Anterior-only decompression insufficient in concentric stenosis in degenerative cervical myelopathy

Abstract

Aims & objectives

The compression of the spinal cord can result in degenerative cervical myelopathy (DCM), a condition characterized by progressive neurological dysfunction.

This study examines clinical outcomes after anterior decompression and predictors of secondary posterior decompression after initial anterior surgery.

Materials & methods

We conducted a retrospective review of 94 consecutive patients who underwent anterior-only cervical decompression in DCM. Preoperative variables (demographics, neurological and clinical status, spine anatomy) were entered into a multivariate logistic regression to determine associations with secondary posterior decompression. Neurologic function was assessed by the JOA score.

Results

Nine of 94 patients (9.6 %) underwent secondary posterior decompression for persistent myelopathic symptoms, of which all presented with a concentric spinal canal stenosis. In multivariate analysis, higher ASA score (p = 0.042), concentric stenosis (p = 0.008) and osteoporosis (p = 0.044) were independently associated with increased revision risk. At six weeks, the anterior-only cohort achieved a mean JOA score of 15.44, compared to 12.60 in those undergoing revision (measured pre-revision).

Conclusion

Concentric stenosis morphology, compromised bone quality, and elevated ASA status independently predict failure of anterior-only decompression. In patients exhibiting this high-risk profile, a primary combined anterior–posterior approach should be strongly considered to ensure circumferential decompression and optimize early neurologic recovery.

1. Introduction

Degenerative cervical myelopathy (DCM) is a degenerative condition resulting in a variety of neurological symptoms: gait disorder, fine motor impairment, sensory deficits, and muscle weakness.^1,2 Common cause is a cervical spinal canal stenosis compressing the spinal cord. Pathophysiologically, this is caused by degenerative structural changes of the cervical spinal canal such as disc degeneration, osteophyte formation, ossification of the posterior longitudinal ligament (OPLL) or hypertrophy of the ligamenta flava.³

The narrowing of the spinal canal can be either anterior, posterior, or antero-posterior (e.g. concentric). In concentric stenoses, the spinal cord is compressed from both sides, which can significantly impair neurological function.

Surgical decompression remains the mainstay of treatment for DCM.⁴ Both anterior approaches (e.g., anterior cervical discectomy and fusion, multilevel corpectomy)⁵ and posterior techniques (e.g., laminoplasty, laminectomy with or without fusion) achieve comparable neural recovery.⁶ Deciding which route to pursue requires careful integration of multiple factors: the anatomical pattern and number of levels of cord compression, sagittal alignment and segmental instability, patient comorbidities (including osteoporosis, rheumatoid arthritis, ankylosing spondylitis).

To systematize this decision-making, the “7-Letter Code” (7LC) was introduced by Papavero as a didactic tool.⁷ Seven key imaging and clinical parameters are each categorized as favoring an anterior, posterior or combined approach.

These parameters include 1.) the type of spinal cord compression (e.g. anterior, posterior or concentric), 2.) the extension of the spinal canal compression, 3.) the sagittal alignment of the cervical spine (e.g. physiological lordosis vs. pathological kyphosis), 4.) segmental instability, 5.) overall patient morbidity (e.g. assessed by the ASA classification), 6.) underlying bone pathology such as osteoporosis, rheumatoid arthritis or ankylosing spondylodiscitis and 7.) ventral pathology characteristics inside the spinal canal.

In this study, we aim to analyze the clinical outcomes and identify the predictors for the need of a secondary posterior decompression following an initial anterior surgical approach in patients with DCM. By focusing on preoperative factors such as the neurological status, the type of spinal cord compression, and the application of the 7-Letter Code, we seek to provide insights into optimizing surgical decision-making and improving patient outcomes.

2. Materials and methods

2.1. Data collection and study design

In this retrospective study, we collected data from our clinic who underwent anterior decompression of the cervical spine between 2020 and 2024. The data was extracted from the intrahospital information system (ORBIS, Agfa Healthcare GmbH, Bonn, Germany).

