Strategies to attenuate micro-vascular obstruction during P-PCI: the randomized reperfusion facilitated by local adjunctive therapy in ST-elevation myocardial infarction trial

Abstract Background Microvascular obstruction (MVO) following primary percutaneous coronary intervention (PPCI) treatment of ST-segment elevation myocardial infarction (STEMI) contributes to infarct expansion, left ventricular (LV) remodelling, and worse clinical outcomes. The REFLO-STEMI trial tested whether intra-coronary (IC) high-dose adenosine or sodium nitroprusside (SNP) reduce infarct size and/or MVO determined by cardiac magnetic resonance (CMR). Methods and results REFLO-STEMI, a prospective, open-label, multi-centre trial with blinded endpoints, randomized (1:1:1) 247 STEMI patients with single vessel disease presenting within 6 h of symptom onset to IC adenosine (2–3 mg total) or SNP (500 μg total) immediately following thrombectomy and again following stenting, or to standard PPCI. The primary endpoint was infarct size % LV mass (%LVM) on CMR undertaken 24–96 h after PPCI (n = 197). Clinical follow-up was to 6 months. There was no significant difference in infarct size (%LVM, median, interquartile range, IQR) between adenosine (10.1, 4.7–16.2), SNP (10.0, 4.2–15.8), and control (8.3, 1.9–14.0), P = 0.062 and P = 0.160, respectively, vs. control. MVO (% LVM, median, IQR) was similar across groups (1.0, 0.0–3.7, P = 0.205 and 0.6, 0.0–2.4, P = 0.244 for adenosine and SNP, respectively, vs. control 0.3, 0.0–2.8). On per-protocol analysis, infarct size (%LV mass, 12.0 vs. 8.3, P = 0.031), major adverse cardiac events (hazard ratio, HR, 5.39 [1.18–24.60], P = 0.04) at 30 days and 6 months (HR 6.53 [1.46–29.2], P = 0.01) were increased and ejection fraction reduced (42.5 ± 7.2% vs. 45.7 ± 8.0%, P = 0.027) in adenosine-treated patients compared with control. Conclusions High-dose IC adenosine and SNP during PPCI did not reduce infarct size or MVO measured by CMR. Furthermore, adenosine may adversely affect mid-term clinical outcome. Clinical Trial registration ClinicalTrials.gov Identifier: NCT01747174; https://clinicaltrials.gov/ct2/show/NCT01747174


Background
Microvascular obstruction (MVO) following primary percutaneous coronary intervention (PPCI) treatment of ST-segment elevation myocardial infarction (STEMI) contributes to infarct expansion, left ventricular (LV) remodelling, and worse clinical outcomes. The REFLO-STEMI trial tested whether intra-coronary (IC) high-dose adenosine or sodium nitroprusside (SNP) reduce infarct size and/or MVO determined by cardiac magnetic resonance (CMR).

Introduction
Primary percutaneous coronary intervention (PPCI) is the default reperfusion therapy for ST-segment elevation myocardial infarction (STEMI). 1 However, residual mortality (up to 12% at 6 months 1 ) and morbidity may be partially attributable to sub-optimal microvascular perfusion. 2,3 Microvascular obstruction (MVO) can occur in up to 70% patients when detected with cardiovascular magnetic resonance (CMR) imaging. 4,5 Microvascular obstruction may contribute to infarct expansion, adverse left ventricular (LV) remodelling and adverse clinical outcomes, independent of infarct size. 4,5 Various strategies have been proposed to attenuate MVO 6 and have included calcium channel blockers, nicorandil, atrial natriuretic peptide, glycoprotein IIb/IIIa (GPIIb/IIIa) inhibitors, thrombolytics, and, perhaps most studied, vaso-active agents such as sodium nitroprusside (SNP) 2,7,8 and adenosine. 8 -12 Adenosine inhibits neutrophil-related processes central to the evolution of MVO and in a canine model, reduced ischaemia-reperfusion injury (IRI), limited infarct size, and improved ventricular function. 13 Sodium nitroprusside is a direct nitric oxide donor that mediates potent arteriolar vasodilation, inhibits platelet adhesion and promotes antiinflammatory processes, 14 which effectively reduce MVO in animal IRI models. 15 Previous clinical trials with these agents varied in their design and they mostly lacked a sensitive method to detect MVO, which may have contributed to the multiple conflicting results seen. 7 -9,12,16,17 Uncertainty remains regarding the potential therapeutic impact of these agents. 1 We hypothesized that, in patients undergoing PPCI for STEMI, appropriate doses of intra-coronary (IC) adenosine or SNP delivered distally to the coronary bed, and again after stent deployment, was the optimal administration that would reduce infarct size and MVO as determined by the sensitive measure of in-patient CMR.

