Three-dimensional printed artificial vertebral body versus titanium mesh cage and nanohydroxyapatite/polyamide-66 cage in anterior cervical corpectomy and fusion: a retrospective cohort study with a minimum 5-year follow-up

Abstract

Background

Anterior cervical corpectomy and fusion (ACCF) is a commonly performed procedure for cervical spondylotic myelopathy (CSM) and ossification of the posterior longitudinal ligament (OPLL), but the choice of reconstructive device after corpectomy remains unsettled. Titanium mesh cages (TMCs) and nanohydroxyapatite/polyamide-66 (n-HA/PA66) cages are both widely used in clinical practice, although subsidence and postoperative loss of cervical alignment remain important concerns. Three-dimensional printed artificial vertebral bodies (3DP-AVBs) have been introduced as an alternative reconstructive option. This study assessed the long-term radiographic and clinical outcomes of 3DP-AVB, n-HA/PA66, and TMC reconstruction after single-level ACCF.

Methods

This retrospective cohort study included 101 patients with CSM or OPLL who underwent single-level ACCF at two centres. Reconstruction was performed using a 3DP-AVB (n = 33), an n-HA/PA66 cage (n = 34), or a TMC (n = 34). Clinical and radiographic data were reviewed with a minimum follow-up of 5 years. Radiographic outcomes included fusion status, subsidence, fused segment height (FSH), and cervical sagittal alignment. Clinical outcomes included the Visual Analog Scale (VAS), Japanese Orthopaedic Association (JOA) score, and Neck Disability Index (NDI). Complications and revision procedures were also recorded.

Results

At final follow-up, the 3DP-AVB group had less loss of FSH than the n-HA/PA66 and TMC groups (1.99 ± 0.73 vs. 2.80 ± 0.70 vs. 3.07 ± 0.81 mm; P < 0.001) and better preservation of cervical sagittal alignment (C2-7 Cobb angle, 19.47 ± 3.31° vs. 15.19 ± 3.25° vs. 15.75 ± 3.54°; P < 0.001). The unweighted subsidence rate was lower in the 3DP-AVB group (15.2% vs. 32.4% vs. 44.1%; P= 0.036), whereas IPTW-weighted analysis showed a numerically lower but nonsignificant trend (17.9% vs. 35.7% vs. 45.8%; P = 0.067). Fusion was achieved in all patients by 12 months. All three groups improved after surgery. Differences in JOA and NDI were observed at 3 months and remained significant at final follow-up (final JOA, 15.09 ± 1.28 vs. 13.59 ± 1.37 vs. 13.53 ± 1.44; final NDI, 10.03 ± 1.57 vs. 11.88 ± 1.45 vs. 11.88 ± 1.61; both  P < 0.001). VAS scores were similar among groups.

Conclusions

In patients undergoing single-level ACCF, 3DP-AVB reconstruction was associated with better preservation of fused segment height and cervical sagittal alignment during long-term follow-up than n-HA/PA66 or TMC reconstruction. Although subsidence was less frequent in the 3DP-AVB group in the unweighted analysis, the IPTW-weighted comparison showed only a nonsignificant trend and should be interpreted cautiously. All three reconstructive methods achieved fusion and postoperative clinical improvement. Compared with the other two implants, 3DP-AVB may provide greater radiographic durability. Although JOA and NDI scores were statistically more favorable in the 3DP-AVB group, the clinical relevance of these differences should be interpreted cautiously.

Introduction

Anterior cervical corpectomy and fusion (ACCF) is an established surgical option for cervical spondylotic myelopathy (CSM) and ossification of the posterior longitudinal ligament (OPLL) [1, 2]. By directly removing ventral compressive lesions, it can provide effective decompression of the spinal cord and may result in satisfactory neurological recovery in appropriately selected patients [3]. Nevertheless, reconstruction after corpectomy remains challenging, particularly because graft-related complications and postoperative loss of sagittal alignment may still occur [4].

The reconstructive strategy used in anterior cervical surgery has gradually changed over time. Autologous iliac crest bone grafting was historically regarded as the standard method, but donor-site morbidity limited its wider acceptance [5]. Cage-based reconstruction was subsequently adopted to avoid graft harvest. Titanium mesh cages (TMCs) reduced donor-site complications, yet implant subsidence remained a persistent concern. Nanohydroxyapatite/polyamide-66 (n-HA/PA66) cages were later introduced because of their osteoconductive properties, although radiographic height loss and interface-related problems have not been completely eliminated [6].

