Measuring Treatment Effects in Clinical Trials Using Cardiac MRI
- First Online:
- Cite this article as:
- de Waha, S., Fuernau, G., Eitel, I. et al. Curr Cardiovasc Imaging Rep (2011) 4: 98. doi:10.1007/s12410-011-9069-5
Cardiac MRI (CMR) offers the potential to assess valid and reliable parameters associated with cardiac diseases and the corresponding clinical prognosis. It has therefore emerged as a tool for outcome measures, resulting in lower study sample sizes, shorter duration of clinical studies, and subsequently less trial costs. This review focuses on the theoretical and practical background of CMR in measuring treatment effects in clinical trials.
KeywordsCardiac magnetic resonance imagingClinical trialsOutcome measures
The goal of clinical research identifying new therapeutic strategies is to reduce mortality and morbidity by altering the natural course of disease. The selection of appropriate end points in clinical trials is of crucial importance in the assessment of the effects of therapeutic interventions. Meaningful end points should 1) link the treatment effect to the underlying pathophysiology; 2) be clinically relevant and provide prognostic information to subsequent patient outcome; and 3) be determined accurately, with high specificity and sensitivity, as well as high reproducibility. Moreover, the choice of the primary end point influences the size of the projected treatment effect, which is a major determinant of study sample size [1••].
Cardiac MRI (CMR) provides a multifaceted view of the heart by assessing cardiac function and structure in vivo and noninvasively. CMR parameters have been linked to specific cardiac pathophysiological states and can be assessed with high intraobserver and interobserver reproducibility [2•]. Several CMR parameters have been linked to adverse clinical outcome in a diversity of cardiac diseases [3•]. Therefore, CMR offers the potential to assess clinical prognosis and therapeutic success, resulting in lower study sample sizes, shorter duration of clinical studies, and less trial costs. Consequently, CMR has emerged as an important tool for outcome measurements.
This review focuses on the role of CMR in measuring treatment effects in clinical trials. We will discuss the level of evidence and report on data from recent studies.
Statistical Background: Primary End Point, Treatment Target Effect, and Study Sample Size
Primary End Point
Clinical trial end points used to assess treatment effects should be causally linked to the underlying pathology . However, some end points might be directly linked to the pathophysiological elements but nonetheless not influence mortality or morbidity . Therefore, an end point should also be related to the patient’s clinical outcome. Furthermore, an ideal end point should be easy to measure and demonstrate high intraobserver and interobserver reproducibility, partly influencing sensitivity and specificity [1••].
The basis for the definition of the target treatment effect is the “minimum clinically important difference” of measures leading to the subsequent minimum change in an end point a trial is designed to detect . It is often defined by clinical judgment, based on current literature and guidelines or expert consensus. In cardiovascular medicine, clinical parameters such as death, myocardial infarction, or new congestive heart failure are often being used as primary end points, reflecting mortality and morbidity. The incidence of “hard” clinical end points is often relatively low, and long follow-up periods or large study sample sizes are required to detect treatment effects (e.g. reduction in mortality). Thus, surrogate parameters known to correspond with clinical outcome are often used as alternative end points to assess the effect of a new therapeutic strategy.
Statistical Power and Study Sample Size
For a superiority trial with a continuous surrogate end point, sample size is based on the effect size (the expected absolute difference between treatment groups), its variability, and the alpha and beta levels set by the study design . By convention, alpha and beta are usually fixed. However, by choosing an appropriate surrogate end point it is often possible to detect small effect sizes, thus leading to significant reductions in study sample sizes and an increase in statistical power.
Cardiac Magnetic Resonance
Assessment of cardiac function, structure, and tissue characterization using cardiac MRI
SE and double IR GRE/SSFP
- LV/RV/LA/RA dimensions
- Myocardial masses
- LV/RV/LA/RA volumes
- Ejection fraction
- Wall motion abnormalities (at rest + during stress)
- Valvular heart disease
- Ischemia evaluation, detection of functionally significant stenoses
- Cardiac mass
- Acute myocardial infarction
- Acute myocarditis/inflammation
- Iron overload cardiomyopathy
- Myocardial infarction (acute + chronic)
- Hypertrophic cardiomyopathy
Recent randomized clinical trials using CMR parameters as primary or secondary outcome measures
CMR end point
Volumes and function
Jahnke et al. 
