Analysis of Retinal and Choroidal Circulation During Central Retinal Vein Occlusion Using Indocyanine Green Videoangiography

Objective  To explore the hemodynamic changes and their correlation with clinical presentation during central retinal vein occlusion.

Materials and Methods  Retrospective, 2-center study. The medical records of 27 patients with central retinal vein occlusion were reviewed. For each patient, the plasma transit in central retinal vessels and in peripapillary choroidal vessels was analyzed using indocyanine green videoangiography.

Results  The incidence of alteration in retinal plasma transit, ie, pulsatile filling of arteries and/or veins and increased arteriovenous filling time, was inversely correlated to duration from the onset but not to funduscopic features. Among the 14 patients with less than 1 month's duration of symptoms, 3 developed chronic macular edema, and impairment of arterial flow preceded its onset. Among the 10 patients with opticociliary circulation, choroidal drainage routes were identified in 5 cases, with pulsatile filling in 3.

Conclusions  Early in the course of central retinal vein occlusion, arterial and/or venous flow alterations are present, irrespective of visual acuity, vein dilation, or fundus hemorrhages. These alterations are less frequent in chronic than in recent-onset central retinal vein occlusion. The mechanisms of these alterations remain uncertain but may involve arterial constriction and/or intermittent venous compression. The relationship between these alterations in retinal flow and the secondary onset of macular edema or capillary nonperfusion deserves further investigation.

ALTHOUGH CENTRAL retinal vein occlusion (CRVO) frequently occurs, it is still poorly understood.¹ In specific clinical presentations the description of decreased plasma velocity in arteries^2,3 and the development of opticociliary collateral vessels are indications that complex retinal and choroidal hemodynamic changes occur during the course of CRVO. However, despite the frequency of the disease, few studies have dealt in detail with the natural history of retinal and choroidal hemodynamics. Retinal circulation during CRVO has been evaluated by dynamic angiography^4,5 or Doppler techniques.^6-11 The authors of these studies reported a decrease in arterial, capillary, and/or venous blood flow in the weeks following the onset of visual symptoms. In this study, we qualitatively and quantitatively analyzed the plasma transit in major retinal vessels and in the peripapillary choroid during CRVO using indocyanine green (ICG) videoangiography. In addition, we explored possible correlations between the functional and fundoscopic presentations on the one hand and retinal and choroidal plasma transit on the other in an attempt to clarify their clinical significance and to formulate a pathophysiologic hypothesis regarding the consequences of CRVO.

We reviewed the medical records of patients with CRVO who were examined at 2 university-based ophthalmology departments between January 1, 1999, and January 31, 2001. Central retinal vein occlusion was defined by the presence of scattered retinal hemorrhages and dilation of retinal veins in 4 quadrants. Among the patients seen, those who had undergone ICG videoangiography were selected for inclusion in our study. Their medical records were extensively reviewed by 2 of us (M.P. and V.G.). In patients with recent-onset CRVO, its duration was calculated from the date of occurrence of the first visual symptoms. All patients underwent a complete ophthalmological examination, including visual acuity (VA) measurement, slitlamp biomicroscopy, applanation intraocular pressure measurement, fundus examination and photography, and conventional fluorescein angiography. Optical coherence tomography (Humphrey Instruments, San Leandro, Calif) was performed in selected cases.

