Three-dimensional printing of patient-specific plates for the treatment of acetabular fractures involving quadrilateral plate disruption

This study was implemented with the approval of the Ethics Committee of The Third Affiliated Hospital of Southern Medical University (approval No. 201704006). It was performed in strict accordance with the recommendations in the Guide for the Care and Use of Humans of the National Institutes of Health. The 3DPPS plates in this study have already been approved and certificated by the CFDA (Class III medical device, NO. 20163460576).

Between January 2016 and June 2017, 50 patients in The Third Affiliated Hospital of Southern Medical University Clinical Center with acetabular fractures were included retrospectively according to our inclusion and exclusion criteria. All patients were treated with ORIF at our trauma center. Preoperatively, all patients were informed that they could choose 3DPPS plates or reconstruction plates for internal fixation. They were then divided into a control group and an experimental group based on their choice. The control group (Group A) comprised 35 patients treated by the conventional method of intraoperative contouring of reconstruction plates. The experimental group (Group B) consisted of 15 patients who underwent internal fixation with a 3DPPS plate.

All patients underwent radiographic examinations including X-rays (anterior–posterior (AP) view and Judet view) and computed tomography (CT) scan (slice thickness of 1 mm) before operation. The fractures were classified according to the Judet and Letournel classification [1]. Preoperative data including age, gender, time from injury until surgery, injury mechanism and fracture classification were recorded (Table 1).

In this study, the inclusion criteria were: acute fracture (< 21 days), and unilateral acetabular fracture associated with QLP disruption. The exclusion criteria were: open fractures of the acetabulum, patients who were younger than 18 years or older than 65 years of age at the time of the injury, and fractures involving the posterior wall.

Radiographic data including X-rays and CT scans (Fig. 1a) were imported into Mimics 15.0 software (Materialise, Leuven, Belgium) to generate a 3D model of the pelvis. The threshold was automatically set as Bone (Min226-Max1476). Then the femur and spine were removed using the command “Edit Mask in 3D”. The mask of the pelvis was separated and reconstructed (Fig. 1b). In consideration of the bony symmetry of the pelvis, the uninjured side was mirrored as a supportive model for plate design (Fig. 1c, d).

The patient-specific acetabular plate on the mirrored pelvis was developed using Geomagic Studio 2012 (3D systems, Rock Hill, SC, USA) and Solidworks professional 2015 software (Dassault Systèmes Solidworks Corp, Waltham, MA, USA). First, the STL file of the mirrored pelvis was imported into Geomagic Studio software. Then, the model was repaired and the noise was eliminated. The contour of the plate was then designed. Subsequently, the initial designed surface was extracted using the command “Trim with curve” (Fig. 1e). Taking the fracture pattern into account, the plate was designed in the most optimal position on the mirrored pelvis to achieve stable fixation. A buttress for preventing medial displacement of the QLP was incorporated into the design of the plate, and the surface of the plate was extracted and shelled to construct a plate with thickness of 3.0–3.5 mm to ensure adequate plate strength (Fig. 1f). Finally, the models of the plate and the mirrored pelvis were imported into Mimics 15.0 and Solidworks software after boundary smoothing. Cylinders of 3.5 mm and 6.5 mm were created to simulate the insertion path of screws, and the simulations of the screw insertions included position, orientation, length and number of screws (Fig. 2g–l). The screw path must pass through the major fracture fragments and avoid penetration into the pelvic cavity or hip joint. After the insertion path of screws was determined feasible by surgeons and engineers, the corresponding screw holes were generated in the plate. The final model of the plate was obtained and saved as an STL file.

The model of the plate was imported into the selective laser melting (SLM) 3D printer (DiMetal-100, SCUT, Guangzhou, China) and printed into a real plate using Ti-6AL-4 V powder as the raw material. The biocompatibility and mechanical properties of Ti-6AL-4 V 3DPPS plates have already been proved to be safe for clinical application, as reported in our previous study [13, 14]. Compared to the traditional plate, the 3DPPS plate shows superior biomechanical properties in biomechanical tests [12, 13]. The 3DPPS plate and the acrylonitrile butadiene styrene plastic pelvic model were matched to carry out a rehearsal of the operation before post-processing (Fig. 2a-f). Post-processing of the 3DPPS plate included heat treatment, roll casting, oil cleaning, acid pickling, polishing, anodizing and cleaning. Finally, the plate was packaged and sterilized according to the routine standards for clinical application.

