Does Bone Quality Predict Pseudarthrosis,Subsidence,or Loss of Lordosis After Anterior Cervical Discectomy and Fusion? Minimum 5-Year Cohort Study

Abstract

Study Design

Retrospective cohort study.

Objectives

To determine whether preoperative cervical bone quality assessed by computed tomography Hounsfield units (CT HU) predicts pseudarthrosis, cage subsidence, or loss of cervical lordosis after anterior cervical discectomy and fusion (ACDF) and to identify perioperative predictors.

Methods

Consecutive patients who underwent 1-3-level ACDF with ≥5-year follow-up were reviewed. Outcomes included neck/arm pain visual analog scale scores, Neck Disability Index, and C2-C7 lordosis and segmental height. Bone quality was assessed by preoperative CT HU. Subgroup comparisons, multivariable regression, and receiver operating characteristic (ROC) analyses evaluated predictors.

Results

Among 253 patients, subsidence, pseudarthrosis, and loss of lordosis occurred in 15 (5.9%), 26 (10.3%), and 41 (16.2%), respectively. Index HU and clinical outcomes did not differ between subgroups defined by subsidence or pseudarthrosis. HU ≤ 231.5 was associated with a higher incidence of loss of lordosis. On regression, higher BMI and greater segmental lengthening independently predicted subsidence, while multilevel surgery predicted pseudarthrosis. Greater preoperative C2-C7 lordosis predicted loss of lordosis; very low HU (≤231.5) was significant in univariable analysis but not after multivariable adjustment. On ROC analysis, Index HU did not predict subsidence (AUC: 0.511), pseudarthrosis (AUC: 0.414), or loss of lordosis (AUC: 0.551). In contrast, segmental lengthening predicted subsidence (AUC 0.695) with threshold around 3 mm (3.12 mm).

Conclusion

Preoperative cervical CT HU did not predict pseudarthrosis or subsidence over ≥5 years after ACDF; low HU showed an unadjusted association with loss of lordosis. Bone quality alone may be insufficient to justify avoiding ACDF. Limiting segmental lengthening to ∼3 mm may mitigate subsidence risk.

Keywords

anterior cervical discectomy and fusion osteoporosis bone quality computed tomography Hounsfield unit cage subsidence pseudarthrosis complication

Introduction

Anterior cervical discectomy and fusion (ACDF) is a commonly performed surgical procedure for degenerative cervical radiculopathy and myelopathy. Pseudarthrosis and cage subsidence are common complications following ACDF. These complications may result in suboptimal clinical outcomes and frequently necessitate revision surgeries.^1,2

Low bone mineral density (BMD) is a well-recognized risk factor for implant failure and subsidence in lumbar spine fusion.^3-6 However, BMD measured by dual-energy X-ray absorptiometry (DXA) has limitations because degenerative changes, such as osteophytes, can artificially increase the measured values.⁷ Computed tomography–derived Hounsfield units (CT HU) measured directly within the vertebral body have emerged as a practical surrogate marker of bone quality^8-11 and are increasingly used to predict postoperative complications.^12-14

However, the extent to which poor bone quality predisposes to complications such as pseudarthrosis, cage subsidence, and loss of lordosis after cervical surgery remains unclear. Several recent studies have reported that osteoporosis or osteopenia do not significantly predict pseudarthrosis or subsidence following ACDF.^15-17 Cadaveric studies have demonstrated substantial biomechanical differences between cervical and lumbar vertebrae, particularly in trabecular architecture and distribution.^18,19 Therefore, the relationship between bone quality and complications after ACDF may differ from that observed in lumbar spine surgery.

This study aimed (1) to determine whether poor cervical bone quality, assessed by CT HU, is associated with increased risk of pseudarthrosis, subsidence, or loss of lordosis following ACDF; and (2) to identify perioperative factors associated with these complications.

Methods

Study Design and Population

This retrospective study was approved by the Institutional Review Board of the participating institution (2025-07-009), and informed consent was waived because of the study’s retrospective design. We reviewed 1231 patients (age range, 27-82 years) who underwent one- to three-level ACDF performed by a single surgeon at a single center between February 2006 and January 2020. Inclusion criteria were: (1) a diagnosis of degenerative cervical disease (eg, degenerative disc disease, cervical spondylosis) causing radiculopathy, myelopathy, or myeloradiculopathy; (2) receipt of one- to three-level plated ACDF; (3) availability of preoperative CT suitable for HU measurement; and (4) a minimum follow-up of 5 years. Exclusion criteria were: (1) prior cervical surgery; (2) concomitant procedures such as posterior fusion, vertebral body sliding osteotomy, or corpectomy; (3) ACDF using a cage alone (without plate fixation); (4) surgery for tumor, trauma, or infection. Based on these criteria, 253 patients were included in the final analysis (Figure 1).

Figure 1.

Flow diagram of patient selection for the study

Surgical Technique

One surgeon performed all operations at a single center. The patient was positioned supine, and a standard left-sided Smith–Robinson approach was used to remove the intervertebral disc and osteophytes. Minimal decortication of the superior and inferior endplates was performed to minimize the risk of cage subsidence. Interbody cages were inserted at each level and packed with local autograft harvested during osteophytectomy and endplate preparation. When arm pain caused by foraminal stenosis was present, an additional uncoforaminotomy was performed. All operated levels were stabilized with an anterior cervical plate and variable-angle screw constructs to promote postoperative graft compression and gradual settling. The cage was positioned 2-3 mm posterior to the anterior vertebral body margin. A cervical collar was worn postoperatively for 6-12 weeks.^20-22

Variables

Patient-reported outcomes included neck pain visual analog scale (VAS), arm pain VAS, and the Neck Disability Index (NDI) scores, which were assessed preoperatively, at 2 years postoperatively, and at the final follow-up (≥5 years). Neck and arm pain were measured using a 10-point VAS.

