Ординатура / Офтальмология / Английские материалы / The Retina and its Disorders_Besharse, Bok_2011

Figure 4 Right eye. Total rhegmatogenous retinal detachment secondary to a small superior retinal tear. Proliferative vitreoretinopathy is evidenced by posterior starfolds involving approximately 11 clock hours (PVR CP11). HST, horseshoe tear; PVR, proliferative vitreoretinopathy. Reproduced with kind permission from Paul M. Sullivan.

Table 1 Proliferative vitreoretinopathy classification by the Retina Society Terminology Committee

From Machemer, R., Aaberg, T. M., Freeman, M., et al. (1991). An updated classification of retinal detachment with proliferative vitreoretinopathy. American Journal of Ophthalmology 112: 159–165, with kind permission from Elsevier.

Management

The primary aim in the management of RRD is to prevent complete loss of central and peripheral vision, that is, progression to no perception of light. In patients with symptomatic RRD, the secondary aim of management is restoration/improvement of peripheral visual field and central vision. Depending on the type of retinal break, size, and rate of progression of the RRD and PVD status, treatment may be either conservative or active. Conservative management involves observation for progression of RRD without intervention. Active management includes either barrier laser demarcation to wall off the detachment

and prevent further progression, or surgery to reattach the retina. The various treatment options are discussed in turn, categorized by type of causative retinal break.

Conservative

Conservative management consists of either selfmonitoring by the patient for development or progression of RRD symptoms, and/or observation by regular clinical examinations charting any signs of an enlarging RRD. This option is usually reserved for patients with chronic RRD without clinical signs of recent progression who are often asymptomatic and discovered to have RRD as an incidental finding. Another group of patients, where conservative management may be appropriate, comprises those with such significant systemic comorbidity that active treatment cannot be safely administered. When considering conservative management regardless of the specific indication, the ophthalmologist should discuss in detail its risks and benefits versus active therapies with the patients, allowing them to make informed decisions regarding their care. In addition, it is important to ensure that the patient understands the symptoms of RRD progression, and are able to reliably monitor their vision monocularly to detect changes before macula detachment occurs.

RRD due to retinal tear

RRD due to retinal tears occurs following PVD and tends to be rapidly progressive in nature due to persistent vitreoretinal traction on the anterior edge of the tear. It is therefore unusual for this to be asymptomatic or present as chronic RRD, and so much less likely to be managed conservatively.

RRD due to retinal hole or dialysis

Both retinal holes and retinal dialyses are usually associated with an attached vitreous, that is, without a PVD. This is thought to account for the relatively slower progression of RRD compared to those caused by retinal tears. As such, these patients often present with signs of chronicity and nonprogression, and many are asymptomatic until the macula becomes involved. Conservative management may be considered with appropriate patient selection.

Laser Demarcation

Laser photocoagulation results in the formation of a chorioretinal scar, conferring enhanced adhesion between the neurosensory retina and RPE. Laser demarcation for RRD is applied to the area of attached retina immediately adjacent to the most posterior edge of the RRD, walling off the area of detachment and preventing further

progression. Two to three contiguous rows of laser are required, extending anteriorly up to the ora serrata on either side of the RRD. This is usually achieved using indirect laser ophthalmoscopy with scleral indentation. Increased adhesion at the site of laser photocoagulation is evident within 24 h in histological studies of animal and human eyes, and is twice normal by 2–3 weeks. Therefore, laser demarcation is only appropriate for RRD with signs of slow progression or recent stasis. Cryotherapy is not appropriate for RRD demarcation as a large area of retinopexy is required, causing more extensive tissue destruction and is less precise than laser.

RRD due to retinal tear

As discussed previously, acute RRD due to retinal tears tends to progress rapidly and is therefore not usually suitable for laser demarcation. Possible exceptions are localized RRD from a single small tear, particularly if within inferior retina.

RRD due to retinal hole or dialysis

Patients with limited or no symptoms from RRD secondary to retinal hole or dialysis, with associated signs of chronicity, may be suitable candidates for laser demarcation. Exceptions are patients with significant symptoms, especially visual field loss, signs of recent rapid progression, or when the SRF has progressed to the major vascular arcades. Laser demarcation limits rather than alleviates RRD symptoms. Laser-induced chorioretinal scars in the posterior pole can expand up to 13% per year, potentially causing a symptomatic central scotoma, if this is close to the macula.

