In the present study, we generated 3D maps of the distal femur fractures with coronal fragments and measured the angles of the coronal fractures. Coronal fractures occurred in approximately 32% of the distal femur fractures, involving 67% of type 33-C3 fractures. The coronal fractures were more likely to occur at the lateral condyle , which may result from the valgus of the distal femur. It is widely acknowledged that compared with plain radiographs, CT scan imaging makes it easier to identify coronal fragments [5, 14, 15].
Overwhelming evidence suggests that the incidence of coronal fractures is relatively high. Nork et al. identified coronal fragments in 38.1% of supracondylar–intercondylar fractures, and open fractures were more likely to have coronal fragments than closed fractures . Richards et al. found that 53% of supracondylar–intercondylar fractures had coronal fractures . Li et al.  reported that coronal fractures were present in 41.1% of articular fractures, being involved in 36.8% of AO/OTA type C and comprising 48.1% of type B fractures. In the present study, 67% of type 33-C3 fractures were diagnosed with coronal fractures. Biplanar plain radiographs were diagnostic only for 69% of coronal plane fragments , emphasizing that a preoperative CT scan is mandatory to avoid missed diagnoses.
A three-dimensional fracture map can demonstrate the characteristic of fracture morphology and has been used to study fractures in distal radius , intertrochanter, distal femur , patella , tibial plateau , and ankle , providing more details and the overall characteristics of fractures compared to plain radiography or two-dimensional CT images. Based on the 3D fracture map, the fracture heat map also displays the fracture incidence of each part. Although the fracture heatmap highlights frequently fractured sites, some rare fracture types can be ignored. It should be borne in mind that the 3D fracture map also has its shortcomings: if too many types or numbers of fractures are included, the characteristics of each type of fracture cannot be clearly displayed. Therefore, it is necessary to effectively classify fractures when drawing the map.
Richards et al. divided the intercondylar coronal fractures into four regions: anteromedial, posteromedial, anterolateral, and posterolateral . In their study, the comminuted area of the articular surface was in the weight-bearing area, similar to comminution in the Busch–Hoffa fracture [12, 16]. However, in our 3D maps, the comminution zone and the fractures in type 33-C3 coronal fractures were anterior to that in the Busch–Hoffa fracture. The reconstruction of type 33-C3 fractures showed that most coronal fracture lines ended at the metaphyseal and patellofemoral articular fractures, suggesting that both metaphyseal and patellofemoral fractures occurred prior to fracture of the posterior condyle and that the isolated Busch–Hoffa fracture occurred with both regions intact. This discrepancy may be caused by the different injury mechanisms and resulted in two kinds of comminution and fractures.
The fracture lines and the comminuted area in type 33-B3 in our study were similar to those reported by Xie et al. , who found that the angles between the Busch–Hoffa fracture and the posterior condyle axis were 34.4° (range, − 8.4° to 52.7°) at the lateral condyle and 29.0° (range, − 19.4° to 59.4°) at the medial condyle; the fracture angles with the distal femoral shaft were 23.1° (range, − 8.2° to 68.2°) at the lateral condyle and 19.3° (range, − 10.8° to 58.6°) at the medial condyle. The lateral angles in this study were similar, but the medial angles differed, which may be caused by the dispersion of the distribution of the fractures lines in the medial condyle and the small number of the samples (n = 26 in Xie et al.'s study, n = 14 in the present study).
A mechanical study by Jarit et al. demonstrated that posterior–anterior screws were more stable than anterior–posterior screws for AO/OTA type 33-B3 . Their samples were obtained by osteotomy along the posterior cortex of the femur. However, in the present study, Busch–Hoffa fractures were anterior to the posterior cortex and extended in an anteroinferior direction, consistent with Xie et al.'s findings. The fractures of Jarit's samples were posterior to the concentrated area on the fracture map. Currently, there is no standard osteotomy or fracture model for distal femur biomechanical studies. Importantly, fracture mapping enables visualization of the fracture characteristics and can be used as a reference for making fracture models. Li et al.  suggested that fracture models should be made according to their morphology results.
Our study has several limitations and shortcomings. Injury mechanism and patient outcomes were not analyzed. Patients were excluded from our research if they were diagnosed by plain radiography or CT scan imaging data was unavailable. Finally, this was a single-center study, which compromises the robustness of our findings.