Pooled Analysis of VA, ACAS, ACST-1 & GALA Trials Risk factors for procedural MI, stroke and death among asymptomatic patients undergoing carotid endarterectomy Pooled Analysis of VA, ACAS, ACST-1 & GALA Trials Dylan MORRIS, Hongchao PAN, Alison HALLIDAY, Richard BULBULIA On behalf of the VA, ACAS, ACST-1 and GALA Collaborators Today I’m going to present some very early results on risk factors for procedural MI, stroke and death among asymptomatic patients undergoing carotid endarterectomy. MRC Population Health Research Unit Clinical Trial Service Unit & Epidemiological Studies Unit University of Oxford

Decision for carotid endarterectomy Procedural Hazards VS Long-term Benefit The decision for carotid intervention may not always straight forward in asymptomatic patients. There’s lots of different factors that come into play, such as the patients age and life expectancy, their absolute stroke risk which Peter will talk about later on, the potential benefits from successful carotid intervention which we now know is about a halving in the long-term risk of stroke from the trials, but also the procedural hazards of surgery. As the absolute benefits from carotid surgery are more moderate in asymptomatic patients, the procedural risks in this population are particularly importance. If someone has a very high procedural risk, then there may only be a small or no net benefit from carotid intervention. However if someone has a very low procedural risk then the net benefit from carotid endarterectomy may be large.

Purpose of this Study AIM: to identify important procedural risk factors for asymptomatic patients undergoing CEA The aim of this study, was to identify important risk factors for procedural MI, stroke and death for asymptomatic patients undergoing carotid endarterectomy.

Methods IPD of ‘surgically treated’ asymptomatic patients from 4 large CEA trials VA ACAS Restricted to surgical patients, events 0-30 days Major operative events: MI, stroke, death within 30 days of operation Odds ratios (RR) from logistic regression ACST-1 GALA We analysed individual patient data of surgically treated patients from four large carotid surgery trials, The VA trial, ACAS, ACST-1 and GALA. We conducted analyses looking at the risk of major operative events, including MI, stroke and death within 30 days of operation. The association of risk factors with these events was assessed using logistic regression, adjusting for age and sex, and stratifying by trial.

Trial Characteristics   VA ACAS ACST-1 GALA Recruitment 1983 - 1987 1987 - 1993 1993 - 2003 1999-2007 Participants 444 1 662 3 120 1 362 (asymptomatic) Region USA Europe Europe, Australasia Follow-up, Median [IQR] 5.7 [4.5-7.0] 4.8 [3.7-5.0] 9.0 [6.1-11.1] Up to 1y Just looking at these trials in more detail. As you might remember, the VA trial, ACAS trial and ACST-1 trial compared early carotid endarterectomy plus medical therapy to medical therapy alone in asymptomatic patients, across about 3 decades. The VA trial recruited 444 males from the USA between 1983-87, with just over 5-years median follow-up. The ACAS trial recruited over 1600 people from the US in the following years, from 1987 to 1993 with slightly shorter follow-up, and ACST-1 recruited over 3000 patients from Europe with very long follow-up. The GALA trial was different – it recruited over 3500 patients who were both symptomatic and asymptomatic, among whom a decision had already been made for carotid surgery. Patients were randomised to undergo carotid endarterectomy under local anaesthetic vs general anaesthetic. The primary outcome was 30-day MI, stroke or death. For the purposes of this study, we only included asymptomatic patients from the trial. Procedure: LA vs GA CEA + MT vs MT Alone

Trial Characteristics (Cont’d)   VA ACAS ACST-1 GALA Age ± SD 64.5 ± 6.8 67.2 ± 6.9 68.6 ± 7.5 70 ± 8.8 Sex (% Male) 100% 66% 70% IHD (%) 41% 42% 34% 36% Diabetes (%) 28% 23% 20% 25% As you can see here the mean age of participants in these trials range from about 65 to 70 years. The VA trial only included males, whereas the other trials included about one third females. Over a third of participants had ischaemic heart disease, and diabetes was relatively common. Regarding medical therapy, antiplatelet therapy and blood pressure lowering therapy were relatively common across the trials, however statin use changed. Statins weren’t used in VA. They were used infrequently in ACAS, and they were used increasingly across ACST-1 and GALA. Procedure: LA vs GA CEA + MT vs MT Alone