Pneumatic Retinopexy

Pneumatic retinopexy for RRD was introduced by Hilton and Grizzard in 1986 as a two-step outpatient procedure without conjunctival incision. This involves an intravitreal

gas injection to temporarily close the retinal break preventing further recruitment of SRF from liquefied vitreous, followed by retinopexy (laser or cryotherapy) to permanently seal the break. Postoperative posturing for a minimum of 5 days is required in all patients. Careful case selection is required as not all patients are suitable for pneumatic retinopexy.

This procedure is not suitable in the following cases:

1.multiple breaks spanning over 3 clock hours;

2.giant retinal breaks;

3.inferior breaks involving inferior 4 clock hours;

4.presence of PVR grade C; and

5.inability to maintain postoperative posturing.

RRD due to retinal tear

Pneumatic retinopexy is most suitable for RRD with a single superior tear situated in the superior 8 clock hours. Patients must be able to posture, and attend clinic regularly in the early postoperative period to monitor the outcome of treatment. Additional gas injection or further surgery with scleral buckling or vitrectomy may be required in the event of treatment failure.

RRD due to retinal hole or dialysis

Pneumatic retinopexy is not suited to either of the following pathologies: RRD due to retinal hole or dialysis. The presence of an attached vitreous in both retinal holes and dialyses increases the risk of complications of the procedure, including new retinal break and formation of multiple small gas bubbles (see the section titled ‘Complications’).

Scleral Buckling

The concept of scleral buckling was introduced by Custodis in 1949. Prior to the advent of noncontact wideangle viewing systems for vitrectomy, scleral buckling was the most commonly performed operation for RRD, and remains so in certain countries due to high primary success rates of over 90%. It is typically performed under general anesthesia, although local anesthesia may be used.

The basic principle of scleral bucking (Figure 5) is to close retinal breaks by indentation of the sclera, preventing further recruitment of fluid into the subretinal space. Scleral indentation is achieved by using a variety of explant materials sutured externally to the scleral surface. Encircling explants afford a permanent 360 indent, while segmental explants provide a localized indent, which supports the break for a period of several months until the indent fades. Segmental explants are preferred due to lower associated ocular comorbidity. Permanent break closure must be ensured by application of retinopexy, typically cryotherapy. Intraoperative drainage of SRF

may be performed either to aid more rapid resolution of the RRD, or to create space within the vitreous cavity for the scleral indent by substituting SRF volume for the volume of indent, thus preventing excessive IOP elevation. Drainage of SRF involves surgical penetration of the sclera and choroid to the subretinal space. Indications for drainage remain controversial as this converts scleral buckling surgery from an external to an internal procedure, with associated increased risk of intraoperative complications (see the section titled ‘Complications’).

RRD due to retinal tear

RRDs due to a single tear, or a cluster of tears spanning 2–3 clock hours at the same anteroposterior position, are suitable for scleral buckling. Scleral buckling should be considered in patients with inferior tears, particularly as these are more difficult to effectively tamponade with a vitrectomy and gas procedure. Relatively young phakic patients are ideal candidates for scleral buckling as accommodation is preserved and the risk of developing cataract is small. Scleral buckling is not suitable for patients with significant media opacity, large tears, GRTs, or posterior breaks.

RRD due to retinal hole or dialysis

Patients with retinal holes or dialyses respond well to scleral buckling, typically with segmental explants. Primary reattachment rates of up to 100% have been reported.

Vitrectomy

Machemer performed the first pars plana vitrectomy in 1971. Since then, significant progress has been made in our understanding of vitreoretinal pathology alongside improvements in microsurgical instrumentation and techniques. In the United Kingdom, vitrectomy is now the most commonly performed procedure for both simple and complex RRD.

Vitrectomy (Figure 6) is performed via three pars plana sclerostomy ports. Standard 20-gauge (0.9 mm) instruments are most widely used. The technique involves removal of vitreous with relief of vitreous traction at the site of retinal breaks, followed by intraocular drainage of SRF, application of retinopexy (cryotherapy or laser), and intravitreal injection of gas or silicone oil tamponade. Noncontact wide-angle viewing systems allow for superior depth of field, visualization, and break localization compared to scleral buckling. This is particularly useful in pseudophakic patients and those with media opacities.

RRD due to retinal tear

In both the UK and USA, vitrectomy has become the treatment of choice for RRD due to retinal tears as there is usually a preexisting PVD. This simplifies

vitrectomy surgery as induction of a PVD is not required and direct relief of vitreoretinal traction at the tear can be achieved. A PVD is present in approximately 75% of people aged over 65. Therefore, the majority of patients with retinal tears is already presbyopic, minimizing the consequences of postoperative cataract formation.