Advances in additive manufacturing have enabled the clinical use of three-dimensional printed artificial vertebral bodies (3DP-AVBs). Their porous architecture can be adjusted to better approximate the mechanical behavior of native bone while also facilitating osseointegration [7–9]. From a theoretical perspective, this design may reduce elastic modulus mismatch, improve load transfer, and support implant-bone integration; however, these mechanisms were not directly tested in the present cohort and should be regarded as biologically plausible explanations.

Although these implants are increasingly used in cervical reconstruction, direct comparisons among TMCs, n-HA/PA66 cages, and 3DP-AVBs remain limited. The present retrospective cohort study was therefore performed to compare the long-term radiographic and clinical outcomes of these three reconstructive options after single-level ACCF. We expected that 3DP-AVB would show better radiographic performance, particularly in terms of subsidence control and maintenance of cervical sagittal alignment.

Methods

Study design, aim, and setting

This retrospective cohort study was conducted at the Departments of Spinal Surgery of China-Japan Friendship Hospital and Beijing Hospital of Traditional Chinese Medicine (Beijing, China), from September 2016 to September 2020. The study was approved by the institutional ethics committees of both hospitals. Written informed consent was waived because of the retrospective design, in accordance with national regulations and institutional requirements.

Participants

Consecutive patients who underwent single-level anterior cervical corpectomy and fusion (ACCF) for cervical spondylotic myelopathy (CSM) or ossification of the posterior longitudinal ligament (OPLL) during the study period were screened for inclusion. Before surgery, all patients were informed in detail about the three available reconstructive implants, including the 3DP-AVB, n-HA/PA66 and TMC. The explanation included the material characteristics, potential advantages and disadvantages, expected clinical use, and relative costs of each implant. After receiving this information, patients made an autonomous choice of reconstructive implant according to their individual condition, economic considerations, and personal preference.

The inclusion criteria were: (1) age 18–80 years; (2) diagnosis confirmed by magnetic resonance imaging (MRI) and computed tomography (CT); (3) complete clinical and radiographic data; and (4) a minimum follow-up of 60 months. The exclusion criteria were: multilevel corpectomy, previous cervical surgery, traumatic or infectious pathology, severe osteoporosis (T-score <  − 2.5 on dual-energy X-ray absorptiometry), and systemic or psychiatric disorders that might affect outcome assessment.

Surgical procedure

All procedures were performed by a senior spine surgeon with more than 15 years of surgical experience. A standardized right-sided Smith-Robinson anterior cervical approach was used [10]. After complete corpectomy, the endplates were prepared carefully with curettes to preserve the subchondral bony structure while removing the cartilaginous endplate. Implant size was selected intraoperatively to obtain an appropriate fit without excessive distraction. Autologous bone harvested from the resected vertebral body was morselized and tightly packed into the implant.

Patients in the 3DP-AVB group received a 3D-printed artificial vertebral body (Beijing AK Medical Co., Ltd., Beijing, China), with implant dimensions selected according to intraoperative measurements. The n-HA/PA66 group received nanohydroxyapatite/polyamide-66 cages (Sichuan National Nano Science and Technology Co., Ltd., Chengdu, China). The TMC group received titanium mesh cages (Fule Science & Technology Development Co., Ltd., Beijing, China), which were trimmed intraoperatively to improve endplate conformity (Fig. 1). An anterior cervical plate was applied in all patients. After surgery, all patients wore a rigid cervical collar for 12 weeks.

Fig. 1.

Representative reconstructive implants used in this study

Outcome assessment

The primary radiographic outcomes included fused segment height (FSH), implant subsidence, the C2-7 Cobb angle measured on standing lateral radiographs, and fusion status at 12 months. FSH was measured on the lateral cervical radiographs as the mean of the anterior and posterior heights of the fused segment. AB represented the anterior distance between the inferior endplate of the cranial adjacent vertebra and the superior endplate of the caudal adjacent vertebra, whereas CD represented the corresponding posterior distance. FSH was calculated as (AB + CD)/2 (Fig. 2). Subsidence was defined as a decrease of ≥ 3 mm in FSH compared with the immediate postoperative measurement [11].

Fig. 2.

Radiographic measurement methods: Fused segment height (FSH) calculated as (AB + CD)/2; the angle β formed by the lower edge of C2 vertebra and the lower edge of C7 vertebra is considered as the C2–7 Cobb angle

Fusion was evaluated using the available postoperative imaging. Initial assessment was based on anteroposterior and lateral, and dynamic flexion–extension cervical radiographs. Fusion was considered present when continuous bridging bone crossed the operated segment, no radiolucent gap was identified between the grafted area and the adjacent vertebral endplates, and no abnormal motion was observed on dynamic radiographs [12]. When plain radiographs were inconclusive, CT with sagittal and coronal reconstruction was reviewed. Fusion was then confirmed when continuous bridging bone, particularly extragraft bridging bone, was seen across the operated segment without intervening lucency [13].

