Right ventricular myocardial deoxygenation in patients with pulmonary artery hypertension

Patients undergoing treatment in the PAH clinics at two South Australian hospitals were invited to participate in this study. Inclusion criteria included right heart catheter (RHC) proven PAH (defined as pulmonary capillary wedge pressure (PCWP) \(\le\) 15 mmHg, and mean pulmonary artery pressure (mPAP) ≥ 25 mmHg). Exclusion criteria included severe RV dysfunction on echocardiography (determined by tricuspid annular plane systolic excursion < 1.7), echocardiographic LV ejection fraction < 50% and/or coronary artery disease (defined as > 70% luminal stenosis in an epicardial coronary artery at angiography or prior myocardial infarction) as well as contraindications to CMR and/or adenosine (second or third degree heart block, obstructive pulmonary disease or dipyridamole use). Eleven healthy subjects also underwent the same CMR imaging procedures and served as a control group. The healthy subjects had no known cardiac/respiratory disease or symptoms and were free of any cardiac risk factors, including hypertension, smoking and diabetes. This study was approved by the Southern Adelaide Clinical Human Research Ethics Committee (HREC/15/SAC/397), and all participants provided written and informed consent to participate in the study.

The full protocol has been described elsewhere [18]; in brief, CMR was performed using a 3 T clinical CMR scanner (MAGNETOM Skyra, Siemens Healthineers, Erlangen, Germany) with 18 channel torso phased array coil in conjunction with a spinal coil posteriorly). Electrographically (ECG) gated balanced steady-state free precession (bSSFP) sequence was used to acquire cine images in vertical and horizontal long-axis, and ten short-axis images covering the entire RV and LV. For OS imaging, a single mid-ventricular slice was acquired at mid-diastole using a single-shot T2-prepared ECG-gated bSSFP sequence (FOV of 340 × 340 cm, matrix 168 × 192 m, slice thickness 6 mm, producing a voxel size 2.02 × 1.77 mm. The repetition time (TR) was 256 ms and with echo time (TE) of 1.21 ms, flip angle 44˚). Each OS-CMR image was acquired during a single breath-hold over six heart beats. Prior to adenosine infusion, four resting OS-CMR images were acquired. Stress OS-CMR stress images acquisition commenced 2 min after adenosine infusion. Four to six OS-CMR images were acquired over the stress period with the typical duration of stress OS-CMR image acquisition was 3 min with the total adenosine infusion duration of about 5 min (Fig. 1).

Native T1 mapping were acquired using Shortened Modified Look-Locker Inversion recovery (ShMOLLI) on a single mid-ventricular short-axis slice as previously described [19]. The ShMOLLI sequence was performed with prospective ECG triggering performed until optimal image quality was obtained. This sequence has a FOV of 360 × 360 cm, matrix 192 × 144, slice thickness 8 mm, producing an interpolated voxel size 0.9 × 0.9 mm. The TR was 379.40 and with TE of 1.07. Typically, the flip angle was set to 35 degrees and IPAT (GRAPPA) factor of 2, with inversion time (TI) of 260 ms. The T1 maps were acquired after the resting OS-CMR images prior to adenosine infusion (Fig. 1). The mid-ventricular slice location selected was matched to the selected mid-ventricular resting OS-CMR images. Stress heart rate and blood pressure were obtained every minute of adenosine infusion. Patients were monitored throughout the study by ECG, sphygmomanometry and pulse oximetry.

CMR analysis was performed using cvi⁴² (Circle CVI, Calgary, Canada). The short-axis bSSFP cine images were used to quantify LV and RV systolic function. Standard function parameters were calculated including LV ejection fraction (LVEF), RV ejection fraction (RVEF), cardiac index, and body surface area (BSA) indexed end-diastolic (EDVI), end-systolic (ESVI) volumes for LV/RV and LV indexed myocardial mass. End-diastolic RV mass and volumes was used to calculate RV mass/volume ratio.

