Background
Up to 70% of patients with hypertrophic cardiomyopathy (HCM) exhibit left ventricular (LV) outflow tract (LVOT) obstruction [
1]. LVOT obstruction is frequently associated with significant mitral regurgitation as a result of systolic anterior motion of mitral leaflets [
2]. Less commonly, mitral regurgitation is related to intrinsic valve abnormalities [
2]. LVOT obstruction together with systolic anterior motion-related mitral regurgitation leads to progressive deterioration of clinical status. Significant symptomatic LVOT obstruction requires septal reduction therapy, which may vary depending on the mechanisms of LVOT obstruction, mitral valve disease status and alterations in papillary muscles [
3,
4]. The coexistence of mitral regurgitation may influence the decision between intervention and conservative treatment, as some centers will consider septal reduction therapy in HCM patients with mild symptoms and significant LVOT obstruction provided that they have moderate-to-severe systolic anterior motion related mitral regurgitation [
3]. The mechanism of mitral regurgitation is crucial for deciding between alcohol septal ablation and surgical myectomy since intrinsic mitral valve abnormalities leading to regurgitation can be adequately addressed only by surgery. Additionally, the extent of surgery (myectomy alone or myectomy with adjunctive procedures ranging from anterior mitral leaflet plication to mitral valve replacement) depends on, among other things, the severity of mitral regurgitation [
2‐
4]. Lastly, mitral regurgitation is one of the factors responsible for left atrial enlargement and is an important cause of atrial arrhythmias, particularly atrial fibrillation [
3].
Over the past decade, cardiovascular magnetic resonance (CMR) imaging has emerged as a valuable tool for the non-invasive assessment of patients with HCM [
5,
6]. CMR imaging provides prognostic information and valuable data on cardiac anatomy, mechanisms of LVOT obstruction, and differential diagnoses of LV hypertrophy [
3,
5,
7‐
9]. CMR imaging has also been shown to be an accurate and reproducible method of flow quantification, including valve regurgitation grading [
8,
10‐
13]. In the case of the mitral valve, the mitral regurgitant volume (MRvol) is usually quantified as the difference between LV stroke volume (LVSV) – measured by volumetric assessment – and aortic forward flow (Ao) – measured by phase-contrast (PC) velocity mapping [
12]. However, several factors have been shown to limit the accuracy of the PC technique, including complex flow patterns, as shown in patients with bicuspid aortic valves and aortic valve stenosis [
14,
15]. Similarly, LVOT obstruction in patients with HCM also leads to turbulent flow and flow measurement errors; therefore, it may have important consequences for the calculation of MRvol. These issues, however, have not yet been adequately addressed. We hypothesized that LVOT obstruction causing turbulent, non-laminar flow in the ascending aorta, leads to alterations in Ao measurements and, hence, impacts MRvol quantification in HCM patients with LVOT obstruction. In patients without intra- or extracardiac shunts, net aortic flow is almost equal to pulmonary net flow, and the latter may alternatively be used for the calculation of mitral regurgitation severity [
16]. To verify this hypothesis, we measured MRvol using both aortic and pulmonary flow determined by CMR imaging in patients with HCM with and without LVOT obstruction.
Discussion
The main findings of our study are as follows: 1) in HCM patients with LVOT obstruction, aortic flow was significantly lower than MPA flow and, consequently, MRvol calculated as the difference between LVSV and Ao was significantly higher than MRvol calculated as the difference between LVSV and MPA flow; 2) the exclusion of papillary muscles and trabeculations from the blood pool and the inclusion of them in the LV mass calculation led to significantly lower MRvol than when papillary muscles and trabeculations in the blood pool were included; and 3) using the grading system for the severity of mitral regurgitation based on MRvol, there was substantial discordance between aortic flow-based and pulmonary-flow-based MRvol estimates, and there was no consistency between mitral regurgitation quantification using different methods of LVSV calculation (LVSVincl vs. LVSVexcl).
The most common approach for mitral regurgitation quantification using CMR imaging is based on two parameters: LVSV and Ao (MRvol = LVSV − Ao). Changes in either of these parameters will affect calculated MRvol. We demonstrated that the second parameter of this equation (i.e., Ao) was significantly affected by the presence of LVOT obstrution, leading to underestimated measurements when compared to pulmonary flow-based measurements. This finding has important clinical implications, as the estimated MRvol and mitral regurgitation grades may be higher than the true values. Under normal conditions, net aortic flow and net pulmonary flow should be almost equal, with a slightly greater pulmonary flow (taking into consideration that coronary flow reduces the net aortic flow when measured distal to the sinuses of Valsalva), and the ratio of pulmonary to systemic flow (Qp:Qs) should be close to 1. These calculations performed using CMR data are based on PC images and depend on a variety of factors, including turbulent flow. Previous studies showed that both bicuspid aortic valve and aortic valve stenosis cause turbulent flow in the ascending aorta, leading to flow measurement errors [
14,
15]. LVOT obstruction, which is frequently observed in patients with HCM, is another mechanism leading to exaggerated turbulence in the LVOT, which spreads upstream to the level of the aortic valve and ascending aorta [
27,
28]. Thus, we hypothesized and provided data confirming that LVOT obstruction may result in altered ascending aorta flow, which was found to be lower than pulmonary flow. This finding has important clinical implications regarding discrepancies in MRvol values. In our study, the maximal difference between aortic and pulmonary forward flow reached 47 ml in the extreme case, which translated into a 47-ml overestimation of MRvol and a Qp:Qs ratio of 1.74. Furthermore, the highest Qp:Qs ratio (1.90) was observed in a patient with an aortic net flow that was 41 ml higher than the pulmonary net flow.
