Risk factors for subsidence of titanium mesh cage following single-level anterior cervical corpectomy and fusion

The present retrospective cohort study included 73 consecutive patients who underwent single-level anterior cervical corpectomy and fusion with the use of TMC (DePuy Spine, New Brunswick, New Jersey) between March 2012 and October 2015. To achieve stability, anterior plate (DePuy Spine, New Brunswick, New Jersey) was also employed in all the patients. Written and informed consent has been obtained from each patient. Corpectomies were performed for various pathologies such as posterior osteophytes of vertebral, OPLL and prolapse of free nucleus pulposus. There were 58 patients with cervical spondylotic myelopathy and 15 patients with OPLL. The enrolled patients included 37 males and 36 females and the mean age at surgery was 58.18 years (range 30–77). The level of corpectomy and fusion was as follows: C3 in 1 case, C4 in 14 cases, C5 in 29 cases and C6 in 29 cases. The exclusion criteria were: trauma, infection, tumor, posterior fusion, revision surgery and inability to measure due to the overlap of shoulders. Fifteen patients had one comorbid illness whereas 10 patients had more than one comorbidity. Hypertension was diagnosed in 16 patients, diabetes mellitus in 13 patients, and coronary disease in 6 patients.

All operations were performed by an experienced spine surgeon (H. T.). Under general anesthesia, the patient was maintained in the supine position with neck slightly extended. We approached the cervical spine through a standard channel technique in all patients. Once the vertebral levels were carefully identified by intraoperative radiography, the intervertebral space was expanded with a distractor system. Following necessary discectomies, the vertebral bodies were removed and autologous bone chips obtained from corpectomy were used as bone graft material. Then remove the posterior longitudinal ligament and osteophytes; and carefully prepare the endplate in each case to minimize subsidence. After complete decompression, an appropriate length of TMC was selected and filled with autologous bone fragments taken from the excised vertebra, with end caps employed in some cases. The TMC was then inserted into the corpectomy defect and the location of the cage was confirmed using intraoperative fluoroscopy. Finally, an appropriately sized anterior cervical locking plate system was applied in all patients for further stabilization. The patients were instructed to wear Philadelphia collars for 6 weeks after surgery.

Anteroposterior, lateral radiographs taken with the patient in a standing position were obtained before operation, immediately after operation. In addition, anteroposterior, lateral, flexion–extension lateral radiographs were taken at 3, 6, 12, 18, 24 months postoperatively and then annually thereafter. All the patients were observed for at least 24 months after surgery. Radiological assessment included segmental angle (SA), cervical sagittal angle (CSA), height of anterior (HAB) and posterior border (HPB), height of anterior (HAE) and posterior endplate (HPE), ratio of anterior (RAE) and posterior endplate (RPE), the alignment of TMC, the extent of subsidence (ES) and the global cervical Hounsfield Units (HU) value. The SA was defined as the angle between the borders of endplates above and below the affected segment (Fig. 1a). The CSA was defined by the Cobb angle formed between the lower endplate of C2 and C7 (Fig. 1b). The HAB and HPB were measured as the distance between the anterior and posterior points of the upper endplate of the superior vertebra and the lower endplate of the inferior vertebra (Fig. 1b). The HAE and HPE were measured as the distance between the anterior and posterior points of the lower endplate of the superior vertebra and the upper endplate of the inferior vertebra (Fig. 1c). The RAE (RPE) was the ratio of immediately postoperative HAE (HPE) to preoperative HAE (HPE), representing the extent of intervertebral space expansion during operation (Fig. 2). The alignment of TMC was defined as the angle formed between the central axis of TMC and the line through the midpoints of both endplates of adjacent vertebrae (Fig. 1d). Subsidence was defined as loss of more than 3 mm in any of the HAB and HPB measured heights compared with that immediately after operation. ES was the ratio of HAB (HPB) at the final follow-up to the HAB (HPB) immediately after operation, representing the extent of subsidence (Fig. 2). HU values obtained from CT scanning have been reported to be firmly correlated with bone mineral density (BMD) and we used technique described by Schreiber to measure the global cervical HU value [12, 13]. Using standard picture archiving and communication system (PACS), an elliptic region of interest (ROI) encapsulating only cancellous bone was drawn on three non-sequential axial images: just inferior to the superior endplate, mid-body, and just superior to the inferior endplate. The software automatically calculated the mean HU value of each ROI and the average of the 3 measurements was regarded as the HU for individual vertebral body (Fig. 3). The global cervical HU value was defined as the mean HU value of each vertebral body from C2 to C7. Bone fusion was judged by the absence of motion more than 2° between the spinous processes on flexion–extension lateral radiographs, the absence of radiolucent gap between the graft and end plate, and the presence of continuous bridging bony trabeculae at the graft-endplate interface. If there was doubt about the satisfaction of criteria on X-ray and the patient was symptomatic, sagittal reconstructed CT was performed for the judgement of fusion. In our research, 42 CTs were performed to examine the bone fusion. Movement of ≥2° on flexion/extension radiographs was regarded as a pseudarthrosis. These parameters were observed by a single spine surgeon who was not involved in the surgery.

