The randomised Complete versus Lesion-only PRimary percutaneous coronary Intervention Trial: Cardiovascular Magnetic Resonance imaging substudy (CvLPRIT-CMR)

Current guidelines

The guidelines from the European Society of Cardiology (ESC)3 and American Heart Association (AHA)4 recommend percutaneous coronary intervention (PCI) of the IRA only at the PPCI unless the patient is in cardiogenic shock, where complete revascularisation (CR) is permitted for critical (> 90% stenosis) or unstable lesions (ESC/AHA: class IIa, level of evidence B). The two alternative options at the PPCI are CR during the index admission (of all significant lesions, including in non-IRAs) and planned outpatient PCI of the non-IRA at a later date (usually < 6 weeks) as a staged procedure. 12 Where outpatient revascularisation to non-IRAs is being considered, this should be preceded by non-invasive ischaemia assessment [e.g. myocardial perfusion scintigraphy (MPS), stress cardiovascular magnetic resonance imaging (MRI)] (ESC: class I, level of evidence A; AHA: class IIa, level of evidence B). 3,4 Fractional flow reserve (FFR) assessment of non-IRAs at the PPCI is not currently undertaken as it is felt that potential microvascular dysfunction in non-IRA territories could render FFR inaccurate. However, one study has investigated this and showed that FFR at the PPCI was unchanged when reassessed 35 days later (FFR 0.77 at both time points), suggesting that non-IRA lesion severity may be accurately measured by FFR in acute STEMI. 13 The resulting lack of consensus regarding management of MVD at the PPCI was reflected by the US Cardiovascular Data Registry, demonstrating wide variation in practice, with 0–38% of MVD patients undergoing CR at the PPCI on registry studies. 14,15

Evidence base for revascularisation strategies for multivessel disease at primary percutaneous coronary infarction

At the time of this grant application (July 2010) the evidence base for the management of MVD at the PPCI was weak, largely based on retrospective analyses and registries. 3,4 There were only a small number of randomised clinical trials, with clinical outcomes that were of limited quality, lacking statistical power, and with varying design and outcomes (Table 2).

By far the best-quality study to date is the Preventative Angioplasty In Myocardial Infarction (PRAMI) trial, the results of which were published in August 2013,19 at which time the Complete versus Lesion-only PRimary percutaneous coronary Intervention Trial (CvLPRIT) was in follow-up. This trial demonstrated an extremely large reduction [hazard ratio (HR) 0.35, 95% confidence interval (CI) 0.21% to 0.58%; p < 0.001] in major adverse cardiovascular events [MACEs; death from cardiac causes, non-fatal myocardial infarction (MI) or refractory angina], which was driven by all components of the primary end point. 19 However, the trial could be criticised for randomising patients after the PPCI, potentially introducing selection bias, and an excess of anterior MI in the IRA-only arm and the inclusion of the ‘soft’ end point of refractory angina in the primary outcome. The benefits of a CR strategy remained in doubt, particularly with concern over the safety of such a strategy undertaken at same sitting as the PPCI.

Potential benefits

A CR strategy at the time of the PPCI in STEMI patients with MVD may:

1. Limit infarct size (IS) and increase the amount of salvaged myocardium by increasing collateral flow to the at-risk, but non-necrotic, peri-infarct zone. There are no specific data (either observational or from clinical trials) available to confirm such a benefit.

2. Reduce overall hospital stay and total cost of care.

3. Reduce ischaemic burden,20 which appears to be an important determinant of outcome following MI, at least in the era before the PPCI. 21

4. May reduce further PCI at a later date, either for symptoms or silent ischaemia as per current guidelines,3,4 reducing subsequent hospitalisation for the patients and with resultant economic benefits.

5. Reduce the risk of recurrent MI/death, as has been observed for non-STEMI,22 although this finding has not been replicated in chronic stable angina. 23

6. Reduce vascular complications by having all PCI performed during the index intervention through a single-access site.

Potential risks

Potential risks of a CR strategy are detailed below.

1. IS may be increased. Approximately one-third of patients undergoing even seemingly uncomplicated elective PCI experience a rise in troponin levels consistent with the diagnosis of myocardial necrosis. 24 The risk for unstable angina patients is higher, with 53% experiencing a post-PCI troponin elevation. 24 There is debate whether or not such troponin increases are of independent prognostic significance. 25 The mechanisms of injury are likely to include necrosis of myocardial tissue adjacent to stent insertion and microembolisation to the distal vasculature. 26 IS in the 30% of patients with new infarction was 5% of LV mass (LVM) and 1.3% of LVM for the entire cohort. 26 In a mixed cohort of patients undergoing elective PCI (n = 92) or coronary artery bypass grafting (n = 60), those who experienced new myocardial injury, detectable on cardiac magnetic resonance (CMR) imaging, had a threefold increase in MACEs. 27 There is also the not insignificant, but impossible to quantify, risk of complete no-reflow in the non-IRA which could have devastating consequences in a patient undergoing the PPCI. There are no data available that tell us the frequency of ‘new’ injury in the non-IRA with multivessel PPCI, but we can anticipate this will be significantly higher than in patients undergoing elective PCI. 24 Even without myocardial necrosis, resting perfusion28 and myocardial perfusion reserve29 are reduced following PCI, potentially impairing collateral flow to the area at risk (AAR) and decreasing the amount of salvaged myocardium.

