With carotid artery stenting becoming a more acceptable and commonplace procedure, the issue now moves on to the question of cerebral protection. Speaking at the 19th International Congress for Endovascular Interventions (Scottsdale, AZ) Dr Michael Makaroun gave his views of cerebral protection and presented preliminary findings of a new randomised trial. A few months later, the Charing Cross International Symposium (London, UK) also highlighted the protection question with a debate between Alberto Cremonesi (Cotignola, Italy) and Sumaira Macdonald (Newcastle, UK).
At the International Congress, Makaroun opened by saying: “The technique of carotid artery stenting with a filter device has been standardised for the last few years. I do agree with everybody that cerebral protection with stenting does make sense: it’s intuitive and only a criminal would want emboli in the patient’s brain.” However, he observed that there is a dearth of proof that cerebral protection actually works, commenting that the filters are only marginally better than charms sold to tourists ‘to cure disease’ in Japan and are in fact based on “the name of the first band of Eric Clapton: Blind Faith”.
Makaroun highlighted Global Registry data presented by Wholey et al (Catheter Cardiovascular Interventions, 2003), which shows that “the incidence of stroke in those without protection is twice as high as for those with protection. However, the groups are not concurrent and there are many advances that happened between the two.” Looking at other data, for example that from the Germany registry (Theiss et al; Stroke, 2004), then the differences actually vanish.Therefore, the investigator-sponsored Investigational Device Exemption (IDE) study that Makaroun has been involved in had this purpose: “to evaluate the effectiveness of filter cerebral protection devices in reducing cerebral emboli during carotid artery stenting using diffusion-weighted MRI”. This was a collaboration between four specialties at the University of Pittsburgh – vascular surgery, neurosurgery, neurology and neuroradiology – and included only patients with more than 70% stenosis at high risk for carotid endarterectomy (CEA). Subjects in the trial underwent carotid stenting and were randomised by symptoms to either receive protection or not. “The endpoint was the presence and size of defects on diffusion-weighted MRI, read by a ‘blinded’ neuroradiologist,” he added.
The trial ran for two years starting December 2003. Of more than 260 patients treated in that time, only 36 were recruited for the IDE study; “18 were treated without cerebral protection and all were technically successful”, he noted. Eighteen patients were treated with protection but in two the procedure could not be completed. Two patients had a stroke following the procedure: one 82-year old female with protection who was asymptomatic, and an 84-year old asymptomatic female without protection.
Preliminary results show that MRI showed defects in 13 out of 18 (72%) patients with protection and 8 out of 18 (44%) without protection, while ipsilateral diffusion defects were seen in 12 (67%) of patients with protection and only 7 (39%) of those without. “Neither of these have reached significance yet,” Makaroun stated. Also non-significant were the average number of defects (6.1-6.2 across the two groups) and the mean defect size (16.63mm3 vs 15.61 mm3 respectively). These findings are not unknown in the literature. Other studies have also shown a higher defect rate for patients with filters than for those without. “In conclusion, cerebral protection devices may not reduce the incidence of new ischemic defects on diffusion weighted MRI; further investigation is definitely warranted. However, the clinical conclusion I’d like to make is – when needed – stenting without a filter is acceptable if the patient is at high risk for CAE.” However, Makaroun added a personal comment that outside of clinical trials he still uses protection.
At Charing Cross, Cremonesi defended protection devices by drawing attention to the fact that, using scientific data, “we have to demonstrate that by applying a protection device we can prevent stroke and complication rate during a procedure”. He highlighted a paper from 2003 that reviewed single center studies on protected and unprotected stenting published between January 1990 and June 2002. This included 2,357 patients undergoing 2,537 stenting procedures. “We have a 3.7% minor stroke rate without protection and a 0.56% rate with cerebral protection – at a highly significant p-value. Similarly, the major stroke rate was 1.01% against 0.33%, again significant.” Cremonesi noted that there were weaknesses with the paper: such literature analyses are necessarily retrospective and contain reported cases with heterogeneous study designs.
