Myocardial Infarct Size by CMR in Clinical Cardioprotection Studies

Objectives The aim of this study was to review randomized controlled trials (RCTs) using cardiac magnetic resonance (CMR) to assess myocardial infarct (MI) size in reperfused patients with ST-segment elevation myocardial infarction (STEMI). Background There is limited guidance on the use of CMR in clinical cardioprotection RCTs in patients with STEMI treated by primary percutaneous coronary intervention. Methods All RCTs in which CMR was used to quantify MI size in patients with STEMI treated with primary percutaneous coronary intervention were identified and reviewed. Results Sixty-two RCTs (10,570 patients, January 2006 to November 2016) were included. One-third did not report CMR vendor or scanner strength, the contrast agent and dose used, and the MI size quantification technique. Gadopentetate dimeglumine was most commonly used, followed by gadoterate meglumine and gadobutrol at 0.20 mmol/kg each, with late gadolinium enhancement acquired at 10 min; in most RCTs, MI size was quantified manually, followed by the 5 standard deviation threshold; dropout rates were 9% for acute CMR only and 16% for paired acute and follow-up scans. Weighted mean acute and chronic MI sizes (≤12 h, initial TIMI [Thrombolysis in Myocardial Infarction] flow grade 0 to 3) from the control arms were 21 ± 14% and 15 ± 11% of the left ventricle, respectively, and could be used for future sample-size calculations. Pre-selecting patients most likely to benefit from the cardioprotective therapy (≤6 h, initial TIMI flow grade 0 or 1) reduced sample size by one-third. Other suggested recommendations for standardizing CMR in future RCTs included gadobutrol at 0.15 mmol/kg with late gadolinium enhancement at 15 min, manual or 6-SD threshold for MI quantification, performing acute CMR at 3 to 5 days and follow-up CMR at 6 months, and adequate reporting of the acquisition and analysis of CMR. Conclusions There is significant heterogeneity in RCT design using CMR in patients with STEMI. The authors provide recommendations for standardizing the assessment of MI size using CMR in future clinical cardioprotection RCTs.

reperfusion injury to reduce MI size in patients with STEMI treated with PPCI (3).
Late gadolinium enhancement (LGE) by CMR is considered the gold standard for MI size quantification (4). MI size (5), microvascular obstruction (MVO) (6), and myocardial salvage (7) assessed by CMR performed in the first few days post-PPCI have all been shown to be strongly prognostic. As a result, CMR is increasingly being used for surrogate endpoints in RCTs (3). A recent meta-analysis of 2,632 patients from 10 RCTs found that MI size measured by CMR or singlephoton computed tomography within 1 month post-PPCI showed that for every 5% increase in MI size, there was a 20% increase in the relative hazard ratio for 1-year hospitalization for heart failure and all-cause mortality (5).
Despite CMR endpoints being quite robust and their ability to keep sample size small (3), there is limited guidance on its use. In this study, we reviewed all published RCTs in this field so far, and we provide recommendations for standardizing the use of CMR in future clinical cardioprotection RCTs.  [53%]), as shown in Figure 3.
Four RCTs (7%) used a combination of 1.5-and 3-T scanners. Ten RCTs (16%) did not specify the field  center RCTs. Figure 5 shows the distribution of the quantification techniques for MI size used in these RCTs. The other major findings were as follows: 1) acute CMR was most commonly performed at 3 to 5 days and follow-up CMR at 6 months; 2) Gd-DOTA, gadobutrol, and Gd-DTPA were most commonly used at a similar dose of 0.20 mmol/kg, with LGE acquired at 10 min;

DISTRIBUTION OF CULPRIT VESSEL AND TIMI
3) MI size was quantified manually in most RCTs, for positive RCTs, the most commonly seen effect sizes were 34% for acute MI and 18% for chronic MI ($1 month) size reduction; 6) using the control arms  (8,9,11,12) and are strongly linked to prognosis (5,7,(13)(14)(15)  There was a wide range of timings for both scans, and the most common timings were 3 to 5 days for the acute scan and 6 months for the follow-up scan.   LV remodeling (17)(18)(19)22), and prognostic information (20). Unlike single-photon emission computed tomography, CMR has superior spatial resolution (14), does not involve radiation, and requires only a single examination, when the patient is relatively stable.
Furthermore, MVO and MI size by CMR have been shown to be more prognostic compared with MI size by single-photon emission computed tomography (14).
The superior spatial resolution of CMR also allows detection of small MIs that could be missed by relying on wall motion abnormalities on echocardiography alone (23), interrogation of the peri-infarct zone (24), and hypointense core of the MVO (25), which are all prognostic (24)(25)(26). Therefore, it is not surprising that CMR endpoints have gained popularity for use in several RCTs (3).