A dynamic angiogram of ICG transit in the peripapillary area was recorded for each patient using a digital fundus camera equipped with diode laser (795-nm) illumination (TRC 50 IAL; Topcon, Tokyo, Japan).¹² All of the patients examined gave informed consent to undergo this procedure. Blood pressure and pulse rate were measured beforehand, and patients with a blood pressure exceeding 160/100 mm Hg were excluded from the study. After pupil dilation, focusing was adjusted on the optic disc using a 20° examination field in the affected eye. In some cases, the tortuousness of veins outside the retinal plane and/or the instability of fixation did not allow a satisfactory focus on the retinal veins within the 20° field, and a 35° examination field was used. Five milligrams of ICG in 1 mL of a glucose solution was injected in less than 1 second into the antecubital vein using a syringe connected to tubing, while at the same time videotaping was started. Care was taken to avoid overexposure by modulating the illuminating power. To maintain corneal transparency, the patient was occasionally asked to blink. After complete washout of the dye from retinal vessels, ie, about 30 minutes later, the contralateral eye was examined, unless the first examination was considered unsatisfactory; in that case, the same eye was reexamined. The video sequences were recorded on 1 of 2 tape recorders (Hi8 or DV; both from Sony, Tokyo, Japan).

Fundoscopic features were evaluated on 50° field, red-free photographs by 2 independent observers (M.P. and V.G.) in a masked fashion using a semiquantitative scale. Retinal hemorrhages were classified as absent (0), moderate (1), numerous(2), or covering most of the retinal surface (3). Hemorrhages in the nerve fiber layer in temporal arcades were classified as absent (0), few (1), or numerous (2). Retinal vein tortuousness was classified as normal (0), moderate(1), severe but limited to the retinal plane (2), or severe with tortuous veins outside of the retinal plane (3). Papilledema was classified as absent(0), moderate with visible disc outlines (1), severe with no visibility of disc outlines (2), or diffuse papillomacular edema (3).

Videoangiograms were analyzed offline in a masked fashion. Of the 43 patients concerned, 16 were excluded because of the suboptimal quality of their angiograms. The most frequent causes of poor-quality angiograms were lacrymal film defects and/or cataract and/or unstable fixation. Qualitative analysis of the retinal and choroidal plasma circulation was performed by iteratively playing the video sequence on a monitor, either at normal speed or in slow motion. The degree of uneven filling of retinal vessels was assessed. An example of pulsatile filling of retinal arteries is shown in Figure 1. During the initial phase of venous filling, the boli of dye arising from side-branching venules were defined as the pulsatile venular outflow (PVO) (Figure 2). The motion of PVO was synchronous with the central retinal artery pulse, with an acceleration phase corresponding to the diastole of the central retinal artery and a deceleration during systole (see arrowhead in Figure 2). Pulsatile venular outflow was distinguished from physiologic venous pulsatility, because physiologic venous pulsatility is associated with variation in venous diameter.

After digitalization by iMovie software on a personal computer (iMacDV; Apple, Cuppertino, Calif), the arteriovenous filling time (AVFT), defined as the time elapsing between the first appearance of the dye in the central retinal artery and the complete filling of the temporal veins 1 disc diameter from the optic disc, was measured by frame-by-frame analysis. The AVFTs in the temporal veins were averaged. We used this protocol in an earlier series to assess the mean (SD) AVFTs of 17 normal eyes of 17 patients, which was 13.15 (2.2) seconds. For the latter series, the mean intersession difference was 1.8 seconds (6 eyes; 12 measures) (M. P., unpublished data, 19XX).

For statistical comparison, the Mann-Whitney, χ², and t tests were performed when appropriate, using StatView 4.0 software (Abacus Concepts, Berkeley, Calif). Differences were considered statistically significant when P<.05.

A 60-year-old man had loss of vision in the right eye that had started 7 days previously. His medical history was remarkable for arterial hypertension. When we saw him, his VA was 20/200 OD. Fundus examination showed the presence of dilated veins in 4 quadrants, generalized narrowing of arteries, scattered hemorrhages, and papillomacular edema. There was no capillary dropout. Optical coherence tomography showed the presence of the foveolar depression (Figure 3). Arteriovenous filling time was 20.2 seconds in the affected eye, and systolic retrograde PVO was noted (Figure 4). One month later, VA was counting fingers; fundus examination showed a decrease in vein tortuousness (Figure 5). Fluorescein angiography revealed conversion to ischemia combined with cystoid macular edema. Arteriovenous filling time was 17 seconds and PVO had disappeared. Panretinal photocoagulation was performed. At 6 months, VA was counting fingers OD.