Study design
The trial design and methods have been published previously. 16 Briefly, REFLO-STEMI was a multi-centre, prospective, randomized, open-label, controlled trial, with blinded endpoint analysis, conducted in accordance with the Helsinki Declaration. Ethical approval for the study (reference 11/H0405/10) was obtained from the National Research Ethics Service (UK). ST-segment elevation myocardial infarction patients with single vessel disease and thrombolysis in myocardial infarction (TIMI) flow grades 0 -1 in the IRA were enrolled between October 2011 and April 2014 in four regional cardiac centres in the UK. All patients .18 years age, presenting within 6 h of symptom onset, with ST-segment elevation ≥2 mm in ≥2 contiguous leads and with a baseline corrected QT interval (QTc) ,450 ms on admission electrocardiogram (ECG) were eligible and assented. 16 Full eligibility criteria are shown in Supplementary material online, Table S1.

Randomization and treatment
Patients were randomized using a 24-h interactive voice recognition service with concealed allocation. All patients were pre-treated with aspirin 300 mg and prasugrel 60 mg or ticagrelor 180 mg 1 as well as bivalirudin 0.75 mg/kg bolus plus infusion of 1.75 mg/kg/h until completion of PPCI or continued for 4 h if clinically indicated. Bivalirudin, recommended in all patients, was standard of care in the enrolling centres at the time of trial initiation. Allocation was 1:1:1 to adenosine, SNP, or control (standard PPCI alone) with stratification by 'symptoms to balloon either less than or ≥3 h' and 'anterior infarction or not'. Manual thrombectomy was mandated in all patients and the thrombectomy catheter then used to deliver the first drug bolus (adenosine 1 mg or SNP 250 mg) IC distal in the coronary bed, after thorough flushing of the catheter. Immediately following stent deployment, and providing a repeat QTc was ,450 ms (or ,60 ms increased over baseline value), the second dose (adenosine 1 mg if IRA was the right coronary artery otherwise 2 mg or SNP 250 mg) was delivered over 1 min via the guide catheter (see Figure 2).

Study outcomes
Cardiac magnetic resonance was performed at 24 -96 h on a 3.0T scanner ( Figure 1) to determine infarct size (primary endpoint). 16 Secondary outcomes included: MVO extent assessed on early-and late-gadolinium enhanced (EGE, LGE) images and the presence of intra-myocardial haemorrhage (IMH) on CMR; ST-segment resolution (STR) on ECG performed 90 min post-PPCI; major adverse cardiovascular events (death, MI, heart failure, target lesion revascularization, and stroke; major adverse cardiac events, MACE) within 6 months. Important adverse events and study outcome measures are defined online (see Supplementary material online, Tables S5 and S6, respectively). A clinical events committee, blinded to patient details and treatment allocation, reviewed and adjudicated key trial adverse events using original source documents. An independent DSMB, with an independent statistician, periodically reviewed trial conduct and outcome data.

