cover the retinal surface, giving the impression of a membrane originating from outside the neuroretina (Fig 8). However, GFAP staining demonstrated that the proliferation was exclusively composed of glial retinal elements coming from the inner part of the retina (Fig 9).
|The pathophysiology of the astrocytic gliosis|
The causes for Astrocytic gliosis
In neuropathology, astrocytic gliosis is found mostly in the case of ischemia. In conditions of chronic ischemia in mice, transitional limited hypertrophy of fibrous astrocytes localized on the brain surface without neuronal cell death has been observed . Conversely, in the ischemic area, a decrease in GFAP staining of the astrocyte signifies increased permeability of the BBB for macromolecules and then quick cell death. In experimental in-vitro studies, reduction of GFAP in astrocytes’ perivascular endfeet also precedes an increase in BBB permeability .
Regarding the iERM, the same ischemic hypothesis can be applied. This hypothesis is actually reinforced by the following arguments :
- It is our biomicroscopic observation that most eyes with iERM show vascular anomalies (pathological arterio-venous crossings, vascular constriction), compatible with chronic ischemia.
- Some edema always accompanies iERM, as a result of vascular leakage.
- Multifocal ERG recordings in iERM-affected eyes show a delay of P1 implicit time, usually found in retinal ischemic diseases .
However, astrocytic gliosis can also be triggered by other types of retinal noxa, such as vitreous traction, neurodegenerative disorders (Alzheimer’s disease, prion’s disease), AIDS, Parkinson’s disease, or inflammation. The following arguments support the vitreous traction hypothesis :
- iERM rarely occurs in young individuals ; with age the vitreous loses its uniform gel consistency, and the mixture of gel and fluid (syneresis) allows vitreous movement (acceleration/deceleration). It is therefore possible for vitreous traction to exist even in the absence of a PVD.
- Partial PVD or anomalous PVD (vitreoschisis) acts in a similar way : partial vitreous attachment to the retina with coexisting vitreous movement.
We must also consider the possibility of a more complex pathomechanism. One can hypothesize that initialy a non tractional noxa causes astrocytic gliosis. The consequent arms ingrowth that penetrate the PH creates a strong(er) vitreoretinal adhesion, a condition sine qua non for vitreoretinal traction. Similarly, it is possible that the edema is a secondary pathology.
Whatever the cause, the astrocytes proliferate, developing a horizontal gliosis (Fig 11) in the inner retina level (video15) and this seems to protect this level from CME progression (Fig 12).
The causes of loss of function in eyes with iERM
As mentioned before, there are several potential reasons for the visual disturbance (VA drop, metamorphopsia). It is not known precisely which one is responsible for what extent of the functional consequences.
- Light absorption/scattering by the membrane. The membrane is not completely transparent as is obvious to the surgeon by its appearance.
- Metabolic changes due to the presence of the membrane. The thickened ILM, sandwiched between glial proliferation can reduce the molecular transport between retina and vitreous, including oxygen and potassium .
- Wrinkling/folding of the retina (depending on whether it is superficial or full-thickness). This is responsible for the metamorphopsia.
- Edema. Responsible for interruption of axoplasmic flow by synaptic disconnection and/or compression leading to neuronal cell death in absence of timely intervention.(video16)
- Ischemia ( ?). Can be a causative (noxa) or secondary pathology ; vitreous removal has been found to improve retinal blood circulation .
The obvious goals of treatment are then the following :
- Elimination of the membrane’s light-blocking and light-scattering effect.
- Restoration of unhindered metabolic interchange between the vitreous and the retina.
- Recreation of the normal smoothness of the retina, including of its surface.
- Elimination of the edema and prevention of its recurrence.
- Improving retinal oxygenation (this is primarily due to removal of the vitreous).
- Table 3 : Causes of fonction loss and the treatment options
The ideal treatment addresses all five consequences of the pathology.
Medical treatment targets only one of the five problems, the edema. Although intraocular steroid injection is able to rapidly improve the function , the drug has a limited half-life in the eye. Repeated reinjections are thus necessary, which makes its use impreactical. Similarly, continual oral therapy of diuretics is not a viable option because of the medication’s side effects.
Removal of the preretinal part of the structure without the ILM does not address the surface wrinkling, nor the edema. In addition, recurrence of the iERM in up to 21% of eyes has been reported .
