Introduction

Central retinal artery occlusion (CRAO) usually causes abrupt and irreversible visual loss. Although multiple therapeutic regimens have been proposed for it, the effect seems to be limited. The visual prognosis for most patients was poor. Visual acuity increased by three or more lines in only 15% of the patients with or without treatment.1 Selective intra-arterial lysis has been proposed for the treatment of CRAO in recent years. A meta-analysis of the published data showed that there may be marginal visual benefit, and it has potential serious adverse effects, such as embolization of other parts of the vascular system, which might lead to stroke or even death.2 In this study, vitreous surgery with direct central retinal artery massage as a new technique for the treatment of CRAO was investigated. To our knowledge, there have been no similar reports in the literature previously.

Patients and methods

Patients

This study was approved by the Committee Review Board of Beijing Tongren Hospital. It consisted of 10 consecutive patients with acute CRAO who were treated with vitreous surgery with direct central retinal artery massage between October 2005 and April 2007 in Beijing Tongren Hospital. Informed consent was obtained from all the patients before treatment.

All the eyes included in our study showed fundus features typical of acute CRAO, with abrupt painless visual loss. Criteria of exclusion: CRAO of more than 6 days' duration at initial examination; patients with a foveal sparing cilioretinal artery; the caliber of the retinal vessels had returned to normal and angiographic images revealed a normal retinal circulation time; initial visual acuity better than counting fingers.

Before the surgery, best-corrected visual acuity, anterior and posterior segments of the eye were examined. Photography of the posterior pole of all the patients and fundus fluorescein angiography (FFA) in some patients were performed. Laboratory tests were conducted for blood clotting, electrolytes, and blood sugar.

Ten patients with an average CRAO period of 73 h (24 h to 6 days) were enrolled. The age of the patients ranged from 34 to 70 years, with an average of 48 years (Table 1). Nine patients had a history of arterial hypertension, in which, one had a history of stroke 2 years ago, and three had a history of myocardial infarction. These patients were under drug therapy for high blood pressure. Other history of special medical treatment was denied. Only one patient had no history of systemic diseases (Table 1). Temporal arteritis was excluded by examining the blood sedimentation rate.

Table 1 Clinical characteristics of the patients with CRAO and visual changes before and after treatment
Full size table

Retinal edema with cherry red spot and narrowed retinal arteries were found in all the cases (Figure 1a). Initial visual acuity in these cases ranged from no light perception to counting fingers (Table 1). Emboli of the central retinal artery were not detected in these cases. Small yellow-white emboli were found in the retinal arterioles but not on the optic disc in two cases. FFA was performed in five patients. The arteriovenous transit times ranged from 18 to 311 s, with an average of 138 s.

Figure 1
figure 1

(a) Colored fundus photograph of a 44-year-old woman with central retinal artery occlusion for 24 h and visual acuity of hand movements. The retinal arterioles are severely narrowed, and there is heavy retinal edema. Small patch of normal looking retina was seen near the optic disc due to an efficient cilioretinal arteriole, but the macular was not spared. (b) First day after operation, the caliber of the retinal arterioles looks greater, especially at the macular region. (c) Fifth day after operation, the retinal vessels look normal and the retinal edema decreased prominently. (d) Thirty days later, although the retinal edema subsided and visual acuity improved to 20/200, thinned retinal arterioles are also left with optic disc pallor.

Full size image

Technique of vitreous surgery with direct central retinal artery massage

After a standard 3 port pars plana vitrectomy, a special probe made of nickel titanium and extended from a 19 gauge support shaft (Figure 2) was put on the central bifurcation of the retinal artery, and slight pressure was applied until entire flow cessation of the artery occurred, followed by 2 s of release. This was repeated for 5–6 times. If small pieces of embolus were seen moving downstream from the central retinal artery and the caliber of the retinal vessels increased, then the operation could be finished. If no change of the retinal circulation occurred after this procedure, a microvitreoretinal blade was used to make a small incision along the central retinal artery in the optic nerve head, and then the special probe was introduced carefully into the incision and advanced into the optic nerve head toward the lamina cribrosa along the central retinal artery. The site of the incision was carefully selected to avoid major branches of the retinal arterioles and venules. As the probe moving toward the lamina cribrosa, slight pressure was applied against the central retinal artery until entire blood flow cessation occurred, followed by 2 s of release. This procedure was repeated for 5–6 times. Approximately 3–4 mm of the probe entered the optic nerve head. If the caliber of the retinal arteries increased, then the operation could be finished. If no change of the retinal circulation occurred following this procedure, then a sclera depressor was used to press the optic nerve head until slight deformation of the optic nerve head was observed under the surgical microscope, followed by 2 s of release. Again, this procedure was repeated for 5–6 times before the surgery was completed. This procedure was achieved by inserting a sclera depressor retrobulbarly through the superonasal conjunctival incision. The whole movement of the sclera depressor along the superonasal surface of the sclera toward the optic nerve head was observed under the surgical microscope.

Figure 2
figure 2

(a) Fundus photograph of a patient during operation. The arrow indicates the probe being used for direct central retinal artery massage. (b) Fundus photograph of the probe (patent has been obtained).

Full size image

Patient follow-up

Follow-up examinations were performed at 24 h, 48 h, 1 week, 2 weeks, 3 weeks, 4 weeks and 2 months after intervention. This included test of the best-corrected visual acuity, fundus photography, and FFA.

