Flow diagram depicting the 5 categories of specimens examined in this series.
Vitreous strands. A, Vitreous strand (arrows) without cells. B, Vitreous strand with rare inflammatory cells. C, Vitreous strand with lymphocytes (cellulose membrane filter, modified Papanicolaou stain, original magnification: A, ×20; B, ×35; and C, ×544).
Fibrocellular membrane fragments. A, Fibrocellular membrane fragment. B, Fibrocellular membrane fragment (arrow) with adherent vitreous strand (arrowhead). C, Fibrocellular membrane fragment (cellulose membrane filter, modified Papanicolaou stain, original magnification: A, ×300; B, ×65; C, ×210).
Retinal fragments. A, Definite retinal fragment with photoreceptor inner segment (arrowhead) (cellulose membrane filter, modified Papanicolaou stain, original magnification ×500). B, Probable retinal fragment (cellulose membrane filter, periodic acid–Schiff, original magnification ×500).
Cellular membrane fragment (cellulose membrane filter, periodic acid–Schiff, original magnification ×210).
Internal limiting lamina fragments. A, Crinkled, cellophane-like appearance of an internal limiting lamina fragment. B, Internal limiting lamina fragment (arrow) with an adherent fibrocellular tissue (arrowhead) (cellulose membrane filter, periodic acid–Schiff, original magnification: A, ×85; B, ×130).
Lens capsule (A) and cortex (B) fragments (cellulose membrane filter, modified Papanicolaou stain, original magnification: A, ×340; B, ×165).
Sadda SR, Campochiaro PA, de Juan E, Haller JA, Green WR. Histopathological Features of Vitreous Removed at Macular Hole Surgery. Arch Ophthalmol. 1999;117(4):478-484. doi:10.1001/archopht.117.4.478
To describe the histopathological features of the vitreous removed at surgery for macular holes in 200 consecutive cases.
The complete vitrectomy specimen in each case was concentrated by means of cellulose membrane filters and stained for light microscopy. The cases were organized into 5 categories: (1) all cases (N=200), (2) eyes without previous vitrectomy (n=174), (3) eyes with previous vitrectomy (n=26), (4) idiopathic cases (n=143), and (5) traumatic (accidental or surgical) cases (n=31). The type and frequency of tissue fragments present in the vitreous were determined for each case.
Fibrocellular and cellular membrane fragments were found in a minority of cases in all categories. Retinal fragments were a rare finding, present in only 4 cases. Inflammation was present in 57 (28.5%) of all cases.
The absence of fibrocellular and cellular membrane fragments in the majority of cases suggests that mechanisms other than cellular proliferation are important in the pathogenesis of macular holes. These fragments are, however, the likely histopathological correlate of the opercula that are often observed clinically in patients with macular hole. Opercula rarely if ever contain retinal fragments, and thus are better termed pseudo-opercula, as has been previously suggested. The cellular proliferation and inflammation that are observed in some of the cases are likely a secondary or reactive process.
MACULAR HOLES have become the focus of much interest and controversy in ophthalmology. Much of this renewed interest stems from new theories of pathogenesis1- 4 and the development of a possible surgical treatment5 for macular holes. Despite the numerous proposed theories, the pathogenesis of these lesions is still not well understood. Most current investigators1- 4 believe that tangential vitreous traction plays an important role in their pathogenesis. However, there are several mechanisms by which this tangential traction may be produced. Gass1,3,4 theorized that condensation and contraction of the prefoveal cortical vitreous with glial cell proliferation in this condensed vitreous may generate tangential traction. Guyer and Green2 suggested that fluid movements of the liquefied vitreous in an enlarged premacular bursa can exert traction on the remaining formed cortical vitreous, with that traction transmitted tangentially to the fovea.
Two large histopathological series of postmortem cases of macular hole were reported by Frangieh et al6 and Guyer et al.7 These investigators observed that full-thickness holes were characterized by the absence of all retinal layers with rounding of the retina at the margins of the hole. Cystoid macular edema, epiretinal membranes, and retinal pigment epithelial (RPE) changes were the most frequent associated features.6,7 A localized area of retinal detachment was often present, and there was variable degeneration of the photoreceptors adjacent to the hole. Cystoid macular edema and/or retinal detachment is believed to be the histopathological correlate of the "subretinal fluid cuff" observed clinically. Frangieh et al6 and Guyer et al7 also reported the features of spontaneously healed macular holes. These holes were sealed by hyperplastic RPE or glial cells.
