This is a single centre, prospective diagnostic accuracy study to assess the comparative sensitivity of dobutamine versus adenosine for detection of severe non-infarct epicardial coronary artery stenosis in subjects with reduced LV EF ( EFed referred for clinical coronary angiography for investigation of symptoms or to establish the cause heart failure. Study participants will be identified from hospital angiography referral waiting lists, or already known with CAD and from heart failure outpatient clinics at Glenfield General Hospital. Clinical and CMR data will be collated on-site (at Glenfield General Hospital) from medical records stored and stress cardiac MRI scans at Glenfield General Hospital. The analysis will occur over a 12-month period following study commencement.

Aim The primary aim of this study is to assess the comparative sensitivity of adenosine and dobutamine for detection of severe CAD in patients with HFrEF. Secondary aims are to evaluate how the addition of stress perfusion may alter management compared to CMR without stress.

Primary hypothesis:

In patients with HFrEF (EF≤40%), dobutamine perfusion has superior sensitivity to adenosine for the detection of non-infarct severe CAD (>70% epicardial coronary artery stenosis).

Secondary hypotheses The addition of stress perfusion imaging to CMR cine and LGE imaging does not significantly increase the accuracy for determining whether the aetiology of HFrEF is due to CAD or not.

The response to adenosine (haemodynamics/splenic T1) will be inversely related to markers of LV dysfunction/injury (high sensitivity troponin I/B-type natriuretic peptide) and directly to EF.

Reduced myocardial perfusion reserve (<2.0) increases the accuracy of adenosine perfusion to detect CAD.

The biphasic wall motion response to dobutamine is not specific for the detection of severe coronary artery stenosis.

Experimental details and design of proposed investigation Design: prospective cross-sectional study. Subjects: participants will be recruited from the Cardiology Department at Glenfield General Hospital, Leicester, one of the largest secondary and tertiary care units in the UK. Patients referred for invasive coronary angiography to determine the cause of HFrEF or symptomatic angina and EF≤40% will be recruited. Patients will be contacted directly at hospital appointments or by letter after reviewing of case records to ensure eligibility.

Inclusion criteria: 1. HFrEF (LV EF≤40%, consistent with recent guidelines for the diagnosis of HFrEF)[38] referred for investigation of CAD; 2. Age ≥18 years. Exclusion criteria: 1. Absolute contraindication to MRI. 2. Absolute contraindication to dobutamine OR adenosine; 2. Stage III-V renal disease (estimated glomerular filtration rate <30ml/min/1.73m2; 3. Previous coronary artery bypass grafting; 4. Unstable angina or MI within 6 weeks.

Investigations:

1. History taking and examination to confirm signs and symptoms of heart failure and obtain a current medical history.
2. ECG to determine rhythm, ST-segment and conduction abnormalities. Venepuncture (20mL) for full blood count, biochemistry, high-sensitivity troponin I and B- type natriuretic peptide.
3. CMR scanning performed at Glenfield Hospital on a dedicated NIHR 3T research scanner. The investigators have extensive experience of stress CMR for the assessment of CAD, including dobutamine stress. Each CMR study will be supervised by the fellow after appropriate training including safe administration of dobutamine. Patients will be asked to withhold beta-blocker therapy for 48 hours to minimise antagonism to dobutamine. For this study, The investigators will use newly installed software with dual-sequence perfusion that allows fully automated inline quantification of myocardial blood flow utilising advanced image reconstruction algorithms, Gadgetron, in conjunction with our collaborator (Dr Peter Kellman, NIH, Bethesda, USA). This sequence allows both visual analyses for perfusion defects and quantification of myocardial blood flow and perfusion reserve on a pixel by pixel basis. A modified protocol will be undertaken, incorporating:

Cine functional assessment to determine ventricular volumes and function; Adenosine stress perfusion, dobutamine stress wall motion and perfusion, and rest perfusion imaging to assess qualitative perfusion defects, inducible wall motion abnormalities and myocardial perfusion reserve. Dobutamine will be reversed with intravenous metoprolol before undertaking LGE and then rest perfusion.

Delayed contrast enhancement for assessment of LV fibrosis and evidence of previous myocardial infarction. The investigators did consider undertaking adenosine stress CMR and dobutamine stress CMR scans on separate days, to mitigate the effects of contrast remaining affecting perfusion quantification during dobutamine stress testing. However, this would double the expense of CMR scanning in the study and reduce the acceptability to potential participants. The investigators are confident that performing both adenosine and dobutamine stress in one CMR study will not have any material impact on image assessment, as there will be 20 minutes between the two stress sequences.

All patients will have invasive coronary angiography performed as to clinical standards to assess for the presence of CAD, which will be performed within 4 weeks of CMR scanning. Fractional flow reserve will be performed as clinically indicated and given that this may be equally influenced by adenosine down-regulation is not mandated in the protocol.

