Prognostic factors for acute central retinal artery occlusion treated with hyperbaric oxygen: The Hong Kong study report number five

Abstract

BACKGROUND

Central retinal artery occlusion (CRAO) is a potentially blinding disease, and hyperbaric oxygen therapy (HBOT) is becoming increasingly popular with the support of scientific evidence. Despite the presence of various acute management measures, there is no clear evidence on the gold standard treatment for CRAO.

AIM

To identify factors and imaging parameters associated with good visual outcome, which guide ophthalmologists in the triage of CRAO patients for HBOT.

METHODS

Patients who suffered from CRAO and had a symptom onset ≤ 6 h were recruited for a course of HBOT in a tertiary hospital after failing bedside treatment. Patient demographics, onset time, CRAO eye parameters, and past medical history were prospectively collected. Visual outcomes after HBOT were also analyzed.

RESULTS

A total of 26 patients were included; the female-to-male ratio was 1:1.6, and the mean age was 67.5 years ± 13.3 years (range 44–89 years). The mean duration of follow-up and mean visual acuity (VA) improvement were 10.0 mo ± 5.3 mo and 0.48 logarithm of minimal angle of resolution (logMAR) ± 0.57 logMAR (approximately 9 letters in ETDRS) (P = 0.0001, Z = -3.67), respectively. The 1 mm zone of central macular thickness (CMT) on optical coherence tomography was not associated with VA changes (P = 0.119); however, the 1-to-3 mm circular rim of CMT was fairly associated (P = 0.02, Spearman's coefficient = 0.45). Complete retinal perfusion time during fundus fluorescein angiography (FFA) was moderately associated (P = 0.01, Spearman's coefficient = 0.58) with visual outcome.

CONCLUSION

A thinner 1-to-3 mm circular rim of CMT, but not the central 1 mm zone, is associated with better visual outcome. A shorter perfusion delay on FFA is also associated with better visual outcome.

Key Words: Central retinal artery occlusion; Fundus fluorescein angiography; Hyperbaric oxygen therapy; Optical coherence tomography; Stroke

Core Tip: Patients’ eyes with central retinal artery occlusion demonstrated a mean visual acuity improvement of 0.48 logarithm of minimal angle of resolution (logMAR) ± 0.57 logMAR (approximately 9 letters in ETDRS) after hyperbaric oxygen therapy. A thinner 1-to-3 mm zone of central macular thickness, but not the central 1 mm zone, was associated with better visual outcome. A shorter perfusion delay on fundus fluorescein angiography was also associated with better visual outcome.

INTRODUCTION

Central retinal artery occlusion (CRAO) is a potentially blinding ophthalmic emergency[1], also known as ocular stroke[2]. Despite the presence of various acute management measures, including breathing into a paper bag, carbogen inhalation[3], topical and/or systemic medical treatment for lowering intraocular pressures, ocular massage, anterior chamber paracentesis[4], or even thrombolytic therapy[5], there is no clear evidence on the gold standard treatment for CRAO[6].

With the support of scientific evidence[7,8], hyperbaric oxygen therapy (HBOT) for CRAO was started in Hong Kong in November 2018[9]. This HBOT center is situated in a tertiary hospital that receives referrals from both public and private practitioners 24 h every day on a territory-wide basis. During the coronavirus disease 2019 (COVID-19) pandemic, the HBOT service is still continuously serving more than 7 million people in the city[10,11]. A study on the outcome of this novel treatment in Hong Kong was established, which is called the HBOT for CRAO (HORA) study[12]. Previously we reported the absence of severe acute respiratory syndrome coronavirus 2 among our cohort, and delayed hospital presentation of CRAO patients during the local COVID-19 crisis[12,13]. This study aimed to identify prognostic factors associated with visual outcome, which might guide ophthalmologists in the triage of CRAO patients for HBOT.