Radiographic outcomes were evaluated across three domains at predefined time points. The literature employs varying definitions of cage subsidence, often based on reductions in intervertebral disc height.^23,24 Previous studies have demonstrated that changes in fused segmental body height correlate strongly with true subsidence²⁵; therefore, subsidence was defined as a >3-mm reduction in segmental height at the fused level(s), consistent with previous studies.^26-29 Segmental height was measured on 150%-magnified neutral lateral radiographs obtained preoperatively, within 1 week postoperatively, and at the final follow-up.¹⁶ Immediate postoperative segmental lengthening (increase from preoperative to immediate postoperative values) was prespecified as a surrogate of intraoperative distraction, and cage subsidence as the decrease from immediate postoperative to final follow-up^16,30 (Table 1) (Figure 2).

Table 1.

Variables Definitions and Assessment Timings

Variables	Definition	Timing
Bone quality (index HU)	CT-measured mean vertebral HU at the index level	Preoperative
Segmental lengthening	Immediate postoperative increase of segmental height (mm)	Preoperative–Immediate postoperative
Subsidence	≥3 mm loss of segmental height	Immediate postoperative–Final
Pseudarthrosis	Non-union confirmed by ISM or bridging bone	Final (5-year)
Loss of lordosis	≥5° loss of C2-C7 lordosis	Preoperative–Final
Revision	Reoperation due to pseudarthrosis or mechanical failure	≥ Final follow-up

HU, Hounsfield unit; CT, computed tomography; ISM, interspinous motion.

Figure 2.

Radiographic measurement of segmental height, segmental lengthening, and subsidence.

Pseudarthrosis was defined using two complementary criteria, consistent with prior validated methods^20,30-34: (1) interspinous motion (ISM) >1 mm on 150%-magnified dynamic lateral radiographs (flexion–extension), or (2) absence of bridging bone across the graft connecting adjacent endplates or outside the graft on CT. ISM and CT bridging were evaluated at 2 years and at the final follow-up, and union status or pseudarthrosis was recorded at each time point (Table 1) (Figure 3).

Figure 3.

Evaluation of pseudarthrosis using CT and dynamic radiographs.

Cervical alignment was assessed as C2-C7 lordosis on neutral lateral radiographs obtained preoperatively, within 1 week postoperatively, and at the final follow-up. Loss of lordosis was defined as a decrease of ≥5° at the final follow-up relative to the preoperative value^35,36 (Table 1) (Figure 2).

Cervical bone quality (Index HU) was derived from preoperative noncontrast CT by measuring HU on mid-sagittal vertebral body reformats using small trabecular regions of interest (ROIs) that excluded the cortical rim, endplate sclerosis, and osteophytes. HU values were obtained from each operative-level vertebral body, and the Index HU was defined as the mean of these values. For multilevel surgeries, the Index HU represented the average HU across all vertebral bodies included in the fusion construct²⁵ (Table 1) (Figure 4).

Figure 4.

Measurement of preoperative bone quality using Hounsfield units (HU).

Two fellowship-trained attending spine surgeons independently performed all radiographic measurements to assess interobserver reliability, with intraclass correlation coefficients (ICCs) ranging from 0.842 to 0.953.

Statistical Analysis

The normality of continuous variables was assessed using the Shapiro–Wilk test. Subgroup analyses were conducted based on the presence or absence of subsidence, pseudarthrosis, and loss of lordosis at the final follow-up, using independent t-tests for continuous variables and chi-square tests for categorical variables. Intergroup analyses also used published CT HU thresholds for cervical osteoporosis—231.5 in a Korean cohort and 326.5 in a U.S. cohort.^37,38

We conducted univariable and multivariable regression analyses to identify predictors of complications (subsidence, pseudarthrosis, and loss of lordosis). Candidate covariates included demographic variables and relevant preoperative and intraoperative factors. Variables with a screening P-value of ≤.100 in univariable analyses were entered into multivariable models.

Receiver operating characteristic (ROC) curves were constructed to evaluate the ability of the preoperative Index HU to predict pseudarthrosis, cage subsidence, and loss of lordosis. For each outcome, the area under the curve (AUC) and 95% confidence intervals (CI) were calculated. The optimal cutoff was determined by maximizing the Youden index (sensitivity + specificity − 1), and the corresponding sensitivity and specificity were reported. An identical ROC analysis was performed for postoperative segmental lengthening. All analyses were performed using IBM SPSS Statistics, version 29.0 (IBM Corp., Armonk, NY, USA), with statistical significance set at P < .05.

Results

Baseline Patient Characteristics

Among 253 patients, the distribution of operated levels was 1-level in 113 patients (44.7%), 2-level in 98 (38.7%), and 3-level in 42 (16.6%). The mean Index HU was 324.67 ± 77.07. At the final follow-up, subsidence occurred in 15 patients (5.9%). Pseudarthrosis was identified in 79 patients (31.2%) at 2 years; 67.1% (53/79) subsequently achieved delayed union, resulting in a final pseudarthrosis rate of 26 patients (10.3%) (Figure 5). Loss of lordosis was observed in 41 patients (16.2%). Subgroup analysis by subsidence status showed no significant difference in Index HU between groups (subsidence: 328.85 ± 70.76 vs no subsidence: 324.40 ± 77.58; P = .829) or in other demographic factors. Similarly, when stratified by pseudarthrosis status at the final follow-up, Index HU values did not differ significantly (pseudarthrosis: 307.53 ± 87.77 vs union: 323.63 ± 75.72; P = .232). When stratified by loss of lordosis, the mean Index HU was comparable between groups (P = .269). Notably, the proportion of patients with HU ≤ 231.5 was higher in the loss subgroup than in the no loss of lordosis subgroup (19.5% vs 8.5%; P = .047). The pseudarthrosis subgroup had a higher body mass index (BMI) (P = .047) and a greater proportion of smokers (P = .036). The distribution of underlying pathology differed between groups (P = .035), and multilevel procedures were significantly more frequent in the pseudarthrosis subgroup than in the union subgroup (P < .001) (Table 2, Supplemental Table 1).