RRD due to retinal hole or dialysis

Vitrectomy is rarely indicated in RRD secondary to retinal holes or dialyses, which usually have an attached vitreous. Surgically inducing a PVD in these typically young patients is often difficult, adding an unnecessary layer of complexity with increased risk of complications, for example, iatrogenic break formation. In addition, in the case of a dialysis, PVD induction would convert the break into a GRT.

Outcomes

Pneumatic Retinopexy

Pneumatic retinopexy is less invasive than scleral buckling or vitrectomy, with less local tissue damage and inflammation. Primary success rate in a selected patient group has been shown to be noninferior to scleral buckling in a multicenter, randomized, controlled trial by Tornambe and co-workers. A recent comprehensive review of 4128 eyes reported in the literature over a 21-year period (1986–2007) showed that pneumatic retinopexy has an overall primary retinal attachment rate of 74.4%, lower than scleral buckling or vitrectomy, with a final reattachment rate of up to 96.1%. Aphakic or pseudophakic patients appear to do less well, with primary retinal reattachment rates of 41–67%.

Scleral Buckling

Primary retinal reattachment occurs in over 90% of cases with scleral buckling alone, although up to 100% has been reported in small series. In patients with RRD and an attached macula, 90% achieve visual acuity of 20/30 or better at 6 months, although reduction in two Snellen lines of visual acuity occur in 10% despite anatomical success. Causes include macular epiretinal membrane and cystoid macular edema. Patients, who have their macular detached preoperatively, achieve a visual acuity of 20/50 or better in 38–71% of cases. A recent study involving foveal optical coherence tomography (OCT) imaging, by one of the authors (DGC), showed that postoperative resolution of foveal SRF can be delayed by over 6 weeks in 55% of patients, and complete resolution can take over 12 months. These patients have worse visual outcomes.

Pars Plana Vitrectomy

Primary retinal reattachment rates with vitrectomy are similar to scleral buckling, ranging from 65% to 100% with a mean of 85%. Final reattachment rate is 96–99%. Median visual acuity is 20/30 in macula-attached RRD, and 20/40 in macula-detached RRD. In an OCT imaging study by the same author (DGC), delayed resorption of foveal SRF was less common following vitrectomy, occurring in 15% of patients.

Complications

Intraoperative and postoperative complications for each procedure are discussed below.

Pneumatic Retinopexy

The main intraoperative complication of pneumatic retinopexy is formation of multiple small intravitreal gas bubbles or fish eggs during injection of gas, rather than a single large bubble as desired. This can result in subretinal gas migration, inadequate break tamponade, and treatment failure. This is principally managed by postoperative posturing to allow the gas bubbles to coalesce.

Postoperatively, new retinal break formation occurs in up to 20% of patients, and is a significant cause of surgical failure. PVR occurs in approximately 5% of patients.

Scleral Buckling

Intraoperative complications include inadvertent scleral perforation during suture placement, or SRF-drainage- related problems. Consequences include hypotony, choroidal or subretinal hemorrhage (with or without subfoveal tracking of blood), and retinal incarceration.

Postoperative complications include glaucoma, anterior segment ischemia (following encirclement), diplopia secondary to ocular dysmotility, change in refractive error and astigmatism, explant extrusion/intrusion or infection, and PVR. Contact lens users should be forewarned about possible lens-wear intolerance due to an irregular tear film and ocular surface following surgery, particularly as many patients who undergo scleral buckling are young myopes.

Pars Plana Vitrectomy

Intraoperative complications include iatrogenic retinal breaks, vitreous and retinal incarceration into the sclerostomy ports, lens trauma, and suprachoroidal hemorrhage.

Postoperatively, cataract formation is common and occurs in 81% of patients by 6 months. Raised IOP in the early postoperative period is common, and there appears to be an associated risk of late secondary glaucoma. PVR complicates between 5% and 12% of RRDs. Endophthalmitis following 20-gauge vitrectomy is rare.

Treatment Failure

In approximately 1 in 100 patients, it is not possible to reattach the retina despite multiple surgical attempts resulting in loss of vision. Consequences include chronic uveitis, rubeosis iridis, glaucoma, hypotony, band keratopathy, leukocoria, phthisis bulbi, and chronic ocular discomfort which may be severe enough to necessitate evisceration or enucleation of the eye.