Removal of the preretinal part of the structure as well as the ILM, along with the sub-ILM gliosis, is able to deal with all five direct consequences of the problem.
The case for systematic ILM removal in all eyes with iERM
Several arguments can be made in favor of a systematic ILM removal.
- 1) In the absence of enzymatic PVD, ILM peeling is the only effective way to achieve complete removal of the collagen fibers from the retinal surface. As described earlier, even in the case of a true PVD (rather than vitreoschisis), and even in the absence of a posterior hyaloid plaque, a few collagen fibers remain attached to the macula, contributing to the contraction of the proliferative tissue. This means that the first four variables in Table 3 have been addressed.
- 2) ILM removal, although it interferes with a repairing process (astrocytic gliosis), it has a very powerful action against the edema. As detailed later in the chapter on ME, our hypothesis is that the removal of the endfeet of the Müller cells, which occurs during ILM removal, induces a “vertical” (perpendicular to the surface) Müller cell gliosis. As part of the cellular response at the Müller cell level, they become filled with gliofibrils of GFAP. (video17) This results in :
- An anti-apoptotic action that fights against the blood-retinal barrier breakdown.
- A neuronal reconnection thanks to GFAP’s synaptogenetic properties.
- Consolidation of the retinal architecture, due to the homeostasis properties by the Müller cells, reinforced by the gliosis. This is easily demonstrated by the following.
- The ILM removal’s effect on stage I macular holes, whose tendency to progress to fullthickness holes is not reduced by the removal of the PH . Conversely, we have shown that ILM removal arrested this progress .
- Once the ILM cover has been removed from the macular surface, recurrence of the iERM does not occur. In an iERM, however, although the ILM itself never regenerates, astrocytic reproliferation usually develops, as shown by the astrocytic proliferation reappearance of a reflex simulating a cellophane membrane.
In 1998 we conducted a statistical analysis on 1413 eyes that underwent iERM removal - with the intention of ILM removal. After a mean follow-up of 45 months, we found a visual improvement from an average LogMAR line of 0.47 preoperatively to an average LogMAR line of 0.25 postoperatively.
While the 2.2 mean line-improvement does not sound as an exceptional improvement, a more valid conclusion can be made if we analyze the final result based on the level of preoperative visual deterioration (Graph A).
The lower the preoperative VA, the greater the number of lines of improvement that can be achieved. However, the lower the preoperative VA, the less likely that the final outcome will approach full vision (Graph B). In other words, the vision lost as a result of delaying the operation is not compensated by the higher number of lines of improvement.
This is shown in Graph C : the lower the preoperative vision, the lower the likelihood that the patient will achieve reading vision.
|Indications for surgery|
Due to the similarities between iERM and ME, the indications will be discussed together with those for ME, see below.
Surgery for ME, whether it involves astrocytic gliosis or (more commonly) not, should be considered as a first-line treatment option. While the cause and intraretinal location of ERM-related ME are different compared to that caused by other diseases such as diabetes or venous occlusion, the treatment is similar.
However, certain questions need to be answered.
|Does ILM removal work ?|
Currently, there is no proven effective therapy for CRVO-related, and only limited success for BRVOrelated ME (we restricted our analysis to this indication) , , . However, we found the following with our approach with systematic ILM removal in 154 eyes with CRVO and 82 eyes with BRVO.
- Average improvement 4.15 LogMAR lines.
- Improvement increasing with follow-up ; after one year, more than 5 LogMAR lines.
- 70% of eyes with improvement in VA with 6% of eyes showing decreased vision - some of which
due to cataract formation.
- All eyes with at least 3 lines of visual loss had ischemic CRVO.
No treatment modality has been able to achieve so significant and permanent improvement.
|Why are the individual functional results so different ?|
The essence of treatment is not the removal of a pathological process, as in the case of cataract surgery. This means that the results depend on the following conditions :
- The Müller cell’s capacity to fill itself with gliofibrills of GFAP, and then to fight against blood-retinal barrier breakdown, and to repair the interrupted axoplasmic flow. This explains why the visual improvement can take 6 months or more to reach its highest level.
- The persistence of the original noxa. This is why the improvement in diabetic edema is less longlasting : the diabetic microangiopathy is not cured by surgery. Obviously, the post-vein occlusion edema is a one-time event, without a long-term postoperative decrease in vision.
