Introduction
Over the past few years, the use of cardiac magnetic resonance (CMR) imaging has steadily increased in patients with hypertrophic cardiomyopathy (HCM). Although CMR is not yet part of the current risk stratification [
1], several studies have been undertaken to address the impact of CMR for future patient management [
2,
3]. In recent publications it was demonstrated that left ventricular (LV) ejection fraction (EF) and LV mass can be used to identify HCM patients at high risk of adverse cardiac events [
2,
4‐
6]. CMR is the current gold standard for quantification of these parameters, because of its superior accuracy and reproducibility [
7].
In daily practice, uniformity in CMR image analysis is lacking on how to deal with measurements of the papillary muscles, despite recommendations in the current CMR guidelines. The guidelines state that LV volumes and LV mass should be quantified according to the same protocol as used for the reference ranges [
8,
9]. In general, studies on normal values of LV parameters used to include the papillary muscles in the LV mass [
10‐
12]. Surprisingly, in most general hospitals exclusion of the papillary muscles has become the standard [
13].
In HCM patients the papillary muscle mass is higher than in normal healthy volunteers [
14]. Therefore, especially in the HCM population, the impact of the papillary muscles on quantification of LV parameters might be substantial. We studied the impact of inclusion and exclusion of the papillary muscles on the quantification of LVEF and LV mass in clinical HCM patients with overt hypertrophy. In addition, we studied the impact in a group of subclinical HCM mutation carriers without hypertrophy.
Discussion
To our knowledge this is the first study that addresses the basic question of how much inclusion or exclusion of the PMs affects everyday CMR image analysis in patients with HCM in particular, but also in relation to a group of controls without hypertrophy. In HCM patients, inclusion of the PMs resulted in significant relative increases of 9 % in LV mass and 4 % in LVEF. Despite the involvement of the PMs in the disease process of HCM, the relative differences were in the same order of magnitude in the control subjects without hypertrophic myocardium. This underscores the general need for uniform protocols, given the potential impact for clinical decision-making based upon values of LV mass, volume and function.
In recent studies on different CMR image analysis protocols, the PM mass was combined with that of the trabecularisations (Fig.
1c). Inclusion or exclusion of both the PMs and the trabecularisations as part of the LV mass led to significant differences in LVEF and LV mass [
20‐
22]. This was confirmed in HCM patients [
13,
23]. Reports on normal values with CMR imaging are limited and refer to studies that considered the PMs and trabecularisations to be part of the LV mass [
10‐
12]. In daily practice, however, these structures are often not included, resulting in underestimation of LVEF and LV mass. Normal values for this more practical approach of CMR image analysis are scarce. Our study points out the need for either uniform CMR image analysis including these structures, or the need for studies on normal values of the more practical approach.
This is illustrated when our results are put in the context of the excellent reproducibility of CMR image analysis, with a coefficient of variation of 4 % [
13,
17,
20,
24]. Based on the coefficient of variation we can make statistical inferences about the expected results when a particular LV mass is measured repeatedly, provided the same methodology of CMR image analysis is followed. Appreciating the 4 % coefficient of variation, the relative differences of the repeated measurements of that particular LV mass will lie between − 8 % and + 8 % in 95 % of cases. In a setting of uniform CMR analysis, this means that a true change in LV mass will be present if the results differ by more than a relative 8 %.
Our finding that the two different approaches of CMR analysis result in a 9 % relative difference in LV mass should be interpreted in this context. Given the mean relative difference of 9 %, the exact same LV mass will seem to have changed in at least half of the cases. This indicates that non-standardised assessment of LV mass may have important implications for both research and daily clinical practice. As for LVEF, the relative difference of 4 % seems less important.
In the setting of clinical trials, for example, there is a growing interest in studies reporting LV mass and the extent of fibrosis relative to LV mass [
25,
26]. In case of CMR follow-up studies on this subject, CMR image analysis should be performed according to the same protocol at baseline and follow-up. Otherwise, differences in LV mass can be interpreted as a true change, while in fact it could be the result of two CMRs analysed according to two different protocols.
As for daily clinical practice, the use of CMR in HCM patients has increased over the years, even though it is not yet part of the routine clinical work-up for risk stratification [
1]. This development is partly related to its excellent reproducibility, but may also be influenced by the growing body of evidence reported by CMR studies in the field of HCM. For example, increased LV mass assessed by CMR (i.e. > 91 or > 69 g/m
2 for men and women respectively) has been suggested as a more sensitive marker of adverse outcome than LV maximal wall thickness [
6]. A cut-off value (< 50 %) has been reported to predict an adverse prognosis in HCM, not only for LV mass but also for LVEF [
4,
5]. In the latter case, the 4 % difference we observed could have implications for patients with values near the cut-off. In that regard, it has previously been demonstrated that depending on the imaging technique, differences in LVEF assessment may have important clinical consequences in ICD candidates [
27]. In addition, the amount of fibrosis relative to LV mass has the potential to become an additional risk factor in the selection of candidates for primary prevention with an ICD [
2,
3]. Again, this underscores the importance of uniform assessment of LV mass.
In order to achieve uniformity, a guideline statement with regard to the preferred method of analysis should be a first initiative. In the above-mentioned context of increasing use of CMR data, the recent initiative to validate the accuracy of SSFP CMR imaging should be appreciated [
28]. Importantly, these figures on accuracy were obtained with inclusion of both the PMs and the trabecularisations. On the other hand, algorithms incorporating quantification of trabecularisations have been questioned with regard to reproducibility, given the possible overestimation of LV mass due to partial volume artefacts [
17,
29]. If exclusion of the PMs were to become the method of choice, normal values obtained with contemporary sequences in an adequately sized series of healthy controls would be required [
30]. In the present era of non-uniform CMR analysis, reliable comparison between previous and future studies is difficult. This underscores the need for investigators to report the CMR analysis algorithm, and clinicians should be aware of the protocol used at their centre when interpreting CMR parameters.
In the present comparison of the two algorithms, the associated absolute differences in LV mass were higher in clinical HCM patients than in subjects who were HCM mutation carriers without hypertrophy. On the other hand, the relative changes in LVEF and LV mass were similar in both groups. This can be explained by the fact that in HCM patients the proportion of the PM mass in relation to the total LV mass was similar to that in subjects without hypertrophy. Subsequently, after transformation of the absolute differences to relative differences, the impact of inclusion or exclusion of the PMs was comparable in both groups. This suggests that our results apply not only to patients with overt hypertrophy, but may apply to the general population as well, although more study subjects are necessary to substantiate this. This is supported by a study in healthy controls, which reported a similar impact of the PMs on LV mass [
31].
It should be appreciated that this study did not address accuracy, due to lack of a comparative standard. We merely described the impact of a more practical CMR algorithm in terms of quantification of LV parameters, and with regard to potential consequences for research questions and daily clinical practice. In order to put our findings into context, we referred to the well-accepted 4 % coefficient of variation, i.e. a relative difference of 8 % or more indicates a true change [
13,
17,
20,
24]. In a random set of 20 of our participants, intraobserver and interobserver variability were 4.69 and 4.49 % for LVEF and 4.76 and 4.86 % for LV mass, respectively.
In summary, our findings indicate that non-uniform CMR image analysis will render incorrect conclusions with regard to the presence or absence of changes in important LV parameters in about half of patients with two CMR assessments. This holds true not only for HCM patients with overt hypertrophy but also for subjects without hypertrophy. Given the potential impact for both research and daily clinical practice, our data underscore the importance of a standardised approach, either with or without the PMs as part of the LV mass.