New Developments

Smaller 25-gauge vitrectomy instruments were first introduced in 1990. Since then, techniques have evolved to enable sutureless transconjunctival surgery, with the benefits of reduced surgical time, less postoperative inflammation, more rapid wound healing, and improved patient comfort. As yet, small-gauge sutureless vitrectomy has not replaced 20-gauge systems, principally due to concerns regarding higher rates of endophthalmitis. A large retrospective series of 8601 patients from the Wills Eye Hospital demonstrated a 12-fold higher incidence of endophthalmitis with 25-gauge compared to 20-gauge vitrectomy. A smaller difference has been supported by other studies. Until this concern is adequately addressed, 20-gauge systems with conjunctival peritomy and suturing of ports will remain the gold standard.

The need for gas tamponade and postoperative posturing following vitrectomy for RRD has recently been challenged. A small prospective study of 60 patients by Martı´nez-Castillo and co-workers demonstrated a 98.3% primary reattachment rate without tamponade following complete drainage of SRF and laser retinopexy. This has not yet been supported by larger studies.

PVR remains to be the most important cause of failure of RRD treatment. Novel prophylactic treatment with adjunctive intraoperative 5-fluorouracil and low- molecular-weight heparin showed promise with decreased incidence of PVR in high-risk cases. However, this has not been reflected in unselected RRD undergoing primary vitrectomy, and may in fact be detrimental to visual acuity outcomes in patients with macula-attached RRD. Reducing the risk of PVR formation remains an important area of research.

See also: Proliferative Vitreoretinopathy.

Further Reading

Aylward, G. W., Sullivan, P. M., and Vote, B. (2007). Vitreoretinal. Volume 1: Basic Techniques (DVD). Surrey: Eye Movies.

Chan, C. K., Lin, S. G., Nuthi, A. S., and Salib, D. M. (2008). Pneumatic retinopexy for the repair of retinal detachments:

A comprehensive review (1986–2007). Survey of Ophthalmology 53: 443–478.

Charteris, D. G. and Wong, D. (2007). The role of combined adjunctive 5-fluorouracil and low molecular weight heparin in proliferative vitreoretinopathy prevention. In: Kirchhof, B. and Wong, D. (eds.) Vitreo-Retinal Surgery, 1st edn., pp. 33–37. New York: Springer.

Kunimoto, D. Y. and Kaiser, R. S. (2007). Wills eye retina service. Incidence of endophthalmitis after 20and 25-gauge vitrectomy. Ophthalmology 114: 2133–2137.

Lincoff, H. and Gieser, R. (1971). Finding the retinal hole. Archives of Ophthalmology 85: 565–569.

Machemer, R., Aaberg, T. M., Freeman, M., et al. (1991). An updated classification of retinal detachment with proliferative

vitreoretinopathy. American Journal of Ophthalmology 112: 159–165.

Martı´nez-Castillo, V., Zapata, M. A., Boixadera, A., Fonollosa, A., and Garcı´a-Arumı´, J. (2007). Pars plana vitrectomy, laser retinopexy, and aqueous tamponade for pseudophakic rhegmatogenous retinal detachment. Ophthalmology 114: 297–302.

Peyman, G. A., Meffert, S. A., and Conway, M. (eds.) (2007). Vitreretinal Surgical Techniques, 2nd edn. London: Informa UK.

Ryan, S. (ed.) (2006). Retina, Vol. III: Surgical Retina, 4th edn., 3 vols. Philadelphia, PA: Elsevier.

Wilkinson, C. P. and Rice, T. A. (eds.) (1997). Michels Retinal Detachment, 2nd edn. St. Louis, MO: Mosby.

Glossary

Cilium – An organelle that projects from a cell and contains a defined array of microtubules. Electron microscope – A microscope that uses a particle beam of electrons to illuminate a specimen and create a highly magnified image. An electron

microscope has much greater resolving power than a light microscope because the wavelength of an electron is much smaller than that of visible light. Glycoconjugate – A class of carbohydrates covalently linked to proteins, lipids, and other types of molecules.

Mitochondrion – A cellular organelle that provides most of the chemical energy, in the form of adenosine triphosphate, for metabolism in eukaryotic cells.

Plasma membrane – A bilayer of lipid molecules that physically separates the interior of cells (i.e., cytoplasm) from the extracellular environment.