Cardiac magnetic resonance analysis
Cardiac magnetic resonance was performed as previously described 16 and analysis, blinded to all patient details and treatment allocation, was undertaken offline in a central core lab (University of Leicester) using cmr42 (Circle Cardiovascular Imaging, Calgary, Canada). Earlygadolinium enhanced and LGE images were acquired after 1 -3 and 10 -15 min, respectively, following 0.15 mmol/kg gadolinium-DTPA (Magnevist, Bayer, Germany) administration using a single-shot or segmented inversion-recovery gradient-echo sequence. Early-and late-MVO (E-MVO, L-MVO) were defined as hypoenhancement within the infarct territory on EGE and LGE imaging, respectively. Left ventricular volumes, infarct size, myocardial oedema (area at risk, AAR), intramyocardial haemorrhage (IMH), and MSI were determined as described previously. 16

Sample size
To detect a reduction in infarct size from 20 to 15% of LV mass, assuming a standard deviation of 10%, 18 -22 a of 0.05 and two tailed, with 80% power, a total of 192 patients were required. Allowing for a drop-out rate of 20% between PPCI and CMR resulted in a final sample size requirement of 240 patients.

Statistical methods
Primary analysis was by intention-to-treat with a secondary perprotocol analysis in those patients who received both doses of the investigational drugs. Continuous variables, including infarct size, were investigated for normality and were log-transformed when found to be non-normally distributed. Normally distributed continuous variables were expressed as mean + standard deviation and compared using t-tests and analysis of variance (ANOVA). Non-normally distributed data were presented as median (25th -75th quartiles) and compared using non-parametric methods (Mann -Whitney or Kruskal -Wallis). Comparison between groups for categorical outcomes was undertaken Strategies to attenuate microvascular obstruction during P-PCI using x 2 tests. Potential significant confounders of infarct size (age, sex, diabetes, anterior MI, ischaemia time and collateral blood flow to the infarct territory determined by the Rentrop score 23 ) were assessed using a forward selection procedure (statistical significance level of 5%). Multivariable analysis using linear regression was used to adjust for significant confounders. Time-to-event Cox proportional hazards regression models were used to assess predictors of first MACE (both at 30 days and at the study end) and to obtain unadjusted and adjusted hazard ratios (HRs) with 95% confidence intervals (CIs) for adenosine/SNP vs. control. In addition, actuarial event rates of MACE by 1 and 6 months for each intervention were reported with 95% CIs. The assumption of proportional hazards was assessed by a global hypothesis test. 24 No formal adjustment for multiple hypothesis testing was used, but P-values were interpreted cautiously. All statistical analyses were performed using SPSS (Version 20.0. Armonk, NY: IBM Corporation) and R (Version 3.1.2. The R Foundation for Statistical Computing, Vienna). Figure 2. Two hundred and fortyseven patients were randomized with 207 (84%) undergoing CMR. Ten patients did not complete the CMR (due to claustrophobia or musculoskeletal discomfort) so that the primary outcome of infarct size was assessed in 197 patients (80% of those randomized).

Baseline characteristics
Baseline characteristics for randomized patients are presented in Table 1 (see Supplementary material online, Table S7 for patients completing CMR by treatment allocation). There were no significant differences in characteristics of those randomized and those who completed CMR.

Angiography and primary percutaneous coronary intervention procedure details
Radial access, thrombectomy, and drug-eluting stent use were uniformly high ( Table 2). Intra-procedural complications were similar across all groups. However, the incidence of transient atrioventricular (AV) block not requiring pacing was greater in the control arm. There was a low incidence of AV block requiring pacing with the study drugs (adenosine 2.4% vs. SNP 1.3% vs. 0% in control arm). The rate of transient hypotension (not requiring vasopressor or intra-aortic balloon-pump) was almost three-fold greater (16.5 vs. 5.8%, P ¼ 0.028) in the SNP group compared with control.