- In edema with large cysts synaptogenesis is not possible.
|How does ILM removal work ?|
This question is impossible to precisely answer. However, in enucleated specimen we have found (see Yvette’s work) that for a given noxa, such as ischemia, there can be three different retinal glial reaction patterns (Table 4).
- Table 5 : gliosis type according to the pathology
Our findings show that without astrocytic and Muller cell gliosis, there is full-thickness retinal edema. In the presence of significant astrocytic gliosis, as mentioned before, the inner retina is dry. Finally, if there is Muller cell gliosis, the entire retina is dry. All eyes with RD showed Muller cell gliosis and no ME ; all eyes with an attached retina showed
no Muller cell gliosis but did have ME. (Table 5)
Our hypothesis is that as a result of RD, the Muller cell is stimulated - its fibrils extend from the ELM all the way to the RPE - as the neuroretina
gets separated from the RPE. (video18) Muller cell is considered radial glia ; if traumatized (such as by a tearing of its fibrils), radial glia reacts by gliosis in order to restore homeostasis . We believe a similar event occurs during ILM removal : as its endfeet are separated with the ILM , (video19) this traumatic event induces Muller cell gliosis, which in turn helps fight the edema .
|Why is ILM peeling more difficult in edema than in iERM ?|
We demonstrated that in iERM there is a horizontal gliosis. This gliosis is, most of the time, a dissecting gliosis, which facilitates the dissection : the cleavage is already present. When the horizontal gliosis is not present, we have to create the cleavage plane ourselves.
However, the use of Brilliant Peel facilitates the dissection such that, nowadays, no major great difficulty is found whatever the case : idiopathic ME (video17), diabetic ME (video20) or post venous thrombosis ME. (video21)
Contrary to commonly held belief, our hypothesis is that the iERM is not simply a pathological epiretinal (i.e., on top of the ILM) structure whose removal restores the normal anatomy and function. The iERM is a periretinal structure : with a gliosis that occurs both underneath the ILM and on top of it.
We interpret the gliosis as a repair mechanism, a healing attempt, by the body to a noxa (ischemia ? traction ? unknown ?).
While removal of the epiretinal part of the structure (leaving the ILM) is beneficial in respect of the light reflection/scattering and the fold-formation effects, it does not address the wrinkling effect, nor does it bring the intraretinal benefit ILM removal induces. Removal of the ILM, represents a stimulus for the Muller cells whose gliosis-response will have a much greater effect that the superficial - astrocytic - gliosis. Muller cells gliosis effects the entire depth of the retina, fighting edema, maintaining homeostasis, using its synaptogenesis properties : the Muller cell is increasingly understood to exhibit progenetic/stem cell characteristics .
iERM and ME
|Based on clinical findings|
1. Reflection by the membrane
The intensity of the reflection is not proportional to the severity of the disease, but rather to the retinal capacity to produce an astrocytic proliferation to fight against the BRB breakdown. In addition, if posterior hyaloid fibers remain attached, the astrocytic arms expansions enter these fibers, thus increasing the membrane’s reflection. This explains why there is a complete dissociation between the intensity of the reflection and the vision drop.
Commonly, despite a 20/25 VA, highly reflective membranes surrounding macular pseudoholes can be observed. This indicates that the retina had the time to react by inducing a gliosis at the ILM level, and that an adherent vitreous plaque is present around the fovea. This is why the intensity of the membrane reflection alone is not a major argument for indicating surgery.
2. Macular folding and ectopia
These are much more important in the decision-making on whether to operate.Without removal of the contracted structure, the retina will never unfold and the fovea will remain repositioned. The decision to operate is all the more easy to make since in the vast majority of eyes there is a significant drop in vision and a high degree of metamorphopsia.
|Based on angiographic and OCT characteristics|
The use of imaging systems is very useful in the following cases :
PH traction without visible retinal consequences ;
less advanced pathology (e.g., iERM with low reflection) ;
high myopia (e.g., low contrast, foveoschisis) ;
ophthalmologist-patient communication (counseling).
Spectral-domain OCT allows the surgeon to visualize fine details never seen before ; consequently, these abnormalities (such as photoreceptor destruction) could not be entered into the decision-making process.
|Based on visual acuity and metamorphopsia|
Even if metamorphopsia is less commonly mentioned by the patient than the VA - and less often discussed by the ophthalmologist - when the functional consequences of the iERM are discussed, they are equally important factors to consider. Metamorphopsia - if it is recognized - can be very disturbing, even in the presence of good VA. (This explains why surgery can be indicated even if excellent VA is maintained and why patients whose VA did not improve with surgery are nevertheless satisfied with the outcome.) However, VA loss threatens to cause permanent function loss.
