- Email: [email protected]
Fundus photographic, fluorescein angiographic, and indocyanine green angiographic signs in successful laser chorioretinal venous anastomosis for central retinal vein occlusion1
Fundus Photographic, Fluorescein Angiographic, and Indocyanine Green Angiographic Signs in Successful Laser Chorioretinal Venous Anastomosis for Central Retinal Vein Occlusion David J. Browning, MD, PhD Objective: To describe the fundus signs and angiographic signs that accompany development of a laserinduced chorioretinal venous anastomosis in central retinal vein occlusion and to describe the chronology of the signs. Design: Noncomparative, consecutive case series. Participants: Fifteen eyes of 15 patients were treated. Intervention: The argon laser was used in the original method of McAllister and Constable to form an anastomosis in five eyes, and the modified method of McAllister involving the argon laser followed by the YAG laser was used for ten eyes. Main Outcome Measures: Changes in vessel diameters, retinal blood flow, and morphology of anastomosis over time as documented photographically and angiographically. Results: The earliest fluorescein angiographic sign of success is a hyperfluorescent spindle at 1 week. The earliest indocyanine green angiographic sign is direct connection of retinal venous and choroidal venous circulations at 2 weeks. The earliest fundus photographic and, hence, ophthalmoscopic sign is asymmetry in venous diameter at the disc at 3 weeks. No sign is present in all successful cases. The most commonly observed sign is fluorescein flow around a corner in a retrograde direction toward the anastomosis in 80% of cases. Drainage of only a fraction of the retina occurred in 93% of cases. Fifteen eyes with successful anastomoses had mean improvement of 2.3 ⫾ 2.4 (standard deviation [SD]) Snellen lines of best-corrected visual acuity compared to 0.2 ⫾ 2.3 (SD) lines for 9 eyes with unsuccessful anastomoses (P ⫽ 0.0439). Conclusion: Recognition of the variety and typical chronology of postoperative fundus and angiographic signs in laser-induced chorioretinal anastomosis will help prevent premature retreatment and guide appropriately timed additional treatment for failed initial attempts. Fluorescein angiography and indocyanine green angiography are necessary components of intensive postoperative follow-up of these patients. The follow-up care is more difficult than the technical aspects of the surgery itself. Successful anastomoses help by taking part of the flow away from the compromised central vein, not by providing global venous bypass. This technique remains controversial, unproven, and in need of a randomized clinical trial to determine its role in the management of nonischemic central retinal vein occlusion. Ophthalmology 1999;106:2261–2268 The technique of laser-induced chorioretinal venous anastomosis for central retinal vein occlusion was invented by
Originally received: November 10, 1998. Revision accepted: July 27, 1999. Manuscript no. 98577. From Charlotte Eye, Ear, Nose, and Throat Associates, Charlotte, North Carolina. Presented in part at the American Academy of Ophthalmology annual meeting, New Orleans, Louisiana, November 1998. The author has no proprietary interest in any of the materials used in this study. Reprint requests to David J. Browning, MD, PhD, Charlotte Eye, Ear, Nose, and Throat Associates, 1600 East Third Street, Charlotte, NC 28204, E-mail: [email protected].
McAllister and Constable1 to provide an alternate drainage route for retinal venous blood. The technique has produced successful anastomoses in variable fractions of treated patients,1–3 and the method has been in a state of evolution as the initial technique using argon laser has been superseded by a serial technique using higher power argon laser followed by yttrium–aluminum– garnet (YAG) laser.4 There have been incidental remarks in publications regarding the pattern of blood flow and the fundus appearance in successful cases, but a detailed study of the fundus signs and blood flow characteristics of successful cases is lacking. Such a study is needed to provide insight as to how the method works and to guide the ophthalmologist as to the need for and the timing of retreatment.
2261
Ophthalmology Volume 106, Number 12, December 1999
Figure 1. Narrow parent trunk sign. Color fundus photograph showing anFigure 3. Fluorescein angiogram showing flow of blood from the disc toward the anastomosis. As more tributaries empty into the vein and carry example of daughter venous branches (filled arrows) having greater diameter than the parent trunk (large arrowhead). The photograph also showsblood distally, the venous diameter increases as in Figure 2. Note the an instance in which the venous segment connected to the anastomosis venous tributary (light arrow) empties into the inferior hemicentral vein, and proximal to it (open arrow) is thinner than the segment distal to itwith abrupt increase in the vein diameter distal to the intersection, implying blood flow away from the disc, not toward it. Blood flows toward (small arrowhead). The disc venous asymmetry sign is also illustrated in which the superior vein stemming from the central vein has a greaterthe anastomosis (triangle). diameter than the inferior vein. At the same time, dilated collaterals are present on the superior disc margin (e.g., at the 1:30- and 2:30-o’clock and two of these cases have been reported.3 The subsequent ten meridians) but not on the inferior disc margin. 4
Methods This study reports on fundus photographs, sequential fluorescein angiograms, and indocyanine green angiograms in 15 consecutive cases of successful laser-induced chorioretinal venous anastomosis for central or hemicentral retinal vein occlusion treated by 1 retina specialist in a private practice. These cases are drawn from a larger population of 24 consecutive cases treated for this condition, of which 9 were unsuccessful. The time period for case accrual was June 20, 1995, through June 9, 1998. Twenty-two of the patients were treated by the author and 2 by Andrew Antoszyk, MD (coauthor of ref. 3). The first five successful patients were treated in the originally described technique of McAllister and Constable,1
cases were treated in a modified technique described later. In this technique, the argon laser is used to rupture Bruch membrane adjacent to a retinal vein chosen preferably in the inferior fundus, usually 2- to 4-disc diameters away from the optic disc. Higher powers than in the original technique are chosen, from 2.5 to 4.5 W, depending on the degree of lens opacity. The number of argon burns ranged from 1 to 13 depending on the number of anastomoses attempted. Then, the patient was treated with the YAG laser. Treatment is directed to the edge of the retinal vein adjacent to the argon burn. The YAG laser powers used ranged from 2.5 to 5.2 mJ depending on lens opacity. The number of YAG applications ranged from 1 to 15, until retinal venous bleeding occurred. Color fundus photographs and fluorescein angiograms were obtained
Figure 4. An example of the venous asymmetry sign. This eye has a single central vein, yet there is venous diameter asymmetry as the branches enter Figure 2. Reverse tapering sign. Fundus photograph showing increased the disc. The superior branch (large arrowhead) has a larger diameter than diameter of a retinal vein as one traverses it heading away from the disc,the inferior branch (small arrowhead). The successful anastomosis is a reversal of the usual pattern. The diameter at the arrowhead is smallerlocated in the inferior fundus and is not seen in this view. The daughter than the diameter at the arrow. The corresponding fluorescein angiogramvenous branches are thicker than the parent venous trunk at the arrow, a sign of blood flow away from the disc and out the anastomosis. explaining the appearance is shown in Figure 3.
2262
Browning 䡠 Laser Chorioretinal Anastomosis before surgery in all patients and at 2- to 4-week intervals in the postoperative period until the anastomosis was functioning. Indocyanine green angiograms were obtained in 9 of 15 patients when this method became available to us. All analysis of the photographs and angiograms was performed by the author. Informed consent was obtained from all patients.
