Serial T1 mapping of right ventricle in pulmonary hypertension: comparison with histology in an animal study

All experiments were approved by the Animal Care and Use Committee of our institution, and we performed experiments according to the National Institutes of Health guidelines. A total of 46 6-week-old Wistar male rats were used. A single subcutaneous injection of monocrotaline (40 mg/kg) (Sigma Chemical, St. Louis, Missouri, USA) was used to induce PH [16, 17]. Animals were divided into 5 groups according to disease duration; baseline before injection, and 2, 4, 6 and 8 weeks after injection, respectively. Among 46 animals, 4 expired before the study was completed (1 at 4 weeks, 1 at 6 weeks, and 2 at 8 weeks), and 6 animals allocated to each of the 5 disease-duration groups. The remaining 12 6-week-old Wistar male rats were used to control for confounding factors thought to arise from aging, and 6 animals were assigned to each group after 4 weeks and 8 weeks after the baseline time point without monocrotaline injection. The weight of each animal was measured immediately before CMR scanning to calculate body surface area [18]. Venous blood samples were obtained from the tail vein for hematocrit levels to calculate ECV.

All CMR images were acquired through a 9.4 T small animal CMR scanner (Bruker Biospin, Billerica, Massachusetts, USA) with an 86-mm volume transmit coil and a 4-channel phased-array surface coil. CMR was performed under spontaneous ventilation with anesthesia by inhalation of 2% isoflurane. Cine images in the short-axis planes were acquired using a self-gated fast low-angle shot sequence (IntraGate, Bruker BioSpin, ParaVision 5.1) with the following parameters: TR/TE, 5/2.3 ms; flip angle, 10°; field of view, 40 × 40 mm; pixel size, 0.32 × 0.32 mm²; section thickness, 2 mm; number of slice, 8—9; and cardiac phase, 25. Pre-contrast T1 maps were acquired in the short axis planes using a respiratory navigated Look-Locker inversion recovery imaging [19] (Additional file 1: Fig. S1). The sequence was proposed to provide higher motion robustness so that the myocardial T1 maps could be accurately measured in small animals with higher heart rates and respiratory rates. The sequence is based on the Look-Locker sequence, which continuously acquires with a spoiled gradient-echo readout sequence during T1 recovery following the inversion pulse (IR) without respiratory gating. Therefore, the sequence is more suitable for small animals with heart rates of 300–400 bpm. In this T1 mapping sequence, image data was acquired consecutively during T1 recovery process with electrocardiographic (ECG) gating. Respiratory motion compensated IR-weighted images were reconstruced using a self-gating method that estimates respiratory motion from the acquired image data itself [20]. Scanning parameters were as follows: TR/TE, 15/1.52 ms, flip angle, 8°; field of view, 40 × 40 mm, pixel size = 0.2 × 0.2 mm²; section thickness, 1.5 mm; slice gap, 1.5 mm; acquisition duration, 16 heart beats; number of slice, 3; and relaxation duration, 5000 ms. The RV myocardial thickness of normal Wistar male rats was approximately 0.6 mm in diastole and 0.9 mm in systole. The relative spatial resolution was 0.2 × 0.2 mm². Post-contrast T1 maps were obtained with the same protocol 15 min after administration of the contrast agent (gadoterate meglumine, 0.2 mmol/kg; Dotarem; Guerbet, Roissy, France). T1 maps were generated by applying three-parameter non-linear least square fitting on a pixel-by-pixel basis for the motion-corrected phase-sensitive inversion recovery images and then Look-Locker correction [21].

