Session S81.3

Reproducibility of IVUS Measurements in Heart Transplant Recipients: Increased Quality of Data by Using a Dedicated Software for Image Analysis

V D'Errico*, L Potena, D Fiore, F Fabbri, F Grigioni, G Magnani,
P Ortolani, I Bianchi, I Corazza, R Zannoli, A Branzi

Hospital S. Orsola-Malpighi
Bologna, Italy

Background: Cardiac allograft vasculopathy (CAV) is the major cause limiting long term graft survival after heart transplantation (HT), and is characterized by changes in coronary artery geometry, like intimal thickening and vessel remodeling. Given the limited strategies available to reduce its impact on outcome, early diagnosis of CAV – for which intravascular ultrasound (IVUS) is the gold standard– is crucial to appropriately modulate therapy and to reduce contributing risk factors. However, a highly reproducible image-analysis method is required to capture the complex mechanisms beyond CAV-related changes in coronary geometry.
Methods: Volumetric reconstruction of the proximal 30mm of left anterior descending artery was performed on IVUS images (Eagle Eye Gold imaging Catheter, Volcano corporation) prospectively acquired from 13 HT recipients 1 and 12 months after transplant. Manual and semi-automatic (by QIvus Clinical Edition Medis, Rotterdam, NL) border detection techniques were compared to obtain vessel, lumen and plaque volumes. We compared the measurements obtained with the two techniques and assessed interobserver variability of the software-based technique, by repeating the same randomly chosen measures. In addition, we compared reproducibility of maximal intimal thickness (MIT), a linear measure obtained manually and with proven prognostic relevance, with volume measurements obtained with the software based method.
Results: Changes in plaque, vessel and lumen volumes detected manually poorly correlated with the software measures (i.e. R=0.78 for lumen volume after 12 months, R=0.74 for intimal volume after 12 months, ). Interobserver variability of software-based measurements was lower than 5%, both for volumetric measurements and for MIT change (i.e. R=0.96 for intimal volume after 12 months)
Conclusion: Measures of coronary volume, obtained with the same instrument, significantly varied depending on the analysis method. A software analysis based on semi-automated border detection provided a better accuracy and measurements repeatability than manual analysis. Volumetric measures obtained by the software-based technique resulted as reproducible as MIT measures. These findings allow wide use of volumetric parameters for prognostic validation and for a comprehensive description of CAV development.

(Abstract Control Number: 282)