In this article, we focus on the organization of the photoreceptor inner and outer segments, with an emphasis on the structural similarities and differences between rods and cones. The process by which the photoreceptor outer segments replace their disk membranes; soma and synapse; and cilim, phototransduction, soluble protein dynamics, and the visual cycle are described elsewhere in this encyclopedia.

In the late nineteenth century, Schultze proposed an organizational framework for vertebrate photoreceptors that, today, has become a cornerstone of visual science. According to this duplicity theory, photoreceptors may be anatomically and functionally divided into two main groups: rods and cones. These terms emerged from early anatomical observations showing that the distal portions of photoreceptors, the so-called outer segments, are either cylindrical or conical in shape. However, it soon became clear that there were exceptions to the outer segment shape criterion, most notably in the foveas of some species where cone outer segments possess a distinct rod-like configuration. Therefore, with the passage of time, the definitions of rods and cones were relaxed somewhat to include the combined shapes of the outer and inner segments (Figure 1). More recently, light-sensitive ganglion cells without obvious outer segments have been identified.

Duplicity theory also stipulated that rods and cones may be distinguished functionally by the light levels to which they are tuned. Rods were regarded primarily as a

sensitivity mechanism; whereas cones were considered to be an acuity mechanism, and a prerequisite for color vision. The predominance of rods in nocturnal species and of cones in diurnal species provided compelling evidence in support of this generalization.

With the advent of electron microscopy in the latter half of the twentieth century, it became apparent that there were also ultrastructural differences between rods and cones. In longitudinal sections, rod outer segments are visualized as a stack of hundreds of free-floating disks that resemble a stack of coins. The disk stack in rods is enclosed by, and separated from, the cell’s plasma membrane, except at the base where new disks are formed (Figure 2, left). In cones, however, many and perhaps all of the disk membranes appear to be in continuity with the plasma membrane (Figure 2, right; Figure 4). This ultrastructural difference between rod and cone outer segments has been confirmed in a wide variety of vertebrates, and still remains a distinction with no known exceptions. In cross section, single-rod disks have a scalloped margin all of which are in alignment; in contrast, cone disks possess a single incisure that extends from the margin to the center of the disk.

The photoreceptor cell is the epitome of a specialized neuronal cell. It is dedicated to the absorption of light energy (photons) and transduction of that energy into an electrochemical signal that is transmitted throughout the retina and, ultimately, to the brain. The packing density of the inner and outer segments defines the spatial resolution of the retina. Where visual acuity is at a premium, the segments form a fine array, the density of which is limited only by the wave nature of light. Hence, in many retinas, the inner and outer segments are long, narrow structures that may be as small as 1 mm in diameter.

The photoreceptors form the outermost layer of the neural retina (Figure 3). Rod and cone outer segments are enveloped by microvilli (mv) that emanate from the apical surface of the retinal pigment epithelium (RPE). In rods, the outer segments abut the apical surface; but in many species, including humans, cone outer segments are slightly recessed from the apical surface (see Figure 1). A highly organized array of mv known as the cone sheath extends down from the apical RPE surface and ensheaths the cone outer segments. Cone outer and inner segments are also ensheathed by an extracellular matrix rich in glycoconjugates known as the cone matrix sheath (Figure 4). The inner segments lie between the outer segments and the photoreceptor cell nuclei. Broadly speaking, the inner segment provides the metabolic support for the outer segment.

RPE

*

ROS

COS

RIS

CIS

Figure 1 Light micrograph showing the photoreceptor cell layer in a monkey retina. RPE, retinal pigment epithelium; ROS, rod outer segment; RIS, rod inner segment; COS, cone outer segment; CIS, cone inner segment; *, cone sheath.

Figure 2 Electron micrographs illustrating the organization of disk membranes at the base of mammalian rod and cone outer segments. The only cytoplasmic link between inner and outer segments is through a connecting cilium. Arrowheads indicate growth points for the evagination of new disk membranes from the plasma membrane adjacent to the centric face of the connecting cilium. In cones, most disks retain a connection to the enclosing plasma membrane; whereas, in rods, only the most basal disks (below the asterisk) retain a plasma membrane connection. bb, basal body; c, ciliary process; cc, connecting cilium; CIS, cone inner segment; m, mitochondrion; RIS, rod inner segment.