The OS-CMR assessment of the RV and LV was performed using cvi⁴² (Circle Cardiovascular Imaging, Calgary, Canada) as previously described [18]. In brief the RV was segmented into RV anterior, RV free-wall and RV inferior segments. OS-CMR analysis of RV anterior and RV free-wall were limited by thinned myocardium and partial volume effects [18]. Reliable OS signal was only obtained in RV inferior [18], hence a region of interest (ROI) contour was traced manually in this RV segment. For the LV OS-CMR assessment, the epicardium and endocardium borders were manually traced and the LV was sub-divided into 6 equiangular segments based on a standard American Heart Association (AHA) segmentation of the mid-ventricular slice as previously reported [20]. The mean myocardial Signal Intensity (SI) within the RV and LV segments was obtained as previously described [18]. As it is a cardiac gated sequence, the mean signal intensity at rest and stress was corrected for the variations of heart rate and its effects on T1 relaxation as previously described [21]. The relative SI change was calculated as ΔSI (%) = (SI stress-SI rest)/SI rest × 100. Reproducibility of the OS-CMR of the RV has been previously published [18].

The T1-values of the RV were assessed using cvi⁴² (Circle Cardiovascular Imaging). As previously described [22], a ROI contour was traced manually on the T1 maps of inferior RV region avoiding the RV insertion point (Fig. 2); attempts were made to draw ROI in RV anterior and RV free-wall regions but this was not feasible. The ROI was carefully drawn within myocardial walls to avoid partial volumes and epicardial fat (Fig. 2). For the LV, the endocardial and epicardial contours were manually traced and was sub-divided into 6 equiangular segments based on a standard AHA segmentation [20]. A mean myocardial T1 value was obtained within each segment.

All analyses were performed using the Stata statistical software version 15.1 (StataCorporation, College Station, Texas, USA). Categorical data are described using frequencies and percentages and continuous data using median and range. Differences between groups in the mean change in myocardial oxygenation response and native T1-values were assessed using Mann–Whitney test. Changes in these same outcomes within patient groups were assessed using the Sign-rank test. Reproducibility of measurements for inter and intra-observers was assessed by using the coefficient of variation and by assessing the absolute and proportional bias. Absolute bias was assessed using a t-test of the differences in the 2 measures versus zero and proportional bias by regressing the differences in the 2 measures on the mean for the 2 measurements. Associations between test parameters were assessed using the Spearman’s rho (ρ) correlation coefficient. A 2-sided Type 1 error rate of alpha = 0.05 was used for assessing statistical significance.

The subject characteristics are summarised in Table 1. Both PAH and healthy control groups were well matched for age and gender. The aetiology within PAH were idiopathic pulmonary artery hypertension (IPAH) (n = 20, 48%), systemic sclerosis associated PAH (SScPAH) (n = 17, 40%) and chronic thromboembolic pulmonary hypertension (CTEPH) (n = 5, 12%). All patients had RHC and coronary angiogram proven diagnosis of PAH with a median (range) mPAP of 33 (16–67) mmHg and PCWP of 12 (4–15) mmHg. Mean duration between RHC with coronary angiogram and CMR study was 17 months. The mean 6 min walk distance in the patient group was 384 (100–600) meters. World Health Organisation functional class was between class II (n = 27, 64%) and III (n = 15, 36%). The PAH patients had an echocardiography-estimated resting systolic pulmonary artery pressure of 41 (27–128) mmHg. Furthermore, on echocardiography 8 patients (19%) had mild RV dilatation and 5 patients (12%) had moderate RV dilation. All PAH patients were receiving treatment with pulmonary vasodilators (e.g., bosentan, macitentan, sildenafil, riociguat) or a combination of macitentan and sildenafil. At recruitment, the PAH patients were stable and the pulmonary vasodilators therapy was continued prior to the CMR research scan.

