The study received approval from the Ethics Committee of Huashan Hospital, Fudan University under approval number KY2022-683, with informed consent obtained from all participants. We consecutively enrolled patients who underwent anterior or posterior cervical decompression and fusion for CSM between March and September 2023. Inclusion criteria required at least two myelopathic signs (e.g., hand clumsiness, gait disturbance, hyperreflexia) combined with magnetic resonance imaging (MRI)-confirmed spinal cord compression at one or more C3–C7 levels. Exclusion criteria encompassed traumatic or pathologic conditions (tumors, infections, deformities, prior cervical surgery), systemic diseases (rheumatoid arthritis, cerebral palsy), or concurrent thoracolumbar pathology. From an initial screening of 230 patients, 43 were excluded due to trauma history (n=9), prior spinal surgery (n=12), systemic diseases (n=5), or incomplete imaging (n=17), resulting in a final cohort of 187 patients.

Demographic and clinical data were extracted from the hospital’s electronic medical records, including age, gender, symptom duration, and lesion characteristics. Cervical spine lesions were analyzed from C2 to C7, with disc degeneration defined as Pfirrmann grade III or higher on T2-weighted mid-sagittal MRI sequences, and spinal canal stenosis determined by a Pavlov ratio <0.8 on lateral radiographs [20]. Surgical details (anterior vs. posterior approach, instrumented levels) were recorded from operative reports.

Standardized lateral cervical radiographs were acquired using a Philips DigitalDiagnost C50 system (Philips, Amsterdam, Netherlands) with patients in a neutral standing position and the Frankfort plane parallel to the floor. All measurements were performed using INFINITT PACS ver. 5.0 software (INFINITT Healthcare, Seoul, Korea). The radiographic evaluation protocol included three key components as follows: (1) EOP: EOP morphology was classified into three types based on lateral X-ray measurements (Fig. 1): type I: flat type (Fig. 1A); type II: crest type (Fig. 1B); and type III: spine type (Fig. 1C). Measurements were performed at 200% magnification to minimize interobserver error. (2) ONL: ONL was assessed using the classification proposed by Wang et al. [21], which categorizes ossification into four types based on lateral radiograph morphology. On lateral cervical radiographs, the extent of cervical levels involved by ONL and the maximum osteophyte length were measured. (3) Cervical sagittal alignment parameters: Assessment involved categorizing cervical sagittal alignment based on the method proposed by Donk et al. [22]. Sagittal alignment was quantified using six parameters defined in Table 1: T1 slope (TS), spino-cranial angle (SCA), C0–C2 Cobb angle, cervical lordosis (CL), C2–C7 sagittal vertical axis (SVA), and forward head posture (FHP).

All assessment data were evaluated twice, 4 weeks apart, by two experienced spinal surgeons. The inter-rater reliability between two senior spine surgeons was excellent for all parameters (intraclass correlation coefficient: 0.85–0.94).

All statistical analyses were performed using IBM SPSS ver. 20.0 (IBM Corp., Armonk, NY, USA). Categorical variables—including gender distribution, diagnosis subtypes, cervical sagittal alignment categories (lordotic/straight/kyphotic), EOP types (I–III), and ONL classifications—were compared using Pearson’s chi-square or Fisher’s exact tests as appropriate. Continuous variables—such as age, symptom duration, EOP hyperplasia length (mm), ONL osteophyte length (mm), and cervical sagittal parameters—demonstrated normality as confirmed by Shapiro-Wilk tests (all p>0.05), intergroup comparisons between two cohorts were analyzed using independent two-sample t-tests. For three-group comparisons, one-way analysis of variance was employed to assess between-group differences, followed by Bonferroni-adjusted post hoc comparisons to control for multiple testing. For ordinal/non-normally distributed data—specifically the number of pathology-affected segments, surgically intervened levels and ONL-involved vertebral levels—Mann-Whitney U tests were applied. To identify factors independently associated with EOP length, a multiple linear regression model was constructed with EOP length as the dependent variable. Independent variables included: demographics: age, gender (coded as 1=male, 2=female); ONL characteristics: osteophyte length; sagittal parameters: C0–C2 Cobb angle, CL, TS, SCA, CGH–C7 SVA, C2–C7 SVA, and FHP. Statistical significance threshold was set at α=0.05 (two-tailed).