Secondary outcomes were clinical measures, including the Visual Analog Scale (VAS; 0–10) for neck and arm pain, the Japanese Orthopaedic Association (JOA) score (0–17) for myelopathy severity, and the Neck Disability Index (NDI; 0–50) for functional impairment. Assessments were performed preoperatively and at 3 months and the final follow-up (≥ 60 months).

Radiographic assessment

Radiographic measurements were performed using the picture archiving and communication system. To reduce measurement error, all standing lateral radiographs were obtained using a standardized protocol, with patients standing in a neutral cervical position and looking straight ahead. Radiographic magnification was corrected using a fixed-length calibration ruler placed at the imaging plane. Measurements were independently performed by two spine surgeons who were blinded to the clinical outcomes. Interobserver reliability was evaluated using intraclass correlation coefficients, and all ICC values were greater than 0.85. Any disagreement between the two observers was resolved through discussion and consensus; if consensus could not be reached, a third senior spine surgeon was consulted for final adjudication.

Statistical analysis

All statistical analyses were performed using SPSS version 27.0 (IBM Corp., Armonk, NY, USA). Continuous variables are presented as mean ± standard deviation and were compared using one-way analysis of variance (ANOVA) with Tukey's post hoc test, or the Kruskal–Wallis test when appropriate. Categorical variables are expressed as frequencies and were compared using the chi-square test or Fisher's exact test. All tests were two-sided, and P < 0.05 was considered statistically significant. Because this was a retrospective study, no a priori sample size calculation was performed; the final sample size was determined by the number of eligible patients available during the study period. Longitudinal changes in radiographic and clinical outcomes were further evaluated using linear mixed-effects models, with group, time, and the group × time interaction included as fixed effects and patient identity included as a random intercept. In addition, inverse probability of treatment weighting was performed as a sensitivity analysis to improve baseline comparability among the three non-randomized groups. The propensity score model included clinically relevant baseline variables, including age, sex, BMI, BMD, diagnosis, operated level, follow-up duration, and surgical year.

Results

Patient demographics

A total of 101 patients met the inclusion criteria, including 34 in the TMC group, 34 in the n-HA/PA66 group, and 33 in the 3DP-AVB group. The baseline demographic and preoperative characteristics, including age, sex, BMI, bone mineral density (BMD), diagnosis, duration of follow-up, operated level and surgical year, were comparable among the three groups.

The operative parameters also varied significantly across the groups. The 3DP-AVB group had the shortest operative time (97.70 ± 11.29 min), followed by the TMC group (110.15 ± 8.86 min) and the n-HA/PA66 group (118.41 ± 11.20 min) (P < 0.001). Similarly, intraoperative blood loss was lowest in the 3DP-AVB group (86.61 ± 8.43 mL), followed by the n-HA/PA66 group (91.62 ± 10.17 mL) and the TMC group (96.71 ± 11.67 mL) (P < 0.001; Table 1).

Table 1.

Baseline demographic, clinical, and perioperative characteristics of the three groups

Characteristic	3DP-AVB (n = 33)	n-HA/PA66 (n = 34)	TMC (n = 34)	P value
Age (years)	57.15 ± 6.96	54.35 ± 6.21	55.32 ± 7.73	0.257
Sex (M/F)	20/13	17/17	18/16	0.668
BMI (kg/m2)	23.57 ± 2.44	24.66 ± 2.24	23.56 ± 2.20	0.080
BMD (T-score)	-0.50 ± 0.31	-0.44 ± 0.33	-0.37 ± 0.36	0.303
Diagnosis (CSM/OPLL)	23/10	27/7	24/10	0.608
Blood loss (mL)	86.61 ± 8.43	91.62 ± 10.17	96.71 ± 11.67	< 0.001
Operative time (min)	97.70 ± 11.29	118.41 ± 11.20	110.15 ± 8.86	< 0.001
Follow-up duration (months)	80.73 ± 15.54	76.00 ± 12.69	73.18 ± 15.85	0.106
Operated level, n (%)				0.467
C3	1(3.0)	2(5.9)	2(5.9)
C4	8(24.2)	3(8.8)	8(23.5)
C5	14(42.4)	18(52.9)	18(52.9)
C6	10(30.3)	11(32.4)	6(17.6)
Surgical year, n (%)				0.497
2016	2 (6.1)	7 (20.6)	7 (20.6)
2017	5 (15.1)	6 (17.6)	5 (14.7)
2018	9 (27.3)	4 (11.8)	9 (26.5)
2019	8 (24.2)	11 (32.4)	8 (23.5)
2020	9 (27.3)	6 (17.6)	5 (14.7)
Fusion at 12 months, n (%)	33 (100.0)	34 (100.0)	34 (100.0)
Fusion assessed by plain radiographs alone, n (%)	0 (0.0)	5 (14.7)	3 (8.8)	0.093
Fusion confirmed by CT, n (%)	33 (100.0)	29 (85.3)	31 (91.2)
Subsidence, n(%)	5 (15.2)	11 (32.4)	15 (44.1)	0.036