A 50-year-old man had had a loss of vision in the left eye that had started 3 days previously. His medical history was remarkable for arterial hypertension. When we saw him, his VA was 20/20 OS. Fundus examination showed the presence of dilated veins in 4 quadrants and a moderate amount of hemorrhages(Figure 6). Using optical coherence tomography, the macular profile was normal. Fluorescein angiography did not reveal any blood-retinal barrier breakdown or capillary closure. Videoangiography showed that a cilioretinal artery filled 1.84 seconds before the central retinal artery, that PVOs were present, and that AVFT was 19.1 seconds. Two months later, VA was 20/50 OS, and fundus examination showed a decrease in vein tortuousness(Figure 7). Cystoid macular edema was present. Videoangiography demonstrated simultaneous filling of the central retinal and cilioretinal arteries, suggesting an improvement in retinal arterial perfusion. Arteriovenous filling time was 12 seconds; there was no PVO.

The recent-onset group (<4 weeks' duration of visual symptoms) included 14 eyes of 14 patients. The median duration of symptoms was 5.5 days (range, 1-23 days). Initial VA ranged from counting fingers to 20/20. Two patients(cases 10 and 13) had papillomacular edema. The others had no detectable rupture of the blood-retina barrier at the initial examination. All patients were intially considered as perfused, except for patient 5 who had numerous hemorrhages that prevented the appreciation of capillary perfusion status. On videoangiograms, pulsatile filling of retinal arteries and/or of disc capillaries was noted in 6 eyes (43%); in 2 of the 6 patients concerned, frame-by-frame analysis evidenced that arterial flow was reversed during diastole (case 9, Figure 1). Pulsatile venular outflow was noted in 13 eyes (93%); in 1 case venous flow was reversed during systole of the central retinal artery (case 10). Mean (SD) AVFT in eyes with CRVO(n = 14) was 24.5 (4.3) vs 12.5 (2.3) seconds in 8 contralateral eyes (P = .001, Wilcoxon rank sum test). No specific choroidal abnormalities were detected.

Ten of these 14 patients were examined repeatedly, and follow-up data were obtained for 3 others. Eight patients (cases 1, 2, 6-9, 11, and 13) recovered completely without functional sequelae. In 3 patients (cases 4, 10, and 14) fundus hemorrhages, vein tortuousness, and videoangiographic features improved, but chronic macular edema occurred in patient 10 with concomitant conversion to peripheral ischemia. Patient 12 had a final VA of 20/200; patient 5 was lost to follow-up.

The group with more than 1 month's duration of symptoms was composed of 13 eyes of 13 patients. Median duration of symptoms was 9 months (range, 30 days to 8 years). Visual acuity ranged from counting fingers to 20/25. Eleven patients had macular edema, and 2 had ischemic CRVO. On videoangiograms, none had pulsatile filling of arteries, but PVO was detected in 3 patients. In 10 patients with opticociliary collateral circulation, the drainage route of the collateral vessels in the choroid was clearly visualized in 5 patients(Figure 8); 3 had pulsatile filling of these opticociliary vessels. The mean (SD) AVFT was 15.8 (3.6) seconds in eyes with CRVO (n = 14), and 13.1 (1.8) seconds in 5 contralateral eyes(P = .08, Wilcoxon rank sum test). The 3 patients with ischemic CRVO (cases 10, 19, and 20) had no detectable abnormalities of dynamic angiography, apart from a moderately increased AVFT. The clinical and angiographic characteristics of all patients are given in Table 1.