Clinical results were evaluated using the Japanese Orthopedic Association (JOA) scoring system for functional neurological status and the Visual Analog Scale (VAS) for arm and neck pain before operation, immediately after operation and at the final follow-up. All surgery-related complications including wound infection, dysphonia, dysphagia and implant-related complications were recorded immediately after operation, 3, 6, 12, 18, 24 months postoperatively and then annually thereafter.

Statistical evaluation was performed using SPSS software. Results were presented as the mean ± standard deviation. Variables between the subsidence and non-subsidence group were compared using the Student’s t-test, Mann-Whitney U test, χ² test and Fisher’s exact test. Factors with p < 0.05 were selected into the multivariate logistic model to identify the risk factors for subsidence of TMC. The best cut-offs of parameters were established using the receiver operating characteristics (ROC) curve. A linear regression analysis was performed to evaluate the relationship between the RAE and the extent of subsidence. The p value < 0.05 was considered statistically significant.

Comparison between the subsidence and non-subsidence group (Table 1) showed that there were no statistically significant differences in the CSA and SA. However, the value of RAE of the patients who experienced subsidence was 1.24 ± 0.08, which was significantly (p < 0.001) larger than that of patients in the non-subsidence group (1.12 ± 0.08). The alignment of TMC and the global cervical HU value were also found significantly associated with the risk of subsidence (p = 0.002, p < 0.001). The fusion rate in the subsidence and non-subsidence group was 27 patients (87.1%) and 39 patients (92.9%) 6 months after the surgery, respectively (p = 0.448). Bone fusion was achieved in all patients at the last follow-up.

The JOA and VAS scores for arm and neck significantly improved in both groups immediately after operation and at the final follow-up (P < 0.001). However, there were no significant differences of the VAS and JOA scores between the subsidence and non-subsidence groups immediately after operation and at the final follow-up (Tables 2, 3). Moreover, no significant differences were observed in VAS and JOA scores immediately after operation and at the final follow-up in the subsidence group (p = 0.393, p = 0.385, p = 0.579). No patients showed wound infection, dysphonia after surgery. Dysphagia occurred in 3 patients (4.1%), and spontaneously resolved within 3 months postoperatively. Three patients underwent revision surgery for reconstruction failures (Fig. 4).

The ROC curves (Fig. 5) demonstrated that the areas under the curve (AUC) for RAE, alignment of TMC, global cervical HU value were 0.803 (p < 0.001), 0.739 (p = 0.001), 0.795 (p < 0.001) and the best thresholds for RAE, alignment of TMC, global cervical HU value were 1.18 (sensitivity:71.9%; specificity:75.6%), 3 (sensitivity:68.8%; specificity:77.5%) and 333 (sensitivity:82.9%; specificity:65.6%). In multivariate logistic regression analysis, RAE > 1.18 (OR = 6.116, 95%CI = 1.613–23.192, p = 0.008), alignment of TMC > 3° (OR = 5.355, 95%CI = 1.474–19.454, p = 0.011) and the global cervical HU value< 333 (OR = 11.238, 95%CI = 2.844–44.413, p = 0.001) were independently associated with subsidence (Table 4).

Moreover, the value of RAE had a significant effect on the extent of subsidence. As evidenced by the simple linear regression (Fig. 6), the subsidence of TMC became more serious with the increase of RAE (r = − 0.502, p = 0.006).