MI, Stroke and Death with 30 Days of CEA VA ACAS ACST-1 GALA* Procedures 229 1036 1841 1335 Events† 14 31 69 40 Event Risk 6.1% 3.0% 3.7% 3.2% This figure shows the number of asymptomatic carotid procedures across the trials, along with the risk of procedural events. Altogether there were about 4441 procedures, and 154 events. The risk of MI, stroke or death within 30 days of endarterectomy was 6% in the early VA trial, whereas it was closer to 3% in the ACAS, ACST and GALA trials. *Only asymptomatic patients included

Types of Events Event Type* Number Percent Stroke 111 72.1% MI 34 22.1% Death 9 5.8% Any Event 154 100% Regarding the types of events –stroke accounted for a large proportion of major procedural events, followed by MI and death. I’ve only included one type of event here per patient, so if someone suffered multiple events then they were listed as having a stroke or death. So in absolute terms there were probably a few more myocardial infarctions here. *Note. If multiple events, stroke taken as main event

Association of Traditional CV Risk Factors with Procedural MI, Stroke & Death (day 0-30) First lets look at the association between cardiovascular risk factors and procedural MI, stroke and eath. As you can see, ischaemic heart disease was the only risk factor significantly associated with major procedural events. People with a history of ischaemic heart disease had about a 50% high risk of event. Age and sex were not associated with procedural events, although keep in mind that these patients were already deemed fit for surgery, so there is some selection bias here. Systolic blood pressure and cholesterol were surprisingly of little relevance after adjusting for treatment of these conditions.

Association of Cerebrovascular Factors with Procedural MI, Stroke and Death (day 0-30) Next looking at cerebrovascular risk factors. Both high grade contralateral stenosis, and prior stroke were strong risk factors for procedural events. Both of these were associated with more than a 50% risk of MI, stroke or death. Interestingly, the percentage ipsilateral stenosis did not predict procedural events as you can see here. Odds ratio of 1.

Association of Surgical Factors with Procedural MI, Stroke & Death (day 0-30) Finally, looking at surgical factors – keep in mind that this is not a randomized comparison, so there may be reasons why some patients do and do not receive some of these interventions such as shunts, anaesthesia type, and eversion endarterectomy. None of the surgical factors were significantly associated with procedural MI, stroke or death which is consistent with previous randomized evidence.

Important Risk Factors for Procedural MI, Stroke & Death (day 0-30) So, just highlighting the most important risk factors in this study -High grade contralateral stenosis >80% -Prior stroke -and prior ischaemic heart disease were each associated with about a 50% higher risk of procedural MI, stroke or death.

Number of Risk Factors and Procedural Risk Well how about if you have more than one of these risk factors. This figure shows that the more risk factors you have, the higher the procedural risk. Compared to people with no risk factors, those with one risk factor had about an 80% higher risk of an event. Those with two risk factors had more than a doubling in the risk of stroke, and there were very few people in this study with all three risk factors. As you can see here the procedural risk increases from about 2% to 4% to 5% in this population based on risk of stroke.

Risk Prediction What if the risk is lower? ‘New benchmarks for CEA are a major stroke or death risk of 1.2%’* No risk factors: ~1% One risk factor: ~2% Two+ risk factors: ~3% Well we can make some generalisations. If the risk really is this low, and most of these people had few risk factors, then we may see absolute procedural risks as follows. Those with no risk factors, which I am assuming is most of the people in these contemporary registries, might have a risk of 1.2%. One risk factor increases this risk to 2.2%, two or more risk factors increases this risk to about 2.9%. Now these are very semi-quantitative estimates –what we’d like to do is apply this to some contemporary registry data to see how well these risk factors predict procedural events. *Munster et al. (2015), Neurology

Conclusion Simple characteristics (contralateral stenosis, prior stroke, IHD) may identify asymptomatic patients with higher risk of procedural MI, stroke & death So in conclusion, simple characteristics including contralateral stenosis, prior stroke and ischaemic heart disease may identify asymptomatic patients with higher risk of procedural MI, stroke and death.

ACST, ACAS, VA, GALA Trialists and to the participants who took part Acknowledgements ACST, ACAS, VA, GALA Trialists and to the participants who took part