Anatomy of Arteriovenous Crossings in Branch Retinal Vein Occlusion David Weinberg, M.D., David G. Dodwell, M.D., and Steven A. Fern, B.S.
We studied the photographic records of 292 eyes, including 103 eyes with branch retinal vein occlusion, 90 fellow eyes, and 99 control eyes without branch retinal vein occlusion. All arteriovenous crossings within three disk diameters of the optic disk, including the crossings at the sites of branch retinal vein occlusions, were studied. The relative positions of the crossing artery and vein could be determined at 1,939 crossings in all eyes. Crossings at which a vein crossed over an artery were a common finding (22.3% to 33.0% of crossings), but were rare at the crossings where branch retinal vein occlusions were found (2.4%). A greater proportion of arterial overcrossings was found in eyes with branch retinal vein occlusions (77.7%) compared to fellow eyes (70.6%) or control eyes (67.0%). Our data indicate that arterial overcrossings are at relatively higher risk of branch retinal vein occlusion than venous overcrossings, and that the risk of branch vein occlusion in an eye is proportional to the number of arterial overcrossings in the eye.
The crossing of retinal vessels such that the artery lies over (that is, anterior to or innermost to) the vein is considered the normal anatomic configuration. In 1936, however, Jensen observed that venous overcrossings occur at 30% of all crossings in the retinas of normal eyes.' The presence of both types of crossings has been demonstrated histologically." We have rarely seen a branch retinal vein occlusion at an intersection at which the vein crosses over the artery (Fig. 1). References in published studies to similar occlusions are also rare.v" In one series that specifically addressed the anatomy of the crossings at 25 branch vein occlusions, no venous overcrossings were found at the sites of occlusion." We undertook this study to examine the anatomy of the retinal vessels at the sites of branch vein occlusions and to determine if it differs from the anatomy of other arteriovenous intersections in the same eyes or uninvolved fellow and control eyes.
T
The files at the retinal photography laboratories of Northwestern University and the New York Hospital, Cornell University from 1985 to 1988 were reviewed for patients with a diagnosis of branch retinal vein occlusion. Only vein occlusions that occurred within three disk diameters of the edge of the optic disk were studied. Central and hemispheric central retinal vein occlusions were excluded. One hundred and three eyes of 100 patients met these criteria. Using color transparencies, red-free photographs, and fluorescein angiograms of the involved and fellow eyes, the anatomy (artery over vein or vein over artery) of all photographed arteriovenous crossings within three disk diameters of the edge of the optic disk was studied. Intersections occurring on the surface
HE OBSERVATION that branch retinal vein occlusion occurs at arteriovenous intersections was made over 100 years ago by Leber.' The associations of branch retinal vein occlusion with systemic hypertension and with the retinal vascular manifestations of hypertension have been documented.v"
Accepted for publication Jan. 5, 1990. From the Department of Ophthalmology, Northwestern University Medical School, Chicago (Dr. Weinberg), and the Department of Ophthalmology, New York Hospital, Cornell Medical Center, New York (Dr. Dodwell and Mr. Fern). This study was supported in part by an unrestricted grant from Research to Prevent Blindness, Inc. Reprint requests to David G. Dodwell, M.D., Springfield Eye Center, 301 N. 8th St., Springfield, IL 62701.
298
Material
and Methods
©AMERICAN JOURNAL OF OPHTHALMOLOGY
109:298-302,
MARCH,
1990
Vol. 109, No.3
Branch Retinal Vein Occlusion
Fig. 1 (Weinberg, Dodwell, and Fern). Example of a branch vein occlusion at a venous overcrossing. Late frame of the angiogram shows staining of the vein as it crosses over the artery. of the optic disk were excluded. The vessel crossing over was defined as lying innermost or most anterior at the arteriovenous intersection. At the site of each vein occlusion, the anatomy was designated "artery over vein" or "vein over artery" if the positions of the vessels could be determined with certainty. If the positions of the vessels could not be ascertained, the site was designated "undetermined." Arteriovenous crossings other than the site of occlusion were designated "artery over vein" or "vein over artery" if the anatomy could be determined with certainty. The number of crossings at which the anatomy could not be determined was not tabulated for these crossings without vein occlusions. As a control population, the photographic records of 53 consecutive patients referred for fluorescein angiography with diagnoses other than branch retinal vein occlusion were studied. These eyes were analyzed exactly as the fellow eyes of patients with branch retinal vein occlusion. Statistical testing was performed using chisquare analysis. Statistical significance was defined as P < .05, unless otherwise stated.
