The TOTAL trial, the design of which was previously published, was an international, investigator-initiated, multi-centre prospective randomized trial of routine manual thrombectomy with the Export catheter (Medtronic Inc., Santa Rosa, CA, USA) vs. PCI alone. Patients with STEMI who were referred for primary PCI within 12 h of symptom onset were eligible to participate. Although the trial design was open label, the adjudication of outcomes was blinded and neurologists specifically adjudicated all stroke events.
In the thrombectomy group, aspiration thrombectomy was to be the first procedure after wire crossing. The guide catheter was to be fully engaged with the coronary ostium during removal of the thrombectomy catheter to avoid embolizing thrombus to the systemic vasculature. Following each thrombectomy run, the guide catheter was to be aspirated to ensure removal of air or thrombus. It was recommended that a minimum of two syringes be aspirated during thrombectomy.
In the PCI alone group, thrombectomy was allowed as a bailout procedure if there was failure of the initial PCI alone strategy, defined as either thrombolysis in myocardial infarction (TIMI) 0 or 1 flow with large thrombus after balloon pre-dilation or large thrombus persisting after stent deployment.
The trial protocol recommended use of guideline recommended PCI pharmacotherapies but did not mandate a specific regimen.
The primary efficacy outcome was the occurrence of CV death, recurrent MI, cardiogenic shock, or new or worsening New York Heart Association class IV heart failure within 180 days. The key safety outcome was stroke within 30 days. A central events committee, blinded to treatment allocation, adjudicated all of the primary outcome events, strokes, transient ischaemic attacks (TIAs), major bleeding, target vessel revascularization, and stent thromboses. Neurologists who were members of a central adjudication committee, blinded to treatment assignment, specifically adjudicated all stroke events.
Stroke was defined as the presence of a new focal neurologic deficit thought to be vascular in origin, with signs or symptoms lasting >24 h. It was strongly recommended that an imaging procedure such as computerized tomography (CT) or magnetic resonance imaging (MRI) be performed to confirm the diagnosis. Transient ischaemic attack was defined as a transient episode of a new focal neurologic deficit thought to be vascular in origin, with signs or symptoms lasting <24 h.
The modified Rankin score (Supplementary material online, Table S1 http://eurheartj.oxfordjournals.org/content/suppl/2015/06/29/ehv296.DC1) was used to assess disability from stroke over a range from 1 (full recovery) to 6 (fatal stroke). Stroke was categorized as stroke with minor or no disability if the modified Rankin score was ≤2 and as stroke associated with major disability or fatal if the Rankin score was 3–6 as previously published.
Stroke subtype was categorized by neurologists as either ischaemic, or primary haemorrhagic (intra-cerebral or subarachnoid haemorrhage) based on CT or MRI, or as uncertain (no neuroimaging or autopsy). Ischaemic strokes that had had secondary haemorrhagic transformation were categorized as ischaemic in aetiology. The diagnosis of haemorrhagic stroke subtype was based on local interpretation of imaging and reviewed by a senior stroke neurologist adjudicator, with images obtained and reviewed by a neuroradiologist when local interpretation was equivocal. Images were not routinely reviewed by a central core laboratory for neurologic imaging. Neurologic events which include TIA were categorized as ischaemic (ischaemic, unknown stroke, or TIA) or primary haemorrhagic.
The primary analysis was conducted using a modified intention-to-treat approach that was pre-specified to include only randomized patients who had undergone primary PCI. Other pre-specified sensitivity analyses included full intention-to-treat, on-treatment, and per-protocol analyses as described previously.
A two-sided, log-rank test was used to compare the two randomized groups with a P-value <0.05 considered significant. The hazard ratios (HRs) and their 95% confidence intervals (CIs) were estimated using a Cox proportional hazards regression model with treatment group as the predictor variable.
To better understand the time course of stroke in relation to the procedure, a series of cut points were used for landmark analyses (0–48 h, >48 h to <7 days to <30 days, 30 to <90 days, 90 to 180 days). This was performed for overall stroke, ischaemic and haemorrhagic neurologic events.
The subgroup analyses for stroke examined procedural factors that may impact stroke as well as traditional risk factors for stroke and included: TIMI thrombus grade <4 vs. 4 or greater, initial TIMI flow (0–1 vs. 2–3), age (65 or younger vs. over 65 years), centres divided into tertiles of annual primary PCI volume, radial vs. femoral access, prior stroke, history of hypertension and diabetes.
A stepwise multivariable analysis using a Cox proportional hazards model was performed to determine independent predictors of stroke after primary PCI for STEMI. Known predictors of stroke after PCI were used including: age, sex, prior stroke, peripheral arterial disease, diabetes, intra-aortic balloon pump use, coronary bypass surgery during follow-up, prior hypertension, smoking, and additional PCI variables were tested including PCI procedure time, staged PCI, initial TIMI flow, radial access, GP IIb IIIa use, TIMI thrombus grade, and thrombectomy group.
We performed an updated meta-analysis of manual thrombectomy vs. PCI alone in patients with STEMI (see Supplementary material online, Appendixhttp://eurheartj.oxfordjournals.org/content/suppl/2015/06/29/ehv296.DC1 for detailed methods). A comprehensive systematic search strategy was performed of Medline, EMBASE, and Cochrane for all published studies up to April 12, 2015. Randomized trials of manual thrombectomy vs. PCI alone in patients with STEMI were included. Two reviewers independently reviewed abstracts and extracted data. The outcomes of CV mortality, myocardial infarction, and stroke were examined. When CV mortality was not available all-cause mortality was used. Odds ratios were calculated using the Mantel Haenszel method with a fixed effects model. Sensitivity analyses were performed using a random effects model.