Results To help in orienting the reader, results will be presented in a parallel manner for the three methods of examination: fundus photography, fluorescein angiography, and indocyanine green angiography. For each of these methods, new signs will first be described and illustrated. The relative frequency of the new and previously reported signs will be reported next. Finally, the temporal sequence of development of the signs will be reported.
Fundus Photography Previously described and illustrated signs include asymmetric disc collaterals,3 asymmetric vein diameters at the disc,3 a thinner venous segment proximal to the anastomosis,2,3,5 a thicker venous segment proximal to the anastomosis,1,2 enlargement of a draining choroidal vein over time,2 and a venous loop at the anastomosis site.2,3 In addition to the previously documented signs, we observed three previously undescribed fundus signs. The first we have termed the narrow parent trunk sign (Fig 1). It occurs at the next retinal venous branching more proximal than the anastomosis. In this sign, the daughter veins at the involved branching have larger diameters than their parent trunk. This appearance correlates with the appearance in the fluorescein or indocyanine green angiogram of dye flow from one daughter branch to the other and not toward the parent trunk as expected (see Fig 15 for an example of this). This finding suggests more blood flow through the daughter branches than through the parent trunk. The second sign has been termed the reverse tapering sign and occurs at the venous segment at the optic disc, which leads toward the anastomosis (Fig 2). In this sign, one observes that the diameter of the venous segment increases as one traverses the vein distally, a reversal of the usual pattern of venous narrowing. This correlates with tributary venules emptying into the parent vein with the flow directed peripherally out the anastomosis rather than centrally out the disc (Fig 3). The third new sign is termed the disc venous asymmetry sign. Although previously published photographs have shown asymmetry in the thickness of retinal veins at the disc, these cases have reflected hemicentral retinal venous anatomy.3 We have observed that this fundus sign can be associated with an unequivocal single central vein as shown in Figure 4. The reason for greater venous engorgement superiorly in a situation in which the superior and inferior hemicentral veins visibly join is unclear. No fundus photographic sign occurred in every successful anastomosis. Table 1 lists the relative frequency of various signs. The most frequently observed sign is an abrupt increase in the thickness of the more distal venous segment attached to the anastomosis compared to the central segment attached to the anastomosis. This correlates with flow of blood entering the anastomosis from two directions, with a presumed greater volume of blood coming through the distally attached venous segment than the proximally attached segment. The chronology of first appearances of various fundus photographic signs is listed in Table 2. The earliest of these signs is asymmetry of disc collateral vessels and of veins as they pass into the disc. These earliest signs can appear as early as 3 weeks after
Table 1. Relative Frequency of Observed Signs in Eyes with Successful Laser-induced Chorioretinal Venous Anastomosis* No. (%) with Sign Fundus photographic signs Venous segment proximal to anastomosis is thinner Asymmetric vein diameters at the disc Narrow parent trunk (daughter venous branches larger than parent trunk) Reverse tapering (increasing venous diameter further from disc) Venous segment proximal to anastomosis is thicker Venous loop at anastomosis site Fluorescein angiographic signs Flow around the corner Gap Abrupt hyperfluorescence Trilaminar flow Retrograde venous flow between disc and anastomosis Spindle Late wall staining Indocyanine green angiographic signs Connection of retinal vein and choroidal vein Abrupt hyperfluorescence Enlargement of draining choroidal vein over time
11 (73) 9 (60) 6 (40) 6 (40) 4 (27) 2 (13) 12 (80) 7 (47) 5 (33) 3 (20) 3 (20) 3 (20) 2 (13) 4 (44) 4 (44) 2 (22)
* The denominator in the fundus photographic sign and fluorescein angiographic sign sections is 15; in the indocyanine green angiographic section it is 9.
laser treatment. A relatively late sign is a venous loop at the anastomosis. The development of a venous loop over time is shown in Figures 5 and 6. The finding correlates with presumed increasing blood flow through the anastomosis. As this example illustrates, sequential comparison of photographs is informative regarding a developing anastomosis. In another such example displaying a different facet of shunt development, Figures 7 and 8 show progressive thinning of a proximal venous trunk as the anastomosis matures and diverts more venous blood away from the disc.
Fluorescein Angiography Previously described and illustrated signs include trilaminar flow,1 retrograde venous flow between disc and anastomosis,1 and nondrainage of the entire retina by virtue of a hemicentral retinal vein.3 Previously unreported signs were discovered. The first of these is a hyperfluorescent spindle-shaped lesion occurring at the anastomosis site. It is the earliest sign of a successful anastomosis, occurring 1 to 2 weeks after the laser treatment. It can be followed within 2 weeks by neovascularization, which arises from the retinal vein or from the choroid (Figs 9 and 10). Because of the appearance of the lesion, it has been called the spindle sign.6 A later sign is the appearance of a gap in the dye stream occurring between the first retinal venous branching more proximal than the anastomosis and the next retinal venous branch closer to the disc (Fig 11). This gap shows that flow occurs out the anastomosis rather than toward the disc for blood arriving at the involved venous branching. It also shows that blood flow is not occurring across the disc toward the anastomosis from sections of the fundus separated from the anastomosis by the disc. Instead, it shows slow flow of blood out the central vein for part of the fundus more proximal than the anastomosis to the disc.
2263
Ophthalmology Volume 106, Number 12, December 1999 Table 2. Relative Chronology of Observed Signs in Eyes with Successful Laser-induced Chorioretinal Venous Anastomosis* Time of Earliest Appearance (wks) Fundus photographic signs Asymmetric collateral disc vessels Asymmetric vein diameters at the disc Venous segment proximal to anastomosis is thicker Venous segment proximal to anastomosis is thinner Narrow parent trunk (daughter venous branches larger than parent trunk) Reverse tapering (increasing venous diameter further from disc) Venous loop at anastomosis site Fluorescein angiographic signs Spindle Abrupt hyperfluorescence Retrograde venous flow between disc and anastomosis Gap Flow around the corner Trilaminar flow Late wall staining Indocyanine green angiographic signs Connection of retinal vein and choroidal vein Abrupt hyperfluorescence Enlargement of draining choroidal vein over time
3 3 3 4 5 7 1 5 5 5 5 9 24 2 6 12
* The denominator in the fundus photographic sign and fluorescein angiographic sign sections is 15; in the indocyanine green angiographic section, it is 9.
A sign seen in a well-developed anastomosis is a sharp change in hyperfluorescence at the anastomosis. This sign suggests a relative discontinuity in the thickness of the fluorescein column occurring at an anastomosis site (Fig 12); this interpretation suggests the name abrupt fluorescence change sign. A rare sign is the increased staining of the venous walls of an underperfused segment of the vein connecting the disc to the
Figure 5. Red-free photograph showing anastomosis site before significant flow through the shunt (arrow). Notice that the inferior hemicentral vein is more distended (arrowhead) at this time than in Figure 6, when the shunt is working, decompressing the venous drainage out the disc.
