1356872 - Diastolic dysfunction evaluation by CMR-derived E, A, e’ and pulmonary vein flow

Background: Diastolic dysfunction (DD), reflecting impaired filling of the LV in diastole due to myocardial stiffness, and subsequent compensatory increases in LV end-diastolic pressure, is an important cause of heart-failure, with  equal morbidity and mortality compared to systolic heart failure (1). Despite this, while echocardiography (ECHO) is commonly used to assess DD, CMR is not in use for the clinical evaluation of diastolic dysfunction. Several phase contrast (PC) CMR studies have categorized DD, using phase-contrast for mitral inflow and annular velocities (2). Our recent approach (3) used 1) feature tracking for measurement of the mitral annular velocities (e’), replacing 2D PC. In this study, we used fully automated measurement of mitral velocity (4) and included pulmonary vein (PV) flow evaluation. PV flow by ECHO is considered valuable but is limited by its far-field location.

Methods: In this prospective IRB approved study, 24 patients with recent ECHO exam (< 3 months, 46±31days) and 12 controls underwent a standard CMR exam complemented by a 4 chamber in-plane PC with flow-encoding in the LV long axis direction and a through-plane right inferior pulmonary vein PC. Image analysis was performed using the freely available software Segment (http://segment.heiberg.se).
Mitral valve flow maximal velocities E, deceleration time (DT), and A were evaluated from a circular ROI in 4ch PC images; e’ was automatically evaluated by a machine learning plugin (MVnet) on standard 4ch cine images (4). The phasic PV maximum velocities were measured (S2, D (and DT), Ar, and Ar reversal time), from automatically propagated PV segmentation (Fig. 1).

Results: Eight patients had abnormal findings by ECHO (n=2 Grade 1, Grade 2, abnormal DD and indeterminate). Table 1 shows that almost all CMR exhibited significant differences in patients with and without DD. There were strong correlations between ECHO and CMR (r >0.6) for average e’, A, E/A and E/e’, and the biases were low (Tab. 1, Fig. 2), except for lower CMR e’ velocities; also CMR LA volumes were higher, as previously reported (5). e’ is known to correlate with age, and CMR-derived e’ correlation to age was stronger compared to ECHO (ECHO: r=-.62, p=.0011, CMR: r=-.75, p< .0001). When analyzing only CMR/ECHO studies occurring very close in time (< 45days, n=12), associations between CMR and ECHO parameters became stronger (r=.80, .88, .78 for A, E/A and e’ respectively vs. r=.76, .70, .65). PV indices correlated with ECHO diastology; D DT to E DT (.59, p = .0029) and D with E (.62,p=.0021), A (.53,p=0071) E/A (.51, p =.0117) and e’ (.50, p=.0135) respectively. S2/D decreased with E/A (r=-65,p=.0001) (Fig. 2). ROC analysis showed that CMR (AUC >.76) and ECHO (AUC >.79) e’, E and E/A differentiated DD, as well as PV diastolic D peak (AUC=.78) (Fig.2).

Conclusion: Diastolic function parameters can be measured by CMR. Phase contrast PV flow parameters differentiated patients with DD from normal patients.