The recent interest in surgery (vitrectomy with gas tamponade) for the treatment of macular holes has provided additional material for histopathological study. We studied the histopathological features of the vitreous from eyes with macular holes by examining aspirates obtained at the time of vitrectomy. We sought to investigate the relative importance of cellular proliferation in macular hole formation by examining the frequency of fibrocellular and cellular membrane fragments. We compared the features of the vitreous in idiopathic vs traumatic macular holes. We also sought to characterize better the composition of macular hole opercula, which were observed clinically in many of the cases in this series. Previous studies of macular hole opercula8,9 have been selective—ones in which the structure believed to be an operculum was larger and easily visualized by the surgeon.
We examined 200 consecutive vitreous aspirates obtained during pars plana vitrectomy for macular hole performed at the Wilmer Eye Institute, Baltimore, Md, between June 6, 1990, and August 7, 1996. Eighteen surgeons were involved. Of the total specimens, 153 (76.5%) were submitted by 3 surgeons. The clinical records of each case were examined to determine the patient's age, sex, race, and ocular history (specifically a history of trauma or surgery). Every patient underwent a standard 3-port pars plana vitrectomy. Twenty-one patients also underwent a simultaneous extracapsular cataract extraction or pars plana lensectomy.
The complete vitrectomy specimen from each case was concentrated by means of mixed cellulose acetate and nitrate membrane filters (5 µm pore size, 47 mm diameter, Millipore filter; Millipore Corporation, Bedford, Mass) and stained with a modified Papanicolaou stain, and some were also stained by the periodic acid–Schiff technique. The entire filter specimen was examined by light microscopy in a systemic fashion, with the use of both low (×10) and high (×40) magnification.
In 15 cases, additional material was collected by the surgeon and submitted separately for special studies. This additional material was identified by the surgeon as an epiretinal membrane in 9 cases, posterior hyaloid in 4 cases, an operculum in 1 case, and an operculum with cortical vitreous in 1 case. In 2 other cases, isolated material was taken for special studies by 1 of us (P.A.C.). In 1 of these cases, the tissue was prepared for immunohistochemical staining by the technique described by Vinores and coworkers10 and stained for RPE-specific markers TUJ1, RPE9, and RPE15.11- 13
The cases were categorized into 5 groups (Figure 1). Of the total 200 eyes, 26 were found to have had a previous vitrectomy. Of these 26 cases, 24 had undergone a previous vitrectomy for idiopathic macular hole (in 13 cases the hole persisted after surgery, and in 11 the hole reopened after an initially successful closure period). Of the remaining 2 cases, the previous vitrectomy was performed for retinal detachment (1 case) or macular pucker (1 case). The remaining 174 cases were further subdivided into idiopathic (n=144) and traumatic (n=30) groups. Of the 143 idiopathic cases, 5 patients had diabetes, 1 had age-related macular degeneration, 3 had idiopathic high myopia, and 2 had previous cystoid macular edema (1 pseudophakic case and 1 secondary to uveitis). In both cases with a history of cystoid macular edema, the edema had completely resolved ophthalmoscopically and angiographically at least 1 year before the development of the macular hole.
The average age of the patients in this series was 66.5 years, and 57.5% of the patients were female. The duration between onset of symptoms and pars plana vitrectomy for macular hole was determined by the clinician in 134 of the 200 cases, and ranged from 1 week to 14 years. In 62% (83/134) of the cases, the hole was present for 4 months or less (based on symptoms).
The stage of the hole was determined by the surgeon in 124 of the 200 cases. There was 1 stage I hole, 15 stage II holes, 53 stage III holes, and 51 stage IV holes. Two holes were classified as being "stage II or III," and 2 were classified as "stage III or IV."
The types of tissue fragments present in the vitreous included vitreous strands (with or without cellular proliferation, and with or without inflammatory cells) (Figure 2), fibrocellular membrane fragments (Figure 3), cellular membrane fragments (Figure 4), internal limiting lamina fragments (with or without a cellular proliferation) (Figure 5), lens fragments (Figure 6), and retinal fragments (Figure 7). The degree of cellular proliferation and inflammation in the vitreous strands was graded quantitatively as being mild, moderate, or marked.
The frequency of these various fragments was tabulated for each of the 5 categories, and the results are summarized in Table 1. Retinal fragments were observed in only 1 of the 4 traumatic cases. The fragment could be determined to be a definite retinal fragment in only 1 (a traumatic case) of the 4 cases. In 1 case, the fragment was classified as a probable retinal fragment and in the remaining 2, as a possible retinal fragment.
As shown in Table 1 (column 4), lens fragments were observed in the vitreous in 8 cases of idiopathic macular hole. In all 8 cases the crystalline lens was removed at the time of macular hole surgery, in 2 cases by lensectomy and in 6 by extracapsular cataract extraction. Lens fragments were also present in 14 cases of traumatic macular hole and in 3 cases of patients with a macular hole who had undergone a previous vitrectomy. In all cases, either the patient had a history of previous cataract surgery or the lens was removed at the time of macular hole surgery.
