MRI has previously demonstrated increased lung water content in patients with heart failure (HF), but has not yet been used to distinguish between intravascular and extravascular water in these patients. This study evaluated dynamic contrast-enhanced MRI (DCE-MRI) for measuring pulmonary oedema and endothelial permeability in healthy volunteers (HV) and chronic HF patients at rest and post-exercise. DCE-MRI showed a redistribution of lung water towards the interstitial space in chronic HF, as compared to HV, suggesting this method may have value as a novel endpoint for dose-ranging and proof-of-mechanism studies in chronic HF. No exercise-induced change was seen in either group.
37 subjects were enrolled in the study (23 HV and 14 chronic HF patients). The evaluable population was defined as patients with MRI data and age >40 yrs, leaving 29 subjects – 17 HV (15 male, 2 female; mean(SD) age = 61.2 (11.4) yrs; mean(SD) body mass index = 26.4 (2.9) kg/m2) and 12 HF patients (10 male, 2 female; mean(SD) age = 67.8 (13.4) yrs; mean(SD) body mass index = 29.2 (2.9) kg/m2). Chronic HF patients had a diagnosis of mild or moderate HF of any aetiology with symptoms corresponding to New York Heart Association (NYHA) Class II or III.
Subjects had three whole-chest DCE-MRI scans each – a baseline scan (Session 1), a repeat baseline ~1 week later for evaluation of repeatability (Session 2), and a post-exercise challenge scan (within 1-3 days of Session 2) to explore effect of exercise on DCE-MRI parameters (Session 3). Imaging was performed on a 1.5 T GE MR450 scanner (software version DV24), using an 8-channel cardiac coil, with subjects in a supine position and free-breathing throughout. The DCE-MRI protocol included variable flip angle (FA) T1-mapping acquisitions, using a 3D coronal spoiled gradient echo sequence with core parameters of: TR/TE = 2.03/0.83 ms, FOV (in plane) = 450 mm x 450 mm, slice thickness = 10 mm and number of slices = 18 (interpolated to 28 slices at 5 mm), acquisition matrix = 112x88, FA = 4/2/7/10° (10 repeated volumes). This was followed by a 7-minute dynamic series (parameters as for T1-mapping, FA = 10°, 170 repeated volumes, temporal resolution = 2.5 s/volume). Intravenous contrast agent was injected at the 15th volume (half-dose Gadovist administered by power injector at 1.5 ml/s, with 25 ml saline flush). Total scan time was ~10 mins, allowing for inclusion of patients with mild orthopnoea.
Data were quality controlled, and breathing motion corrected using a non-linear image registration2. Lungs were segmented from the thorax using a semi-automatic thresholding technique. DCE-MRI analysis was then performed with pharmacokinetic modelling using the extended Tofts model3 to quantify ve, vp and Ktrans on a voxel-wise basis. Analysis allowed production of 3D maps of DCE-MRI parameters in the lung from which summary statistics could be extracted. Data were fitted using a repeat measure analysis of variance (ANOVA) model with terms including patient population (HF or HV), visit, interaction of patient population and visit, with subject ID as block of the repeat factor.
Study procedures were tolerated well by both groups. Increased ve was observed in HF at each scan (Figure 1), with clear separation between HF and HV subjects (Table 1, Figure 2). Analysis of Ktrans, vp, T1 and proton density in the lungs showed no clear difference between HF and HV subjects at each scan. Good reproducibility was demonstrated in measurements of ve and Ktrans between Session 1 and 2 – no statistically significant variability for either HV or HF subjects, and magnitude of the differences observed between the two sessions were generally small. Analysis of DCE-MRI data pre-/post-exercise showed no clear effect of exercise on DCE-MRI parameters for either group (e.g. ve (Session1 vs. Session3), p=0.48 (HV) and p=0.23 (HF)).