OPTIMAL TIMING OF ACUTE CMR POST-STEMI.
Preclinical studies have shown that performing CMR too early post-reperfusion (day 1) leads to an overestimation of MI size because of a combination of edema and partial volume effect (27). In the clinical setting, acute MI size has also been shown to be dynamic and to decrease significantly in size between days 1 and 7 (28,29) but is stable between days 3 and 4 (8).
Recently, Carrick et al. (30) showed that acute MI size was stable between days 1 and 3 and subsequently reduced in size by day 10. Furthermore, late MVO has also been shown to be stable between days 1 and 3 and to reduce in size by day 10 (20), and the persistence of late MVO at 1 week following STEMI was more prognostic (31). The detection of intramyocardial hemorrhage has been shown to peak at day 3, and reduced in size and incidence by day 10, and intramyocardial hemorrhage was more prognostic than MVO (20).
There is no established method to assess the AAR If the CMR scan is performed at >5 days, the edemabased AAR may be underestimated. Therefore, acquiring the acute CMR scan at 3 to 5 days following STEMI, as performed in most cardioprotection RCTs in this review, may be the optimal time to undertake the acute CMR scan, as illustrated in Figure 6.

Manual 38%
Not specified 27% The most common myocardial infarct (MI) size quantification method was manual delineation, followed by 5-SD and full width half maximum (FWHM). However, 27% of randomized controlled trials did not specify the method used.

OPTIMAL GBCA DOSE AND TIMING OF LGE
ACQUISITION. Gadobutrol has been shown to delineate the infarcted myocardium better from the blood pool (better contrast-to-noise ratio) compared with Although FWHM has emerged as being the most reproducible (39,40), it has been shown to underestimate acute and chronic MI size (40). Some studies showed that 5-SD was promising (39-41), but others The references for the RCTs included in each row are provided in Online Table 5.
showed that it overestimated MI size (42,43). The n-SD technique requires the remote myocardium to be appropriately nulled and free of artifacts. A manual region of interest is required in the remote myocardium, and this can be a source of variability.
The Otsu technique does not require a region of interest as a reference and has been shown to accurately delineate MI size (40). But 2 subsequent studies showed that Otsu overestimated MI size (42,43).
The most common method used to quantify MI size in the RCTs was manual contouring, followed by 5-SD and FWHM. We recently showed that FWHM underestimates chronic MI size in those with MVO on the acute scan because of very high extracellular volume in the area previously occupied by MVO and should be avoided for RCTs assessing chronic MI size (42).
The 6-SD method appears the most promising and has been shown to have the highest accuracy to predict segment wall recovery in patients with chronic myocardial infarction (44) and is similar to manual quantification in patients with both acute and chronic myocardial infarction (39,42). The 6-SD approach also performed well when using 2 different LGE sequences (42).

IMPACT OF PATIENT SELECTION AND TIMING OF
CMR ON SAMPLE SIZE. In Table 1  Acute MI size should be preferred to chronic MI size as a surrogate endpoint, because acute MI size is already prognostic (5), and this would reduce sample size and result in fewer dropouts.
Acute CMR should ideally be performed on day 3, 4, or 5. When a follow-up scan is planned, 6 months would be the optimal timing for data on both chronic MI size and LV remodeling.
Manual quantification of MI size by experienced operators at a core laboratory level is considered the gold standard (4). When this is not practical, the 6-SD threshold would be an alternative option. However, any semiautomated technique is likely to be influenced by the LGE image quality, and therefore, each center may need to validate the performance of these semiautomated techniques at the respective center or core laboratory.
In the absence of a gold-standard method for the AAR, MI size should be reported as %LV, which has also been shown to be prognostic (5