The CMR database was searched for all patients with (i) a dobutamine stress CMR and (ii) chronic kidney failure as defined by a GFR of <15 ml/min/1.73 m² or patients on permanent dialysis for at least 3 months according to the current guidelines (19). In all patients, the hospital information system was searched for complementary information on cardiovascular comorbidities, cardiovascular risk factors, and etiology of kidney failure. As per standard, information with regard to dobutamine dosage, stress CMR results, and vital parameters before, during, and after the exam as well as periprocedural complications are documented for all patients during their CMR. Periprocedural occurrences of symptoms were categorized as symptoms that are mild or moderate (angina, shortness of breath, or palpitations), severe symptoms with a need for medication (beta blockers intravenously or glyceryl trinitrate sublingually), and symptoms with need to abort the exam. Major complications were defined as death, life-threatening arrhythmia, myocardial infarction, or other life-threatening event with potential permanent impact on the health of the patients. The patients were followed up for up to 1 year after the exam, and the outcome was categorized into the following: no event, major adverse cardiac event (MACE) including myocardial infarction or cardiovascular death, major non-cardiac event (e.g., stroke, non-cardiac death), and incomplete follow-up (<12 months had passed since the exam or lost to follow-up). To assess the prognostic impact of stress CMR, a true negative stress exam was assumed if no MACE, no coronary revascularization therapy (stent, bypass), and no invasive coronary angiography with an evidence of a >70% stenosis occurred during the 12 months of follow-up. Similarly, if any of the before-mentioned events did occur, a true positive stress exam was assumed.

The examinations were performed on a 1.5 or 3 T MRI scanner (Ingenia and IngeniaCX, Philips Healthcare, Best, Netherlands). For the stress test, cine images of three long axis (two-, three-, four-chamber view) and of three short axis (basal, midventricular, apical) views were obtained using a steady-state free precession with 35 phases per cardiac cycle during a breath-hold. The maximum heart rate for each patient was calculated using the formula HR_max= 220 – patient age. The target heart rate was defined as ≥85% of the maximum heart rate. The dobutamine dosage was chosen according to weight starting at 10 µg/kg/min and incrementally increased by 10 µg/kg/min every 3 min up to a maximum of 40 µg/kg/min until the target heart rate was reached. If the heart rate response was insufficient (<85% of the maximum heart rate), additional medication of up to 2 mg atropine was used intravenously in the absence of contraindications. The vectorcardiogram, oxygen saturation, and blood pressure of all patients were monitored at any moment, and a cardiologist was always present. The patients were able to communicate with the technician or doctor during the exam via intercom and were regularly asked for the occurrence of symptoms. Stress testing was aborted in the incidence of severe, non-tolerable symptoms such as chest pain or dyspnea, a decrease in systolic blood pressure of >40 mmHg, hypertension of >220/120 mmHg, severe arrhythmias, or an evidence of a positive stress result. A positive stress result was defined as new or worsening wall motion abnormality in ≥1 segment according to literature (20).

This retrospective study was approved by the Ethical Committee of the University of Heidelberg (S-151/2019). Due to the retrospective nature of the study, the ethical committee did not require an additional informed consent.

Analyses were carried out using the R language and environment for statistical computing (version 4.2.1) with the user interface R Studio (version 2022.07.0/548) (21). Normal distribution was assessed using the Shapiro–Wilk test. Parametric variables are given as mean ± standard deviation (SD), and non-parametric variables are given as median with interquartile range. For the comparison of normally distributed variables between two groups, the Welch two sample t-test was used. Non-parametrically distributed variables were tested for differences using the Wilcoxon rank-sum test. To analyze the survival probability, a log-rank test (single predictors) and a cox proportional hazard model (multiple predictors: target heart rate achieved, stress result, ejection fraction) were employed, and a Kaplan–Meier estimator was calculated. The a priori significance level was set to p < 0.05.