2264
Figure 6. Red-free photograph of same eye as shown in Figure 5 several weeks later. Now, flow through the shunt is greater and a venous loop 6 (arrow) has developed. The inferior hemicentral vein is less distended than in Figure 5 (arrowhead).
anastomosis. This sign may reflect relative ischemia of this venous segment. It correlates with a smaller than normal diameter for such a proximal part of the venous tree (Figs 13 and 14). The relative frequency of fluorescein angiographic signs is shown in Table 1. The flow around the corner sign is most commonly seen (80% of cases), whereas the more publicized trilaminar flow sign is only a fourth as commonly observed. The spindle sign is an observed sign seen earliest (Table 2). The trilaminar flow sign occurs late. It was seen earliest at 9 weeks after laser surgery, when relatively large volumes of blood drain through a more developed shunt. Other listed signs occur at intermediate intervals.
Indocyanine Green Angiography One previously reported ICG sign is a visualized connection of the retinal vein and the choroidal vein.2 This sign, shown in Figure 15, is particularly useful because of its early appearance at 2 to 3 weeks after laser treatment (Table 2). One other useful sign, not described previously, is a sharp change in hyperfluorescence of the
Figure 7. Red-free photograph showing greater thickness of proximal venous segment (arrow) before the anastomosis site has begun to function (arrowhead); compare to Figure 8.
Browning 䡠 Laser Chorioretinal Anastomosis
Figure 8. Red-free photograph of same eye as in Figure 7 several weeks later showing decreased thickness of the proximal venous segment (arrow) now that the shunt is working (arrowhead); compare to Figure 7.
dye column at the anastomosis. The appearance and probable explanation are likely analogous to the similar observation in the fluorescein angiogram (Fig 16). This is a relatively late sign, seen at 6 weeks or later, and seems to require relatively high flow through the shunt to be seen. The initial description of enlargement of a draining choroidal vein over time was made from fundus photographs,2 but this is only possible in an eye with relatively blonde coloration. Indocyanine green angiography reveals the finding even in eyes with darker fundi, in which it was seen in two (22%) of nine cases and only as a late sign (12 weeks at earliest).
Evolution, Concordance, and Mutual Exclusivity of Photographic Signs An eye can evolve from one set of signs to a later distinct set. An example of this is shown in Figures 17 and 18. In this eye at 9 weeks after laser treatment, an anastomosis has formed in which the only venous input is distal to the anastomosis (Fig 17). How-
Figure 9. An example of the spindle sign. The hyperfluorescent, spindleshaped structures at two anastomosis sites are seen 8 days after laser treatment (arrows).
Figure 10. Same patient as in Figure 9 12 days after the fluorescein angiogram in Figure 9 and 20 days after laser treatment. The spindle structures have evolved into neovascular fronds. In this case, the blood supply was choroidal, but in other cases it can be from the retinal circulation.
ever, by 16 weeks after surgery, a prominent proximal input to the anastomosis has developed, as shown in Figure 18. The photographic signs cluster in groups. The gap sign is frequently seen in cases showing the flow around the corner sign. Retrograde flow from disc to anastomosis is seen with the reverse thickening sign. A spindle sign is a harbinger of retinal neovascularization to follow in 1 to 2 weeks. Asymmetry in disc collaterals is seen in association with asymmetric venous diameters at the optic disc. There is a correlation between time to development of a successful anastomosis and other signs of venous decompression such as clearing of retinal hemorrhage and resolution of macular edema. When time to half clearing of retinal hemorrhage as determined
Figure 11. Example of the flow around the corner sign and the gap sign. Fluorescein dye turns around at a retinal venous branching (arrow) more proximal than the anastomosis site (triangle) and flows distally rather than proximally. This leaves a gap without fluorescein. Fluorescein appears again in this unperfused venous segment closer to the disc when tributaries to this segment contribute their flow (check mark).
2265
Ophthalmology Volume 106, Number 12, December 1999
Figure 12. Abrupt hyperfluorescence sign. At the site of a functioning anastomosis, the fluorescein column has an abrupt increase in brightness representing a discontinuity in the amount of fluorescein in adjacent segments of the vein at the anastomosis. A small amount of blood comes to the anastomosis from the right, where two branches supply the site (arrowheads) and a larger amount of blood comes to the site from the left where backflow from the disc out the shunt is occurring (arrow).
from sequential fundus photographs was plotted versus time to a successful anastomosis after initial laser, the coefficient of determination r2 ⫽ 0.776 (P ⬍ 0.0001). Comparison of Figures 7 and 8 illustrates this correlation. Similarly, when time to resolution of macular edema as determined by clinical examination was plotted versus time to successful anastomosis after initial laser, r2 ⫽ 0.510 (P ⫽ 0.0061). A more qualitative correlation was seen with development of a successful anastomosis and lessening of venous tortuosity in the decompressed sector of the fundus. A quantitative assessment of the concept of tortuosity is difficult to formulate, making a more precise analysis on this point elusive. Certain signs are mutually exclusive. For instance, in the case of distal obstruction occurring at the anastomosis site, one cannot have an abrupt increase in hyperfluorescence at the anastomosis. In addition, one will not see a thinner venous segment attached to the
Figure 13. Late wall staining sign. This frame during the arteriovenous phase shows the underperfused proximal venous segment (filled arrow), which will show the wall staining late (see Fig 14). The anastomosis is shown at the open arrow.
2266
Figure 14. Late wall staining sign. Same eye as shown in Figure 13 but in the late phase of the angiogram. The walls of the underperfused proximal venous segment show staining (arrow), probably reflecting relative ischemia.
anastomosis and proximal to it together with a reverse thickening sign. Neither can one see the narrow parent trunk sign in combination with a thinner venous segment proximal to the anastomosis than distal to it.
Persistence in Attempts, Site Location, and Argon Power as Related to Anastomosis Success When the initial anastomosis was unsuccessful, repeat attempts were made. Of the 15 successful cases, 9 had 1 attempt, 5 had 2 attempts, and 1 had 4 attempts. Of the nine unsuccessful cases, seven had one attempt and two had four attempts. More than one site was treated at some of the attempts. The number of sites attempted per laser session for the group of patients with a suc-
Figure 15. Indocyanine green angiogram showing direct connection of a retinal vein (filled arrow) and a choroidal vein (open arrow) at an anastomosis site. This frame also shows the reversal of flow at a retinal venous branching (large arrowhead) more proximal than the anastomosis with transit of dye distally out the anastomosis rather than centrally toward the disc. As a result, the proximal parent venous trunk is smaller in diameter and relatively hypofluorescent (small arrowhead).
