For each patient of the template group, the preoperative CT scan was imported into the commercially available software Mimics 17.0 (Materialize, Belgium) for the 3D reconstruction. Each vertebra from C3 to C7 was reconstructed and exported as “.stl” file. The “.stl” files were processed using the 3-Matic 9.0 software (Materialize, Belgium) for surgical simulation and template designing. Two cutting paths, slightly wider than the diameter of the bur used on the hinge side and the piezosurgery used on the open-door side, were simulated at the junction between the lamina and the lateral mass. The width and depth of the cutting path on the hinge side were based on the length and diameter of the bur (Figures 1A,B). The width and depth of the cutting path on the open-door side were based on the length and thickness of the piezosurgery (Figures 1C,D). These two cutting paths were also simulated to be 5 to 10 degrees deviated from the midline, according to the surgeon's preference. Next, part of the posterior surface of the lamina was extracted to create the body of the EOLP template, which was then subtracted by the two cutting paths. Thereafter, the EOLP template was exported as “.stl” file, which was processed by the 3D printer to fabricate the actual EOLP template in acrylate resin by stereolithography. Then, the templates were thoroughly washed and vacuum dried (Figure 1E). The 3D printed templates were sterilized with ethylene oxide and stocked in plastic bags.

The procedure was carried out as described previously (10). Briefly, after general anesthesia, the patient was put in the prone position with the head fixed by the Mayfield holder. A midline incision was made to expose the spinous process and bilateral lamina from C3 to C7. The supraspinous ligament and interspinous ligament were then cut before the spinous processes were amputated. Thereafter, in the template group, the template was firmly pressed to the lamina. On the hinge side, the bur was placed through the path of the 3D-printed template and the trough was made by sliding the bur along the path. On the open side, the lamina was completely cut through the cutting path with piezosurgery (Figure 2). In the control group, the hinge side and open side were cut based on the bony landmarks and the surgeons' experience. Then the lamina was opened and carefully fixed with the Centerpiece mini-plate. The Centerpiece mini-plate was anchored to the lamina and lateral mass by two 5 mm screws and two 7 mm screws, respectively.

The neurological function was assessed by the Japanese Orthopedic Association (JOA) scores. Visual analogue scale (VAS) scores were used to evaluate neck pain intensity. The JOA and VAS scores were evaluated before surgery and at 12 months postoperatively. The occurrence of complications, including wound infection, cerebrospinal fluid (CSF) leakage, axial symptoms, and C5 palsy, were recorded and dealt with accordingly.

The C2–C7 Cobb angle and the C2–C7 range of motion (ROM) were measured using the Cobb method on the x-ray films before surgery and at 12 months postoperatively (11). The anteroposterior diameter of the canal and the Pavlov's ratio were determined on the lateral view of x-ray films before surgery, at 3 days postoperatively and 12 months postoperatively. The open angle was assessed using the axial view of CT scans taken 3 days postoperatively and 12 months postoperatively. The fracture type on the hinge side and the satisfaction of the trough position on the hinge side and open side were assessed using the axial view of CT scans taken 3 days postoperatively. It was deemed incomplete fracture if the ventral cortex was continuous. Otherwise, it was deemed complete fracture (6). The position of the hinge side and open-door side were evaluated in the axial view of CT scans before surgery and 3 days postoperatively. The satisfactory position was simulated at the axial view of CT scans before surgery in both groups. The final position was marked at the axial view of CT scans 3 days postoperatively. The deviation of the final position from the simulated position was recorded (Figure 3). If the deviation was less than 2 mm, it was recorded as satisfactory. Otherwise, it was recorded as unsatisfactory. The radiological evaluation was performed by two independent spine surgeons, and the mean values were used for statistical analysis.

Statistical analysis was performed using SPSS software (version 19.0, SPSS Inc., IL, USA). The findings are presented as the mean values ± standard deviation (SD) or counts, as indicated. The paired Student's t-test was performed to test the difference before and after surgery within the groups. The independent Student's t-test and chi-square test were performed for analyzing the differences between the two groups. A value of P < 0.05 was considered statistically significant. A chi-square test or Fisher's exact test was used to analyze categorical data between the two groups.

The study started in August 2019 and ended in December 2020. A total of 40 patients were recruited. One patient in the template group and 2 patients in the control group were excluded because of the loss of follow-up. Other patients were all followed up for 12 months. The baseline demographic information for the two groups, including age, sex, body mass index (BMI), diagnosis, operative time, blood loss, size of mini-plate, preoperative VAS of neck and preoperative JOA were evaluated, and no significant differences were observed between the characteristics in the template group and the control group (P > 0.05). The occurrence rate of complications, including CSF leakage, axial symptoms and C5 palsy was a little higher in the control group, but no significant differences were observed (P > 0.05). All the complications were recovered within 1 month after surgery. No wound infection was observed in both groups. The details are described as follows (Table 1).

In the template group, The VAS scores decreased significantly from 4.26 ± 1.59 preoperatively to 1.32 ± 0.86 at 12 months postoperatively (P < 0.05). The JOA scores increased significantly from 8.26 ± 2.16 preoperatively to 15.00 ± 1.73 at 12 months postoperatively (P < 0.05). In the control group, the VAS scores decreased significantly from 4.50 ± 1.29 to 2.06 ± 1.21, and the JOA scores increased significantly from 8.50 ± 2.09 to 14.61 ± 1.33 (P < 0.05). The significant difference of VAS scores between the two groups was observed at 12 months postoperatively (P < 0.05). The details are described as follows (Table 2).

The C2–C7 Cobb angle and the C2–C7 ROM were decreased with no significant difference at 12 months postoperatively compared with that before surgery, and no significant difference was observed between the two groups (P > 0.05). The anteroposterior diameter and Pavlov's ratio increased significantly after surgery and remained well during the follow-up in both groups (P < 0.05). Besides, the data increased more in the template group compared with that in the control group with significantly differences at 3 days postoperatively and 12 months postoperatively (P < 0.05). The Open angle was also significantly higher in the template group compared with that in the control group at 3 days postoperatively and 12 months postoperatively (P < 0.05). The details are described as follows (Table 3).

A total of 185 lamina were analyzed. Based on the CT scans taken 3 days after surgery, 37 out of 90 (41.11%) were completely fractured on the hinge side in the control group, which was significantly higher compared with that in the template group (P < 0.05) (Table 4). The satisfaction of the trough position on both sides was 87.37% (166/190) in the template group, which was significantly higher than that in the control group (P < 0.05) (Table 5).

Nine patients in the template group and 9 patients in the control group were performed by the senior surgeon, and 10 patients in the template group and 9 patients in the control group were performed by the junior surgeon. The incomplete fracture rate and satisfaction of the trough position on both sides had no significant difference between the two groups when the EOLP was performed by the senior surgeon (P > 0.05). However, to the junior surgeon, the incomplete fracture rate and satisfaction were significantly higher in the template group compared with that in the control group (P < 0.05). The details are described as follows (Tables 6, 7).