Patients were eligible for inclusion if they were diagnosed with DCM and were above 18 years of age. Exclusion criteria included cases of acute trauma, infection, or neoplasms as well as any history of prior cervical-spine surgery.

We recorded the demographic data of the patients, the preoperative JOA score, neurologic or bone-related diseases, the narrowed segments, the instability of the vertebral bodies and the type of spinal canal stenosis (Table 1). Details of the collected parameters are provided in the Supplementary Materials.

Table 1.

Continuous variables—including age, height, weight, body mass index (BMI), ASA physical status score, and preoperative Japanese Orthopaedic Association (JOA) score—are reported as mean ± standard deviation. Categorical data are given as absolute numbers with corresponding percentages. Sex is recorded as the patient's biological sex.

Spine anatomy parameters encompass radiographically confirmed segmental instability on flexion–extension views, K-line status (negative when the posterior aspect of the ossified or hypertrophied ligamentum flavum does not cross the K-line on lateral radiographs), and the pattern of cervical canal stenosis classified as anterior, posterior or concentric. The extent of stenosis is indicated by the number and specific levels involved (C3/4 through C6/7). Coexisting neurologic or bone-related conditions—such as lumbar spinal canal stenosis, polyneuropathy, rheumatoid arthritis, Klippel–Feil syndrome, carpal tunnel syndrome, osteoporosis, fibromyalgia, and ossification of the posterior longitudinal ligament (OPLL)—are likewise tabulated by number and percentage.

Demographic date	Mean ± SD - value	Spine anatomy	Number (percentages)
Age (years)	63.07 ± 12.24	Patients with Spinal instability	11 (11.7 %)
Height (m)	1.73 ± 0.09	K-Line positive	2 (2.1 %)
Weight (kg)	81.93 ± 17.68	Anterior stenosis type	51 (54.2 %)
BMI (kg/m2)	27.28 ± 5.36	Posterior stenosis type	0
Preoperative JOA Score	13.93 ± 2.04	Concentric stenosis type	43 (45.8 %)
Sex	55 male, 39 female	One-level spinal canal stenosis	47 (50 %)
ASA Score	Median value: 2	Two-level spinal canal stenosis	37 (39.4 %)
		Three-level spinal canal stenosis	10 (10.6 %)

Neurologic or bone-related diseases	Number (percentages)	Affected cervical segments	Number (percentages)
Lumbar Spinal Canal stenosis	18 (19.15 %)	C3/4	12 (12.8 %)
Polyneuropathy	3 (3.2 %)	C4/5	45 (47.9 %)
Rheumatoid Arthritis	3 (3.2 %)	C5/6	70 (74.5 %)
Klippel-Feil-Syndrome	2 (2.1 %)	C6/7	26 (27.7 %)
Carpal Tunnel Syndrome	1 (1.1 %)
Osteoporosis	6 (6.4 %)
Fibromyalgia	1 (1.1 %)
OPLL	1 (1.1 %)
Thoracic Spinal Canal Stenosis	1 (1.1 %)

All patients were assessed with standardized clinical and radiological examinations following surgery.

For those undergoing anterior cervical decompression, follow-up visits were scheduled at discharge, 6 weeks and 12 months postoperatively.

At each time point, neurological function was quantified using the Japanese Orthopaedic Association (JOA) score, and MRI was used to evaluate both the anterior and posterior epidural spaces.

In cases requiring a supplementary posterior procedure, this was carried out at a median of 1.5 months (range 0.5–7 months) after the initial anterior surgery. In these cases, the clinical and radiological examinations were done before posterior surgery.

These patients then returned for additional follow-up at 6 weeks and again at 12 months after their posterior intervention, with the same battery of JOA scoring and epidural-space imaging.

2.2. Outcome measures

The primary outcome of this study was the need for a second posterior surgery following initial anterior decompression.

We recorded the incidence, timing, and indications for any subsequent posterior procedures, and analyzed the influence of multiple demographic, clinical and radiographic variables on the risk of secondary surgery.