Drug delivery
Both doses of investigational drugs were administered in 66/82 (80%) and 53/79 (67%) patients (adenosine and SNP, respectively). The commonest reasons for withholding the second dose were QTc prolongation in 30 (adenosine: 9; SNP: 21) and hypotension in 3 (adenosine: 2; SNP: 1) patients. Failure to cross the lesion or deploy a stent and prohibiting intra-procedural complications accounted for the remainder.
Angiographic, electrocardiogram, and enzymatic assessment of myocardial injury There was no difference in the occurrence of post-PPCI TIMI flow grade ,3 between groups ( Table 2). Likewise ECG (STR.70%) assessment of microvascular tissue perfusion was similar across the groups. There was no statistically significant difference between groups in peak creatine kinase levels.
Cardiac magnetic resonance assessment of myocardial injury: primary end point There was no statistically significant difference in the primary outcome measure of unadjusted infarct size between groups ( Table 3). On multivariable regression analysis, adjusting for significant confounders, mean infarct size was increased in the adenosine group Strategies to attenuate microvascular obstruction during P-PCI (mean difference 2.73, 95% CI: 20.18 to 5.64, P ¼ 0.066) when compared with controls. This was not seen in the SNP group.
Late-MVO was present in 67% patients and was significantly more prevalent in the SNP arm vs. control (75.4 vs. 56.9%, P ¼ 0.029). However, the extent of L-MVO was small and not significantly different between any groups. Other CMR parameters of microvascular injury (E-MVO and IMH) were also similar between groups and none of the potential confounders were identified as being of statistical importance by the forward selection procedure. Diagnostic quality T2-weighted (oedema) imaging was only obtainable in 109 patients (55%). There were no significant differences in AAR or MSI between groups.
Left ventricular end-diastolic and end-systolic volumes indexed to BSA (LVEDVI, LVESVI) were significantly increased with adenosine compared with control and this was accompanied by a nonsignificant reduction in ejection fraction (EF). Left ventricular  volumes and function were similar in the SNP and standard PPCI arms ( Table 3).

Clinical outcomes
Two hundred and thirty-three (94%) patients completed follow-up (12 patients withdrew consent/refused follow-up and 2 patients were lost to follow-up). There was a significant increase in MACE at 6 months (15.6 vs. 2.5%, P ¼ 0.004) in the adenosine-treated group compared with control (see Figure 3) driven by incidence of heart failure (10.4 vs. 1.2%, P ¼ 0.016); HR 6.53 (95% CIs 1.46 -29.2), P ¼ 0.01. When adjusted for potential confounders, the observed MACE difference remained with similar hazard ratios with P-values: P ¼ 0.018 at 30 days and P ¼ 0.01 at 6 months (see Supplementary material online, Table S9). There was no statistically significant difference in bleeding between groups (see Table 4).

Per-protocol analysis
In patients who actually received both doses of adenosine there was, when compared with controls, a significant increase in infarct size (%LVM, 12.0 vs. 8 Figure S1).

Sub-group analyses
In patients with anterior STEMI, those who received adenosine had significantly greater IS %LVM (16.  Table S8). Furthermore, the composite secondary clinical endpoint of death, MI, and heart failure was significantly greater in these patients (21.2 vs. 3.1%, P ¼ 0.025) compared with controls.

Discussion
Sub-optimal microvascular perfusion in STEMI confers additional morbidity and mortality despite TIMI 3 flow in the epicardial vessel. 25 Adenosine and SNP are powerful vasodilators with pleiotropic effects, including anti-inflammatory, anti-platelet, and immunemodulatory actions, that have been shown to reduce IRI, particularly when delivered IC targeting the microvascular bed. 16

