Myocardial Strain Analysis With CMR in Breast Cancer Patients with Iatrogenic Cardiotoxicity Using Heart Deformation Analysis: Comparison to DENSE
Abraham Bogachkov1, Kai Lin2, Ahmadreza Ghasemiesfe2, Amir Ali Rahsepar3, Bruce Spottiswoode3, Ben Freed4, Michael Markl2, James Carr2, and Jeremy Collins2

1Northwestern University Feinberg School of Medicine, Chicago, IL, United States, 2Radiology, Northwestern University, Chicago, IL, United States, 3Cardiovascular MR R&D, Siemens Healthcare, Chicago, IL, United States, 4Cardiology, Northwestern University, Chicago, IL, United States

Synopsis

Strain imaging at cardiac MR has been shown to be a powerful tool in the pre-clinical detection of early cardiac dysfunction in the heart failure population, but has been only minimally studied in cardiotoxicity patients. This study evaluated a semi-automatic heart deformation analysis (HDA) tool in the assessment of left ventricular myocardial strain in patients with known cardiotoxicity, and found very good to excellent agreement with global strain values calculated using displacement encoding with stimulated echoes (DENSE). HDA analysis of conventional cine sequences has the potential to play a significant role in the evaluation of patients at risk for cardiotoxicity.

Background

Cardiac function measures in patients with cardiotoxicity are of prime importance as a gauge of disease severity and progression. Strain imaging at cardiac MR (CMR) has been shown to be a powerful tool in the pre-clinical detection of early cardiac dysfunction in the heart failure population, but has been only minimally studied in the assessment of cardiotoxicity patients. Displacement encoding with stimulated echoes (DENSE) CMR imaging encodes tissue motion into the image phase, and has been previously validated for the evaluation of regional myocardial strain with good spatial and temporal resolution and low intraobserver and interobserver variability. Although DENSE is an effective method of evaluating myocardial strain, downsides of this approach include the requirement for dedicated acquisitions and time-intensive post-processing. In contrast, heart deformation analysis (HDA) uses a deformable image registration (DIR) algorithm to track myocardial motion on balanced steady state free precession (bSSFP) cine images. Using an inverse consistent DIR algorithm to recover deformation fields, this technique tracks variations in myocardial deformation over time, generating myocardial strain values. The purpose of this study is to compare radial and circumferential LV strain values calculated via DENSE and HDA in patients with known cardiotoxicity from breast-cancer chemotherapeutic agents.

Methods

The study cohort comprised 7 female prospectively-recruited breast cancer patients (avg. age 54.6±14.3). CMR imaging was performed at 1.5 T (MAGNETOM Avanto, Siemens Medical Systems, Erlangen, AG) using GRAPPA factor 2 acceleration. DENSE myocardial strain analysis was performed using CIM heart deformation post-processing prototype software (Auckland MRI Research Group, University of Auckland, Auckland, NZ), and HDA analysis was done using semi-automatic prototype software calculating Lagrangian strain from deformation field analysis (Siemens Corp, Corporate Technology, Princeton, NJ). Left ventricular midwall global peak and average segmental peak systolic radial and circumferential strains were calculated from short-axis acquisitions for the base, mid-chamber, and apex. Strain data between groups was compared using univariate analysis of variance (ANOVA) to assess statistical equivalence, and the two methods were compared using linear regression with associated R2 values.

Results

Strain analysis was successful in all subjects. No statistically significant difference (p < 0.05) was noted between mean global peak or mean average-segmental peak strains between HDA and DENSE (Figure 1). The R2 values for linear regression of global peak radial and circumferential strain values of HDA vs. DENSE were 0.48 and 0.63, and 0.77 and 0.65 for average-segmental peak radial and circumferential strain values, respectively (Figure 2).

Discussion

This preliminary study validates the use of a semi-automatic prototype HDA strain analysis tool in the assessment of myocardial strain in cardiotoxicity patients. Specifically, mean peak global and average segmental peak strain values generated with HDA analysis were found to be statistically equivalent compared to DENSE derived strain data. Furthermore, regression analysis of peak global and average segmental peak strain values indicated a strong correlation between these two methods of evaluating myocardial strain. Given the possibility that early changes in myocardial strain may precede global changes in cardiac function, strain analysis is positioned to play a significant role in evaluating patients for early signs of cardiotoxicity.

Conclusion

Use of a semi-automatic prototype HDA strain analysis tool in the assessment of myocardial strain yielded statistically equivalent results compared to DENSE derived strain data. Work is ongoing to elucidate a more defined clinical role for CMR in the evaluation of patients at risk for myocardial injury from cardiotoxic medications.

Acknowledgements

No acknowledgement found.

References

• Guetter, Christoph, et al. "Efficient symmetric and inverse-consistent deformable registration through interleaved optimization." Biomedical Imaging: From Nano to Macro, 2011 IEEE International Symposium on. IEEE, 2011.

• Kuijer, Joost, et al. "DENSE and HARP: Two views on the same technique of phase-based strain imaging." Journal of Magnetic Resonance Imaging 24.6 (2006): 1432-1438.

• Collins, Jeremy D. "Global and Regional Functional Assessment of Ischemic Heart Disease with Cardiac MR Imaging." Radiologic Clinics of North America 53.2 (2015): 369-395.

• Zhong J, Yu X. “Strain and torsion quantification in mouse hearts under dobutamine stimulation using 2D multiphase MR DENSE.” Magn Reson Med 2010;64(5):1315–22.

• Smith, Peter M., et al. "Myocardial strain analysis in patients with Heart Failure with preserved Ejection Fraction using bright blood cine MR images: A comparison with speckle-tracking echocardiography." Journal of Cardiovascular Magnetic Resonance 16.1 (2014): 1-2.

Figures

Figure 1: A – mean global peak and mean average segmental peak radial strain values; B – mean global peak and mean average segmental peak circumferential strain values. No statistically significant (p < 0.05) differences between groups were noted.

Figure 2: A,B – linear regression of peak global and average segmental peak radial strain values using HDA vs. DENSE; C,D – linear regression of peak global and average segmental peak circumferential strain values using HDA vs. DENSE.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3125