2. Contrast-induced nephropathy, as a result of the increased volume load of contrast, could be increased.

3. Stenting of bystander lesions in the non-IRAs which are neither causing ischaemia nor symptoms may lead to no symptomatic or prognostic benefit to the patient and with increased costs to the NHS.

4. There may be an increased risk of both early, especially in the thrombogenic milieu of acute infarction, and late stent thrombosis and restenosis.

5. Non-IRA revascularisation may not reduce ischaemia more effectively than by intensive medical therapy following MI. 30

Prognosis following acute myocardial infarction

Left ventricular systolic dysfunction has long been recognised as an important sequelae in survivors of MI. 2 In 605 male survivors of acute MI, LVEF, LV end-diastolic volume (LVEDV) and LV end-systolic volume (LVESV) were all predictive of mortality during an average of 78 months’ follow-up, but only LVESV was an IP on multivariate analysis. 2 The importance of reductions in ejection fraction, IS and increases in LV volumes have been confirmed in over 2300 survivors of STEMI receiving reperfusion therapy. 31 Limiting IS with reperfusion leads to improved LV function and attenuation of subsequent cardiac remodelling (LV dilatation > 20%), and is a key aim of current treatment strategies in STEMI.

Cardiac magnetic resonance imaging assessment of myocardial injury following ST-segment elevation myocardial infarction

Cardiac MRI is the gold standard technique for the quantification of LV volumes and function and late gadolinium enhancement (LGE) imaging can detect and quantify MI with unique precision. 32,33 In a dog model of experimental reperfused MI, microvascular obstruction (MVO) and IS were strongly related to early LV remodelling (r = 0.89 and r = 0.81, respectively). 34 MVO was the best IP (r² = 0.71; p < 0.001) of remodelling. 34 In patients, CMR-measured MVO correlates strongly with ST-segment resolution in patients undergoing the PPCI but relatively weakly with myocardial blush grade and not significantly with thrombolysis in MI (TIMI) flow. 35 However, CMR-measured MVO is not simply an oversensitive measure of small vessel obstruction. Larger infarcts on CMR are consistently associated with larger ventricular volumes, reduced ejection fraction and increased MVO, which occurs in 40–60% of patients treated by the PPCI. 35–37 IS and MVO have consistently been related to adverse ventricular remodelling. In all the PPCI studies in which MVO has been included in a multivariate model, MVO predicts remodelling independently of IS, ejection fraction and cardiac volumes. 35,36,38–40

Prognostic value of cardiac magnetic resonance imaging following primary percutaneous coronary intervention

Infarct size and MVO have been shown to be related to medium-term prognosis, even in relatively small studies. For example, in 122 patients with STEMI undergoing the PPCI, LVEF, LVEDV and LVESV were all associated with IS (r = –0.75, r = 0.42 and r = 0.69, respectively; all p < 0.001) and outcome. IS on CMR was the only IP of MACEs (one death, one MI and 16 heart failure admissions) at 2 years. 40 In 184 patients undergoing successful PPCI, the presence of MVO on CMR was independently predictive of MACEs (five deaths, 13 heart failure, 18 re-infarction and eight unstable angina) at 1 year. 40 Larger studies have consistently shown that IS and MVO are independently related to prognosis, even when other clinical variables, and LV volumes and ejection fraction are considered. 41–43

Myocardial salvage index

The extent of myocardial necrosis after an acute coronary occlusion is variable and dependent on a number of factors including the time to reperfusion, collateral blood flow, metabolic demand of the tissue and the total AAR, as determined by amount of tissue that is acutely hypoperfused at the time of coronary artery occlusion, probably being the most important. 44 The efficacy of reperfusion strategies can be assessed by calculating myocardial salvage index (MSI) (AAR IS/AAR), which may be an important measure of outcome. 45

Cardiac magnetic resonance imaging and myocardial salvage index

Cardiac MRI can accurately quantify MSI. During ischaemia or infarction, myocardial tissue develops oedema that can be detected as high signal intensity on T2-weighted (T2W) images, and the area of oedema is greater than the area of irreversibly damaged, necrotic myocardium. 46,47 Myocardium with high T2 signal closely correlates with the AAR, confirmed in experimental models of both reperfused48 and non-reperfused MI. 49 As expected, the size of salvaged myocardium decreases with increased IS, as measured by CMR imaging. 46 Two small clinical studies have validated the MSI with CMR imaging against single-photon emission computed tomography (SPECT). 50,51 A major advantage of CMR imaging-measured salvage index is that it can be measured during a single examination in addition to quantification of volumes, function, IS and MVO. Although black-blood T2W imaging has been prone to artefact, recent advances including increased slice thicknesses, use of coil signal intensity correction algorithms and motion correction have made the assessment of oedema much more robust. 45

Study design

The CvLPRIT was a multicentre, open, randomised controlled clinical trial comparing inpatient IRA-only and CR for the management of MVD at the PPCI for STEMI. The trial was funded by the British Heart Foundation (BHF) in 2010, as a pilot study aiming to recruit 250 patients in four centres (Leicester, Leeds, Harefield and Southampton). The embedded CMR imaging substudy (CvLPRIT-CMR) was funded by the Efficacy and Mechanism Evaluation (EME) programme, a Medical Research Council (MRC) and National Institute for Health Research (NIHR) partnership, following a fast-track application in August 2010. The design was a pragmatic, multicentre, prospective, randomised controlled, open, clinical trial with blinded end-point (CMR imaging) analysis (PROBE design). 52 It was intended to complete recruitment, follow-up and data analysis within 2 years of study initiation.