These variables included the following clinically and radiographically defined categories:

• •Stenosis configuration: concentric versus non-concentric stenosis patterns
• •Number of levels decompressed
• •Preoperative sagittal alignment: preserved lordosis versus kyphotic alignment
• •K-line status: positive (cord remains dorsal to the K-line) versus negative (cord encroaches ventral to the K-line)
• •Presence of OPLL: ossification of the posterior longitudinal ligament identified on CT
• •Segmental instability: dynamic instability on flexion–extension radiographs
• •Baseline myelopathy severity: as measured by JOA Score
• •ASA physical status
• •Osteoporosis: defined as a bone mineral density T-score of −2.5 or lower

Neurological function was assessed using the Japanese Orthopaedic Association (JOA) score.

The JOA score evaluates motor and sensory function in the upper and lower extremities as well as bladder function on a scale from 0 to 17.

As a secondary outcome, the extent of postoperative anterior and posterior epidural space of the spinal cord was measured using a 1.5 T scanner (Siemens Healthineers, Erlangen, Germany). T2-weighted sagittal and axial sequences were acquired in all cases.

Fig. 1 shows the measurement of anterior and posterior epidural space.

Fig. 1.

Anterior epidural space is measured by the distance from the approximated anterior limitation of the spinal cord to the anterior spinal canal. Posterior epidural space is measured by the distance from the approximated posterior limitation of the spinal cord to the posterior spinal canal. a) preoperative, b) postoperative, blue arrows show anterior and posterior epidural space.

Anterior epidural space was quantified by measuring the perpendicular distance from the approximated anterior limitation of the spinal cord to the anterior spinal canal. Posterior epidural space is measured by the distance from the approximated posterior limitation of the spinal cord to the posterior spinal canal.

For patients with multilevel stenosis, this measurement was performed at each narrowed segment, and the mean was calculated.

To ensure measurement reliability, two independent observers (a senior and a junior spine surgeon) performed the measurements for all patients. Inter-rater reliability was assessed using intraclass correlation coefficient (ICC). The resulting ICC value was 0.88 (95 % CI 0.84–0.95), indicating very good agreement between raters.

2.3. Statistical analysis

The present assessments were performed using computer-assisted analyses. The necessary programming was performed in the R language using the R software package available on the Comprehensive R Archive Network (CRAN) at https://cran.r-project.org and Microsoft Excel. The distribution of data was examined using the Shapiro–Wilk test for normality. As the data did not follow a normal distribution, we applied a non-parametric test, the Wilcoxon test. In addition, we used a linear and logistic multivariable regression. Statistical significance was defined as p < 0.05.

A full multivariable logistic regression model including all candidate predictors (Table 2) was initially fitted using the glm(…, family = binomial) function in R. To screen each predictor's contribution, univariate likelihood‐ratio tests were performed via drop1(full_model, test = "Chisq"). Noting warnings of quasi‐complete separation (fitted probabilities numerically 0 or 1 occurred), we then refitted the model using Firth's penalized maximum likelihood approach (package logistf, profile‐penalized log‐likelihood method) to obtain bias‐reduced coefficient estimates and reliable confidence intervals.

Table 2.

Multivariable logistic regression analysis. The table summarizes the output of a multiple logistic regression for each predictor: the odds ratio (OR) estimates the change in odds of an unfavorable outcome per unit increase (or presence vs. absence) of the variable; the 95 % confidence interval (CI) indicates the precision of the estimate; and the p-value tests whether the association differs from no effect.

Variable	Odds Ratio	Confidence Interval	p-value
Age	0.998	0.991–1.005	0.579
ASA	0.857	0.741–0.992	0.042
Sex	1.036	0.906–1.184	0.609
BMI	1.002	0.99–1.015	0.71
Preoperative JOA Score	0.977	0.943–1.012	0.195
Sagittal C2/7°	0.999	0.992–1.005	0.67
Concentric stenosis pattern	1.211	1.055–1.39	0.008
Spinal instability	0.967	0.797–1.173	0.736
K-line positive	0.895	0.477–1.678	0.729
Neurologic or bone-related disease	0.917	0.797–1.055	0.228
Diabetes	1.053	0.863–1.286	0.611
Osteoporosis	1.234	1.009–1.509	0.044
Smoker	1.067	0.923–1.234	0.382
Number of affected segments	1.055	0.952–1.169	0.309
Length-of-hospital-stay	0.999	0.986–1.012	0.912

Abbreviations: ASA – American Society of Anesthesiologists; JOA score – Japanese Orthopaedic Association score; BMI – body mass index.