The patients in Group B were informed that they would be treated with the 3DPPS plate before surgery, and all of them agreed to participate and signed the informed consent form for the 3DPPS plate. All patients received antibiotic 30 min prior to induction of general anesthesia. Patients were placed in a supine position on a radiolucent operating table and the principal surgeon stood on the opposite side to the affected hip. The single lateral-rectus abdominis approach was used to anteriorly expose the fracture site [9]. All operations were performed by one senior surgeon. The surgical technique used in the lateral-rectus abdominis approach is described in the Supplementary materials. To repair fractures, it was first necessary to reduce the medial dislocation of the femoral head by lateral traction with a Schanz pin in the lateral trochanter or by manual reduction. After the fracture of the anterior column was reduced, K-wires were used as temporary fixators to fix the anterior column in Group B if necessary. The anterior column was stabilized with a reconstruction plate in Group A. Subsequently, reduction of the QLP and posterior column was performed under direct visualization using reduction clamps or ball-spiked pushers with footplates and held with K-wires. In Group B, the 3DPPS plate was placed in the predetermined position after reduction of the fractures. By pushing the buttress design of the QLP we were able to press the quadrilateral surface back to its anatomical position. Then surgeons were able to insert all screws based on their positions in virtual pre-operative surgical planning. In Group A, an infrapectineal plate was used to buttress the posterior column and the QLP. Fixation of the posterior column was achieved by insertion of a posterior column lag screw. Fractures of the peri-sacroiliac joint and iliac crest were fixed with plates and/or lag screws. Fracture reduction and implant position were checked by fluoroscopy before wound closure. The wound was then closed in layers over drains. The drains were removed when drainage volumes were less than 50 mL per day (Fig. 3).

Postoperative X-ray and CT scans with 3D reconstruction were conducted, and prophylactic intravenous antibiotics were administered for 48 h. To prevent DVT, low-molecular weight heparin was administered daily for 2 weeks.

Physical therapy, including range of motion and muscle strength training, was initiated as early in recovery as was appropriate. Toe-touch weight-bearing was allowed for the first 8 weeks. Progressive weight bearing was allowed after radiological evidence of fracture consolidation was presented for each patient.

Blood loss, operative time, postoperative residual displacement (axial image on CT), reduction quality, and complications were evaluated and statistically compared between two groups. Time for preparation of the 3D-printed model and the plate in Group B and time taken to contour the plates were also assessed. Patients received routine postoperative follow-up assessments at 4 weeks, 3 months, 6 months, 1 year and every 6 months thereafter. The evaluation criteria for reduction quality were according to the Matta scoring system (anatomic < 1 mm, satisfactory: 2–3 mm, or poor > 3 mm) [15].

As shown in Table 1, among all preoperative data, no statistically-significant difference was exhibited between the two groups (P > 0.05). Meanwhile, due to the time cost of construction and transportation, the mean preparation time of the 3DPPS plates was 3.5 ± 0.7 days (range: 3–5 days) in Group B. The construction time of the 3D-printed pelvic models was 3.3 ± 0.5 days (range: 3–4 days). Simultaneously, in Group A, the mean time taken for intraoperative contouring of plates was 11.1 ± 3.4 min (range: 4–18 min).

When compared to Group A, the blood loss in Group B was found to be significantly lower, and the operative time of Group B was also reduced compared to Group A (Table 1). Meanwhile, compared to Group A (2.38 ± 1.10 mm; range: 0.76–4.73 mm), a significant reduction of postoperative residual displacement was exhibited in Group B (1.51 ± 0.97 mm; range: 0.63–4.32 mm). Further, according to the Matta Scoring System, in Group B 10 of the cases (66.7%) were graded as anatomic, four cases (26.7%) as satisfactory and one case (6.7%) as poor, while in Group A, 18 (51.4%) cases were graded anatomic, 13 (37.1%) satisfactory and four (11.4%) as poor, with no significant difference between the groups (Table 2).

All cases in this study achieved radiological evidence of fracture healing within 18 weeks. For those with follow-up of at least 1 year, none of them had loss of reduction at the final follow-up. Representative radiographic data of two cases in Group A are shown in Figs. 3 and 4.

The complications of each group were recorded. In Group B, one patient experienced loosening of a pubic screw although the patient suffered no discomfort and radiological healing was observed at the 3-month follow-up. However, in Group A there was one wound infection, one instance of deep vein thrombosis (DVT), one case of traumatic arthritis of the hip joint and two obturator nerve injuries (Table 3). Regarding the obturator nerve injuries, both patients exhibited temporary weakness and pain of the hip adductors, but the patients recovered eventually in about 1 month without any further medical intervention. In addition, 1 patient was diagnosed with traumatic arthritis at 12 months post-operation.