Abstract
Introduction
Patient specific instrumentation (PSI) generates customized guides from an MRI- or CT-based preoperative plan for use in total knee arthroplasty (TKA). PSI software executes the preoperative planning process. Several manufacturers have developed proprietary PSI software for preoperative planning. It is possible that each proprietary software has a unique preoperative planning process, which may lead to variation in preoperative plans among manufactures and thus variation in the overall PSI technology. The purpose of this study was to determine whether different PSI software generate similar preoperative plans when applied to a single implant system and given identical MR images.
Methods
In this prospective comparative study, we evaluated PSI preoperative plans generated by Materialise software and Zimmer Patient Specific Instruments software for 37 consecutive knees. All plans utilized the Zimmer Persona™ CR implant system and were approved by a single experienced surgeon blinded to the other software-generated preoperative plan. For each knee, the MRI reconstructions for both software programs were evaluated to qualitatively determine differences in bony landmark identification. The software-generated preoperative plans were assessed to determine differences in preoperative alignment, component sizes, and resection depth. PSI planned bone resection was compared to actual bone resection to assess the accuracy of intraoperative execution.
Results
Materialise and Zimmer PSI software displayed differences in identification of bony landmarks in the femur and tibia. Zimmer software determined preoperative alignment to be 0.5° more varus (p=0.008) compared to Materialise software. Discordance in femoral component size prediction occurred in 37.8% of cases (p<0.001) with 11 cases differing by one size and 3 cases differing by two sizes. Tibial component size prediction was 32.4% discordant (p<0.001) with 12 cases differing by 1 size. In cases in which both software planned identical femoral component sizes, Zimmer software planned significantly more bone resection compared to Materialise in the medial posterior femur (1.5 mm, p<0.001) and lateral posterior femur (1.4 mm, p<0.001).
Discussion
The present study suggests that there is notable variation in the PSI preoperative planning process of generating a preoperative plan from MR images. We found clinically significant differences with regard to bony landmark identification, component size selection, and predicted bone resection in the posterior femur between preoperative plans generated by two PSI software programs using identical MR images and a single implant system. Surgeons should be prepared to intraoperatively deviate from PSI selected size by 1 size. They should be aware that the inherent magnitude of error for PSI bone resection with regard to both planning and execution is within 2–3 mm. Users of PSI should acknowledge the variation in the preoperative planning process when using PSI software from different manufacturers. Manufacturers should continue to improve three-dimensional MRI reconstruction, bony landmark identification, preoperative alignment assessment, component size selection, and algorithms for bone resection in order to improve PSI preoperative planning process.