Participants

Patients presenting to the participating centres with acute STEMI and MVD being treated by the PPCI.

Participating centres and recruitment dates

Recruitment started in May 2011 and was slower than anticipated. The study was rolled out to three additional centres with support from the NIHR Comprehensive Local Research Networks. The BHF awarded a 1-year extension in 2013 to allow the recruitment target to be extended (285 patients, ensuring) and the 12-month follow-up to be completed. The NIHR EME also awarded a 9-month time extension with a small cost extension. The participating hospitals and recruitment dates are given below.

The seven centres undertaking 24/7 PPCI in this multicentre study were:

1. Glenfield Hospital (recruited May 2011–April 2013).

2. Southampton General Hospital (recruited August 2011–April 2013).

3. Leeds General Hospital (recruited September 2011–April 2013).

4. Harefield Hospital (recruited November 2011–April 2013).

5. Kettering General Hospital (recruited July 2012–April 2013).

6. Royal Derby Hospital (recruited August 2012–April 2013).

7. Royal Bournemouth Hospital (recruited February 2013–April 2013).

Inclusion criteria

1. Suspected acute STEMI: ST-segment elevation (≥ 2 mm in two or more adjacent chest leads, ≥ 1 mm in two or more adjacent limb leads, ≥ 1 mm in leads V7–V9) or left bundle branch block (LBBB) on 12-lead electrocardiography (ECG).

2. Symptom duration < 12 hours. 3,4

3. Scheduled for the PPCI.

4. Verbal assent followed by written informed consent. 53

5. MVD: defined as IRA plus ≥ 1 non-IRA with significant disease (> 70% stenosis in one plane or > 50% in two planes). The non-IRA must be a stentable epicardial coronary artery or a major branch of > 2 mm in diameter. 53

Exclusion criteria

• Age < 18 years.

• Clear indication for or against CR according to operator.

• Previous Q-wave MI.

• Previous coronary artery bypass graft (CABG).

• Cardiogenic shock.

• Ventricular septal defect or moderate/severe mitral regurgitation.

• Severe chronic kidney disease [estimated glomerular filtration rate (eGFR) < 30 ml/minute].

• Stent thrombosis.

• The only significant non-IRA lesion is a chronic total occlusion.

• Standard MRI contraindications (pacemaker, implantable cardiac defibrillator, intracranial implant incompatible with magnetic field, severe claustrophobia, weight > 200 kg).

Initial assessment and assent

The coronary care unit at each centre was alerted by paramedics of incoming STEMI patients. On arrival at hospital, the CvLPRIT research team discussed the study with patients once acute STEMI of < 12 hours’ duration was confirmed on history and ECG. Prior to the PPCI, an ethically approved, short study narrative was read to the patient (see Appendix 1). Where eligible patients provided verbal agreement (assent) to enter the randomised controlled trial, this was documented in the medical records. Assent allowed delivery of key information to patients within expected time constraints during STEMI and sufficient opportunity for patients to ask questions. Verbal information is understood and retained significantly better by patients compared with written information in acute MI trials. 54–56 The assent procedure was successfully used in the Strategic Reperfusion Early after Myocardial Infarction (STREAM)57 and the Reperfusion Facilitated by Local adjunctive therapy in STEMI (ReFLO-STEMI)58 multicentre acute STEMI studies. If patients met the inclusion criteria after angiography they were asked to give further verbal assent before randomisation.

Randomisation

Patients were randomised on-table, pre PCI, via a dedicated interactive voice recognition telephone service to either in-hospital IRA-only revascularisation or CR. Randomisation was concealed to all investigators and stratified using minimisation, by anterior or non-anterior STEMI (ECG-guided), and symptom time (time to reperfusion) less than or equal to, or greater than 3 hours, as these are strong prognostic indicators post STEMI. 59 Randomisation was run through an independent company (Sealed Envelope^TM, London, UK) and took less than 90 seconds.

Consent

Randomised patients were given patient information leaflets within 24 hours, assuming they were medically fit, and asked to provide full written informed consent to continued study participation, including the optional CMR imaging substudy. At all times, patients were informed that they were under no obligation to continue study participation.

Infarct-related artery-only revascularisation

The PPCI to the IRA only was regarded as the standard of care and was performed in accordance with the ESC3 and American College of Cardiology Foundation (ACCF)/AHA4 guidelines. Multiple angiographic views of the left and right coronary artery systems were acquired in standard radiographic projections using digital fluoroscopic angiography systems at 15 frames per second.

PeriPCI adjuncts were administered at the operator’s discretion: dual antiplatelet loading with aspirin plus clopidogrel (Plavix^®; Sanofi-aventis Ltd, UK) or prasugrel (Efient^®; Eli Lilly and Company Ltd, UK) or ticagrelor (Brilique^®; AstraZeneca, UK) for P2Y₁₂ inhibition pre angiography; heparin, bivalirudin (Angiox^®; Medicines Company, USA), glycoprotein IIb/IIIa inhibitors [e.g. abciximab (ReoPro^®; Eli Lilly & Co Ltd, UK)], thrombus aspiration devices (e.g. Export^®; Medtronic, USA), vasodilators [e.g. adenosine (Adenoscan^®; Sanofi-aventis Ltd, UK)] and isosorbide dinitrate (Isoket^®; UCB Pharma Ltd, Belgium) during the PPCI. The choice of stent and stent implantation technique were at the operator’s discretion but drug-eluting stent (DES) use was strongly encouraged.