Results Age ranged from 21 to 88 years (mean, 66.5 years), and 54 patients were men and 46 were women. The right eye was involved in 53 patients (51.5%), and the left eye was involved in
299
50 patients (48.5%). Three patients had bilateral branch retinal vein occlusions (two men and one woman). The locations (by quadrant) of the occlusions were as follows: 62 superotemporal (60.2%),39 inferotemporal (37.9%), one superonasal (1.0%), and one inferonasal (1.0%). The results are summarized in Figure 2 and the Table. At the site of the branch retinal vein occlusion, the artery lay anterior to the vein in 82 eyes (79.6%). The vein lay anterior to the artery in two eyes (1.9%). In 18 eyes (17.4%), the crossing was undetermined. In one eye (1.0%), the occlusion occurred at an anomalous vein as it exited from the edge of the disk rather than at an arteriovenous crossing. For the 84 vein occlusion sites at which the anatomy could be determined, 82 (97.6%) were arterial overcrossings and two (2.4%) were venous overcrossings. Determination of the anatomy could not be made in 18 eyes: seven eyes had media opacity, five eyes had hemorrhage, and in six eyes, the intersections were too small to characterize with certainty. The anatomy of a total of 728 arteriovenous crossings, including the 84 determinable branch vein occlusion sites, was tabulated in the 103
100 90 til
Ol
c
80
'iii
70
0
60
til
0 '0 C Q) ~
Q)
C.
50 40 30 20 10 0
BRVO sites BRVO eyes Fellow eyes Control eyes n=84 n=728 n=487 n=724
o ~
Venous overcrossings Arterial overcrossings
Fig. 2 (Weinberg, Dodwell, and Fern). Distribution of types of arteriovenous crossings among the groups. Unshaded and shaded areas represent percentage of venous and arterial overcrossings, respectively. BRVO sites indicates sites of occlusion in eyes with branch vein occlusion; BRVO eyes, all crossings in eyes with branch vein occlusion; fellow eyes, crossings in fellow eyes of those with branch vein occlusion; control eyes, crossings in eyes of patients without branch vein occlusion; n, the number of arteriovenous crossings at which the positions of the vessels were determined for each group.
300
March, 1990
AMERICAN JOURNAL OF OPHTHALMOLOGY
TABLE COMPARISON OF FEATURES OF ARTERIOVENOUS CROSSINGS NO. OF
NO. OF
NO. OF
NO. OF
DETERMINABLE
CROSSINGS
NO. OF ARTERIAL
NO. OF VENOUS
EYES
CROSSINGS
CROSSINGS
PER EYE
OVERCROSSINGS (%)
OVERCROSSINGS (%)
Branch retinal vein occlusion sites
103
102
82 (97.6)
2 (2.4)
Branch retinal vein occlusion eyes
103
7.1
566 (77.7)
162 (22.3)
84 728
Fellow eyes
90
487
5.4
344 (70.6)
143 (29.4)
Control eyes
99
724
7.3
485 (67.0)
239 (33.0)
eyes (mean, 7.1 per eye). Of the arteriovenous crossings, 566 (77.7%) were arterial and 162 (22.3%) were venous. The difference in frequency of arterial and venous overcrossings between the vein occlusion sites and all crossings in the involved eyes was statistically significant (P = .0001). Of the 103 branch retinal vein occlusions studied, six eyes of three patients had bilateral occlusions and were not included in the analysis of fellow eyes. In another seven cases, photographs of the fellow eyes were unavailable or inadequate for analysis, leaving 90 fellow eyes of 90 patients. Among these eyes, the anatomy of 487 crossings was determined (mean, 5.4 per eye). Fewer crossings per eye were counted in the fellow eyes than in the involved eyes because, in general, fewer photographic frames were available for the fellow eyes. The early frames of the fluorescein angiograms, which were useful for determining the anatomy in the involved eyes, were not available for the fellow eyes. Of the 487 crossings counted, 344 (70.6%) were arterial overcrossings and 143 (29.4%) were venous overcrossings. Compared to the vein occlusion sites and to the vein occlusion eyes, the differences were statistically significant (P = .0001 and .0063, respectively), with a greater prevalence of venous overcrossings in the fellow eyes. Of 53 consecutive patients referred for fluorescein angiography with diagnoses other than branch retinal vein occlusion, 99 eyes had photographic records adequate for study. The diagnoses were as follows: diabetes (16), agerelated macular degeneration (11), cystoid macular edema (three), optic neuropathy (three), macular hole (three), presumed ocular histoplasmosis (two), myopia (two), macular pucker (two), and other (11). In the last category, no diagnosis was represented more than once. Patient age ranged from 14 to 88 years (mean,
63.9 years), and 22 (41.5%) patients were men and 31 (58.5%) were woman. The anatomy of 724 crossings was determined in the 99 eyes (mean, 7.3 per eye). Of the crossings, 485 (67.0%) were arterial and 239 (33.0%) were venous overcrossings. The proportion of venous overcrossings in the control eyes was greater than in any other group. This difference was statistically significant compared to the vein occlusion sites (P = .0001) and the vein occlusion eyes (P = .0001), but not compared to the fellow eyes (P = .20).