The first parameter of the equation mentioned above (MRvol = LVSV − Ao) that is used for the calculation of MRvol is LVSV, which can be calculated in two ways: with or without the inclusion of papillary muscles and trabeculations in the blood pool. There is no consensus regarding the optimal method of ventricular segmentation, although some authors have suggested the superiority of the detailed (
excl) segmentation method [
29,
30]. Discrepancies in LV parameters calculated with the inclusion or exclusion of papillary muscles and trabeculations from the blood volume have been previously demonstrated in various populations, including patients with HCM [
30‐
39]. Our results are in line with those of previous studies demonstrating that the exclusion of papillary muscles from the blood pool decreases LVSV with a resultant MRvol smaller than that obtained with the inclusion of papillary muscles in the blood volume. MRvol calculated with the inclusion of the papillary muscles/trabeculae in the blood pool is particularly overestimated in patients with coarse papillary muscles and trabeculae, such as patients with Fabry disease [
33]. Similarly, patients with HCM exhibit more prominent papillary muscles than healthy subjects [
40]. Among the controls in our study, who had no more than mild mitral regurgitation as assessed using echocardiography, median MRvol calculated with LVSV
incl was approximately 20 ml (up to 35 ml in the extreme case); similarly, Han et al. [
30] demonstrated an average MRvol value of 20 ml (and a maximal value of 40 ml) in HCM patients with no evidence of mitral regurgitation or trivial mitral regurgitation on echocardiography. Therefore, one may state that inclusion of papillary muscles in the blood pool results in overestimation of MRvol (approximately 20 ml) compared to the true MRvol value (absent or trivial). In contrast, MRvol values close to zero calculated by subtracting aortic or pulmonary flow from LVSV
excl are in agreement with the values of MRvol predicted intuitively in patients with up to mild mitral regurgitation. This result supports the notion that the approach with the exclusion of papillary muscles/trabeculations from the blood volume should be preferred for LVSV calculations.
Taken together, our results indicate that alterations in the two necessary elements of MRvol quantification must be taken into consideration when assessing mitral regurgitation in HCM patients. Mitral regurgitation severity in HCM patients is important in clinical decision-making processes. We demonstrated that MRvol differed substantially based on the method of quantification used and that the grading system based on MRvol showed significant discrepancies independent of the thresholds used. We tested the agreement between the expanded five-stage grading system (<15 ml, 15–29 ml, 30–44 ml, 45–59 ml, and ≥60 ml) and the simplified dichotomous categorization (non-severe vs. severe). Regardless of the grading scale used, discordance was observed between the quantification methods used to define the grades. Particularly, discrepancies in grading stages based on LVSVincl vs. LVSVexcl were extremely prominent and resulted in the reclassification of up to 44.4% of patients in the ≥ 60 ml MRvol category to the < 15 ml MRvol category. The total number of patients with MRvol < 15 ml increased by more than 5-fold simply by changing the ventricular segmentation method (from LVSVincl to LVSVexcl), and for patients with MRvol < 30 ml, there was an approximately 2-fold increase. Additionally, the utilization of dichotomous categorization with the most recent threshold of 60 ml as a cut-off value for both primary (degenerative) and secondary (functional) severe mitral regurgitation resulted in the reclassification of a substantial percentage (almost 90%) of patients from severe to non-severe mitral regurgitation. These data explicitly demonstrate that the same thresholds should not be used with different methods of MRvol quantification when grading the severity of mitral regurgitation.
The differences in LV parameters between two ventricular segmentation methods may seem to be larger than those previously published [
30‐
39]. This is attributable to several factors. First, several previous studies considered papillary muscles and trabeculations separately – thus some studies excluded from the blood pool only trabeculations and some excluded only papillary muscles. In contrast, we excluded both papillary muscles and trabeculations from the blood pool, which resulted in more pronounced differences. The study by Chuang et al. revealed similar findings as those of our study when papillary muscles and trabeculations were simultaneously excluded from the LV volume [
32]. Additionally, the papillary muscle and trabeculation volume expressed as a fraction of LVEDV was similar to that in our study (0.23 ± 0.03 vs. 0.25 ± 0.05). The second possible reason is the inverse relationship of age with the size of papillary muscles and trabeculae, as previously shown [
32]. In other words, older patients have less pronounced papillary muscles and trabeculations; consequently, the difference between segmentation methods is smaller (decays) in older patients than in younger patients. Our population was rather young compared to that of some other studies. Different approaches to small trabeculations may also be the source of discrepancies between studies, as trabeculations smaller than 1.5 mm were not considered in some studies [
33,
36,
38,
39]. Finally, interindividual differences in the degree of trabeculation may exist even among healthy subjects, as demonstrated by Chuang et al. in a large, community-dwelling population [
32]. Our control group was rather small, and the vast majority was male. Considering all the above-mentioned factors and the wide normal reference range (e.g., 50th percentile of 32.3 ml and 95th percentile of 45.6 ml) for trabeculation and papillary muscle volume in healthy men [
32], it is possible that some studies included more heavily trabeculated individuals than other studies.