Figure 5.

Change in union status from 2 years to ≥5 years after ACDF.

Table 2.

Baseline Characteristics of Included Patients and Subgroup Comparisons by Subsidence and Pseudarthrosis at Final Follow-Up

Characteristics	Included patients (n = 253)	Subsidence subgroup			Pseudarthrosis subgroup
Characteristics	Included patients (n = 253)	Subsidence (n = 15)	No subsidence (n = 238)	P value	Pseudarthrosis (n = 26)	Union (n = 227)	P value
Age (year)	57.23 ± 11.20	61.53 ± 11.02	56.95 ± 11.18	.125	57.85 ± 9.05	57.15 ± 11.43	.722
Sex (Female)	120 (47.4%)	6 (40.0%)	114 (47.9%)	.552	10 (38.5%)	110 (48.5%)	.334
BMI (kg/m²)	24.48 ± 3.22	26.14 ± 5.14	24.37 ± 3.05	.105	25.66 ± 3.72	24.34 ± 3.14	.047*
DM	42 (16.6%)	3 (20.0%)	39 (16.4%)	.721	4 (15.4%)	38 (16.7%)	.560
HTN	73 (28.9%)	5 (33.3%)	68 (28.6%)	.770	11 (42.3%)	62 (27.3%)	.110
Smoking	98 (38.7%)	7 (46.7%)	91 (38.2%)	.516	15 (57.7%)	83 (36.6%)	.036*
Index HU	324.67 ± 77.07	328.85 ± 70.76	324.40 ± 77.58	.829	307.53 ± 87.77	323.63 ± 75.72	.232
Pathology				.605			.035*
Myelopathy	118 (46.6%)	6 (40.0%)	112 (47.1%)		7 (26.9%)	111 (48.9%)
Radiculopathy	77 (30.4%)	5 (33.3%)	72 (30.3%)		10 (38.5%)	67 (29.5%)
Myeloradiculopathy	58 (22.9%)	4 (26.7%)	54 (22.7%)		9 (34.6%)	49 (21.6%)
Number of levels operated				.774			<.001*
1–level	113 (44.7%)	7 (46.7%)	106 (44.5%)		4 (15.4%)	109 (48.8%)
2–level	98 (38.7%)	6 (40.0%)	92 (38.7%)		13 (50.0%)	85 (37.4%)
3–level	42 (16.6%)	2 (13.3%)	40 (16.8%)		9 (34.6%)	33 (14.5%)
Follow-up period (month)	89.25 ± 22.70	98.13 ± 35.26	88.69 ± 21.66	.322	90.12 ± 25.73	89.15 ± 22.39	.838

BMI, body mass index; DM, diabetes mellitus; HTN, hypertension; Index HU, CT-measured mean vertebral Hounsfield Unit at index level.

*P < 0.050.

Outcome Comparison Based on Complications

Clinical and radiographic outcomes were compared between subgroups stratified by each complication. Stratification by subsidence revealed no preoperative or final follow-up differences in neck pain VAS or NDI, although the subsidence group showed higher preoperative arm pain VAS (P = .006). Radiographically, C2-C7 lordosis did not differ between groups at either time point, and the proportion of patients with loss of lordosis was comparable (P = .780). Segmental height was similar between subgroups at baseline and the final follow-up; however, postoperative segmental lengthening was significantly greater in the subsidence group (2.81 ± 1.64 vs 1.59 ± 1.68 mm; P = .007). The subsidence group also showed higher rates of pseudarthrosis at final follow-up (40.0% vs 8.4%; P = .002) and revision surgery (20.0% vs 5.0%; P = .050) (Table 3).

Table 3.

Clinical and Radiologic Outcomes at Final Follow-Up After ACDF; Comparison Based on Subsidence and Pseudarthrosis

Variables		Subsidence subgroup			Pseudarthrosis subgroup
Variables		Subsidence (n = 15)	No subsidence (n = 238)	P value	Pseudarthrosis (n = 26)	Union (n = 227)	P value
Clinical outcomes
Neck pain VAS	Preoperartive	4.27 ± 3.53	3.96 ± 2.73	.371	4.65 ± 2.80	3.89 ± 2.77	.187
Neck pain VAS	Final	2.36 ± 2.58	2.43 ± 2.17	.461	2.56 ± 2.41	2.41 ± 2.18	.794
Arm pain VAS	Preoperartive	6.80 ± 2.14	5.02 ± 2.70	.006*	6.35 ± 2.21	4.98 ± 2.72	.014*
Arm pain VAS	Final	3.64 ± 2.29	3.45 ± 2.64	.410	3.22 ± 2.34	3.49 ± 2.65	.685
NDI	Preoperative	18.60 ± 12.20	14.96 ± 8.28	.096	17.59 ± 9.69	14.91 ± 8.40	.219
NDI	Final	10.45 ± 9.09	8.34 ± 6.56	.157	9.56 ± 8.10	8.35 ± 6.58	.473
Radiologic outcomes
C2-C7 lordosis	Preoperartive	5.65 ± 14.55	5.91 ± 11.19	.466	2.54 ± 12.40	6.28 ± 11.23	.113
C2-C7 lordosis	Final	9.85 ± 12.70	9.97 ± 9.70	.482	7.94 ± 12.53	10.20 ± 9.52	.270
Loss of lordosis		1 (6.7%)	40 (16.8%)	.501	4 (16.0%)	37 (16.2%)	1.000
Segmental height	Preoperartive	33.27 ± 2.36	32.33 ± 2.81	.103	33.14 ± 2.20	32.30 ± 2.84	.152
Segmental height	Final	32.54 ± 2.35	33.15 ± 2.88	.209	32.97 ± 2.49	33.13 ± 2.89	.786
Segmental lengthening		2.81 ± 1.64	1.59 ± 1.68	.007*	1.59 ± 1.80	1.67 ± 1.69	.826
Subsidence	Final	NA	NA	NA	6 (23.1%)	9 (4.0%)	.002*
Pseudarthrosis	Final	6 (40.0%)	20 (8.4%)	.002*	NA	NA	NA
Revision	After final	3 (20.0%)	12 (5.0%)	.050*	4 (5.9%)	11 (4.8%)	.055

VAS, visual analog scale; NDI, neck disability index.