Discussion The ages of the patients with branch retinal vein occlusion and the controls were comparable. We would not expect age to be an important variable, because the anatomy of the crossings is determined prenatally and should not influence longevity. The difference in gender between the branch vein occlusion eyes and control eyes was not statistically Significant (P = .19). The location of the occlusions was similar to those in previous series.v" with most occlusions in the superotemporal quadrant and most of the remainder in the inferotemporal quadrant. Branch retinal vein occlusions in the nasal quadrants were rare. Our data confirm the observations of Ienserr' that venous overcrossings are not a rare finding in normal fundi. Jensen used direct ophthalmoscopy to examine the eyes of 50 normal patients and found venous overcrossings at 30% of all arteriovenous intersections. Duker and Brown/ using color photographs and fluorescein angiograms, found no venous overcrossings at the site of occlusion in their series of 25 branch retinal vein occlusions. For controls, they observed a corresponding crossing in the opposite arcade (superior or inferior)
Vol. 109, No.3
Branch Retinal Vein Occlusion
in the same eye, and all first- and second-order crossings in one eye of 26 patients without branch retinal vein occlusions. The frequencies of venous overcrossings in the two groups were 35% and 32%, respectively. We used color stereoscopic photographs, redfree photographs, and fluorescein angiograms to examine 292 eyes, and we tabulated 1,939 arteriovenous intersections. We found venous overcrossings at two of 82 vein occlusion sites (2.4%). Venous overcrossings were present in 162 of 728 (22.3%) crossings in involved eyes, in 143 of 487 (29.4%) crossings in fellow eyes, and in 239 of 724 (33.0%) crossings in control eyes. If branch retinal vein occlusions occur randomly among all crossings, the proportion of venous overcrossings in a series of branch retinal vein occlusions should not differ from their frequency in the fundus overall (between 22% and 35% based on the data of jensen," Duker and Brown," and ourselves). We found the frequency of venous overcrossings lower and the frequency of arterial overcrossings higher at branch retinal vein occlusion sites than would be predicted based on their overall frequency in the involved eyes, fellow eyes, and control eyes. Thus, arterial overcrossings are at higher risk for branch retinal vein occlusion than venous overcrossings. The overall frequency of venous overcrossings increased from the involved eyes to the fellow eyes, and from the fellow eyes to the control eyes, with the highest incidence of venous overcrossings (and concomitantly lowest incidence of arterial overcrossings) in the control eyes (Fig. 2). This makes sense statistically, because eyes with a greater frequency of arterial overcrossings have more high-risk crossings, and thus would be expected to have a higher incidence of branch retinal vein occlusion. Several sources of bias in our data were considered. It is possible that the frequency of venous overcrossings at the site of vein occlusion was underestimated because of an overrepresentation of these crossings in the "undetermined" group. Even if half of the undetermined crossings were venous overcrossings, 91 of 102 occlusions (89.2 %) would be arterial overcrossings, and 11 of 102 occlusions (10.8%) would be venous overcrossings. These figures, although they almost certainly overestimate the frequency of venous overcrossings in the "undetermined" group, are still statistically significantly different compared to all other groups (branch retinal vein occlusion eyes, P = .011;
301
