Infarct size following complete revascularization in patients presenting with STEMI: a comparison of immediate and staged in-hospital non-infarct related artery PCI subgroups in the CvLPRIT study

The study design and main results have been published previously (3, 6). CvLPRIT CMR was a prespecified substudy of a multicenter, prospective, randomized, controlled, open- label, clinical trial with blinded CMR endpoint analysis (PROBE design) conducted in 7 UK centers between May 2011 and May 2014 (5). Inclusion and exclusion criteria were as for the main trial with absolute contraindications to CMR imaging as an additional exclusion.

After verbal assent patients were randomized after coronary angiography but before IRA PCI, to IRA-only or in-hospital complete revascularization. If there were no clinical contraindications, immediate CR was recommended but the non-IRA procedure could be staged, at the operator’s discretion, but completed during the index admission. Reasons for staging revascularization were not recorded. Recruitment is shown in Fig. 1. Ninety-eight patients in the substudy were randomised to in-hospital CR, of which 63 were performed immediately and in 30 the procedure was staged. Five patients crossed over into the IRA-only treatment arm.

Pre and post-PPCI epicardial coronary flow was assessed using Thrombolysis In Myocardial Infarction scoring (7). Collateral flow to the IRA pre-PPCI was graded using the Rentrop system (8). Quantitative Coronary Angiography (QCA) was undertaken using QAngioXA v1.0 software (Medis, Leiden, Netherlands). Myocardium at risk was angiographically quantified using the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH score) (9, 10).

The CMR methods have been described in detail previously (5). In brief, CMR was performed pre-discharge and after any staged procedure and at 9 months (follow-up CMR).

After localisers and long axis cine images, complete stacks of short axis images covering the entire left ventricle (LV) were acquired with (1) T2w-STIR to determine the area at risk, (2) cine images for LV volumes, mass and ejection fraction and (3) late gadolinium enhanced (LGE) images to determine infarct size and MVO after administration of 0.2 mmol/kg of Magnevist (Bayer, Leverkusen, Germany).

Follow-up CMR was performed at 9 months (±4 weeks) post-PPCI. The protocol for follow-up CMR was similar to the pre-discharge scan, but with T2w-STIR imaging omitted and assessment of reversible ischemia with first-pass perfusion after pharmacological stress with adenosine included.

Analysis was performed as previously described by physicians blinded to all clinical data including treatment allocation at the University of Leicester core lab (5). Briefly, infarct size was quantified on LGE imaging using the Full-Width Half-Maximum technique (11). On the pre-discharge CMR scan, ischaemic area-at-risk (oedema) was assessed using Otsu’s Automated Technique (12) and myocardial salvage index (MSI) was calculated as the percentage of the area at risk that was not infarcted on LGE (5). If infarction was seen in >1 coronary territory in the pre-discharge CMR, this was recorded as being in the IRA territory (associated oedema and/or MVO) or the non-IRA territory with the consensus of three observers (JNK, GPM, JPG). Non-IRA infarcts were additionally classified as likely to be acute or chronic (presence of wall thinning and no oedema/MVO). Infarct size was recorded for both IRA and non-IRA LGE and total infarct size was the sum of all LGE. On the follow-up CMR, perfusion images were visually assessed for defects and reversible ischaemia burden calculated as a percentage expression of the summed difference score (13).

MACE comprised a composite of all-cause mortality, recurrent MI, heart failure and ischemia-driven revascularization. Secondary endpoints included cardiovascular death and individual components of the primary endpoint. Safety endpoints comprised stroke, major bleeding and contrast-induced nephropathy. Data were collected by an independent clinical trials unit (Royal Brompton Hospital, London) and events adjudicated by blinded clinicians.

The primary outcome of the CMR substudy was infarct size (expressed as % of LV mass) on pre-discharge CMR, analysed on a log-transformed scale due to right skew. This was adjusted for known baseline predictors of infarct size (anterior MI, time to revascularization, diabetes, TIMI flow pre-PPCI) and important baseline variables that significantly differed between the two groups (TIMI flow post-PPCI, SYNTAX score, dual antiplatelet therapy choice, glycoprotein inhibitor/bivalirudin use for N-IRA PCI) using generalized mixed models. Normally distributed continuous variables were expressed as mean ± standard deviation and comparison was with student’s t-tests. Non-normally distributed data were expressed as median (25^th–75^th quartiles) and analysed using Mann-Whitney testing. Categorical variables were compared using Chi-squared testing. Clinical outcomes were assessed using time-to-first event survival analysis (log-rank test with right censoring). Kaplan-Meier curves were plotted for the period of randomization to the occurrence of the clinical outcomes and compared using log-rank test, and Cox proportional hazard models were fitted to estimate hazard ratios and 95 % confidence intervals for treatment comparisons.

Baseline characteristics and comorbidities were closely matched in the in-hospital staged and Immediate CR subgroups and were similar to those in the overall CvLPRIT study population (Table 1). Four patients in the immediate CR group versus none in the staged group had a history of non-STEMI and previous PCI.

The median time to staged non-IRA PCI was 34.2 h post-PPCI (IQR 24.8–48.9). There was increased visible thrombus, subsequent thrombectomy catheter use, a higher incidence of IRA no-reflow and reduced TIMI grade post-PPCI in staged CR patients (Table 2). There was a small but significant increase in CAD complexity in the staged group (SYNTAX score 18.3 vs. 16, p = 0.021) involving the IRA (p = 0.043). The prevalence of well collateralised IRA territory and LAD IRA were similar in both groups. The angiographically derived AAR on APPROACH score was similar in the groups. Patients with right coronary artery IRA were more likely, and those with circumflex IRA less likely, to have a staged procedure. There was less glycoprotein IIb/IIIa inhibitor and bivalirudin use during the non-IRA PCI in the staged compared to the immediate CR group. When the staged and PPCI procedures were added, there was significantly increased cumulative screening time, contrast dose, number of stents (non-IRA PCI and total number of stents) and total procedure lengths in staged versus immediate CR (Table 2).