In the recent-onset group, the AVFT (P<.001, Mann-Whitney test) and the incidence of pulsatile filling of arteries (P = .04) and of PVO (P<.001, Fisher exact test) were significantly higher than in the older-onset group. There was no significant correlation between AVFT and duration of symptoms when all patients were included in the analysis. However, when the patients with more than 18 months' duration of CRVO were excluded, AVFT in the affected eye correlated strongly with the duration of symptoms (r = −0.79, P<.001, Figure 9). No correlation was found among the number of fundus hemorrhages, degree of vein tortuousness, or papilledema on the one hand and presence of pulsatile filling of arteries or PVO on the other. The presence of pulsatile filling of arteries correlated with the presence of PVO (P = .05, Fisher exact test); indeed, all patients with pulsatile filling of arteries had PVO. The association of PVO with increased AVFT had strong statistical significance (P<.01, Mann Whitney test), whereas the association of pulsatile arterial filling with increased AVFT was of borderline significance (P = .08). Among recent-onset cases, there was no correlation between the presence of arterial or venous flow alteration and final outcome.

The nature of the retinal hemodynamic changes consecutive to CRVO is not yet clear. Echo Doppler studies have led to conflictual results; compared with normal eyes, central venous velocity during CRVO was decreased^7,9 venous outflow resistance was either decreased⁷ or increased,^6,9 and the velocity in the central retinal artery was found to be either decreased^7-9 or normal.¹⁰ By laser Doppler velocimetry, a progressive decrease in flow resistance was detected in a small series of patients with CRVO of less than 2 months' duration,¹¹ that is increased venous diameter combined with constant velocity. Williamson and Baxtor⁷ indicated that systolic velocity in the central retinal vein was a strong predictor of conversion to ischemia. Yet the clinical relevance of these results remains uncertain since the correlation to clinical features has received little attention. Because the use of dynamic angiographic techniques is becoming more widespread, the definition of a clinically applicable scheme of videoangiographic interpretation during CRVO is of interest. To our knowledge, only one group of authors reported the videoangiographic alterations during the course of CRVO.^4,5 In the present study, we noted that duration from onset is the strongest clinical feature associated with altered plasma transit. Indeed, compared with chronic CRVO, there is, in its acute phase, a significantly higher incidence of pulsatile arterial filling and of PVO, and the AVFT is significantly higher. Conversely, no association was evidenced here between functional or funduscopic features and dynamic angiographic features.

The pathologic significance of these plasma transit alteration remains speculative. The pulsatile filling of arteries may be due to a decrease in perfusion pressure, an increase in arteriolar resistivity, or the transmission of an elevated venous pressure through the capillaries. Even if our data do not allow us to determine which of these hypotheses is the correct one, in most patients with pulsatile filling of arteries, the venous dilation was moderate. Therefore, we suggest that the most likely explanation for pulsatile filling of arteries is an increase in arterial resistivity, either within the retina or in the optic nerve. Follow-up study of the vessels' diameters should help to assess this point. In some of the patients with PVO the boli of dye were seen arrested within the vein lumen during arterial systole (arrowhead in Figure 2) suggesting an extremely slow venous flow during arterial systole. In patient 10, PVO was associated with reverse flow during the systole of the central retinal artery (Figure 4). Since this reflux was synchronous to the arterial systole, the most likely explanation is that the central vein is compressed by an artery, most probably the central retinal artery, within the optic nerve. Finally, PVO was associated with increased AVFT. Pulsatile venular outflow, thus, appears correlated to a decrease in venous blood velocity. Overall 7 of 14 patients with recent-onset CRVO presented evidence of retinal arterial dysfunction, pulsatile filling of arteries, constriction, delayed filling, and/or systolic compression of the central retinal vein.

