Right ventricular volume-strain loops using 3D echocardiography-derived mesh models: proof-of-concept application on patients undergoing different types of open-heart surgery

Abstract

Background: Right ventricular (RV) function can be quantified by right heart catheterization-derived pressure-volume loops. While this technique is invasive, echocardiography-based volume-strain loops (VSLs) potentially reflect a non-invasive alternative. In this study, an approach to generate VSLs from volume and multidimensional strain data of 3D echocardiography-derived RV mesh models is evaluated with regard to feasibility and reproducibility. Methods: In a retrospective cohort study design, 3D intraoperative transesophageal echocardiograms of twenty-three patients undergoing aortic valve surgery (AVS) and eighteen patients undergoing off-pump coronary artery bypass (OPCAB) grafting were available prior to sternotomy and after sternal closure. RV meshes were generated using 3D speckle-tracking. Custom-made software quantified the meshes’ volumes, global longitudinal (RV-GLS) and global circumferential strain (RV-GCS) for VSL generation. Linear regression of systolic VSLs yielded slopes, intercepts and systolic areas. Polynomial regression of two orders was used to analyze systolic-diastolic coupling at 10% increments of the RV end-diastolic volume (RVEDV). Reproducibility was analyzed by fourfold double-measurements of four datasets. Results: VSL calculation was feasible from all included 3D datasets. RV-GLS remained unaltered, but RV-GCS worsened in AVS [abs. diff. (∆) 3.9%, P<0.01] and OPCAB patients (∆4.5%, P<0.001). While RV-GCS systolic areas were markedly reduced at the end of AVS (∆268mL%, P<0.01) and OPCAB (∆185mL%, P<0.001), RV-GCS slopes did not change. Systolic-diastolic uncoupling was not observed, but in trend, decreased diastolic RV-GCS after AVS (P=0.06) and increased diastolic RV-GCS after OPCAB (P=0.06) were observed. Intraclass correlation coefficients (0.84–0.98) and coefficients of variation (6.4–11.8%) indicated good reproducibility. Conclusions: RV VSL generation using 3D echocardiography-derived mesh models is feasible. Longitudinal and circumferential strain vectors yield intrinsically different VSL indices. In future investigations, VSLs of multidimensional strains could provide further insight into periprocedural changes of RV mechanics.