Mortality in ST-Segment Elevation Myocardial Infarction With Nonobstructive Coronary Arteries and Mimickers

This cohort study evaluates mortality in ST-segment elevation myocardial infarction with nonobstructive coronary arteries.


Introduction
2][3] The majority of AMI cases, including ST-segment elevation myocardial infarction (STEMI) and non-STEMI (NSTEMI), without any culprit artery are classified as myocardial infarction with nonobstructive coronary arteries (MINOCA). 4NOCA is a syndrome composed of distinct diagnoses based on pathophysiologic mechanisms.The 2019 American Heart Association (AHA) statement defines MINOCA as coronary artery plaque disruption, epicardial coronary spasm, or coronary embolism/thrombosis and excludes takotsubo cardiomyopathy, myocarditis, and nonischemic cardiomyopathy due to their failure to meet the fourth universal definition of MI and classifies them as MINOCA mimickers. 5,6Our current understanding of MINOCA is limited for the following reasons: (1) the term MINOCA is used interchangeably with AMI without a culprit artery; however, not all patients without a culprit artery have nonobstructive coronary arteries; (2) inclusion of cases without troponin elevation; (3)   discordance between the AHA 2019 and the European Society of Cardiology (ESC) 2016 statement, which includes MINOCA mimickers in the definition of MINOCA syndrome 6,7 ; (4) lack of transparency in published studies on the MINOCA definition and diagnostic criteria used; (5) limited use of cardiac magnetic resonance imaging (CMRI) which is the reference standard diagnostic test for evaluating patients presenting with MINOCA to determine the underlying diagnosis; and (6) limited studies to date of patients with MINOCA presenting with STEMI.
MINOCA has historically been considered a low-risk AMI phenotype.5][16] These disparate findings are primarily due to the limitations of current MINOCA studies.Despite worse prognosis in patients with MINOCA presenting with STEMI compared with NSTEMI, 17,18 knowledge on the outcomes in patients with STEMI presenting with MINOCA as compared with obstructive disease is limited to 2 studies that we know of. 19,20Both included patients with and without elevated troponin in the MINOCA group and failed to report whether MINOCA mimickers were included in the MINOCA cohort.Therefore, we examined a large prospective multicenter US cohort of consecutive STEMI activations to evaluate the prevalence, characteristics, and 5-year mortality risk of patients with MINOCA and MINOCA mimickers in comparison with obstructive disease.

Study Population
The Midwest STEMI Consortium is composed of 4 high-volume, tertiary, regional STEMI centers, including Minneapolis Heart Institute in Minneapolis, Minnesota (site 1); The Christ Hospital in Cincinnati, Ohio (site 2); Prairie Heart Institute in Springfield, Illinois (site 3); and Iowa Heart Center in Des Moines, Iowa (site 4). 21Data from site 4 were not included in the index study, given the inability to provide 5-year mortality data.All consecutive patients beginning March 2003 who had a STEMI activation within 24 hours of symptom onset were prospectively included in the database.These regional tertiary centers use similar standardized STEMI protocols and serve as the referral centers for over 100 nonpercutaneous coronary intervention (PCI) hospitals with a spectrum of urban and rural communities.The design of the Midwest STEMI consortium has been described in detail previously. 21The study protocol, data sharing agreements, and relevant information have been approved by institutional review boards (IRB) in each study center.Informed consent or waiver of consent was obtained as required by each institution's IRB.

Data Collection and Follow-Up
At the time of the STEMI event, trained research assistants collected all relevant demographic clinical data (including self-reported race), angiography, laboratory markers performed by the hospital laboratory, and discharge medications by reviewing electronic medical records.Race was included because it is an important social determinants of health that is associated with cardiovascular outcomes.The standardized protocols included serial cardiac troponin I and T assays.A positive troponin was defined as at least 1 troponin measurement above the 99th percentile for the assayspecific reference limit, which varied over time and by site.When available, the following CMRI data were collected: evidence of infarction (late gadolinium enhancement), regional wall motion abnormality, and final impression by reader.
Major adverse cardiovascular events (MACE), defined as cardiac arrest, myocardial infarction, or physician-adjudicated cardiac-related death were collected and analyzed at 1 year at sites 1 and 2.
Time to MACE was defined as time to the first adverse event in the MACE definition.All-cause mortality was collected in-hospital and at 1 and 5-year follow-up at all sites from the electronic medical records and the national death index.The end of follow-up for the mortality outcome was February 2023.Only the initial STEMI event was included in the analysis and all subsequent events were excluded.There were 335 patients without follow-up data (either due to in-hospital mortality or loss to follow-up) who were censored at day 0.

