Ординатура / Офтальмология / Английские материалы / Retinal Vascular Disease_Joussen, Gardner, Kirchhof_2007

underlying disease etiology [104]. This can now be corroborated in situ by OCT imaging in which cystoid spaces appear as round, hyporeflective regions within the neurosensory retina (Fig. 12.3) [15]. These features typically occur in the outer retinal layers, but when they increase in size, they can span the entire thickness of the retina and extend to the internal limiting membrane (Fig. 12.4) [28]. Pseudophakic cystoid macular edema is a notable exception because hyporeflective spaces may occur in the inner

retina instead of the outer retinal layers [11]. Small cystoid spaces often have well-defined boundaries, while larger spaces may have somewhat diffuse boundaries [11, 28].

The distribution of cystoid spaces within the retina can be a clue to the underlying disease etiology. For example, cystoid spaces in CRVO tend to be spread evenly across the macula, whereas cystoid spaces in BRVO typically respect the horizontal raphe (Fig. 12.2) [55]. And although CME usually

12.3.4 Vitreomacular Traction Syndrome

Essentials

OCT has greatly improved the evaluation and description of the vitreomacular interface

Partial posterior vitreous detachments may be more common in patients with DME

Epiretinal membranes may be depicted in better detail with OCT

Measurement of vitreomacular traction may be useful in managing patients with retinal vascular diseases

Retinal vascular diseases are frequently associated with breakdown of the blood-retinal barrier. Fluorophotometry studies suggest that transudate from retinal vessels can migrate not only into the retina, but also into the vitreous cavity [85]. Early histologic and biochemical changes in the vitreous of patients with diabetes include elevated levels of early and advanced glycation end products (AGEs) and greater amounts of collagen cross-linking [85]. Stitt observed that AGEs may provide the substrate for the collagen cross-linking observed in these patients, and suggested that these cross-links could explain the vitreous changes observed clinically in these patients [96]. Changes in the vitreous of eyes with retinal vascular disease may also contribute to the development of macular edema, particularly cystoid macular edema, by providing a substrate for traction on the retina [83, 84]. Schepens described two types of traction: traction with a narrow vitreous strand, and traction from a broad vitreoretinal adhesion [83].

Biomicroscopically, relationships between the posterior hyaloid and the retina can be difficult to visualize, particularly in patients with broad adhe-

sions. OCT has dramatically improved the evaluation of this relationship, and increased the recognition of vitreomacular traction although it may be necessary to adjust the signal-to-noise ratio of com-

mercially available OCT instruments to accentuate II 12 the posterior hyaloid face [23, 51, 55]. In patients

with a completely detached posterior vitreous, the posterior hyaloid may be visible as a thin, hyperreflective signal anterior to and separate from the retina (Fig. 12.2) [11]. This signal will be absent in patients with a completely attached posterior vitreous. In patients with a partially attached vitreous, this thin hyperreflective membrane can be observed with broad or focal insertions to the retinal surface [105]. The resulting anteroposterior distortion of the retinal contour may produce a characteristic peaked retinal appearance (Fig. 12.7) or a focal neurosensory detachment.

The superiority of OCT in imaging the vitreoretinal interface led Gaucher to define a new staging system for vitreous separation in diabetic patients, combining OCT and clinical criteria: Stage 0 was defined as the absence of a posterior vitreous detachment (PVD), with no Weiss ring visible and the posterior hyaloid not visible on OCT; stage 1, as a perifoveolar PVD with foveolar attachment (posterior hyaloid partially attached to macula on at least one OCT scan); stage 2, as an incomplete PVD with residual attachment to the optic nerve (no Weiss ring but posterior hyaloid remains visible on all OCT scans but is detached from macula); and stage 3, as a complete PVD (Weiss ring seen on funduscopy) [23]. Applying this staging system, stage 1 PVDs were found to be more frequent in patients with DME (53 %) compared to those without edema (22 %), consistent with the presumed importance of vitreomacular traction in DME. Catier and colleagues looked at patients with macular edema of various etiologies and concluded that patients with DME were much more likely to have a partial PVD than in other causes of mac-

ular edema [11]. Of interest, however, Uchino’s group found this stage of PVD in more than half of normal patients in their study [105].

