Objectives:Patellar osteochondral allograft (OA) transplantation has been shown to be a successful treatment in patients with isolated patellar cartilage injury. Currently, there is minimal guidance in anatomic and sizing factors that portend similar patellar surface topography. The most commonly utilized patellar sizing criteria to match the donor and recipient is radiographic tibial width. Our hypothesis is that specific patella anatomic factors will better predict surface topography matching. To our knowledge, no prior study has investigated the topography of the patella and what intrinsic factors of the graft and the recipient affect matching of the chondral and osseous layers between the graft and defect.Methods:Computed tomography (CT) images of the specimens were acquired and three-dimensional (3D) CT models of the patella were then created and exported into point-cloud models using a 3D reconstruction software program. Circular articular cartilage and subchondral bone defect models were created in each point-cloud model of the recipient patella with a diameter of 18 mm and 22.5 mm at 3 locations: the medial, central, and lateral portions of the patellar surface. Circular articular cartilage and subchondral bone graft models were created on all possible locations on the articular cartilage surface models of the donor patellae (Figure 1). The graft models were virtually placed on the surface of the defect model. Orientation of the graft model was adjusted so that its axis matched that of the defect site. Least distances between the graft and the defect articular surface models were calculated and were defined as the shortest distance from the point in question to the corresponding point in space. A mean value of the least distances was calculated for each position of the graft model. The mean least distance of subchondral bone surface in each point was calculated simultaneously. The graft model was then rotated 360° around the axis perpendicular to the articular cartilage surface in 1° increments, and the least distance of articular cartilage surface and the resulting least distance of subchondral bone surface were calculated at each rotating angle. This procedure was repeated for all points in the articular surface model of the donor patella. Step-off was then calculated as the least mean square difference between the defect and graft along the periphery. Stepwise linear regression was used for each defect location to analyze which variables predict degree of mismatch in millimeters.Results:A total of 16 patella were utilized in analysis. Comparison of cartilage least mean square distances between locations demonstrated that the lateral location had significantly less surface incongruity compared to the other two locations (vs medial: p = .0038, vs central: p = .0046). In addition, significant differences in subchondral bone distances were observed between the locations (lateral vs medial: p = .0007, lateral vs central: p < .0001, medial vs central: p < .0001) (Table 1). The associations of six anatomic and morphologic variables with cartilage mismatch, bone mismatch, and step-off for 18 mm and 22.5 mm defects are presented in Tables 2 and Table 3. All variables were analyzed as the difference in value between the recipient and donor. For both lesion sizes, cartilage step-off was the most susceptible to variable differences. Compared to the 18 mm defect group, the 22.5 mm defects were more affected (higher coefficients) by the same differences in variables. Differences in tibial width were associated mismatch for central lesions (eg. 22.5mm defect coefficient: -0.026, p < .001), while cartilage width was associated with mismatch for lateral lesions. (eg. 22.5 mm defect coefficient: -0.034, p < .023).Conclusions:Multiple clinically relevant factors were found to affect graft and defect chondral mismatch and to a lesser extent osseous mismatch. For all three locations at both defect sizes, step-off was the most susceptible to differences in patellar morphology between the donor and recipient. In addition, differences in tibial width, a commonly used metric for patellar graft matching, did not significantly predict chondral mismatch for lateral and medial sized lesions. These findings should be considered when selecting and preparing the graft in a patella osteochondral allograft procedure.