Statistical Analyses
Categorical variables are presented as numbers and relative frequencies (percentages computed out of all nonmissing data) and continuous variables as mean (SD) or median (IQR).Outcomes were compared between STEMI presenting with obstructive disease as compared with MINOCA and MINOCA mimickers.Also, outcomes between MINOCA and MINOCA mimickers were compared.
Categorical variables were compared using the Fisher exact test or χ 2 test and continuous variables using 2-sided Wilcoxon rank-sum tests or independent sample 2-sided t tests.
The primary end point of the study was 5-year mortality risk.Survival and MACE-free survival from the time of STEMI presentation to 1 and 5 years were estimated using the Kaplan-Meier method and compared between STEMI presenting with obstructive disease as compared with MINOCA and MINOCA mimickers using the log-rank test.In sensitivity analysis, the Kaplan-Meier method was used In a secondary analysis, a propensity score-matched sample was created using nearestneighbor matching to assess 5-year mortality in patients with STEMI with MINOCA and MINOCA mimickers as compared with obstructive disease.Propensity scores were estimated using logistic regression on the following variables: age, sex, hypertension, diabetes, dyslipidemia, PCI history, study site, and year of admission.Matching variables were selected out of known baseline risk factors according to current literature and availability in the data; therefore, 8310 of the 8560 patients (97.1%) with complete data were included in the matching procedure.Matching was performed without replacement according to the logit of the propensity score with a 5:1 ratio of patients with MINOCA to patients with obstructive disease due to the baseline imbalance in the population sizes to assign relatively greater weight to the MINOCA cohort.To allow for imperfect matching while also enforcing similarity between matched patients, a caliper width of 0.2 standardized differences was used.One patient with MINOCA was unmatched, resulting in a matched sample of 2404 patients.
Matching was conducted using the MatchIt R package, version 4.5.2(R Project for Statistical Computing). 26To assess the resulting covariate balance in the matched data set, a love plot was constructed (eFigure 2 in Supplement 1).This indicated that all matched variables were well balanced after matching (since all standardized mean differences were reduced to below 0.1).Hazard ratios were estimated from the matched sample via a Cox regression model adjusting for the same variables as the unmatched model (eTable 1 in Supplement 1).The interaction term between the groups (obstructive disease or MINOCA) and PCI history was found insignificant and removed from the model.Age was modeled with a natural cubic spline with knots at the 25th, 50th, and 75th percentiles due to evidence of nonlinearity.Since this model violated the proportional hazards assumption, it was stratified by sex, and site was included in the model as a random effect.These models were then built on unmatched and matched data to assess the model's sensitivity to modeling specifications.This procedure was repeated to estimate 5-year differences in mortality percentage among STEMI presenting with obstructive disease as compared with MINOCA mimickers.
Signficance was set at a P value of less than .05.

Results
The study population included 8560 patients with STEMI, of whom 8151 (95.2%) presented with

Outcomes in STEMI Presenting With MINOCA as Compared With MINOCA Mimickers
Compared with STEMI presenting with MINOCA, MINOCA mimicker cases were less frequently discharged receiving aspirin, P2Y12 inhibitors, and statins, but no significant difference was observed in β-blockers and ACEIs or ARBs (eTable 2 in Supplement 1).In-hospital, 1-year, and 5-year mortality risks and 1-year MACE events were similar in STEMI presenting with MINOCA as compared with MINOCA mimickers.In multivariable adjusted analysis, the 5-year mortality risk in STEMI presenting with MINOCA was similar to MINOCA mimickers (HR, 1.65; 95% CI, 0.68-4.01).

CMRI Findings
The frequency of CMRI use in STEMI cases presenting with MINOCA was 30 for site 1, 1 for site 2, and 0 for site 3.Among STEMI cases presenting with MINOCA, the number of MINOCA cases with diagnoses were significantly higher at 76% (34 of 45 patients) at site 1, where CMR was routinely used, compared with site 2 at 42% (8 of 19 patients) and site 3 at 57% (32 of 56 patients; P for trend = .03).At site 1, CMRI was performed in 67% of MINOCA cases (30 of 45 patients).Sensitivity analysis evaluating 5-year mortality among the CMRI-diagnosed MINOCA cases was similar to the overall cohort.Mortality at 5 years in STEMI cases presenting with obstructive disease was similar to MINOCA (χ 2 1 = 1.4;P = .24).At site 1, CMRI was performed in 49% of MINOCA mimicker cases (83 of 169 patients).By contrast, only 3 CMRI were performed among MINOCA mimicker cases at sites 2 and 3.Among the 83 patients with MINOCA mimickers who underwent CMRI at site 1, 37 had takotsubo cardiomyopathy, 42 had myocarditis, and 3 had nonischemic cardiomyopathy.