The distal region of the inner segment, known as the ellipsoid, contains a high concentration of mitochondria that are excluded from the outer segment. The only cytoplasmic link between inner and outer segments is a connecting cilium (Figure 2). The proximal region of the inner segment, the myoid, is the primary site for the synthesis of proteins destined for the outer segment (Figure 3). Projecting from the lateral margin of the ellipsoid is an array of calycal processes containing longitudinally oriented, filamentous actin and myosin. In monkey cones, each ellipsoid contains an array of actin-containing cables that originate in the myoid region, converge as they course through the

RPE

Ellipsoid

segmentInner Myoid

OLM

Nu

Figure 3 Low power electron microscopic autoradiogram of squirrel photoreceptors, illustrating the ellipsoid and myoid regions, the outer segments, and their relationship to the RPE. The black dots over the tissue are developed silver grains that signify sites of incorporation of a sugar residue (fucose) that had been labeled with a radioactive isotope (tritium). mv, microvilli; Nu, nucleus; OLM, outer limiting membrane; RPE, retinal pigment epithelium.

ellipsoid, and terminate within the calycal processes that form a circumferential ring at the base of the outer segments (Figure 5).

Just proximal to the myoid is the outer limiting membrane (OLM) (Figure 3). In longitudinal sections, the OLM appears as a demarcation line between the inner and outer retina. In reality, it is not a membrane but a series of aligned adherens junctions between photoreceptor cells, and between photoreceptor cells and adjacent radial glial cells (i.e., Mueller cells), that are linked to the cells’ actin cytoskeleton. In lower vertebrates, the myoid region of the inner segment is quite labile, and is capable of expansion and contraction in response to changes in ambient lighting. These so-called retinomotor movements are considered to be a light-adaptive mechanism in many species of lower vertebrates.

Approximately 90% of the membrane protein in the rod disk and plasma membranes consists of the visual pigment opsin, a light-sensitive guanine nucleotide-binding protein

OS

IS

Figure 4 Longitudinal section of a human cone photoreceptor inner segment (IS) and outer segment (OS) labeled with a plant lectin (peanut agglutinin) conjugated to a fluorescent probe. The cone matrix sheath (CMS) (shown in yellow) is a distinct domain of the retinal interphotoreceptor matrix that is rich in glycoconjugates and envelops cone outer and inner segments. Modified from Hageman, G. S. and Johnson, L. V. (1991). Progress in Retina Research. In: Osborne, N. and Chader, J. (eds.) Structure, Composition, and Function of the Retinal Interphotoreceptor Matrix, Vol 10, Ch.9, p. 226.

(G-protein)-coupled receptor. Cone outer segments contain homologous visual pigment proteins, also known as photopsins or iodopsins. Numerous other proteins that participate in the phototransduction cascade are also present in the outer segments. Rod outer segments, especially from bovine and rodent retinas, have been used widely in biochemical studies of structure and function. The ciliary connection between the inner and outer segments is quite fragile, such that outer segments are readily broken off in solution by shaking retinas that have been detached from the adjacent RPE. Following purification over a density gradient, relatively pure biochemical preparations of outer segments can be obtained. With additional steps, the disk and plasma membranes of rod outer segments can be separated from each other, and have thus been shown to contain some notable differences in protein composition.

CP

Figure 5 Electron micrograph of a tangential section through a monkey cone outer segment. The calycal processes (small arrows) that project from the ellipsoid form a basket at the base of the cone outer segment. At one point, the rim of the cone disk and the outer plasma membrane appear to be in continuity (large arrow), signaling a potential growth point. cc, connecting cilium; CP, calycal processes.

For example, the light-dependent cyclic nucleotide-gated channel and Na+, K+, Ca2+ exchanger proteins are restricted to the plasma membrane, while the retinal degeneration slow (peripherin/rds), rod outer segment membrane protein 1 (rom-1), and the ATP-binding cassette family A 4 (ABCA4) transporter proteins are present only in disk membranes. Opsin is present in both membrane domains. The disk membranes also contain two distinctive domains. Opsin is a ubiquitous component of rod disk membranes, except at their rims where a complex of peripherin/rds and rom-1 enables the formation of a loop that gives the disk its characteristic bilamellar structure.

In development, the outer segments form by repeated outgrowths or evaginations of the distal plasma membrane of the connecting cilium. In a mature photoreceptor cell, new rod and cone disks are also formed from evaginations originating from the centric face of the connecting cilium (Figures 2 and 6). The mature outer segment remains connected to the inner segment by what corresponds to the transition zone of the cilium, and what has been referred to historically as the ciliary stalk or connecting cilium. In recent years, it has been recognized that, while the photoreceptor cilium is highly unusual with its attached