INTRODUCTION
Laminoplasty is widely recognized as an effective surgical technique for patients with cervical ossification of the posterior longitudinal ligament (OPLL). Rationales for achieving spinal cord decompression through laminoplasty encompass both direct and indirect decompression [24,26]. After the restrictive posterior structures are decompressed, the spinal cord can migrate posteriorly, much like a bowstring. This indirect decompression is largely based on the preoperative lordotic cervical sagittal alignment [8]. However, even when adequate cervical lordosis is present preoperatively, some patients may experience a loss of cervical lordosis (LCL) postoperatively, potentially causing a gradual deterioration in their clinical outcomes over time [2,12,26,27].
Several studies have reported that preoperative radiological parameters, such as T1 slope (T1S), cervical lordotic angle (CLA) minus T1S, and the extension ratio, are significantly correlated with postoperative LCL and could be useful for predicting surgical outcomes, including LCL [6,7,13,16,18,22,25]. However, the correlation between preoperative radiological parameters and clinical outcomes remains controversial, and only a few studies have investigated their relationship with health-related quality of life (HRQOL) [20,25]. Moreover, because these studies included patients with both cervical spondylotic myelopathy and OPLL, the resulting heterogeneity introduced issues of multiplicity, thereby limiting the generalizability of the findings specifically to OPLL [7,12,13,22,25,26].
Therefore, the purpose of this study was to analyze the LCL changes over time after laminoplasty, identify preoperative radiological parameters predicting LCL, and analyze their relationships with clinical outcomes, including HRQOL, in patients with OPLL.
MATERIALS AND METHODS
Patient population
This retrospective study was approved by the Institutional Review Board of Chonnam National University Medical School Research Institute (Approval No. CNUH-2023-315). A total of 484 patients who underwent cervical laminoplasty between January 2013 and December 2022 were retrospectively reviewed. Of these, 202 patients diagnosed with OPLL affecting more than three segments and able to maintain a standing position preoperatively were initially considered. Exclusion criteria were as follows : 1) a history of cervical spine surgery (n=13); 2) concurrent removal of thoracic ossification of the yellow ligament (n=7); 3) a diagnosis of other neurological diseases, such as transverse myelitis and hereditary spastic paraparesis (n=2); 4) a follow-up period of less than 2 years (n=57); and 5) insufficient radiological examination (n=13). Ultimately, 110 patients (93 men and 17 women; mean age, 61.31±10.80 years) were enrolled in the study. Open-door or midline-splitting laminoplasties were performed by three experienced neurosurgeons, according to their preferences. In midline-splitting laminoplasty, paraspinal muscles were bilaterally dissected up to the midline of the lateral masses. In open-door laminoplasty, the opening side was dissected to the midline of the lateral mass, and the hinge side to the medial margin of the lateral mass. The opening of the lamina was fixed with a titanium miniplate. Postoperatively, each patient wore a Miami J cervical collar for 6 weeks. Patients underwent radiological evaluations and clinical outcome evaluations through patient-reported outcome measure (PROM) batteries at baseline (preoperatively), as well as at 1, 3, 6, and 12 months postoperatively, and annually thereafter. Then, final follow-up minus preoperative values (Δ) were calculated to analyze the changes after laminoplasty.
Radiological assessment
Preoperative and postoperative cervical spine lateral radiographs were obtained in neutral, flexion, and extension positions. Radiological parameters were measured using a picture archiving and communication system (INFINITT PACS M6; INFINITT Healthcare, Seoul, Korea). Two observers, blinded to the study hypothesis, independently measured all radiological parameters, and the average of their measurements was used.
On lateral radiographs taken in the neutral position, we measured the occupying ratio, the C2-C7 sagittal vertical axis (C2-C7 SVA), the center of gravity of the head to the C7 sagittal vertical axis, the T1S, and the occiput-C2 angle (Fig. 1A). To assess the parameters related to cervical spine motion, we measured the CLA in each position. The range of motion (ROM) was calculated as the extension CLA minus the flexion CLA. This ROM was further divided into flexion and extension capacities, using the neutral position as a reference. Flexion capacity was defined as the ROM from neutral to full flexion, and extension capacity was defined as the ROM from neutral to full extension. The extension ratio was defined as the degree to which the cervical spine could extend from the neutral position (Fig. 1B).
LCL, which reflects the degree of kyphotic change after laminoplasty, was calculated as the final follow-up CLA minus the preoperative CLA. A negative LCL indicated a kyphotic change Patients were divided into two groups based on their LCL values : a kyphosis group (LCL ≤-10°) and a control group (LCL >-10°) [4,11,13,22,23].