Table 2 shows a comparison of CMR variables between PAH patients and controls. The only significant difference in CMR volumetric and functional indices between PAH and healthy controls were the RV mass index (17 (9–42) vs 13 (9–19)g/m², p = 0.001), RV mass/volume ratio (0.30 (0.22–0.49) vs 0.22 (0.15–0.25), p < 0.001) and inferior RV wall thickness (4 (3–9) vs 3 (2–5)mm, p = 0.002). Table 3 demonstrates the CMR variables between the IPAH and SScPAH subgroups. The inferior RV wall thickness was measured on the short axis cine slice location similar to the mid ventricular slice location chosen for the OS imaging.

Of the total 42 PAH patients recruited, 40 (95%) patients and all the healthy controls completed the OS-CMR study protocol. The mean RV ΔOS-CMR SI change in the inferior RV segments was significantly lower in PAH group compared to the controls (9.5 (-7.4 – 42.8) vs 12.5 (9 – 24.6)%, p = 0.02) (Fig. 3a). The global LV ΔOS-CMR SI change was also significantly lower in PAH group compared to the controls (11.0 (-2.7 – 33.2) vs 21.0 (2.5 – 35.7)%, p = 0.004) (Fig. 3b). The ΔOS-CMR SI in the inferior RV wall of PAH patients was comparable to both the LV septal wall (9.5 (– 7.4–42.8) vs 11.3 (– 2.8–36.5)%, p = 0.32) and LV lateral wall (9.5 (– 7.4–42.8) vs 11.5 (– 9.2–31.8)%, p = 0.57) of PAH patients.

Good quality and analysable RV native T1 maps were obtained from 71% (30/42) PAH patients and 73% (8/11) of controls. Reasons for excluding T1 maps were image artefacts and partial volume effects. Compared to healthy controls, patients with PAH had higher native T1 values in the inferior RV wall: 1303 (1107–1612) vs 1232 (1159–1288)ms, p = 0.049 (Fig. 3c). The LV native T1 for the healthy controls were within previously published ranges [23]. In PAH patients, the native T1 values in the inferior RV wall was comparable to both the LV septal (1303 (1107–1612) vs 1332 (1169–1492)ms, p = 0.71) and LV lateral wall (1303 (1107–1612) vs 1301 (1102–1500)ms, p = 0.85) of PAH patients. The inter-observer absolute bias was 2.0, 95% confidence interval (CI) (– 4.95, 8.95), p = 0.53 and the relative bias was – 0.05, 95% CI (– 0.19, 0.84), p = 0.41. The intra-observer absolute bias was 9.1, 95% confidence interval (CI) (– 39.12, 57.32), p = 0.68 and the relative bias was 0.48, 95% CI (– 0.28, 1.23), p = 0.18.

The RV ΔOS-CMR SI between the IPAH (n = 20) and SScPAH (n = 17) patients was 9.4 (– 5.1–23.6) vs 9.0 (– 7.4- 42.8)%, p = 0.88 (Fig. 4a). The native T1 in the inferior RV wall between the IPAH (n = 14) and SScPAH (n = 14) patients were significantly different: 1242 (1107–1612) vs 1386 (1219–1552)ms, p = 0.007 (Fig. 4b). Among the 5 CTEPH patients, 2/5 (40%) did not complete the OS-CMR protocol and 3/5 (60%) had poor quality RV native T1 maps, hence no comparison was made in this group.

There was a moderate-good correlation between the inferior RV ΔOS-CMR SI and the global LV ΔOS-CMR SI (r = 0.7, p < 0.001) (Fig. 5a). Compared with the CMR volumetric/functional and haemodynamic indices (Additional file 1: Table S1), the inferior RV ΔOS-CMR SI had a moderate-good inverse correlation to RV inferior wall thickness (r = – 0.7, p < 0.001) (Fig. 5b) and moderate inverse correlation to RHC mPAP (r = – 0.4, p = 0.02) (Fig. 5c). This suggests that increased RV hypertrophy is associated with increased myocardial deoxygenation. However, compared to native T1 values, there was no correlation between RV ΔOS-CMR SI and RV native T1 values (r = 0.02, p = 0.91) (Fig. 5D). Furthermore, no correlation between RV ΔOS-CMR SI and RV native T1 values was demonstrated in IPAH and SScPAH subgroups. RV native T1 had a moderate inverse correlation to RV mass/volume ratio (r = – 0.41, p = 0.036). However, there was no other significant correlation between RV native T1 and CMR volumetric/functional or haemodynamic indices.