Browning 䡠 Laser Chorioretinal Anastomosis
Figure 16. Indocyanine green angiogram abrupt hyperfluorescence sign. The dye column to the left of the anastomosis (arrow) is more hyperfluorescent than the column to the right of the anastomosis (large arrowhead). The explanation for the phenomenon is the same as that for the analogous fluorescein sign (see Fig 12). The anastomosis site frequently is relatively hypofluorescent on indocyanine green angiography (small arrowhead).
cessful anastomosis was 1.6 ⫾ SD 0.9 (n ⫽ 23). The number of sites attempted per laser session for the group of patients without successful anastomosis was 1.5 ⫾ SD 0.6 (n ⫽ 15). The difference in number of sites attempted per laser session did not differ between the two groups (P ⬎ 0.5). From postoperative photographs, distances could be measured from the optic disc to the anastomosis site for 45 of the attempted sites. The mean distance of successful sites was 4.2 ⫾ 1.8 mm (SD) compared to a mean distance of unsuccessful sites of 2.6 ⫾ 1.0 mm (SD) (P ⬍ 0.01). The mean power of argon laser treatment for the group of successful anastomoses was 3.1 ⫾ 0.4 W (SD) (n ⫽ 16). Power was not listed on the operative note for two of the sites. The mean power for the group of unsuccessful anastomoses was 3.0 ⫾ 0.8 W (SD) (n ⫽ 42). There was no statistically significant difference in the mean power for the two groups (P ⬎ 0.05).
Figure 18. Evolution of changes at an anastomosis site in the same eye as shown in Figure 17, 7 weeks later. In this fluorescein angiogram frame, the anastomosis is fed by branches both proximal (arrow) and distal (triangle) to the shunt site (compare Fig 17). Notice that the venous diameter increases as the vein travels from disc to anastomosis, implying reversal of blood flow from its usual direction.
Of the 15 patients with successful anastomoses, 3 patients had 2 successful anastomoses and 12 patients had 1 successful anastomosis. Of the 18 successful anastomoses, 12 were in the inferotemporal quadrant, 2 were inferonasal, 2 were superotemporal, and 2 were superonasal. Of the 42 failed anastomoses distributed among the 24 patients, 15 were in the inferotemporal quadrant, 14 were inferonasal, 2 were superotemporal, and 11 were superonasal. The proportions of anastomoses located in the superonasal and superotemporal quadrants did not differ to a statistically significant extent between the success and failure groups. The proportion of anastomoses located in the inferotemporal quadrant was higher in the success group (P ⬍ 0.05). The proportion of anastomoses located in the inferonasal quadrant was lower in the success group (P ⬍ 0.05).
Correlation of Anatomic Success and Visual Outcome Although the primary intent of this study is to describe signs helpful in recognizing an anatomically successful anastomosis, the intent of clinical relevance requires some reporting of relationship of anatomic success to visual outcome. In comparing the change in the number of Snellen lines of best-corrected visual acuity between the preoperative state and the last visit, the 15 patients with successful anastomoses had a mean improvement of 2.3 ⫾ 2.4 (SD) lines compared to a mean improvement of 0.2 ⫾ 2.3 (SD) lines for the 9 patients with unsuccessful anastomoses (P ⫽ 0.0439, Kruskal–Wallis test).
Discussion Figure 17. Evolution of changes at an anastomosis site. In this fluorescein angiogram frame, the anastomosis (arrow) is fed only by venous branches (triangles) more distal than the shunt site (compare Fig 18). The time from laser surgery is 9 weeks.
The technique of laser chorioretinal venous anastomosis has evolved since its first description in humans by McAllister and Constable in 1995.1 Prominent among the changes has been recognition of better outcomes with higher argon laser power and abandonment of sole argon use in favor of a
2267
Ophthalmology Volume 106, Number 12, December 1999 sequential technique in which the argon laser breaks through Bruch membrane, and the YAG laser punctures the adjacent, overlying retinal vein.4 It is recognized that earlier intervention probably is favorable to later to prevent macular scarring from longstanding edema.3,5 Anastomosis success rates and visual outcomes have improved with these changes. The success rate in McAllister’s hands has increased from 33% of patients in the original series1 to 54% in the most recent one.5 In our hands, the originally reported success rate of 25%3 has increased to 63% (15 of 24 patients, current study). In light of these improvements, the hypothesis of benefit relative to the natural history of the disease seems a reasonable one, and it is now appropriate to consider a randomized, controlled clinical trial of the technique with these modifications. A key element in such a trial will be recognition of success or failure of the anastomosis attempt. The current study is meant to systematize the fundus photographic, fluorescein angiographic, and indocyanine green angiographic findings to make follow-up as uniform as possible in such a study and to help clinicians who choose to selectively apply the technique before a definitive study of its efficacy has been completed. Because follow-up of the patients is more difficult than performing the procedure, the greatest detail possible regarding the postoperative sequence of events is worth reporting. The earliest sign of anastomosis success is the fluorangiographic hyperfluorescent spindle sign. This occurs at 1 to 4 weeks after laser treatment and is often followed within 2 more weeks by retinal neovascularization. If the spindle sign is witnessed, closer follow-up is indicated, with application of scatter photocoagulation if retinal neovascularization develops. The retinal neovascularization in this setting is exquisitely sensitive to scatter photocoagulation and readily regresses. However, if not treated, it can lead to severe visual loss from vitreous hemorrhage.7 A second helpful sign at this early time is a directly visualized connection between the retinal vein and the choroidal vein on indocyanine green angiography. Later signs of anastomosis success are retrograde flow, flow around the corner, the gap sign, and the abrupt change in fluorescence of the retinal venous column at the anastomosis site. At the same time, the indocyanine green angiogram will show an abrupt change in fluorescence in the retinal venous column at the anastomosis site. The fundus photographic signs are relatively late in developing and are less helpful in the management of the patient. This situation arises in part from the obscuring effects of surrounding choroidal hemorrhage, retinal and subretinal hemorrhage, and overlying vitreous hemorrhage that accompany the treatment itself and are slow to resolve. Moreover, the dynamic nature of fluorescein and indocyanine angiography allows one to detect informative changes in blood flow patterns, which develop first and only later are accompanied by morphologic correlates such as venous diameter changes. Thus, fluorescein angiography and indo-
2268
cyanine green angiography are valuable and probably necessary in the follow-up of these patients. Attention to the described patterns will help the clinician avert premature further attempts at anastomosis creation when promising early signs of shunt formation are present and will help in recognizing neovascularization when it can be stopped by scatter laser treatment and before vitreous hemorrhage or secondary traction retinal detachment occurs. The mechanism by which these anastomoses work to improve visual outcome in nonischemic central retinal vein occlusion has been illuminated in this study. As originally conceived, it was assumed that the anastomosis provided an outflow path for the entire retinal venous drainage.1 The idea was that blood from retinal quadrants remote from the anastomosis could actually flow across the disc to get to the shunt. It was shown later that this idea was too simple for the 20% of eyes with hemicentral venous anatomies, in which a single anastomosis decompresses only half the fundus.3 The data obtained herein show that the story is more complex in eyes with a central venous anatomy as well. In 14 (93%) of the 15 eyes we studied, the successful anastomosis decompressed only a sector or half of the retinal venous system. Some blood continues to leave the eye through the compromised central vein. Thus, a laserinduced anastomosis acts much as a disc collateral, helping to decompress the central vein, but not providing global venous bypass. Laser chorioretinal venous anastomosis for nonischemic central retinal vein occlusion is a promising new technique, but it remains controversial and unproven. A randomized clinical trial will be necessary to establish its proper role in the management of this disease.