Model discrimination was assessed by plotting the receiver operating characteristic (ROC) curve and calculating the area under the curve (AUC) using the pROC package in R.

3. Results

3.1. Preoperative data

Ninety-four patients met inclusion criteria during the study period. The mean age was 62.4 ± 12.3 years, and 41 % (39/94) of patients were female.

The mean JOA score was 14.1 ± 1.9 for the cohort which was operated from anterior only (anterior—surgery-only-cohort or ASC); while the other cohort (which was also operated from posterior; combined-surgery-cohort or CSC) had a mean JOA score of was 12.4 ± 2.7.

Radiologically, 38 patients (40.4 %) had stenosis primarily from anterior structures, no patients had solely posterior compression, 13 patients (13.8 %) had a “hill-type” compression (e.g. ventral compression from one side), and 43 patients (45.8 %) had a concentric stenosis.

One-level disease was present in 47 patients, two-level in 37, and three-level in 10 patients.

Notably, there was a significant association between stenosis morphology and preoperative JOA score with non-concentric cases averaging 14.4 ± 1.8 versus 12.4 ± 2.1 in concentric cases (p = 0.027).

3.2. Predictors for insufficient anterior-only decompression

Nine patients (9/94, 9.6 %) required a secondary posterior decompression due to persistent or recurrent stenosis causing persisting myelopathic symptoms. All nine cases occurred in patients with a concentric stenosis representing 20.1 % (9/43) of that subgroup.

At six weeks post-operatively, the ASC demonstrated a mean JOA score of 15.44 ± 1.34, whereas patients in the CSC—assessed just before undergoing posterior surgery—had a mean score of 12.6 ± 1.32.

In a multivariate logistic regression analysis of factors influencing the postoperative outcome (Table 2), a higher ASA score was independently associated with a lower likelihood of a favorable outcome (OR 0.857; 95 % CI 0.741–0.992; p = 0.042). A concentric stenosis pattern conferred 21 % increased odds of an unfavorable outcome (OR 1.211; 95 % CI 1.055–1.390; p = 0.008), and the presence of osteoporosis was likewise a significant risk factor (OR 1.234; 95 % CI 1.009–1.509; p = 0.044). All other variables showed no statistically significant association with outcome (all p > 0.05).

In the logistic regression model, VIFs ranged from 1.15 to 1.89, with the highest VIF of 1.89 for age. All values fall well below the standard threshold of 5, indicating negligible multicollinearity.

The likelihood‐ratio tests identified ASA (p = 0.030), concentric stenosis pattern (p = 0.001), neurologic and bone-related diseases (p = 0.047), and osteoporosis (p = 0.042) as significant contributors.

In the Firth‐penalized model (n = 94), ASA (OR 0.22; 95 % CI 0.04–0.98; p = 0.046), concentric stenosis (OR 24.3; 95 % CI 2.8–3187; p = 0.001) and osteoporosis (OR 7.3; 95 % CI 1.2–66.7; p = 0.031) remained independent predictors of revision.

The global profile‐likelihood ratio test for this four‐variable model was highly significant (χ² = 19.10, df = 4, p = 0.00075), confirming that it fit the data substantially better than an intercept‐only model.

The ROC analysis yielded an AUC of 0.914, indicating excellent discrimination between patients who did and did not undergo secondary posterior surgery (Fig. 2).

Fig. 2.

Receiver operating characteristic (ROC) curve for the Firth‐penalized logistic regression model. The area under the curve (AUC) was 0.914, demonstrating excellent model discrimination.