Radiographic outcomes

The 3DP-AVB group demonstrated more favorable continuous radiographic outcomes compared with the n-HA/PA66 and TMC groups. Specifically, the 3DP-AVB group had the lowest subsidence rate (15.2% vs. 32.4% vs. 44.1%, P = 0.036) (Table 2) and the smallest loss of fused segment height (1.99 ± 0.73 mm vs. 2.80 ± 0.70 mm vs. 3.07 ± 0.81 mm, P < 0.001). In addition, cervical sagittal alignment was better preserved in the 3DP-AVB group, which showed a significantly greater C2–7 Cobb angle at final follow-up than the n-HA/PA66 and TMC groups (19.47 ± 3.31° vs. 15.19 ± 3.25° vs. 15.75 ± 3.54°, P < 0.001) (Figs. 3, 4 and 5). Fusion was achieved in all patients by 12 months, resulting in a 100% fusion rate in all three groups. Fusion was assessed by plain radiographs alone in 0 of 33 patients in the 3DP-AVB group, 5 of 34 patients in the n-HA/PA66 group, and 3 of 34 patients in the TMC group. CT confirmation was used in the remaining 33 of 33, 29 of 34, and 31 of 34 patients, respectively (Table 2). (Figs. 6, 7 and 8) (Figs. 9, 10, 11, 12 and 13).

Table 2.

Radiographic outcome of the three groups

Outcome	3DP-AVB (n = 33)	n-HA/PA66 (n = 34)	TMC (n = 34)	P value
Fusion at 12 months, n (%)	33 (100.0)	34 (100.0)	34 (100.0)
Fusion assessed by plain radiographs alone, n (%)	0 (0.0)	5 (14.7)	3 (8.8)	0.093
Fusion confirmed by CT, n (%)	33 (100.0)	29 (85.3)	31 (91.2)
Subsidence, n(%)	5 (15.2)	11 (32.4)	15 (44.1)	0.036

Fig. 3.

Comparison of C2–7 Cobb angle among the three groups

Fig. 4.

Comparison of fused segment height among the three groups

Fig. 5.

Comparison of subsidence among the three groups

Fig. 6.

Changes in VAS scores among the three groups

Fig. 7.

Changes in JOA scores among the three groups

Fig. 8.

Changes in NDI scores among the three groups

Fig. 9.

Representative case of a 67-year-old male patient treated with single-level ACCF using a 3DP-AVB. a Preoperative lateral radiograph. b Preoperative sagittal CT image showing index-level compression. c Immediate postoperative lateral radiograph. d Lateral radiograph at final follow-up showing complete fusion. e Flexion lateral radiograph at final follow-up. f Extension lateral radiograph at final follow-up

Fig. 10.

Representative case of a 52-year-old female patient treated with single-level ACCF using a 3DP-AVB. a Preoperative lateral radiograph. b Immediate postoperative lateral radiograph. c Final follow-up lateral radiograph showing maintenance of segmental height/alignment and fusion

Fig. 11.

Representative case of a 47 year-old male patient treated with single-level ACCF using a 3DP-AVB. a Preoperative lateral radiograph. b Immediate postoperative lateral radiograph. c Final follow-up lateral radiograph demonstrating postoperative alignment and implant stability

Fig. 12.

Representative case of a 54-year-old female patient treated with single-level ACCF using an n-HA/PA66 cage. a Preoperative lateral radiograph. b Immediate postoperative lateral radiograph. c Final follow-up lateral radiograph showing fusion status and sagittal alignment

Fig. 13.