MATERIALS AND METHODS

The management of patients in this prospective study adhered firmly to the tenets of the Declaration of Helsinki, and the research ethics committee approval number was HKECREC-2020-116. CRAO patients with ≤ 6 h of symptom onset were first given emergency bedside treatments, such as ocular massage, topical and systemic intraocular pressure lowering medications, rebreathing into paper bags, etc. In addition, if all possible means failed to reverse the CRAO, they were recruited for a course of HBOT at a tertiary hospital[9]. The first session of HBOT would follow the United States Navy Treatment Table (USNTT) 5. The protocol started with slow pressurization to a maximum pressure of 2.8 atmospheres absolute (ATA), which was maintained for 45 min, followed by slow depressurization to 1.9 ATA, which lasted for 30 min. The total treatment time was approximately 141 min. The subsequent planned 9 HBOT sessions (twice daily) followed the USNTT 9. A protocol with a maximum pressure of up to 2.4 ATA was used for 100 min with slow depressurization to level ground pressure lasting for 14 min. The total treatment time was approximately 124 min. HBOT might be terminated early if the patient could not tolerate treatment, refused further treatment, or was complicated with other medical conditions such as seizure and further cerebral stroke etc. Basic ophthalmic examinations including visual acuity (VA), intraocular pressure, slit lamp and dilated fundus examinations, were performed by ophthalmology specialists recognized by the College of Ophthalmologists of Hong Kong to confirm the diagnosis[14]. Eyes with pathologies other than CRAO were excluded from the HBOT and HORA study. Eyes with concurrent retinal vein occlusion, diabetic macular edema, age-related macular degeneration, pigment epithelial detachment, central serous chorioretinopathy, retinal detachment, hereditary retinal or macular dystrophy, epiretinal membrane, vitreomacular traction, lamella hole, macular hole, giant cell arteritis patients, or CRAO cases with cilioretinal artery spared were all excluded. Optical coherence tomography (OCT) and fundus fluorescein angiography (FFA) were performed by vitreoretinal practitioners to evaluate the severity of CRAO. Each CRAO patient was managed using a multidisciplinary approach together with internal medicine and emergency department HBOT practitioners.

Patient data were retrieved from the electronic health records system held by the Hospital Authority[15,16]. Patient demographic information (age, gender, ethnicity), follow-up duration, onset-to-attendance time, symptom-to-HBOT time, diseased eye characteristics (laterality, best corrected VA, OCT parameters, FFA perfusion time in seconds), past ophthalmic history, and past medical and drug history (any brands of anti-platelet/anti-coagulant use) were retrieved. The best corrected VA was measured using Snellen’s charts and converted to the logarithm of minimal angle of resolution (logMAR) unit for analysis[17]. The following logMAR denotations were used for non-numerical VA measurements: (1) Finger count = 1.7 logMAR; (2) Hand movement = 2.0 logMAR; (3) Light perception (LP) = 2.3 logMAR; and (4) No LP = 3.0 logMAR.

OCT and FFA were performed with images captured by Heidelberg Spectralis (Heidelberg Engineering, Heidelberg, Germany), and a central macular thickness (CMT) map was generated from 19 horizontal line scans (Figure 1A). The central 1 mm zone of CMT data in micron was directly extracted from the printout (Figure 1B), whereas the circular rim of 1-to-3 mm circular rim of CMT was calculated by averaging the numerical data shown for each quadrant (superior, temporal, inferior, nasal) of the circle (Figure 1C).

Figure 1 An optical coherence tomography scan report. A: An optical coherence tomography scan report example with machine generated thickness map for a left eye suffering from acute central retinal artery occlusion; B: The encircled central 1 mm zone thickness data was extracted from the printout; C: The highlighted area indicates the central 1-to-3 mm zone on the thickness map.

FFA was stopped 15 min after the injection of dye if retinal artery perfusion was still not observed, and in such cases, the perfusion time would be marked as > 900 s. Post-HBOT VA outcomes and HBOT-related adverse events and complications were reviewed. Normal distribution of the data was tested by the Shapiro-Wilk test given the small sample size, and statistical analyses for the non-parametric data were conducted using Wilcoxon, Fisher’s exact, and Spearman's correlation tests via Statistical Package for the Social Sciences version 27 (IBM, Armonk, NY, United States)

RESULTS

A total of 37 CRAO patients attended our hospital during the study period, of which 11 were out of our treatment timeframe of 6 h (Figure 2). Eventually, 26 patients were included, and the female-to-male ratio was 1:1.6, with a mean age 67.5 years ± 13.3 years (range 44–89 years). Only one patient was not Chinese. There were no patients lost to follow-up. Only 7 patients were too ill to undergo FFA, or refused FFA; otherwise there were no missing data for other parameters. The mean follow-up period and mean best corrected VA improvement were 10.0 mo ± 5.3 mo and 0.48 logMAR ± 0.57 logMAR (approximately equal to 9 letters of improvement in ETDRS, details in Table 1) (P = 0.0001, Z = -3.67), respectively. The data were not normally distributed; hence, non-parametric tests were used. Age (P = 0.49), pre-HBOT VA (P = 0.42), anti-platelet/anti-coagulant medication history (P = 0.42), onset-to-attendance time (P = 0.36), and symptom-to-HBOT time (P = 0.42) were not correlated with VA outcomes. The mean, SD, and median of each parameter are listed in Table 1.

