The incidence of postoperative visual loss ranges from 0.0008%-0.002% in non-ocular, non-cardiac surgery to 0.06-4.5% in CPB series [1–9]. The most common post-operative visual loss defects are related to ischaemic optic neuropathy (ION) CRAO and cortical blindness [1]. There is still much controversy surrounding this injury and it appears it is not always preventable. Should we however inform all patients of the risk given the incidence of post-operative visual loss is rare but devastating? The lack of certainty of the mechanism of the insult limits our understanding combined with the multiplicity of risk factors. The rodent model described by Bernstein is of uncertain importance as most evidence is from case-series or case-control studies in cardiac, spinal, trauma and non-ocular surgery [2, 4–6, 8, 10, 11].
In this case report, we will restrict our discussion to the aetiology and mechanisms pertaining to ION describing the clinical findings and mechanism of both AION and PION. AION occurs at the optic nerve head as the optic nerve and retinal vessels enter the globe. PION occurs anywhere posterior from the optic nerve head to the optic chiasm. Early fundoscopic examination will demonstrate disc oedema in AION (Figure 1), whereas the fundus is entirely normal in early PION (Figure 2). Days to weeks later fundoscopy reveals optic nerve pallor in both conditions (Figure 3). Pupillary light reflexes are absent or delayed in both conditions and patients may complain of total blindness or altitudinal field defects (visual loss above or below the visual horizon). Delayed or worsening visual loss usually implicates AION as the probable diagnosis and the delay can be up to a few weeks after surgery.
In just over half of the ION cases in the American Society of Anaesthesiologists Post Operative Visual Loss (ASA POVL) Registry, both eyes were affected, suggesting a systemic event [1]. Complete recovery from either AION or PION is rare.
AION has been primarily associated with CPB procedures but can also occur during prone spine operations and spontaneously. PION is most commonly associated with lengthy prone spine operations and bilateral radical neck operations. Numerous contributory factors for AION and PION have been proposed (hypotension, anaemia, oedema, venous congestion, variations in ocular anatomy and physiology, effects of fluid resuscitation) but none has been causally linked [11, 12]. Older patients are more at risk of ION due to the natural reduction in nerve fibres of approximately 5,000/yr [13].
Blood flow to the anterior optic nerve is autoregulated by endovascular and metabolic factors similar to cerebral circulation. Autoregulation is effective over a critical range of perfusion pressure - not clearly defined in humans. Diseases of the circulatory system (hypertension, diabetes mellitus, atherosclerosis) damage autoregulation and therefore increase the critical perfusion pressure below which autoregulation of the anterior optic nerve fails. PION arises from the suboptimal blood supply from branches of the ophthalmic artery and the central retinal artery. The risk factors for both AION and PION in non-surgical patients include: high serum cholesterol and triglycerides, lipidaemia, fibrinogenaemia, DM and smoking. Risk factors in cardiac surgical patients include: intrinsic small vessel disease (DM), hyperglycaemia, hypertension, prolonged hypotension, vasospasm secondary to inotropes, the inflammatory process induced by CPB, anaemia and lower than normal cup:disc ratio which correlates with small scleral canal and possibly emboli [1]. A combination of risk factors in a patient may lead to additive or synergistic effects on blood flow and autoregulation.
A large retrospective time-matched, case-control study of 27,915 patients who underwent CPB was undertaken using the Mayo Clinic Surgical and Medical Indexes [6]. ION was identified in 17 (0.06%) cases. 2 controls for each ION patient were chosen by selecting 2 patients who underwent CPB exactly 2 weeks before each ION patient. They found significant risk factors for post-operative ION following cardiac surgery using CPB to be: advanced age, lower minimum post-operative haemoglobin concentrations, history of clinically severe vascular disease, pre-operative angiogram within 48 hours of CPB, longer pump times, surgical disruption of particulate matter, RBC transfusions and use of non-RBC blood components. Visual field and acuity loss was present in 100% of the cases and no improvement was found on follow-up.
Anaemia is an amenable risk factor to efficient correction. Nevertheless optimum haemoglobin concentration is a controversial subject especially in patients with cardiovascular disease. By contrast Spiess found that a higher haematocrit on intensive care admission was associated with Q wave MI [15]. Brown et al reported a series of 6 patients over a 10-year period that developed AION related to anaemia and hypotension after various surgical procedures [16]. There is debate as to whether cardiac patients behave similarly to other surgical patients [12]. Roth investigated the effect of isovolaemic haemodilution on ocular blood flow in cats [17]. He found a 71% increase in retinal blood flow and retinal O2 delivery remained approximately constant as the increased blood flow countered a significant decrease in arterial O2 content. Neely et al found feline retinal tissue oxygen tension increased initially during isovolaemic haemodilution to 50% above baseline at about 2/3 of the original haematocrit level and beyond this a steady decline in tissue oxygen tension [18]. Among cardiac surgical patients Mansour carried out a retrospective chart review of 1594 consecutive CABG patients over a five-year period (1995-1999) in one medical centre [19]. 3 patients experienced visual loss due to anterior ischaemic optic neuropathy - all had diabetes mellitus and 2 suffered severe postoperative anaemia. Among diabetics the risk of anterior ischaemic optic neuropathy was significantly higher in subjects with a postoperative haematocrit below 22% compared with the non-diabetic population (28.6% v 0.21%, p = 0.001).
