Study design

Single-center, prospective interventional case series with partially-blinded outcome assessment. Outcomes were compared to the EAGLE CST-arm. Institutional Review Board approval was obtained from the independent ethics committee of Tübingen University (protocol-no 564/2016BO2) and all clinical investigation have been conducted according to the principles expressed in the Declaration of Helsinki. Written informed consent was obtained from all patients prior to IVT.

Setting, patient eligibility and treatment

The study was conducted at the Eye Hospital and the Department of Neurology & Stroke of Tübingen University Hospital, a tertiary care facility.

Diagnosis of NA-CRAO was confirmed by an experienced ophthalmologist independent of the study through assessment of BCVA using Early Treatment Diabetic Retinopathy Study (ETDRS) charts, intraocular pressure, relative afferent pupil defect, slit-lamp biomicroscopy and fundoscopy to exclude rare differential diagnoses associated with sudden, painless, and severe visual loss, e.g. retinal detachment or hemorrhage. To avoid delay of IVT, fluorescein angiography and assessment of visual fields were not performed prior to treatment.

For inclusion, BCVA had to be LogMAR ≥1.3 (Snellen equivalent: 6/120), corresponding to functional blindness according to the World Health Organization International Classification of Diseases (WHO-ICD) [14]. Categories of low vision (counting fingers, hand motion, light perception and no light perception) were assessed at a distance of 30 cm and translated into LogMAR values following suggestions from Lange [15].

In addition to ophthalmological assessment, all patients received a thorough neurological examination including blood sampling, National Institutes of Health Stroke Scale score (NIHSS) and a cranial computed tomography (CT) or magnetic resonance imaging (MRI) to exclude intracranial hemorrhage (ICH), (sub-acute) cerebral infarction, and other IVT contraindications (see supplemental Methods).

In analogy to acute ischemic stroke, eligible patients received 0.9 mg of tPA (Actilyse®, Boehringer Ingelheim, Germany) per kilogram of bodyweight (max. 90 mg, 10% as bolus and the remainder over one hour) within 4.5 hours of self-reported symptom onset.

All patients were admitted to the Stroke Unit of the Department of Neurology & Stroke. Brimonidine or dorzolamide/timolol was applied to normalize increased intraocular pressure in two cases. No further CST was administered.

Contraindications for intravenous thrombolysis

Patients with acute non-arteritic central retinal occlusion were excluded from intravenous thrombolysis according to the official contraindications provided by the European Medicines Agency for use of tissue plasminogen activator within 4.5 hours of ischemic stroke onset [16]; exception was made for patients “over 80 years of age”, “with any history of prior stroke and concomitant diabetes”, and “on vitamin K antagonists and INR ≤1.7” (all ex- and inclusion criteria are listed in S1 Table).

Ophthalmological follow-up

All patients received ophthalmological follow-up (FU) examinations at pre-specified time points, i.e. at day 5±2 after IVT or prior to discharge (whichever occurred first) (d5) and at day 30±5 (d30). BCVA was determined at each visit by an experienced ophthalmologist independent of the study. Functional recovery was pre-defined as a BCVA of LogMAR ≤0.5 (Snellen equivalent: 6/20) (reading ability according to the WHO-ICD [14]).

Blinding

All ophthalmologists performing FU examinations were blinded to prior BCVA.

Neurological follow-up

All patients underwent at least 24 hours of stroke unit monitoring. FU brain imaging (CT or MRI) was performed 24±12 hours after IVT as part of clinical routine and evaluated by board-certified neuroradiologists independent of the study for presence of ICH, new ischemic lesions, and subcortical leukoencephalopathy.

All patients underwent thorough etiological work-up including neurovascular ultrasound, 24-hour electrocardiographic monitoring, echocardiography and laboratory inflammatory markers for exclusion of arteritic CRAO.

NIHSS was assessed at d5. Modified Rankin Scale score (mRS) and the occurrence of SAE were assessed at d30.

Comparison to conservative standard treatment

For estimation of IVT effect, we compared visual outcomes with individual patient data taken from the EAGLE CST-arm provided by the EAGLE Study Group (see original publication for patient baseline characteristics [7]). Of the 40 patients in the EAGLE CST-arm, 39 received FU at d5 and 37 at d5 and d30.