Figure 2  Flowchart of patient recruitment.

Table 1 Data on baseline parameters investigated, n (%).

	Mean	Inter quartile range	SD	Median
Age (years)	67.5	22	13.3	70.5
Pre-HBOT VA (logMAR)	2.0	0.4	0.3	2.0
Pre-HBOT VA (Snellen acuity)	Hand movement	--	--	Hand movement
Post-HBOT VA (logMAR)	1.5	0.4	0.7	1.7
Post-HBOT VA (Snellen acuity)	20/640	--	--	Finger count
VA changes (logMAR)	-0.48	0.65	0.57	-0.3
VA changes (ETDRS letter score)	9	--	--	3
Onset-to-attendance time (min)	139	157	116	105
Symptom-to-HBOT time (min)	789	511	450	585
1 mm zone of CMT (µm)	334	96	64	332
1-to-3 mm circular rim of CMT (µm)	430	76	51	437
Fundus fluorescein angiography complete retinal perfusion time (s)	221	177	212	111
	Anti-platelet	Anticoagulant	Nil
Anti-platelet/anticoagulant history	8 (30.8)	2 (7.7)	16 (61.5)

CMT: Central macular thickness; HBOT: Hyperbaric oxygen therapy; LogMAR: Logarithm of minimal angle of resolution; VA: Visual acuity.

Concerning OCT parameters, there were no failed scans, and thickness maps were successfully generated for all patients. The 1 mm zone of CMT (334 µm ± 64 µm) was not associated with VA changes (P = 0.119), but the 1-to-3 mm circular rim of CMT (430 µm ± 51 µm) was fairly associated (P = 0.02, Spearman's coefficient = 0.45) with a positive correlation, i.e., the thicker the CMT was, the less negative the logMAR VA change. By categorizing the VA changes into two groups, “responder” for those with VA improvement after HBOT and “non-responder” for those without any VA improvement, 18 responders and 8 non-responders were identified. The mean 1 mm zone of CMT for responders and non-responders was 332 µm ± 70 µm and 338 µm ± 51 µm, respectively. For the 1-to-3 mm circular rim of the CMT, the means for responders and non-responders were 423 µm ± 59 µm and 447 µm ± 24 µm, respectively. A full dose of 5 mL of 10% Fluorescite (Alcon Laboratories, Fort Worth, TX, United States) was given to each patient who underwent FFA, and the complete retinal perfusion time (221 s ± 212 s, range 44–710 s) was moderately associated (P = 0.01, Spearman's coefficient 0.58) with visual outcome.

One patient experienced cerebrovascular stroke shortly after being diagnosed with CRAO, and HBOT had to be terminated after 2 sessions for the treatment of stroke. No other ischemic cerebrovascular events, nor contralateral eye CRAO, were detected during the follow-up period. Other HBOT-related complications included hypoglycaemia in 3 patients (11.5%), 2 of whom were known to have diabetes mellitus before the CRAO; and barotrauma in 4 patients (15.4%). All these patients continued to finish the whole course of 10 sessions of HBOT (Table 2).

Table 2 Complications after hyperbaric oxygen therapy.

	n	Percentage
Cerebrovascular stroke	1	3.8
Hypoglycaemia	3 (2 with diabetes mellitus)	11.5
Barotrauma	4	15.4

DISCUSSION

HBOT is based on the theory that hyperbaric oxygenation of the choroidal circulation helps to perfuse the inner retina while allowing the retinal circulation to be restored by different means. The HORA study is the first territory-wide study in Hong Kong that provided data on the prognostic factors for CRAO patients. There was no loss to follow-up in our patients which minimized bias.

Our study demonstrated a comparable VA gain in CRAO patients treated with HBOT to that in previous series at different countries of approximately 0.53 logMAR[18,19]. However, we did not find a significant correlation between symptom onset to HBOT time and VA outcome, which is considered one of the most critical factors in determining the best visual prognosis[20]. This may be a type II error given that our sample size was small. Previous animal studies have shown that irreversible retinal tissue loss occurs with retinal ischemia for 240 min or more[21]. Most human studies have demonstrated the best visual outcome if HBOT is initiated within 12 h of symptom onset. The Undersea and Hyperbaric Medical Society recommended CRAO patients with symptom onset < 24 h and not responding to normobaric oxygen therapy for a course of HBOT[22].