Increasing the transfusion threshold in an effort to correct anaemia may minimize the chances of ION. However, transfusion in cardiac surgery has been shown to increase morbidity and mortality. Engoren demonstrated that transfused cardiac surgical patients had a five-year mortality double that of non-transfusion group [20]. Similarly Kudavalli found increased one-year mortality in the transfusion group [21]. Habib found increased incidence of renal injury and Banbury found increased risk of infection after transfusion in cardiovascular surgery [22, 23]. Surgenor found that treatment of anaemia with transfusion increased the risk of low output cardiac function requiring treatment with inotropes and intra-aortic balloon counter pulsation [24]. In Hebert's multicentre RCT, he found no significant difference in 30-day or 60-day mortality in ICU patients with significant cardiovascular disease whether restrictive or liberal transfusion triggers were used [14]. However restrictive transfusion triggers appeared to be associated with significantly lower mortality in younger patients with lower APACHE scores.
So, with regard to correction of anaemia it would seem that we are "damned if we do and damned if we don't - withholding transfusion may have a cost in terms of increased incidence of ION or renal failure but transfusion itself may increase morbidity and mortality [25]. The Society of Thoracic Surgeons and the Society of Cardiovascular Anaesthesiologists recommend transfusion thresholds >7 g.dl-1[26].
Given the concerns surrounding transfusion as a treatment for perioperative anaemia, it has been proposed that prophylactic measures to minimize bleeding might be the solution to the problem. In a meta-analysis by Laupacis, tranexamic acid administration was associated with lower transfusion requirements in patients undergoing cardiopulmonary bypass [27]. Nevertheless, a major concern with tranexamic acid use is the possibility of increased risk of thrombembolic events, however the evidence is weak at present [28].
Another risk factor for ION, which may be amenable to intervention, is patient positioning. Prone position is known to increase IOP and although prone positioning is not used routinely in cardiac surgical patients, they may develop a lung injury in the post-operative period either from the cardiopulmonary bypass itself, transfusion, or sepsis and I feel it is worth a mention. Ozcan looked at the effect of prone positioning in awake volunteers and found IOPs 50% higher in prone horizontal patients [29]. The rise in IOP could be reduced with head-up tilt of 10 degrees, but not to normal. Cheng found IOP doubled in prone anaesthetised patients [30]. The episcleral venous system is connected to the central venous system via a valveless arrangement such that an increase in CVP causes an increase in episcleral venous pressure and hence increased IOP. There have been many reports of ION following surgical procedures performed in the prone position. Proning is also increasingly used in intensive care to treat patients with ARDS and the effect on IOP should also be considered in these critically ill patients. A mild degree of reverse Trendelenberg may ameliorate the increased episcleral venous pressure caused by proning. There are as yet no published cases of blindness associated with proning in ARDS, but it is only a matter of time. We are proposing to carry out a survey of practice across hospitals in the geographical area in order to establish practice with regard to proning in ITU and actions taken to defend ocular perfusion. We also wish to carry out a preliminary crossover study measuring intraocular pressure and haemodynamic parameters in supine position before proning procedure and thereafter in prone flat and then with a moderate degree of head-up position.
Perfusion pressure of the anterior optic nerve is the difference between the ciliary artery and venous drainage - approximately MAP - IOP. Increased IOP may substantially reduce the perfusion pressure of the anterior optic nerve. Increased IOP may contribute to the higher incidence of ION after operations involving CPB. Sanjay et al showed that intraocular pressure (IOP) increases on commencement of CPB and maximally at 20 minutes to 21 mmHg [31]. Methods to reduce IOP and therefore maximise intraocular perfusion pressure must be sought, including head position and use of loose endotracheal tube ties. CVP can increase beyond IOP and affect intraocular perfusion as it is a valveless system. This is in addition to external pressure on the globe [12]. Visual loss is not always associated with intraoperative hypotension, however it is quoted in a number of case reports [5, 12]. Often the nadir of perfusion pressure is no different between those affected and unaffected after CPB and a modicum of hypotension is a frequent occurrence in anaesthetised patients, which may be associated with altered autoregulation or anatomical variation of optic nerve blood supply [12, 32, 33].
Embolic phenomena are another possible cause of ION and may be related to aortic cross clamping [34]. In a prospective study of neurological complications of coronary artery bypass surgery in 312 patients, 25.6% developed neuro-ophthalmological complications [35]. Horton showed visible retinal emboli in three patients, two coronary artery bypass graft patients using CPB and one cardiac catheterisation patient [36]. Intra-operative epi-aortic ultrasound may be a solution to minimise embolic risk for placement of the aortic cross-clamp. A "no-touch" technique may be used in off-pump CABG and alternative route CPB [37]. A sutureless proximal aortic anastamotic technique has been shown to reduce micro-embolic counts on transcranial Doppler [38].
The use of inotropes has also been implicated in cases of ION. Shapira et al found the development of AION in 8 patients out of 602 consecutive cardiac surgery patients (1.3%), an increase from their usual 0.5% rate [33]. Although it was impossible to implicate any single factor in the increased incidence, prolonged CPB, lower haematocrit, and peri-operative use of adrenaline and amrinone were all found to be associated. In one case series of 4 patients with AION occurring within a one-month period, high doses of mixtures of inotropes were one of the putative causes [39]. Adrenaline is postulated to cause severe vasospasm of the ocular vessels [40, 41]. This presents anaesthetists with a difficult clinical decision since prolonged hypotension is also listed as a risk factor for ION.
In conclusion perioperative ION is a rare, devastating event, which may be difficult to detect in a sedated post-operative patient. The signs are subtle and are often late to present. Cardiac surgical procedures involving CPB appear to be at increased risk of causing this condition especially if complicated by hypotension, use of inotropes or anaemia. No effective treatment for ION exists, once established, hence prevention is crucial. Despite this, studies to date have only succeeded in demonstrating that minimizing one risk factor for the development of ION may merely serve to introduce another.