Background
Despite prompt reperfusion of acute ST-elevation myocardial infarction (STEMI) by primary percutaneous coronary intervention (PPCI), adverse left ventricular (LV) remodeling still occurs in a significant proportion of patients [
1], and its presence predisposes to heart failure [
2] and worse clinical outcomes [
3]. In contrast, some reperfused STEMI patients develop reverse LV remodeling, which portends to good clinical outcomes [
4].
Cardiovascular magnetic resonance (CMR) is considered the gold-standard imaging modality for quantifying myocardial infarct (MI) size [
5], and measuring LV volumes and LV ejection fraction (LVEF) [
5,
6], given its high reproducibility [
5,
7]. As a result, CMR is increasingly being used to assess surrogate clinical end-points following STEMI in cardioprotection studies. Adverse LV remodeling following STEMI has been conventionally defined as ≥ 20% increase in LV end-diastolic volume (LVEDV) from baseline. This cut-off value was determined using echocardiography, and was based on the upper limit of the 95% confidence interval of intra-observer variability for the percentage change (%Δ) in LVEDV following STEMI [
8,
9]. Reverse LV remodeling has been defined as ≥10% decrease in LV end-systolic volume (LVESV) by echocardiography following cardiac resynchronization therapy, and was derived using receiver operator characteristic (ROC) curves for the optimal cut-off for the %ΔLVESV to predict mortality [
10]. So far, no cut-off values for adverse and reverse LV remodeling following STEMI have been defined by CMR, and studies using CMR to assess post-STEMI LV remodeling have relied upon using these cut-off values defined by echocardiography for adverse [
11,
12] and reverse LV remodeling [
13].
Therefore, the first aim of this study was to perform intra-observer and inter-observer measurements of LV parameters in paired acute and follow-up CMR scans in reperfused STEMI patients, in order to determine the minimal detectable changes (MDCs) that could be used as cut-off values for defining post-STEMI remodeling. Secondly, we aimed to identify the cut-off values for clinically important %ΔLVEDV and %ΔLVESV to predict LVEF <50% at follow-up [
14], as a surrogate for poor clinical outcome [
15]. Finally, cut-off values for %ΔLVEDV and %ΔLVESV were then applied to a large cohort of STEMI patients with paired acute and follow-up scans to assess different patterns of post-STEMI LV remodeling.
Discussion
The main findings of this study are as follows: (1) The MDC95 in %ΔLVEDV and %ΔLVESV of 12% was higher than the corresponding cut-off values for predicting LVEF < 50% at follow-up (11% for %ΔLVEDV, and 5% for %ΔLVESV), providing cut-off values for assessing adverse and reverse LV remodeling following STEMI by CMR; (2) The MDC95 for %ΔLVM and %ΔLVEF from the acute to follow-up CMR scan were 12% and 13%, respectively, providing cut-off values for assessing changes in these LV parameters following STEMI by CMR; (3) By assessing the combined %ΔLVEDV and %ΔLVESV between the acute and follow-up CMR, we observed 4 different patterns of LV remodeling following STEMI.
In this study, we measured both intra-observer and inter-observer variability, and as expected, the MDC95s for all these LV parameters were greater for inter-observer than intra-observer measurements. Our analyses on the whole cohort mainly focused on the inter-observer rather than the intra-observer measurements because different operators analyzed the scans from each study. We found that the inter-observer MDC95s for %ΔLVEDV and %ΔLVESV between the acute and the follow-up CMR were 12% each. Using these cut-off values for defining LV remodeling following STEMI, a combination of an increase in LVEDV (≥12%) and in LVESV (≥12%) could be used to identify adverse LV remodeling, whereas a decrease in LVESV (≥12%) with or without a decrease in LVEDV (≥12%) could be used to identify reverse LV remodeling. However, further studies are required to investigate the prognostic implications of these proposed cut-off values for defining adverse and reverse LV remodeling following STEMI.
As expected, the cut-off value of 12% or more for %ΔLVEDV to define adverse LV remodeling obtained in our cohort is significantly lower than that defined by echocardiography (20% or more for %ΔLVEDV). This is due to the better spatial resolution of CMR and superior intra-observer and inter-observer variability [
6]. On the other hand, the cut-off value for defining reverse LV remodeling as ≥12% for %ΔLVESV from our study is higher than the 10% cut-off value currently proposed by echocardiography [
10]. The echocardiography-based method was derived using ROC curve for the optimal cut-off for decrease in ESV to predict mortality in patients undergoing cardiac resynchronization therapy and they did not perform inter-observer and intra-observer variability for change in LVESV. It is highly likely that the inter-observer and intra-observer for %ΔLVESV by echocardiography in STEMI patients would be higher than the CMR cut-off value we obtained.