*P < .050.

Stratification by pseudarthrosis revealed no between-group differences in neck pain VAS or NDI, except for higher preoperative arm pain VAS in the pseudarthrosis subgroup (P = .014). Radiographically, there were no significant between-group differences preoperatively or at the final follow-up. Subsidence occurred more frequently in the pseudarthrosis subgroup than in the union subgroup (23.1% vs 4.0%; P =.002), while revision surgery rates did not differ (P = .055) (Table 3).

When stratified by loss of lordosis, no significant between-group differences were observed in preoperative or final patient-reported outcomes (VAS and NDI). Radiographically, the loss subgroup had greater preoperative C2-C7 lordosis (16.01 ± 8.98 vs 3.94 ± 10.76; P < .001), whereas the final C2-C7 lordosis was comparable between groups (P = .061). Postoperative segmental lengthening was modestly smaller in the loss subgroup (1.14 ± 1.61 vs 1.76 ± 1.70; P = .032). Revision surgery was numerically more frequent in the loss subgroup, but the difference was not significant (12.2% vs 4.7%; P = .075) (Supplemental Table 2).

Comparison of Postoperative Complications by HU Cutoff Subgroups

Previously proposed cervical CT HU thresholds for poor bone quality (231.5 and 326.5) were used to compare complication rates across subgroups. For the 231.5 cutoff, 26 patients had HU ≤ 231.5 and 227 had HU > 231.5; by the 326.5 cutoff, 128 patients had HU ≤ 326.5 and 125 had HU > 326.5. Stratification by the 231.5 cutoff showed no significant between-group differences in subsidence, pseudarthrosis, or revision surgery rates. Loss of lordosis was more frequent in patients with HU ≤ 231.5 (30.8% vs 14.5%; odds ratio [OR]: 2.613; 95% CI: 1.051-6.497; P = .039). Stratification by the 326.5 cutoff revealed no significant between-group differences in complication rates, including loss of lordosis (Table 4).

Table 4.

Postoperative Complications According to HU Cutoffs of 231.5 and 326.5

Characteristics	HU cutoff 231.5			HU cutoff 326.5
Characteristics	HU ≤ 231.5 (n = 26)	HU > 231.5 (n = 227)	P value	HU ≤ 326.5 (n = 128)	HU > 326.5 (n = 125)	P value
Subsidence	1 (3.8%)	14 (6.2%)	1.000	8 (6.3%)	7 (5.6%)	.827
Pseudarthrosis	5 (19.2%)	20 (8.8%)	.154	15 (11.7%)	10 (8.0%)	.322
Loss of lordosis	8 (30.8%)	33 (14.5%)	.039*	22 (17.2%)	19 (15.2%)	.796
Revision	2 (7.7%)	13 (5.7%)	.657	7 (5.5%)	8 (6.4%)	.754

HU, Hounsfield Unit.

*P < .050.

Regression and ROC Analysis for Predicting Complications

Univariable and multivariable regression analyses were performed to identify predictors of complications. Higher BMI (OR: 1.194; 95% CI: 1.031-1.383; P = .018) and greater segmental lengthening (OR: 1.721; 95% CI: 1.193-2.483; P = .004) independently predicted subsidence. Multilevel surgery significantly predicted pseudarthrosis (OR: 2.585; 95% CI: 1.476-4.527; P < .001). Greater preoperative C2-C7 lordosis predicted loss of lordosis (OR: 1.122; 95% CI: 1.077-1.169; P < .001). Although very low HU (≤231.5) was significantly associated with loss of lordosis in univariable analysis (OR: 2.613; 95% CI: 1.051-6.497; P = .039), this association was not significant after multivariable adjustment (OR: 1.294; 95% CI: 0.411-4.074; P = .660). Moreover, the Index HU and HU below cutoff were not associated with subsidence or pseudarthrosis (Tables 5-7).

Table 5.

Multivariate Regression Analysis for the Predictors of Subsidence

Predictors	Odds ratio	95% CI	P-value
Univariate analysis
Age	1.039	0.989-1.091	.127
Sex	0.725	0.250-2.101	.554
Pathology	1.184	0.625-2.243	.605
Operation level	0.899	0.434-1.860	.773
BMI	1.163	1.005-1.344	.042*
DM	1.276	0.344-4.732	.716
HTN	1.250	0.412-3.792	.694
Smoking	1.413	0.496-4.029	.517
Preoperartive C2-C7 lordosis	0.998	0.953-1.045	.931
Preoperartive SVA	0.988	0.935-1.044	.672
Segmental lengthening	1.629	1.141-2.327	.007*
CT HU at index level	1.001	0.994-1.008	.828
HU ≤ 326.5	1.124	0.395-3.198	.827
HU ≤ 231.5	0.609	0.077-4.826	.638
Multivariate analysis
BMI	1.194	1.031-1.383	.018*
Segmental lengthening	1.721	1.193-2.483	.004*

VAS, visual analog scale; NDI, neck disability index; BMI, body mass index; DM, diabetes mellitus; HTN, hypertension; SVA, sagittal vertical axis.

*P < .050.

Table 6.

