IABP is a mechanical support device that consists of a flexible 30–50-cc helium-filled balloon catheter attached to a console that times periodic inflation and deflation according to the cardiac cycle. The distal tip of the balloon should be placed in the descending aorta, approximately 1 cm distal to the origin of the left subclavian artery. The IABP was first placed by surgical cut-down of the femoral artery by Dr. Adrian Kantrowitz in the 1960s. Currently, implantation is usually done by a percutaneous (Seldinger) technique via the femoral approach, although surgical insertion in the subclavian artery [3‐5, 6•] or percutaneous introduction via the axillary artery [7•] is also possible.

Its physiological effect is dual. By inflating the balloon immediately after aortic valve closure, diastolic and mean arterial pressures rise and coronary perfusion improves. On the other hand, a vacuum effect—caused by rapid deflation of the balloon just before aortic valve opening—provides a reduction in left ventricle afterload and thereby passively augments cardiac output (CO) [8]. The hemodynamic effect will vary based on the clinical setting and the overall stroke volume. In vivo left pressure-volume loops, measured invasively with a conductance catheter, show an acute decrease in left ventricular end-systolic volume by 6%, a decrease in left ventricular end-systolic pressure by 18%, and an increase in stroke volume by 14% (see Fig. 1b) [9]. Left ventricle stroke work is reduced [10]. The primary objectives of the IABP are an increase in myocardial oxygen supply, a decrease in oxygen demand, and optimization of end-organ perfusion [10]. The bedside effects on aortic pressure curves are generally characterized by a decrease in systolic blood pressure, an increase in diastolic blood pressure, and an increase in mean arterial pressure (Fig. 1a) [8]. A reduction in pulmonary capillary wedge pressure (PCWP) and an increase in stroke volume can be measured with right heart catheterization or estimated with echocardiography [8].

IABP has been applied in a wide spectrum of indications.

Compared with other MCSDs like micro-axial pVADs (Impella, Abiomed, Danvers, MA; USA) and Tandem Heart (CardiacAssist Inc., Pittsburgh, PA, USA), extracorporeal centrifugal-flow LVAD, and VA-ECMO, complication rates of IABP are low. The reported incidence of adverse events in femoral IABP implantation ranges between 0.9 and 31.1% [26•, 27, 28•, 29, 30••], but these rates also include minor adverse events (e.g., access site hematoma, transient loss of pulsations, or need for blood transfusion). The most frequent device-related complication is (most often reversible) limb ischemia with a roughly estimated incidence of 5% (range from 0.9 to 26.7%) [27, 29, 31]. However, we have to consider that complications may be the result of the CS itself, since the complication rate in IABP supported patients was equal compared with controls in IABP-SHOCK [31]. When the IABP is implanted by an axillary or subclavian approach, the following complications have been reported: malfunction due to kinking, rupture, or migration requiring removal or reposition (15–37%), stroke (0–3%), upper limb ischemia (0–4%), transient brachial plexus injury (0–2%), mesenteric ischemia (0–3%), local vascular complications (0–7%), bacteremia requiring antibiotics (0–9%), and bleeding needing transfusion (0–16%) [4, 5, 7•, 32].

While cardiogenic shock following acute myocardial infarction (AMICS) was the main indication for an IABP for many years, the results of the IABP-SHOCK II trial in 2012, the largest IABP trial so far, caused a severe decline in its routine use [2, 33, 34]. In this RCT, 600 patients with AMICS were randomized to IABP or conservative therapy, both including routine revascularization [31]. No difference in all-cause mortality after 30 days was observed. On the other hand, IABP was not associated with increased adverse events like re-infarction, stent thrombosis, bleeding, sepsis, or stroke. In 2015, a meta-analysis of 7 RCTs including 790 patients with AMICS showed similar results of no survival benefit by the routine placement of an IABP in this population [35]. As a consequence of these results, both European and American guideline recommendations were downgraded (ESC: III/B; ACC/AHA: IIb/B) [24, 36, 37]. Because only 13% of patients in the IABP group of the IABP-SHOCK II trial received the IABP before revascularization, a meta-analysis including 1348 patients with AMICS was performed in order to clarify the role of the timing of its placement [38]. However, no difference was seen with respect to short- or long-term (≥ 6 months) survival between patients supported upstream or only after primary PCI. Also, no significant outcome difference in terms of re-infarction, repeat revascularization, stroke, renal failure, and major bleeding was seen.