fellow eyes, P = .002; control eyes, P = .0001), with a lower frequency of venous overcrossings at sites of branch retinal vein occlusion. By including the occlusion site in the evaluation of the anatomy of all crossings in the involved eyes, bias may have been introduced, because particular effort was made to characterize these crossings, which had a different frequency distribution. By excluding the 84 determinable occlusion sites from the analysis of crossings in the involved eyes, the proportion of venous overcrossings increases from 22.3% (162 of 728) to 24.8% (160 of 644). This is still less than the proportion in the fellow eyes or the control eyes. The difference remains statistically significant compared to the control eyes (P = .0011), but not compared to fellow eyes (P = .10). Differences in the adequacy of the photographic records among the three groups resulted in differences in the mean number of crossings counted per eye. This may have introduced some bias if the locations of the two varieties of crossings are not randomly distributed. For instance, if one type of crossing tends to occur nearer to the disk, or at the intersections of larger vessels, it might be overrepresented in the fellow eyes, because fewer crossings were counted in these eyes, and those that were counted tended to be larger and nearer to the disk. We are unable to determine if such bias exists in our data. The higher relative risk of branch retinal vein occlusion at arterial overcrossings suggests a hemodynamic difference between arterial and venous overcrossings. We and others" have observed that characteristic hypertensive changes seen at arteriovenous crossings are less prominent at venous overcrossings than at arterial overcrossings. Seitz" has studied the histologic characteristics of both anatomic variations in normal and hypertensive patients. In the hypertensive patients, there was a thickening of the walls and a narrowing of the lumina of both vessels. At the site of the crossing, the vessels shared a common vascular wall and a common, thickened, adventitial and glial sheath. These changes were independent of which vessel was innermost. In both variants, the course of the artery was unchanged. The vein deviated around the artery, dipping deep into the retina in arterial overcrossings, and bulging against the internal limiting membrane in venous overcrossings. No compression of the underlying vessel was observed. Seitz attributed the more prominent clinical appearance of crossing phe-
302
AMERICAN JOURNAL OF OPHTHALMOLOGY
nomena (nicking and obscuration of the underlying vessel) in arterial overcrossings to the deeper position of the vein, rather than to any true compression of the vessel. Thus, by light microscopy, hemodynamic differences between the two anatomic configurations are not obvious. Perhaps there is a difference in the distensibility of the vein lying beneath the artery within the retina in an arterial overcrossing, as opposed to between the internal limiting membrane and the artery in a venous overcrossing. Such physiologic differences could explain the disparity in risk of branch retinal vein occlusion.
References
Handbuch der Gesammten Augenheilkunde, Pathologie und Therapie. Leipzig, Verlag Von Wilhelm Engelmann, 1877, pp. 521-535. 2. Gutman, F. A., and Zegarra, H.: The natural course of temporal retinal branch vein occlusion. Trans. Am. Acad. Ophthalmol. Otolaryngol. 78: 178, 1974.
3. Blankenship, G. W., and Okun, E.: Retinal tributary vein occlusion. Histology and management by photocoagulation. Arch. Ophthalmol. 89:363, 1973. 4. Jensen, V. A.: Clinical studies of tributary thrombosis in the central retinal vein. Acta Ophthalmol. 1 (sup pI. X): 1, 1936. 5. Seitz, R. (Blodi, F. c.. translator): The "crossing phenomenon." In The Retinal Vessels. St. Louis, C. V. Mosby, 1964, pp. 20-74. 6. Clemett, R. S.: Retinal branch vein occlusion. Changes at the site of obstruction. Br. J. Ophthalmol. 58:548,1974.
1. Leber, T.: Die Krankheite der Netzhaut und des
sehnerven. In Graefe, A., and Saernisch, T. (eds.):
March, 1990
7. Duker, J. 5., and Brown, G. c.: Anterior location of the crossing artery in branch retinal vein occlusion. Arch. Ophthalmol. 107:998, 1989.