Complete revascularisation

Complete revascularisation was the investigational intervention. It was recommended that revascularisation be completed during the index PPCI procedure (Figure 1). Where this was not possible (at the operator’s discretion), non-IRA PCI was performed during the index admission, within 36 hours of the PPCI and prior to CMR imaging.

FIGURE 1.

Multivessel coronary artery disease at the PPCI managed with CR. (a) RCA is IRA in this patient with inferior MI (two significant stenoses: *≈70% stenosis in proximal RCA and **complete occlusion); (b) angiogram of left coronary system, with two stenosis of > 70% (*) in the mid-LAD and ostium of the second diagonal branch of the LAD (D2) artery demonstrating non-IRA disease and confirming MVD (white arrows show collateral flow from septal branches of the LAD artery to the territory of the RCA); (c) no significant LCX disease; (d) RCA after the PPCI with both lesions successfully stented; (e) post stent dilatation (* in D2 and ** in mid-LAD); and (f) both lesions in LAD artery system successfully treated, confirming CR angiographic success. D2, second diagonal; LAD, left anterior descending artery; LCX, left circumflex artery; LMS, left main stem; OM1, obtuse marginal branch 1 of LCX; RCA, right coronary artery; S, septal branches of LAD.

All patients were treated with optimal medical treatment as per ESC and ACCF/AHA guidelines [dual antiplatelet therapy (DAPT), angiotensin-converting enzyme (ACE) inhibition, beta blockade, high-dose statin]. 3,4 Repeat coronary angiography was recommended only for (1) recurrent ischaemic symptoms with confirmation on non-invasive imaging (e.g. stress CMR imaging, SPECT) or (2) at the discretion of the local investigator following a positive non-invasive test at 6–8 weeks post PPCI.

Angiographic analysis

Pre- and post-PPCI epicardial coronary flow was assessed using TIMI scoring (Table 4). 61

The degree of stenosis in each significant IRA and non-IRA lesion was graded visually by local investigators on a five-point scale (1, 1–49%; 2, 50–74%; 3, 75–94%; 4, 95–99%; and 5, 100%). Additionally, after CMR analysis had been completed and the database locked, core laboratory angiographic analysis was performed by a single operator (JNK) to determine (1) collateral flow to the IRA pre PPCI (graded using the Rentrop system);62 (2) the percentage diameter stenosis of lesions by two-dimensional (2D) quantitative coronary angiography (QCA) using QAngioXA version 1.0 (Medis, Leiden, the Netherlands) (Figure 2); and (3) the complexity and extent of coronary artery disease (CAD) using the validated SYNTAX score (sum of SYNTAX scores for each lesion) by two observers (JNK and SAN). 63

FIGURE 2.

Two-dimensional QCA (patient X642).

The percentage diameter stenosis of lesions was assessed by 2D QCA using QAngioXA version 1.0. The artery diameter is quantified by comparison with the reference catheter (white arrow, here 6 French, 2.0 mm). Loci proximal and distal to the lesion under assessment are manually identified. The software then automatically contours the artery and lesion. Manual adjustment of contours can be performed if needed. In Figure 2, the distal right coronary artery lesion (blue arrow) is of 80% stenosis in a segment of 2.4 mm diameter.

Blood sampling

Twenty millilitres of venous blood was collected with the patient lying semirecumbent. For the assessment of serum creatinine, eGFR and creatine kinase (CK), blood was collected in clot activator tubes (BD Diagnostics, Oxford, UK) as summarised in Table 3. These were routine clinical bloods analysed using the Clinical Pathology Accreditation Service (CPA)-accredited (United Kingdom Accreditation Service, Middlesex, UK) laboratories at each centre, which were to ISO 15189 standard (International Organisation for Standardisation, UK).

Acute cardiac magnetic resonance imaging

The detailed protocol for the acute CMR scan is summarised in Figure 3 and explained in subsequent sections. All imaging was performed with retrospective electrocardiographic gating using dedicated cardiac receiver coils, unless atrial fibrillation or frequent ectopy was present, or for tagging images, where prospective gating was used. Parallel imaging (factor 2) was used to shorten breath-holds for all imaging, except T2W short-tau inversion recovery (T2W-STIR).

FIGURE 3.

Magnetic resonance imaging protocol for the acute CMR scan. 4/3/2C, four-, three-, two-chamber long axis; FOV, field of view; FWHM, full width, half maximum; IMH, intramyocardial haemorrhage; RV, right ventricular; SAX, short axis; TE, echo time; TR, repetition time. Reproduced under Creative Commons CC BY license from McCann GP, Khan JN, Greenwood JP, Sheraz N, Dalby M, Curzen N, et al. Complete versus lesion-only primary PCI: the randomized cardiovascular MR CuLPRIT substudy. J Am Coll Cardiol 2015;66:2713–24.

Cine imaging

After the acquisition of localising images, balanced steady-state free precession cine imaging (bSSFP) was performed in four-, two- and three-chamber long-axis views. The field-of-view was optimised to achieve in-plane spatial resolution of ≈1.1–1.7 mm × 1.3–1.9 mm. The number of segments was adjusted according to heart rate [heart rate < 50 beats per minute (b.p.m.), 17 segments; heart rate 50–70 b.p.m., 15 segments; heart rate 71–90 b.p.m.; 13 segments; heart rate > 90 b.p.m., 11 segments] at the discretion of the supervising investigator.