Statistics

Statistical analysis was performed using SPSS Statistics 24 (IBM, USA). One-factorial repeated-measurements analysis of variance (ANOVA) was used to assess intra-individual differences from baseline to d5 and d30 in the respective treatment group. To analyze differences between groups, ANOVA with two independent factors (IVT and CST) and time was used. Significant values were corrected for inhomogeneous variance according to Greenhouse-Geisser. Student’s t-test was performed for dependent means; significant values were corrected for multiple testing according to Bonferroni. Fisher’s exact test was employed to determine differences in the proportions of functional recovery between study groups. P <0.05 was considered significant. Power calculation for a suggested future randomized trial was done using the function power.prop.test from R version 3.2.2 stats package (The R Foundation, Austria). Reporting in accordance with the TREND guidelines for non-randomized interventional trials.

Patient population

Between January 2014 and August 2016, 20 consecutive patients with acute NA-CRAO received IVT according to our SOP (see Table 1 for baseline characteristics, and S2 Table for individual patient data). Patient flow is indicated in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Patient flow.

BCVA = best corrected visual acuity, CRAO = central retinal artery occlusion, IVT = intravenous thrombolysis, INR = international normalized ratio, LogMAR = logarithm of the minimum angle of resolution.

https://doi.org/10.1371/journal.pone.0198114.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Patient characteristics.

https://doi.org/10.1371/journal.pone.0198114.t001

Treatment and visual outcome

Median onset-to-treatment time was 210 minutes (interquartile range (IQR) 120–240). Mean BCVA was LogMAR 2.46±0.33 (standard deviation) (light perception) on admission and improved to 1.52±1.09 (Snellen equivalent: 6/200) at d5 (p = 0.002) and 1.60±1.08 (Snellen equivalent: 6/240) at d30 (vs. baseline p = 0.004, vs. d5 n.s.) (Fig 2A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

Evolution of best corrected visual acuity over time (BCVA): (A) mean BCVA of our intravenous thrombolysis (IVT) cohort (N = 20) and of the conservative standard treatment (CST) group of the EAGLE-trial [7]. (B) individual BCVA of our IVT-cohort and (C) of the EAGLE CST-arm. Functional blindness (LogMAR >1.3) and functional recovery (LogMAR ≤0.5) are indicated by a gray and blue background, respectively. LogMAR = logarithm of the minimum angle of resolution.

https://doi.org/10.1371/journal.pone.0198114.g002

Ophthalmological FU examinations were performed beyond d30 in 14 patients. Median time between NA-CRAO and latest FU was 6 months (IQR 4–11.5). Mean BCVA at the latest FU was LogMAR 1.39±1.15 (Snellen equivalent: 6/150) and did not differ from BCVA at d30 (1.51±1.11, n = 14) (Snellen equivalent: 6/200).

Improvement of BCVA from baseline to FU was also observed in the EAGLE CST-arm (N = 37; baseline LogMAR 2.09±0.51 (counting fingers) vs. d5 LogMAR 1.78±0.60 (Snellen equivalent: 6/380), p = 0.001; baseline vs. d30 LogMAR 1.63±0.62 (Snellen equivalent: 6/240), p<0.001; d5 vs. d30, n.s.). However, the gain in BCVA from baseline to d30 was greater in our IVT-treated cohort (F(1.459, 110) = 6.601, p = 0.005; Fig 2A).

Despite mean baseline BCVA being worse in our group compared to the EAGLE CST-arm (p = 0.018; Fig 2A), the rate of functional recovery to LogMAR ≤0.5 (Snellen equivalent: 6/20) was higher: 6/20 (30%) vs. 1/39 (3%) at d5 (p = 0.005) and 5/20 (25%) vs. 2/37 (5%) at d30 (p = 0.045); compare Figs 2B, 2C and 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Categorical presentation of best corrected visual acuity.

Categorical presentation of best corrected visual acuity (according to the current version of the WHO International Classification of Diseases [14]) at baseline and at day 30 of our intravenous thrombolysis cohort and of the conservative standard treatment group of the EAGLE-trial [7]. We defined favorable outcome as mild or no visual impairment (LogMAR ≤0.5, indicated in blue). Unfavorable outcome includes moderate or severe visual impairment (LogMAR >0.5 to ≤1.3) and functional blindness (LogMAR >1.3). LogMAR = logarithm of the minimum angle of resolution.

https://doi.org/10.1371/journal.pone.0198114.g003

One patient in our cohort, who had recovered to LogMAR 0.0 (Snellen equivalent: 6/6) at d5, suffered a second NA-CRAO in the same eye two weeks later. IVT was not repeated due to multiple sub-clinical small acute and subacute cerebral infarcts seen on MRI. The patient’s final BCVA was LogMAR 1.5 (Snellen equivalent: 6/200) at d30.