OPHTHALMIC MINIATURE
"Such a day for cold you never did see, Finn," he said. "I came on a boy with warts standing in the snow. He was frozen to death nearly by the looks of him. His lips were so stiff he couldn't get a word out when I asked him why he didn't come in by the fire with the rest of us. Then I saw the reason plain as day. The cold had had him weeping and his tears had frozen hard clear to the ground. He was tethered there by his two eyes and would have perished surely if I hadn't broken the silver icy streams of his grief with a stick and freed him." Frederick Buechner, Brendan New York, Atheneum, 1987, p. 12
We studied the photographic records of 292 eyes, including 103 eyes with branch retinal vein occlusion, 90 fellow eyes, and 99 control eyes without branch retinal vein occlusion. All arteriovenous crossings within three disk diameters of the optic disk, including the crossings at the sites of branch retinal vein occlusions, were studied. The relative positions of the crossing artery and vein could be determined at 1,939 crossings in all eyes. Crossings at which a vein crossed over an artery were a common finding (22.3% to 33.0% of crossings), but were rare at the crossings where branch retinal vein occlusions were found (2.4%). A greater proportion of arterial overcrossings was found in eyes with branch retinal vein occlusions (77.7%) compared to fellow eyes (70.6%) or control eyes (67.0%). Our data indicate that arterial overcrossings are at relatively higher risk of branch retinal vein occlusion than venous overcrossings, and that the risk of branch vein occlusion in an eye is proportional to the number of arterial overcrossings in the eye.
The crossing of retinal vessels such that the artery lies over (that is, anterior to or innermost to) the vein is considered the normal anatomic configuration. In 1936, however, Jensen observed that venous overcrossings occur at 30% of all crossings in the retinas of normal eyes.' The presence of both types of crossings has been demonstrated histologically." We have rarely seen a branch retinal vein occlusion at an intersection at which the vein crosses over the artery (Fig. 1). References in published studies to similar occlusions are also rare.v" In one series that specifically addressed the anatomy of the crossings at 25 branch vein occlusions, no venous overcrossings were found at the sites of occlusion." We undertook this study to examine the anatomy of the retinal vessels at the sites of branch vein occlusions and to determine if it differs from the anatomy of other arteriovenous intersections in the same eyes or uninvolved fellow and control eyes.
T
The files at the retinal photography laboratories of Northwestern University and the New York Hospital, Cornell University from 1985 to 1988 were reviewed for patients with a diagnosis of branch retinal vein occlusion. Only vein occlusions that occurred within three disk diameters of the edge of the optic disk were studied. Central and hemispheric central retinal vein occlusions were excluded. One hundred and three eyes of 100 patients met these criteria. Using color transparencies, red-free photographs, and fluorescein angiograms of the involved and fellow eyes, the anatomy (artery over vein or vein over artery) of all photographed arteriovenous crossings within three disk diameters of the edge of the optic disk was studied. Intersections occurring on the surface
HE OBSERVATION that branch retinal vein occlusion occurs at arteriovenous intersections was made over 100 years ago by Leber.' The associations of branch retinal vein occlusion with systemic hypertension and with the retinal vascular manifestations of hypertension have been documented.v"
Accepted for publication Jan. 5, 1990. From the Department of Ophthalmology, Northwestern University Medical School, Chicago (Dr. Weinberg), and the Department of Ophthalmology, New York Hospital, Cornell Medical Center, New York (Dr. Dodwell and Mr. Fern). This study was supported in part by an unrestricted grant from Research to Prevent Blindness, Inc. Reprint requests to David G. Dodwell, M.D., Springfield Eye Center, 301 N. 8th St., Springfield, IL 62701.
298
Material
and Methods
©AMERICAN JOURNAL OF OPHTHALMOLOGY
109:298-302,
MARCH,
1990
Vol. 109, No.3
Branch Retinal Vein Occlusion
Fig. 1 (Weinberg, Dodwell, and Fern). Example of a branch vein occlusion at a venous overcrossing. Late frame of the angiogram shows staining of the vein as it crosses over the artery. of the optic disk were excluded. The vessel crossing over was defined as lying innermost or most anterior at the arteriovenous intersection. At the site of each vein occlusion, the anatomy was designated "artery over vein" or "vein over artery" if the positions of the vessels could be determined with certainty. If the positions of the vessels could not be ascertained, the site was designated "undetermined." Arteriovenous crossings other than the site of occlusion were designated "artery over vein" or "vein over artery" if the anatomy could be determined with certainty. The number of crossings at which the anatomy could not be determined was not tabulated for these crossings without vein occlusions. As a control population, the photographic records of 53 consecutive patients referred for fluorescein angiography with diagnoses other than branch retinal vein occlusion were studied. These eyes were analyzed exactly as the fellow eyes of patients with branch retinal vein occlusion. Statistical testing was performed using chisquare analysis. Statistical significance was defined as P < .05, unless otherwise stated.
