The majority of published studies on thrombectomy in patients with AMI used aspiration catheters [11–17]. A major advantage of manual aspiration catheters is the ease of use. Two major limitations of these devices are the unpredictability of the efficacy, since in 30 % of cases of successful lesion crossing by the catheter, the aspiration is completely negative, and the high profile of the catheters that may promote embolization when the occlusion is crossed or prevent their utilization in tortuous, calcified, or small vessels. The routine use of manual aspiration catheters in the setting of AMI was supported by the positive results of some single center studies. In the Thrombus Aspiration during Percutaneous coronary intervention in Acute myocardial infarction (TAPAS) trial 1,071 patients were randomized to manual aspiration or conventional PCI, and 10 % of patients randomized to thrombus aspiration crossed to conventional PCI since the operator considered the target vessel too small or tortuous to allow the use of the aspiration catheter [15]. Thus, in this study any traumatic attempt to cross the lesion with the aspiration catheter in patients with a difficult anatomy was avoided. Despite the exclusion from aspiration of patients with difficult anatomy, particles could be retrieved in only 72.9 % of cases randomized to aspiration, and manual aspiration was associated with a better myocardial reperfusion as assessed by the surrogate angiographic primary end point of myocardial blush [15]. At 1 year follow-up, patients randomized to manual aspiration had a better survival than patients randomized to conventional PCI: the mortality rates were 3.6 and 6.7 %, respectively [16]. However, it should be outlined that the study was not powered for survival and the differences in survival could have been due to chance.

The results of the TAPAS trial were not confirmed by the TASTE trial, that was the largest trial comparing manual aspiration with conventional angioplasty in patients with AMI and sufficiently powered for clinical outcome [18, 19]. This multicenter study randomized 7,244 patients to manual aspiration or conventional PCI. Randomization to manual aspiration was not associated with improved survival at 1 month and at 12 months. The 30-day mortality rates were 3.0 % in the standard PCI arm and 2.8 % in the manual aspiration arm (HR 0.94 95 % CI 0.72–1.20, p =0.63). At 1 year, there were no difference in the composite of death, myocardial infarction and stent thrombosis between the 2 arms (17.7 % in the standard PCI arm, and 16.3 % in the manual aspiration arms). Another ongoing trial comparing manual aspiration with standard PCI is enrolling 10,700 patients and will provide a definite answer to the usefulness or futility of routine use of manual aspiration catheter in AMI (TOTAL; ClinicalTrials.gov number, NCT01149044).

The rheolytic thrombectomy (RT) system (AngioJet, Boston Scientific, Minneapolis, MN) consists of a dual lumen catheter with an external pump providing pressurized saline solution via the effluent lumen to the catheter tip. Multiple saline jets from the distal part of the catheter travel backwards at 390 mph, and create a localized negative pressure zone that draws thrombus where the jets fragment it and propels the small particles to the evacuation lumen of the catheter. The first 5 F generation catheter for coronary use was associated with a substantial device failure rate due to the inability to cross the lesion by the large and poorly trackable catheter, embolization, and vessel perforation. In a post-hoc analysis in a series of 70 patients with AMI enrolled in the VEGAS 1 and 2 trials, the device failure rate was 22 % [20, 21]. The second generation AngioJet catheter (XMI) and the more recent third generation catheter (Spiroflex) that are available are 4 F in size and have an improved design of the profile and of the opening of the jets allowing easy and nontraumatic navigation also in complex anatomy (tortuous or calcified vessels), and the ability to remove quickly large amount of fresh thrombus. The last generation catheter can cross the lesion without the need for pre-dilation in more than 95 % of the cases.