It is hypothesized that the Impella device, by direct unloading, may reduce infarct size, particularly when starting pre-PCI in patients with AMICS who are revascularized [39]. Patients with CS who were treated with pVAD (Tandem Heart® or Impella®) had a significantly higher mean arterial pressure and a faster decrease in lactate levels compared with patients treated with IABP [40]. However, in the same meta-analysis including 148 patients, no significant difference in 30-d mortality was seen, whereas bleeding occurred more frequently in patients with pVAD (RR 2.50; P < 0.001) [40]. Of note, sample sizes of the 4 RCTs included in this meta-analysis were small. Critics also emphasize that 92% of patients in the latest IMPRESS (IMPella versus IABP REduces mortality in STEMI patients treated with primary PCI in Severe cardiogenic SHOCK) study had been resuscitated from cardiac arrest, resulting in a 46% death rate due to anoxic brain damage [41].

Two important observational studies were recently published. First, Schrage and colleagues retrospectively matched 237 patients with AMICS treated with Impella to an equal number of patients from the IABP-SHOCK II trial treated with medical therapy or IABP [42]. The authors found no significant difference in 30-d all-cause mortality, while severe or life-threatening bleeding and peripheral vascular complications occurred significantly more often in the Impella group. Second, in a large US retrospective study including 1680 propensity-matched paired patients with AMICS undergoing PCI, there was a significantly higher risk of in-hospital death and major bleeding associated with the use of pVADs compared with treatment with IABP (45% vs 34% and 31% vs 16% respectively; P for both < .001) [43]. These findings were remarkable since patients with pVADs were significantly younger and less likely to have STEMI compared with patients treated with IABP.

In the past decade, a larger-capacity (50-cc) IABP was introduced into clinical practice. Compared with previously used 40-cc IABPs, patients who received a 50-cc IABP showed higher-peak augmented diastolic pressure, higher magnitude of diastolic augmentation, and a greater slope and magnitude of deflation pressure from peak augmented diastolic pressure to reduced aortic end-diastolic pressure [44]. In 50-cc IABP recipients, diastolic pressure and PA occlusion pressure were reduced, and CO, cardiac index, and PA oxygen saturation were increased, while these PA catheter–derived measurements did not significantly change in patients with a 40-cc IABP. The absolute increase in CO was 1.4 ± 1.0 L/min in the 50-cc IABP group versus 0.7 ± 0.9 L/min in the 40-cc IABP group, which represented a relative increase of CO compared with baseline of 40% and 18% respectively (P = .08). Fifty cubic centimeters IABP also resulted in a greater systolic unloading and a larger reduction in pulmonary capillary occlusion pressure, compared with 40-cc IABP. The magnitude of systolic unloading correlated directly with the magnitude of diastolic augmentation and inversely with the PA occlusion pressure [44]. Also in later studies, 50-cc IABP caused significant diastolic pressure augmentation (Δ + 42 mmHg), systolic unloading (Δ − 15 mmHg), increased CO (Δ + 1.03 L/min), and decreased cardiac filling pressures in the majority of patients [45, 46].

Although the use of IABP in patients with AMICS is now controversial, 20–70% of all CS is not caused by an ACS [2, 47‐49]. This non-ACS CS group (also defined as ADHF-CS: acute decompensated HF with cardiogenic shock) includes acute decompensated chronic HF but also CS as a presentation of de novo HF. Importantly, this group seems to be a different population with regard to age, gender, ventricular function, and ventricular dimensions [2, 47, 49, 50••]. Patients with non-ACS CS also have less atherosclerotic cardiovascular risk factors and are more likely to have chronic kidney disease and pre-existing HF, compared with patients with AMICS [47, 48, 50••]. In contrast to AMICS, the etiology of non-ACS CS is diverse, reaching from temporary cardiac disturbances like arrhythmias (responsive to interventions or even self-limiting) until expressions of end-stage HF without any traceable provoking events. Although the role of IABP in this population remains insufficiently defined, several small uncontrolled studies have been performed in order to elucidate its feasibility in this subgroup. These studies are summarized in Table 1.