3.3. Anterior and posterior epidural space following anterior decompression

After anterior decompression: in the CSC, the mean anterior and posterior epidural space were below our measurement resolution (nominally calculated as 0.16 and 0.22 mm, respectively). In the ASC the mean anterior and posterior epidural space was 1.6 and 0.6 mm, respectively. The difference in anterior epidural space between the two cohorts was highly significant (p = 0.00035), whereas posterior epidural space did not differ significantly (p = 0.2092).

3.4. Secondary posterior surgeries

The posterior surgeries were laminectomies with instrumented fusion in all nine cases, performed at a median of 1.5 months (range 0.5–7 months) after the initial anterior surgery.

No intraoperative complications occurred during or after the posterior surgeries.

After secondary surgery, the mean JOA score improved to 14.4 at 6 weeks follow-up and 14.7 at 12-months follow-up.

4. Discussion

In this study, we aimed to identify independent predictors of secondary posterior decompression following an initial anterior cervical approach and to characterize early postoperative neurologic recovery. Using a multivariate logistic regression model, we found that three factors—ASA score, concentric stenosis pattern, and pre-existing osteoporosis—were significantly associated with the need for secondary posterior surgery.

Patients with a higher ASA classification had a 14.3 % reduction in odds of a favorable outcome after anterior-only decompression (OR 0.857; p = 0.042), indicating that poorer preoperative systemic health increases the likelihood of persistent or recurrent compression requiring posterior intervention. The ASA score not only reflects comorbid burden but also correlates with physiological reserve and postoperative mobilization.

Similarly, the presence of a concentric stenosis pattern more than doubled the odds of revision (OR 1.211; p = 0.008) and accounted for all nine secondary decompressions—20.9 % of patients in that subgroup—underscoring the importance of stenosis morphology in surgical planning.

The difference in early functional recovery, as assessed by JOA score, further illustrates the clinical impact of these predictors. Patients who remained in the ASC achieved a mean JOA of 15.44 ± 1.34, whereas those who ultimately required posterior revision had a lower interim score of 12.60 ± 1.32 (measured immediately prior to the second surgery).

Osteoporosis also emerged as an independent risk factor (OR 1.234; p = 0.044). While statistically significant, the link between osteoporosis and early revision should be interpreted with caution: its primary impact is likely on longer‐term fusion, and other mechanical or surgical factors may better explain revisions occurring within the first few months.

These findings align closely with Papavero et al.’s 7-Letter-Code, which stratifies cervical stenosis patterns to guide surgical approach selection.⁷ It has been developed as a standardized system for classifying the anatomical features of cervical stenosis. This tool is intended to help surgeons determine the most effective surgical approach—whether an anterior, a posterior, or a combined procedure.

In this system, concentric stenoses call for a combined anterior–posterior strategy, given that ventral decompression alone may leave significant dorsal compression unaddressed. Our data validate this recommendation by illustrating that an anterior-only approach fails to restore sufficient spinal canal patency in concentric cases, precipitating symptomatic persistence and the need for reintervention.

Importantly, the 7-Letter-Code goes beyond stenosis morphology to incorporate patient-specific factors such as osteoporosis and elevated ASA status, and it recommends a combined anterior–posterior approach for this high-risk constellation.

The choice of surgical approach must be individualized based on the precise anatomy of cord compression, overall cervical alignment and patient-specific risk factors. In particular, concentric stenosis—where both ventral osteophytic ridges and dorsal elements such as ligamentum flavum hypertrophy or facet arthropathy contribute substantively to cord impingement—poses a high risk for residual compression if addressed only from the front.⁸ While anterior decompression alone can achieve excellent results in cases of focal ventral disease with preserved or lordotic alignment, its efficacy diminishes as the number of levels increases or when posterior hypertrophic changes are present; beyond three levels, it is also technically more demanding and carries greater risk of graft subsidence or construct failure, especially in osteoporotic bone.^5,9

Conversely, posterior laminoplasty or laminectomy (with or without fusion) offers more extensive canal enlargement across multiple segments⁶ and is therefore generally preferred for extensive, multisegmental or truly circumferential stenosis.10, 11, 12 In selected high-risk patients—those with significant anterior disease compounded by poor bone quality or elevated ASA status—a combined anterior–posterior strategy can maximize decompression while mitigating the risks of isolated approaches.¹⁰ Ultimately, integrating detailed imaging-based stenosis classification, sagittal balance assessment and individual systemic factors ensures that the surgical plan provides sufficient canal patency and structural support from the outset.