Representative case of a 72 year-old male patient treated with single-level ACCF using a TMC. a Preoperative lateral radiograph. b Immediate postoperative lateral radiograph. c Final follow-up lateral radiograph showing complete fusion

Clinical outcomes

All three groups showed significant postoperative improvement in JOA, NDI, and VAS scores. Baseline clinical scores were comparable among the groups, including preoperative JOA, NDI, and VAS scores (P = 0.156, P = 0.551, and P = 0.093, respectively).

At 3 months postoperatively, significant between-group differences were observed in JOA and NDI scores (P = 0.022 and P = 0.002, respectively), whereas VAS scores remained comparable across the groups (P = 0.952). These between-group differences in JOA and NDI persisted at the final follow-up. The 3DP-AVB group showed the highest final JOA score (15.09 ± 1.28), compared with the n-HA/PA66 group (13.59 ± 1.37) and the TMC group (13.53 ± 1.44) (P < 0.001). The 3DP-AVB group also showed the lowest final NDI score (10.03 ± 1.57), compared with the n-HA/PA66 group (11.88 ± 1.45) and the TMC group (11.88 ± 1.61) (P < 0.001). Final VAS scores remained comparable among groups (P = 0.864) (Figs. 6, 7 and 8).

Longitudinal changes in radiographic and clinical outcomes

Linear mixed-effects models were used to further evaluate changes in radiographic and clinical outcomes over time. In each model, group, time, and the group × time interaction were included as fixed effects, with patient identity included as a random intercept.

For radiographic outcomes, the C2-7 Cobb angle showed a significant time effect (χ² = 505.623, df = 3, P < 0.001) and a significant group × time interaction (χ² = 364.312, df = 6, P < 0.001), whereas the overall group effect was not significant (χ² = 4.143, df = 2, P = 0.126). FSH showed a similar pattern, with a significant time effect (χ² = 241.757, df = 3, P < 0.001) and a significant group × time interaction (χ² = 44.623, df = 6, P < 0.001), but no significant overall group effect (χ² = 3.928, df = 2, P = 0.140).

For clinical outcomes, JOA scores showed significant group (χ² = 10.883, df = 2, P = 0.004), time (χ² = 409.438, df = 2, P < 0.001), and group × time interaction effects (χ² = 31.071, df = 4, P < 0.001). NDI scores showed a significant time effect (χ² = 748.221, df = 2, P < 0.001) and a significant group × time interaction (χ² = 17.708, df = 4, P = 0.001), although the overall group effect was not significant (χ² = 4.765, df = 2, P = 0.092). By contrast, VAS scores showed only a significant time effect (χ² = 605.507, df = 2, P < 0.001), with no significant group effect (χ² = 2.111, df = 2, P = 0.348) or group × time interaction (χ² = 3.774, df = 4, P = 0.437). Thus, neurological function, disability, and pain improved over time, but group-related differences in longitudinal improvement were evident for JOA and NDI scores rather than for VAS scores (Table 3).

Table 3.

Linear mixed-effects model analysis of longitudinal radiographic and clinical outcomes

Outcome	Effect	Likelihood ratio χ2	df	P value
C2–7 Cobb angle	Group	4.143	2	0.126
	Time	505.623	3	< 0.001
	Group × Time	364.312	6	< 0.001
Fused segment height	Group	3.928	2	0.140
	Time	241.757	3	< 0.001
	Group × Time	44.623	6	< 0.001
JOA score	Group	10.883	2	0.004
	Time	409.438	2	< 0.001
	Group × Time	31.071	4	< 0.001
NDI score	Group	4.765	2	0.092
	Time	748.221	2	< 0.001
	Group × Time	17.708	4	0.001
VAS score	Group	2.111	2	0.348
	Time	605.507	2	< 0.001
	Group × Time	3.774	4	0.437

Linear mixed-effects models were fitted separately for each outcome, with group, time, and the group × time interaction included as fixed effects and patient ID included as a random intercept. Likelihood ratio tests were used to evaluate fixed effects. For C2–7 Cobb angle and fused segment height, four time points were analyzed: preoperative, immediate postoperative, 3 months postoperative, and final follow-up. For JOA, NDI, and VAS scores, three time points were analyzed: preoperative, 3 months postoperative, and final follow-up. P < 0.05 was considered statistically significant

IPTW sensitivity analysis

IPTW-weighted analyses were performed as a sensitivity analysis and generally yielded results consistent with the primary analyses for continuous radiographic and clinical outcomes. At final follow-up, weighted FSH loss remained lowest in the 3DP-AVB group (2.06 ± 0.73 mm), compared with the n-HA/PA66 group (2.86 ± 0.73 mm) and the TMC group (3.07 ± 0.80 mm) (P < 0.001). The weighted final C2-7 Cobb angle was also better preserved in the 3DP-AVB group (19.89 ± 3.32°) than in the n-HA/PA66 group (14.69 ± 3.04°) or the TMC group (15.35 ± 3.47°) (P < 0.001).