The cohort comprised 187 surgically treated CSM patients (120 males, 64.2%; 67 females, 35.8%) with comparable age between genders (males: 56.7±10.8 years vs. female: 55.9±11.3 years, p=0.491). Key gender disparities emerged in symptom duration (males: 15.8±22.4 months vs. females: 25.3±32.7 months, p=0.002) and EOP hyperplasia length (males: 9.4±6.8 mm vs. females: 4.6±3.4 mm, p<0.001).

Anterior cervical discectomy and fusion (ACDF) was the predominant surgical approach (149 cases, 79.7%), with posterior laminoplasty performed in 38 patients (20.3%), showing equivalent gender distribution (p=0.614). The median number of pathologically involved segments was 4 (interquartile range [IQR], 1), while surgically intervened levels averaged 2 segments (IQR, 1), with no gender-based differences in segmental distribution patterns (pathology: p=0.561; surgery: p=0.662) (Table 2).

Stratification by EOP morphology revealed pronounced sexual dimorphism. Type III EOP demonstrated male predominance (82.4% vs. type I: 31.8%, p<0.001; vs. type II: 51.4%, p<0.001), whereas type I was predominantly female (68.2%). Age and EOP length differed significantly: type III patients were older than type II (58.3±10.8 years vs. 54.2±11.2 years, p=0.016) with greater EOP length (11.6±6.6 mm vs. 5.1±1.9 mm, p<0.001) (Table 3).

EOP morphology significantly correlated with ONL. The prevalence of ONL escalated with EOP severity: 45.5% (type I), 41.9% (type II), and 64.8% (type III), with type III demonstrating significantly higher rates than both type I (p=0.043) and type II (p=0.005). No significant differences were observed in the distribution of ONL morphology among the three EOP-type patient groups (p=0.099). However, type III EOP exhibited ONL osteophytes with both a greater number of involved segments and longer osteophyte length: three segments (IQR, 1) in type III vs. two segments (IQR, 1) in type II (p=0.046); 18.8±9.8 mm (type III) vs. 14.2±8.1 mm (type I, p=0.047) and 14.2±9.4 mm (type II, p=0.033) (Table 3).

EOP morphology showed no significant association with cervical sagittal alignment profiles (Table 4). The distribution of cervical curvature types (lordotic/straight/kyphotic) remained comparable across EOP subtypes (p=0.420). Similarly, quantitative analyses revealed no intergroup differences in C0–C2 Cobb angle (p=0.642), SCA (p=0.431), TS (p=0.965), CL (p=0.557), CGH–C7 SVA (p=0.821), C2–C7 SVA (p=0.356), and FHP (p=0.537).

The multivariable model integrating demographic characteristics, ONL profiles, and sagittal alignment parameters explained 38.2% of the variance in EOP hyperplasia length (adjusted R²=0.382, p=0.028). Significant associations were observed as follows: male sex demonstrated 3.8 mm greater hyperplasia compared to females (β, −3.82; 95% CI, −6.672 to −0.975; p=0.009); ONL osteophyte length showed a positive graded relationship (0.16 mm increase in EOP per 1 mm ONL elongation; 95% CI, 0.029–0.283; p=0.017); T1 slope was positively correlated with EOP hyperplasia length (0.27 mm per degree increase; 95% CI, 0.011–0.533; p=0.041); SCA exhibited an inverse association with EOP hyperplasia length (−0.19 mm per degree decrease; 95% CI, −0.333 to −0.050; p=0.009) (Table 5).

The findings of this study demonstrate strong concordance with previously documented sexual dimorphism in the developmental characteristics of the EOP [2–4,25]. The observed male predominance (82.4%) and greater EOP length (11.6±6.6 mm) in type III EOP cases corroborate forensic anthropological evidence indicating that EOP morphology achieves 79% accuracy in sex determination [3]. Notably, the detection rate of type III EOP in our cohort reached 48.7%, significantly exceeding the 6.5% reported in the cadaveric study by Singal et al. [2]. This discrepancy primarily stems from three methodological considerations: (1) The study population comprised exclusively surgical candidates with CSM, exhibiting a higher mean age (56.4 years) than baseline values in prior EOP investigations; (2) Gender distribution skewed toward males (64.2%), consistent with CSM epidemiological patterns [26]; (3) The elevated detection rate of type III EOP and its pronounced hyperplasia magnitude in our cohort suggest a potential correlation between EOP proliferation and the severity of cervical degenerative changes.