Intravenous contrast was administered before short-axis cine stack acquisition to minimise scans time before acquiring the LGE images (primary end point). Cine imaging was performed in contiguous short-axis slices covering the entire left ventricle and right ventricle (Figure 4). The basal short-axis slice was planned at the mitral valve annulus perpendicular to the interventricular septum to minimise partial volume at the atrioventricular boundary.

FIGURE 4.

Steady-state free precession cine imaging in a contiguous short-axis stack. (a) In end-diastole; and (b) corresponding end-systolic image. Left to right images: basal to apical, complete left and right.

Oedema (area-at-risk) imaging

The AAR was assessed using black-blood T2W-STIR imaging. T2W-STIR imaging was performed using coil signal intensity correction in four-, two- and three-chamber long-axis views and contiguous short-axis slices covering the entire left ventricle (Figure 5). Slices 10-mm thick were acquired to optimise signal-to-noise ratio. The echo train length (ETL) was adjusted with heart rate (heart rate < 50 b.p.m., ETL 40; heart rate 50–70 b.p.m., ETL 30; heart rate 71–90 b.p.m., ETL 25; heart rate > 90 b.p.m., ETL 20).

FIGURE 5.

T2-weighted short-tau inversion recovery imaging showing myocardial oedema (AAR). Myocardial oedema seen as hyperenhancement on T2W-STIR imaging in the anteroseptal segments (left anterior descending artery IRA) as indicated by *. (a) Four-chamber long-axis view; (b) two-chamber long-axis view; (c) basal LV short-axis view; (d) mid-LV short-axis view; and (e) apical LV short-axis view.

Late gadolinium enhancement imaging

Late gadolinium enhancement imaging was commenced 10 minutes after intravenous administration of 0.2 mmol/kg gadolinium diethylenetriaminepentaacetate (Gd-DTPA; Magnevist, Bayer, Germany) using a segmented inversion-recovery gradient-echo sequence with a two-beat trigger. This was preceded by a bSSFP Look–Locker inversion time (TI) scout to determine the optimal TI to null unaffected myocardium. The TI was progressively adjusted to maintain nulling of unaffected myocardium. LGE imaging was performed in four-, two- and three-chamber long-axis views and contiguous short-axis slices covering the entire left ventricle (Figure 6). T2W-STIR, cine and LGE short-axis images were acquired at identical slice positions. The number of segments was adjusted with heart rate (heart rate < 50 b.p.m., 40 segments; heart rate 50–70 b.p.m., 30 segments; heart rate 71–90 b.p.m., 25 segments; heart rate > 90 b.p.m., 20 segments).

FIGURE 6.

Late gadolinium enhancement imaging showing infarction. MI seen as hyperenhancement on LGE imaging in the anteroseptal segments as indicated by * in the oedematous territory (AAR) (see Figure 5). (a) Four-chamber long-axis view; (b) two-chamber long-axis view; (c) basal LV short-axis view, (d) mid-LV short-axis view; and (e) apical LV short-axis view.

Follow-up cardiac magnetic resonance imaging

The protocol for follow-up CMR imaging was similar to the acute scan, but with oedema (T2W-STIR) imaging omitted and assessment of reversible ischaemia included with perfusion assessment included.

Oedema (area at risk) quantification

Oedema (AAR) was quantified as hyperenhancement on T2W-STIR imaging using CMR imaging40 (Circle Cardiovascular Imaging, Calgary, AB, Canada) using Otsu’s automated threshold (OAT). 66 Endocardial and epicardial borders were manually contoured on contiguous LV short-axis slices, excluding papillary muscles, trabeculae, epicardial surfaces and blood pool artefact (Figure 7).

FIGURE 7.

Exclusion of blood pool artefact from oedema quantification. Blood pool artefact (hyperenhancement) is caused as stagnant blood in regions of severe hypokinesia receives all inversion pulses in T2W-STIR imaging, and needs to be excluded from LV myocardium during endocardial contouring. (a) T2W-STIR image showing oedema (hyperenhancement) in the anteroseptal segments (*); (b) hyperenhancement on the cavity side of the endocardial contour (red) is blood pool artefact; and (c) hyperenhancement is correctly excluded from LV myocardial contours.

Otsu’s automated threshold automatically calculates a unique signal intensity threshold for each slice by dividing the greyscale signal intensity histogram into two groups (enhanced and normal) based on the threshold giving the least intraclass variance within each group,66 without the need for a user-defined region of interest (ROI). Oedema was calculated as a percentage area for each of the 16 AHA segments67 (Figure 8). Total AAR was expressed as percentage of LVM. The most apical T2W-STIR slice was excluded to minimise partial volume.

FIGURE 8.

Oedema quantification on T2W-STIR imaging. (1) T2W-STIR image showing oedema (*) in basal anteroseptum and anterior segments; (2) endocardial (green) and epicardial contours (red) drawn; (3) OAT automatically highlights enhanced myocardium (oedema) in blue; (4) final image after exclusion of noise artefact; and (5) percentage segmental area extent of oedema.

Two manual corrections were applied to AAR measurements: (1) inclusion of hypointensity within enhancement corresponding to intramyocardial haemorrhage (IMH);68 and (2) and exclusion of small, isolated enhanced regions without interslice continuity in non-IRA territories deemed noise artefact (Figure 9).

