Three-dimensional Feature Tracking Cardiac Magnetic Resonance in Hypertrophic Cardiomyopathy: Comparison with Two-Dimensional Algorithm

Compared to the range values obtained in healthy volunteers¹¹, patients with HCM showed significantly reduced GLS 3D (9.9 ± 3.5 vs. -14.6 ± 2.7, p<0.0001) and GRS 3D (38.7 ± 13.9 vs. 47.4 ± 12.9, p=0.0001), while GCS 3D only showed a trend to reduced values (-16.3 ± 6.6 vs. 17.6 ± 2.6, p=0.08), as seen in Table 3.

Table 4 shows the comparison of CMR characteristics between patients with septal and apical HCM as the two most common subtypes in our cohort. There was no difference regarding age or BMI between the two HCM phenotypes. In terms of standard CMR parameters, patients with apical HCM had a higher number of segments with LVH (5.3 ± 0.7 vs. 3.6 ± 2.2, p=0.006) but were less likely to exhibit a spiral pattern of LVH (6.7 vs. 35.3%, p=0.04). A high proportion of patients in both groups was found to have focal fibrosis (60 vs. 58.8%, p=0.97), but there was no difference regarding the number of segments with LGE or the presence of LGE in segments with LVH (66.7 vs. 55.9%, p=0.55). Right and left ventricular volumetric measurements were similar between groups. Both groups showed borderline LA dilatation but without significant difference according to phenotype (30 [26.3-30.1] vs. 28.5 [25-32] cm², p=0.72). Regarding FT parameters, those with apical HCM only had significantly more reduced GLS 3D compared to patients with septal LVH (-7.4% ± 4.1 vs. -10.5% ± 2.9, p=0.004), while there was no difference in terms of GCS 3D (-18.4% ± 2.7 vs. -15.1% ± 8.3, p=0.14), GRS 3D (38.8% ± 15.9 vs. 39.8% ± 12.3, p=0.8), and 2D parameters.

We demonstrated that myocardial strain evaluated using 3D FT-CMR was consistently lower compared to 2D FT in patients with HCM. This is in line with previously published data in 100 healthy volunteers using the same software as in our study¹¹, and replicating the results of comparisons between 2D and 3D STE^12,13. In patients with HCM, characterized by abnormal geometry from asymmetric LVH, the observed differences may be attributed to variations in tracking algorithms as well as to the assessment of myocardial deformation across multiple dimensions provided by 3D FTI.

We found that the highest level of agreement was achieved between 2D and 3D GCS (mean of differences of -0.3%). It is worth mentioning that although there was a statistically significant difference between 2D and 3D GLS strain, a difference of -2.7% may represent an acceptable level of agreement in clinical practice. Our results arein accordance with the majority of studies, which demonstrate that GCS measurements provide the highest level of agreement between 2D and 3D methods and greatest reproducibility. For instance, a previous study that evaluated the agreement and reproducibility of 3D FT-CMR in healthy volunteers demonstrated that 3D FT-CMR exhibited superior reproducibility as compared to 2D FT-CMR, with the highest degree of reproducibility observed for GCS¹¹. In contrast, another study that evaluated STE in HCM found that there was only modest agreement between 2D and 3D GLS and GCS values¹³. Nevertheless, this may be associated with the limitations of 3D echocardiography.

Both the better reproducibility and the lower values obtained with 3D FT-CMR can probably be linked to the through-plane loss of features, which limits 2D strain analysis. This phenomenon occurs mainly during LV twist, where the out-of-plane myocardial motion accentuates the observed magnitude of contraction, leading to an inaccurate estimation of deformation. 3D FT-CMR has the potential to overcome this limitation and potentially provide more accurate results for deformation assessment. Nevertheless, while improved reproducibility is promising, further investigations are required to assess whether it provides supplementary clinical information in comparison to 2D strain.

In accordance with our research, a recent meta-analysis including multiple vendors conducted on 659 healthy subjects demonstrated that GCS was the most consistent FT-CMR parameter, whereas the highest inconsistency was observed for GRS¹⁴. This large variation regarding GRS measurements has been identified in most of the studies, including those incorporating other techniques to assess deformation, such as myocardial tagging^11,15,16. This observation is most likely attributable to the radial direction being the most sensitive parameter for myocardial tracking of LV twist.

A study evaluating the inter-vendor agreement of FT-CMR in healthy volunteers, using two different software, reported that GCS was the most consistent parameter as well. Conversely, the measurements obtained with the same software utilized in our study were significantly lower than those obtained from the alternative software, suggesting limited interchangeability between vendors¹⁷. The highest variability was observed in radial strain across all techniques in an investigation that sought to determine strain values in patients with cardiomyopathies and healthy controls using tagged imaging as a reference method for multiple techniques and vendors¹⁸. This suggests that reference values should be defined for every technique and software product. Overall, the major drawback of other techniques in the quantification of deformation is the need for acquiring additional sequences, which results in a prolonged scanning time. The post-processing technology used for FT-CMR eliminates the necessity of supplementary scanning time, hence conferring an important advantage to this technique¹⁹.