A study of particular interest is the one by Malick and colleagues, in which the effect of IABP placement was directly compared between patients with AMICS (n = 73; 36%) and those with non-ACS CS (n = 132; 64%) [50••]. Baseline characteristics showed that patients with non-ACS CS had significantly higher PAP (mean 38 ± 9 vs 31 ± 8 mmHg), lower LVEF (18 ± 9 vs 30 ± 12%), higher left ventricular end-diastolic dimension (7 ± 1 vs 5 ± 1 cm), higher serum creatinine (1.97 ± 1.06 vs 1.59 ± 1.11 mg/dL), lower serum lactate (2.54 ± 2.50 vs 4.92 ± 4.21 mmol/L), higher PA pulsatility index (2.91 ± 3.35 vs 2.00 ± 1.69), and more vasoactive agents (1.7 ± 1.0 vs 1.4 ± 0.8). Interestingly, patients with non-ACS CS experienced a 5-fold greater CO augmentation compared with patients with AMICS (0.58 ± 0.79 L/min vs 0.12 ± 1.00 L/min; P = 0.0009). Patients with non-ACS CS experienced an increase by almost a quarter (24%) of their baseline CO, while the increase in patients with AMICS was only 10% (P = 0.02). Systemic vascular resistance decreased significantly in non-ACS CS patients but remained equal in patients with AMICS (P < 0.05).

We recently performed the first RCT regarding IABP therapy versus inotropy in the early phase of non-ACS CS [30••]. The population included both de novo and acute on chronic HF patients without signs of acute ischemia. All patients (n = 32) had a systolic blood pressure of < 100 mmHg, fluid retention, at least moderate tricuspid valve regurgitation and/or mitral valve regurgitation, a dilated inferior cava vein, high filling pressure, low CO, a neutral or positive fluid balance despite fluid restriction, and high-dose intravenous loop diuretics, together with dysfunction of at least 1 other organ. Sixteen patients were treated with a 50 cc IABP and 16 with inotropes. After 48 h, those treated with IABP had significant higher central venous oxygen saturation (+ 17 vs. + 5%), a better increase in cardiac power output (+ 0.27 vs + 0.09 W/m²), lower N-terminal pro B-type natriuretic peptide levels (− 59 vs − 16 ng/L), a more negative cumulative fluid balance (− 3.066 vs − 1.198 L), and a better decrease in dyspnea severity score (− 4 vs − 2). In addition, mean arterial pressure increased more in the IABP group, and mean PAP and PCWP decreased more in the IABP group. Fewer patients in the IABP group ended up with moderate to severe mitral valve regurgitation. Finally, patients treated with an IABP tended to have lower major adverse cardiovascular events (a combined end point of crossover or other escalation of therapy, death, HF, re-hospitalization or TIA/stroke) (38% vs 69%), and mortality at 90 days (25% vs 56%), when compared with the group of patients who were treated by inotropes only.

Although other MSCDs like Impella, Tandem Heart, or VA-ECMO provide more hemodynamic support, (first-line) IABP has multiple advantages. First of all, it is relatively cheap [1] and IABPs are largely available and applicable, also in non-tertiary centers. Insertion of an IABP device is more straightforward and can be performed in the intensive care unit without the need for fluoroscopy. Compared with other devices, IABP placement is associated with fewer adverse events like vascular complications [58] or hemolysis [39]. Although mobilization of patients with femoral IABPs is compromised, placement in the axillary or subclavian artery allows mobilization and early physical rehabilitation [3‐5, 6•, 7•]. When the IABP fails or cannot be weaned, rapid escalation is possible to percutaneous MCSDs, VA-ECMO, or advanced HF therapies like durable MCSDs (e.g. LVAD) or orthotopic heart transplant (OHT) [59]. Finally, an IABP is easily removed and the presence of an IABP does not complicate native heart excision in case of bridging to OHT.

The hemodynamic effects of an IABP stand out better with larger balloon size. Several recent studies demonstrate that the use of larger 50-cc balloons resulted in a greater reduction in cardiac filling pressures and increased CO compared with the 40-cc IABPs [44‐46]. Unfortunately, 50-cc IABPs were generally not used in the major landmark studies so far, since the 50-cc IABP was only introduced in 2012. Since the number of patients achieving optimal hemodynamic benefit from IABP activation may be < 50% with the older 30–40-cc IABPs, this could potentially have contributed to the failure of previous IABP studies [44].