In PAH, RV myocardial ischemia is acknowledged to be a significant contributor to adverse remodelling and progressive RV dysfunction [4]. Reduced right coronary artery perfusion pressure [24] and microvascular ischemia leads to progressive RV dysfunction causing RV failure and death. Abnormal pressure gradient between the aorta and the RV due to escalating pulmonary vascular resistance, reduces myocardial blood flow in the RV branch of the right coronary artery causing ischemia [7, 25]. RV myocardial ischemia is further exacerbated by inadequate compensatory increase in the myocardial vasculature owing to progressive RV hypertrophy (RVH) [7, 26, 27]. Occlusive microvascular damage and impairment in angiogenesis (known as capillary rarefaction) has been demonstrated in animal models and SScPAH, is thought to cause microvascular ischemia [5, 28]. Furthermore, a study by Meloche et al. proposes common pathobiology mechanisms between PAH pulmonary vasculature and coronary arteries [29]. These pathobiology mechanisms trigger coronary vascular remodeling in both the LV and RV leading to microvascular ischemia.

In this study, we have demonstrated a blunted myocardial oxygenation response to vasodilator stress in the inferior RV segment on PAH patients. Specifically our study demonstrates the presence in-vivo myocardial deoxygenation in the adaptive stage of PAH. The findings of in-vivo RV myocardial deoxygenation in PAH are novel because abnormalities in myocardial blood flow or microvascular dysfunction may not necessarily lead to myocardial deoxygenation [11]. Arnold et al. had demonstrated that in coronary artery disease, 50% of hypoperfused segments on quantitative myocardial perfusion demonstrate no evidence of deoxygenation [11]. Furthermore in dilated cardiomyopathy, Dass et al. had demonstrated disassociation between microvascular dysfunction and oxygenation [30]. Although we had not performed any direct comparison between myocardial blood flow and myocardial oxygenation, these findings provide further mechanistic insight into pathophysiology of myocardial ischemia in the RV in PAH.

The OS-CMR SI changes in the RV had good correlation with global OS-CMR SI changes seen in the LV, an area of myocardium in which there is good validation with the OS-CMR technique. In our study, the OS-CMR technique has demonstrated significant changes in the inferior RV segment. Thinned myocardium (causing partial volume effects), off-resonance artefacts and image artefacts are known limitations of the OS-CMR technique in the RV [18]. However, these limitations have previously been known to affect the accuracy of OS-CMR SI in LV myocardium [31]. Consistent voxel-size was applied during OS-CMR imaging, hence the measured inferior RV wall thickness provided sufficient coverage to avert partial volume effects. Additionally, it has been shown that the inferior RV segment has the greatest end-diastolic wall thickness once the afterload, serum biomarkers and RV structural adaptation in pre- and post-capillary pulmonary hypertension have been taken into consideration [32].

Furthermore, our study has demonstrated a moderate inverse correlation between RV OS-CMR SI and mPAP. The moderate correlation highlights that there are other pathophysiological factors (such as microvascular ischemia) that affect the RV besides elevated mPAP. Patients with microvascular ischemia often present with typical effort-induced angina, but also with atypical symptoms such as dyspnea on exertion [33]. Exertional limitation is the dominant symptom in PAH, however it is contributed by many factors such as respiratory mechanics/ventilation, cardiovascular response and psychological/emotional aspects [34]. Even though RHC resting hemodynamic measures such as mPAP and cardiac output correlate with severity of symptoms, there are considerable inter-individual variability that is not explained by RHC hemodynamic severity [35]. The findings of myocardial deoxygenation provide further mechanistic insights into the pathophysiology of myocardial ischaemia in PAH population.