Results Age ranged from 21 to 88 years (mean, 66.5 years), and 54 patients were men and 46 were women. The right eye was involved in 53 patients (51.5%), and the left eye was involved in
299
50 patients (48.5%). Three patients had bilateral branch retinal vein occlusions (two men and one woman). The locations (by quadrant) of the occlusions were as follows: 62 superotemporal (60.2%),39 inferotemporal (37.9%), one superonasal (1.0%), and one inferonasal (1.0%). The results are summarized in Figure 2 and the Table. At the site of the branch retinal vein occlusion, the artery lay anterior to the vein in 82 eyes (79.6%). The vein lay anterior to the artery in two eyes (1.9%). In 18 eyes (17.4%), the crossing was undetermined. In one eye (1.0%), the occlusion occurred at an anomalous vein as it exited from the edge of the disk rather than at an arteriovenous crossing. For the 84 vein occlusion sites at which the anatomy could be determined, 82 (97.6%) were arterial overcrossings and two (2.4%) were venous overcrossings. Determination of the anatomy could not be made in 18 eyes: seven eyes had media opacity, five eyes had hemorrhage, and in six eyes, the intersections were too small to characterize with certainty. The anatomy of a total of 728 arteriovenous crossings, including the 84 determinable branch vein occlusion sites, was tabulated in the 103
100 90 til
Ol
c
80
'iii
70
0
60
til
0 '0 C Q) ~
Q)
C.
50 40 30 20 10 0
BRVO sites BRVO eyes Fellow eyes Control eyes n=84 n=728 n=487 n=724
o ~
Venous overcrossings Arterial overcrossings
Fig. 2 (Weinberg, Dodwell, and Fern). Distribution of types of arteriovenous crossings among the groups. Unshaded and shaded areas represent percentage of venous and arterial overcrossings, respectively. BRVO sites indicates sites of occlusion in eyes with branch vein occlusion; BRVO eyes, all crossings in eyes with branch vein occlusion; fellow eyes, crossings in fellow eyes of those with branch vein occlusion; control eyes, crossings in eyes of patients without branch vein occlusion; n, the number of arteriovenous crossings at which the positions of the vessels were determined for each group.
300
March, 1990
AMERICAN JOURNAL OF OPHTHALMOLOGY
TABLE COMPARISON OF FEATURES OF ARTERIOVENOUS CROSSINGS NO. OF
NO. OF
NO. OF
NO. OF
DETERMINABLE
CROSSINGS
NO. OF ARTERIAL
NO. OF VENOUS
EYES
CROSSINGS
CROSSINGS
PER EYE
OVERCROSSINGS (%)
OVERCROSSINGS (%)
Branch retinal vein occlusion sites
103
102
82 (97.6)
2 (2.4)
Branch retinal vein occlusion eyes
103
7.1
566 (77.7)
162 (22.3)
84 728
Fellow eyes
90
487
5.4
344 (70.6)
143 (29.4)
Control eyes
99
724
7.3
485 (67.0)
239 (33.0)
eyes (mean, 7.1 per eye). Of the arteriovenous crossings, 566 (77.7%) were arterial and 162 (22.3%) were venous. The difference in frequency of arterial and venous overcrossings between the vein occlusion sites and all crossings in the involved eyes was statistically significant (P = .0001). Of the 103 branch retinal vein occlusions studied, six eyes of three patients had bilateral occlusions and were not included in the analysis of fellow eyes. In another seven cases, photographs of the fellow eyes were unavailable or inadequate for analysis, leaving 90 fellow eyes of 90 patients. Among these eyes, the anatomy of 487 crossings was determined (mean, 5.4 per eye). Fewer crossings per eye were counted in the fellow eyes than in the involved eyes because, in general, fewer photographic frames were available for the fellow eyes. The early frames of the fluorescein angiograms, which were useful for determining the anatomy in the involved eyes, were not available for the fellow eyes. Of the 487 crossings counted, 344 (70.6%) were arterial overcrossings and 143 (29.4%) were venous overcrossings. Compared to the vein occlusion sites and to the vein occlusion eyes, the differences were statistically significant (P = .0001 and .0063, respectively), with a greater prevalence of venous overcrossings in the fellow eyes. Of 53 consecutive patients referred for fluorescein angiography with diagnoses other than branch retinal vein occlusion, 99 eyes had photographic records adequate for study. The diagnoses were as follows: diabetes (16), agerelated macular degeneration (11), cystoid macular edema (three), optic neuropathy (three), macular hole (three), presumed ocular histoplasmosis (two), myopia (two), macular pucker (two), and other (11). In the last category, no diagnosis was represented more than once. Patient age ranged from 14 to 88 years (mean,
63.9 years), and 22 (41.5%) patients were men and 31 (58.5%) were woman. The anatomy of 724 crossings was determined in the 99 eyes (mean, 7.3 per eye). Of the crossings, 485 (67.0%) were arterial and 239 (33.0%) were venous overcrossings. The proportion of venous overcrossings in the control eyes was greater than in any other group. This difference was statistically significant compared to the vein occlusion sites (P = .0001) and the vein occlusion eyes (P = .0001), but not compared to the fellow eyes (P = .20).