Results

Ten patients with acute CRAO were treated by vitreous surgery with direct central retinal artery massage. In four cases, small pieces of yellow-white embolus were seen dislodged from the central retinal artery and moved downstream along one branch of the retinal arterioles toward the periphery of the fundus following the direct central retinal artery massage. At the same time, the vessel caliber of both the retinal artery and vein increased, indicating that the retinal blood flow improved. The central retinal artery became narrowed again 2 days after operation in one patient, but reopened on the third day postoperatively. In other six cases, no change of retinal circulation was found during the operation, among which, retinal circulation improved in four cases 24–48 h after operation, remained the same in two cases. The arteriovenous transit times on FFA performed within the first 2 days postoperatively were normal or nearly normal as shown in Table 2 (normal arteriovenous transit times were ⩽11 s).3 FFA was not performed for the two patients with no change in the ophthalmoscopic appearance. During follow-up visits, retinal artery occlusion did not recur in the eight cases with improved retinal circulation (Figures 1, 3 and 4). Visual acuity improvement occurred within the first week after operation in most patients. In some patients, visual acuity gradually improved during the first 2 months postoperatively (Table 1).

Table 2 Characteristics of the patients with CRAO during and after operation
Full size table
Figure 3
figure 3

Fundus fluorescein angiograph of the same patient as in Figure 1. (a) Before operation, at 12 s after injection, the cilioretinal arterioles are filled with dye, but the central retinal artery is not filled yet. (b) Before operation, at 33 s after injection, there is retinal vascular hypoperfusion. (c) On the first day after operation, at 19 s after dye injection, the central retinal artery, retinal arterioles, and the cilioretinal arterioles are filled with dye. (d) On the first day after operation, at 32 s after injection, the retinal vessels are well filled with dye.

Full size image
Figure 4
figure 4

Late stage postoperative fluorescein angiograph of the same patient as in Figure 1. At 10 min after injection, there is slight staining of fluorescence at the operation site on the optic nerve head.

Full size image

During the operation, localized retinal hemorrhage on the optic nerve head occurred in most cases due to injury of the capillaries at the operation site, which was controlled by increasing the infusion pressure. Postoperative FFA showed slight staining of fluorescence at the operation site on the optic nerve head (Figure 4). In one case, central retinal vein occlusion occurred 5 days after operation. Iris neovascularization occurred 20 days later (Table 2). Cryotherapy of the peripheral retina for 360 degrees was performed and the neovascularization of the iris regressed gradually after operation. No other serious complication occurred.

Discussion

Clinical and experimental studies have indicated that there was a time limit for CRAO to cause irreversible damage to the retina. If retinal circulation restored within this time limit, retinal function could recover. In primates, the time limit was 105 min.4 In the clinic, this time limit could be much longer, because the occlusion in CRAO is seldom complete.5 It was reported that vision recovered to different degrees within 12–33 h after CRAO.5,6 There was even a report that showed that vision could recover after 4 days of CRAO.7 In our study, for all the patients with an occlusion time of 4 days or less, visual acuity improved in the cases that retinal blood circulation restored after surgery. For the patient with 6 days of occlusion, visual acuity remained the same although retinal circulation restored postoperatively (Table 1). We cannot comment on the correlation between the occlusion time and the degree of visual recovery, as the number of cases is too small.

The common cause of CRAO is an embolus lodging in the central retinal artery. The actual position of the embolus is difficult to determine. It may be immediately posterior to the lamina cribrosa5 or more posterior.8 In our study, pressing the central retinal artery in the optic nerve head with a probe or retrobulbarly with a sclera depressor led to deformation and bending of the blood vessel wall. The vessel wall of the retinal artery is elastic, therefore, slight deformation and bending may not cause any damage to the vessel wall. The emboli were made of cholesterol in 74%, calcified material in 10.5%, and platelet-fibrin in 15.5%.9 The emboli may be less flexible and more fragile in comparison with the wall of the retinal arteries, so during the mechanical movement of the vessels induced by the probe, the embolus may be broken into smaller pieces and then be dislodged. During the operation, emboli were seen dislodged and moving downstream with immediate restoration of retinal circulation after the direct central retinal artery massage in four of our cases, indicating that mechanical movement may be effective to dislodge the embolus. In the other four cases, no embolus was seen during the operation, but the retinal circulation improved 24–48 h postoperatively. This may be because the emboli were more adherent to the vessel wall or less fragile, which made it more difficult to be dislodged immediately. However, the direct retinal artery massage may had loosened the embolus and made it readily dislodged or absorbed gradually, so 24–48 h after operation, retinal circulation restored. For two cases, circulation did not recover, this may be due to that the position of the emboli was further posterior to the optic nerve head or there was no embolus at all.

The diameter of the probe was 0.1 mm, similar to the diameter of the central retinal artery (Figure 2). It was advanced into the optic nerve head in a parallel direction to the optic nerve fiber, and was pressed directly upon the blood vessel wall, so it may cause little damage to the optic nerve. One day after operation, there was only slight staining at the operation sites on FFA, and most patients had an increased visual acuity during follow-up study (Figure 4). This indicated that the operation caused no or negligible damage to both the central retinal artery and the optic nerve. Only one patient developed central retinal vein occlusion 5 days after operation. This may be due to damage of the central retinal vein during the operation. As retinal circulation did not recover in this case during follow-up study, slow blood flow in the vein might gradually caused thrombus formation at the site of injury, and finally resulted in central retinal vein occlusion. This indicated that the operation had a potential risk of central retinal vein injury.

Visual acuity improved by three or more lines in 6 out of 10 cases (60%). Forty percent of our patients had a final visual acuity of 20/200 or better (Table 1). The visual outcome was better than that of conservative forms of treatment or natural course of the disease.1 Our study showed that vitreous surgery with direct central retinal artery massage was relatively safe and may be effective in restoring blood circulation in patients with CRAO and improving the visual prognosis. Limitations of our study included: noncomparative nature, limited number of patients, and delayed treatment due to patient referral. A randomized controlled trial is necessary to prove the effectiveness of the treatment.