Visual deterioration caused by an iERM should be compared to one caused not by cataract but by one caused by glaucoma. The visual loss is reversible only when it is mild and fresh ; but if there are many non-functioning cells at the time of the surgery, the operation will not be able to completely reverse the lost function.
There are three important issues to discuss when considering surgery based on the VA.
- 1.The condition of the fellow eye
Should the surgeon deny the patient an operation for an iERM if the fellow eye has full vision ? The short answer is : No. The fellow eye may lose vision in the future to an unrelated cause ; by then so much may have been lost in the first eye that only limited final vision can be achieved.
- 2.Low cut-off visual level
Improving vision from 20/500 to 20/250 is as beneficial as improving from 20/50 to 20/25 ; therefore there is no justification not to perform surgery even if the VA is very low. However, there are a few exceptions. For instance, in an elderly patient in very poor general health, having 20/500 vision for an extended period, and presenting with a macrocystic edema that has destroyed the macula, surgery should be discouraged.
- 3.High cut-off visual level
If there is metamorphopsia that disturbs the patient, surgery is indicated even if the vision is 20/25 ; many active patients, despite having 20/25 VA, must close the affected eye to be able to drive a car.
If no metamorphopsia is present, the following needs to be taken into consideration :
- Is there evidence of progressive visual loss ?
- a patient whose vision has remained unchanged 20/40 for 3 years will not necessarily benefit
from an operation ;
- a patient who experienced a drop from 20/25 to 20/35 in the last 4 months will.
- a patient whose vision has remained unchanged 20/40 for 3 years will not necessarily benefit
- Is there angiographic and/or OCT evidence of ME ? The greater the edema (i.e., increased dyepooling and retinal thickness/volume) the more important to induce a GFAP filling of the Müller cells by ILM removal.
- Is there evidence of progressive visual loss ?
In summary, crucial factors to consider :
- Natural history : with few exceptions, the process is progressive.
- The longer the pathology persists, the higher the probability of sustaining irreversible visual loss.
- Surgery has certain risks.
- The decision whether to operate is not the ophthalmologist’s ; it is the patient’s.
The benefits and risks of surgery are summarized in table 6 .
The ultimate goal of surgery is to restore/maintain reading vision (>20/50). An analysis of our own cases clearly demonstrates the risk of waiting. The probability for the patient not to have at least 20/40 vision four years after surgery is :
5% when the preoperative vision was 20/40 ;
10% when the preoperative vision was 20/50 ;
30% when the preoperative vision was 20/100.
Surgical technique (in 2008) - General concepts
The overall goal is to perform the minimal intervention necessary and do what is necessary to obtain the best VA, as long as it does not increase the rate of complications . All strategic choices are based on this principle.
|The slit lamp-microscope|
The advantages of this approach include the following : (learn more on slit lamp advantages)
- The optical cut improves tissue recognition just as the slit lamp does.
- Due to the 5 to 7-degree illumination angle, a higher amount of light is reflected back to the surgeon’s eye, compared to the 25-degree angle with the intraocular light pipe. (video22), (video23)
- The light delivery device remains external, which makes for three advantages.
- The risk of photoxicity, which increases in proportion to the square of the distance, is eliminated since the light source is at 200 mm from the macula. (video24)
- No hand of the surgeon is required to hold an illuminating pipe.
- No extra sclerotomy is required.
- The liberated hand of the surgeon allows additional (finger-)stabilization of the forceps (The stabilizing effect by the finger of the liberated hand decreases tremor and increases precision and speed) center the contact lens, or turn the angle of the forceps-jaw platforms. (video25)
- Reduced cost (no need for disposable intraocular light pipe for each surgery).
|Viewing of the macula|
In macular surgery, no need to have a wide field ; there is however, need for the highest possible resolution. The plano-concave contact lens provides an image with a much higher resolution and optical clarity than with any wide angle system (video26), available.
The classical 20 g technique has the following advantages.
- Both 23 and 25 g forceps have smaller platforms ; they necessarily have a much higher tendency to tear the tissue (To remove a wallpaper, I will use my two hands to have a larger grasping effect ; I will not use two fingers).
