Real-time cardiovascular magnetic resonance T1 and extracellular volume fraction mapping for tissue characterisation in aortic stenosis

Between January 2016 and August 2019 patients who underwent transcatheter aortic valve replacement (TAVR) following heart team consensus decision were approached for study participation. Ethical approval was obtained from the local ethics committee. All patients gave written informed consent before participation. The study was conducted according to the principles of the Helsinki Declaration. The study was funded by a grant from the German Research Foundation (DFG, CRC 1002, D1).

CMR imaging was carried out using a standard protocol including balanced steady state free precession (bSSFP) functional imaging, native mapping prior to gadolinium application and post-contrast mapping 20 min after contrast administration as well as late gadolinium enhancement (LGE) imaging. Cardiac volumes were assessed in short axis (SAX) stacks acquired using electrocardiogram(ECG) gated bSSFP cine imaging. Mapping sequences comprised commercially available MOLLI (native: 5(3)3, post-contrast: 4(1)3(1)2) sequences [19‐21] (Fig. 1) (MOLLI T1 mapping) with typical imaging parameters comprising a field-of-view (FOV) of 360 × 306.6 mm², in-plane resolution 1.41 × 1.41 × 8 mm³, TR/TE = 2.7/1.12 ms, nominal flip angle 35°, bandwidth 1085 Hz/pixel and a total acquisition of 11 heart beats. Additionally, RT T1 single-shot myocardial T1 maps [16] (RT T1 mapping) of the mid-LV slices were acquired (Fig. 1) at a nominal in-plane resolution of 1.0 × 1.0 mm² and 8 mm slice thickness using a FOV 256 × 256 mm² in combination with a resolution of 512 complex data points per radial spoke (two-fold oversampling). Other parameters were TR/TE = 2.67/1.66 ms, nominal flip angle 4°, bandwidth 850 Hz/pixel and total acquisition time of 4 s following a non-selective adiabatic inversion pulse. Correct overlapping of underlying T1 weighted images was checked by the operator prior to drawing of the respective region of interest (ROI). The ROI was then propagated to all underlying T1 weighted images to create a midventricular SAX slice coverage and a septal ROI separately. To avoid partial volume effects of the blood pool or adjoining non-myocardial structures ROI delineation was carefully reviewed on all underlying T1 weighted images. ECV was calculated for the entire myocardium as well as the septal ROI separately as recently published [22]. To specifically evaluate diffuse processes, T1 and ECV ROIs excluded focal fibrosis/scars represented by focal LGE. Calculation of matrix volumes were performed using the product of LV myocardial volume (LV mass divided by specific gravity of myocardium [1.05 g/mL]) and ECV or (1 – ECV) for cellular volumes [13]. LGE imaging for the evaluation of ischemic scar was performed using inversion-recovery-gradient echo sequences 10–20 min after the administration of gadolinium-based contrast agents (0,15 mmol/kg). Post-processing analyses were performed using Medis (version 3.1.16.8, Medical Imaging Systems, Leiden, Netherlands). Quantifications were performed by observers blinded to clinical and histological or imaging characteristics respectively.

LV endomyocardial biopsies were taken during TAVR procedures prior to replacement and from the anteroseptal basal myocardial segment using a Proflex-Bioptom (7 F, Medical Imaging Systems). The study procedure accounted for additional 5 to 10 min of intervention time. TAVR and CMR were performed within 48 h. Biopsies were fixed in 10% paraformaldehyde (PHA) and embedded in paraffin. Fibrosis was assessed using quantitative morphometry (cell-Sens 1.6, Olympus Corporation, Tokyo, Japan) and defined as blue area in Masson trichrome staining (MTS, section with positive staining for collagen) in relation to total tissue area. Collagen volume fraction (CVF) was defined as the amount of total fibrosis, including subendocardial fibrosis (ischemic), interstitial diffuse fibrosis (reactive) and replacement (scar) fibrosis [17] (Fig. 1), in relation to the total area of the biopsy. To precisely define the amount of interstitial myocardial fibrosis areas of subendocardial or replacement fibrosis were then excluded from the analysis volume in a second step with the final volume only including cardiomyocytes and interstitial fibrosis. Subsequently, diffuse CVF (dCVF) was defined as the percentage of interstitial fibrosis in relation to the remaining volume respectively.