Radiological evaluation of postoperative epidural space further demonstrate the shortcomings of anterior-only decompression in concentric stenosis. Postoperative MRI revealed that patients requiring secondary posterior decompression had persisting spinal canal stenosis. This highlights that anterior decompression alone inadequately addresses circumferential compression, leaving the spinal cord tethered posteriorly.

Our results corroborate earlier work by Alafifi et al., who identified combined anterior and posterior compressive pathology on MRI as a predictor of suboptimal outcomes following surgical intervention for DCM.¹³

5. Limitations

Several factors temper the interpretation of our findings. First, the retrospective nature of the study inherently exposes it to selection bias and unmeasured confounding. Although we adjusted for known covariates in the multivariate model, there may be additional patient- or disease-related factors (e.g., severity of cord signal changes on MRI, nuances of surgical technique) that were not captured in our dataset and could influence reoperation risk.

Second, this is a single-center series with surgeries performed by a limited number of surgeons under institution-specific protocols. As a result, our outcomes may reflect local practice patterns and surgeon expertise, which could differ elsewhere, limiting the generalizability of the results.

Third, the absolute number of revision events was small (n = 9), reducing statistical power and widening confidence intervals for some estimates. This may have contributed to the lack of significance for variables that may show trends in univariate analysis but did not reach multivariate significance.

Finally, there was no standardized protocol for when to proceed to secondary posterior decompression; decisions were based on individual surgeon judgment and patient preference. This introduces potential treatment-selection bias, as thresholds for revision may vary between clinicians.

In sum, although our data identify important predictors of revision surgery, these limitations underscore the need for prospective, multicenter investigations with standardized imaging, surgical, and follow-up protocols to validate and refine risk-stratification algorithms.

6. Conclusion

In summary, our analysis identifies concentric stenosis morphology, osteoporosis, and elevated ASA status as independent predictors of failure following anterior-only cervical decompression. Patients exhibiting this high-risk profile experienced significantly lower early neurologic recovery and a markedly increased likelihood of requiring secondary posterior decompression. These findings underscore the critical importance of a tailored surgical strategy: in cases of multilevel concentric stenosis—particularly when compounded by poor bone quality or systemic comorbidity—a primary combined anterior–posterior approach should be strongly considered to achieve comprehensive, circumferential decompression and optimize stability.

Future work should focus on prospective validation of these predictors in multicenter cohorts. Ultimately, refining risk-stratification algorithms will enable spine surgeons to individualize surgical approaches more effectively, reduce reoperation rates, and improve functional recovery for patients with complex cervical myelopathy.

Author contributions

YN- information retrieval, data analysis, conceptualization and design of study, conceptualization of the figures, interpretation of the results, writing of the manuscript.

MR-initiation of the project, supervision of the data acquisition, conceptualization and design of study, interpretation of the results, writing of the manuscript, critical revision of the manuscript for important intellectual content.

PH - critical revision of the manuscript for important intellectual content.

PD - critical revision of the manuscript for important intellectual content.

YA - critical revision of the manuscript for important intellectual content.

MN- supervision of the data acquisition, interpretation of the results, writing of the manuscript, critical revision of the manuscript for important intellectual content.

Ethical approval

The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the ethics committee Ethikkommission Landesärztekammer Hessen, application number 2025-3982-evBO. As this study is retrospective in nature a need for informed consent was waived by the ethics committee.

Data availability statement

All data analyzed during this study are included in the Supplementary Information files.

Funding

This research received no external funding.

Declaration of competing interest

The authors declare no conflict of interest.