Epiretinal membranes (ERMs) are another form 12 II of preretinal traction that may be present in retinal vascular diseases and can be visualized with OCT. Typically, ERMs can be distinguished from the posterior hyaloid membrane by their thicker, more consistent appearance and higher reflectivity. They may be easier to detect when separated from the inner retina, but can also be recognized by secondary retinal distortion or disruption of the normal foveal depression [55]. OCT can be used to document the opacity and thickness of an epiretinal membrane, its distance from the inner retina, its effects on the underlying retina, such as distortion, edema, or neurosensory detachment, or a patient’s response to

treatment.

An advanced form of epiretinal traction, fibrovascular proliferation secondary to proliferative diabetic retinopathy, can lead to two distinct anatomic configurations: traction retinal detachment and tractional retinoschisis [36]. Traction retinal detachments are diagnosed by observing optically clear subretinal spaces in the presence of epiretinal proliferation. Retinoschisis, on the other hand, can be described as a separation of the neurosensory retina into two layers connected by bridging columnar bands of tissue without accompanying subretinal fluid. Differentiating between these two conditions may be of prognostic importance in choosing the best management option [36].

Description of the vitreoretinal interface by OCT has affected the management of patients with DME. For instance, early vitrectomy may be considered in patients with evidence of significant vitreomacular traction [51, 88]. Shah and coworkers studied the prognostic significance of vitreomacular traction in 33 patients undergoing vitrectomy in their prospective study [88]. They observed that patients with evi-

dence of clinical and/or OCT macular traction showed significant improvements in postoperative visual acuity compared with patients without traction. Yamada and coworkers performed a similar study comparing the prognostic significance of two types of partial PVDs which they distinguished by OCT: (1) incomplete V-shaped detachments, and (2) partial detachments temporal to the fovea but with nasal attachment [113]. They found that anatomic outcomes appeared to be more favorable in the former group.

Beyond its ability to confirm the release of vitreomacular traction forces after vitrectomy, OCT has also been used to evaluate the resolution of neurosensory detachments and to study vitreomacular relationships following YAG capsulotomy [11, 107].

12.3.5 Miscellaneous Findings

Essentials

Structures such as blood vessels, hemorrhages, neovascularization, and lipid often appear as hyperreflective interfaces with posterior shadowing

Posterior shadowing may confound automated retinal thickness calculations in these areas

Sparse sampling with conventional radialline scanning protocols makes detection of microaneurysms difficult

Blood vessels can be identified on OCT as circular or semicircular, hyperreflective or hyporeflective regions with varying degrees of posterior shadowing (Fig. 12.8). Retinal hemorrhages, particularly if thick, may also appear as areas of hyperreflectivity with posterior shadowing (Fig. 12.9). Retinal neovascularization can present as a focal area of hyperref-

lectivity on the retinal surface, but may be difficult to distinguish from a non-vascularized epiretinal membrane. Inner retinal ischemia may be evident on OCT as increased reflectivity in the nerve fiber layer and inner retinal layers (Fig. 12.10). Although retinal artery occlusions may have this appearance in the acute phase, they often progress to longer-term retinal atrophy (Fig. 12.11).

Traditionally, other features typical of retinal vascular diseases such as microaneurysms and intrare-

tinal microvascular abnormalities have been difficult to visualize on OCT [28]. This is likely due, at least in part, to the poor sampling density of the peripheral macula that is characteristic of conventional radialline scanning protocols. Next-generation OCT devices employing high-speed, Fourier-domain technology can overcome this limitation by significantly increasing the density of the scans in the macula, while also increasing the axial resolution of the image (Fig. 12.12).

