Detection of articular perforations of the proximal humerus fracture using a mobile 3D image intensifier – a cadaver study

Correct fracture classification, anatomical reduction, and stable fixation, along with avoidance of iatrogenic and material-related complications, may provide the basis for a good functional outcome following proximal humerus fracture surgery. In recent years, locking plates have been widely used for the treatment of proximal humerus fractures [1‐4]. Nevertheless, high complication rates, comprising primary and secondary screw perforation, malreduction, malunion, nonunion, avascular necrosis, and infection, have been observed [5]. One area of particular concern involves the reportedly high rates of intraoperative humeral head screw perforation or screw cutout in the follow-up period [6‐10]. This is likely the result of several factors, including diverging and converging locking screw vectors, the convex morphology of the humeral head, and poor bone quality limiting the tactile feedback of the drill bit, among other things. Iatrogenic articular screw penetration can lead to the destruction of the glenoid, which has been found to be unsatisfactorily treatable [7, 11]. The reduction is usually assessed intraoperatively while utilizing fluoroscopy in the anterior-posterior radiographic (AP) and Velpeau axillary views [12, 13]. The standard postoperative radiological control involves anteroposterior scapular, lateral scapular, and axillary radiographs.

The introduction of mobile 3D fluoroscopy has made intraoperative multiplanar imaging possible and it is used in navigated spinal surgery [14], pelvic operations [15], and for fractures of several extremeties [16‐19]. Thus, the purpose of this study was to determine the AP views that are necessary to detect primary screw perforation of the humeral head under a controlled “in-vitro” setup. The secondary goal was to investigate the accuracy of perforation detection via multiplanar reconstructions using a mobile 3D image intensifier.

The principal findings of this study show that a combination of three AP views – neutral, 30° internal rotation, and 30° external rotation – permit the identification of 100% of articular perforations in an in-vitro setup. Additionally, coronal reconstruction of a 3D fluoroscopic scan provided a 100% rate of detection of the perforating K-wires.

The use of locking plates in the surgical treatment of proximal humerus fractures is associated with an unexpectedly high rate of screw cutouts and revision surgery [9]. Biomechanical studies have emphasized the value of anchoring screws in the subchondral bone of the humeral head to improve implant stability [20, 21]. However, the spherical shape of the proximal humerus and the limited tactile sensation of its soft cancellous bone make it difficult to determine an accurate screw length, and reported rates of intraoperative screw penetration are high. Iatrogenic screw penetration, even if recognized and corrected before leaving the operating room, may lead to late failure [11].

The protocol for using locking plates and the attention placed on the technical aspects of applying them have been emphasized in the past [11, 22]. Nevertheless, only a few studies have investigated the potential for optimizing the recognition of early or late stage screw perforation [23, 24].

Complications following proximal humerus fracture surgery may likely be a result of inadequate use of intraoperative radiographs or fluoroscopy [25]. Accordingly, the use of routine intraoperative fluoroscopy to confirm hardware placement and a stable anatomical reduction are recommended. The anteroposterior view is a key component of a basic shoulder series. Often, two AP projections are obtained, namely one with the arm in an external rotation and one in an internal rotation [26]. Nevertheless, the algorithm for adequate intraoperative imaging remains inconsistent. Bengard and Gardner suggested placing the arm at 20° to 30° flexion and 80° to 90° internal rotation to aid in visualizing the difficult-to-assess posterosuperior region of the humeral head [11]. However, in contrast to our study they did not provide experimental data to corroborate their suggestion. Spross et al. identified a combination of four projections to account for all cut outs and to establish the correct screw position [23]. In a cadaver study, they determined that the axial view with 30° abduction was the best radiographic projection (76% sensitivity), and that a combination of four views (APIR/AP0°/APER/ax30°) had a sensitivity of 100%. They too examined a combination of the external rotation, neutral position and internal rotation (sensitivity 96%), though the degree of internal rotation (sling position) appeared variable. With intraoperative 3D fluoroscopy and multiplanar reconstruction standardized imaging with CT-like quality can be obtained. At the same time the findings of our study suggest that the investigated procedure holds the potential to detect 100% of primary screw perforations as one of the most common intraoperative complications. Recently, Lowe et al. described the use of a combination of nine fluoroscopic images to identify eight of nine intra-articular screws with a sensitivity of 100% [24]. Their recommendation to use nine C-arm views to evaluate screw placement may be realistic under in-vitro conditions with standardized placement of the screws.

Moreover, all screws may be scrutinized under fluoroscopy in varying degrees of internal and external rotation in order to verify that there is no need for intra-articular hardware [12].

We suggest the use of intraoperative 3D scans to detect 100% of screw perforations independent of screw placement. The advantage of the intraoperative fluoroscopic 3D scan compared to conventional live fluoroscopy is the defined number of images together with multiplanar reconstruction. Moreover, the operating personnel can leave the operating room which reduces the radiation exposure. For the analyzed five screw configuration at least four determined images (30° IR, 30° ER, AP, axial) are needed to examine all screws properly. Finding the right plane involves several control images especially as an exact axial plane is not reproducible with certainty. Altogether this would lead again to a higher amaount of radiation exposure at least for the personnel in the operating room.

The important role of intraoperative multiplanar imaging after osteosynthesis is supported by the high rate of immediate corrections for 11–39% of other regions of the body [16, 18, 19, 27]. Hence, the additional medical benefit seems undisputed [28].

In a feasibility study of the intraoperative use of a mobile 3D C-arm with multiplanar imaging for operating on acute proximal humerus fractures in 20 patients, screw replacements due to perforation or subchondral positioning were performed in 25% of cases [29]. Overall, the complication and revision rates due to technical errors after locking plate osteosynthesis [30, 31] may be drastically reduced if flaws were discovered intraoperatively. The question of whether or not the making of intraoperative corrections using 3D scans leads to superior immediate and long-term functional results has not yet been sufficiently investigated for other joints [32].

Our study has the same inherent weakness of many cadaveric studies. However, the mean age of the cadavers used in our study was 76.8 years, corresponding to the typical age of patients undergoing surgery after proximal humerus fractures. In addition, the results of our study are only valid for the tested plate design with the five screw configuration and are not generalizable. Other proximal humeral plate systems and screw configurations would need separate testing to determine the necessary X-ray views for the detection of all perforations. Whereas most implant designs have at least eight options for screw placement, a screw configuration with five screws has been chosen for the present study. This is in accordance with Erhardt et al. [33] who suggested that at least five screws in the humeral head fragment are necessary for the stabilization of proximal humeral fractures. The screw configuration that was used was defined by the angle stable locking plate and the necessary targeting device. Nevertheless, additional X-ray images may be necessary for other screw configurations [23, 24] and plate designs. Despite the positive aspect of our findings, namely the successful identification of all screw perforations with three AP views, additional X-ray images are often required to properly assess reduction and fixation. Therefore, the intraoperative 3D scan with multiplanar reconstruction is optimal in that it detects all perforations, independent of the screw configuration, and provides critical information regarding reduction and fixation. Finally, the radiation exposure of the intraoperative 3D scan may be a major concern, though in comparison to computed tomography, the radiation dose is significantly reduced. However, to our knowledge comparative data are not available in the literature.

In order to reveal all of the intraoperative and postoperative screw perforations in a “five screw configuration”, conventional AP images should be established in both the neutral positions (0°), at 30° internal rotation and 30° external rotation. Alternatively, the intraoperative 3D scan with multiplanar reconstructions enables a 100% rate of detection of the screw perforations.