Computer-based analysis to optimize prosthesis sizing during aortic valve surgery

The choice of the best prosthesis size during aortic valve sparing (AVS) and replacement (AVR) procedures is still debated. Postoperative performances either in case of AVR using valve prosthesis (especially stentless bioprosthesis) or in case of AVS using Dacron graft, are indeed strongly related to an appropriate choice of prosthesis size. We have recently developed a patient-specific approach to optimize prosthesis sizing in such situations using Finite Element Analysis (FEA) to virtually simulate post-operative valve behavior following surgery. The finite element method is a numerical technique for finding approximate solutions of partial differential equations introduced by Hrennikof and Courant during the 2nd World War and then developed by Zienkiewicz and Taylor [1]. Since the late 90s FEA has been also applied in cardiovascular medicine in the attempt to better understand the mechanism of aortic valve insufficiency [2] and to evaluate the results of surgical procedures in the treatment of aortic root pathology [3,4]. Our experience, however, represents the first application of FEA to obtain a patient-specific prediction of post-operative results. Our FEA-based simulation includes four steps: 1) Creation of the patient aortic root finite element model obtained, based on preoperative patient measurements. The model of valve leaflet was created following the geometrical guidelines proposed by Labrosse [5]. 2) Creation of the prosthesis model obtained, based on company specification, either for stentless bioprosthesis or for Dacron graft tube. Once more for the stentless prosthesis the model was created according to the guidelines proposed by Labrosse [5] and assuming that the three leaflets have the same geometrical features of a healthy aortic valve [6]. Different models were realized according to different sizes of both types of prosthesis. 3) Computer-based simulation of the surgical procedure: obtained, according to the type of surgery performed, after the definition of the ideal suture line of the prosthesis within the aortic root (in case of AVR procedure) or, conversely, of the native aortic valve inside the Dacron graft in case of AVS. A different simulation was obtained for each size of the prosthesis considered. Abaqus Explicit solver v6.10 (Dassault Systemes, Providence, RI, USA) was used to perform FEA. 4) Computer-based simulation of the diastolic behavior of the aortic root following surgery: to model the diastolic loading an 80 mm Hg uniform pressure was applied on the leaflets. The evaluation of the optimal size to be implanted was obtained using the prediction of postoperative physiology of aortic root function, based on two post-operative parameters (see Fig. 1): (i) height of coaptation ratio (Hc=Hc/Sh 100) defined as the level of the sinus height where the coaptation occurs (Hc) correlated to the total sinus height (Sh); (ii) length of coaptation ratio (Lc=Lc/Lh 100) defined as the effective coaptation length (Lc) correlated to the leaflet height (Lh). According to our pre-study evaluation in patientwithout aortic root disease, themost physiological conditions were defined as: a) Hc approaching 100%; and b) Lc approaching 40%. Following the validation phase of our model, we obtained a complete patient-specific simulation either in case of AVR or AVS. In both cases the FEA predictive model shows as, for a specific patient, the size of the graft used significantly influence post-operative parameters. In case of AVR, both Hc and Lc increased when increasing the size of the prosthesis, regardless of the different aortic root configurations represented by different patients. In case of AVS, conversely, both values decrease when increasing the size of the graft. According to the criteria previously described we were able to identify the graft of choice either in AVR or AVS. Furthermore, in all cases considered in our preliminary phase, the prosthesis size predicted to be the best option was the one actually chosen at the operation by the surgeon, completely blinded in respect to the results of simulation. We also obtained, therefore, a clinical validation of our model as post-operative in vivo transthoracic echo evaluation showed a good correlation with FEA prediction in terms of Hc and Lc. The application of engineering models to the medical science is surely a matter of interest. Our work was surely inspired by previous application of FEA to study valve sparing procedures [3,4] but the model we propose is quite dissimilar from the model previously proposed. Grande et al. [3] considered only the diastolic phase of cardiac function and, therefore, their model is based only on the final closing phase of valve leaflets. Furthermore, their study is mainly focused on the stress/strain pattern on the leaflets and on the aortic wall and does not take into account the length and height of coaptation. Soncini et al. [4], on the other hand, used a model based on both closing and opening phases of the valve and focused also on post-operative leaflet coaptation. As we mentioned above, the innovative aspect of our model consists of the extension of FEA application from the descriptive and diagnostic area to the extremely more complex and potentially valuable area of surgical result prediction in a patient-specific relationship. The second innovative aspect of our study is related to the application of our model to the evaluation of AVR surgical procedure, achieved by isolating patient-