Another patient, with a history of age-related macular degeneration and prior BCVA of LogMAR 1.0 (Snellen equivalent: 6/60) in the right eye and 2.3 (perception of hand motion) in the left eye, presented with an acute NA-CRAO of the right eye, which further reduced BCVA to LogMAR 2.3. It was decided to apply IVT in an attempt to avert even more severe visual impairment. At d5 and d30 FU, the patient had recovered to pre-NA-CRAO BCVA. For statistical analysis, we classified both patients as not recovered.

Safety and neurological outcome

IVT-related SAE occurred in two patients. One patient suffered orolingual angioedema, which was treated with single-dose antihistamine and prednisolone and did not require invasive airway-management. Another patient suffered hemorrhage from an abdominal aortic aneurysm of which the attending neurologist was unaware at the time of IVT. The bleed led to a relevant drop in hemoglobin levels (from 4.97 to 3.35 mmol/L) and required a single transfusion of packed red blood cells. Both patients completely recovered from the respective SAE.

No intracranial or intraocular hemorrhage was observed. Silent cerebral infarcts were seen in 3/20 (15%).

Baseline NIHSS was 0 in all patients apart from two, who presented with an NIHSS of 1. One patient with known dementia scored 1 due to impaired orientation. The other patient reported mild sensory loss of the right face and arm, which had occurred simultaneously to NA-CRAO and resolved within 24 hours after IVT.

Median pre-NA-CRAO mRS was 0 (IQR 0–1). Median mRS at d30 was 2 (IQR 1.5–2) due to visual loss.

Safety

We observed two IVT-related SAE; both required intervention but did not cause permanent damage. No intracranial or intraocular hemorrhage was observed in our IVT-cohort.

Overall, we would expect the rate of IVT-related adverse events in NA-CRAO to be equal to those seen in minor stroke (2 to 2.4% [18–19]) or even lower. Increased ICH rates as observed in the EAGLE- (5%) [7] and the Chen-trial (12.5%) [8] may be due to delay of thrombolytic treatment far beyond 4.5 hours after symptom onset. Previous IVT-trials in stroke excluded patients exceeding the time windows of 4.5 to 6 hours to avoid an assumed increase in ICH risk associated with progressive blood-brain barrier disruption. MRI studies have revealed concurrent subclinical cerebral infarction in up to 25% of NA-CRAO cases [20].

Risk and benefit of IVT in NA-CRAO

Patients not recovering from NA-CRAO suffer sustained disability due to visual loss, reflected by an mRS of 2. This instance justifies aggressive treatment despite an increased risk for adverse events.

We agree with Schrag that only a randomized placebo-controlled double-blinded phase III trial may conclusively answer the question whether IVT initiated within 4.5 hours is efficacious and safe for NA-CRAO treatment.

Important considerations for a clinical phase III trial

First and foremost, the choice of clinical endpoint is crucial to the success of such a trial. Analyses of changes in mean BCVA–recently presented in [10]–are not suited to evaluate IVT efficacy in NA-CRAO, as significant changes in LogMAR do not necessarily reflect clinically relevant improvement or deterioration: a change from LogMAR 2.5 to 1.5 (as observed in both our IVT-cohort and the EAGLE CST-arm) (Snellen equivalent: light perception to 6/200) basically represents no relevant improvement as both values correspond to functional blindness [14], whereas a change from 1.5 to 0.5 (Snellen equivalent: 6/200 to 6/20) signifies recovery from functional blindness to reading ability. We consequently suggest a dichotomized analysis of visual outcome data and the use of a clinically relevant primary efficacy endpoint, i.e. regain of reading ability (LogMAR ≥0.5) (Snellen equivalent: 6/20), to distinguish favorable from unfavorable outcome.

It will be necessary to limit enrolment to patients with previously healthy eyes (LogMAR ≤0.5; Snellen equivalent: 6/20) and severely impaired vision on admission (LogMAR ≥1.3; Snellen equivalent: 6/120), i.e. functional blindness according to WHO-ICD [14]). Normal vision prior to NA-CRAO is the precondition for reaching the chosen clinical endpoint of functional recovery and will avoid potential bias through unbalanced baseline BCVA. Additionally, only severe visual loss in a previously healthy eye with realistic chances of full recovery and relevant disability as an alternative outcome, justifies the use of a potentially harmful therapy. Furthermore, a sudden and painless steep drop in BCVA most reliably indicates the presence of NA-CRAO with a complete and proximal vessel occlusion and reduces the risk of including patients with other differential diagnoses to an absolute minimum.