Weighted subsidence rates showed a trend toward a lower incidence in the 3DP-AVB group (17.9%) than in the n-HA/PA66 group (35.7%) and the TMC group (45.8%), although the difference did not reach statistical significance (P = 0.067). Weighted clinical outcomes showed higher JOA scores and lower NDI scores in the 3DP-AVB group (P < 0.001), whereas VAS scores were similar among groups (P = 0.879) (Table 4).

Table 4.

IPTW-weighted sensitivity analysis of key outcomes at final follow-up

Outcome	3DP-AVB (n = 33)	n-HA/PA66 (n = 34)	TMC (n = 34)	P value
Final FSH loss (mm)	2.06 ± 0.73	2.86 ± 0.73	3.07 ± 0.80	< 0.001
Final C2–7 Cobb angle (°)	19.89 ± 3.32	14.69 ± 3.04	15.35 ± 3.47	< 0.001
Final subsidence (%)	17.9	35.7	45.8	0.067
Final JOA	15.03 ± 1.20	13.65 ± 1.33	13.54 ± 1.43	< 0.001
Final NDI	9.87 ± 1.72	11.78 ± 1.37	11.92 ± 1.58	< 0.001
Final VAS	1.36 ± 0.77	1.46 ± 0.79	1.40 ± 0.99	0.879

Values are IPTW-weighted means ± standard deviations for continuous variables and IPTW-weighted percentages for subsidence. The numbers in parentheses in the column headings indicate the original sample sizes before weighting. P values were obtained from IPTW-weighted linear regression for continuous outcomes and IPTW-weighted logistic regression for subsidence. IPTW, inverse probability of treatment weighting

Complications

No infections, cerebrospinal fluid leakage, or perioperative deaths were observed (Table 5). The most common complications were transient dysphagia and adjacent segment disease (ASD) requiring revision. Dysphagia occurred in 6.1%, 8.8%, and 14.7% of patients in the 3DP-AVB, n-HA/PA66, and TMC groups, respectively, with no significant between-group difference (P = 0.480). Hoarseness or recurrent laryngeal nerve palsy was observed in 3.0%, 5.9%, and 11.8% of patients, respectively (P = 0.356). ASD was numerically more frequent in the TMC group (8.8%) than in the 3DP-AVB group (3.0%) or the n-HA/PA66 group (2.9%), although the difference was not statistically significant (P = 0.442).

Table 5.

Postoperative complications during follow-up

Complication	3DP-AVB (n = 33) n (%)	n-HA/PA66 (n = 34) n (%)	TMC (n = 34) n (%)	P value
Infection	0 (0)	0 (0)	0 (0)	–
Dysphagia	2 (6.1)	3 (8.8)	5 (14.7)	0.480
Hoarseness / recurrent laryngeal nerve palsy	1 (3.0)	2 (5.9)	4 (11.8)	0.356
CSF leakage	0 (0)	0 (0)	0 (0)	–
New/worsened neurological deficit	0 (0)	0 (0)	0 (0)	–
Adjacent segment disease (ASD)	1 (3.0)	1 (2.9)	3 (8.8)	0.442

Discussion

In the present cohort, reconstruction with 3DP-AVB was associated with more favorable preservation of fused segment height and cervical sagittal alignment over long-term follow-up than reconstruction with either n-HA/PA66 cages or TMCs after single-level ACCF. The main differences were observed in fused segment height loss and maintenance of cervical sagittal alignment. Linear mixed-effects models further showed significant time effects and group × time interactions for both FSH and C2-7 Cobb angle, indicating that the radiographic trajectories differed among the three implants over time rather than being explained only by isolated final measurements. IPTW-weighted analyses yielded consistent results for these two continuous radiographic outcomes. Although the IPTW-weighted comparison of subsidence showed only a nonsignificant trend toward a lower rate in the 3DP-AVB group, this threshold-based binary outcome should be interpreted together with the significant difference in continuous FSH loss. Dichotomizing FSH loss at ≥ 3 mm may reduce statistical sensitivity and obscure quantitative differences in height preservation. At the same time, all three implants achieved fusion and were associated with postoperative clinical improvement. These findings suggest that the main advantage of 3DP-AVB lies in preservation of structural support over time, particularly FSH and cervical sagittal alignment, rather than in uniform superiority across all measured outcomes.