Discussion The ages of the patients with branch retinal vein occlusion and the controls were comparable. We would not expect age to be an important variable, because the anatomy of the crossings is determined prenatally and should not influence longevity. The difference in gender between the branch vein occlusion eyes and control eyes was not statistically Significant (P = .19). The location of the occlusions was similar to those in previous series.v" with most occlusions in the superotemporal quadrant and most of the remainder in the inferotemporal quadrant. Branch retinal vein occlusions in the nasal quadrants were rare. Our data confirm the observations of Ienserr' that venous overcrossings are not a rare finding in normal fundi. Jensen used direct ophthalmoscopy to examine the eyes of 50 normal patients and found venous overcrossings at 30% of all arteriovenous intersections. Duker and Brown/ using color photographs and fluorescein angiograms, found no venous overcrossings at the site of occlusion in their series of 25 branch retinal vein occlusions. For controls, they observed a corresponding crossing in the opposite arcade (superior or inferior)
Vol. 109, No.3
Branch Retinal Vein Occlusion
in the same eye, and all first- and second-order crossings in one eye of 26 patients without branch retinal vein occlusions. The frequencies of venous overcrossings in the two groups were 35% and 32%, respectively. We used color stereoscopic photographs, redfree photographs, and fluorescein angiograms to examine 292 eyes, and we tabulated 1,939 arteriovenous intersections. We found venous overcrossings at two of 82 vein occlusion sites (2.4%). Venous overcrossings were present in 162 of 728 (22.3%) crossings in involved eyes, in 143 of 487 (29.4%) crossings in fellow eyes, and in 239 of 724 (33.0%) crossings in control eyes. If branch retinal vein occlusions occur randomly among all crossings, the proportion of venous overcrossings in a series of branch retinal vein occlusions should not differ from their frequency in the fundus overall (between 22% and 35% based on the data of jensen," Duker and Brown," and ourselves). We found the frequency of venous overcrossings lower and the frequency of arterial overcrossings higher at branch retinal vein occlusion sites than would be predicted based on their overall frequency in the involved eyes, fellow eyes, and control eyes. Thus, arterial overcrossings are at higher risk for branch retinal vein occlusion than venous overcrossings. The overall frequency of venous overcrossings increased from the involved eyes to the fellow eyes, and from the fellow eyes to the control eyes, with the highest incidence of venous overcrossings (and concomitantly lowest incidence of arterial overcrossings) in the control eyes (Fig. 2). This makes sense statistically, because eyes with a greater frequency of arterial overcrossings have more high-risk crossings, and thus would be expected to have a higher incidence of branch retinal vein occlusion. Several sources of bias in our data were considered. It is possible that the frequency of venous overcrossings at the site of vein occlusion was underestimated because of an overrepresentation of these crossings in the "undetermined" group. Even if half of the undetermined crossings were venous overcrossings, 91 of 102 occlusions (89.2 %) would be arterial overcrossings, and 11 of 102 occlusions (10.8%) would be venous overcrossings. These figures, although they almost certainly overestimate the frequency of venous overcrossings in the "undetermined" group, are still statistically significantly different compared to all other groups (branch retinal vein occlusion eyes, P = .011;
301