Exudations of lipid and protein are common features of retinal vascular diseases. Due to their relative optical opacity, hard exudates, which accumulate in the retina, can be frequently identified on OCT images as focal areas of high optical reflectivity with dense shadowing that extends posteriorly (Fig. 12.13) [28, 70]. Although they occur most commonly in the outer plexiform layer, hard exudates can be observed in almost any retinal layer as well as in the subretinal space. One important consequence of the posterior shadowing caused by these and other structures is obscuration of the posterior retinal boundary which, in turn, can confound calculations of the retinal thickness by naive OCT algorithms [28].

12.4 Management

OCT has many potential applications in the management of patients with retinal vascular disease, particularly in patients with macular edema. Quantification of the extent of retinal thickening by OCT can assist in the selection of patients who may benefit from treatment, identify which treatment is indicated, guide the application of the treatment, and allow precise monitoring of the response to treatment [28,

70]. The morphologic detail available in OCT scans can also be used to define classification or staging systems for retinal disease which, in turn, can guide management. Panozo et al. devised a classification system for DME which takes into account five parameters: retinal thickness, diffusion, volume, morphology, and presence of vitreous traction [71]. Applying this classification system, in a retrospective study of 169 patients undergoing one of three treatments (laser, vitrectomy, or intravitreal triamcinolone) for diabetic macular edema, patients with earlier OCT stages (e.g., diffuse edema without subretinal fluid) were observed to have better outcomes compared to those with more advanced stages [72]. OCT findings may also have prognostic implications which can affect treatment decisions. For instance, in patients with diabetic retinopathy and good vision (20/20) undergoing panretinal photocoagulation (PRP), patients with preoperative parafoveal thickness higher than 300 μm had a higher incidence of persistent decreased vision after PRP [92]. This observation may help identify a group of patients who may benefit from a less aggressive laser strategy, with a staged delivery of treatment and a longer interval between PRP sessions.

12.4.1 Focal and Panretinal Photocoagulation

Essentials

OCT does not yet replace the clinical definitions of CSME

It may be possible to exclude some patients with clinical foveal edema from unnecessary laser treatment by demonstrating normal foveal retinal thickness by OCT

Focal laser can cause a significant reduction in OCT-measured retinal thickness

PRP may cause a transient increase in macular thickening by OCT

The Early Treatment Diabetic Retinopathy Study (ETDRS) demonstrated that focal laser photocoagulation could reduce the risk of moderate vision loss (defined as a doubling of the visual angle) in patients with CSME diagnosed by contact lens biomicroscopy and stereoscopic fundus photography [18]. As described in Sect. 12.2.2, correlating OCT measurements with existing definitions of CSME has proven difficult due to the limitations in existing scanning protocols and analysis algorithms, which are designed to provide a dense sampling of the fovea, but rely on extensive interpolation in the extrafoveal areas that may be involved by CSME. The development of grid scanning protocols for the Stratus OCT [80] and dense scanning patterns facilitated by FDOCT may eventually address this limitation for nonfoveal zones. In the interim, many clinicians have utilized OCT for assessment of foveal edema not detected by the clinician [5, 8]. It is uncertain, however, whether patients with OCT-detected “subclinical edema” are at risk for moderate vision loss or whether these patients would benefit from focal laser therapy. Brown et al. deemed that “it does not seem reasonable at this time to extrapolate the results of the Early Treatment Diabetic Retinopathy Study to cases of subclinical edema” [5]. OCT may be of some value, however, in excluding patients who in fact do not have significant foveal thickening (i.e., a falsepositive clinical exam), thereby averting the risks of treatment toxicities such as paracentral scotoma or choroidal neovascularization (CNV) [8].