Multivariate Regression Analysis for the Predictors of Pseudarthrosis

Predictors	Odds ratio	95% CI	P-value
Univariate analysis
Age	1.006	0.970-1.043	.765
Sex	0.665	0.289-1.527	.336
Pathology	1.694	1.029-2.788	.038*
Operation level	2.585	1.476-4.527	<.001*
BMI	1.126	1.000-1.269	.050*
DM	0.904	0.295-2.774	.860
HTN	1.952	0.850-4.480	.115
Smoking	2.366	1.038-5.391	.040*
Preoperartive C2-C7 lordosis	0.971	0.937-1.007	.113
Preoperartive SVA	0.989	0.948-1.032	.616
Segmental lengthening	0.975	0.769-1.236	.834
CT HU at index level	0.997	0.991-1.002	.231
HU ≤ 326.5	1.643	0.715-3.774	.242
HU ≤ 231.5	2.336	0.798-6.834	.122
Multivariate analysis
Pathology	1.540	0.933-2.541	.091
Operation level	2.585	1.476-4.527	<.001*
BMI	1.128	0.993-1.281	.063
Smoking	1.884	0.797-4.453	.149

VAS, visual analog scale; NDI, neck disability index; CI, confidence interval; BMI, body mass index; DM, diabetes mellitus; HTN, hypertension; SVA, sagittal vertical axis.

*P < .050.

Table 7.

Multivariate Regression Analysis for the Predictors of Loss of Lordosis

Predictors	Odds ratio	95% CI	P-value
Univariate analysis
Age	1.002	0.973-1.033	.882
Sex	0.750	0.381-1.475	.404
Pathology	1.130	0.748-1.708	.562
Operation level	0.687	0.422-1.120	.132
BMI	1.043	0.942-1.155	.413
DM	1.269	0.540-2.985	.585
HTN	1.345	0.660-2.740	.415
Smoking	1.636	0.835-3.207	.152
Preoperartive C2-C7 lordosis	1.121	1.076-1.166	<.001*
Preoperartive SVA	0.982	0.947-1.017	.304
Segmental lengthening	0.812	0.669-0.985	.035*
CT HU at index level	0.997	0.993-1.002	.268
HU ≤ 326.5	1.158	0.592-2.263	.668
HU ≤ 231.5	2.613	1.051-6.497	.039*
Multivariate analysis
Preoperartive C2-C7 lordosis	1.122	1.077-1.169	<.001*
Segmental lengthening	0.800	0.630-1.017	.068
HU ≤ 231.5	1.294	0.411-4.074	.660

VAS, visual analog scale; NDI, neck disability index; CI, confidence interval; BMI, body mass index; DM, diabetes mellitus; HTN, hypertension; SVA, sagittal vertical axis.

*P < .050.

ROC analysis showed that preoperative Index HU did not predict subsidence (AUC: 0.511; 95% CI: 0.372-0.649; P = .882), pseudarthrosis (AUC: 0.414; 95% CI: 0.298-0.530; P = .147) or loss of lordosis (AUC: 0.551; 95% CI: 0.447-0.655; P = .338). In contrast, segmental lengthening significantly predicted subsidence (AUC: 0.695; 95% CI: 0.550-0.840; P = .008), with an optimal cutoff of 3.12 mm (Figure 6).

Figure 6.

Receiver operating characteristic curve analysis for postoperative segmental lengthening.

Discussion

This study assessed whether complications after ACDF—pseudarthrosis, subsidence, and loss of lordosis—were predicted by preoperative bone quality measured by CT HU over at least 5 years of follow-up. Cervical bone quality did not predict pseudarthrosis or subsidence, but lower HU values were associated with postoperative loss of cervical lordosis. Immediate postoperative segmental lengthening—used as a surrogate for intraoperative distraction—was associated with subsidence; a threshold of approximately 3 mm may guide operative planning. Overall, these findings highlight that differences in cervical and lumbar biomechanics warrant region-specific interpretation of bone quality.

Although lumbar fusion studies consistently linked osteoporosis-related bone quality to adverse outcomes and implant complications,^4-6 recent cervical studies report no significant association between bone quality and pseudarthrosis or subsidence after ACDF.^15-17 These findings support this site-specific distinction. Cadaveric studies by Schröder et al showed that cervical vertebrae have thicker, denser trabeculae and less age-related trabecular loss than lumbar counterparts.^18,19 MRI-based vertebral bone quality studies similarly reveal distinct bone-quality profiles between the cervical and lumbar regions.³⁹ These microarchitectural features likely increase load-bearing capacity per unit area, suggesting greater cervical spine resilience to postoperative mechanical failure, even in osteoporotic bone.

Mechanically, ACDF involves a short moment arm between the plate and the fusion interface (cage–endplate), and the plate acts as a tension band limiting micromotion at the interbody surface.^40-42 In contrast, posterior lumbar constructs have longer distances between pedicle screw–rod anchors and the fusion segment, transmitting greater bending moments to the anterior column under similar loads.^43,44 These structural differences may explain why reduced bone quality does not directly lead to higher pseudarthrosis or subsidence rates following ACDF (Figure 7).

Figure 7.

Moment-arm contrast between ACDF and posterior lumbar constructs.

In this cohort, pseudarthrosis, subsidence, and loss of lordosis had limited effects on clinical outcomes, including pain scales and NDI. Notably, 67.1% of pseudarthroses identified at 2 years achieved delayed union by final follow-up, consistent with previous ACDF reports.^20,32 Although the subsidence group had a higher revision rate, interpretation is limited by the small subgroup size (n = 15) and the few revision events. Revision rates did not differ significantly when outcomes were stratified by pseudarthrosis or loss of lordosis. Overall, the anatomic advantages of the cervical spine and the stability provided by the short moment arm in ACDF may mitigate both the incidence and clinical impact of mechanical complications, even in patients with low bone quality.

