3D-printed aortic roots may help to predict paravalvular leak post-TAVR
Click Here to Manage Email Alerts
A method of preprocedural modeling using 3D printing and ex vivo implantation successfully predicted incidence of paravalvular leak and site of leak after transcatheter aortic valve replacement, according to research published in Catheterization and Cardiovascular Interventions.
Using preprocedural CT scans of the aortic roots of patients (n = 20; median age, 78 years; 70% men) who underwent TAVR for calcified aortic stenosis, researchers created 3D-printed models made of thermoplastic polyurethane.
These models were subsequently implanted ex vivo with Sapien balloon-expandable frames (Edwards Lifesciences), matched to those implanted in each patient, and scanned with flash dual-source CT (Siemens). Researchers then evaluated the scans for relative stent appositions. Findings were compared with post-TAVR echocardiograms to confirm paravalvular leak.
In 10 patients with echocardiographic paravalvular leak, the analysis of the 3D model correctly identified the site of the leak in eight cases.
Moreover, in 10 patients without echocardiographic paravalvular leak, the 3D model analyses correctly predicted the absence in nine cases.
“The surgeons have always relied on direct visualization of the anatomy to perform these procedures and therefore have frequently relied on three-dimensional preprocedure planning, with 3D printing only in extreme scenarios or as a means of communicating with patients,” Sergey Gurevich, MD, assistant professor of medicine of the cardiovascular division at University of Minnesota Health, told Healio. “However, structural interventional cardiologists are unable to directly visualize the anatomy and rely on multiple imaging modalities to piece together the true cardiac anatomy (fluoroscopy, preprocedure CT, transesophageal echocardiography, etc). It was, therefore, a natural evolution of preprocedure planning to advance to the one option that truly mimics what the surgeons rely on to perform their operations: preprocedure 3D printing of the human heart.”
Sensitivity and specificity
In other findings, the sensitivity and specificity of the 3D models for the prediction and detection of paravalvular leak was greater than annular calcium (40%) or annular eccentricity index (50%), respectively. According to the study, the C statistic for the 3D models was significant at 0.85 (95% CI, 0.69-1).
“The key to our study was the degree of paravalvular leak. Our patients had only mild paravalvular leak. There are data showing that even mild paravalvular leak in these patients can lead to poor outcomes. However, the majority of the data has stemmed from moderate and greater paravalvular leak,” Gurevich said in an interview. “As a result, the 3D prediction models seemed to achieve a similar success to more conventional prediction models such as asymmetric annular calcium and annular eccentricity index. Yet our study showed that these predictors fare much worse — some not much better than flipping a quarter — when trying to predict mild paravalvular leak. The 3D printed method, however, was able to predict these with high fidelity. Our overall better prediction, when compared to other studies evaluating 3D printing, likely stems from post printing implants with real valves and the computer analysis of those prints for evidence of leak.”
Plans for future research
“Second and third generation TAVR valves are designed to overcome many of these limitations such as high rates of prohibitive paravalvular leak and permanent pacemaker implants,” Gurevich told Healio. “Yet, certain populations continue to suffer a high rate of these less favorable outcomes. Some, in particular, due to an unfavorable anatomy (ie, nontricuspid aortic valves — bicuspids, unicuspids, quadricuspids). Our next study will look at evaluating primarily bicuspid TAVR implants for the degree of paravalvular leak. As mentioned above, our ultimate progression is to an entirely digital/computerized approach for predictive 3D modeling. The proof of concept for this should be published imminently in the near future.” – by Scott Buzby
For More Information:
Sergey Gurevich, MD, can be reached at gure0011@umn.edu.
Disclosures: The authors report no relevant financial disclosures.
Read more about
Click Here to Manage Email Alerts