We conducted a longitudinal retrospective study by enrolling patients who had no inducible ischemia on adenosine or dobutamine stress CMR and who had known CAD with moderate (50%–75%) to severe (>75%) stenosis assessed by ICA or coronary computed tomography angiography (CTA). Patients with moderate stenosis were classified as group 1, and those with severe stenosis were classified as group 2. These patients underwent stress CMR examination 180 days before or after ICA at the Department of Cardiology, Angiology and Pneumology of the Heidelberg University Hospital between February 2015 and December 2016. Figure 1 presents a flowchart of how the patients are selected. ICA or CTA was performed to detect suspected obstructive CAD and MI. Obstructive CAD was defined as the visual diagnosis of >50% stenosis in the coronary arteries on ICA.

The reasons for performing dobutamine stress CMR were as follows: contraindications for adenosine stress CMR, such as renal insufficiency defined by an estimated glomerular filtration rate of <30 ml/min/1.73 m², severe bronchoconstrictive or bronchospastic lung disease (e.g., asthma) and adenosine hypersensitivity, history of coronary artery bypass grafting (CABG) surgery, and chronic total or subtotal coronary occlusion (11).

A subgroup analysis was performed for patients with moderate and severe stenoses in the proximal segments of the coronary arteries on ICA or CTA. The proximal segments consisted of the proximal right coronary artery (P-RCA), left main (LM) coronary artery, proximal and middle left anterior descending artery (P-LAD and M-LAD), and proximal left circumflex artery (P-LCX). SYNTAX I score was calculated to determine the complexity of CAD using an online tool (https://syntaxscore.org) (12).

The follow-up consisted of a clinical visit as part of usual care or by direct contact with the patient, and the follow-up time was at least 3 years. Data collection ended in October 2022. The primary combined endpoint was defined as the occurrence of a major adverse cardiovascular event (MACE), such as cardiac death, non-fatal MI, percutaneous coronary intervention (PCI), or CABG. Furthermore, a secondary combined endpoint that consisted of cardiac death and non-fatal MI was analyzed.

Cardiac death was defined as sudden cardiac death with documented fatal arrhythmias or death preceded by acute MI or acute or decompensated chronic heart failure. According to the fourth universal definition of acute MI, non-fatal MI was defined as typical angina lasting for more than 20 min, a rise in troponin above the 99th percentile, and ECG changes (13). In cases where several events occurred, the first event was counted, except for one patient who presented with non-ST-elevation MI (non-STEMI) who was treated with CABG 6 days later; here, CABG was counted.

The patients underwent ICA or CTA either for suspected obstructive CAD or for acute MI, such as ST-elevation MI (STEMI) and non-STEMI. Those with acute MI received revascularization therapy and were referred for stress CMR because of the remaining moderate-to-severe stenosis in other coronary artery segments. The examinations were visually analyzed by two experienced readers. The coronary artery segments were defined according to the current guidelines issued by the European Society of Cardiology (1).

Statistical analysis was performed with MedCalc Version 20.114 (MedCalc Software, Ostend, Belgium). The Shapiro–Wilk test was used to determine the normal distribution of continuous variables, which were described as mean ± standard deviation or median and interquartile range (IQR), as appropriate. Variables with normal distribution were tested for statistical significance using the t-test. Variables without normal distribution were analyzed using the Mann–Whitney test and were presented as median and IQR. Categorical variables were compared using the chi-squared test. Kaplan–Meier curves were used for the visual representation of survival estimation as a function of follow-up time, and p-values were determined by performing log-rank testing as previously described (14). A p-value of <0.05 was considered statistically significant.

We included 169 patients in the study (24% females), with a median age of 69 (IQR 60–75) years. A total of 85 (50.3%) patients had moderate stenosis (group 1), and 84 (49.7%) patients had severe stenosis (group 2). The baseline characteristics are presented in Table 1. Most patients underwent stress CMR after ICA. Twenty-four (14.2%) patients underwent ICA for acute MI, of which nine (5.3%) were for STEMI and 15 (8.9%) were for non-STEMI. The median time between ICA and subsequent stress CMR was 7 (2–97) days. Five (3.0%) patients underwent stress CMR before ICA. These five patients were referred for ICA after negative stress CMR due to persistent symptoms of angina. Another 19 (11.2%) patients underwent coronary CTA 87 ± 55 days before undergoing stress CMR. Approximately one-third in each coronary stenosis group underwent adenosine stress CMR [group 1, n = 29 (34.1%), vs. group 2, n = 24 (29.5%), respectively; p = 0.55].