FIGURE 9.

Manual inclusion of IMH and MVO in AAR and IS. Top row: T2W-STIR images demonstrating oedema in inferolateral segments. Region of hypoenhancement within oedema corresponding to IMH (*): (a) without oedema detection; (b) with automated oedema detection; and (c) IMH manually included in total AAR and labelled as IMH. Corresponding images demonstrating MVO on LGE imaging in the same patient: (d) no detection; (e) automated detection; and (f) manually included MVO.

Late gadolinium enhancement

Infarct was defined semi-automatically on magnitude LGE images using CMR imaging. 40 Endocardial and epicardial borders were manually contoured on contiguous short-axis LV slices, excluding papillary muscles, trabeculae and epicardial surfaces and the full-width half-maximum (FWHM) technique69 applied. Here, a 2-cm² ROI was manually drawn in the infarct core and enhancement calculated as pixels of > 50% of the automatically determined maximum signal intensity in the ROI (Figure 10). Total IS was expressed as a percentage of LVM and segmental area extent of LGE was calculated. 67 The apical LGE slice was excluded to minimise partial volume effect. Total IS was manually corrected by including hypointensity within enhancement (MVO) to total IS, and exclusion of noise artefact as per AAR quantification.

FIGURE 10.

Full-width half-maximum infarct quantification method on LGE. (1) LGE image showing infarct (*) in basal anteroseptum and anterior segments; (2) endocardial (green) and epicardial contours (red) drawn; (3) a 2-cm² ROI (pink) drawn in infarct core; (4) FWHM enhancement with signal intensity threshold > 50% of maximum in infarct core; and (5) percentage area of each myocardial segment with infarct.

Perfusion analysis

Perfusion images were visually, semiquantitatively assessed for perfusion defects (visible defect for five or more heartbeats) by the consensus of two experienced observers (JNK and GPM). Stress perfusion, rest perfusion and LGE images were coregistered to allow accurate assessment based on all available data. Three perfusion patterns were possible: (1) no perfusion defect – normal perfusion of myocardium during stress and rest; (2) reversible perfusion defect – perfusion defect seen only during stress perfusion, in viable, non-infarcted myocardium; and (3) matched perfusion defect – stress perfusion defect in infarcted myocardium (Figure 11). Perfusion defects and areas of infarction were graded as subendocardial (≤ 50% transmurality) or transmural (> 50% transmurality) and given a score of 1 or 2, respectively, per segment, whereas normal myocardium was scored 0. A modified summed difference score was calculated (maximum score 32),71 defined as the difference between the sum of segmental stress perfusion defects and LGE. 23 The summed difference score was expressed as percentage of the maximum possible to give an estimate of ischaemic burden (% LVM). 72 Examples of no perfusion defect, reversible perfusion defect and matched defect to IS are shown in Figure 11.

FIGURE 11.

The range of perfusion patterns possible. (a) Stress; (b) rest perfusion imaging, no perfusion defect (SDS = 0 for segments in image); (c) stress and (d) rest perfusion imaging, with reversible transmural defect in anterolateral and inferolateral segments (*SDS = 2 for both segments); and (e) and (f) matched subendocardial perfusion defect in infarcted basal inferior/inferolateral segments (*SDS = 0 for segment). SDS, summed difference score.

Primary outcome

The primary outcome of the CMR imaging substudy was total IS (% LVM) on acute CMR imaging (pre discharge).

Secondary cardiac magnetic resonance imaging outcomes

The following outcomes were compared in the treatment arms at both CMR scans, except for those underlined [at acute CMR scan only (pre discharge)] or in italic (at follow-up scan only):

• IS (% LVM) at 9 months

• number of discrete infarcts on CMR scan

• new MI (CMR imaging detected) at 9 months compared with acute CMR imaging

• LV volumes, LVEF and right ventricular (RV) ejection fraction

• IMH and MVO (% LVM)

• AAR (% LVM)

• baseline and final MSI

• proportion of patients with ischaemia and global ischaemic burden (% LV)

• visual presence of RV infarction, LV thrombus.

Clinical outcomes

The following clinical end points were recorded (time points in brackets) and definitions are detailed in Appendix 2:

• contrast-induced nephropathy (inpatient)

• vascular access injury requiring surgical repair (inpatient)

• all-cause mortality (all: inpatient, 6 week, 6 month, 12 month)

• MI (all)

• planned or repeat revascularisation (CABG or PCI) (all)

• heart failure admission (all)

• transient ischaemic attack/cerebrovascular event (all)

• major bleed (TIMI)73 (all).

The primary clinical outcome for the main CvLPRIT was first combined MACE at 12 months (all-cause mortality, MI, planned or repeat revascularisation, heart failure admission) and this was assessed for all patients in the CMR imaging substudy. Secondary outcomes included individual clinical end points at 12 months and inpatient events (safety analysis).

Data handling

Cardiac MRI data were recorded in a lockable, validated74 Research Electronic Data Capture version 5.0 (REDCap) database (Vanderbilt University, Nashville, TN, USA). No clinical data were released to the CMR imaging core laboratory until the database was complete, checked for errors and a locked copy provided to the Clinical Trials and Evaluation Unit (CTEU). The CMR imaging database was locked on 13 June 2014 (see Appendix 3). Data entry into the REDCap database was automated, using data transposition from automatically produced data files from CMR imaging analysis software. Complete data sets for 5% of randomly selected patients were manually checked and 100% of these data were correct compared with raw data files from CMR imaging software.