- Both 23 and 25 g forceps have a higher flexibility than a 20 g forceps does. The surgeon wants to fight the disease, not the instrument.
- For surgeons using slit lamp illumination and aspiration flow control, the time saved by faster vitrectomy is greater than the time saved during the opening and conclusion of 23 and 25 g procedures (video Vitreous Nascar). This is also true for the speed of the ILM-peeling time.
- The cost of a 20 g set is substantially lower than either a 23 or a 25 g’s.
It is true that traditional 20 g surgery creates more scarring than transconjunctival surgery. However, our conjunctival incision is only 50 degrees wide, is underneath the superior lid, and postoperative appearance is similar to that after transconjunctival surgery. Furthermore, in many cases suturing is necessary even in 23 or 25 g surgery. Finally, transconjunctival surgery is also possible with 20 g.
Completeness of vitreous removal
The decision must take into consideration the efficacy of the procedure and the risk it entails. We prefer core vitrectomy, which, using our parameters, is both efficient and, by avoiding exerting traction at the vitreous base, has not increased the postoperative RD rate (2%). In iERM, the vitreous is typically healthy, obviating the need of more complete vitreous removal, except in eyes with widespread vitreous pathology such as floaters.
Surgical technique - The actual surgery
|Conjunctival opening and sclerotomies|
We systematically perform a 2-port vitrectomy, either with or without conjunctival opening.
- Traditional 20 g technique. (video27) • 50-degree conjunctival incision from 10:30 o’clock to 12 o’clock. • Light diathermy with a bipolar curve forceps at 11 and 12 o’clock. • Two scleral incisions are created, which are trapezoid. (video28) • A self-maintaining, disposable, 4.5 mm long infusion cannula is placed at 12 o’clock • Its intravitreal position is verified with postero-centripetal movement.
- If a transconjunctival technique is chosen, the same entry sites are selected, except when previous or subsequent glaucoma surgery requires individual sclerotomy placement. (video29) But even in these cases I will suture the sclerotomies at the end of the surgery. (video30)
- The maximum aspiration flow is 20 ml/min (2 ml/min below the maximum infusion flow) ; the cut rate is 600 to 800 per minute. These parameters avoid inducing strong traction at the vitreous base. If a sudden hypotony occurs, there can be two reasons : the infusion tubing is pinched, in which case this needs to be untangled ; or the infusion tip is covered by the detached ciliary pigment epithelium, in which case the tip needs to be freed.
- The vitrectomy probe is introduced directly towards the macula. (video31)
- A posterior vitrectomy is performed, which usually takes less than one minute.
- The vitrectomy probe is placed close to the retina and with the aspiration flow and cutting action
activated, the retina is examined.
- If there is no PVD, the retina will flicker (with a very high cut rate, this phenomenon will not occur, regardless of the status of the vitreous). (video32) The posterior vitrectomy is stopped at 3 mm from the retina, leaving thus a sufficient mattress of vitreous attached to the retina to allow PVD creation by engaging this mattress. Aspiration flow is applied without cutting and without inducing traction on the periphery to detach the PH. The detached PH is removed.
|Staining of the ILM|
1.The rationale for using a dye and for dye selection
The question then is “to dye or not to dye”. My position has been changing slowly from 1985, when I started to remove the ILM, and I must explain why :
Before staining became available, I already performed a lot of ILM removals, thanks to the visual benefits of surgical slit lamp illumination. I gained significant experience with ILM peeling without any staining.
Staining, though, does have tangible benefits, but not all dyes are created equal.
- Triamcinolone is not a true dye ; it simply adheres to the collagen fibers. I use it only to mark the PH in cases where it is difficult to identify and to remove it, such as in highly myopic eyes or vitreoschisis.
- Membrane Blue, stains the vitreous fibers as well as the ILM and even the posterior capsule. It has thus no selectivity and obliges the surgeon to perform a fluid-gas exchange before Membrane Blue injection.
- ICG is the most efficient stain for the ILM ; it does not stain the epiretinal gliosis but it does stain the vitreous. Toxicity is a potential problem, but this can be addressed by using glucose as a solvent and avoiding ICG in eyes with a macular hole (subretinal ICG pooling). ICG is a great tool for those with less experience to learn how to remove the ILM and even for more experienced surgeons to increase control and precision. However, for me its use was never considered an option because of the lack of official approval.