References 1. McAllister IL, Constable IJ. Laser-induced chorioretinal venous anastomosis for treatment of nonischemic central retinal vein occlusion. Arch Ophthalmol 1995;113:456 – 62. 2. Fekrat S, Goldberg MF, Finkelstein D. Laser-induced chorioretinal venous anastomosis for nonischemic central or branch retinal vein occlusion. Arch Ophthalmol 1998;116:43–52. 3. Browning DJ, Antoszyk AN. Laser chorioretinal venous anastomosis for nonischemic central retinal vein occlusion. Ophthalmology 1998;105:670 –7. 4. McAllister IL. Discussion. Ophthalmology 1998;105:677–9. 5. McAllister IL, Douglas JP, Constable IJ, Yu DY. Laserinduced chorioretinal venous anastomosis for nonischemic central retinal vein occlusion: evaluation of the complications and their risk factors. Am J Ophthalmol 1998;126:219 –29. 6. Browning DJ. Outcomes and complications of laser-induced chorioretinal anastomosis. Vitreoretinal Update 1997. San Francisco, CA: American Academy of Ophthalmology, 1997; 112–5. 7. Browning DJ, Rotberg MH. Vitreous hemorrhage complicating laser-induced chorioretinal anastomosis for central retinal vein occlusion. Am J Ophthalmol 1996;122:588 –9.
Originally received: November 10, 1998. Revision accepted: July 27, 1999. Manuscript no. 98577. From Charlotte Eye, Ear, Nose, and Throat Associates, Charlotte, North Carolina. Presented in part at the American Academy of Ophthalmology annual meeting, New Orleans, Louisiana, November 1998. The author has no proprietary interest in any of the materials used in this study. Reprint requests to David J. Browning, MD, PhD, Charlotte Eye, Ear, Nose, and Throat Associates, 1600 East Third Street, Charlotte, NC 28204, E-mail: [email protected].
McAllister and Constable1 to provide an alternate drainage route for retinal venous blood. The technique has produced successful anastomoses in variable fractions of treated patients,1–3 and the method has been in a state of evolution as the initial technique using argon laser has been superseded by a serial technique using higher power argon laser followed by yttrium–aluminum– garnet (YAG) laser.4 There have been incidental remarks in publications regarding the pattern of blood flow and the fundus appearance in successful cases, but a detailed study of the fundus signs and blood flow characteristics of successful cases is lacking. Such a study is needed to provide insight as to how the method works and to guide the ophthalmologist as to the need for and the timing of retreatment.
2261
Ophthalmology Volume 106, Number 12, December 1999
Figure 1. Narrow parent trunk sign. Color fundus photograph showing anFigure 3. Fluorescein angiogram showing flow of blood from the disc toward the anastomosis. As more tributaries empty into the vein and carry example of daughter venous branches (filled arrows) having greater diameter than the parent trunk (large arrowhead). The photograph also showsblood distally, the venous diameter increases as in Figure 2. Note the an instance in which the venous segment connected to the anastomosis venous tributary (light arrow) empties into the inferior hemicentral vein, and proximal to it (open arrow) is thinner than the segment distal to itwith abrupt increase in the vein diameter distal to the intersection, implying blood flow away from the disc, not toward it. Blood flows toward (small arrowhead). The disc venous asymmetry sign is also illustrated in which the superior vein stemming from the central vein has a greaterthe anastomosis (triangle). diameter than the inferior vein. At the same time, dilated collaterals are present on the superior disc margin (e.g., at the 1:30- and 2:30-o’clock and two of these cases have been reported.3 The subsequent ten meridians) but not on the inferior disc margin. 4
Methods This study reports on fundus photographs, sequential fluorescein angiograms, and indocyanine green angiograms in 15 consecutive cases of successful laser-induced chorioretinal venous anastomosis for central or hemicentral retinal vein occlusion treated by 1 retina specialist in a private practice. These cases are drawn from a larger population of 24 consecutive cases treated for this condition, of which 9 were unsuccessful. The time period for case accrual was June 20, 1995, through June 9, 1998. Twenty-two of the patients were treated by the author and 2 by Andrew Antoszyk, MD (coauthor of ref. 3). The first five successful patients were treated in the originally described technique of McAllister and Constable,1
cases were treated in a modified technique described later. In this technique, the argon laser is used to rupture Bruch membrane adjacent to a retinal vein chosen preferably in the inferior fundus, usually 2- to 4-disc diameters away from the optic disc. Higher powers than in the original technique are chosen, from 2.5 to 4.5 W, depending on the degree of lens opacity. The number of argon burns ranged from 1 to 13 depending on the number of anastomoses attempted. Then, the patient was treated with the YAG laser. Treatment is directed to the edge of the retinal vein adjacent to the argon burn. The YAG laser powers used ranged from 2.5 to 5.2 mJ depending on lens opacity. The number of YAG applications ranged from 1 to 15, until retinal venous bleeding occurred. Color fundus photographs and fluorescein angiograms were obtained
Figure 4. An example of the venous asymmetry sign. This eye has a single central vein, yet there is venous diameter asymmetry as the branches enter Figure 2. Reverse tapering sign. Fundus photograph showing increased the disc. The superior branch (large arrowhead) has a larger diameter than diameter of a retinal vein as one traverses it heading away from the disc,the inferior branch (small arrowhead). The successful anastomosis is a reversal of the usual pattern. The diameter at the arrowhead is smallerlocated in the inferior fundus and is not seen in this view. The daughter than the diameter at the arrow. The corresponding fluorescein angiogramvenous branches are thicker than the parent venous trunk at the arrow, a sign of blood flow away from the disc and out the anastomosis. explaining the appearance is shown in Figure 3.
2262
Browning 䡠 Laser Chorioretinal Anastomosis before surgery in all patients and at 2- to 4-week intervals in the postoperative period until the anastomosis was functioning. Indocyanine green angiograms were obtained in 9 of 15 patients when this method became available to us. All analysis of the photographs and angiograms was performed by the author. Informed consent was obtained from all patients.
Results To help in orienting the reader, results will be presented in a parallel manner for the three methods of examination: fundus photography, fluorescein angiography, and indocyanine green angiography. For each of these methods, new signs will first be described and illustrated. The relative frequency of the new and previously reported signs will be reported next. Finally, the temporal sequence of development of the signs will be reported.