Pulmonary vein isolation in atrial fibrillation
Gaddam et al. 
Spironolactone in patients with resistant hypertension and hyperaldosteronism
LV volumes and LV mass at 3 and 6 months
Cowan et al.  (ONTARGET CMR substudy)
Combination of telmisartan and ramipril vs ramipril alone in high-risk CAD patients
LV mass and volumes at 24 months
Kirschbaum et al. 
Complete vs incomplete revascularization in multivessel CAD
LV ejection fraction at 6 months
Chan et al. 
Endomyocardial implantation of autologous bone marrow cells in patients with severe CAD
LV ejection fraction and myocardial perfusion reserve at 6 months
Thiele et al.  (EPACCO trial)
Application of circulating progenitor cells in CTO
LGE and LV ejection fraction at 3 and 15 months
Pegg et al. 
Comparison of a hybrid technique on-pump beating heart CABG with conventional on-pump CABG in patients with CAD
LGE, LV ejection fraction, and LV volumes at 6 months
Yao et al. 
Intracoronary administration of bone marrow mononuclear cells in chronic myocardial infarction
LV ejection fraction and LV volumes at 6 months
Pegg et al. 
Off-pump CABG vs on-pump CABG in patients with CAD and CHF
RV ejection fraction and RV volumes at 6 months
Webb et al. 
Oral testosterone in men with low plasma testosterone and CAD
Myocardial perfusion reserve at 2 months
Cheng et al. 
PCI vs conservative treatment in CTO and nonoccluded CAD
Myocardial perfusion reserve at 2 months
Acute myocardial infarction
Thiele et al.  (LIPSIA-N-ACC study)
High-dose N-acetylcysteine in addition to primary PCI in STEMI
Myocardial salvage index at days 2–4
Abbate et al. 
Interleukin-1 blockade with anakinra in STEMI
LV end-systolic volume index at 10–14 weeks
Wöhrle et al. 
Autologous intracoronary bone marrow cell therapy after STEMI
LV ejection fraction at 6 months
Lonborg et al. 
Ischemic postconditioning in STEMI
Myocardial salvage at 3 months
Haeck et al. 
Proximal embolic protection and thrombus aspiration in STEMI
Infarct size at 4–6 months
Patel et al.  (APEX-AMI trial)
Pexelizumab (anti-C5 complement antibody) in STEMI
Infarct size and LV ejection fraction at 90 days
Weir et al. 
Eplerenone in CHF after STEMI
LV end-systolic volume index at 24 weeks
Dill et al.  (REPAIR-AMI trial)
Intracoronary administration of bone marrow–derived progenitor cells after STEMI
LV ejection fraction at 4 and 12 months
Tendera et al.  (REGENT study)
Intracoronary infusion of bone marrow–derived selected CD34 + CXCR4+ cells vs non-selected mononuclear cells after STEMI
LV ejection fraction at 6 months
Song et al. 
Upstream high-dose tirofiban treatment in STEMI
Infarct size at 1 month
Atar et al.  (FIRE study)
Administration of FX06 in AMI
Infarct size at 5 days
Sardella et al.  (EXPIRA trial)
Thrombus aspiration in STEMI
Microvascular obstruction at 4 days and infarct size at 3 months
Thiele et al.  (LIPSIAbciximab)
Intracoronary vs intravenous bolus abciximab application in patients with STEMI
Infarct size and microvascular obstruction at 2 days
Hahn et al. 
Distal protection device in STEMI
Infarct size at 3 days
Engelmann et al. 
Granulocyte colony-stimulating factor in subacute STEMI
Global and regional myocardial function at 3 months
Kang et al.  (MAGIC Cell-3-DES trial)
Intracoronary infusion of mobilized peripheral blood stem cells in STEMI
LV ejection fraction at 6 months
Ripa et al.  (STEMMI trial)
Granulocyte colony-stimulating factor after STEMI
Systolic wall thickening at 6 months
Janssens et al. 
Transfer of autologous bone marrow–derived stem cells after STEMI
LV ejection fraction at 4 months
Thiele et al. 
Pre-hospital combination-fibrinolysis plus conventional care vs pre-hospital combination-fibrinolysis plus facilitated PCI in STEMI
Infarct size at 6 months
Gick et al. 