- When Brilliant Peel appeared, all my reluctance toward stain use disappeared. It colors the astrocytic proliferation as well as the ILM sufficiently and is officially approved.
• Half of the original bottle of Brilliant Peel is injected over the macula to stain the ILM, and to demonstrate :
- The presence/absence of the PVD (if the diagnosis has not be done during the core
- Any remaining vitreous plaque - which is in “negative staining”. (video34)
- Sometimes the presence of a spontaneous tear of ILM (video35), avoiding then to make retinal damage. • A tuberculine syringe (1 ml) and a 25 g needle are used to avoid creating a high jet-force that could result in the dye damaging the retina or entering the subretinal space. • The dye is immediately aspirated by holding the vitrectomy probe’s port horizontally over the dye. (video36) This maneuver induces a rotation of the colorant as in a washing machine. (video37), (video38) While turning, the dye “brushes against” the retina, increasing the dying effect. If heavy Brilliant Peel is used, there is no need to use this washing machine technique. (video39) The heavy Brilliant Peel gives a deeper staining. (video40) • Dye reinjection, using the remaining 50% in the bottle, may become necessary if there is any doubt regarding the completeness of ILM removal. (video41), (video42)
If possible, the remaining vitreous plaque, the epiretinal gliosis, and the ILM are removed in a single piece. This decreases both the surgical time and the number of entries through the sclerotomy.
- Use of the Tano diamond scraper is contraindicated ; its crystals will cause permanent nerve damage if used over retinal areas devoid of the ILM.
- A sharp pick or the tip of a bent needle is used to initiate a plane of cleavage in the posterior hyaloid, before continuing with a hook and/or a forceps (two-step procedure) ; the risk is to cause retinal damage. To reduce the risk, three recommendations should be considered :
- Work at the highest magnification of the microscope.
- Use a very sharp instrument. If the instrument is blunt, you will push the tissue in front of the instrument, creating a retinal fold (see below), and risking entering too deep.
- Start in an area without retinal folds in order to avoid entering the retina deeper than the ILM. If you cannot find a suitable area without retinal folds, never move your sharp instrument perpendicular to the folds.
- Many forceps are available :
- The Tano or Eckardt models are very precise. They allow selective ILM grasping as they have a small contact area, but they tend to tear the ILM. (video43) In addition, they are ineffective for any ill-defined tissue.
- The crocodile-type of forceps has a very high grasping power but has lower selectivity.
- A forceps that combines these advantages is a crocodile style one (thus providing a huge grasping force) but at the tip of one of the jaws there is a very thin hook to engage the edge of the membrane (In fact working with this forceps is similar to detaching a wallpaper : you first use the hook as you do with
your fingers when engaging the edge of the paper, and then you close the jaws and use the large platform as
you do when using your two hands to detach as much as possible of the wallpaper in one piece.). (video44).This is why my favorite forceps is the one produced by Synergetics under the reference of my own name.
- a. Where to start
I usually recommend to start the peeling at a site that allows convenient access and comfortable hand positioning. It should not be too far from the macula, and in a location where the retina is flat. Whether the “point of attack” is temporal or nasal to the fovea varies according to laterality (right or left eye) and to which hand the surgeon uses, as well as on how dilated the pupil is, the presence of a crystalline lens opacities , etc.
You can also select different locations according to what the staining shows : if there is a plaque of attached vitreous you can either decide to start in an area where the ILM is free of vitreous cover and then remove both the PH & ILM in a single piece (video45), or decide to remove first the PH and then the ILM in a second step. (video46)
There are locations that should be avoided, especially by the beginners to start the peeling :
- The foveola ;
- The maculopapillary bundle as retinal damage here could have a major deleterious visual effect ;
- The area close to the vascular arcades as the vitreoretinal adherence is very strong there ;
- In the area of retinal folds (see above) ;
- In an area where the ILM-cover is missing, due to it having been spontaneously torn by the contracting force of the vitreous plaque. The ILM will be dragged toward the plaque, rolled, and leave a non-stained (by Brillant Peel) area behind, with a sharp edge . Any attempt to engage
the dye - free zone implies damage to the axons. (video35)
- b. Basic manipulations/considerations
The overriding principle is to minimize traction exerted on the retina.
- Point of attack : Engage the membrane at the point of least adherence (to the retina) and move towards the point of most adherence.
- Vector of forceps movement : The intravitreal path of the forceps should be parallel to the retina while staying relatively close to it.