Fundus Photography Previously described and illustrated signs include asymmetric disc collaterals,3 asymmetric vein diameters at the disc,3 a thinner venous segment proximal to the anastomosis,2,3,5 a thicker venous segment proximal to the anastomosis,1,2 enlargement of a draining choroidal vein over time,2 and a venous loop at the anastomosis site.2,3 In addition to the previously documented signs, we observed three previously undescribed fundus signs. The first we have termed the narrow parent trunk sign (Fig 1). It occurs at the next retinal venous branching more proximal than the anastomosis. In this sign, the daughter veins at the involved branching have larger diameters than their parent trunk. This appearance correlates with the appearance in the fluorescein or indocyanine green angiogram of dye flow from one daughter branch to the other and not toward the parent trunk as expected (see Fig 15 for an example of this). This finding suggests more blood flow through the daughter branches than through the parent trunk. The second sign has been termed the reverse tapering sign and occurs at the venous segment at the optic disc, which leads toward the anastomosis (Fig 2). In this sign, one observes that the diameter of the venous segment increases as one traverses the vein distally, a reversal of the usual pattern of venous narrowing. This correlates with tributary venules emptying into the parent vein with the flow directed peripherally out the anastomosis rather than centrally out the disc (Fig 3). The third new sign is termed the disc venous asymmetry sign. Although previously published photographs have shown asymmetry in the thickness of retinal veins at the disc, these cases have reflected hemicentral retinal venous anatomy.3 We have observed that this fundus sign can be associated with an unequivocal single central vein as shown in Figure 4. The reason for greater venous engorgement superiorly in a situation in which the superior and inferior hemicentral veins visibly join is unclear. No fundus photographic sign occurred in every successful anastomosis. Table 1 lists the relative frequency of various signs. The most frequently observed sign is an abrupt increase in the thickness of the more distal venous segment attached to the anastomosis compared to the central segment attached to the anastomosis. This correlates with flow of blood entering the anastomosis from two directions, with a presumed greater volume of blood coming through the distally attached venous segment than the proximally attached segment. The chronology of first appearances of various fundus photographic signs is listed in Table 2. The earliest of these signs is asymmetry of disc collateral vessels and of veins as they pass into the disc. These earliest signs can appear as early as 3 weeks after
Table 1. Relative Frequency of Observed Signs in Eyes with Successful Laser-induced Chorioretinal Venous Anastomosis* No. (%) with Sign Fundus photographic signs Venous segment proximal to anastomosis is thinner Asymmetric vein diameters at the disc Narrow parent trunk (daughter venous branches larger than parent trunk) Reverse tapering (increasing venous diameter further from disc) Venous segment proximal to anastomosis is thicker Venous loop at anastomosis site Fluorescein angiographic signs Flow around the corner Gap Abrupt hyperfluorescence Trilaminar flow Retrograde venous flow between disc and anastomosis Spindle Late wall staining Indocyanine green angiographic signs Connection of retinal vein and choroidal vein Abrupt hyperfluorescence Enlargement of draining choroidal vein over time
11 (73) 9 (60) 6 (40) 6 (40) 4 (27) 2 (13) 12 (80) 7 (47) 5 (33) 3 (20) 3 (20) 3 (20) 2 (13) 4 (44) 4 (44) 2 (22)
* The denominator in the fundus photographic sign and fluorescein angiographic sign sections is 15; in the indocyanine green angiographic section it is 9.
laser treatment. A relatively late sign is a venous loop at the anastomosis. The development of a venous loop over time is shown in Figures 5 and 6. The finding correlates with presumed increasing blood flow through the anastomosis. As this example illustrates, sequential comparison of photographs is informative regarding a developing anastomosis. In another such example displaying a different facet of shunt development, Figures 7 and 8 show progressive thinning of a proximal venous trunk as the anastomosis matures and diverts more venous blood away from the disc.
Fluorescein Angiography Previously described and illustrated signs include trilaminar flow,1 retrograde venous flow between disc and anastomosis,1 and nondrainage of the entire retina by virtue of a hemicentral retinal vein.3 Previously unreported signs were discovered. The first of these is a hyperfluorescent spindle-shaped lesion occurring at the anastomosis site. It is the earliest sign of a successful anastomosis, occurring 1 to 2 weeks after the laser treatment. It can be followed within 2 weeks by neovascularization, which arises from the retinal vein or from the choroid (Figs 9 and 10). Because of the appearance of the lesion, it has been called the spindle sign.6 A later sign is the appearance of a gap in the dye stream occurring between the first retinal venous branching more proximal than the anastomosis and the next retinal venous branch closer to the disc (Fig 11). This gap shows that flow occurs out the anastomosis rather than toward the disc for blood arriving at the involved venous branching. It also shows that blood flow is not occurring across the disc toward the anastomosis from sections of the fundus separated from the anastomosis by the disc. Instead, it shows slow flow of blood out the central vein for part of the fundus more proximal than the anastomosis to the disc.
2263
Ophthalmology Volume 106, Number 12, December 1999 Table 2. Relative Chronology of Observed Signs in Eyes with Successful Laser-induced Chorioretinal Venous Anastomosis* Time of Earliest Appearance (wks) Fundus photographic signs Asymmetric collateral disc vessels Asymmetric vein diameters at the disc Venous segment proximal to anastomosis is thicker Venous segment proximal to anastomosis is thinner Narrow parent trunk (daughter venous branches larger than parent trunk) Reverse tapering (increasing venous diameter further from disc) Venous loop at anastomosis site Fluorescein angiographic signs Spindle Abrupt hyperfluorescence Retrograde venous flow between disc and anastomosis Gap Flow around the corner Trilaminar flow Late wall staining Indocyanine green angiographic signs Connection of retinal vein and choroidal vein Abrupt hyperfluorescence Enlargement of draining choroidal vein over time
3 3 3 4 5 7 1 5 5 5 5 9 24 2 6 12
* The denominator in the fundus photographic sign and fluorescein angiographic sign sections is 15; in the indocyanine green angiographic section, it is 9.
A sign seen in a well-developed anastomosis is a sharp change in hyperfluorescence at the anastomosis. This sign suggests a relative discontinuity in the thickness of the fluorescein column occurring at an anastomosis site (Fig 12); this interpretation suggests the name abrupt fluorescence change sign. A rare sign is the increased staining of the venous walls of an underperfused segment of the vein connecting the disc to the
Figure 5. Red-free photograph showing anastomosis site before significant flow through the shunt (arrow). Notice that the inferior hemicentral vein is more distended (arrowhead) at this time than in Figure 6, when the shunt is working, decompressing the venous drainage out the disc.
2264
Figure 6. Red-free photograph of same eye as shown in Figure 5 several weeks later. Now, flow through the shunt is greater and a venous loop 6 (arrow) has developed. The inferior hemicentral vein is less distended than in Figure 5 (arrowhead).
anastomosis. This sign may reflect relative ischemia of this venous segment. It correlates with a smaller than normal diameter for such a proximal part of the venous tree (Figs 13 and 14). The relative frequency of fluorescein angiographic signs is shown in Table 1. The flow around the corner sign is most commonly seen (80% of cases), whereas the more publicized trilaminar flow sign is only a fourth as commonly observed. The spindle sign is an observed sign seen earliest (Table 2). The trilaminar flow sign occurs late. It was seen earliest at 9 weeks after laser surgery, when relatively large volumes of blood drain through a more developed shunt. Other listed signs occur at intermediate intervals.
Indocyanine Green Angiography One previously reported ICG sign is a visualized connection of the retinal vein and the choroidal vein.2 This sign, shown in Figure 15, is particularly useful because of its early appearance at 2 to 3 weeks after laser treatment (Table 2). One other useful sign, not described previously, is a sharp change in hyperfluorescence of the
Figure 7. Red-free photograph showing greater thickness of proximal venous segment (arrow) before the anastomosis site has begun to function (arrowhead); compare to Figure 8.