Clinical assessment
Clinical outcomes and HRQOL were assessed using PROM batteries, including the visual analog scale for the neck (nVAS), the VAS for arm radiculopathy (aVAS), the Neck disability index score, the Japanese Orthopaedic Association (JOA) score, and the EuroQol-5 Dimension (EQ-5D) index. Changes in clinical outcomes after laminoplasty were determined by calculating the Δ values of these PROMs.
Statistical analysis
Between-group comparisons were performed using the Student’s t-test or Mann-Whitney U-test for continuous variables and the chi-square test or Fisher exact test for categorical variables. A linear mixed-effects model was used to analyze the changes in CLA over time between the two groups. Univariate and multivariate linear regression analyses were conducted to identify predictors of LCL after laminoplasty. Variables with a p-value <0.1 in univariate analyses were included in the multivariate linear regression model. The optimal cutoff value for predicting LCL was determined through receiver operating characteristic (ROC) curve analysis using the Youden index. The ROC curve was interpreted using the area under the curve (AUC) as follows : AUC <0.7, poor discrimination; 0.7≤ AUC <0.8, acceptable discrimination; 0.8≤ AUC <0.9, good discrimination; and AUC ≥0.9, excellent discrimination. Additionally, the Pearson correlation coefficient test and multivariate linear regression analysis were used to examine the relationships between LCL risk factors and clinical outcomes. All analyses were conducted using SPSS Statistics for Windows, version 27.0 (IBM Corp., Armonk, NY, USA). A p-value less than 0.05 was considered statistically significant.
RESULTS
The clinicodemographic characteristics of the patients are summarized in Table 1. A total of 110 patients were enrolled, consisting of 93 men and 17 women. At the time of surgery, the mean patient age was 61.31 years (range, 35-87), and the mean follow-up period was 42.95 months (range, 24-110). Of these patients, 82 underwent midline-splitting laminoplasty, and 28 underwent open-door laminoplasty, according to the surgeon’s preference. Preoperatively, 27 patients (24.5%) had a negative Kline, 18 patients (16.4%) had an occupying ratio ≥60%, and 20 patients (18.2%) had a preoperative CLA <0°.
Comparisons between kyphosis and control groups
Based on their LCL after laminoplasty, patients were divided into two groups: the control group (LCL >-10°; n=83) and the kyphosis group (LCL ≤-10°; n=27). No significant differences were found between the groups with respect to age, sex, follow-up duration, body mass index, smoking status, operation duration, operation method, operation level, proportion of patients with a negative K-line, or proportion of patients with an occupying ratio ≥60%. However, the kyphosis group exhibited significantly worse clinical outcomes than the control group, as reflected by ΔnVAS (-2.37±2.73 vs. -0.89±2.14; p=0.012), ΔJOA (2.51±2.17 vs. 1.52±1.65; p=0.032), and ΔEQ-5D (0.123±0.179 vs. 0.013±0.179; p=0.007).
Table 2 provides a summary of the perioperative radiological parameters for the kyphosis and control groups. Before surgery, the kyphosis group had a significantly higher CLA (13.40°±8.94° vs. 8.53°±8.82°; p=0.015), flexion capacity (27.56°±8.55° vs. 20.45°±9.44°; p=0.001) and T1S (29.94°±4.61° vs. 24.68°±4.08°; p<0.001) compared to the control group. In contrast, the kyphosis group had a lower extension capacity (8.10°±7.66° vs. 14.17°±7.69°; p<0.001), extension ratio (21.11±15.36 vs. 40.35±16.45; p<0.001) and occiput-C2 angle (16.94°±7.39° vs. 21.18°±8.15°; p=0.018). Regarding changes in radiological parameters, the kyphosis groups had a greater decrease in CLA (-12.81°±2.23° vs. -1.11°±4.34°; p<0.001) and flexion capacity (-15.41°±9.53° vs. -3.25°±8.21°; p<0.001). At the same time, this group had greater increase in extension capacity (1.62°±8.72° vs. -4.61°±6.52°; p<0.001), extension ratio (29.61±46.13 vs. -4.36±23.42; p=0.001), T1S-CLA (12.23°±3.35° vs. 1.11°±4.64°; p<0.001), C2-C7 SVA (7.10±12.49 mm vs. 1.62±9.29 mm; p=0.043), and occiput- C2 angle (10.70°±6.71° vs. 2.15°±7.05°; p<0.001) compared to the control group. However, T1S itself changed minimally in both groups after laminoplasty.
The change in CLA over time after laminoplasty
The numbers of observed patients at each period were 110, 89, 51, 46, 88, 89, 93, 52, and 35 at preoperative period, 1, 3, 6, 12, 18, 24, 36, 48, and over 60 months postoperatively. The preoperative mean CLA in the kyphosis group was higher than in the control group (13.40°±8.94° vs. 8.53°±8.82°; p=0.015). Both groups experienced a decrease in CLA during the first 6 months and reached a plateau pattern. In the kyphosis group, the CLA declined more abruptly than in the control group, reaching -0.42°±8.92° at 6 months and remaining at 0.79°±9.06° after 5 years. In contrast, the control group demonstrated a more gradual reduction, reaching 6.10°±9.76° at 6 months and maintained at 5.33°±8.88° after 5 years (Fig. 2).
