Clinical Research & Data, Patient Management, Safety & Efficacy, Unloading, Surgical Applications

What is the STEMI DTU Pilot Trial?

1. What is the STEMI DTU pilot trial?

Key points from the Door-To-Unload (DTU) in STEMI pilot trial:


  • The STEMI DTU pilot trial demonstrated for the first time that LV unloading using the Impella CP® heart pump with a 30-minute delay before reperfusion is safe and feasible
  • No prohibitive safety signals that would preclude proceeding to a larger pivotal study of LV unloading and delaying reperfusion for 30 minutes were identified
  • Since it was an exploratory trial with a small sample size, no difference in infarct size was expected and observed with LV unloading for 30 minutes and delayed reperfusion vs. LV unloading for 30 minutes and immediate reperfusion
  • In patients with sum STE>6mm, infarct size normalized to the myocardial area at risk was significantly lower with LV unloading for 30 minutes and delayed reperfusion compared to LV unloading and immediate reperfusion

Note: STEMI = ST-segment elevation myocardial infarction; DTU = Door-to-unload

The current guideline-recommended therapy for acute myocardial infarction (AMI) due to occlusion of a coronary artery is rapid revascularization to enable primary coronary reperfusion and limit infarct size.1 However, reperfusion of an occluded artery also contributes to reperfusion injury and increases infarct size. 2

Recent pre-clinical studies suggest that primary ventricular unloading using LV assist devices such as the Impella® heart pump prior to coronary reperfusion reduces infarct size.1,3

The STEMI DTU pilot trial (NCT03000270) was a multicenter, prospective, randomized trial to assess the impact of primary unloading with Impella CP® heart pump prior to revascularization on infarct size in patients with anterior wall STEMI. Patients with cardiogenic shock were not included in the trial.


2. What is the rationale for STEMI DTU pilot trial?

Despite significant decrease in death due to acute myocardial infarction (AMI) since 1960, there has been a parallel increase in incidence of heart failure.4 The pooled patient-level analysis from 10 randomized trials showed that infarct size after AMI is strongly associated with mortality and hospitalization for heart failure during 1-year follow-up (Figure 1).5 Hence, there is a major clinical need for interventions that may reduce myocardial damage during myocardial infarction.


Braunwald et al. in 1971 suggested “that measures designed for reduction of myocardial oxygen demands and improvement of coronary perfusion, when effected promptly might reduce the ultimate size of (a myocardial) infarction”.6 Since 1978, multiple preclinical studies have tested whether reducing myocardial oxygen consumption by implementing a circulatory pump limits myocardial damage in AMI. Studies testing mechanical unloading using mechanical devices have demonstrated that unloading before, not after reperfusion, is required to reduce infarct size.4

Recently, 5 pre-clinical studies provided insights into the mechanisms through which LV unloading by the Impella® heart pump limits myocardial damage.

In 2015, using a porcine model of AMI, Kapur et al. demonstrated that first unloading the LV using Impella CP® and then delaying reperfusion for 60 minutes resulted in reduced LV mean wall stress and peak wall stress compared with immediate reperfusion.3 In fact, the magnitude of wall stress reduction correlated directly with the magnitude of infarct size reduction. In addition, the study demonstrated that LV unloading with delayed reperfusion activated cardioprotective signaling system that decreased cell death and resulted in a reduction in myocardial infarct size by 43%.

Sun et al. explored the effects of the surgically implanted Impella LD® heart pump on LV remodeling and function at 1 month postinfarction in a porcine model.7 LV unloading with Impella LD was initiated 30 minutes before reperfusion and support was maintained until 120 minutes after reperfusion. LV unloading resulted in lower LV pressure and wall stress and higher mean arterial pressure and cardiac output. Importantly, at 1-month follow-up, the infarct size in the LV unloaded pigs was reduced by 50%. Based on these results, the authors concluded that early assistance with Impella LD in AMI limited LV remodeling, thereby improving prognosis.

Further, Saku et al. assessed the effects of different degrees of LV unloading using Impella CP in a canine model of AMI.8 LV unloading with Impella CP was initiated at 60 minutes after the onset of ischemia and support was continued until 60 minutes after reperfusion. At 4 weeks after ischemia and reperfusion, the infarct size was reduced by 48% with partial support and by 87% with total support.

Watanabe et al. compared the effects of mechanical unloading with Impella CP to pharmacological unloading using intravenous sodium nitroprusside (SNP) in a porcine model of AMI.9 The results showed that LV wall stress decreased in both groups, although SNP-treated pigs developed severe hypotension. Additionally, microvascular perfusion increased 2-fold within the infarct zone with Impella CP with no difference in the SNP group. In fact, microvascular perfusion to the infarct area correlated inversely with LV end-diastolic wall stress (EDWS), demonstrating the role of wall stress in regulating tissue perfusion during MI. Taken together, this study showed that LV unloading using Impella CP during AMI, in addition to reducing LV wall stress and myocardial oxygen consumption, may improve microcirculatory blood flow, thus promoting myocardial recovery.