Although the ultimate utility of OCT in the identification of candidates for focal laser treatment is still uncertain, OCT has been used in studies evaluating the anatomic effects on the retina following photocoagulation [44]. In a series by Rivellesse et al., patients undergoing focal laser for diabetic CSME had a significant reduction in retinal thickness after therapy [71]. The anatomic effects of laser on macular thickness have also been studied in patients receiving PRP for

proliferative diabetic retinopathy. Shimura et al. evaluated 72 eyes of 36 patients undergoing PRP treatment spread over 4 sessions [90]. Patients were treated weekly in one eye and biweekly in the other. Macular

thickness by OCT was observed to transiently II 12 increase in both eyes, but to a greater degree in weekly

treated eyes. Recovery of macular thickening to normal levels was also faster in biweekly treated eyes, but no statistically significant difference in visual outcomes was observed between the two groups. Although this was a small series, the observations are consistent with previous clinical observations in the ETDRS that showed that macular edema may worsen after PRP [19]. The OCT findings may also explain the apparent reduction in macular function on multifocal electroretinography (mfERG) in the absence of a change in visual acuity observed by Lovestam-Adrian [46]. Although Lovestam did not observe changes on OCT, OCTs were not obtained in the immediate perioperative period, but only 6 months after PRP.

12.4.2 Pharmacologic Therapy

Essentials

OCT is useful in monitoring the response to pharmacologic therapies for retinal vascular disease

Changes in OCT appear to correlate well with changes in angiographic leakage

OCT retinal thickness parameters are becoming important as entry criteria and outcome variables in clinical trials for retinal vascular disease

Longitudinal changes in OCT are useful for guiding retreatment decisions

The development of pharmacotherapies for retinal vascular disease has been a major focus of clinical research since the discovery of the role of vascular endothelial growth factor (VEGF) in disease pathophysiology. Anatomic improvements in retinal thickness on OCT have been an important variable in studies evaluating the effectiveness of these agents.

Although it likely has multiple mechanisms of action in addition to its anti-VEGF effect, intravitreal triamcinolone acetonide (IVTA) has been widely used for the treatment of various retinal vascular diseases. A number of small studies have demonstrated a reduction in central retinal thickening (foveal center thickness or average foveal subfield thickness) following a single intravitreal injection of triamcinolone in patients with macular edema associated with diabetes [14, 37, 42, 47, 52] and retinal venous occlusive disease [108]. Follow-up was variable in these studies, but a

50 % reduction in central thickening on OCT was a typical response by 3 months after treatment [47]. In some series, the reduction in retinal thickness correlated with reduced leakage on angiography and an

12 II improvement in visual acuity [53]. OCT has also been used to document the time course of retinal thickness changes following triamcinolone injection, and has demonstrated that macular edema tends to recur after 3 months [99]. Recurrence of edema as identified by OCT is a useful criterion for clinicians when considering potential retreatment. Large-scale randomized clinical trials are currently in progress (SCORE, DRCR) to evaluate the efficacy of IVTA for patients with macular edema from retinal vascular disease. Retinal thickness measurements from OCT are being used as outcome variables as well as criteria for enrollment and retreatment in these studies.

Studies using OCT findings as outcome variables are currently underway for other anti-angiogenic therapies (including pegaptanib, ranibizumab, and bevacizumab) used for retinal vascular disease. Rosenfeld reported an improvement from 20/200 to 20/50 with resolution of CME in one patient with a CRVO just 1 week after an intravitreal injection of 1.0 mg of bevacizumab (Avastin) [79]. A randomized, sham-controlled study of 172 patients with DME (two-thirds of whom received treatment) revealed a mean reduction of OCT central retinal thickness of 68 μm and a modest visual benefit in patients receiving a 0.3 mg dose of pegaptanib [16].

12.4.3 Surgery

Surgical approaches for retinal vascular diseases have gained increasing popularity and OCT findings have proven to be useful not only in monitoring the response to therapy, but also in selecting suitable candidates for surgery.