A total of 18,057 patients were screened, and 176 patients met the inclusion criteria. The study cohort consisted of 127 men and 49 women with a mean age of 60.9 ± 14.7 years. Of those patients, 156 were on permanent dialysis. The underlying pathology for end-stage chronic kidney disease was diabetic nephropathy in 39, hypertensive nephropathy in 17, glomerulonephritis in 50, polycystic kidney disease in 20, tubulointerstitial kidney disease in 13, and other or unknown cause in 37 patients. In 138 patients (78.4%), an invasive coronary angiography was available for direct comparison, which was performed at a median of 123 days (21; 472) before or after CMR. This showed a one-vessel coronary artery disease in 23 patients (15 left anterior descending, five left circumflex, and three right coronary artery), a two-vessel disease in 23 patients, a three-vessel disease in 78 patients, and no coronary artery disease in 14 patients. An involvement of the left main coronary artery was present in 52 patients. The detailed patient characteristics are given in Table 1.

An occurrence of new onset of symptoms during stress CMR was frequent (22.1%). Those were tolerable in 5.1% of the patients, needed medication in 9.1% or led to the abortion of the stress exam in 7.9%. The reasons for the abortion of the stress exam were severe angina pectoris in six patients, a hypertensive crisis in four patients, severe shortness of breath in two patients, severe palpitations in two patients, and malaise in two patients. All symptoms diminished minutes after the exam. Major complications were rare with one case of a periprocedural non-ST elevation myocardial infarction (NSTEMI, 0.6%). Stress CMR was successfully completed in 161 patients (91.5%), of which 28 patients (17.4%) had a positive stress result and 133 patients (82.6%) had a negative stress result. A positive stress CMR was highly predictive for the presence of MACE, coronary revascularization, or an evidence of severe coronary artery stenosis (25 true positive patients, positive predictive value of 92.6%). Similarly, a negative stress CMR was highly predictive for the absence of MACE, coronary revascularization, or an evidence of severe coronary artery stenosis for 1 year after the exam (negative predictive value of 92.2%). A coronary revascularization during the 1-year follow-up was necessary in eight of the 133 patients with negative stress CMRs (four NSTEMI, one STEMI, three with high symptom burden and decision for percutaneous coronary intervention despite a negative stress CMR). The predictive value of a negative stress CMR further increased if only the patients with achieved target heart rate were included (a negative predictive value of 95.0%). The testing accuracy of dobutamine stress CMR was excellent if the target heart rate was achieved (sensitivity of 83.3%, specificity of 97.9%), and it was good if the target heart rate was not achieved (sensitivity of 71.4%, specificity of 98.4%). The CMR characteristics, dosages of stress medications, and vital parameters before and during peak stress are given in Table 2. Figure 1 illustrates the different results and stress test accuracy, and Figure 2 shows the exemplary result of a positive stress exam.

A 1-year follow-up was obtainable for 171 out of 176 patients (97.2%). Only one patient was lost to follow-up due to relocation to another country. For the remaining four patients, >9 months but <12 months had passed since the exam and the analysis of this study, which is the reason for their exclusion in the outcome evaluation. However, no events had occurred so far in these patients, who all had negative stress CMRs.

The clinical event rate was high in the study group as 16.4% of patients experienced a major cardiac (9.4%) or non-cardiac (7.0%) event. A non-fatal NSTEMI was the main contributor to the cardiac events (4.7%). The 1-year mortality rate was 10.6% (18 patients) with cardiac events accounting for 4.1% (seven patients) of events. A cumulative incidence of adverse events is shown in Table 3.

No differences in the 1-year outcome (MACE and non-cardiac death) were found between patients with aborted vs. completed stress CMR (p = 0.51). Similarly, there was no difference in the patient outcome after a positive vs. negative stress result (p = 0.61, see Figure 3B). However, the therapeutic regimen differed significantly as the positive stress tests were followed by an invasive revascularization in 13 out of 25 patients. Apart from the stress CMR results, the requests of the patient for further investigations, technical feasibility, symptoms, and expected benefit on the quality of life were among the factors influencing the shared decision making of further treatment. The patients with an invasive revascularization had a significantly better outcome than the patients not undergoing an invasive revascularization despite the positive stress results (p = 0.02).

Regarding the multivariate analysis, a left ventricular ejection fraction (LV-EF) of ≤35% at rest was an independent predictor for a poor outcome (non-cardiac and cardiac events) in both the univariate (p = 0.037) and multivariate (p = 0.047) analysis (see Figure 3A). Equally, the failure to achieve the target heart rate in non-aborted exams was an independent predictor for a poor outcome (see Figure 3C, univariate p = 0.029, multivariate p = 0.047), and the occurrence of MACE was lowest in the patients with a negative stress result and reached target heart rate (4.9% vs. 16.1%, p = 0.03).