All CMR images were evaluated with commercially available software (cvi42; Circle Cardiovascular Imaging, Calgary, Alberta, Canada). Two radiologists (Y.J.H. and H-J.L. with 10 and 12 years of experience in cardiovascular imaging) reviewed the images in consensus and were blinded to disease duration. From the short-axis cine images, the end-diastolic and systolic volumes of both ventricles were calculated by multiplying the areas enclosed by the endocardial contours by slice thickness. The end-diastolic volume indices (EDVI) were calculated by dividing the end-diastolic volume by body surface area (m²). The ejection fraction (EF) was calculated with the following equation: (end-diastolic volume − end-systolic volume) × 100/end-diastolic volume (%). The ECV was measured using pre- and post-contrast T1 maps as follows: (ΔR1_m/ΔR1_b) × (1 − hematocrit) × 100 (%), for which R1_m is R1 in the myocardium, R1_b is R1 in the blood, and ΔR1 is the change in T1 relaxivity before and after the contrast agent was administered. T1 measurements were performed on the middle third of the RV free wall, septum and LV lateral wall right at the midway between the apex and base of the LV on short axis plane. When measuring the middle third of RV free wall, the inner junctional points of both ventricles were used as positional reference points to match the imaging findings with the histologic findings (Fig. 1). The endo- and epicardial borders of the myocardium were delineated manually for T1 measurements, and the very edges were excluded to reduce partial volume averaging from the myocardial-blood interface. Additionally, T1 measurements were done at the anterior and inferior SIPs with circular regions of interest (ROI) of approximately 2 mm². Blood T1 values were also measured in the LV cavity while avoiding papillary muscles.

After CMR imaging, all hearts were resected and fixed in 4% formaldehyde, and sliced in sections approximately 2-mm thick right at the midway between the apex and base of the LV on the short-axis plane. The sections were embedded in paraffin and were sliced further into 5-µm-thick sections. Afterwards, the specimens were stained with H&E and Masson’s trichrome staining. Histologic analysis was performed by a pathologist H.S.S. with 13 years of experience in thoracic pathology blinded to disease duration and CMR findings. H&E-stained sections were used for the overall examination. Masson’s trichrome-stained sections were digitally scanned with high-power magnification (400X) (ImageScope v10.0; Aperio Technologies, Vista, California, USA), and the pixel numbers of the myocardial tissue and collagen were counted using a macro written in ImageJ (National Institute of Health, Bethesda, Maryland, USA) based on the thresholding algorithms [22, 23]. Collagen density for the RV was calculated as the percentage of pixel numbers of collagen to the myocardial tissue within a defined ROI, which was the same on T1 maps located at the middle one third level of the RV free wall.

Baseline characters are summarized in Table 1. The mean baseline body weight was 267.8 ± 8.1 g, and body weight increased according to age and disease duration (P < 0.001). The mean hematocrit level for ECV calculations was 53.8 ± 2.0% at baseline, and increased according to age and disease duration (P < 0.001), with 6 weeks and 8 weeks showing significantly increased levels compared to the baseline (P = 0.004 and 0.002, respectively). The mean heart rate was not significantly different according to age and disease duration (P = 0.648).

The results of cardiac function are described in Fig. 2 and Additional file 1: Table S1. The mean baseline values of RVEDVI and LVEDVI were 10.1 ± 0.4 ml/m² and 10.4 ± 0.9 ml/m², and of RVEF and LVEF were 67.6 ± 2.0% and 66.3 ± 2.2%, respectively. The mean values of RVEDVI increased and the mean values of LVEDVI decreased according to disease duration (P < 0.001 for both) (Fig. 2a). The mean values of RVEF and LVEF decreased according to disease duration (P < 0.001 for both) (Fig. 2b). For RVEDVI and RVEF, significant differences were first observed at 4 weeks compared to the baseline (P = 0.004 for both), and the values did not show significant differences between 4 and 6 weeks (P = 0.109 and 0.150, respectively); However, significant differences were observed between 6 and 8 weeks (P = 0.004 and 0.025). For LVEDVI, significant differences were observed at 6 weeks compared to the baseline (P = 0.004), and no significant differences were observed between 6 and 8 weeks (P = 0.180). For LVEF, significant differences were first observed at 4 weeks compared to the baseline (P = 0.037). In addition, no significant differences were observed between 4 and 6 weeks (P = 0.055), but a significant difference was observed between 6 and 8 weeks (P = 0.030).

