Discussion
In this study we report an overall incidence of perioperative stroke of 1.99% in a large cohort of contemporary patients undergoing the breadth of adult cardiac surgery. A diagnosis of stroke had a significant impact on outcomes. Patients required prolonged hospital stays, experienced greater complications and experienced a 3-fold higher in-hospital mortality compared with matched controls. The 1-year survival was almost 28% lower.
The incidence of stroke varies significantly depending on the nature of the surgery, being most prevalent following major aortic surgery (8.14%) and lowest following isolated CABG (0.85%). The high prevalence following aortic surgery likely reflects the significant aortic manipulation that occurs and possible manipulation of the proximal carotid arteries leading to embolization of atherosclerotic material. It is also important to highlight that 55.7% of these patients had undergone emergency repair of type A aortic dissection. Other studies have also identified this association with aortic surgery7, and they also note higher incidence following double procedures, presumably related to longer CPB times and greater likelihood of atherosclerotic embolisation. In some centres, epi-aortic scanning is performed to assess the ascending aorta prior to cannulation. This is not currently available at our centre but may facilitate reduced intraoperative embolisation. It is notable that the incidence of stroke following thoracic surgery at our centre over the same period was only 0.2%, highlighting that patients undergoing cardiac surgery are at a substantially greater risk, presumably related to factors specific to cardiac surgery and cardiopulmonary bypass.
In our series, over ¾ of the strokes were early, detected on waking from anaesthesia – likely reflecting an intra-operative event. In keeping with this, the vast majority of strokes were ischaemic. A meta-analysis reported almost equal incidence of early and delayed stroke across a large number of studies2. Many of the included studies were from the early 2000’s and our lower proportion of delayed strokes – which are thought to be related to pre-existing cerebrovascular disease (chronic small vessel disease and carotid atherosclerosis) and atrial fibrillation3, may reflect better pre-operative assessment and optimisation and improved anaesthetic management in the modern era leading to lower risk of stroke despite operating on higher risk patients8.
As could be expected, risk factors for atherosclerotic disease were prevalent in our population of patients experiencing stroke including diabetes, smoking history, peripheral vascular disease and end stage renal impairment. We performed propensity matching in order to compare these patients to similar individuals undergoing similar procedures rather than comparing to the entire cohort as most other studies have. In terms of the atherosclerotic risk factors, there were no significant differences between those patients going on to experience a stroke, likely reflecting the similar pathophysiology for their cardiac disease and predisposition to stroke, making it challenging to use these factors to predict preoperatively patients at risk of developing stroke, although some have attempted to create risk scores. The relatively small number of patients in our study provided insufficient power to generate a scoring system from our data. The notable difference between the groups was that there was a significantly higher prevalence of previous TIA in the patients going on to experience a stroke. Interestingly, the authors of the meta-analysis discussed above, performed a meta-regression and identified previous neurological event as being the only factor significantly associated with development of perioperative stroke2.
Carotid atherosclerosis is thought to be an important risk factor for perioperative stroke. However, there has been much debate on routine use of carotid doppler to diagnose significant carotid stenosis preoperatively, and this is not currently routinely used at our centre, although is used in selected cases. The prevalence of significant carotid stenosis in unselected patients is found to be low, and the incidence of stroke in patients with severe carotid stenosis is found to be too low to warrant widespread screening and is not recommended by current guidelines9-12. Some suggest selective screening (for example, patients >70 years with previous TIA/stroke) may be useful as there is some evidence that severe carotid stenosis is a significant predictor for perioperative stroke following cardiac surgery in these patients13. It is notable that even in studies describing preoperative carotid screening there is a low rate of carotid intervention even when severe stenoses are detected. The options include pre- or simultaneous carotid endarterectomy or carotid artery stenting. However, knowledge of carotid atherosclerosis is useful as it can lead to interventions to attempt to reduce the incidence of cerebrovascular ischemia, for example by ensuring high pressures and pulsatility whilst on CPB.
The reason for interest in identifying patients at risk preoperatively is reinforced by examining patient outcomes following stroke. In our patients, there was significant prolongation of ICU and hospital length of stay and patients experienced much greater incidence of common complications including AKI and respiratory tract infections. The in-hospital mortality was 3-times higher than the matched controlled patients, at 17.0%, and significantly inferior longer-term survival was observed. It is worth emphasising that even when the significant early mortality is excluded, by imposing 30-day conditional survival, the survival of patients who suffered a stroke remains significantly inferior highlighting that the impact of stroke on mortality is not simply in the early phase but persists long-term. In addition to the financial burden placed on our institution, it is notable that 59.4% of these patients were transferred to their local hospital for ongoing stroke rehabilitation. It is also worth highlighting that the ICU and hospital stays refer to duration at our centre, and not the full inpatient stay which would be even greater considering the number being transferred. These outcomes are similar to other series reporting on patients following perioperative stroke. The 1- and 3-year survival rate from the meta-analysis were 80.2% and 73.0% respectively2. Our rates were significantly lower at 61.5% and 53.8%. We believe that this reflects the increase in older patients with more comorbidities undergoing cardiac surgery. Indeed, our matched cohort had survival rates of 89.4% and 86.1% which are significantly inferior to those presented in the meta-analysis of 99.5 and 99.2%.
The challenge of managing perioperative stroke has in part been due to the limited treatment options traditionally available, since thrombolysis is contraindicated in this population. There only option has been to administer high-dose anti-platelet medication and refer the patient for stroke rehabilitation. Recently though, mechanical thrombectomy has emerged as an important development in stroke management. Mechanical retrieval of thromboemboli from the cerebral arteries by neurointerventional radiology has been demonstrated to result in significantly greater neurological recovery in selected patients14,15. This is not contraindicated in surgical patients and there are now many reports of early postoperative patients undergoing mechanical thrombectomy with good clinical outcomes both in cardiac surgery and other surgical specialties16-18. Mechanical thrombectomy may offer cardiac surgical patients experiencing perioperative stroke an opportunity to achieve much greater neurological recovery than has been observed in the past.