Image analysis: All imaging will be analysed blinded to patient details. CMR analysis will be undertaken by using dedicated software for research (cmr42, Circle Cardiovascular Imaging, Calgary, Canada). CMR studies will be divided into three series: i) cine and LGE images, ii) cine, adenosine stress and rest perfusion and LGE images, and iii) cine, dobutamine stress and rest and LGE images, and allocated a computer-generated random number identifier prior to analysis. Each of the 16 segments of the left ventricle will be scored for function, infarction, perfusion defects and contractile response as previously described. Qualitative analysis will be performed by the consensus of two readers (the fellow and Dr Arnold) and in cases of disagreement or uncertainty with a third observer (Prof McCann). Splenic switch-off will be assessed qualitatively and quantitatively with T1 mapping for evaluation of the adequacy of vasodilator stress with adenosine. The random number key will not be accessible to staff involved in the analysis until all analyses have been completed and the database has been locked. Image analysis of series I, ii and iii will be performed separately, at least two weeks' apart. Invasive angiography images will be anonymised and reported by a senior clinician or the fellow (after appropriate training and under supervision).

Aetiology of heart failure: will be decided by a "gold" group consisting of an interventional cardiologist, heart failure specialist and cardiac imager (Dr Adlam, Prof Squire and McCann, respectively) who will be presented with all investigations, as per previous studies.

Data handling: To reduce costs, the NIHR Leicester BRC staff have developed electronic databases using the validated and open source REDCap system and CiviCRM for management of recruitment of subjects into clinical studies, research, governance and monitoring. This infrastructure will be available during this programme of study. All imaging data will be entered into REDCap in a blinded fashion and will only be unblinded for statistical analysis when the database is locked. This will ensure the integrity of the data and any changes may be audited.

Statistical analysis: Data will be examined for normality and summarised. Non-normally distributed data will be transformed to achieve normality prior to analysis. CMR image series will be compared by paired t-tests, Mann-Whitney and Chi-squared tests as appropriate. The sensitivities, specificities and accuracy of adenosine stress and dobutamine stress imaging for identifying non-infarct related ischemia will be compared. The kappa statistic will be used as a measure of agreement in diagnoses between series. Correlation coefficients will be assessed for adenosine response v markers of LV dysfunction.

Power calculations The power calculation has been performed with the assistance of Dr C Nelson, Lecturer in Medical Statistics. From our pilot data, the sensitivity of adenosine for detecting non-infarct related myocardial perfusion abnormalities as our baseline standard is 62%. The investigators estimate that the sensitivity of dobutamine for detecting perfusion abnormalities is 82%, as shown by Mordi et al. in patients with left bundle branch block at a maximum dobutamine dose of only 20μg/kg/min. Assuming the prevalence of CAD is 50% and dobutamine has sensitivity 20% higher than adenosine (Delta=0.82-0.62), 86 patients are required for 80% power at the 5% alpha level. To allow for technical issues with failure of either of adenosine, dobutamine or LGE or for patients who may decide not to proceed with invasive coronary angiography, investigators will recruit 100 patients. If all 100 patients completed the assessments investigators would have 80% power to detect an 18% increase in sensitivity with dobutamine.

Bias and confounders The investigators will aim to recruit consecutive patients to avoid selection bias. Ascertainment bias will be minimised by robust blinding of CMR scans as outlined above. As the dobutamine and adenosine stress images will be acquired in the same patients, there is no confounding in selection between the two perfusion arms. The investigators are confident that the accuracy of the dobutamine perfusion will not be compromised by prior administration of contrast for adenosine perfusion due the relatively small dose used (0.05mmol/kg at 3T) and with the long time interval (~20mins) between contrast injections, allowing clearance.

Timeline The investigators perform over 1000 clinical stress examinations and over 1200 invasive coronary angiograms annually, many in patients meeting the inclusion criteria. The investigators aim to recruit a minimum of one patient per week and the study will be completed in under two years.

The expected value of results This study will be the first to compare the diagnostic accuracy of adenosine versus dobutamine stress CMR for the detection of CAD in patients with HFrEF. If the results confirm that adenosine has poor sensitivity for the detection of CAD and that dobutamine is superior then the study will challenge the current paradigm of combined adenosine perfusion and CMR viability assessment. However before clinical practice is changed the investigators would plan to conduct a multicentre trial through BSCMR research group, which has a strong track record in multicentre clinical trials.

Inclusion Criteria:

- HFrEF (LV EF≤40%, based on either echocardiography, CMR or nuclear perfusion imaging) consistent with recent guidelines for diagnosis of HFrEF [38] referred for investigation of CAD.

2. Age ≥18 years ≤90 years old 3. Understand written and verbal English. 4. Patients are not involved in other research studies within the last year.

Exclusion Criteria:

• Absolute contraindication to MRI. 2. Absolute contraindication to dobutamine OR adenosine; 3. Stage III-V renal disease (estimated glomerular filtration rate <30ml/min/1.73m2.

  4. Previous coronary artery bypass grafting. 5. Unstable angina or MI within 6 weeks. 6. Persistent atrial fibrillation 7. Participants who are involved in current research or have recently been involved in any research prior to recruitment 8. Participants who don't speak or understand verbal or written English

yES