The study population showed a significant number of patients with cardiovascular risk factors, predominantly hypertension and hypercholesterolemia. No significant difference was found in medication during follow-up between patients with moderate stenosis and those with severe stenosis, specifically not concerning treatment with antiplatelet therapy and cholesterol-lowering drugs (see Supplementary Material Table S1). Patients being treated for STEMI at initial presentation had residual severe stenosis significantly more often than residual moderate stenosis [n = 8 (9.5%) vs. n = 1 (1.2%) respectively; p = 0.02] (Table 1).

No significant difference in LVEF, LVEDV, LVESV, and LV mass was reported between the two groups (Table 1). The median LVEF was preserved with 57.3% (IQR 51.3%–62.3%) and 59.9% (IQR 56.0%–63.5%) in group 1 and group 2, respectively. A total of 52 patients (30.7%) demonstrated subendocardial late gadolinium enhancement (LGE) on CMR.

Among patients in group 2, severe stenoses were identified in proximal segments in 20 (23.8%) patients and in distal segments in 64 (76.2%) patients. Thirty-nine (23.1%) patients showed a diameter stenosis of >90%. SYNTAX I scores differed significantly between patients with moderate stenoses (nine, IQR 5–14.25) and patients with severe stenoses (IQR 14–20) (Table 1).

After a median follow-up of 5.5 (IQR 4.1–6.6) years, the primary combined endpoint occurred in 16 (18.8%) patients with moderate stenosis and 23 (27.4%) patients with severe stenosis (HR 1.36, CI 0.73–2.56, p = 0.33), showing no significant differences between the two groups (Figure 2). MACE was primarily accountable to PCI, with 13 (15.3%) patients with moderate stenosis and 15 (17.9%) patients with severe stenosis receiving elective PCI (HR 1.09, CI 0.52–2.30, p = 0.81), due to unstable angina (Table 2). Acute MI occurred in three (3.5%) patients with moderate stenosis and eight patients with severe stenosis (9.5%) (HR 2.36, CI 0.72–7.70, p = 0.16). Excluding revascularization by PCI or CABG from the primary combined endpoint, we found no significant difference in the outcomes between the two groups (HR 2.34, CI 0.72–7.66, p = 0.16).

All-cause death occurred in two (2.4%) patients with moderate stenosis and in eight (9.5%) patients with severe stenosis (HR 3.18, CI 0.92–11.0, p = 0.07). No cardiac death was recorded in patients with moderate stenosis, compared with one cardiac death (1.2%) in patients with severe stenosis. Arrhythmic events such as atrioventricular block or pulmonary vein isolation were rare during follow-up (Supplementary Material Table S3). These events occurred after a mean time of 3.0 ± 2.2 years in patients with moderate stenosis and after a mean time of 3.7 ± 2.0 in patients with severe stenosis (p = 0.35). The Kaplan–Meier curve for the primary combined endpoint started to separate after approximately 270 days. In patients with severe coronary stenosis, the probability of an event-free period within the first 9 months was found at 95%. Similarly, within the first 12 months after stress CMR, the probability of an event-free period was at 94%.

In a subgroup analysis of patients with severe lesions, no significant difference was noted in outcomes between patients with proximal stenosis and those with distal stenosis (HR 0.65, CI 0.23–1.83, p = 0.42) (Figure 3). In addition, a subgroup analysis was performed on patients with a stenosis of >90% compared with those with a stenosis of <90%. The outcome was similar between these two groups (HR 1.12, CI 0.56–2.25, p = 0.75). There was also no significant difference in the outcomes between patients with subendocardial LGE and those without LGE (HR 0.90, CI 0.46–1.79, p = 0.77). The outcome did not differ significantly between patients in whom stress CMR was performed with adenosine and in those in whom stress CMR was performed with dobutamine (HR 0.74, CI 0.38–1.43, p = 0.37). The extent of symptoms during follow-up did not differ significantly between the two groups. Specifically, 12 patients with moderate stenoses and 11 patients with severe stenoses suffered from angina (19.0% and 15.3%, respectively, with a p-value of 0.56), while 15 patients with moderate stenoses and 12 patients with severe stenoses had dyspnea (23.8% and 16.7%, respectively, with a p-value of 0.30) (Supplementary Material Table S2).

