To our knowledge, this is the first comprehensive study comparing conventional and new RV function parameters derived from CMR to their echocardiographic equivalents in patients with pulmonary hypertension. Our results indicate strong to moderate correlation of CMR measures of RV function to corresponding echocardiographic parities. However, there are biases that needs to be addressed. As such, lowest relative bias was shown for FAC, then FWS, S’ and highest was for TAPSE/AVPDlat. In comparison with RVEFCMR, a strong correlation was demonstrated for FACCMR and FWSecho, whilst a moderate correlation was demonstrated for FACecho, AVPDlat, FWSCMR and TAPSE, respectively. Only a weak correlation was demonstrated for S’ measured by both modalities.
Comparison between corresponding echocardiographic and CMR measurements
Our results demonstrated a moderate correlation for AVPD
lat/TAPSE. These are measures of longitudinal contraction, and the longitudinal contribution accounts for approximately 80% of the RV stroke volume in normal hearts [
4]. TAPSE is widely used for echocardiographic assessment of RV function [
3], however the value as a prognostic marker in pulmonary hypertension has been debated and has been withdrawn as a risk marker [
28]. As the amplitude of TAPSE can be caused by a passive translation movement (i.e. pseudo-normalisation), it can be questioned as a marker of true RV systolic function in pulmonary hypertension [
29]. Other reasons to defer using TAPSE as a risk marker in the guidelines is the angle dependency. AVPD
lat is a relatively new and sparsely evaluated method, and not referenced in consensus documents for CMR [
22]. We found AVPD
lat to be decreased in patients with pulmonary hypertension in alignment with a prior study [
15,
17], however, the prognostic value of altered AVPD
lat in pulmonary hypertension remains to be investigated.
In our study, the highest bias was demonstrated for TAPSE/AVPD
lat with lower absolute values by CMR compared to echocardiography. This is an unexpected finding since AVPD
lat is considered to be less angle dependent than the measurements of TAPSE. Consequently, the bias is probably methodological since they are measured from differently obtained 4-chamber views. In CMR, acquisition of the 4-chamber view is planned from the short-axis view to perfectly rotated into the RV focused view [
3]. These adjustments are not possible with echocardiography. Furthermore, the differences in temporal resolution (i.e. framerate is higher using echocardiography) could explain some of the bias between modalities.
S’ showed a moderate correlation between echocardiography and CMR. S’
CMR has only been described twice previously [
16,
23] and this is the first study to compare S’
CMR to the echocardiographic equivalent. The disadvantage of S’
echo not being representative for RV global function is substantial since it only measures the velocity in one specific point in the basal segment of the RV lateral wall [
3]. Both S’
echo and TAPSE have in previous studies been presented within normality in pulmonary hypertension even when other RV parameters were impaired [
10,
11]. Methodologically, S’
CMR is derived from the AVPD curve and is defined as the highest measured instantaneous velocity measured as the steepest systolic slope of the AVPD curve. The velocity at peak emptying (S’
CMR) could hence be considered as in parity to the peak systolic annular velocity from echocardiography S’
echo [
16]. S’
CMR diminishes some of the limitations for S’
echo such as angle-dependency and image quality. On the other hand, S’
CMR is acquired with a lower framerate than S’
echo and with a subsequent post processed smoothing of the curve. In comparison, S’
echo is derived from pulsed-wave tissue-doppler imaging and with a high frame rate. The two methods cannot be considered methodologically fully interchangeable. In our study, S’ showed the best agreement with narrow limits in absolute values between the methods. However, as with the other methods there was a systematic substantial relative bias and broad limits of agreement.
The strongest correlation between the modalities were shown by FAC and to our knowledge this is the first study to compare the two modalities. FAC
echo reflects both radial and longitudinal contribution to the overall RV function [
3] and predicts both morbidity and mortality [
30] in patients with pulmonary hypertension. Compared to echocardiographic assessment, delineation of the heavily trabeculated compact endocardial border is less challenging by CMR. One could argue that even if FAC is a surrogate for RVEF and there seems to be little point in measuring FAC by CMR. However, if RVEF is not applicable from CMR images owing to substantial artefacts or missing slices, FAC appears to be a good alternative to RV functional assessment.
In this study, FWS showed a moderate correlation between modalities. Prior studies have shown conflicting evidence as some are in concordance with ours [
31], while others have found a stronger correlation [
32,
33]. Strain values were on average 20% higher when measured by CMR with a greater spread of observed values (Fig.