Filter-based distal protection in STEMI
Infarct size at 3 days
Wollert et al.  (BOOST trial)
Intracoronary transfer of autologous bone marrow cells after STEMI
LV ejection fraction at 6 months
Valvular heart disease
Kim et al. 
Surgery in TR
RV volumes and RV ejection fraction at 1 and 27 months
Gelfand et al. 
Surgery in MR
LV volumes and LV ejection fraction at 27 months
Lurz et al. 
Bare metal stenting and percutaneous pulmonary valve implantation for treatment of right ventricular outflow tract obstruction
RV stroke volume and pulmonary regurgitation
Tanner et al. 
Deferoxamine monotherapy vs deferiprone and deferoxamine in patients with iron-overload cardiomyopathy
LV ejection fraction and T2* at 6 and 12 months
Hughes et al. 
Agalsidase in patients with Fabry disease
LV mass at 6 months
Shimada et al. 
Prednisolone in cardiac sarcoidosis
LGE at 1 month
Volumes and Function
Ventricular volumes, masses, and function can be assessed with CMR by acquisition of a three-dimensional (3D) stack of short-axis images covering the whole left and right ventricle. In comparison to other imaging techniques, such as echocardiography, assessment of cardiac function using CMR is more accurate and reproducible with lower intraobserver and interobserver variability .
Impaired right (RV) and left ventricular (LV) function is associated with adverse clinical outcome  and thus has been used repeatedly as a target of new therapeutic strategies, such as in the CMR substudy of the Reinfusion of Enriched Progenitor Cells and Infarct Remodeling in Acute Myocardial Infarction (REPAIR-AMI) trial . This study could demonstrate a significant improvement of LV function after 12 months through intracoronary administration of bone marrow–derived progenitor cells in 54 patients with ST-elevation myocardial infarction (STEMI) and impaired LV function. Likewise, in stable coronary artery disease (CAD) in the setting of recanalized chronic total occlusions it has been shown that application of circulating progenitor cells results in a significant improvement of LV function as compared to control at short-term and long-term follow-up .
LV mass and volumes are known to be influenced by different pathological states, such as arterial hypertension. In the Ongoing Telmisartan Alone and in Combination With Ramipril Global End Point Trial (ONTARGET), the combination of the angiotensin receptor blocker telmisartan with the angiotensin-converting enzyme inhibitor ramipril did not lead to a decrease in morbidity and mortality in comparison to ramipril alone . In the CMR substudy including 287 patients, the combination of telmisartan and ramipril had a similar effect on LV mass and volumes than ramipril alone . These results strongly support the major outcome findings and even give insight into the potential pathophysiological explanations for the results of the primary study.
Assessment of RV function using echocardiography is limited, mainly due to the complex and asymmetric shape of the right ventricle. 3D echocardiography has overcome some of these limitations, although assessment of RV function using CMR is easier to standardize, less prone to registration errors, and has better resolution than 3D echocardiography . CMR was used in a study of 31 patients undergoing surgery for severe functional tricuspid regurgitation for the assessment of RV volumes. The trial could demonstrate a significant reduction in RV volumes and preserved RV function post-surgery. Furthermore, RV end-diastolic volume index before surgery could effectively discriminate between patients with post-surgical normal RV function from those with impaired RV function, enlightening the importance of accurate pre-surgical assessment of RV volumes and function .
Apart from RV and LV function and volumes, CMR can also assess atrial volumes. A study by Jahnke et al.  could demonstrate that pulmonary vein isolation for atrial fibrillation with subsequent restoration of sinus rhythm leads to a significant decrease of left atrial systolic and diastolic volumes, indicating structural reverse remodeling.
Myocardial Ischemia and Scar in Stable CAD
Late gadolinium enhancement (LGE) can be assessed by T2-weighted imaging 10 to 15 min after gadolinium injection and appears as a hyperenhancement. In patients with CAD it corresponds to myocardial fibrosis and scar tissue. LGE has been shown to be associated with adverse clinical outcome in patients with suspected CAD even without clinical evidence of prior myocardial infarction and provides incremental prognostic information beyond common clinical, angiographic, and functional predictors . A small clinical CMR substudy demonstrated a significant reduction of LGE in patients with stable CAD after recanalization of chronic total occlusions and application of endothelial progenitor cells as compared to controls .