In 13 (6.5%) of the 200 cases, the surgeon recorded in the office record or operative report that he or she had observed an operculum. In all of these cases the operculum was noted intraoperatively by the surgeon. In 11 cases, the operculum was submitted in the vitreous cassette. In 2 cases, the operculum was removed with forceps and submitted as a separate specimen for transmission electron microscopy. However, in both of these cases, no tissue was recovered after processing for transmission electron microscopy.
Isolated specimens were obtained in 15 cases. Thirteen of these were submitted for electron microscopy. Of these 13, 9 were labeled "epiretinal membrane." Three of the 9 contained no tissue. Of the remaining 6 cases, transmission electron microscopy disclosed internal limiting lamina only in 1 case, collagen fibrils (15 nm) only in 1 case, and a fibrocellular membrane fragment in 4 cases. Cellular elements present in the fibrocellular membranes included fibrocytes (2 cases), myofibrocytes (1 case), fibrous astrocytes (3 cases), fibrous astrocytes with myoblastic differentiation (1 case), macrophages (1 case), and RPE (1 case). Of the 4 specimens labeled "posterior hyaloid," 2 contained no tissue, 1 was lost during processing, and 1 contained native collagen fibrils (15 nm).
Two additional epiretinal membranes were taken by one of us (P.A.C.) for special studies. In 1 of these 2 cases, light microscopy disclosed a thin fibrocellular membrane in which some of the cells stained positively with RPE-specific immunohistochemical markers (TUJI, RPE9, and RPE15). Examination of the second case also disclosed a thin folded fibrocellular membrane composed of spindle cells, round cells, and rare pigmented cells. Marker studies were not performed.
The features of the vitreous were also classified according to the duration (Table 2) and stage (Table 3) of the hole. Inflammation was more frequent in older holes (42.3% of holes present longer than 4 months compared with 24.4% of holes present for 4 months or less). Fibrocellular and cellular membrane fragments were more frequent in more advanced-stage holes (31.4% of stage IV holes compared with 6.7% of stage II holes).
We report the histopathological features of the vitreous removed from patients at the time of macular hole surgery. We observed cellular and fibrocellular membrane fragments in only 23.5% of these patients. Vitreous strands containing a cellular proliferation were present in 37.5% of cases. Usually, the cellular proliferation was mild or moderate. Internal limiting lamina fragments with a cellular proliferation were present in 12 (6.0%) of cases. Overall, some degree of cellular proliferation (cellular or fibrocellular membrane fragments, internal limiting lamina with cellular proliferation, or vitreous strands with a cellular proliferation) was present in 51.5% (103/200) of cases.
Previously, Gass4 suggested that glial cell proliferation in the condensed prefoveal cortical vitreous could contribute to tangential tractional forces leading to macular hole formation. Gass's1 theory was supported by the findings of Yoon et al,8 who examined epiretinal tissue removed at the time of surgery in 12 selected patients with unilateral idiopathic macular hole. Yoon and coworkers observed a cellular proliferation (fibrocytes, myofibrocytes, RPE, and/or fibrous astrocytes) in 1 of 2 stage II holes, 4 of 7 stage III holes, and in all of 3 stage IV holes. Yoon et al concluded that idiopathic macular holes formed because of contraction of the prefoveal cortical vitreous with progressive enlargement of the hole caused by contracting myofibroblasts on the inner surface of the internal limiting lamina. The study, however, was limited by its small sample size. Although Yoon et al reviewed a consecutive series of cases received at the eye pathology laboratory, some of the cases were selected and sent to the laboratory by the referring surgeons; thus, the study is subject to selection bias. Madreperla et al9 examined cells of pseudo-opercula harvested at macular hole surgery and found no myoblastic or contractile features. In addition, there was no evidence of a strong attachment between the external surface of the pseudo-operculum and the retina; a firm attachment would be expected if this tissue were to generate enough force to create a hole in the retina. Finally, as noted by Madreperla et al,9,14 other contracting epiretinal membranes that overlie the fovea do not usually produce macular holes.
The lack of cellular proliferation in many cases in our series suggests that other mechanisms, such as the fluid countercurrents proposed by Guyer and Green,2 may play an important role in the development of macular holes. Moreover, it is possible that cellular proliferation is not of primary pathogenetic importance, but is rather a reactive or secondary process that occurs in response to the hole. A number of investigators6,7,15- 17 have found that healed macular holes (both spontaneously healed holes and surgically treated holes) are sealed by a proliferation of cells (glial or RPE). The proliferation of cells in our series and that of Yoon et al8 may represent an unsuccessful or incomplete healing response to the hole. In some cases, rather than remaining incomplete, this healing response may become overly exuberant, resulting in reopening of the hole. Fekrat et al18 reported the histopathological features of an epiretinal membrane from a macular hole that recurred after initially successful surgical closure. The authors concluded that the reparative process that had initially led to closure of the hole went awry, resulting in reopening of the hole.