CT HU-based bone quality did not meaningfully predict pseudarthrosis or subsidence, suggesting that bone quality alone may be insufficient as a rationale to avoid ACDF. In clinical practice, low bone quality identified by screening measures (eg, BMD) should not, by itself, justify forgoing an anterior approach (potential undertreatment) or planning overly extensive combined procedures (potential overtreatment). In additional analyses, very low HU (≤231.5) was associated with an increased risk of postoperative loss of lordosis. The prevalence of HU ≤ 231.5 was higher in the loss subgroup, and a significant association was observed in univariable regression; however, this association was not retained after multivariable adjustment, suggesting limited independent predictive value. Few studies have directly linked preoperative bone-quality metrics to postoperative loss of lordosis after ACDF. These findings indicate that bone quality is more closely associated with alignment maintenance than traditional mechanical complications. Because postoperative loss of lordosis is strongly influenced by baseline alignment and extent of alignment correction, bone quality may be most appropriately interpreted alongside alignment-related variables rather than as a stand-alone metric.

Excessive segmental lengthening was associated with increased subsidence risk, consistent with previous studies^30,45; this study corroborates that association in a large cohort and identifies a threshold around 3 mm (3.12 mm). Limiting segmental lengthening relative to preoperative disc height to approximately 3 mm may serve as a pragmatic guardrail to mitigate subsidence risk. Because higher BMI also independently predicted subsidence in this cohort, this threshold may be particularly relevant for patients with high BMI. However, univariable analyses suggested a directional trade-off in which smaller segmental lengthening was associated with more frequent loss of lordosis, indicating that the 3-mm threshold is best viewed as a guide to avoid excessive distraction rather than a universal rule for all patients. Accordingly, flexible application tailored to individual alignment, target correction, and clinical necessity may be required, and further studies are needed to refine distraction strategies that balance these competing risks.

Limitations

This study has several limitations. First, although the sample size was large and the follow-up was long-term, the retrospective design could not eliminate selection bias. Second, because the study was conducted at a single center, its findings may have limited generalizability. Third, although DXA-derived BMD is a widely used index of bone quality, it is not routinely obtained in cervical surgery candidates; therefore, it was not included in the analyses. Fourth, adjacent segment degeneration, an important long-term complication, was not systematically assessed, precluding evaluation of its relationship with bone quality and postoperative alignment. Finally, the study design did not allow determination of the temporal sequence among pseudarthrosis, subsidence, and loss of lordosis, precluding causal inference; future prospective studies are needed to clarify these relationships.

Conclusion

In this 5-year longitudinal cohort, preoperative CT HU did not predict pseudarthrosis or subsidence after ACDF. Very low HU (≤231.5) was associated with postoperative loss of lordosis in univariable analysis, but did not retain independent predictive value after multivariable adjustment. Notably, these complications had limited impact on 5-year patient-reported outcomes. Immediate postoperative segmental lengthening (a surrogate of intraoperative distraction), and higher BMI independently increased cage subsidence risk, whereas multilevel surgery independently predicted pseudarthrosis. These findings indicate that bone quality in the cervical spine should be interpreted differently from that in the lumbar spine, particularly with respect to mechanical failure. Accordingly, bone quality alone should not be used as a primary rationale to avoid ACDF. A pragmatic guardrail of approximately 3 mm of segmental lengthening relative to preoperative disc height may help mitigate subsidence risk.

Supplemental Material

Supplemental Material - Does Bone Quality Predict Pseudarthrosis, Subsidence, or Loss of Lordosis After Anterior Cervical Discectomy and Fusion? Minimum 5-Year Cohort Study

Supplemental Material for Does Bone Quality Predict Pseudarthrosis, Subsidence, or Loss of Lordosis After Anterior Cervical Discectomy and Fusion? Minimum 5-Year Cohort Study by San Kim, MD, Sehan Park, MD, PhD, Dong Ki Ahn, MD, PhD, Dong-Ho Lee, MD, PhD, Gumin Jeong, MD, Ki Chol Park, MD, Yong Kim, MD, and Changmin Choi, MD, PhD in Global Spine Journal

Footnotes

ORCID iDs

San Kim

Sehan Park

Dong-Ho Lee

Gumin Jeong

Ethical Considerations

This retrospective study was approved by the Institutional Review Board of the participating institution (IRB approval number: 2025-07-009).

Consent to Participate

Informed consent was waived because of the study’s retrospective design.

Author Contributions (CRediT)

Conceptualization: San Kim, Dong-Ho Lee.

Methodology: San Kim, Dong Ki Ahn, Ki Chol Park.

Formal analysis: San Kim, Sehan Park.

Investigation: San Kim.

Data curation: San Kim, Sehan Park, Gumin Jeong, Ki Chol Park, Yong Kim, Changmin Choi.

Validation: Sehan Park, Dong-Ho Lee.

Visualization: Changmin Choi.

Project administration: Dong Ki Ahn, Dong-Ho Lee.

Funding acquisition: San Kim.

Resources: Gumin Jeong.

Software: Yong Kim.

Supervision: Dong Ki Ahn, Dong-Ho Lee.

Writing – original draft: San Kim.

Writing – review & editing: Sehan Park, Dong-Ho Lee.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by a research grant from Jeju National University Hospital in 2025 (grant number 2025-22).

Declaration of Conflicting Interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Supplemental Material

Supplemental material for this article is available online.