Diagnosing CRAO is not always straightforward. It was shown that up to 27% of CRAO patients had a normal-looking fundus upon presentation[23]. Different imaging modalities, such as FFA and OCT, were utilized to aid timely diagnosis of CRAO, and FFA is considered the gold standard for confirmation. Typical findings include a delay in filling of retinal arteries, delayed arteriovenous transit time, and sluggish blood flow in arteries[24]. However, these methods are subjective based on the operator’s experience, and comparisons with the assumed normal fellow eye are more appropriate. A study revealed that a delayed arteriovenous phase > 23 s was observed in more than half of the CRAO cases only[23]. In addition, the exact transit time should be interpreted with caution, as the transition of fluorescein dye from the peripheral to the eye circulation may be affected by various systemic conditions, such as heart failure and carotid stenosis. Moreover, the machines used to capture FFA images are not often available under emergency settings. By the time an FFA is performed, there might already be partial reperfusion of the retinal vasculature, which might create ambiguity in the diagnosis. Additionally, FFA provides no information regarding the viability of the affected retinal tissues.

Emerging evidence has suggested OCT as a promising modality for diagnosing CRAO. OCT is sensitive for detecting changes in retinal layers during ischemia, even before fundoscopic changes appear. Yilmaz and Durukan[25] reported that retinal thickness increases in a near-linear progression within the first hour of CRAO[25], which allows an estimate when exactly the retinal insult was initiated. Pathognomonic signs in the acute phase of CRAO include an increase in retinal reflectivity, thickness of the inner retina, and a prominent middle limiting membrane[26]. Our study provided additional insight into the 1 mm and 1-to-3 mm circular rim of CMT findings in CRAO eyes. The central 1 mm zone of the CMT consists of mainly photoreceptors and their nuclei layers, particularly at the foveal pit. Theses layers exhibited a smaller increase in thickness compared with the inner retinal layers in acute CRAO. Therefore, our results demonstrated no correlation between the 1 mm zone of the CMT and VA changes. In contrast, the 1-to-3 mm circular rim of the CMT consists of all layers of the neurosensory retina. The increase in thickness of the 1-to-3 mm circular rim upon CRAO insult was more uniform than that of the 1 mm zone where the foveal pit was located. OCT is a faster investigation than FFA without the need for vascular access, and does not carry the risk of fluorescein dye. This approach may have a role in identifying patients who are more likely to benefit from HBOT.

HBOT is a safe treatment modality with only rarely reported serious adverse effects. The most commonly reported complication is barotrauma to the middle ear. Incisional myringotomies may sometimes be needed during HBOT if the patient fails to equalize middle ear pressure[27]. Other reported side effects include sinus pain, anxiety, and oxygen toxicity. Interestingly, ocular complications such as reversible myopia and nuclear sclerosis cataract development have also been reported[20]. Hypoglycemia, which was reported in our study, is not a frequent side effect of HBOT. A systemic review demonstrated a reduction in blood glucose levels following a single session of HBOT in patients with type 2 diabetes mellitus, potentially mediated by an increase in insulin sensitivity[28].

Being classified as ocular stroke, CRAO might represent part of the systemic stroke spectrum. It is not surprising that studies have shown CRAO patients to be at risk of stroke at different points in life. In a Canadian cohort, approximately one-third of patients experienced symptomatic stroke before the diagnosis of CRAO, and approximately 5% of the remaining CRAO patients suffered from stroke within 3 years of ocular diagnosis[29]. Magnetic resonance imaging findings corresponding to ischemic stroke were found in nearly 40% of patients in another American cohort at the time of CRAO diagnosis[30]. These findings emphasize the importance of a comprehensive cardiovascular workup and a multi-disciplinary approach with physicians and neurologists in the management of CRAO patients.

The main limitation of the current study lies in its small sample size. This was partially attributed to the 6-h cutoff time for HBOT from symptom onset to diagnosis. Furthermore, bias was inevitable because the study was not randomized or controlled, which also occurred in other researches focusing on rare diseases without known effective alternative treatments. With the ongoing HBOT for CRAO patients in Hong Kong, future studies with larger sample sizes could give more solid evidence.

CONCLUSION

HBOT is promising for CRAO patients to regain vision for navigation. A thinner 1-to-3 mm circular rim of CMT, but not the central 1 mm zone, is associated with better visual outcome. OCT alone, without FFA, may be used by ophthalmologists for triage of HBOT on CRAO cases. However, further studies with larger sample sizes are necessary to validate these findings and ensure their acceptance.