Our findings also suggest that macular hole opercula are rarely, if ever, composed of true retinal tissue. Only 1 case in our series contained a definite retinal fragment, and in this case, the margins were angular and there was no adherent vitreous or fibroglial proliferation. It is possible that this retinal fragment came from the margin of the macular hole and was pulled free during peeling of the prefoveal cortical vitreous at the time of surgery. Gass19 provided a similar explanation for the high incidence of retinal tissue and internal limiting lamina in selected opercula studied by electron microscopy.20
Condensed cortical vitreous (with or without cellular proliferation) or cellular and fibrocellular membrane fragments are the likely histopathological correlates of macular hole opercula. This theory is also supported by Madreperla et al,9 who studied the histological features of 2 opercula removed at the time of vitrectomy for macular hole. The opercula were composed of native collagen (cortical vitreous), fibrous astrocytes, and Mueller cells. No distinct retinal neuronal tissue was present. Thus, macular hole opercula are better termed pseudo-opercula as suggested by Gass,1,3,4 Guyer and Green,2 and Fekrat et al.18
Inflammation is another feature of the vitreous in patients with macular holes in our series. The inflammatory infiltrate was composed of lymphocytes and macrophages and was mild in most cases. Also, inflammation was more common in holes of shorter duration. The significance of these findings is unclear. Although it is possible that these inflammatory cells play a primary role in the pathogenesis of macular holes, we suspect that they are more likely a secondary response to the hole. These inflammatory cells could potentially liberate cytokines and other mediators that could stimulate the proliferation of fibrocytes, fibrous astrocytes, and RPE cells.
The composition of the vitreous in patients with traumatic and idiopathic macular holes is similar. However, the vitreous of traumatic macular holes contains more cellular and fibrocellular membrane fragments than the vitreous of idiopathic macular holes (33.3% vs 22.5%, respectively). Moreover, the traumatic cases more frequently have inflammatory cells. One possible explanation for this difference is that the traumatic insult (accidental trauma or surgery) itself caused an inflammatory reaction in the eye, with subsequent cellular proliferation in response to the inflammation. It is unclear whether the inflammatory response has a significant role in the pathogenesis of macular holes in traumatic cases.
Twenty-four of the 26 patients who had a previous vitrectomy (Table 1, column 3) had the initial vitrectomy for an idiopathic macular hole and in this study underwent a repeat vitrectomy for a persistent or recurrent hole. The features of the vitreous in these cases were largely similar to those of patients with idiopathic macular holes who did not have a previous vitrectomy (Table 1, column 4). There was, however, a slightly higher incidence of inflammation and a lower frequency of cellular or fibrocellular membrane fragments in the repeat vitrectomy cases. The significance of this difference is unclear. The increased inflammation may have been a direct result of the previous surgical trauma. The relative scarcity of fibrocellular or cellular membrane fragments (pseudo-opercula) in these cases may have been the result of the removal of these fragments during the initial vitrectomy. Indeed, the removal of much of the vitreous during the initial vitrectomy would be expected to significantly alter the composition of the vitreous, and may make interpretation of the findings in these cases difficult.
Our study of the vitreous in eyes with macular holes suggests that cellular and fibrocellular membrane fragments are scant or absent in the majority of cases. This finding suggests that mechanisms other than cellular proliferation are important in the generation of tangential traction leading to macular hole formation. We also found retinal fragments to be a rare feature of the vitreous in these patients. They are thus unlikely to be a constituent of macular hole opercula. Opercula are therefore better termed pseudo-opercula, as has been previously suggested.1- 4,18 Cellular and fibrocellular membrane fragments are the likely histopathological correlates of these pseudo-opercula. A mild chronic inflammatory infiltrate (lymphocytes) is present in some cases of macular holes. Inflammation and cellular or fibrocellular membrane fragments are more frequent in traumatic holes than in idiopathic holes. The significance of this difference is unclear, but it is likely a direct result of the initial trauma.
Accepted for publication October 14, 1998.
This study was supported in part by the International Order of Odd Fellows, Winston-Salem, NC; core grant EY 01765-21 from the National Eye Institute, National Institutes of Health, Bethesda, Md; and the Macula Foundation, New York, NY.
Reprints: W. Richard Green, MD, Eye Pathology Laboratory, Maumenee Bldg 427, The Johns Hopkins Hospital, 600 N Wolfe St, Baltimore, MD 21287-9248.