Appendix

References

Lee

Kim

Park

. Risk factors for postoperative subsidence of single-level anterior cervical discectomy and fusion: the significance of the preoperative cervical alignment. Spine. 2014;39(16):1280-1287. doi:10.1097/brs.0000000000000400

Brenke

Dostal

Scharf

Weiß

Schmieder

Barth

. Influence of cervical bone mineral density on cage subsidence in patients following stand-alone anterior cervical discectomy and fusion. Eur Spine J. 2015;24(12):2832-2840. doi:10.1007/s00586-014-3725-9

Chin

Park

Yoon

, et al. Prevalence of osteoporosis in patients requiring spine surgery: incidence and significance of osteoporosis in spine disease. Osteoporos Int. 2007;18(9):1219-1224. doi:10.1007/s00198-007-0370-8

Keaveny

Adams

Fischer

, et al. Increased risks of vertebral fracture and reoperation in primary spinal fusion patients who test positive for osteoporosis by biomechanical computed tomography analysis. Spine J. 2023;23(3):412-424. doi:10.1016/j.spinee.2022.10.018

Levy

Magera

Messer

, et al. Lumbar endplate Hounsfield units enhance transforaminal lumbar interbody fusion subsidence prediction compared to trabecular Hounsfield units alone. Glob Spine J. 2025;16:21925682251356986. doi:10.1177/21925682251356986

Löffler

Sollmann

Burian

, et al. Opportunistic osteoporosis screening reveals low bone density in patients with screw loosening after lumbar semi-rigid instrumentation: a case-control study. Front Endocrinol. 2021;11:552719. doi:10.3389/fendo.2020.552719

Rand

Seidl

Kainberger

, et al. Impact of spinal degenerative changes on the evaluation of bone mineral density with dual energy X-ray absorptiometry (DXA). Calcif Tissue Int. 1997;60(5):430-433. doi:10.1007/s002239900258

Choi

Kim

Lim

. Diagnostic efficacy of Hounsfield units in spine CT for the assessment of real bone mineral density of degenerative spine: correlation study between T-scores determined by DEXA scan and Hounsfield units from CT. Acta Neurochir. 2016;158(7):1421-1427. doi:10.1007/s00701-016-2821-5

Ahmad

Crawford

3rd Glassman

Dimar

2nd Gum

Carreon

. Correlation between bone density measurements on CT or MRI versus DEXA scan: a systematic review. N Am Spine Soc J. 2023;14:100204. doi:10.1016/j.xnsj.2023.100204

10.

Nagata

Glassman

Dimar

2nd , et al. Comparison of bone mineral density in children and adolescents on CT versus DEXA scan. Spine. 2024;49(19):E322-e326. doi:10.1097/brs.0000000000004877

11.

Zaidi

Danisa

Cheng

. Measurement techniques and utility of Hounsfield unit values for assessment of bone quality prior to spinal instrumentation: a review of current literature. Spine. 2019;44(4):E239-E244. doi:10.1097/brs.0000000000002813

12.

Egu

Najjar

Komaitis

, et al. Evaluating the prognostic role of computed tomography Hounsfield units in anticipating spinal outcomes post-instrumentation: a systematic review and meta-analysis. Eur Spine J. 2025;34(4):1420-1432. doi:10.1007/s00586-025-08737-9

13.

Chanbour

Steinle

Chen

, et al. The importance of Hounsfield units in adult spinal deformity surgery: finding an optimal threshold to minimize the risk of mechanical complications. J Spine Surg. 2023;9(2):149-158. doi:10.21037/jss-22-102

14.

Duan

Mummaneni

Rivera

, et al. The association between lower Hounsfield units of the upper instrumented vertebra and proximal junctional kyphosis in adult spinal deformity surgery with a minimum 2-year follow-up. Neurosurg Focus. 2020;49(2):E7. doi:10.3171/2020.5.Focus20192

15.

Baumann

Trager

Gong

Anaspure

Strony

Aleem

. Osteoporosis is not associated with reoperation or pseudarthrosis after anterior cervical discectomy and fusion through 4-years’ follow-up: a retrospective cohort study of US academic health centers. Spine J. 2025;25(2):290-298. doi:10.1016/j.spinee.2024.09.031

16.

Hong

Nam

Lee

. Effect of osteoporosis on clinical and radiological outcomes following one-level anterior cervical discectomy and fusion. Asian Spine J. 2024;18(2):182-189. doi:10.31616/asj.2023.0207

17.

Ahmed

Duey

Rajjoub

, et al. Role of preoperative Hounsfield units in predicting subsidence after anterior cervical discectomy and fusion in the United States: a retrospective analysis including osteopenia diagnosis. Asian Spine J. 2025;19(4):609-618. doi:10.31616/asj.2024.0214

18.

Schröder

Reichel

Spiegel

, et al. Breaking strength and bone microarchitecture in osteoporosis: a biomechanical approximation based on load tests in 104 human vertebrae from the cervical, thoracic, and lumbar spines of 13 body donors. J Orthop Surg Res. 2022;17(1):228. doi:10.1186/s13018-022-03105-5

19.

Schröder

Baginski

Schulze

, et al. Regional variations in the intra- and intervertebral trabecular microarchitecture of the osteoporotic axial skeleton. Anat Sci Int. 2023;98(4):566-579. doi:10.1007/s12565-023-00726-6

20.

Lee

D-H

Cho

Hwang

, et al.

What is the fate of pseudarthrosis detected 1 year after anterior cervical discectomy and fusion?

Spine. 2018;43(1):E23-E28. doi:10.1097/brs.0000000000002077

21.

Park

Lee

Hwang

Yang

Cho

. Anterior decompression and fusion for the treatment of cervical myelopathy caused by ossification of the posterior longitudinal ligament: a narrative review. Asian Spine J. 2023;17(3):582-594. doi:10.31616/asj.2022.0003

22.

Lee

D-H

Lee

Riew

. An algorithmic roadmap for the surgical management of degenerative cervical myelopathy: a narrative review. Asian Spine J. 2024;18(2):274-286. doi:10.31616/asj.2023.0413

23.

W-J

Jiang

L-S

Liang

Dai

L-Y

. Cage subsidence does not, but cervical lordosis improvement does affect the long-term results of anterior cervical fusion with stand-alone cage for degenerative cervical disc disease: a retrospective study. Eur Spine J. 2012;21(7):1374-1382. doi:10.1007/s00586-011-2131-9

24.