Risk factors for LCL after laminoplasty
Univariate linear regression analysis revealed that operation method (β=1.238; p=0.060), CLA (β=-0.185; p=0.006), flexion capacity (β=-0.163; p=0.011), extension capacity (β=0.254; p<0.001), extension ratio (β=0.154; p<0.001), T1S (β=-0.523; p<0.001), and the occiput-C2 angle (β=0.151; p=0.045) were significantly associated with LCL after laminoplasty. Subsequent multivariate linear regression analysis identified the extension ratio (β=0.107; p=0.006) and T1S (β=-0.412; p=0.004) as significant risk factors for LCL, whereas operation method, CLA, flexion capacity, and occiput-C2 angle were not significant after adjustment (Table 3). In the multivariate model, extension capacity was excluded due to multicollinearity with the extension ratio. ROC curve analysis indicated that T1S was an excellent predictor of LCL, with a cutoff value of 27.80° (Fig. 3A; p<0.001; AUC, 0.918; sensitivity, 0.826; specificity, 0.872). An extension ratio below 32.92 was also a significant, albeit acceptable, predictor of LCL (Fig. 3B; p<0.001; AUC, 0.799; sensitivity, 0.709; specificity, 0.826).
Associations between LCL risk factors and HRQOL
Among the PROMs, the EQ-5D index showed the strongest correlation with LCL (Table 4; r=0.294; p=0.002). ΔEQ-5D was negatively correlated with T1S (Fig. 4; r=-0.330; p<0.001) and positively correlated with the extension ratio (r=0.198; p=0.039). Of LCL risk factors, T1S showed a significant correlation with a greater number of PROMs. Furthermore, multivariate analysis examining the relationship between EQ-5D and LCL risk factors identified a high T1S as the only significant LCL risk factor associated with poor HRQOL after laminoplasty (Table 5; β=-0.012; p=0.002).
DISCUSSION
Cervical laminoplasty is recognized as an effective surgical technique for patients with cervical OPLL. However, LCL after laminoplasty is known to be associated with worsening clinical outcomes [2,12,26,27]. Previous research has suggested that this decline may stem from a reduced posterior shift of the spinal cord, increased longitudinal spinal cord tension due to a tethering effect from anterior pathology, and ischemic angiogenic effects due to mechanical compression [7,14,22]. For this reason, anticipating LCL before laminoplasty is critical. In this study, we found that the progression of LCL after laminoplasty mainly occurred within the first 6 months after surgery and then reached a plateau pattern. This finding aligned with the results reported by Choi et al. [1], who observed that LCL developed rapidly during the first 3 months—attributed to the anatomical disruption of the posterior elements—and then stabilized as the posterior structures recovered and acute surgical pain subsided. These findings validate the reliability of using radiological parameters measured more than 2 years postoperatively as representative of the final postoperative status. Furthermore, because LCL tends to follow a plateau pattern over time, investigating the LCL cutoff value that predicts clinical outcomes, including HRQOL, and identifying its preoperative risk factors is crucial for surgeons in formulating effective treatment strategies for patients with multilevel OPLL.
In this study, T1S and the extension ratio emerged as significant risk factors for LCL. T1S is a relatively fixed parameter reflecting the sagittal balance of the cervical spine [6,7,16]. Previous studies have reported that a high T1S correlates with a higher CLA and more severe LCL after laminoplasty [6,7,10,16-18]. In our results, the kyphosis group exhibited a persistently high T1S, as well as significant increases in postoperative C2-C7 SVA and the occiput-C2 angle (Table 2). Lee et al. [11] also proposed that LCL resulting from the posterior muscular-ligamentous complex (PMLC) injury after laminoplasty would be more pronounced in patients with higher T1S, and they reported that the cutoff value of T1S predicting LCL was 29°. Our cutoff value of T1S predicting LCL was similar, at 28°. Furthermore, the AUC for T1S was 0.918, suggesting that T1S is an excellent predictor of LCL.
In addition to T1S, many studies have reported the associations between dynamic cervical parameters and LCL [3,13,22,25]. Extension capacity reflects the functional integrity of the PMLC, while flexion capacity reflects dynamic structural factors—including bone, ligament, and muscle components—that help prevent kyphotic deformity [22]. Accordingly, a low extension capacity and a high flexion capacity are believed to correlate with an increased incidence of LCL [3,13]. Meanwhile, Ono et al. [22] reported that the extension ratio, an index incorporating both flexion and extension capacities, can serve as a risk factor for LCL. In our univariate analysis, flexion capacity, extension capacity, and extension ratio were all significantly associated with LCL. In the multivariate analysis, however, either the extension ratio or extension capacity had to be excluded due to multicollinearity, leaving T1S and extension ratio as risk factors for LCL in the final model (Table 3; R2=0.271). Additionally, we found that T1S was negatively correlated with the extension ratio, with T1S accounting for about 14.3% of the extension ratio variability (data not shown; R=-0.378; R2=14.3%; p<0.001). This suggests that patients with a high T1S need to keep their cervical spine alignment near maximum extension in the neutral position. As a result, these patients tend to have a low extension ratio preoperatively.