Browning 䡠 Laser Chorioretinal Anastomosis
Figure 8. Red-free photograph of same eye as in Figure 7 several weeks later showing decreased thickness of the proximal venous segment (arrow) now that the shunt is working (arrowhead); compare to Figure 7.
dye column at the anastomosis. The appearance and probable explanation are likely analogous to the similar observation in the fluorescein angiogram (Fig 16). This is a relatively late sign, seen at 6 weeks or later, and seems to require relatively high flow through the shunt to be seen. The initial description of enlargement of a draining choroidal vein over time was made from fundus photographs,2 but this is only possible in an eye with relatively blonde coloration. Indocyanine green angiography reveals the finding even in eyes with darker fundi, in which it was seen in two (22%) of nine cases and only as a late sign (12 weeks at earliest).
Evolution, Concordance, and Mutual Exclusivity of Photographic Signs An eye can evolve from one set of signs to a later distinct set. An example of this is shown in Figures 17 and 18. In this eye at 9 weeks after laser treatment, an anastomosis has formed in which the only venous input is distal to the anastomosis (Fig 17). How-
Figure 9. An example of the spindle sign. The hyperfluorescent, spindleshaped structures at two anastomosis sites are seen 8 days after laser treatment (arrows).
Figure 10. Same patient as in Figure 9 12 days after the fluorescein angiogram in Figure 9 and 20 days after laser treatment. The spindle structures have evolved into neovascular fronds. In this case, the blood supply was choroidal, but in other cases it can be from the retinal circulation.
ever, by 16 weeks after surgery, a prominent proximal input to the anastomosis has developed, as shown in Figure 18. The photographic signs cluster in groups. The gap sign is frequently seen in cases showing the flow around the corner sign. Retrograde flow from disc to anastomosis is seen with the reverse thickening sign. A spindle sign is a harbinger of retinal neovascularization to follow in 1 to 2 weeks. Asymmetry in disc collaterals is seen in association with asymmetric venous diameters at the optic disc. There is a correlation between time to development of a successful anastomosis and other signs of venous decompression such as clearing of retinal hemorrhage and resolution of macular edema. When time to half clearing of retinal hemorrhage as determined
Figure 11. Example of the flow around the corner sign and the gap sign. Fluorescein dye turns around at a retinal venous branching (arrow) more proximal than the anastomosis site (triangle) and flows distally rather than proximally. This leaves a gap without fluorescein. Fluorescein appears again in this unperfused venous segment closer to the disc when tributaries to this segment contribute their flow (check mark).
2265
Ophthalmology Volume 106, Number 12, December 1999
Figure 12. Abrupt hyperfluorescence sign. At the site of a functioning anastomosis, the fluorescein column has an abrupt increase in brightness representing a discontinuity in the amount of fluorescein in adjacent segments of the vein at the anastomosis. A small amount of blood comes to the anastomosis from the right, where two branches supply the site (arrowheads) and a larger amount of blood comes to the site from the left where backflow from the disc out the shunt is occurring (arrow).
from sequential fundus photographs was plotted versus time to a successful anastomosis after initial laser, the coefficient of determination r2 ⫽ 0.776 (P ⬍ 0.0001). Comparison of Figures 7 and 8 illustrates this correlation. Similarly, when time to resolution of macular edema as determined by clinical examination was plotted versus time to successful anastomosis after initial laser, r2 ⫽ 0.510 (P ⫽ 0.0061). A more qualitative correlation was seen with development of a successful anastomosis and lessening of venous tortuosity in the decompressed sector of the fundus. A quantitative assessment of the concept of tortuosity is difficult to formulate, making a more precise analysis on this point elusive. Certain signs are mutually exclusive. For instance, in the case of distal obstruction occurring at the anastomosis site, one cannot have an abrupt increase in hyperfluorescence at the anastomosis. In addition, one will not see a thinner venous segment attached to the
Figure 13. Late wall staining sign. This frame during the arteriovenous phase shows the underperfused proximal venous segment (filled arrow), which will show the wall staining late (see Fig 14). The anastomosis is shown at the open arrow.
2266
Figure 14. Late wall staining sign. Same eye as shown in Figure 13 but in the late phase of the angiogram. The walls of the underperfused proximal venous segment show staining (arrow), probably reflecting relative ischemia.
anastomosis and proximal to it together with a reverse thickening sign. Neither can one see the narrow parent trunk sign in combination with a thinner venous segment proximal to the anastomosis than distal to it.
Persistence in Attempts, Site Location, and Argon Power as Related to Anastomosis Success When the initial anastomosis was unsuccessful, repeat attempts were made. Of the 15 successful cases, 9 had 1 attempt, 5 had 2 attempts, and 1 had 4 attempts. Of the nine unsuccessful cases, seven had one attempt and two had four attempts. More than one site was treated at some of the attempts. The number of sites attempted per laser session for the group of patients with a suc-
Figure 15. Indocyanine green angiogram showing direct connection of a retinal vein (filled arrow) and a choroidal vein (open arrow) at an anastomosis site. This frame also shows the reversal of flow at a retinal venous branching (large arrowhead) more proximal than the anastomosis with transit of dye distally out the anastomosis rather than centrally toward the disc. As a result, the proximal parent venous trunk is smaller in diameter and relatively hypofluorescent (small arrowhead).
Browning 䡠 Laser Chorioretinal Anastomosis
Figure 16. Indocyanine green angiogram abrupt hyperfluorescence sign. The dye column to the left of the anastomosis (arrow) is more hyperfluorescent than the column to the right of the anastomosis (large arrowhead). The explanation for the phenomenon is the same as that for the analogous fluorescein sign (see Fig 12). The anastomosis site frequently is relatively hypofluorescent on indocyanine green angiography (small arrowhead).
cessful anastomosis was 1.6 ⫾ SD 0.9 (n ⫽ 23). The number of sites attempted per laser session for the group of patients without successful anastomosis was 1.5 ⫾ SD 0.6 (n ⫽ 15). The difference in number of sites attempted per laser session did not differ between the two groups (P ⬎ 0.5). From postoperative photographs, distances could be measured from the optic disc to the anastomosis site for 45 of the attempted sites. The mean distance of successful sites was 4.2 ⫾ 1.8 mm (SD) compared to a mean distance of unsuccessful sites of 2.6 ⫾ 1.0 mm (SD) (P ⬍ 0.01). The mean power of argon laser treatment for the group of successful anastomoses was 3.1 ⫾ 0.4 W (SD) (n ⫽ 16). Power was not listed on the operative note for two of the sites. The mean power for the group of unsuccessful anastomoses was 3.0 ⫾ 0.8 W (SD) (n ⫽ 42). There was no statistically significant difference in the mean power for the two groups (P ⬎ 0.05).
Figure 18. Evolution of changes at an anastomosis site in the same eye as shown in Figure 17, 7 weeks later. In this fluorescein angiogram frame, the anastomosis is fed by branches both proximal (arrow) and distal (triangle) to the shunt site (compare Fig 17). Notice that the venous diameter increases as the vein travels from disc to anastomosis, implying reversal of blood flow from its usual direction.