fellow eyes, P = .002; control eyes, P = .0001), with a lower frequency of venous overcrossings at sites of branch retinal vein occlusion. By including the occlusion site in the evaluation of the anatomy of all crossings in the involved eyes, bias may have been introduced, because particular effort was made to characterize these crossings, which had a different frequency distribution. By excluding the 84 determinable occlusion sites from the analysis of crossings in the involved eyes, the proportion of venous overcrossings increases from 22.3% (162 of 728) to 24.8% (160 of 644). This is still less than the proportion in the fellow eyes or the control eyes. The difference remains statistically significant compared to the control eyes (P = .0011), but not compared to fellow eyes (P = .10). Differences in the adequacy of the photographic records among the three groups resulted in differences in the mean number of crossings counted per eye. This may have introduced some bias if the locations of the two varieties of crossings are not randomly distributed. For instance, if one type of crossing tends to occur nearer to the disk, or at the intersections of larger vessels, it might be overrepresented in the fellow eyes, because fewer crossings were counted in these eyes, and those that were counted tended to be larger and nearer to the disk. We are unable to determine if such bias exists in our data. The higher relative risk of branch retinal vein occlusion at arterial overcrossings suggests a hemodynamic difference between arterial and venous overcrossings. We and others" have observed that characteristic hypertensive changes seen at arteriovenous crossings are less prominent at venous overcrossings than at arterial overcrossings. Seitz" has studied the histologic characteristics of both anatomic variations in normal and hypertensive patients. In the hypertensive patients, there was a thickening of the walls and a narrowing of the lumina of both vessels. At the site of the crossing, the vessels shared a common vascular wall and a common, thickened, adventitial and glial sheath. These changes were independent of which vessel was innermost. In both variants, the course of the artery was unchanged. The vein deviated around the artery, dipping deep into the retina in arterial overcrossings, and bulging against the internal limiting membrane in venous overcrossings. No compression of the underlying vessel was observed. Seitz attributed the more prominent clinical appearance of crossing phe-
302
AMERICAN JOURNAL OF OPHTHALMOLOGY
nomena (nicking and obscuration of the underlying vessel) in arterial overcrossings to the deeper position of the vein, rather than to any true compression of the vessel. Thus, by light microscopy, hemodynamic differences between the two anatomic configurations are not obvious. Perhaps there is a difference in the distensibility of the vein lying beneath the artery within the retina in an arterial overcrossing, as opposed to between the internal limiting membrane and the artery in a venous overcrossing. Such physiologic differences could explain the disparity in risk of branch retinal vein occlusion.
References
Handbuch der Gesammten Augenheilkunde, Pathologie und Therapie. Leipzig, Verlag Von Wilhelm Engelmann, 1877, pp. 521-535. 2. Gutman, F. A., and Zegarra, H.: The natural course of temporal retinal branch vein occlusion. Trans. Am. Acad. Ophthalmol. Otolaryngol. 78: 178, 1974.
3. Blankenship, G. W., and Okun, E.: Retinal tributary vein occlusion. Histology and management by photocoagulation. Arch. Ophthalmol. 89:363, 1973. 4. Jensen, V. A.: Clinical studies of tributary thrombosis in the central retinal vein. Acta Ophthalmol. 1 (sup pI. X): 1, 1936. 5. Seitz, R. (Blodi, F. c.. translator): The "crossing phenomenon." In The Retinal Vessels. St. Louis, C. V. Mosby, 1964, pp. 20-74. 6. Clemett, R. S.: Retinal branch vein occlusion. Changes at the site of obstruction. Br. J. Ophthalmol. 58:548,1974.
1. Leber, T.: Die Krankheite der Netzhaut und des
sehnerven. In Graefe, A., and Saernisch, T. (eds.):
March, 1990
7. Duker, J. 5., and Brown, G. c.: Anterior location of the crossing artery in branch retinal vein occlusion. Arch. Ophthalmol. 107:998, 1989.
OPHTHALMIC MINIATURE
"Such a day for cold you never did see, Finn," he said. "I came on a boy with warts standing in the snow. He was frozen to death nearly by the looks of him. His lips were so stiff he couldn't get a word out when I asked him why he didn't come in by the fire with the rest of us. Then I saw the reason plain as day. The cold had had him weeping and his tears had frozen hard clear to the ground. He was tethered there by his two eyes and would have perished surely if I hadn't broken the silver icy streams of his grief with a stick and freed him." Frederick Buechner, Brendan New York, Atheneum, 1987, p. 12