12.4.3.1Pars Plana Vitrectomy for Macular Edema

Essentials

Detection of vitreomacular traction by OCT aids in the selection of patients for vitrectomy

Postoperative OCT is used to confirm normalization of retinal morphology following surgery

Diffuse diabetic macular edema on OCT, even in the absence of traction, may benefit from vitrectomy

OCT-detected neurosensory detachment may predict a higher risk of postoperative subfoveal lipid exudates

Pars plana vitrectomy has been advocated as a potential treatment for patients with refractory DME, particularly in patients with suspected VMT. OCT has revolutionized the selection of patients for vitrectomy by providing definitive evidence of tractional effects on the macula. Giovannini et al. identified two patterns of maculopathy in patients with VMT: a thickening of the superior profile of the OCT tomogram, or the disappearance and inversion of the physiologic foveal depression [24]. Following surgery, many of the eyes in this series showed a normalization of the retinal morphology. Additional retrospective series have demonstrated similar results with surgery, showing varying but significant reductions in OCT foveal center retinal thickness following vitrectomy [64, 78, 97]. Several studies have also shown functional improvements [24, 73, 78] in visual acuity and multifocal electroretinography [101, 102] in conjunction with the apparent anatomic benefits of reduction in retinal edema evident on OCT.

Other investigators have advocated the use of PPV for diffuse refractory DME, even in the absence of OCT evidence of macular traction [73, 100]. Patel and coworkers observed a reduction in OCT retinal thickness in patients undergoing PPV for diffuse DME, including a small subgroup of patients who had diffuse low-lying elevation of the retina and no vitreous traction [73]. Despite these favorable anatomic results demonstrated by OCT, the true benefit and role of PPV in diffuse DME remains uncertain. In a randomized trial of vitrectomy versus additional focal laser in patients with persistent (following previous laser) diabetic macular edema without vitreous traction, surgery was not shown to be better than laser in terms of anatomic (macular thickness) or functional (visual acuity) outcomes [100].

In addition to the lack of convincing randomized clinical trial data to support its use, PPV for DME is not without significant risk. Although reduced OCT retinal thickness was observed, Yamamoto and colleagues observed a variety of intraand postoperative complications including retinal tears, retinal detachment, vitreous hemorrhage, neovascular glaucoma, epiretinal membrane formation, and lamellar macular hole [115]. Another complication following surgery that was observed by Yamamoto as well as by Otani was the accumulation of hard exudates in the subretinal space, a finding typically associated with a worse visual prognosis [65]. Otani also noticed that the presence of a serous neurosensory detachment on OCT before or after surgery was associated with an increased tendency toward the formation of these subfoveal exudates.

The use of OCT to quantify reduction in macular edema following vitrectomy has not been restricted to patients with diabetic retinopathy. Sekiryu and

coworkers demonstrated dramatic reductions in OCT retinal thickness in a small series of patients undergoing PPV for massive foveal cystoid macular edema associated with CRVOs [86].

12.4.3.2Radial Optic Neurotomy for Central Retinal Vein Occlusions

Essentials

OCT can demonstrate reduction in macular edema following radial optic neurotomy

Dramatic reductions in edema can be observed on OCT within a few months of surgery

Morphologic response on OCT often does not correlate with visual improvement Retinal ischemia may limit the therapeutic benefit

Ompremcak and coworkers have advocated the use of radial optic neurotomy (RON) for treating patients with persistent macular edema associated with central retinal venous occlusive disease [62, 63]. The procedure involves the use of a blade to make an incision at the nasal edge of the optic nerve. Although the mechanism of action is unknown, many investigators believe that this procedure may increase the development of collateral vessels that serve to decompress the venous bed [20, 22, 94, 103]. In a series of 117 consecutive patients with CRVO and severe visual loss, Ompremcak and coworkers observed anatomical resolution in 95 % and visual improvement in 71 % of patients [63].

OCT has been a useful tool in monitoring the therapeutic response to RON as well as observing postoperative complications such as a macular hole [93]. In several small series, dramatic reductions in OCT retinal thickness were observed by several months after surgery [48, 57, 74]. These beneficial anatomic results, however, often did not correlate with improvements in visual acuity, particularly in patients with ischemic occlusions [48].