Discussion
In this large multicenter prospective cohort of 8560 consecutive patients with STEMI, several observations advance our understanding of MINOCA and MINOCA mimickers: (1) MINOCA was present in 1.4% and MINOCA mimickers in 3.8% of consecutive patients presenting with STEMI; (2)   MINOCA and MINOCA mimickers represent conditions as lethal as patients with obstructive disease in short-term and long-term follow-up; (3) 5-year mortality hazard risk was higher in MINOCA than obstructive disease; (4) MINOCA and MINOCA mimickers were less likely to receive cardiac medications at hospital discharge compared with obstructive disease; and (5) CMRI was useful to establish an underlying diagnosis among patients with MINOCA and MINOCA mimickers.
In our cohort of consecutive patients with STEMI, 120 patients (1.4%) presented with MINOCA and 289 (3.8%) with MINOCA mimicker diagnoses.These findings are consistent with Gue et al 27 who reported 110 cases (4.4%) using the ESC definition, which combined MINOCA and MINOCA mimickers in a smaller cohort of consecutive STEMI cases in the United Kingdom.In their study, the 1-year mortality was 4.5%, which was lower than our 1-year mortality of 12% in MINOCA and 8% in MINOCA mimicker cases. 27We also observed an 18% mortality at 5-year follow-up in both MINOCA and MINOCA mimicker cases presenting with STEMI.By contrast, Andersson et al 20 reported an 8% mortality risk in patients who presented with MINOCA in a single-center study in Denmark among 4793 consecutive patients with STEMI during a 2.6-year follow-up.The higher rates observed in our study may be due in part to the longer follow-up time, the overall higher death rates in this US midwest cohort, and the inclusion of normal cardiac biomarkers in the MINOCA cohort in the Denmark study.
Only 2 prior studies 19,20 that we know of have evaluated outcomes in STEMI patients presenting with MINOCA as compared with obstructive disease with conflicting findings.Importantly, neither of these studies reported whether MINOCA mimickers were included in the MINOCA group, and both studies included patients with normal cardiac biomarkers.The HORIZONS-AMI trial 19 evaluated the safety and efficacy of bivalirudin, and in a post hoc analysis of 3602 patients with STEMI, including 127 (3.5%) with no obstructive disease, at 3-year follow-up, MACE risk was lower in patients without obstructive disease as compared with patients with obstructive disease (mortality risk was not   19 The lower MACE risk in the HORIZONS-AMI post hoc analysis was likely associated with the lower threshold used to define obstructive coronary artery disease (>30% stenosis) and the lack of elevated ischemia biomarkers in nearly 50% of patients.Furthermore, in the Denmark study by Andersson et al, 20 the subgroup of patients with normal coronary arteries and elevated troponin had a 2.65-fold higher risk of death than patients with obstructive disease.Similarly, in our study, we found a 1.93-fold higher risk of 5-year mortality in patients with MINOCA as compared with obstructive disease and also found a similar risk in MINOCA mimickers as compared with obstructive disease.
MINOCA poses a dilemma for clinicians given the multitude of diagnoses that underlie this enigmatic syndrome.Our results suggest that there are remarkable differences in STEMI cases presenting with obstructive disease as compared with MINOCA and MINOCA mimickers.The differences in risk profile, treatment, and outcomes in MINOCA and MINOCA mimickers support the differentiation made by the recent AHA guidelines between coronary (MINOCA) and noncoronary (MINOCA mimicker) causes of MI without a culprit artery. 5In this study, the use of CMRI was infrequent in STEMI cases presenting with MINOCA.We demonstrate that at site 1, where there was a standardized protocol to perform CMRI for MINOCA cases, CMRI substantially improved the ability to identify the underlying diagnosis in 76% of MINOCA cases.These results are consistent with and extend the findings of the HARP study, 28 which demonstrated the utility of multimodality imaging to diagnose the underlying diagnosis of MINOCA events in a cohort predominantly composed of patients with NSTEMI.Coronary intravascular imaging is another important tool in determining the underlying diagnosis of MINOCA, 28 but it was not used at any of the study sites.Our findings support the need for standardized algorithms in the diagnosis of STEMI presenting with MINOCA that includes the use of CMRI in all cases. 29,30tably, we documented that patients with MINOCA and MINOCA mimickers presenting with STEMI were less likely to be prescribed cardiac medications upon discharge compared with obstructive disease, similar to the treatment disparities reported in MINOCA cohorts presenting with NSTEMI. 31,32Patients with MINOCA likely receive suboptimal treatment for several reasons: (1) the popular misconception that MINOCA cases are often false positive troponins; (2) lack of recognition that a MINOCA event is associated with poor outcomes; (3) diversity in the range of underlying diagnoses that cause MINOCA (plaque disruption, coronary vasospasm, and embolism); and (4) limited knowledge regarding effective treatments for MINOCA.There is growing evidence that the nonuse of statins and renin-angiotensin-aldosterone system (RAAS) inhibitors is associated with a 2-fold higher mortality risk in patients with MINOCA. 33,34Yet our study demonstrates that work remains to be done, given the significantly lower use of RAAS inhibitors and statins in STEMI presenting with MINOCA.Therefore, we suggest including MINOCA in STEMI quality metrics.