Currently there is no consensus on whether T&P should be included as part of the LV volume or as part of LV mass during LVEF and LVM assessment by CMR [
5]. We therefore provided MDC95 for %ΔLVEDV, %ΔLVESV, %ΔLVM, and %ΔLVEF using both approaches. It is already known that the T&P can significantly affect LV volumes, LV mass, and LVEF [
23,
24]. We found that LVEDV and LVESV were higher, and LVM and LVEF were lower when T&P were included as part of the LV volume, and this is consistent with previous reports [
23‐
25]. As their inclusion as part of the LV mass is not always practical depending on the software, both methods are currently considered acceptable [
5]. Although the LV parameters differed depending on how the T&P were dealt with, there were no difference in the CoVs both for inter or intra-observer measurements for LVEDV, LVESV, LVM and LVEF when T&P were included as part of the LV volume or LV mass. However, the MDC95 for intra-observer and inter-observer measurements for %Δ in LV parameters varied by 1–2% depending on whether the T&P were included as part of the LV volume or LV mass. We therefore provided the highest MDC95 for each LV parameter in Table
4, irrespective how the T&P were dealt with.
In the absence of clinical outcomes, LVEF <50% in patients with scars have previously been shown to be associated with poor clinical outcomes [
15]. Using this cut-off for LVEF at follow-up as a surrogate marker, we obtained cut-off values for %ΔLVEDV and %ΔLVESV of 11 and 5%, respectively. These figures were lower than that defined by our MDC95 cut-off values of 12% for both %ΔLVEDV and %ΔLVESV. The MDC and the clinically significant change are independent of each other as they are derived in different ways and in our case, the former turned out to be larger than the latter. Therefore we chose the cut-off values of MDC95 to define LV remodeling in the whole cohort.
Using the combination of %ΔLVEDV and %ΔLVESV from the acute to the follow-up CMR, we observed 4 different patterns of post-STEMI remodeling (Figs.
5 and
6) The actual impact of these 4 different patterns of post-STEMI LV modeling on clinical outcome will need to be determined in future studies. Conventionally, adverse LV remodeling post-STEMI has been defined by %ΔLVEDV. Our data, suggests that assessing both %ΔLVEDV and %ΔLVESV, may provide further insights into different patterns of LV remodeling following STEMI, thereby allowing one to customize heart failure therapy to prevent adverse LV remodeling or promote reverse LV remodeling. Orn et al. [
26] described three patterns of LV remodeling based on presence and persistence of MVO by CMR within the first week of an acute STEMI in a serial CMR study of 42 patients. Most LV remodeling occurred by 2 months and continued to at least 1 year. Those with no MVO had a normal pattern of wound healing; those with MVO on day 2 only, they dilated their ventricle but adapted functionally; and the last group were those with persistent MVO at 1 week and they dilated their ventricle without the ability to adapt functionally. These three groups bear some resemblance to the groups of LV remodeling we identified but we did not have serial CMR data on MVO for comparison. Other factors that determine the pattern of LV remodeling post-STEMI also require further study.
Westman et al. [
1] recently showed that there was an imperfect link between MI size and adverse LV remodeling (defined as >10 ml/m
2 increase in indexed LVEDV). Several studies have also shown that MVO was a strong predictor of adverse LV remodeling [
27]. Using the definition in our study for adverse LV remodeling, we also showed that there was an imperfect link between acute MI size and adverse LV remodeling as well between MVO and adverse LV remodeling. Some patients with large MI size and MVO developed reverse LV remodeling and some patients with small MI size and no MVO developed adverse LV remodeling. As eluded by Westman et al. [
1], the development of adverse LV remodeling is complex and multi-factorial, and more work is warranted in this field.
We found the MDC95 in %ΔLVM between acute and follow-up scans to be ≥12%, suggesting that this would be the minimal change in LVM that is unlikely due to inter-observer measurement errors. However, the interpretation of changes in LVM following STEMI is complicated by the fact that on the acute scan, the presence of myocardial edema also contributes to the changes in LVM acutely and therefore we did not investigate %ΔLVM in post-STEMI LV remodeling. However, it would be interesting to determine the MDC95 for assessing %ΔLVM in patients with LV hypertrophy related to hypertension or aortic valve disease, in order to provide cut-off values which can be used in studies assessing the regression of LV hypertrophy.
Finally, we found the MDC95 for % ΔLVEF to be ≥13% in STEMI patients when using CMR. This finding suggests that only a relative change in LVEF of 13% or more can be reliably detected by CMR as being beyond inter-observer measurement errors. This is equivalent to an absolute change of 6.5% in a patient with an acute LVEF of 50%. This needs to be taken into consideration when planning future studies designed to investigate new treatments for improving LVEF following STEMI.
Limitations
Inter-observer and intra-observer measurements were performed in only 40 patients (80 scans) but this is significantly larger than the number of patients used in a previous study (
n = 10) providing the minimal detectable change in LVEF by echocardiography in patients undergoing chemotherapy (10 patients with echocardiography at 2 time-points) [
22]. We only used one analysis tool and LV parameters were quantified using the semi-automated method. We did not have matching echocardiography data for comparison. We did not have complete data on the presence of multi-vessel disease or clinical outcomes and our sample size was relatively small. Therefore, we used an LVEF of <50% at follow-up as a surrogate. [
15] There was heterogeneity in the performance of CMR for acute MI size and MVO (scanner strength, dosage and type of contrast, timing of LGE for MVO and MI, quantification technique used – Additional file
1: Online appendix Table 1) and our findings need to be confirmed by future studies.
Acknowledgements
We express our gratitude to the staff and patients at the UCLH Heart Hospital.