Four randomized trials tested the efficacy and safety of RT in different settings (Table 13.2). The efficacy of RT in decreasing procedural embolization and subsequent clinical adverse events was demonstrated by the VeGAS-2 trial that enrolled patients with a very high risk of embolization, such as patients with diseased venous grafts or native vessels with angiographic evidence of large thrombus [21]. Patients with AMI were excluded. The study, based on a sample of 352 patients compared RT with intravessel infusion of urokinase and showed a > 50 % reduction in 1-month major adverse events in patients randomized to thrombectomy (16 %and 33 % respectively, P < 0.001). The Florence-AngioJet randomized trial was a mechanistic small study based on a sample of 100 patients with a first AMI and the end points of the study were early ST-segment resolution, the corrected TIMI frame count, and the infarct size as assessed by technetium-99 m sestamibi scintigraphy at 1 month [22]. All end points were met. Patients randomized to thrombectomy before direct stenting had a higher incidence of early ST-segment elevation resolution (90 % vs 72 %, P = 0.022), lower corrected TIMI frame counts (18.2 ± 7.7 vs 22.5 ± 11.0, P = 0.032), and smaller infarcts (13.0 ± 11.6 % vs 21.2 ± 18.0 %, P = 0.010) as compared to patients randomized to direct stenting alone. By multivariate analysis, the only variables related to the early ST-segment resolution were randomization to thrombectomy (OR 3.56, 95 % CI 1.11–11.42, P = 0.032), and diabetes mellitus (OR 0.24, 95 % CI 0.07–0.86, P = 0.029). At 1 month, no patient died, or had reinfarction, and the 6-month clinical outcomes were identical in the 2 arms: the mortality rate was 2 % in both groups, and no patient had reinfarction.

The AIMI trial is a multicenter randomized trial that compared RT before stenting of the infarct artery with conventional PCI and was based on a sample of 480 patients [23]. The primary end point of the study was infarct size as assessed by sestamibi scintigraphy at 14–28 days after the procedure. The study showed larger infarcts in the thrombectomy arm as compared to the control arm (12.5 ± 12.13 % and 9.8 ± 10.92 % respectively, P = 0.03), and more importantly, an unexpected higher mortality in the thrombectomy arm at 1 month (4.6 % vs 0.8 %, P = 0.02) and at 6 months (6.7 % vs 1.7 %, P = 0.01). Final TIMI grade 3 flow was seen more frequently in the control arm as compared to the thrombectomy arm (97 % and 91.8 % respectively, P < 0.02). Several concerns in study design and in RT technique may explain the negative and harmful results of the study. The enrollment criteria did not include angiographically visible thrombus, and moderate to large thrombus (grade 3 and 4 according to TIMI thrombus score) was present in an unrealistic minority of patients at baseline angiography (21.3 % in the thrombectomy arm and 19.6 % in the control arm). This figure suggests a selection bias against the enrollment of patients with a large amount of thrombus and who could derive the strongest benefit from thrombectomy before coronary stenting. Unfortunately, the authors did not provide a screen fail registry, but other characteristics of the study patient cohort strengthen this suspicion. More than 1/3 of patients (35 %) had an already open infarct artery at baseline angiography, and more importantly the infarct size was very small in both arms, with similar normal left ventricular ejection fraction at the time of scintigraphic assessment (51.3 ± 11.53 % in the thrombectomy arm, and 52.3 ± 10.89 % in the control arm). Another concern of the study design was the exclusion from enrollment of patients with severe left ventricular dysfunction and cardiogenic shock. The exclusion of these high-risk patients is not easily explained considering that just in this type of patients a no-reflow due to PCI embolization may be immediately fatal. Finally, the nonuniformity of treatment may have introduced confounding effects favoring the control arm. Eight percent of patients randomized to thrombectomy did not have the treatment, procedural variables that may have a significant impact on the risk of no-reflow, such as predilation, or postdilation, or stent type were left to the discretion of the operator, as well as the thrombectomy technique, with a distal-to proximal approach used in 48 % of cases. The thrombectomy retrograde technique should be considered as inappropriate since with this technique, the activation of the thrombectomy catheter is made only after the positioning of the device across the occlusion favoring embolization before thrombectomy.