12.4.3.3Arteriovenous Sheathotomy for Branch Retinal Vein Occlusion

Essentials

Manipulation of the arteriovenous sheath (sheathotomy) may relieve persistent macular edema in BRVO

Sheathotomy may relieve venous compression and mobilize thrombus

Several investigators have also advocated the use of arteriovenous sheathotomy for treating persistent macular edema in patients with BRVO [61, 87]. Supporters of this approach believe that manipulation of the common adventitial sheath shared by the retinal artery and vein at arteriovenous crossing points can relieve the compression of the venous lumen and mobilize the presumed thrombus. Rapid reperfusion of the retinal circulation, including at the time of surgery, has been demonstrated by these investigators. Fujii et al. demonstrated a rapid, dramatic, and sustained reduction (from 450 to 228 μm) in retinal edema by OCT and an improvement in visual acuity, beginning within 1 day after the sheathotomy in a patient with a BRVO [21].

12.5 Future Directions

Despite the tremendous advances in hardware design over the last decade, conventional OCT devices acquire a relatively limited data set and use outdated software fraught with imperfections. Discoveries in the realm of OCT hardware have far outpaced advancements in OCT software. The clinical utility and success of the next generation OCT devices are dependent on careful attention to both hardware and software optimization. Future OCT devices will also likely have broad scale applicability, not only in ophthalmology, but also in many other fields of medicine including cardiology, gastroenterology, oncology, orthopedics, otolaryngology and possibly urology.

12.5.1 Current OCT Limitations

Essentials

Scan acquisition speed and fixation instability can compromise OCT thickness measurements, especially in the foveal center

Low sampling density of existing radial line scan patterns requires interpolation of data over 95 % of the area being mapped

The outer retinal boundary is consistently detected incorrectly by current StratusOCT algorithms

Moderate and severe errors in retinal boundary detection are frequent, especially in cases with subretinal disease

Although OCT technology has spearheaded a quantitative revolution in the diagnosis and management of retinal disease, there are important hardware and software limitations of which the clinician must be aware when interpreting OCT studies. An important limitation of current StratusOCT hardware is the scan acquisition speed. Each “high-resolution” 6-mm macular radial line scan composed of 512 A- scans takes more than 1 s to be acquired. Consequently, the accuracy of the line scan may be affected by eye movements, particularly in patients with macular diseases associated with poor fixation. Fixation instability also hampers registration and correlation of OCT line scans with other imaging modalities such as color fundus photographs or fluorescein angiograms. Although an infrared image of the fundus indicating the approximate location of the line scan is obtained at the end of every line scan, the precise fundus location of the scan is not provided. This limitation is illustrated by Fig. 12.14 in which the infrared image suggests that the line scan passes through the center of an apparent full-thickness macular hole, but the corresponding OCT line scan actually passes through an adjacent area of retina that fails to include the macular hole.

The limitation presented by slow scan acquisition speed is further accentuated when multiple radial line scans are obtained to generate a retinal thickness map. Fixation instability results in misalignment of the centerpoints of the line scans and an inaccurate center retinal thickness. To address this problem, faster scanning protocols such as the Fast Macular Thickness Scan have been developed, but these sacrifice transverse resolution (only 128 A-scans per 6- mm radial line) for higher scanning speed. Slow scan speeds also limit the number of line scans that can be obtained for constructing macular thickness maps, resulting in the need for significant interpolation between sample points (particularly in the peripheral macula). In fact, typical radial-line protocols mea-

Fig. 12.15. Radial line scanning protocols (white lines) cover only a small portion (< 5 % shown as red lines on inset map) of the retinal area being mapped

sure less than 5 % of the retinal area being mapped (Fig. 12.15). This translates into interpolated values covering 95 % of the area being measured. Use of alternative scan patterns such as concentric circles rather than radial lines (Fig. 12.16) may provide more even sampling of the retina and reduce interpolation errors, but is still subject to registration errors related to the slow scanning speed of timedomain OCT.

C-scan OCT has been proposed as a potential hardware solution to the registration problems inherent in StratusOCT. The images obtained by C- scan OCT, however, present retinal anatomy in an orientation that is unfamiliar to most clinicians, thereby limiting the usefulness of this approach.

The StratusOCT analysis software also has significant shortcomings. For example, the generation of a