Although the supposed additional beneficial effect of improved coronary blood flow by IABP would be expected to be extra beneficial for patients with AMICS, IABP-SHOCK II showed no benefit of survival [31]. Several limitations of the IABP-SHOCK II should be mentioned. As discussed previously, most patients were treated with conventional, small-volume IABP-catheters. Besides, 10% of patients in the control group experienced crossover to IABP. Moreover, since almost half of all patients were included after cardiopulmonary resuscitation, a substantial amount might have died due to post-anoxic damage. Finally, a large percentage of patients in this trial were already on vasopressors/inotropes (90%), and thus IABP therapy might have been initiated too late.

Besides the limitations of this study, there are also several possible pathophysiological explanations for the neutral findings of IABP in patients with AMICS. First, ACS-driven (extensive) myocardial damage triggers inflammatory and other systemic responses, which may be insufficiently counter-attacked by an IABP that only passively supports the circulation [37]. Second, the effect of improved coronary blood flow is possibly non-existent in vivo due to intact coronary autoregulation [13]. Hence, Van Nunen and colleagues postulated the hypothesis that IABP only improves coronary blood flow in case of exhausted coronary autoregulation, which was not the case in IABP-SHOCK II, since 90% of the total study population obtained successful reperfusion (i.e., final TIMI flow grade 2 or 3 in the infarct-related artery (IRA)) [13, 31]. Patients with AMI and persistent ischemia despite primary PCI were supposed to have impaired autoregulation and Van Nunen proved that the IABP resulted in more rapid ST-elevation resolution in this subgroup. Also, death, necessity of LVAD implantation, or re-admission for HF tended to occur less frequently after IABP implantation in this subgroup [13]. Hawranek retrospectively evaluated patients with AMICS from the prospective nationwide registry who had unsuccessful PCI (i.e., final TIMI flow grade 0 to 1 in the IRA) [60•]. Although conclusions are limited by its observational design, IABP in this subgroup was associated with lower short-term and 12-month mortality.

Due to improved survival after ACS, the incidence of end-stage HF and non-ACS CS is rising [61]. However, at this time, no large RCTs for the acute mechanical treatment of this subgroup are available [36]. The first (small) RCT showed significant improvement of central venous oxygen saturation, cardiac power output, and urine output by IABP compared with medical therapy [30••]. Baseline hemodynamic parameters were equal to those reported in previous studies on AMICS [62]. Besides, as we show in Table 1, multiple retrospective studies reported that the use of an IABP in non-ACS CS temporarily stabilized hemodynamics and end-organ perfusion and allowed a bridge to recovery of the native cardiac function, decision-making, or more durable heart replacement therapy like OHT and LVAD. The increase of the cardiac index in non-ACS CS ranged from 0.3 to 0.9 L/min/m² [6•, 28•, 32, 50••, 54], and one may imagine that such a (limited) CO augmentation may be sufficient to stabilize patients with chronic HF and CS who are used to have a low CO under stable conditions. Previous studies of patients with AMICS demonstrated less CO augmentation by IABP [62‐64], which probably explains the lack of efficacy in (tachycardic) patients suffering from an acute decrease in stroke volume as included in the IABP-SHOCK II trial [31].

Malick et al. also described that the augmentation of CO occurred to a less extent in patients with AMICS [50••]. The exact reasons for the difference in treatment response between non-ACS CS and AMICS remain unclear. One hypothesis is that IABP support depends on the intrinsic contractile reserve [50••, 65]. Although baseline stroke volume may be identical in AMICS versus non-ACS CS [50••], baseline PAP was higher in non-ACS CS. Since low output may be mainly triggered by high filling pressures in non-ACS CS, and the IABP may be more effective in lowering afterload and optimizing renal perfusion in this subgroup, the IABP may function better in a high-volume status rather than in an acutely developed low-flow contractile state. This explanation is supported by Fried’s finding that non-ACS patients with high baseline mean PAP had the greatest CO augmentation by IABP [28•]. Also in Imamura’s study, patients with higher filling pressures were most likely to benefit from IABP support [6•].