To assess the feasibility of a phase III trial, we performed power-calculations with the clinical endpoint of functional recovery to reading ability at d30 after NA-CRAO.

Based on rates of functional recovery to LogMAR ≤0.5 (Snellen equivalent: 6/20) in our IVT-cohort (25%) and that of Nedelmann (29%) and Schrag (50% to LogMAR ≤0.7; Snellen equivalent: 6/30), we conservatively estimate the success rate in the intervention-arm to be 20%. Based on recovery rates in the EAGLE CST-arm (5.4%) and Schrag’s meta-analysis (7.4%) [9], we optimistically assume a success rate of 10% in the placebo-arm. A two-armed randomized trial with a 10% drop-out rate would require 442 patients in order to detect a treatment effect at the 0.05-level with a power of 80%, using Chi-squared approximation. Our enrolment rate indicates that such a trial could be performed within three years if at least 20 study centers are involved.

In our case series, the majority of NA-CRAO patients with relevant persisting visual loss was ineligible for treatment due to late admission (69/100). In analogy to IVT in ischemic stroke [21], educational efforts will have to be undertaken to decrease admission delays.

The use of prognostic biomarkers, like the previously discussed retrobulbar spot sign [11], may enrich a study population and reduce sample size. However, it may also complicate and slow down enrolment. Additionally, spot sign-positive patients would be left untreated whilst futility of IVT in these patients is not yet established. We therefore strongly recommend, that the evaluation of the retrobulbar spot sign and its prognostic value should be part of the proposed randomized trial, but current data does not support exclusion of patients on its basis.

Strengths and limitations

Several limitations of our study require further discussion. It is a non-randomized prospective interventional case series performed at a single tertiary care center. However, no published data is available from randomized trials studying IVT (or IAT) within a 4.5-hour time window after NA-CRAO.

Including 20 consecutively treated patients, with none lost to FU, this case series is the largest prospective study of early IVT in NA-CRAO. Our outcome data is in line with previously published studies [9, 11].

The control group used for estimation of IVT efficacy was taken from the randomized and CST-controlled EAGLE-trial. Controls consisted of well-characterized, tPA-eligible patients, avoiding a bias inherent to other historic controls. Identical criteria were used for blinded outcome assessment in our cohort and the EAGLE-trial [15].

Mean baseline BCVA observed in our IVT-cohort was significantly worse than in the EAGLE CST-arm. This may have led to a bias in favor of IVT effect expressed by the overall gain in mean LogMAR from baseline to d5/d30, where mean BCVA values did not differ significantly between groups. However, despite worse baseline BCVA, more patients experienced functional recovery, which supports our assumption of a real IVT effect.

We cannot rule out a bias in favor of our IVT-cohort caused by transient NA-CRAO with spontaneous recovery [3]. Mean onset to treatment delay in the EAGLE CST-arm was eleven hours versus three hours in our IVT-cohort, which makes inclusion of transient NA-CRAO more likely in our cohort.

The EAGLE-trial did not address the cause of NA-CRAO; thus, direct comparison to our IVT-cohort is limited since clots’ origin might significantly influence IVT response and visual outcomes.

No fluorescein angiography was performed prior to treatment in our trial. Consequently, there remains a risk of non-CRAO inclusions. However, even guidelines recommend skipping fluorescein angiography in the acute phase in favor of an undelayed hospital admission for immediate treatment and neurovascular work-up [2].

For the same reasons, no initial assessment of the visual field was performed. This may be crucial for distinguishing true recovery of retinal function and apparent recovery, which occurs through adaption to central visual loss by eccentric fixation [3]. However, a BCVA of LogMAR ≤0.5 Snellen equivalent: 6/20) may not be achieved through eccentric fixation as spatial resolution in the peripheral visual field (>2° from the center of foveal vision) is insufficient [22].

Finally, we did not perform retrobulbar ultrasound prior or parallel to IVT in our study. Thus, we can neither make assumptions on the embolus’ composition nor the spot sign’s prognostic value regarding IVT efficacy [11].