3). The bias (absolute and relative) seems to be more pronounced in patients with lower strain values. There are several methodological reasons for strain values to differ between the modalities [
34]. Theoretically, CMR strain assessment is advantageous to echocardiography as it is not vulnerable to poor acoustic windows [
14]. However, this explanation for the bias in our study is unlikely since patients with echocardiographic poor acoustic window were excluded. One other reason could be that the software’s are using different kinds of strain, i.e. CMR measures endocardial strain and the echocardiographic software uses mid-myocardial strain. The results in our study implies that echocardiographic and CMR evaluation of RV strain cannot be regarded as interchangeable.
Comparison with RVEFCMR
While RVEF is invaluable in assessment of pulmonary hypertension, it is not a direct measure of RV contractility, RVEF requires augmentation with other global and regional parameters to assess the function on a myocardial level [
35,
36]. This forms a rationale for investigating the correlation between RVEF
CMR and other measurements parameters.
Our results are in concordance with previous studies on FWS
echo and FAC
echo/CMR [
11,
17] presenting a strong correlation with RVEF
CMR with a modest correlation for TAPSE/AVPD
lat and a weak correlation for S’
echo compared to RVEF
CMR [
11,
17]. As expected, the parameters measuring of area and volume change are closely related. In alignment with our study others have demonstrated a moderate correlation of RVEF
CMR to AVPD
lat [
17] and FWS
CMR [
31,
37]. The poor correlation between longitudinal measurements of RV function and RVEF
CMR could be explained that RVEF is more related to fractional transverse movements than to longitudinal movements in patients with pulmonary hypertension [
38]. This could also explain the strong correlation demonstrated between FAC and RVEF
CMR.
To our knowledge, this is the first study to evaluate the poor relationship of RVEFCMR to S’CMR, and to explore a comprehensive paired comparison of equivalent parameters of echocardiography and CMR.
Clinical implications
RV function is an important determinant of outcome in patients with pulmonary hypertension [
1,
2] and RVEF
CMR is a robust and precise method for risk assessment and clinical follow up. However, RVEF
CMR is an incomplete method of assessing myocardial contraction as it only evaluates relative volume changes. Several studies have shown regional functional assessment of the RV has incremental value beyond EF [
11,
15,
19]. Longitudinal function has been proven of relevance for outcome irrespectively and independently of EF concerning the left ventricle [
35,
39,
40]. Regarding RV, limited prognostic data is available at present. At present in clinical practice FAC
echo, TAPSE, S’
echo and FWS
echo [
3], and to some extent AVPD
lat are used for assessment of systolic function although their prognostic values are not fully explored. FAC
CMR, FWS
CMR and S’
CMR are new measures that are not yet implemented in clinical practice. To some extent the echocardiographic parameters have been showed to be associated to clinical outcome in pulmonary hypertension [
3,
41], while the CMR equivalents are yet to be assessed, although data is accumulating [
35,
36].
According to clinical practice in our center, the majority of patients evaluated for pulmonary hypertension are investigated within CMR of two main reasons: to exclude presence of congenital heart disease (including intracardial shunts) and also to have accurate measurements for RVEF and RV size since CMR is considered as gold standard for this purpose and these measures carry prognostic information [
42,
43]. CMR are included in the current guidelines [
20] when assessing the RV and our center is following the guidelines in that context. Our data implies that the utilization of both imaging modalities of regional measures can helpfully augment global assessment of RV function by RVEF
CMR, but the equivalent measures are not interchangeable at least in pulmonary hypertension. Clinical outcome data was not within the scope of this study. However, investigating which of the different RV functional markers in the present study is most predictive of outcome is of interest for future studies.
Limitations
There are some limitations in our study. This is a rather small study of 55 selected pulmonary hypertension patients since 35 were excluded due to atrial fibrillation (
n = 15) and poor acoustic windows (
n = 20). However, strain analysis on atrial fibrillation patients are at present not recommended and RV assessment by echocardiography is highly dependent on image quality. Therefore, one could debate which imaging modality is preferred for assessing RV function in this highly selected patient population. Echocardiography and CMR were performed contemporaneously, but not immediately so. Since RV function parameters are load-dependent, small changes in treatment (e.g. diuretic) may affect RV function parameters. However, median time was only 1 day with no cardiac events, medical changes, or clinical deterioration between examinations. FWS
echo was measured using an algorithm developed for the left ventricle. However, this method is standard procedure and has been propagated into guidelines [
3,
44]. On the other hand, FWS
CMR was assessed in RV dedicated algorithm. Further standardization between different vendors for both CMR and echocardiography are required [
34] as well as further studies before their utilization in routine clinical practice.