Therefore, Cremonesi and colleagues submitted a paper to Stroke (2005) on pooled analyses of only prospective clinical studies between 1994 and 2005 with very strict eligibility criteria. “This was a trade-off between ideal methodological standards and the need to obtain sufficient data,” he noted. This analysis included 32 studies organised into 36 groups: 10 with cerebral protection (1,222 patients), 13 without protection (1,638), and 13 with CEA (3,369). “The distribution of the large number of patients over a large number of independent studies compensated to some extent for the weaknesses of each individual study,” said Cremonesi.
The CEA group acted as a reference point: 30-day stroke/death rate of 2.6% and 30-day stroke/MI/death rate of 2.8%. In comparison, stenting with protection was at 2.7% and 2.6% respectively, stenting without protection was 5.4% and 6.1%. “If we go to the statistical comparison, we see there is no significant difference between protected carotid stenting and the CEA group, but a really significant difference between the protected and unprotected series.”
The question of the likelihood of plaque detachment and distal embolisation in daily practice is a separate issue, he commented. In one study Cremonesi et al undertook, out of 377 cases 66% resulted in macroscopically visible plaque debris removed from protection devices. In 9% of cases the particles were greater than 2mm in diameter (Euro Intervention, 2005).
“I would like to conclude in a very simple way, protected stenting appears feasible and safe with a major complication rate that compares favorably with that recorded in the CEA trials at Level 1- and 2 evidence. Unprotected carotid stenting demonstrated a highly significant difference in complication rate compared to protected procedures,” he concluded.
Macdonald made many of the same points as Makaroun as she opposed the motion that cerebral protection is mandatory. She showed images of a filter taken from an asymptomatic patient who had been stented by an experienced operator. “Look at the kind of stuff we catch in the filter; compelling. Hell, I’d use one myself and regularly do. But the bottom line is: is there any evidence?” Furthermore, she asked, ‘Does the use of cerebral protection reduce the incidence of adverse neurological events during carotid stenting?’
Criticizing her opponent for accepting ‘trade-offs in evidence’, Macdonald presented her own analysis of the evidence. Referring to the Wholey et al paper and to a large German registry over roughly the same time, she also highlighted the fact that the apparent reduction in adverse events was not a genuine conclusion as the groups were not concurrent. “I’ll show you the problem with that and I’ll show you the influence of confounding variables.”
Since 1996, there has been a huge uptake of cerebral protection devices, she noted. The earlier series of data was from a time when there were no protection devices available. These should not be compared to later studies. Furthermore, in the review that Cremonesi quoted was not, according to Macdonald, “a systematic review as I understand it”. The review did not include any randomised trials, it was mainly self-audited retrospective analyses, it contained very small studies, and there was wide heterogeneity in the study designs, materials used and patient populations. “The study authors themselves conceded that ‘increasing expertise within institutions might influence the complication rate’,” said Macdonald. “Self audit is notoriously unreliable.”
There are independent risk factors that cannot be controlled for using historical data, such as using a dedicated – rather than adapted – stent, having the right pharmacological regimen, and operator experience. “This is not rocket science.” Thus, looking at the data again but taking only post-2002 studies to obtain more homogeneity, there is “no significant difference between protected and unprotected series”, Macdonald stated. “So clearly what we have is a silent evolution that no-one really likes to talk about between the timeframe and use of cerebral protection devices.”
Prospective concurrent data from Theiss et al (2004) show not only no difference in stroke and death rates but also, crucially, an increase in transient symptoms with the use of a protection device. New data from Sheffield suggest that while previous myocardial infarction and taking clopidogrel were independent risk factors for stroke, use of cerebral protection was not. Many randomised prospective studies are showing an increase in lesions in the brain following use of filter devices, she explained.
“To summarise, if we look at registries evaluating historical controls – this is meaningless. If we look at those with concurrent controls there is no difference between protected and unprotected. If we look at systematic reviews that look at contemporary studies, there is no difference in outcome. If we look at multivariate analysis of a sizeable population, protection is irrelevant to outcome. And if we look at randomised trials based on surrogate markers we have increased spots on the brain and increased microembolic signals,” Macdonald stated. She finished by throwing one of Cremonesi’s quotes back at him. “You said that if they could the British would randomise their wives and children. Indeed the UK is a natural home for evidence-based medicine but when you quote series at me based on historical controls, you’re not giving me Level 1 evidence, but at best Level III-3 evidence.”