The lower total infarct size and MVO extent, higher MSI and LV ejection fraction observed with immediate CR may be due to a number of possible factors. There could be real differences arising from treatment strategies; the staged group may have been having larger infarcts and thirdly the decision to stage the procedure, at least in some cases, may have been as a direct result of poor technical success e.g. no-reflow of the IRA. We think it is unlikely that staged patients were having larger infarcts at baseline as the time to presentation, proportion having anterior MI, degree of collateralization of the IRA and Killip Class were not significantly different from the immediate CR group and adjusting for these variables did not significantly alter the results. In addition, the ischaemic area at risk was not significantly different in the two groups. This was the case both when quantified on CMR and on the angiographically derived APPROACH score, which would negate any effect of differing CMR timing. A significant effect of ischemic preconditioning is also unlikely given the low prevalence of anti-anginal medication use in both groups (14).

Immediate CR to non-IRA’s could theoretically reduce infarct size by increasing collateral flow or by improved blood flow to the watershed region of the infarct (15). The severity of the non-IRA lesions (average stenosis diameter 73 % in both groups) also indicates that these were likely to have been flow-limiting stenoses. In support of a real effect of immediate CR is the increase in MSI compared to staged patients. However, and most importantly, differences in angiographic and PPCI results most likely explain the reductions in MSI and increased infarct size in the staged v immediate CR groups. The staged group had significantly more visible thrombus in the IRA (87 % v 50 %), subsequent thrombectomy catheter use and significantly more no-reflow (23 % v 2 %) than the immediate CR group. These factors are likely to be the main reason for the increase in infarct size, reduced salvage and decreased ejection fraction. We did not prospectively record the operators’ reasons for staging the non-IRA procedures in staged patients but we think it is likely that a suboptimal result from the PPCI and the presence of inferior rather than lateral MI influenced the decision to stage the non-IRA PCI.

A surprising finding in this study was that the frequency of non-IRA MI detected by CMR was considerably higher in the staged versus immediate CR groups. PCI related MI (type 4a) are well recognized, (16, 17) although of uncertain clinical significance. In elective PCI patients up to 29 % (18) will have significant increases in troponin and a similar proportion of patients undergoing complex PCI will have evidence of type 4a MI on CMR, even when pre-treated with clopidogrel for >24 h and a glycoprotein IIb/IIIa inhibitor periprocedurally (18). Excluding those patients with evidence of chronic infarction, acute non-IRA MI was seen in 30 % of the staged and only 11 % of the immediate CR groups. Although these type 4a MI were relatively infrequent and small (3.7 and 2.8 % of LV mass for immediate and staged patients respectively) there was considerable variation in size. Revascularization related injury accounting for 4 % of LV mass has been associated with a three-fold increase in MACE (19). Larger randomized studies are required to confirm whether staging CR results in more frequent non-IRA MI and poorer outcomes than immediate CR.

The explanation for the increase in type 4a MI seen with staged CR is likely to be related to greater number of stents implanted in the non-IRA of the staged patients and possibly the different use of adjunctive medication at the time of the non-IRA PCI. Glycoprotein IIb/IIIa inhibitor (8 %) and bivalirudin (11 %) use was low in the staged procedures compared to the immediate CR group (32 and 53 % respectively), which probably reflects clinicians concerns about bleeding with a second in-patient procedure requiring additional vascular access.

The clinical event rate in both groups was similar (immediate 6.3 % and staged 6.7 %) and lower than seen in the main trial for those randomized to CR (10 %). The lack of other significant differences between the two groups in this post-hoc analysis with small numbers mean no conclusions can be drawn. Immediate CR was associated with a shorter inpatient stay of one night compared with staged CR. This finding and the reduction in lab time with second procedures may suggest that an immediate CR is likely to be more cost effective than a staged strategy (20). However these findings could simply be related to the fact the staged patients had larger MI and although cost-effectiveness will be assessed in the entire CVLPRIT population, any differences between staged and immediate CR would have to be confirmed in randomized trials comparing these strategies. The increased frequency of major bleeds with staged CR is likely secondary to the need for two separate procedures and hence two arterial punctures. However, due to the small numbers, this should be confirmed in a larger study.

This is a post-hoc analysis and patients were not randomised to immediate or staged CR. We did not systematically record the reasons for staging the procedure or use of adjunctive medication, which is a significant limitation. The marked differences in angiographic appearances at baseline, and success following PPCI, are likely to contribute to the observed differences in infarct size between the immediate and staged CR groups. However statistical significance persisted after correction for important baseline covariates. Due to the small numbers of patients in this analysis, propensity matching was not possible. The study was not powered for clinical outcomes. Inevitably, patients who died early or who were very ill following PPCI could not participate in the CMR study which likely explains why the clinical event rates are lower than in the main study. The pre-discharge CMR was undertaken later in staged patients (day 4), which is likely to have resulted in a decrease in infarct size and MVO extent compared with scanning at day 2 (21). Hence, the observed differences in CMR outcomes in immediate and staged CR may have been even greater if both groups were scanned at the same timepoint. However, it was important that the CMR was performed after the staged non-IRA procedures to ensure that we captured associated type 4a MI in our results. Finally, as it was not routinely captured, we could not confirm whether the higher incidence of no-reflow in the staged patients was reflected in less ST-segment elevation resolution post PPCI.