Acute Myocardial Infarction
Using T1-weighted and T2-weighted imaging, infarct size, microvascular obstruction (MO), as well as myocardial salvage and myocardial hemorrhage can be assessed (Fig. 1b).
With assessment of LGE using T1-weighted imaging it is possible to directly visualize and quantify areas of infarcted myocardium, both after STEMI and non-STEMI . In an international multicenter study this technique demonstrated high diagnostic sensitivity to detect either acute or chronic infarctions . Due to an in-plane spatial resolution of 2 mm leading to a fourfold to sixfold higher spatial resolution and a higher contrast-to-noise ratio than single photon emission CT, LGE imaging has shown to be consistently more sensitive and specific in detecting and quantifying the extent of infarction . Furthermore, infarct size has been shown to be associated with worse functional and clinical outcome  and has therefore been used as a surrogate end point in various studies examining reperfusion success. In the Randomized Leipzig Immediate Percutaneous Coronary Intervention Abciximab IV Versus IC in ST-Elevation Myocardial Infarction (LIPSIAbciximab) trial including 154 patients with STEMI undergoing primary percutaneous intervention (PCI), intracoronary bolus administration of abciximab led to a significant reduction of infarct size in comparison to intravenous injection . In the first multicenter study analyzing infarct size as primary end point, Atar et al.  could not show a significant reduction of infarct size in patients with STEMI receiving FX06, an anti-inflammatory acting peptide, in comparison to the control group. Various trials analyzing the impact of stem cell administration on infarct size during or early after reperfusion yielded conflicting results (Table 2).
CMR can also display zones of MO, which appear as central dark zones within the infarcted area. MO represents myocardial tissue hypoperfusion corresponding to “no-reflow.” MO can be assessed 1 to 2 min (early MO) and 10 to 15 min (late MO) after gadolinium injection. Recently, a higher prognostic impact of late MO in comparison to early MO independent of traditional angiographic, clinical, and electrocardiographic parameters could be demonstrated . In a substudy of the Thrombectomy with Export Catheter in Infarct-Related Artery During Primary Percutaneous Coronary Intervention (EXPIRA) trial, both the prevalence, as well as the extent of MO could be reduced by performance of thrombectomy prior to PCI as compared to controls . These results highlight the beneficial effect of the studied therapeutic interventions and give insight into the pathophysiological mechanisms leading to the development of MO.
Another part of a comprehensive infarct CMR protocol is T2-weighted imaging. T2-weighted CMR sequences are ultra-sensitive to water-bound protons, delivering information on myocardial edema displayed as areas of high T2-signal. Recently, T2-weighted imaging has demonstrated high accuracy in measuring the area at risk in reperfused infarcts [33, 34]. By comparing T2-weighted imaging to LGE, myocardial salvage can be assessed retrospectively . Importantly, CMR assessment of myocardial salvage can be performed by only one scan within the first week of the index event and does not require injection of an isotope as compared to nuclear scintigraphy. As a consequence, assessment of myocardial salvage has been significantly simplified by the introduction of CMR. Myocardial salvage is a strong predictor for adverse clinical outcome after STEMI  and has been used to test the efficacy of reperfusion modalities. In a study by Lonborg et al. , a significant increase in myocardial salvage through mechanical post-conditioning in patients with STEMI could be observed. In the Prospective, Single-Blind, Placebo-Controlled, Randomized Leipzig Immediate Percutaneous Coronary Intervention Acute Myocardial Infarction N-ACC (LIPSIA-N-ACC) study, the first clinical trial analyzing myocardial salvage as an end point, N-acetylcysteine, which has antioxidant properties, failed to show beneficial effects on myocardial salvage, despite encouraging results in animal models .
Recent studies have also demonstrated the potential of T2-weighted imaging to detect intramyocardial hemorrhage (IMH), a marker of severe reperfusion injury. A hypointense core within the hyperintense area at risk likely corresponds to such IMH. IMH has been shown to be an independent predictor of adverse LV remodelling irrespective of the initial infarct size and MO . However, histological proof of the specificity of hypointense cores in T2-weighted images (vs more specific T2*-weighted CMR) is still awaited. These remaining questions need to be clarified before the use of IMH in clinical trials can be recommended.