The shorter operative time and lower intraoperative blood loss observed in the 3DP-AVB group should also be interpreted cautiously. All procedures were performed by the same senior surgeon, which may reduce but not completely eliminate learning-curve or case-complexity effects. The difference may be partly related to implant-specific technical factors, because TMCs often require intraoperative trimming and adjustment, whereas 3DP-AVBs were selected according to intraoperative measurements and did not require such trimming.

Earlier comparative studies have also reported improved structural stability and radiographic maintenance with 3D-printed vertebral body devices in anterior cervical reconstruction [7–9, 14]. The present study adds longer-term follow-up to that body of evidence, showing that the radiographic differences remained evident after at least 5 years. The significant group × time interactions further indicate that postoperative radiographic changes followed different trajectories among the three reconstructive methods. These findings, together with the IPTW-weighted results, suggest that successful reconstruction after ACCF depends not only on immediate postoperative stability, but also on preservation of segmental height and sagittal alignment throughout the later healing period.

Subsidence remains a major concern after ACCF because it may lead to loss of segmental height, local malalignment, and progressive reduction in construct stability during follow-up [11, 15]. Previous biomechanical studies have shown that implant geometry, endplate matching, and stress distribution at the bone-implant interface play important roles in this process [16, 17]. Clinical and radiographic studies have similarly linked these factors to height loss and late radiographic deterioration [18–20]. In the present study, the 3DP-AVB group showed less FSH loss than the n-HA/PA66 and TMC group in both the primary and IPTW-weighted analyses, indicating more durable preservation of postoperative cervical height. The observed trend remains biologically plausible, given the implant-specific structural characteristics of 3DP-AVB that may improve load transfer and reduce stress concentration at the adjacent endplates.

Comparative evidence in ACCF has increasingly pointed toward a radiographic advantage for 3DP-AVB. Wei et al. reported in a prospective randomized cohort that 3DP-AVBs were associated with less subsidence and better preservation of cervical lordosis than TMCs [7]. Fang et al. and He et al. also described advantages in fused segment height and alignment preservation [8, 9]. More recent comparative studies have shown similar trends in different cohorts [14, 21]. The present study is consistent with these reports, particularly because both the conventional analyses and IPTW-weighted sensitivity analyses showed less final fused segment height loss and better preservation of C2–7 Cobb angle in the 3DP-AVB group. Mechanical studies have demonstrated that newer 3D-printed titanium cage designs may provide stronger resistance to subsidence than conventional titanium cages [22–24]. These biomechanical observations offer a plausible explanation for the present radiographic findings.

One practical advantage of 3D printing is the ability to control implant geometry, pore size, and overall porosity with greater precision [23, 25–28]. Porous titanium constructs can therefore be designed to better balance mechanical strength and elastic modulus, which may reduce stiffness mismatch and improve load transfer at the bone-implant interface [20, 25, 28]. Such structural optimization may partly explain the better maintenance of FSH and sagittal alignment observed in the 3DP-AVB group; however, these mechanisms were inferred from previous biomechanical and biological evidence and were not directly measured in the present cohort.

Long-term implant behavior also depends on biological fixation [29–32]. In the present cohort, fusion was achieved in all patients by 12 months, indicating that all three reconstructive methods were capable of achieving successful arthrodesis under the study conditions. Although n-HA/PA66 constructs can achieve high fusion rates, previous studies have described interface-related radiographic findings, including radiolucent gaps, suggesting that osseointegration may not always be optimal [25, 33]. In contrast, the interconnected porous structure of 3DP-AVB may provide a more favorable scaffold for bone ingrowth and implant-bone integration [26, 34, 35].

All three groups demonstrated substantial postoperative clinical improvement, indicating that adequate decompression and reconstruction were achieved regardless of implant type. While JOA and NDI scores were statistically more favorable in the 3DP-AVB group in both longitudinal and IPTW-weighted analyses, the absolute between-group differences were relatively small, and VAS scores improved similarly across all groups. This suggests that the radiographic advantages of 3DP-AVB, particularly in preserving fused segment height and cervical sagittal alignment, may not fully translate into perceptible differences in pain or functional outcomes for all patients. Postoperative recovery after ACCF is influenced by multiple factors—including baseline neurological status, decompression adequacy, rehabilitation, and patient-specific characteristics [36]—so implant type may more directly affect structural maintenance than symptomatic improvement. Overall, 3DP-AVB appears to offer a measurable radiographic advantage with possible but modest clinical benefit, which should be confirmed in larger prospective studies [37].