Esposito et al. investigated the kinetics of delayed reperfusion and the mechanism of the cardioprotective role of primary unloading in a porcine model of AMI.1 They showed that LV unloading for 30 minutes prior to reperfusion was necessary and sufficient to limit infarct size by 40% after AMI. Activation of Impella support immediately after reperfusion or 15 minutes before reperfusion did not result in the reduction of infarct size. Further, they showed that primary unloading compared to primary reperfusion increased the expression of genes associated with cellular respiration and mitochondrial integrity. Taken together, the results demonstrated that primary unloading and delaying reperfusion by 30 minutes reduced LV scar size (see figure) and expression of biomarkers associated with maladaptive cardiac remodeling and improved cardiac function.

These pre-clinical findings suggesting the benefit of LV unloading and delaying reperfusion by 30 minutes may challenge the contemporary guideline-recommended therapy of rapid reperfusion for STEMI. Hence, a pilot trial assessing the feasibility and safety of primary LV unloading and delaying reperfusion was conducted before attempting a randomized control trial evaluating the efficacy of this approach.

The STEMI DTU pilot trial was the first exploratory study in humans to assess the feasibility and safety of LV unloading and delayed reperfusion as a treatment modality for reducing infarct size in patients with STEMI without cardiogenic shock.10



3. What are the results and clinical implications of the STEMI DTU pilot trial?

In the STEMI DTU pilot trial, 50 patients presenting with anterior STEMI at 14 centers in the United States were randomized to mechanical unloading with the Impella CP® followed by immediate reperfusion (U-IR) or LV unloading with a 30-minute delay to reperfusion (U-DR).10

In this trial, interventional cardiologists were allowed to shorten the time between unloading and reperfusion in the U-DR arm of the study based on clinical judgment. Although permitted by the trial protocol, none of the patients assigned to the U-DR arm received bailout coronary reperfusion before the 30-minute delay. The mean unloading to balloon time in the U-DR group was 34.1 minutes compared to 11 minutes in the U-IR (p = 0.002) (see figure). Consequently, the door-to-balloon (DTB) time was longer in the U-DR group than the U-IR group (96.7 mins vs. 72.6 mins, p = 0.002).


The primary safety endpoint was major adverse cardiovascular and cerebrovascular events (MACCE) at 30 days. MACCE rate at 30 days after STEMI was 8% (2/25) in the U-IR group vs. 12% (3/25) in the U-DR group (p = 0.99). This result suggested no increase in major adverse events including cardiovascular mortality, reinfarction, stroke, and major vascular events at 30 days with LV unloading first followed by a 30-minute delay before reperfusion or immediate reperfusion.

An assessment of infarct size normalized as a percent of total LV mass at 30 days using cardiovascular magnetic resonance (CMR) imaging was also performed. Since it was an exploratory trial (with a small sample size and intent to inform the design of a larger pivotal trial), no statistical difference in infarct size was expected. The infarct size normalized to total LV mass at 30 days was 13.1% in the U-DR group vs. 15.3% in the U-IR group (p = 0.53). No difference in the mean infarct size was observed between the groups both at 3-5 days or 30 days after STEMI.

A subgroup analysis of infarct size normalized to the myocardial area at risk (AAR) was performed in patients with large anterior STEMI (defined by sum ST-segment elevation ≥ 6 mm). The mean infarct size normalized to AAR was lower in the U-DR group (n=14) than U-IR group (n=16) (44.1% vs. 59.9%, p = 0.04). Also, the rates of microvascular obstruction were lower among patients in the U-DR group than U-IR group (1.5% vs. 3.8%, p = 0.12).

The results of the STEMI DTU pilot trial showed that LV unloading using Impella CP with a 30-minute delay before reperfusion is feasible in anterior STEMI. In addition, there was no difference in the rates of adverse events and infarct size with LV unloading and delayed reperfusion compared to LV unloading and immediate reperfusion.

Based on the results of the STEMI DTU pilot trial, an appropriately powered pivotal randomized trial in patients with anterior STEMI is planned.



View more resources


  1. Esposito, M.L., et al. (2018). J Am Coll Cardiol, 72(5), 501-514. 
  2. Yellon, D.M., et al. (2007). N Engl J Med, 357(11), 1121-1135.
  3. Kapur, N.K., et al. (2015).  JACC Heart Fail, 3(11), 873-882.
  4. Kapur, N.K., et al. (2018).  F1000Res, 7(F1000 Faculty Rev), 1852.
  5. Stone, G.W., et al. (2016).  J Am Coll Cardiol, 67(14), 1674-1683.
  6. Maroko, P.R., et al. (1971). Circulation, 43(1), 67-82. 
  7. Sun, X., et al. (2016).  Artif Organs, 40(3), 243-251.
  8. Saku, K., et al. (2018). Circ Heart Fail, 11(5), e004397.
  9. Watanabe, S., et al. (2018).  J Am Heart Assoc, 7(9), e004250.
  10. Kapur, N.K., et al. (2019). Circulation, 139(3), 337-346.