Valvular Heart Disease
Echocardiography represents the gold-standard for valvular morphological assessment and grading of severity of valvular stenosis and/or regurgitation. However, there is an increasing role of CMR regarding timing and assessment of outcome of valvular interventions. Only very limited data are available on valvular interventions and the impact on mortality, both for surgical and transcatheter interventions. Procedural success is most often based on ventricular functional and morphological remodeling. Given its high accuracy in the assessment of ventricular volumes and ejection fraction, CMR has been used to measure the impact of aortic valve replacement in the context of aortic regurgitation and stenosis on LV systolic and diastolic function . Further, CMR plays a major role in clinical trials on timing and outcome of right-sided valvular lesions. Especially in patients with congenital heart disease and pulmonary regurgitation or stenosis, the exact timing for interventions is still unknown and fiercely debated. In recent studies, RV end-diastolic volume thresholds, as assessed by CMR, have been established, beyond which normalization of RV volumes after pulmonary valve replacement is unlikely . Finally, with the extension of minimal invasive and transcatheter valve interventions, the techniques and the patient population undergoing treatment will change fundamentally. CMR is likely to be the key technique to assess the impact of these new techniques on biventricular remodeling .
Using cine-weighted and T2-weighted imaging, as well as LGE, further distinction between different entities of cardiomyopathies can be performed. The main focus is on the differentiation between ischemic or nonischemic origin. Nonischemic cardiomyopathies often display typical patterns in MRI.
Iron overload cardiomyopathy can be directly visualized using T2* imaging. T2* times have been shown to be inversely related to the amount of iron deposition and are associated with adverse clinical prognosis . In a study by Modell et al. , CMR-guided iron chelate treatment led to a significant decrease in mortality by identifying patients at the highest risk, leading to intensification of medical therapy.
In dilated cardiomyopathy LGE typically occurs mid-wall, particularly affecting the basal and mid-septum. Currently, LV ejection fraction is widely used to determine the need for an implantable cardioverter-defibrillator (ICD). In trials with limited patient numbers it has been shown that the amount of LGE is a stronger predictor for cardiac death and inducible arrhythmias than LV ejection fraction . These findings offer the potential for further risk stratification in patients with an LV ejection fraction >35%, although this has not been evaluated in a large study cohort so far.
A similar relation of LGE with sudden cardiac death was observed in patients with hypertrophic cardiomyopathy >40 years of age . However, no data exist using LGE to guide ICD implantation.
In Fabry disease, a condition in which α-galactosidase activity is reduced leading to an accumulation of glycosphingolipid GB3, midwall LGE of the basal lateral wall, as well as prominent LV hypertrophy can be observed. In a study by Hughes et al. , a significant decrease in LV mass could be demonstrated in patients receiving enzyme replacement therapy in comparison to the placebo group, which was consistent with other laboratory and electrocardiographic findings.
In patients with sarcoidosis, cardiac involvement is usually patchy and affects mainly the LV, resulting in low accuracy of endomyocardial biopsies due to sampling errors. In contrast, CMR can detect the often bizarre patterns of LGE and demonstrate areas of inflammation using T2-weighted imaging. This allows early identification of cardiac manifestations leading to an early initiation of corticosteroid therapy, which is supported by the findings of Vignaux et al. .
In suspected myocarditis, CMR can be understood as the first-line imaging technique . However, to date no therapeutic intervention other than symptomatic or unspecific treatment has shown significant improvement in clinical outcome or a change in the corresponding CMR findings. However, this is currently part of extensive clinical research, and CMR parameters might help identify beneficial effects of a treatment earlier than clinical outcome parameters.
In patients following heart transplantation, organ rejection can be assessed with the techniques used for the detection of myocarditis . This is of importance as screening for organ rejection is a critical component of care in post-transplantation patients. Although up to date endomyocardial biopsies are the standard screening tool, CMR might possibly reduce or eliminate the need for endomyocardial biopsies and be of importance for future studies analyzing new treatment strategies to prevent or reverse organ rejection.
With the introduction of CMR parameters as surrogate end points in clinical trials it is possible to assess biologically plausible and reliable data with high prognostic impact often occurring earlier and more frequently than the corresponding clinical outcome measures. This leads to a reduction of the projected follow-up durations as well as study sample sizes. Therefore, CMR is likely to play an important role in future clinical trials.
Drs. de Waha and Fuernau contributed equally to this work.
No potential conflicts of interest relevant to this article were reported.