Some of the patients included in our study had small negative MRvol values. Although highly reproducible, both the PC flow measurements and LVSV pose some estimation errors, which may rarely (in cases with no mitral regurgitation) result in slightly negative MRvol values. Small negative MRvol values could also be due to small (ca. 2–3 beats per minute) differences in heart rate during the acquisition of cine and PC data. Finally, background phase errors cannot be entirely discounted, although we provided data that phase errors did not significantly affect the results. It has recently been shown that offset errors are not clinically relevant for the CMR scanner used at our center [
41]. In a study by Meierhofer et al. [
41], baseline offset correction resulted in no improvement of flow measurements based on PC images acquired using an Avanto scanner. Similarly, there were no significant differences in parameters calculated with or without correction in our study. However, both the population in the study by Meierhofer et al. [
41] and the sample cohort in our study were rather small; therefore, these issues should be addressed in future studies.
There were significant differences in age among patients with LVOT obstruction vs. patients without LVOT obstruction and control subjects. However, the main intent of this study was not to compare Ao and MPA flow between different cohorts but to compare those parameters within the given cohort (i.e., within patients with obstructive/non-obstructive HCM or controls). Using MPA flow, we aimed to internally validate the reliability of Ao PC data in patients with turbulent flow due to LVOT obstruction. Considering the main aim of the study, namely, comparison of Ao and MPA flow within given subjects, age differences should not affect the results, since age similarly affects both aortic and pulmonary flow within the same patient. Nevertheless, we performed multivariate analyses to check whether the differences in MRvol among different cohorts (HCM with LVOT obstruction, HCM without LVOT obstruction, and control group) were attributable to baseline differences between groups. By entering the presence of LVOT obstruction, age, gender, and velocity encoding sensitivity into the regression models, we found that the presence of LVOT obstruction (but not age or other factors) was an independent predictor of MRvol (data not shown). Thus, differences in mitral regurgitation severity between obstructive and non-obstructive HCM patients were not explained by variations in age between study cohorts but by the presence of obstruction.
The degree of LVOT obstruction together with the degree of mitral regurgitation vary depending on preload, LV contractility, and afterload [
3,
42]. Moreover, medications, alcohol intake, and prandial status affect the LVOT gradient. Previous studies have shown that fluctuations in the LVOT gradient are inherent characteristics of HCM patients that stem from the dynamic nature of the obstruction [
43‐
45]. Not only day-to-day but also beat-to-beat and respiratory variability in the LVOT gradient have been demonstrated [
43‐
45]. In our study, the mean time between CMR and echocardiography assessment was 34 days; thus, differences in loading conditions between these two time points cannot be excluded. We are aware that it is not possible to control for a myriad of factors that affect the LVOT gradient, but we have tried to control the factors that are controllable. None of the patients underwent invasive procedures or a change in medical treatment within this interval. Nevertheless, wide spontaneous fluctuations in the LVOT gradient make it almost impossible to accomplish the experiments at two different time points with the same degree of LVOT obstruction. Performing two studies (CMR and echocardiography) in 1 day or in succession may not solve this issue. This limitation is common among all studies on HCM patients.
In addition to the above-mentioned issues, our study has several other limitations that should be acknowledged. Muzzarelli et al. [
14] showed that PC acquisitions performed at levels other than the level of the ascending aorta (at the level of the aortic valve or LVOT) may be the solution to inaccurate flow quantification in the ascending aorta. In HCM patients, measurements in the LVOT in the presence of dynamic obstruction would be challenging and possibly erroneous due to the complex flow pattern caused by LVOT obstruction. Vortex formations originating in the LVOT make measurements proximal to flow turbulence impossible. Moreover, measurements at the aortic valve in the case of subvalvular obstruction (LVOT obstruction) would also be prone to artifacts and errors. Finally, the impact of shorter echo times on aortic flow measurements was not tested. However, we aimed to test our hypothesis using standard real-life parameters. Considering the results of our study, further studies should be considered to determine a solution for resolving inaccurate Ao data caused by LVOT obstruction. Using PC sequences with shorter echo times might be one solution [
15]. The potential utility of pulmonary flow in clinical decision-making in patients with mitral regurgitation should be confirmed or refuted in further studies. In particular, a comparison of pulmonary flow-based vs. aortic flow-based mitral regurgitation volumes calculated using flow measurements performed at different levels of the aorta is warranted.