Patient and public involvement

As this grant application went through a fast-track application there was limited time to involve service users. However, the study was presented, before initiation, to the patient and public involvement (PPI) group of the NIHR Leicester Cardiovascular Biomedical Research Unit, and was welcomed. One patient with a history of MI and previous PCI volunteered to join the Trial Steering Committee (TSC) and regularly attended these meetings. The study progress was presented to the PPI group on two further occasions and the chief investigator spoke to regional PPI meetings on active CMR imaging studies and heart disease, including CvLPRIT-CMR imaging.

The plain English summary was forwarded to our patient representative, our PPI officer and the PPI representatives. No specific concerns or suggestions for improvement were raised. Once the results have been published the study will be presented at our local and regional PPI meetings to help disseminate the findings.

Protocol (version 2) dated 30 March 2011

Protocol (version 2.1) dated 15 December 2011

Protocol (version 2.2) dated 1 February 2013

Interobserver and intraobserver variability

Interobserver and intraobserver agreement were assessed on 10 randomly selected acute CMR scans using two-way mixed-effect intraclass correlation coefficients (ICCs) for absolute agreement76 and Bland–Altman analysis. 77 The ICC agreement was defined as excellent (ICC ≥ 0.75), good (ICC 0.60–0.74), fair (ICC 0.40–0.59) or poor (ICC < 0.40). 78 Intraobserver agreement was assessed using requantification by a single observer (JNK) after a 2-month interval and interobserver agreement was assessed by comparing observations of two independent observers (JNK and SN).

Assessment of clinical outcomes

Inpatient safety outcomes were compared in the two treatment arms using chi-squared testing. Results were presented as odds ratios (ORs) with 95% CIs. The 12-month clinical outcomes were compared in the two treatment arms in CMR imaging substudy patients. These analyses were performed for time to first event with survival analysis using the log-rank test (Cox regression) with right censoring. Results were presented as HRs with their 95% CIs. The Schoenfeld residuals output test was used to confirm the validity of the proportional hazards model. Kaplan–Meier survival curves were produced for the subgroups.

Sample size

There are no published CMR data comparing the revascularisation strategies. There are however numerous data on unselected patients undergoing the PPCI and CMR imaging IS,37,40,79 which are similar to that seen in our centre. A total of 100 patients in each arm had 81% power to detect a 4% absolute difference in IS, assuming a IS ≈20% of LVM, a SD of 10%, α = 0.05 and two-tailed given that either strategy may be associated with a larger IS. A new IS of 4% of LVM is associated with adverse prognosis in CAD patients with revascularisation-related injury. 27

Trial management and governance

The CvLPRIT was sponsored by University Hospitals of Leicester NHS Trust. The University of Leicester was the co-ordinating centre responsible for CMR imaging substudy management, including production of final protocols, case record forms, standard operating procedures, data handling, quality assurance and statistical reporting. Regular progress reports were provided to relevant parties. Close liaison with the Royal Brompton CTEU, that co-ordinated the main CvLPRIT, occurred throughout the study.

The main trial and CMR imaging substudy were overseen by a TSC, with the chairperson and two members being independent of the investigators. There was an independent Data and Safety Monitoring Board (DSMB) that constituted reviewed clinical outcomes during the study. An interim data review performed by the DSMB in October 2012 (16 months after recruitment started, at which point 147 patients had been recruited into the CMR imaging substudy and 36 had undergone 9-month follow-up CMR imaging) was satisfied with progress to date and for the trial to continue (see Appendix 4).

This CMR imaging substudy was funded by the MRC through the EME programmed (project number 09/150/28) and managed by the NIHR on behalf of the MRC–NIHR partnership. The main trial was funded by the BHF.

Baseline characteristics of the cardiac magnetic resonance substudy cohort

The CMR imaging substudy cohort closely represented the overall CvLPRIT group, with similar baseline characteristics, comorbidities and important prognostic predictors including symptom to the PPCI time [time to revascularisation (TTR)], infarct location and Killip class (Table 9).

Anthropometrics and demographics

Baseline characteristics and comorbidities were closely matched in the IRA and CR treatment arms of the CMR imaging substudy cohort (Table 10). There were slightly more males in the CR arm but the difference was not significant (CR 88.8% vs. IRA-only 79.0%; p = 0.06). The proportion of anterior infarcts in each arm was closely matched.

Antiplatelet and discharge medication

Discharge medications were similar in the treatment arms. All patients received DAPT. Two-thirds of patients in each arm received newer DAPT agents, prasugrel or ticagrelor (Table 11).

Angiographic markers

Radial artery access was the preferred technique in both treatment arms. Coronary disease complexity, severity and IRA at baseline angiography were similar in the groups. There was a greater proportion of CR patients with well-collateralised IRA territory, defined as Rentrop grade 2 or 3 (Table 12).

Percutaneous coronary intervention details

In the main trial, 42 out of the 139 CR patients who received the allocated treatment had a staged PCI to the non-IRA and, in the CMR imaging substudy, 30 non-IRA PCI patients were staged. Total screening time, contrast dose, procedure length and number of implanted stents were greater in CR patients. The majority of patients in both arms received DESs, although this proportion was slightly higher in CR patients (see Table 13). Symptom to balloon times (TTR), peri-PCI adjunct usage and post-PPCI CK were similar in both arms (Table 13). No reflow in the IRA patients was more common in the CR group.