Continuous parameters are presented as median values with interquartile range (IQR) after testing for normal distribution using the Shapiro-Wilk test. Non-normally distributed parameters underwent logarithmic transformation to achieve normal distribution. Dependent variables were compared using the Wilcoxon signed-rank test. Correlations were assessed using Pearson correlation coefficients. P-values provided are two-sided, an alpha level of 0.05 and below was considered statistically significant. Statistical agreement and variability were assessed in 20 randomly selected patients using the coefficient of variation (CoV, SD of mean difference divided by the mean) and intra-class correlation coefficients (ICC). Intra-observer reproducibility was performed with 4 weeks in between repeated analyses. ICCs were evaluated based on a model of absolute agreements as well as on a model for consistency [23]. ICC was considered excellent for values ≥0.74, good ≥0.60, fair ≥0.4 and poor < 0.40. Bland-Altman analyses [mean difference between measurements with 95% confidence interval (CI)] were performed to visualize agreements and variability [24]. Statistical analyses were calculated using SPSS (version 24, Statistical Package for the Social Sciences, International Business Machines, Inc., Armonk, New York, USA,) and Microsoft Excel (Microsoft, Redmond, Washington, USA).

Baseline characteristics are reported in Table 1. In total 110 patients agreed for participation in the CMR study. LV endomyocardial biopsies were taken in 46 patients [25]. There was one biopsy induced papillary muscle injury resulting in an increase from moderate to severe mitral regurgitation which was successfully treated with Mitraclip (Abbott, Abbott Park, Illinois, USA) implantation leading to subsequent reduction to mild mitral regurgitation.

Agreements of MOLLI and RT mapping are reported in Tables 3 and 4 as well as Figs. 3 and 4. Statistical agreements comparing assessments by MOLLI and RT sequences are higher for the septum compared to the myocardium. This is also reflected by higher correlation coefficients for septal assessments. Statistical agreements are good for native T1 times (ICC > 0.62), excellent for post-contrast T1 times (ICC > 0.81) as well as excellent and best for ECV (ICC > 0.91) irrespective of midventricular septal or midventricular SAX slice coverage. Intra- and interobserver reproducibility of T1 times and ECV were consistently excellent (Table 5).

Whilst intra- and inter-observer reproducibility was assessed underlining excellent post-processing reproducibility, data on inter-study reproducibility is lacking which would be desirable for follow-up surveys of cardiac remodelling. With this perspective, further clinical follow-up studies are warranted. Mapping was performed in one representative midventricular SAX slice only, future studies should consider an apical, midventricular and basal slice for more comprehensive assessments. We have not demonstrated a significant association of native and post contrast T1 mapping based on RT and MOLLI (native 5(3)3 and post-contrast 4(1)3(1)2) with histology. This may partially be due to the specific sequences used. Both T1 mapping sequences utilized are affected by imperfect inversion due to T2 relaxation. While the MOLLI sequence is additionally affected by T2 and magnetization transfer effects during the bSSFP readout (Robson et al. [38]), this does not affect the RT single-shot T1 mapping sequence which uses a FLASH readout as in the original approach by Deichmann and Haase [39]. However notwithstanding it is important to understand that native and postcontrast T1 based on various T1 mapping techniques did not correlate with outcome in a recent AS multi-center study while ECV did predict outcome in these patients [21]. Furthermore, ECV did correlate with histological fibrosis in the current study, which itself was demonstrated to have strong prognostic implications [33]. Last, only a subset of patients underwent endomyocardial biopsies which limits statistical power.