In PAH, the RV adaptation extends to the cardiomyocytes and interstitium altering the cardiac structure and function [8]. Although the initial adaptive changes of myocardial collagen accumulation serve as a favorable response helping the RV withstand higher pressures, over time maladaptive alteration ensues leading to fibrosis and progressive RV dysfunction [36]. Native CMR T1 parametric mapping has been used as a surrogate for diffuse interstitial fibrosis in the absence of an alternative cause of interstitial expansion (oedema, infiltrations/fibre disarray) [16]. Previous studies of RV native T1 mapping in PAH using Modified Look-Locker Inversion recovery (MOLLI) sequence has produced varying results [17, 22, 37]. One possible explanation is that each of these studies have adapted an altered version of the MOLLI sequence thus producing differing results. Utilising the ShMOLLI sequence, our study demonstrated increased native T1-values in the inferior RV wall compared to healthy controls. The native T1-values in the inferior RV wall were comparable to the LV septum and LV free wall in PAH patients, similar to the findings previously demonstrated by Spruijt et al.[22]. Unexpectedly in our study, we found that the RV native T1-values did not correlate with inferior RV OS-CMR SI changes. This is pertinent in PAH as the pressure overload leads to disruption of the healthy extracellular matrix in the RV due to the excess collagen formation and myocardial interstitial fibrosis. Changes in RV extracellular matrix could potentially disrupt the coronary microvasculature and hence myocardial oxygen supply. However, the lack of association between RV native T1-values and RV OS-CMR SI in our study suggests that the changes in RV extracellular matrix does not directly impact myocardial oxygenation.

Interestingly our study is the first to have demonstrated a significant difference in the native T1-values between the IPAH and SScPAH. We have used the ShMOLLI sequence in contrast to other studies that have used the MOLLI sequence [17, 22, 38]. The ShMOLLI sequence is heart rate independent [39] and as such is able to estimate long T1s which is important for assessing oedematous tissue [40]. Previous ShMOLLI T1 mapping studies of the LV in systemic sclerosis patients suggested the presence of interstitial fibrosis and low-grade inflammation [41]. Therefore, the higher T1-value SScPAH seen in our study could signify a combination of diffuse interstitial fibrosis and myocardial inflammation, a theory that fits with the pathophysiology of systemic sclerosis.

Our study only performed a single mid-ventricular slice on OS-CMR imaging rather than the entire ventricle hence there was incomplete coverage of the full RV myocardium. This is unlikely to be of major relevance in RV assessment of PAH which is a global process. While our study had demonstrated in-vivo myocardial deoxygenation in the RV of PAH, we had not performed any direct comparison with myocardial blood flow, an area for future studies. Late-gadolinium enhancement or extracellular volume assessment was not performed due to concerns of patient safety and tolerability during CMR research scan. While OS-CMR sequence has been studied in many conditions [9], to our knowledge it has not been studied in PAH. The multiple breath-hold during rest/stress OS-CMR image acquisitions could potentially be challenging for PAH patients as dyspnoea is a common symptom in this cohort of patients. The number of subjects in the healthy control groups were relatively small especially with female controls, compared to the PAH group. However the control group were age-matched, minimizing any significant bias. This is important, especially in T1-values whereby age is known to influence native myocardial T1-values [39]. Measuring RV T1 on a curved RV myocardium can be challenging. RV T1 mapping with ShMOLLI sequence at 8 mm slice thickness can be a potential substrate for partial volume effects affecting T1-values. However, every effort was taken to ensure the ROI was drawn in the myocardium. Furthermore the adequate reproducibility and the comparable RV T1-values to the LV T1-values in PAH patients suggest accurate interpretation of RV T1. Although this study was prospective, the PAH patients were stable on pulmonary vasodilator therapy. This could have mitigated against the finding of RV ischemia. A larger study would help to determine the clinical and prognostic utility of these novel CMR techniques in PAH.