Notably, both 2D and 3D FTI were feasible in all patients in our study as image quality was optimal, highlighting their potential utility in cases with suboptimal echocardiographic windows. Augustine et al. likewise observed that FT has a significant ability to track up to 90% of segments¹⁶. Hence, considering that STE allows for the assessment of strain in only approximately 80% of the segments as a result of inadequate image quality in some regions^20,21, FT provides an ideal alternative. Additionally, although 3D STE has been proposed as a solution to address the limitations of 2D imaging, such as out-of-plane motion, it comes with the drawback of a reduced temporal and spatial resolution, which could influence the overall accuracy¹³. A significant benefit of FT-CMR is its ability to efficiently track the segments, even in patients who have limited acoustic windows. Furthermore, it is well-established that individuals with HCM can show abnormal myocardial geometry, characterized by hypertrophy across multiple segments. This might pose challenges for STE and the use of FT could prove particularly advantageous in these patients. Implementing 3D FTI does not require extending the scanning time beyond the standard CMR protocol, and the analysis of 3D FTI can be efficiently completed in a reasonable timeframe.

In our study, patients with HCM exhibited lower values of GLS 3D and GRS 3D compared to the normal range values obtained from a study that evaluated healthy volunteers using the same software¹¹. However, there was no significant difference in GCS 3D values between HCM and healthy volunteers¹¹. This is consistent with another study using STE, which also observed a greater reduction in GLS values in HCM patients compared to controls²². The reduced longitudinal deformation in HCM could be attributed to the vulnerability of the subendocardial layer to perfusion abnormalities and its higher susceptibility to interstitial fibrosis²³. In general, studies using CMR consistently show a decrease in GLS in individuals with HCM, but the results regarding GCS and GRS are variable^8-10. As an example, a study using STE in HCM and examining the response to exercise reported that HCM patients had reduced GLS and GRS at rest, but higher GCS compared to the control group²³. Overall, the absence of a decrease in circumferential deformation in HCM, whether it results in an increase or maintenance of normal levels, may indicate a compensation mechanism that assists in preserving systolic function in HCM. The variations observed between studies are likely related to the use of different imaging methodologies, as well as the inclusion of heterogeneous populations. In terms of direct comparisons between normal FTI and STE values, evaluation is currently constrained by the limited availability of data; however, cvi42 FT data appears to report a trend toward lower normal values in 3D and 2D GLS (-14.6% vs. -19.4% with 4D AutoLVQ, and -19.8% vs. -22.5%), 3D and 2D GCS (-17.6% vs. -19.47% 4D with AutoLVQ and -20.9% vs. -31.9%) and 3D GRS compared to STE (47.4% vs. 50.41% 4D with AutoLVQ), but higher 2D GRS values (47.6% vs. 37.4%), suggesting differences in the assessment of myocardial strain between the two techniques^11,24,25.

Septal and apical HCM are widely recognized as the most common phenotypes of HCM. Our investigation found that although there was no variation in LGE burden between the two phenotypes, apical HCM patients had a significantly larger reduction in GLS 3D, while there was no major difference in terms of GCS 3D and GRS 3D. This may be attributed to the higher proportion of segments with LVH in patients with apical HCM. These segments are likely to exhibit greater impairment in longitudinal deformation, leading to a decrease in the overall global value.

There has been evidence that abnormal GLS is associated with adverse cardiac events in patients with HCM¹. A large study including 2311 individuals with HCM sought to identify factors that can predict abnormal 3D GLS, revealing that several CMR parameters were associated with greater impairment in longitudinal deformation. These parameters include a higher LV mass index and MWT, lower LVEF and RVEF, and a higher extent of LGE²⁶. Consistent with our findings, the same study demonstrated that 3D GLS was better preserved in patients with septal LVH. However, 3D GRS was better preserved in patients with apical HCM.

Previous research suggests that GLS measured by STE can be useful in predicting outcomes in HCM, including sudden cardiac death and all-cause mortality^27,28. Other studies found an association between strain, LGE and native T1, but not ECV^26,29,30, demonstrating that the pathophysiology of HCM involves intricate mechanisms. Notably, myocardial disarray in HCM can also be present in segments that are not hypertrophied, as reported in previous studies using histopathology^31,32. Hence, assessing strain to determine abnormal myocardial contractile function in HCM may offer valuable information about subtle changes potentially associated with higher risk in HCM. Additionally, this could result in an improved understanding of morphological characteristics in HCM, which might aid in distinguishing different hypertrophic phenotypes. Finally, deformation imaging is less affected by loading conditions than LVEF. Hence, deformation imaging has the potential to assist clinical decisions, particularly in individuals with apparently preserved systolic function.

The main limitation of this study is that results were calculated using software from a single vendor, and CMR FT may be subject to variations between different vendors. Another limitation is that the analysis was performed by a single observer, and reproducibility was not tested. This may introduce observer-specific biases and limit the generalisability of the results. Future studies should assess intra- and interobserver variability to enhance the reliability and validity of 3D FTI measurements. The absence of T1 mapping which could have enabled the assessment of interstitial fibrosis in addition to LGE. Furthermore, although the clinical value of 2D strain has been previously demonstrated, it is necessary to collect additional prospective data and include a larger cohort of patients with HCM to assess the predictive value of 3D strain. A possible limitation of this technique is the limited data on its association with major adverse cardiovascular events. Future research should focus on evaluating the potential association of CMR FT with major adverse cardiovascular events in patients with HCM, to to further establish the predictive value of this method within this patient population.