In case the membrane has two areas with strong adhesion, the surgeon should never grasp the membrane in the middle and simply lift it. While this appears very convenient, it exerts traction on the retina at both locations and can easily result in a retinal tear. Instead, the surgeon has two options.
- One, you cut the membrane in the middle and attack each separately.
- Two, you dissect the membrane close to one of the adherences to liberate it from this adhesion and then continue peeling towards the other point of adhesion. (video47)
- c. The area (size) of ILM removal
In order to eliminate the risk of macular contraction recurrence (caused by “leftover” vitreous), the ILM must be removed in a large area around the foveola (i.e., two-thirds of the distance between the foveola and the vascular arcades). Brilliant Peel can be injected to confirm that the ILM has indeed been removed in the entire area and that no ILM island has been left behind.
- d. Per operative incidents and complications :
- Hemorrhages :
- Punctiform hemorrhages are very often seen in the area of ILM removal as the astrocytic cells are connected to the vessels ; it is indeed a signal showing that the ILM has been removed. They soon disappear and are rarely seen the next day.
- Larger hemorrhages can be seen either when pinching in front of a retinal vessel or when tearing the retina during the peel. When this occurs, I usually elevate the intraocular pressure and press slightly with the vitrectomy probe’s tip on the bleeding vessel for 30 seconds. (video48) This is usually sufficient ; there is seldom a need for endodiathermy.
- Retinal tears can be seen mostly in 3 circumstances :
- Strong adherences : this is really exceptional in iERM ; it is mostly seen in inflammatory cases or post-RD cases.
- During pinching of the “membrane” (video49) ; This can be avoided by :
- Using the highest magnification of the microscope ;
- Using the slit lamp of the microscope, thus allowing the free hand to stabilize the forceps. If the surgeon does not have a third point of stabilization for the forceps, having wrist support for the active hand is crucial.
- Losing sight of tip of the forceps and bumping into the retina. The surgeon should - repeatedly if necessary - re-grasp the membrane, close to its point of insertion, to avoid the forceps from bumping into the retina at a point far away from where the membrane is being torn since this is where the surgeon focuses.
- Hemorrhages :
If a retinal tear does occur, I make sure that all remnants of vitreous around the tear have been removed, I inject an air bubble, and ask the patient to position for 6 hours. No laser or cryopexy is performed.
3. Different options according to the configuration of the vitreomacular interface (PVD, anomalous PVD, no PVD)
It must be emphasized again that the presence of a prepapillary (Weiss) ring is evidence only that the posterior hyaloid is detached in front of the disc ; the vitreous can still be attached elsewhere or leave large plaques on the macula.
- a. There is no adherent hyaloid on the macula. (As mentioned before, this can appear in two circumstances : spontaneous PVD, an uncommon occurrence ; or after a surgical PVD).
In this scenario Brilliant Peel uniformly stains the retinal surface. Therefore we are sure that the ILM will be the first structure that our instrument will encounter when touching the retina. Many peeling techniques are then available.
My favored technique is to do everything with the forceps (one-step procedure) : I start by pinching smoothly the retina, exerting small movements sideways in both directions. As the ILM is the only rigid structure of the inner retina, the ILM will easily detach from the underlying tissue.* (video50)
An ILM tear is then created by a larger retina-parallel forceps movement. I perform a centrifugal ILM rhexis around the fovea, in order to remove at least two-thirds of the ILM surface inside the arcade (video51). This is very similar to the manipulation used in macular hole procedure. (video52)
Whatever the technique, the detached specimen, composed of ILM fragments plus glial cells but without vitreous fibers, appears as a thin, non-collapsed, transparent, scrolled membrane, with sharp edges.
- b. There is only a small plaque of hyaloid. The plaque of hyaloid is visible by its higher reflection power and its lower Brilliant Peel (IGC) staining properties. There are two options :
- c. Vitreoschisis or absence of PVD. The hyaloid is a soft tissue composed of collagen fibers. A PVD must be created before the ILM can be attacked. As the PH is rarely rigid, the forceps is not the proper instrument to separate it from the ILM. It will grasp the collagen fibers where they are pinched, but the lack of cohesion does not allow separating it from the ILM even in the adjacent area. (video4), (video55)
To create a plan of cleavage between the PH and the retina, I use a needle with a tip bent at 45 degrees (creating a hook), with the opening facing the retina.*** I gently enter into the fibers and can then observe some fibers covering the tip of the needle ; I then elevate the needle, thus creating an initial detachment of the PH. (video56)
Widening of the PVD area can be done using different instruments : the same hook, forceps, vitrectomy probe. The detached tissue appears as a contracted, shriveled structure, indicating that it contains vitreous elastic fibers. There is always the question : did I remove the PH alone or the PH
together with the ILM, and if the latter, how large a piece of the ILM ?