Infarct size and markers of reperfusion injury
There was no significant difference in infarct size between the groups, although there was a trend to significant increase in infarct size in the adenosine group compared with controls when the results were adjusted for confounders. There was no reduction in MSI, early-or late-MVO measured by CMR with either drug. These results are consistent with the only other randomized trial using CMR, which found that high-dose IC adenosine did not increase myocardial salvage and did not reduce either infarct size or MVO. 9 The per-protocol analysis showed that patients who received both doses of adenosine had significantly increased infarct size, LV volumes, and reduced EF on CMR compared with the control group, which suggests not only a lack of efficacy but potential cardiac toxicity and partial negation of any beneficial effects of reperfusion. This effect was particularly apparent in patients with anterior MI. Reasons for these adverse findings remain unclear but are considered below. The proportion of patients achieving STR .70%, as an ECG surrogate marker of coronary vascular flow, was relatively high and similar across groups compared with previous studies. 8 -11,17 This may reflect our recruitment of only patients who presented within 6 h rather than within 12 h of symptom onset as in some previous studies. 7 -9,17 In REOPEN-AMI, 8 STR .70% occurred in 71% of patients treated by adenosine, in 54% of patients treated by SNP, and in 51% of patients treated by saline (risk ratio: 1.39 [1.07 -1.79]; P ¼ 0.009, and risk ratio: 1.04 [0.78 -1.40]; P ¼ 0.75, adenosine or SNP vs. saline, respectively). Furthermore, in REOPEN-AMI, the enzymatic infarct size was 30% lower in the adenosine compared with the saline group but this is a greater effect size than would be expected by the attenuation of MVO alone, which makes interpretation difficult. However, in the absence of corroborative AAR or infarct size data from robust imaging such as CMR, this may imply that the adenosine group were destined to have smaller infarcts compared with control (saline), potentially over-estimating the benefit of adenosine in that study.
The finding of an adverse effect of high-dose adenosine remains difficult to explain. It could be that high-dose adenosine tends to activate other receptors, which lead to adverse events; adenosine has been speculated to be responsible for excess dyspnoea observed with ticagrelor in the PLATO study. However, any effect of ticagrelor on circulating adenosine levels is likely to be small compared with the relatively high doses of adenosine administered in our study. Furthermore, the distribution of ticagrelor and prasugrel across the treatment arms was comparable (see Table 1) and there were no significant differences in the primary endpoint or MVO for the patients receiving ticagrelor (see Supplementary material online, Table S8). Additionally, adenosine may mediate diuretic resistance; it acts on renal A 1 -AR on afferent arterioles to reduce glomerular flow and filtration, stimulate renin release, and enhance proximal tubular sodium reabsorption. 29 It could be postulated that cross-activation of renal AR promoted fluid retention, increased LV volumes and advanced heart failure pathophysiology in our adenosine-treated cohort. Even though our finding is hypothesis generating, our data strongly suggest high-dose IC adenosine may lead to adverse events (possibly through cross-activation of other receptors) and probably should not be used to prevent MVO, in the circumstances of this trial.

Limitations
The study was open-label, which may have influenced management of patients. However, the primary outcome was assessed on blinded CMR scans. The finding of an increased hazard signal with use of high-dose IC adenosine in our study must be interpreted with caution given our relatively small sample size. The power of the study may have been reduced due to the use of the FWHM technique, which has better observer variability but results in lower estimation of infarct size compared with previous studies using more conservative thresholds. Whilst we did not formally adjust for multiple hypothesis tests, which might not be considered ideal, this was a Phase II trial, and we did not want to exclude identifying potential efficacy of either pharmacological intervention. However, even if we had used a more stringent level of statistical significance, to account for testing both of our two active treatments relative to our control group, the statistical significance of our main results would remain unaltered. Similarly, although there was some evidence against the assumption of proportional hazards for the primary clinical outcome analysis (MACE by 6 months) this was not overwhelming, and our results were further supported by actuarial estimates of the absolute MACE event rate by 6 months.
It is possible that the dose of adenosine used in our study was insufficient to achieve sustained activation of A 2 -AR; 30 We utilized a slow-bolus dosing regimen and it could be that, owing to the shorthalf life of adenosine, concentrations of adenosine were insufficient in the microvascular bed to augment the benefits of reperfusion. A 1 -AR activation at lower adenosine doses may exacerbate MVO by promoting neutrophil chemotaxis. 30 As such, delivering adenosine as a continuous infusion, to ensure more prolonged availability of adenosine in the microvasculature, may be superior to boli injections in antagonizing MVO-related processes. It should be noted that one might still consider low-dose (50 -100 mg) adenosine, as is common in clinical practice, to reverse established slow-flow or no-reflow when it occurs.

Conclusions
Intra-coronary adenosine and SNP did not reduce infarct size nor MVO during PPCI for STEMI. Furthermore, high-dose adenosine appeared to be associated with adverse clinical outcomes, increased infarct size, and reduced EF compared with control. These data suggest that neither agent is effective for sub-clinical no-reflow and should not be used routinely and prophylactically in the setting of PPCI to prevent reperfusion injury.