Of the 15 patients with successful anastomoses, 3 patients had 2 successful anastomoses and 12 patients had 1 successful anastomosis. Of the 18 successful anastomoses, 12 were in the inferotemporal quadrant, 2 were inferonasal, 2 were superotemporal, and 2 were superonasal. Of the 42 failed anastomoses distributed among the 24 patients, 15 were in the inferotemporal quadrant, 14 were inferonasal, 2 were superotemporal, and 11 were superonasal. The proportions of anastomoses located in the superonasal and superotemporal quadrants did not differ to a statistically significant extent between the success and failure groups. The proportion of anastomoses located in the inferotemporal quadrant was higher in the success group (P ⬍ 0.05). The proportion of anastomoses located in the inferonasal quadrant was lower in the success group (P ⬍ 0.05).
Correlation of Anatomic Success and Visual Outcome Although the primary intent of this study is to describe signs helpful in recognizing an anatomically successful anastomosis, the intent of clinical relevance requires some reporting of relationship of anatomic success to visual outcome. In comparing the change in the number of Snellen lines of best-corrected visual acuity between the preoperative state and the last visit, the 15 patients with successful anastomoses had a mean improvement of 2.3 ⫾ 2.4 (SD) lines compared to a mean improvement of 0.2 ⫾ 2.3 (SD) lines for the 9 patients with unsuccessful anastomoses (P ⫽ 0.0439, Kruskal–Wallis test).
Discussion Figure 17. Evolution of changes at an anastomosis site. In this fluorescein angiogram frame, the anastomosis (arrow) is fed only by venous branches (triangles) more distal than the shunt site (compare Fig 18). The time from laser surgery is 9 weeks.
The technique of laser chorioretinal venous anastomosis has evolved since its first description in humans by McAllister and Constable in 1995.1 Prominent among the changes has been recognition of better outcomes with higher argon laser power and abandonment of sole argon use in favor of a
2267
Ophthalmology Volume 106, Number 12, December 1999 sequential technique in which the argon laser breaks through Bruch membrane, and the YAG laser punctures the adjacent, overlying retinal vein.4 It is recognized that earlier intervention probably is favorable to later to prevent macular scarring from longstanding edema.3,5 Anastomosis success rates and visual outcomes have improved with these changes. The success rate in McAllister’s hands has increased from 33% of patients in the original series1 to 54% in the most recent one.5 In our hands, the originally reported success rate of 25%3 has increased to 63% (15 of 24 patients, current study). In light of these improvements, the hypothesis of benefit relative to the natural history of the disease seems a reasonable one, and it is now appropriate to consider a randomized, controlled clinical trial of the technique with these modifications. A key element in such a trial will be recognition of success or failure of the anastomosis attempt. The current study is meant to systematize the fundus photographic, fluorescein angiographic, and indocyanine green angiographic findings to make follow-up as uniform as possible in such a study and to help clinicians who choose to selectively apply the technique before a definitive study of its efficacy has been completed. Because follow-up of the patients is more difficult than performing the procedure, the greatest detail possible regarding the postoperative sequence of events is worth reporting. The earliest sign of anastomosis success is the fluorangiographic hyperfluorescent spindle sign. This occurs at 1 to 4 weeks after laser treatment and is often followed within 2 more weeks by retinal neovascularization. If the spindle sign is witnessed, closer follow-up is indicated, with application of scatter photocoagulation if retinal neovascularization develops. The retinal neovascularization in this setting is exquisitely sensitive to scatter photocoagulation and readily regresses. However, if not treated, it can lead to severe visual loss from vitreous hemorrhage.7 A second helpful sign at this early time is a directly visualized connection between the retinal vein and the choroidal vein on indocyanine green angiography. Later signs of anastomosis success are retrograde flow, flow around the corner, the gap sign, and the abrupt change in fluorescence of the retinal venous column at the anastomosis site. At the same time, the indocyanine green angiogram will show an abrupt change in fluorescence in the retinal venous column at the anastomosis site. The fundus photographic signs are relatively late in developing and are less helpful in the management of the patient. This situation arises in part from the obscuring effects of surrounding choroidal hemorrhage, retinal and subretinal hemorrhage, and overlying vitreous hemorrhage that accompany the treatment itself and are slow to resolve. Moreover, the dynamic nature of fluorescein and indocyanine angiography allows one to detect informative changes in blood flow patterns, which develop first and only later are accompanied by morphologic correlates such as venous diameter changes. Thus, fluorescein angiography and indo-
2268
cyanine green angiography are valuable and probably necessary in the follow-up of these patients. Attention to the described patterns will help the clinician avert premature further attempts at anastomosis creation when promising early signs of shunt formation are present and will help in recognizing neovascularization when it can be stopped by scatter laser treatment and before vitreous hemorrhage or secondary traction retinal detachment occurs. The mechanism by which these anastomoses work to improve visual outcome in nonischemic central retinal vein occlusion has been illuminated in this study. As originally conceived, it was assumed that the anastomosis provided an outflow path for the entire retinal venous drainage.1 The idea was that blood from retinal quadrants remote from the anastomosis could actually flow across the disc to get to the shunt. It was shown later that this idea was too simple for the 20% of eyes with hemicentral venous anatomies, in which a single anastomosis decompresses only half the fundus.3 The data obtained herein show that the story is more complex in eyes with a central venous anatomy as well. In 14 (93%) of the 15 eyes we studied, the successful anastomosis decompressed only a sector or half of the retinal venous system. Some blood continues to leave the eye through the compromised central vein. Thus, a laserinduced anastomosis acts much as a disc collateral, helping to decompress the central vein, but not providing global venous bypass. Laser chorioretinal venous anastomosis for nonischemic central retinal vein occlusion is a promising new technique, but it remains controversial and unproven. A randomized clinical trial will be necessary to establish its proper role in the management of this disease.
References 1. McAllister IL, Constable IJ. Laser-induced chorioretinal venous anastomosis for treatment of nonischemic central retinal vein occlusion. Arch Ophthalmol 1995;113:456 – 62. 2. Fekrat S, Goldberg MF, Finkelstein D. Laser-induced chorioretinal venous anastomosis for nonischemic central or branch retinal vein occlusion. Arch Ophthalmol 1998;116:43–52. 3. Browning DJ, Antoszyk AN. Laser chorioretinal venous anastomosis for nonischemic central retinal vein occlusion. Ophthalmology 1998;105:670 –7. 4. McAllister IL. Discussion. Ophthalmology 1998;105:677–9. 5. McAllister IL, Douglas JP, Constable IJ, Yu DY. Laserinduced chorioretinal venous anastomosis for nonischemic central retinal vein occlusion: evaluation of the complications and their risk factors. Am J Ophthalmol 1998;126:219 –29. 6. Browning DJ. Outcomes and complications of laser-induced chorioretinal anastomosis. Vitreoretinal Update 1997. San Francisco, CA: American Academy of Ophthalmology, 1997; 112–5. 7. Browning DJ, Rotberg MH. Vitreous hemorrhage complicating laser-induced chorioretinal anastomosis for central retinal vein occlusion. Am J Ophthalmol 1996;122:588 –9.