The proportion of patients successfully weaned from IABP in CS is significantly lower in patients with STEMI compared with patients with NSTEMI and congestive HF (P = 0.04) [66]. In this retrospective analysis, even 97.8% of congestive HF patients were weaned from IABP support [66]. In Thiele’s IABP-SHOCK II trial, only 4% of patients who received an IABP were bridged to durable mechanical circulatory support with good long-term outcome [31], and in most other AMICS studies, the rates of successful bridging to durable heart replacement therapy were unfortunately not reported [59]. As shown in Table 1, many patients with non-ACS CS treated with IABP were successfully bridged to durable heart replacement therapy like LVAD or OHT. In our recently published RCT, non-ACS CS patients treated with IABP were significantly more often bridged to LVAD or OHT compared with patients treated with inotropes (31 vs 0% respectively; P < 0.05) [30••]. Recent literature shows that patients with ischemic or non-ischemic heart failure who needed pre-operative IABP have similar short- and long-term survival rates after LVAD implantation (88% and 78% after 3 and 12 months respectively), compared with patients who received LVAD without the need for pre-operative mechanical circulatory support (91% and 82% after 3 and 12 months respectively) [67••]. Also, after OHT, no significant difference in short- or long-term survival post-OHT between pre-OHT IABP and a control group was seen [52]. Unfortunately, most studies looking specifically at IABP in non-ACS CS (Table 1) did not report long-term survival rates.

As already mentioned, CS cannot be seen as one single entity, but rather as a wide spectrum of different aetiologies, hemodynamic characteristics, degree of severity, and response to therapy. This heterogeneity is the main reason that estimating the possible effect of IABP in daily clinical practice remains challenging. Even within the non-ACS CS subgroup, part of the patients appeared to be non-responders [28•]. In 60/75 patients who underwent right heart catheterization in the before-mentioned cohort of Visveswaran, CO and cardiac index increased up to 7 L/min and 3.4 L/min/m² respectively, while in the remaining 20% non-responders CO decreased. Remarkably, the mortality rate between responders and non-responders was equal [46]. In Hsu’s study, all patients showed an initial improvement in CO within the first 24 h, but in patients with adverse events, CO declined after 24–48 h post IABP implantation [26•]. Some authors suggest that the IABP is less effective in patients with non-ACS CS and underlying ischemic cardiomyopathy [26•, 30••]. Others showed that patients with too bad left and/or right ventricle function at baseline were less likely to show clinical stabilization after IABP insertion [10, 26•, 28•, 55, 56, 68]. Many other prognostic parameters at baseline have been proposed (e.g., left ventricular end-diastolic pressure, left ventricle end-systolic pressure, end-systolic pressure-volume relationship, dP/dTmax, right atrial pressure, PAP, right atrial pressure to PCWP ratio, PCWP, left ventricular end-diastolic dimension, heart rate, systemic vascular resistance, absence of biventricular failure, and the degree of inflammation and multi-organ dysfunction), but most study populations were small, sometimes data are conflicting, and underlying mechanisms remain insufficiently understood [6•, 7•, 10, 28•, 30••, 44]. Also, the fact that persisting arrhythmias can cause opposite disadvantageous hemodynamic effects in patients with IABP should always be taken into consideration [4, 8].

Although recommended as first-line therapy of CS [36], the beneficial effect of intravenous positive inotropes and/or vasopressors is never proven and observational data even point towards increased mortality [69, 70]. Possible deleterious effects can be explained by an increased incidence of arrhythmias and aggravation of myocardial ischemia. Since primary IABP placement showed substantial and fast hemodynamic benefit as compared with inotrope therapy [30••], early IABP implantation might result in better outcomes. In Gul’s study, placement of IABP within 1 h of onset of CS showed remarkably lower mortality compared with delayed implantation (35% vs 49% respectively; P < 0.001) [27], suggesting that early IABP placement instead of waiting too long for the possible benefit of inotropes could be beneficial. This is endorsed by the finding that patients who stabilized after IABP were on fewer vasopressors or inotropes in observational studies [28•, 55]. Unfortunately, in the currently available retrospective studies regarding non-ACS CS (Table 1), the timing of IABP insertion and phase of shock is very heterogeneous and sometimes poorly defined. Also in this population, the timing of implantation seems to be a crucial factor, since the time to mechanical support is proportional to the amount of organ preservation. Finally, also the timing of IABP weaning seems to be crucial and is actually poorly defined in previous studies.

Results of randomized trials like the DanGer Shock and ECLS SHOCK are expected to elucidate the effect on LVEF and mortality by respectively Impella CP and ECMO in patients with AMICS. Since IABP might still provide benefit in selected patients with AMICS and unsuccessful revascularization or patients with non-ACS CS, larger RCTs are required to evaluate its effect in those patients. We would recommend hemodynamically guided placement of IABP in those subgroups. Investigators should preferably evaluate not only outcomes like short-term mortality, but also time to reversal of shock, end-organ failure, duration of hospital stay, and long-term mortality and functionality.