Our findings establish a biomechanical continuum between EOP morphology and nuchal ligament pathology, evidenced by two distinct patterns: type III EOP demonstrated a nearly 1.5-fold higher ONL co-prevalence compared to types I/II, aligning with cadaveric studies documenting shared entheseal microtrauma [5,10]; longer ONL osteophytes correlate with EOP hyperplasia magnitude (0.16 mm increase in EOP per 1 mm ONL osteophyte elongation). Experimental evidence from animal enthesopathy models suggests that mechanical loading may activate signaling pathways such as transforming growth factor-β1/bone morphogenic protein-2 to promote ectopic ossification at ligament insertion sites [27]. While human tissue validation remains pending, this proposed mechanism could theoretically explain our observed EOP hyperplasia-ONL co-existence patterns. We hypothesize that the EOP, serving as the anatomical attachment site of the nuchal ligament, may form a biomechanical coupling system with the nuchal ligament. Under sustained abnormal biomechanical loading, this system might induce a cascade of ossification processes, manifesting as EOP hyperplasia and ONL. Therefore, we cautiously hypothesize that concurrent EOP hyperplasia and ONL could serve as composite imaging biomarkers for identifying abnormal biomechanical loading patterns in the cervical spine.

This study investigates the spatial relationship between EOP hyperplasia and cervical sagittal parameters. While no significant differences in baseline sagittal alignment were observed across EOP subtypes (p>0.05), multivariate regression revealed two clinically relevant associations: each 1° increase in TS showed correlation with 0.27 mm EOP hyperplasia (β=0.27, p=0.041), whereas a 1° decrease in SCA correlated with 0.19 mm EOP hyperplasia (β=−0.19, p=0.009). Existing literature provides theoretical foundations for the clinical significance of TS and SCA. Specifically, TS as a crucial biomechanical parameter at the cervicothoracic junction requires dynamic equilibrium with CL. The formula CL=TS–16.5°±2° derived from asymptomatic populations [28] suggests that TS–CL mismatch may accelerate cervical degeneration and predict postoperative kyphotic deformity [29]. Furthermore, Le Huec et al. [30] demonstrated that maintaining SCA within the physiological range of 83°±9° optimizes load distribution, with deviations from this threshold inducing compensatory muscular efforts that accelerate degeneration. The significant correlations between EOP length and TS/SCA parameters revealed in our study suggest that EOP hyperplasia might indirectly correlate with symptom severity and surgical outcomes, requiring future multicenter prospective studies’ validation. Notably, our analysis of FHP parameters demonstrated neither intergroup variations nor correlation with EOP length, contrasting with the findings by Shahar et al. [23]—a discrepancy potentially attributable to population heterogeneity that warrants targeted investigation. A representative case illustrates this biomechanical-degenerative interplay (Fig. 2): A 43-year-old male with type III EOP exhibited 26.9 mm hyperplasia, continuous C3–C6 ONL, and severely reduced SCA (63.6°). The degeneration complex featured multilevel disc herniation, and characteristic “snake-eye” myelopathic changes. Postoperative neurological improvement following C3–6 ACDF underscores the clinical relevance of this biomechanical-degenerative profile.

Above all, building upon established correlations between EOP and demographic factors (gender/age), this study further elucidated morphological characteristics of EOP and their hyperplasia extent in relation to ONL and cervical sagittal alignment parameters. Key findings include: Type III EOP demonstrates not only increased ONL prevalence but also greater ONL osteophyte length. Significant linear correlations were identified between EOP hyperplasia and two cervical sagittal parameters: TS and SCA. These findings suggest EOP may serve as a morphological indicator reflecting biomechanical loading status in the cervical spine, though its precise mechanistic role requires multidisciplinary investigation.

Three primary limitations warrant consideration: First, as a retrospective cross-sectional study, the observed associations cannot establish causality. The temporal relationship and bidirectional interactions between EOP hyperplasia and cervical degeneration necessitate validation through longitudinal cohort studies. Second, while quantitative relationships between EOP hyperplasia and sagittal parameters were established, the absence of biomechanical loading threshold research impedes determination of clinically relevant cut-off values. Third, constrained by current medical data systems, patient-reported outcomes were not incorporated, potentially diminishing clinical translational value. Future investigations should prioritize developing dynamic biomechanical monitoring systems and constructing multimodal assessment models integrating symptomatic data with imaging features, thereby enabling comprehensive elucidation of EOP’s biomechanical significance in cervical degeneration.