Patients who reached the endpoint of cardiac death, MI, or CABG were followed up individually (Table 3). Twelve patients reached one of these endpoints. The mean time between CMR and endpoint was 3.6 ± 2.1 years. In five patients (patients 2, 3, 6, 8, and 11) (41.6%) who presented with MI, the evaluation of their coronary artery segments in stress CMR was causal for the event. In one (patient 6) out of these five patients, the initial lesion was severe, and in four patients (patients 2, 3, 8, and 11), the initially evaluated lesion was moderate. One patient (patient 10) who presented with non-STEMI 3 years after negative stress CMR to evaluate moderate-to-severe LCX stenosis showed a relevant progression of CAD with severe stenoses in proximal LAD and LCX, warranting urgent CABG. Two patients (patients 1 and 4) who presented with MI that was not caused by coronary artery lesions were initially evaluated by stress CMR. Patient 9 revealed no significant progression of CAD on ICA. Thromboembolism was suspected as the cause of non-STEMI because the patient on his own had recently discontinued oral anticoagulation that had been initiated for deep vein thrombosis without an apparent cause 3 years earlier and residual thrombosis in follow-up exams afterward.

Two patients with non-STEMI were not evaluated by ICA. The first patient (patient 12) presented with cardiogenic shock and suffered from cardiac arrest due to asystole before ICA could be performed. In the second patient (patient 5), palliative care was decided due to old age and progressive respiratory failure caused by pneumonia.

We acknowledge several limitations in our study. First, this study might have been underpowered due to the limited number of cases. Unfortunately, due to the absence of complete follow-up data in both groups, the size of the study cohort remained relatively small. Coupled with the retrospective design of our study, these factors allowed us only to hypothesize rather than conclude that patients with myocardial ischemia demonstrated a low rate of cardiac events even with known severe coronary artery stenoses on ICA. Furthermore, the severity of stenosis was assessed visually on coronary angiography performed by two experienced cardiologists, without using an intravascular ultrasound or optical coherence tomography. It is a well-known fact that there is a significant inter- and intraobserver variability in the visual assessment of the severity of coronary artery stenosis on ICA (24, 25). In addition, the hemodynamic significance of the stenosis was solely assessed non-invasively and not with fractional flow reserve measurement. Moreover, our cohort study on the subject is smaller than that of other studies. We included patients who underwent dobutamine and adenosine stress CMR, which are two different agents and vary in sensitivity and specificity in detecting functionally significant CAD. In future studies, a single stress agent is recommended. In addition, previous studies demonstrated that a quantitative perfusion analysis of ischemia performed better than a visual assessment of adenosine perfusion images in terms of the prognostic value in patients with suspected CAD (26, 27). Quantitative perfusion measurement was not performed because of the retrospective nature of this study. It may also improve the prognosis in patients with known severe coronary artery stenoses.

As indicated by the comparable incidence of clinical symptoms, patients with moderate coronary artery stenoses are affected by coronary microvascular dysfunction (CMD). Using a CMR perfusion measurement while both at rest and during adenosine stress, myocardial perfusion reserve index (MPRI) can be measured for evaluating CMD (28, 29). Previously, MPRI was found to be an independent prognostic marker for MACE among patients who experienced chest pain (28). However, perfusion measurement while at rest is not standard practice in stress CMR during regular clinical scans. Therefore, MPRI could not be calculated. Additional studies are needed to investigate the impact of CMD on symptom relief in patients with concurrent CAD.

Furthermore, the main aim of our study was to assess the prognostic value of negative stress CMR. Hence, a control group of matched patients with positive stress CMR was beyond the scope of our research. Moreover, the follow-up process may be biased by self-reporting bias as it involves direct contact with patients via telephone. Finally, it is important to note that our study design was retrospective, resulting in limitations in terms of data collection and potential biases associated with retrospective analyses.