Limitations
This study has limitations common to observational studies.Angiographic images were not evaluated by a core laboratory; however, each case of STEMI without a culprit on angiography was reviewed by the principal investigators at each site, including all available angiograms.Follow-up was performed with the use of NDI and electronic medical records at all the sites, which was a limitation as this can result in biased survival analysis.The lack of a standardized protocol at all sites and resulting underuse of CMRI and nonuse of intravascular imaging greatly limited our ability to further determine the underlying diagnosis of MINOCA cases.However, the low utilization of multimodality imaging in our study is consistent with clinical experience in the US.Furthermore, granular data on cause of death were not available.Because of these limitations, this study should be viewed as hypothesisgenerating to stimulate the need for additional investigations.

Conclusions
In contrast to the general prevailing sentiment, STEMI due to MINOCA and MINOCA mimickers are lethal as obstructive disease.We demonstrate that the 5-year mortality risk is 1.93-fold higher in MINOCA as compared with obstructive disease.The current underutilization of CMRI and intravascular imaging restricts our ability to explore the underlying diagnoses of MINOCA and MINOCA mimickers, which could play a crucial role in guiding future management and prognosis.

JAMA Network Open | Cardiology
The study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies.22Thestudy flow diagram is shown in eFigure 1 in Supplement 1.We identified a total of 10 624 consecutive STEMI activations and excluded 2064 cases.Each STEMI without a culprit artery case was retrospectively reviewed by the principal investigator at each site, including history of presentation, coronary angiogram, left ventriculogram, and echocardiogram images, and CMRI when JAMA Network Open.2023;6(11):e2343402. doi:10.1001/jamanetworkopen.2023.43402(Reprinted) November 16, 2023 2/13 Downloaded from jamanetwork.comby guest on 11/26/2023

JAMA Network Open | Cardiology
JAMA Network Open.2023;6(11):e2343402. doi:10.1001/jamanetworkopen.2023.43402(Reprinted) November 16, 2023 3/13 Downloaded from jamanetwork.comby guest on 11/26/2023 to estimate 5-year mortality in CMRI-confirmed MINOCA cases as compared with obstructive disease using the log-rank test.Additionally, Cox proportional hazard models were built to estimate relative hazard rates of 5-year mortality among patients with MINOCA and MINOCA mimickers (relative to obstructive disease patients), while adjusting for the following predefined clinical indicators: age, sex, hypertension, diabetes, dyslipidemia, previous PCI, history of smoking, body mass index, left ventricle ejection fraction (LVEF), cardiogenic shock pre-PCI, year of presentation, and study site.A Cox proportional hazard model was also built to estimate relative hazard rates of 5-year mortality among patients with MINOCA as compared with MINOCA mimickers.

JAMA Network Open | Cardiology
Compared with STEMI with obstructive disease, patients with MINOCA were younger and more often female (Table1).The frequency of cardiovascular risk factors was similar between STEMI presenting with obstructive disease and MINOCA.STEMI presenting with MINOCA had lower incidences of cardiogenic shock at presentation, along with higher LVEF and lower peak troponin levels than obstructive disease.

Table 1 .
Demographics, Clinical, and Angiographic Characteristics of Overall ST-Segment Elevation Myocardial Infarction (STEMI) Cohort and Obstructive Disease, STEMI With Nonobstructive Coronaries (MINOCA), and MINOCA Mimickers a Percentages were computed out of all nonmissing data.b Other race included American Indian, Alaska Native, and Pacific Islander.c Body mass index is calculated as weight in kilograms divided by height in meters squared.

Table 2 .
Outcomes of ST-Segment Elevation Myocardial Infarction (STEMI) Presenting With Obstructive Disease Compared With STEMI Presenting With Nonobstructive Coronaries (MINOCA) a Percentages were computed out of all nonmissing data.b Log-rank P.