The Efficacy and Associated Learning Curve of Office-Based High-Resolution Ultrasound to Detect Shell Failure in Breast Implants

Goals/Purpose: High-resolution ultrasound (HRUS) has become increasingly prevalent in plastic surgery practices to detect implant rupture prior to revision surgery. MRI has long been the gold standard of breast implant imaging, but it is expensive and lowly compliant. As a result, in 2020 the FDA updated their rupture screening guidance and now recommends initial implant screening between 5-6 years post-surgery and every two years after by either MRI or HRUS. FDA guidance further recommends HRUS for asymptomatic woman as a method of long-term implant surveillance. Published research amongst plastic surgeons using HRUS has found a good overall accuracy of the imaging technique and believe it is a practical alternative to the more expensive MRI. The literature also references a learning curve associated with incorporating this imaging technique into a medical practice. This learning curve is described as predictable and quick, but it is yet to be visualized or defined with use in breast implant imaging. This study will attempt to calculate test statistics for the use of HRUS to detect shell failure, as well as define its associated learning curve with respect to the predicted improvement for all statistics of test and surgeon proficiency.

Methods/Technique: A retrospective review of in-office PS Imaging HRUS on patients with breast implants was conducted across two plastic surgery practices. Pre-operative ultrasound reports and intra-operative findings reported by the surgeons were compared to determine the accuracy of the initial HRUS reading. Due to the retrospective nature of the study, criteria were established for the HRUS readings in which ambiguous or uncertain terms could be clearly categorized and defined. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated for each group of ten subsequent patients to determine test and surgeon accuracy over time. A regression analysis of sensitivity and specificity was completed to determine the learning curve associated with these measures over time.

Results/Complications: Averages for all test values were calculated for the HRUS readings in this study, with overall sensitivity of 77.6%, specificity of 95.1%, positive predictive value of 81.9%, negative predictive value of 93.7%, and overall accuracy of 91.2%. Sensitivity and specificity were then broken down for every group of ten HRUS that were completed, and a regression analysis was performed. The positive predictive value and negative predictive value were not used when looking at the learning curve as these two test measures are based on prevalence. With only two medical practices included in this study and an obviously different prevalence between the two, it was determined that these test measures would not be as useful when determining the learning curve. The regression analysis showed that there was a linear improvement for both sensitivity and specificity over time. The improvement was visible and significant for the sensitivity (p=0.003), and the physicians were able to remain above the standard sensitivity defined by the literature after their 60^th HRUS reading. There was also a linear improvement in specificity, but it was not significant (p=0.816). However, it is important to note that the physicians in this study started at a specificity of 100% and their specificity remained above that of the literature standard throughout the course of their HRUS readings.

Conclusion: The results of this study show a high efficacy of HRUS to detect shell failure in silicone breast implants and supports the idea that in-office HRUS is a reliable alternative to MRI for implant surveillance for asymptomatic patients. The test statistics calculated in this study were also higher than previously reported in literature, suggesting recent improvements in equipment and education have improved the efficacy of the HRUS. There is also a calculable linear improvement for all test values of the HRUS as it becomes incorporated into a plastic surgery practice. This improvement shows that continued use and practice with the imaging technique allows for more accurate findings.