While presence of microhemorrhages and retina whitening are indirect signs of ILM removal, it is adviseable to reinject the Brillant Peel (or use ICG) to visualize whether any ILM remnants are still left behind.
• Close the working sclerotomy with an absorbable monofilament suture whose elasticity provides extra closing power. • Remove the infusion cannula and close this sclerotomy the same way. • Use the bipolar diathermy to close the conjunctival incision. (video57)
We performed in 1998 a the statistical analysis on 1413 eye that underwent iERM removal in order to evaluate the complications rate. With a mean follow up of 45 months, we observed the following : • Induced cataract : Only 52.1% of phakic patients were operated for cataract. Many explanations can be found for this rate, which is lower than usually reported. This can be due to a combination of a number of factors such as the fact that our technique of surgery is perhaps less invasive than others : faster procedure (the entire procedure rarely exceeds 7 to 8 minutes), core vitrectomy (most of the « normal » vitreous is left intact), use of the slit lamp (avoiding any risk of touch by the illuminating system and allowing to visualize the posterior capsule). This low rate is my reason for not routinaly performing combined phaco-vitrectomy. • Post-op retinal detachment : Only 1.90 % were observed in our series, around the same rate as aftera cataract extraction, explaining why I never check the peripheral retina at the end of the procedure. Many explanations can be found for this low rate :
- performing a core vitrectomy only, which acts on two levels :
- The vitreous incarcerations do not induce a traction on the vitreous base ; they are absorbed by the large mattress of vitreous.
- A complete vitrectomy is iatrogenic tears inducer.
- using two sclerotomies only : less risk of incarceration, no traction induced by the manipulation of the intraocular light pipe.
- additionaly reduced risk of vitreoretinal incarceration achieved by :
- a small number of entries/exits (vitrectomy performed without any exit, strong forceps allowing to removal of larger pieces of membrane).
- automatic stop of the infusion each time the vitrectomy probe or the forceps is removed from the eye.
- Vitrectomy machine equipped with flow control, allowing a faster procedure and preventing any risk of having a higher aspiration flow than intended (less risk of traction on the vitreous base, no risk of hypotony).
- No forced PVD : while doing this maneuver we just need to pay attention not to detach the PH too peripherally (video58), avoiding therefore to create peripheral tears. (video59)
This technique personally developed over many years and thousands of cases allows me to perform (video60) a complete operation of iERM removal within 5 to 6 minutes, without increasing the low rate of complications. In our opinion, this is the true way to perform minimally invasive surgery.
 Green WR, Kenyon KR, Michels RG, Gilbert HD, De La Cruz Z. Ultrastructure of epiretinal membranes causing macular pucker after retinal re-attachment surgery. Trans Ophthalmol Soc U K 1979 ;99:65-77.
 Reuss B, Dono R, Unsicker K. Functions of fibroblast growth factor (FGF)-2 and FGF-5 in astroglial differentiation and blood-brain barrier permeability : evidence from mouse mutants. J Neurosci 2003 ;23:6404-6412.
 Becquet F, Zanlonghui X, Libeau L. Pre- and postoperative functional and anatomical changes in patients with idiopathic ERM. Invest Vis Sci 2005 ;Suppl:Poster Board Number : 5442/B5645
 Konstantinidis L, Berguiga M, Beknazar E, Wolfensberger TJ. Anatomic and functional outcome after 23-gauge vitrectomy, peeling, and intravitreal triamcinolone for idiopathic macular epiretinal membrane. Retina 2009 ;29:1119-1127.
 de Bustros S. Vitrectomy for prevention of macular holes. Results of a randomized multicenter clinical trial. Vitrectomy for Prevention of Macular Hole Study Group. Ophthalmology 1994 ;101:1055-1059 ; discussion 1060.
 Ducournau D. Internal limiting membrane removal in stage1 macular hole. The 20th Annual Vitreous Society Meeting, September 30th, 2002, Abstract Book p 162