No infection, cerebrospinal fluid leakage, or perioperative death was observed in any group. The most frequent complications were transient dysphagia, hoarseness or recurrent laryngeal nerve-related symptoms, and adjacent segment disease requiring revision. Although these events were numerically more common in the TMC group, the between-group differences were not statistically significant. This finding is consistent with earlier reports indicating that complications after anterior cervical corpectomy remain clinically relevant and may be influenced by both construct stability and the technical demands of reconstruction [16, 38, 39]. Given the modest sample size and the low number of events, the present study was underpowered to detect reliable differences in complications. Therefore, any possible relationship between improved implant-endplate matching and a lower complication rate should be interpreted with caution.

Limitations

This study has several limitations. First, its retrospective design may have introduced selection bias and residual confounding. Although IPTW analysis was performed to improve group comparability, it could only adjust for measured covariates and could not eliminate unmeasured confounding or replace randomized allocation. Second, the sample size was relatively small, which limited the statistical power for binary outcomes and complication-related comparisons. This may partly explain why the IPTW-weighted subsidence comparison showed only a nonsignificant trend despite a numerically lower rate in the 3DP-AVB group. Third, only single-level ACCF cases were included, so the findings should be extrapolated cautiously to multilevel procedures, revision surgery, severe osteoporosis, or other complex reconstructive settings. Finally, although the minimum 5-year follow-up, longitudinal analysis, and IPTW sensitivity analysis strengthen the study, longer-term and larger prospective multicentre studies are still needed to confirm implant durability, adjacent segment degeneration, and the clinical relevance of radiographic preservation over time.

Clinical implications

The present findings are relevant to implant selection in ACCF reconstruction. When several reconstructive options are available, 3DP-AVB may be considered for patients who are thought to be at greater risk of subsidence or postoperative loss of alignment, particularly when endplate support and load distribution are important surgical considerations. Although the IPTW-weighted subsidence comparison did not reach statistical significance, the consistently smaller FSH loss and better preservation of C2–7 Cobb angle support the potential value of 3DP-AVB when long-term radiographic maintenance is a major surgical priority. More broadly, these results indicate that implant selection after ACCF should not be judged by fusion status alone, but also by how well structural stability is preserved during follow-up. Prospective multicentre studies are still required to further clarify its indications [40].

Conclusions

In this retrospective cohort with a minimum follow-up of 5 years, 3DP-AVB reconstruction was associated with better preservation of fused segment height and cervical sagittal alignment than n-HA/PA66 or TMC reconstruction after single-level ACCF. All three reconstructive methods achieved fusion and were followed by postoperative clinical improvement. Compared with the radiographic findings, the clinical differences were more selective: JOA and NDI favored 3DP-AVB, whereas VAS scores were similar among groups. Overall, these findings indicate that 3DP-AVB may offer better long-term radiographic durability after single-level ACCF, with possible advantages in neurological and functional recovery. Further prospective multicentre studies are required for confirmation.

Abbreviations

ACCF

Anterior cervical corpectomy and fusion

3DP-AVB

Three-dimensional printed artificial vertebral body

n-HA/PA66

Nanohydroxyapatite/polyamide-66

TMC

Titanium mesh cage

ASD

Adjacent segment disease

BMD

Bone mineral density

BMI

Body mass index

CSF

Cerebrospinal fluid

CSM

Cervical spondylotic myelopathy

CT

Computed tomography

FSH

Fused segment height

ICC

Intraclass correlation coefficient

IPTW

Inverse probability of treatment weighting

JOA

Japanese orthopaedic association

MRI

Magnetic resonance imaging

NDI

Neck disability index

OPLL

Ossification of the posterior longitudinal ligament

SD

Standard deviation

VAS

Visual analog scale

Author contributions

C.D., Z.Z., J.Z., S.C., W.L., J.Y., and H.W. contributed to the study conception and design. C.D., Z.Z., J.Z., S.C., and W.L. collected the clinical and radiographic data. J.Y. and H.W. supervised the study. C.D., Z.Z., and J.Y. performed the data analysis and interpretation. C.D. and J.Y. drafted the manuscript. H.W. critically revised the manuscript for important intellectual content. All authors reviewed the manuscript, approved the final version, and agree to be accountable for all aspects of the work.

Funding

This work was supported by the Beijing Capital Health Development Research Special Project (Grant No. 2024–2-2233).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval and consent to participate

This study was approved by the Ethics Committees of both China-Japan Friendship Hospital and Beijing Hospital of Traditional Chinese Medicine. The requirement for written informed consent was waived due to the retrospective nature of the study, in accordance with national ethical regulations.

Competing Interests

The authors declare that they have no competing interests.