The results of myocardial native T1 are described in Fig. 3 and Additional file 1: Table S2. The mean values of native T1 on baseline group were 1541 ± 33 ms, 1552 ± 39 ms, 1562 ± 42 ms, 1529 ± 42 ms, and 1576 ± 37 ms for the RV, septum, anterior SIP, inferior SIP, and LV, respectively. The mean values of native T1 for each section significantly increased according to disease duration (P < 0.001 for the RV and inferior SIP, 0.006 for the septum, and 0.001 for the anterior SIP, respectively) except for the LV (P = 0.349). Significant differences were first observed at 6 weeks for the RV and septum (P = 0.004 and 0.022, respectively), and further significant differences were observed between 6 and 8 weeks (P = 0.004) for the RV, but not for the septum (P = 0.223). For the anterior SIP and inferior SIP, the first significant difference was observed at 4 weeks (P = 0.026 and 0.002, respectively), but no further significant differences were observed between 4 and 6 weeks (P = 0.180 and 0.132, respectively), and between 6 and 8 weeks (P = 0.240 and 0.310, respectively).

The results of ECV are described in Fig. 4 and Additional file 1: Table S3, which showed similar patterns to those of native T1. The mean values of ECV on baseline group were 17.2 ± 1.3%, 17.5 ± 2.0%, 17.9 ± 1.5%, 17.4 ± 2.0%, and 17.7 ± 1.1% for the RV, septum, anterior SIP, inferior SIP, and LV, respectively. These mean values of ECV showed significant increase according to disease duration (P < 0.001 for the RV and inferior SIP, 0.020 for the septum, and 0.001 for the anterior SIP, respectively) except for the LV (P = 0.240). The first significant difference compared to the baseline was observed at 6 weeks for the RV and septum (P = 0.002 and 0.041, respectively). Additional significant differences between 6 and 8 weeks were observed for the RV (P = 0.009), but not for the septum (P = 0.818). For the mean ECV values of the anterior SIP and inferior SIP, significant differences were first observed at 4 weeks compared to the baseline (P = 0.015 and 0.002, respectively), but with no significant differences between 4 and 6 weeks (P = 0.485 and 0.240, respectively), and between 6 and 8 weeks (P = 0.818 and 0.180, respectively).

Animals showed varying degrees of RV hypertrophy macroscopically dependent on disease duration. Microscopic analysis found evidence of varying degrees of myocyte hypertrophy, interstitial fibrosis, muscular fiber disarray, and lymphocytic infiltration with collagen deposition according to disease duration. The mean values of collagen density for the RV were 4.7 ± 0.5%, 4.6 ± 0.3%, 5.2 ± 0.2%, 10.6 ± 0.5%, and 14.0 ± 0.5% at the baseline, 2 weeks, 4 weeks, 6 weeks, and 8 weeks, respectively (Fig. 5). The values significantly increased with disease duration (P < 0.001) and showed similar patterns with native T1 and ECV of the RV free wall. The first significant increase was observed at 6 weeks (P = 0.002) and another significant increase was observed between 6 and 8 weeks (P = 0.002). In addition, significant correlations were observed between native T1 and collagen density (r = 0.770, P < 0.001), and between ECV and collagen density for the RV free wall (r = 0.815, P < 0.001). (Fig. 6 and Fig. 7).

The results of the normal control study are described in Additional file 1: Table S4. Mean body weight increased with age (P < 0.001). However, there were no significant differences for mean hematocrit levels and heart rates (P = 0.264 and 0.076, respectively). For cardiac function, there were no significant differences between RVEDVI and LVEDVI (P = 0.244 and 0.147, respectively) as well as between RVEF and LVEF (P = 0.171 and 0.580, respectively). When characterizing myocardial tissue, native T1 values (P = 0.244 and 0.548 for RV and LV, respectively) and ECV values (P = 0.504 and 0.291 for RV and LV, respectively) of both ventricles showed no significant difference at baseline, 4 weeks, and 8 weeks. In addition, no significant differences were observed for collagen density in the RV according to age (P = 0.331).

Our study has limitations. First, we did not perform a longitudinal observation because frequent administration of anesthesia for CMR scanning often leads to the death of animals with PH, and that would have impeded the progress of our study. However, because the effect of monocrotaline has been reported to be constant throughout time in previous reports [16], the serial changes observed in this study might be acceptable. Second, a variety of different histological methods are currently used to quantify collagen in myocardial fibrosis, and the amount of estimated collagen density can vary slightly depending on the quantification method [28]. Third, we measured native T1 and ECV values with CMR and collagen density with histologic specimens in the middle one third level of the RV to match the corresponding locations. However, slight differences might have been present.