Cardiac magnetic resonance image quality

Acute CMR imaging was undertaken at approximately 3 days post PPCI in both treatment arms. One hundred per cent of cine and LGE images in the final 203 CMR imaging substudy subjects were of very good quality (Table 14). Fifty-two patients’ (26%) short-tau inversion recovery (STIR) data sets were non-diagnostic [no artefact but no oedema discernible (n = 33); STIR not performed because of arrhythmia or suboptimal breath-holding (n = 14); and severe artefact (n = 5)]. Image quality was similar in both treatment arms.

Observer variability

Intra- and interobserver variability were excellent for CMR imaging volumetric and tissue characterisation (AAR and IS). Results are displayed in Table 15.

Cardiac magnetic resonance outcomes

Myocardial and microvascular injury and salvage

There was no difference in the primary CMR imaging outcome of median IS {IRA 13.5% [interquartile range (IQR) 6.2–21.9%] vs. CR 12.6% (IQR 7.2–22.6%); 95% CI –4.09% to 31.17%; p = 0.57}. Adjustment for important covariates including AAR (p = 0.347) or without AAR included (p = 0.501) did not significantly change the results. There was also no difference in IRA IS or number of transmurally infarcted segments in the treatment arms (Table 17). LGE was absent in 13 patients, of whom eight were confirmed aborted infarcts (oedema present but no LGE), with a trend towards fewer aborted infarcts (p = 0.06) with CR. There was a significantly higher prevalence of multiple infarcts and multiple acute infarcts with CR.

There was a trend towards a higher prevalence of acute non-IRA infarcts > 4% LVM with CR. Acute non-IRA infarcts correlated with non-IRA PCI territories in 15 out of 17 CR patients. There was a non-significant trend towards smaller AAR with CR, but no difference in the MSI. Microvascular and RV injury, and LV thrombus were similarly prevalent in both arms.

Cardiac magnetic resonance image quality

Follow-up CMR imaging was undertaken approximately 9.4 months post PPCI in both treatment arms and all scans were analysable. Image quality for all sequences was very good (Table 19). Three patients were unable to undertake adenosine stress perfusion because of airways disease (two in the IRA-only group and one in the CR group). Perfusion imaging was unanalysable in two patients because of severe dark-rim artefact (one in the IRA-only group and one in the CR group).

Cardiac magnetic resonance imaging outcomes

Follow-up

Median follow-up length was 372 days (IRA 378 days vs. CR 366 days; p = 0.37). One hundred and ninety-eight (98%) patients attended the 12-month clinical follow-up (three patients died before this and two patients withdrew consent for follow-up at days 7 and 220).

Safety end points

Length of inpatient stay and incidence of in-hospital clinical events were similar in the treatment arms. There were no adverse effects on safety with CR (Table 23).

The primary clinical outcome of the first combined MACE at 12 months was borderline significantly reduced in patients undergoing CR (see Table 24 and Figure 13). This was driven primarily by reduced revascularisation events (4.1% vs. 9.5%; p = 0.13) and recurrent MI (0% vs. 2.9%; p = 0.09). There was one death in each arm (both cardiovascular) and two-thirds of MIs were type 1 non-STEMI.

FIGURE 13.

Kaplan–Meier survival curves for 12-month clinical outcomes.

Clinical event rates (time to first event) for patients in the CMR imaging substudy versus those who did not take part are shown in Table 25. There was a higher event rate in non-CMR imaging patients that was largely driven by an increase in non-cardiovascular mortality and acute stent thrombosis.

Nurses

Lorraine Shipley, Kathryn Fairbrother, Gemma Turland, Emma Parker, Joanna Hughes, Victoria Meynell and Amanda Swinnerton.

Interventional cardiologists

Ian Hudson, Elved Roberts, David Adlam, Doug Skehan, Nilesh Samani, Jan Kovac, Gail Richardson, Raj Rajendra and Albert Alahmar.

Others

Jamal Khan, Sheraz Nazir, David Monk, Mini Pakal, Anna-Marie Marsh and John McAdam.

Nurse

Paula Rogers.

Interventional cardiologists

Charles Ilsley, Rebecca Lane, Piers Clifford, Tito Kabir and Robert Smith.

Other

Wala Mattar.

Nurses

Charmaine Beirnes, Amanda Chapman, Howard Fairey and Michelle Bilson.

Interventional cardiologists

Kai Hogrefe, Martin Sluka, Mohsin Farooq, Naeem Shaukat, Javed Ehtisham and Salman Nishtar.

Nurses

Kathryn Somers, Michelle Anderson, Charlotte Harland and Natalie Burton-Wood.

Interventional cardiologists

C Malkin, JM Blaxill, SB Wheatcroft and UM Sivananthan.

Nurses

Sarah Orr and Nicki Lakeman.

Nurses

Fiona Robertson, Marie Appleby and Carmen Lisbey.

Interventional cardiologists

Tariq Azeem, Julia Baron, Manoj Bhandari, Kamal Chitkara and Alastair McCance.

Others

Jacqui McCance, Anne Bebbington, Teresa Grieve and Richard Donnelly.

Nurse

Zoe Nicholas.

Interventional Cardiologists

Huon Gray, Iain Simpson, Alison Calver, Simon Corbetta and James Wilkinson.