Based on these findings, we propose the following mechanism for LCL after laminoplasty on patients with high T1S. Patients with a high T1S tend to have a high preoperative CLA to maintain horizontal gaze and neutral position near extension, characterized by a low extension ratio. Such patients place heavy strain on the PMLC to preserve CLA. Thus, injury to the PMLC after laminoplasty causes the deterioration of CLA and an increase in C2-C7 SVA [17]. Consequently, the neutral position after laminoplasty shifts toward flexion, reducing flexion capacity, while extension capacity remains stable or slightly increases, causing a significant rise in the extension ratio. To maintain a horizontal gaze, the patient compensates by increasing the occiput-C2 angle [17]. However, this compensation imposes additional strain on the PMLC and ultimately impairs clinical outcomes.
Several studies have explored risk factors for LCL following laminoplasty, but controversies persist regarding the associations between LCL, its risk factors, and clinical outcomes. Miyazaki et al. [18] reported that while T1S was related to LCL, it was not significantly associated with JOA scores. In turn, Lee et al. [9] reported that cervical sagittal alignment was not significantly associated with clinical outcomes such as nVAS, aVAS, Neck disability index, JOA, and the 36-Item Short-Form Health Survey. However, these studies lacked the statistical power to establish that T1S is unrelated to clinical outcomes due to small subgroup sizes. In other studies, both Lee et al. [12] and Sharma et al. [25] reported that sagittal imbalance after laminoplasty was associated with a lower JOA recovery rate, but they did not investigate its relationship to HRQOL. In our study, we analyzed 83 patients in the control group and 27 patients in the kyphosis group and found that LCL and its risk factors correlated with clinical outcomes, including HRQOL. Furthermore, in contrast to previous studies, we divided patients according to an LCL cutoff of -10°, revealing meaningful differences in clinical outcomes exceeding the minimum clinically important difference [5,15]. Both T1S and extension ratio, as LCL risk factors, were correlated not only with disease-specific outcome measures (VAS, JOA) but also with the EQ-5D index, reflecting HRQOL. In the multivariate linear regression analysis, only T1S, and not extension ratio, predicted HRQOL. These results align with prior research suggesting that HRQOL after laminoplasty is influenced by managing radiating pain and improving lower extremity function [19,21]. As shown in Table 4, the extension ratio had the strongest correlation with nVAS (indicating neck pain), whereas T1S was most strongly correlated with JOA.
This study had several limitations. First, it was a retrospective, non-controlled study with a relatively small sample size. Second, although we had a mean follow-up of about 43 months, longer-term monitoring may be warranted to confirm these findings. Third, missing data resulting from variability in follow-up periods, although adjusted statistically using a linear mixed-effects model, limited our ability to fully characterize outcomes over different durations. Fourth, we did not analyze global sagittal alignment or consider thoracic inlet angles and neck tilt. Fifth, we did not compare laminoplasty with other surgical techniques, such as posterior cervical fusion or corpectomy. Sixth, although operation method (midline-splitting vs. open-door) was included in multivariate analysis and did not show a significant association with LCL, the sample size imbalance between groups may have limited the statistical power to detect subtle effects of posterior muscle disruption. Despite these limitations, our study provides valuable insights into the relationship between LCL and preoperative radiological parameters, as well as clinical outcomes, including HRQOL. We established cutoff values for LCL based on perioperative changes in clinical outcomes, identified a T1S cutoff value for LCL prediction, and explored the correlation between T1S and HRQOL in patients with multilevel OPLL after laminoplasty. A prospective, randomized study comparing alternative surgical techniques is needed to confirm and expand upon our results.
CONCLUSION
In patients with multilevel OPLL undergoing laminoplasty, LCL primarily develops during the first 6 months and then reaches a plateau pattern. This postoperative LCL, when exceeding -10°, was associated with poorer clinical outcomes. LCL occurrence increases with higher T1S and lower extension ratio. Among these risk factors, T1S is significantly correlated with HRQOL, with a higher T1S associated with poorer HRQOL. Regarding the prediction of LCL, the cut-off value for T1S was determined to be approximately 28°. Consequently, a T1S exceeding 28° may represent an important prognostic factor before performing laminoplasty in patients with multilevel OPLL.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