Alterations in retinal plasma transit, whatever their nature, were inversely correlated with the duration of symptoms but not with other clinical features, ie, VA, retinal hemorrhages, or venous dilation. Indeed, the patients with what is known as impending CRVO, ie, a few retinal hemorrhages accompanying venous dilation and preservation of VA (patients 1, 2, 6, 7, and 11), had pulsatile filling of arteries in 2 cases, and all of them had PVO. In CRVO, retinal perfusion impairment detectable by dynamic angiography occurs early in the course of recent-onset CRVO. The present results are in accord with the findings of Wolf et al and Remky et al, who reported an early decrease in plasma velocity both in major vessels⁴ and in macular capillaries,⁵ respectively, again not correlated to fundus features or VA. Our results also agree with the echo Doppler findings of lower velocities and a higher resistive index in the central retinal artery in CRVO of less than 3 months' duration. The presence of increased arterial resistance in the initial phase is also in accord with experimental data indicating that shortly after the occurrence of a branch vein occlusion, arterial constriction is present.¹³ The resulting decrease in blood flow may be linked to a decrease in the release of nitric oxide by the glial cells.¹⁴ To our knowledge, the arterial constriction consecutive to CRVO and the consequences of acute hypoperfusion on retinal metabolism in humans has received little attention. We did not find any correlation between hemodynamic features and final visual outcome, probably because of the few cases studied. However, in our 3 patients with CRVO who ultimately developed cystoid macular edema, the initial videoangiographic features suggested that impaired arterial flow preceded macular edema; one had generalized arterial constriction (patient 10); another, pulsatile arterial filling (patient 14); and the last, delayed filling of retinal arteries (patient 4). The examination of more cases is needed to identify prognostic factors based on hemodynamic features.

None of the patients with CRVO of more than 1 months' duration had a detectable arterial pulse, and despite the presence of vein dilation in all cases, videoangiographic features were normal in most patients, including the 3 cases of ischemia. Therefore, the specific contribution of videoangiography to the analysis of retinal circulation in CRVO of more than 1 month's duration remains uncertain. Chronic CRVO may lead to the development of collateral vessels and dilation of choroidal veins. In a previous ICG angiographic study of collateral vessels after CRVO, the drainage route could be identified in 10 of 13 cases.¹⁵ In our study, the drainage route in the choroid was identified in 5 of 10 cases. In addition to this, ICG dynamic angiography documented the flow in the collateral circulation, which seemed to be impaired in 3 cases.

As regards the exploration of hemodynamics during CRVO, dynamic angiography and echo Doppler each have their own advantages and limitations. Videoangiography allows more detailed examination of the retinal and choroidal circulation than echo Doppler, but retinal hemorrhages may alter the quality of angiograms, and quantification remains presently outside the scope of routine examination, except for the measurement of arteriovenous filling time. Echo Doppler, on the other hand, allows routine quantitative measurement whatever the clinical presentation, but the flow in the central veins is not detectable in a significant number of cases,⁷ the measurements may be altered by the derivation of flow in the collateral circulation, and marked interobserver variability has also been reported.¹⁶ Therefore, these methods seem to be complementary, and in the future it should be of interest to compare, in the same patient, the analysis of arterial and venous flow obtained by each one, to improve our understanding of the hemodynamic alterations occurring upstream of the site of occlusion.

Dynamic ICG angiography is a valuable tool to explore the hemodynamic abnormalities occurring during CRVO. Early in the course of CRVO, arterial and/or venous flow alterations are present, irrespective of VA, vein dilation, or fundus hemorrhages. The mechanisms underlying these hemodynamic alterations remain speculative. Pulsatile filling of retinal arteries may be linked either to arterial constriction or to the retrograde transmission of elevated venous pressure. The presence of retrograde systolic venous flow could suggest the presence of an intermittent compression of the central retinal vein. These alterations are less frequent in chronic CRVOs than in recent-onset ones. In opticociliary vessels, pulsatile flow could be observed, which was suggestive of low perfusion. The relationship between the initial alteration in retinal flow and the onset of neuroretinal complications requires further investigations.

Accepted for publication August 3, 2001.

We thank Topcon France for providing the retinal camera used in this study.

Corresponding author and reprints: Michel Paques, MD, PhD, Department of Ophthalmology, Hôpital Lariboisière, 2 rue Ambroise Paré, 75745 Paris CEDEX 10, France (e-mail: michel.paques@lrb.ap-hop-paris.fr).