Bartels

Donk

Feuth

. Subsidence of stand-alone cervical carbon fiber cages. Neurosurgery. 2006;58(3):502-508. doi:10.1227/01.Neu.0000197258.30821.50

25.

Lee

You

Kim

Sung

Hong

. Significance of cervical spine computed tomography Hounsfield units to predict bone mineral density and the subsidence after anterior cervical discectomy and fusion. Clin Spine Surg. 2021;34(8):E450-E457. doi:10.1097/bsd.0000000000001218

26.

Feng

, et al. Cervical vertebra bone quality score predicts zero-profile anchored spacer interbody fusion cage subsidence after anterior cervical diskectomy and fusion: a retrospective study. Glob Spine J. 2024;15:21925682241280258. doi:10.1177/21925682241280258

27.

Noordhoek

Koning

Jacobs

WCH

Vleggeert-Lankamp

CLA

. Incidence and clinical relevance of cage subsidence in anterior cervical discectomy and fusion: a systematic review. Acta Neurochir. 2018;160(4):873-880. doi:10.1007/s00701-018-3490-3

28.

Wang

Mummaneni

, et al. Lower Hounsfield units on CT are associated with cage subsidence after anterior cervical discectomy and fusion. J Neurosurg: Spine SPI. 2020;33(4):425-432. doi:10.3171/2020.3.SPINE2035

29.

Safdar

Headley

Rommelman

Haseeb

Motiei-Langroudi

. The effect of interbody cage parameters on the rate of subsidence in single-level anterior cervical discectomy and fusion (ACDF): a retrospective analysis of 98 patients. Cureus. 2023;15(12):e50386. doi:10.7759/cureus.50386

30.

Lee

Y-S

Kim

Y-B

Park

S-W

31.

Riew

Yang

Chang

D-G

, et al.

What is the most accurate radiographic criterion to determine anterior cervical fusion?

Spine J. 2019;19(3):469-475. doi:10.1016/j.spinee.2018.07.003

32.

Lee

Park

Seok

, et al. Fate of pseudarthrosis detected 2 years after anterior cervical discectomy and fusion: results of a minimum 5-year follow-up. Spine J. 2023;23(12):1790-1798. doi:10.1016/j.spinee.2023.07.016

33.

Chanbour

Bendfeldt

Johnson

, et al. Longer screws decrease the risk of radiographic pseudarthrosis following elective anterior cervical discectomy and fusion. Glob Spine J. 2023;15:21925682231214361. doi:10.1177/21925682231214361

34.

Zuckerman

Devin

. Pseudarthrosis of the cervical spine. Clin Spine Surg. 2022;35(3):97-106. doi:10.1097/bsd.0000000000001259

35.

Sakai

Yoshii

Hirai

, et al. Cervical sagittal imbalance is a predictor of kyphotic deformity after laminoplasty in cervical spondylotic myelopathy patients without preoperative kyphotic alignment. Spine. 2016;41(4):299-305. doi:10.1097/brs.0000000000001206

36.

X-Y

Wang

Zhu

W-G

Kong

S-B

. Impact of cervical and global spine sagittal alignment on cervical curvature changes after posterior cervical laminoplasty. J Orthop Surg Res. 2022;17(1):521. doi:10.1186/s13018-022-03421-w

37.

Han

You

Lee

Kim

Hong

Sung

. Hounsfield unit measurement method and related factors that most appropriately reflect bone mineral density on cervical spine computed tomography. Skelet Radiol. 2022;51(10):1987-1993. doi:10.1007/s00256-022-04050-4

38.

Fluss

Lo Bu

De la Garza Ramos

Murthy

Yassari

Gelfand

. Nonquantitative CT scan Hounsfield unit as a determinant of cervical spine bone density. J Neurosurg Spine. 2024;41(6):803-812. doi:10.3171/2024.5.Spine24249

39.

Razzouk

Ramos

Ouro-Rodrigues

, et al. Comparison of cervical, thoracic, and lumbar vertebral bone quality scores for increased utility of bone mineral density screening. Eur Spine J. 2023;32(1):20-26. doi:10.1007/s00586-022-07484-5

40.

Dvorak

Pitzen

Zhu

Gordon

Fisher

Oxland

. Anterior cervical plate fixation: a biomechanical study to evaluate the effects of plate design, endplate preparation, and bone mineral density. Spine. 2005;30(3):294-301. doi:10.1097/01.brs.0000152154.57171.92

41.

Fogel

Liu

Liao

Zhou

. In vitro evaluation of stiffness and load sharing in a two-level corpectomy: comparison of static and dynamic cervical plates. Spine J. 2010;10(5):417-421. doi:10.1016/j.spinee.2010.02.004

42.

Brodke

Klimo

Jr Bachus

Braun

Dailey

. Anterior cervical fixation: analysis of load-sharing and stability with use of static and dynamic plates. J Bone Joint Surg Am. 2006;88(7):1566-1573. doi:10.2106/jbjs.E.00305

43.

Chun

Baker

Hsu

. Lumbar pseudarthrosis: a review of current diagnosis and treatment. Neurosurgical Focus FOC. 2015;39(4):E10. doi:10.3171/2015.7.FOCUS15292

44.

Levy

Pinter

Reed

, et al. Transforaminal lumbar interbody fusion subsidence: computed tomography analysis of incidence, associated risk factors, and impact on outcomes. J Neurosurg Spine. 2024;41(4):463-472. doi:10.3171/2024.4.SPINE231007

45.

Yamagata

Takami

Uda

, et al. Outcomes of contemporary use of rectangular titanium stand-alone cages in anterior cervical discectomy and fusion: cage subsidence and cervical alignment. J Clin Neurosci. 2012;19(12):1673-1678. doi:10.1016/j.jocn.2011.11.043

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.12 MB