Spinal fusion surgery is required for diverse degenerative lumbar diseases, such as spondylolisthesis, instability, herniated nucleus pulposus, stenosis, hypertrophic facet arthritis, and kyphosis due to vertebral compression fracture^6,10,18,22). Adjacent segment disease is a representative long-term complication after spinal fusion surgery and is defined as symptomatic degeneration that necessitates second surgery at a level above or below the index fusion level and adversely affects clinical outcomes³⁹⁾. In contrast, radiographic adjacent segment disease or adjacent segment degeneration (ASD) is defined as a pathological process at a disc adjacent to a prior fusion level and includes disc degeneration, segmental instability, or canal stenosis regardless of the presence of clinical symptoms¹⁵⁾.

ASD was first reported about 50 years ago^19-21,24,37). The recent dramatic increase in the number of spinal fusions has led to a corresponding increase in the incidence of ASD. The etiology of ASD is known to be multifactorial. Potential risk factors of ASD include posterior lumbar interbody fusion (PLIF), injury to the facet joint of an adjacent segment, fusion length, the presence of a preexisting degenerated disc, lumbar stenosis, age, osteoporosis, female gender, and a postmenopausal state^{1,4,14,23,28,31)}. Although many authors have evaluated the risk factors of ASD, clearly defined results lack due to study heterogeneities^11,12).

Sagittal balance is closely associated with health-related quality of life, and thus, maintaining sagittal balance with adequate lumbar lordosis (LL) is an important goal of thoracolumbar fusion surgery. However, little is known about how sagittal imbalance affects the onset of ASD. To the best of our knowledge, no study has yet identified the risk factors of early and late-onset ASD. The present study was conducted to determine the incidence of ASD at a single institution over 11 years and to identify the clinical and radiologic risk factors of early and late-onset ASD with a focus on how sagittal imbalance affects ASD onset. To minimize confusion, we restricted enrolment to patients with radiologically confirmed ASD.

During the 11-year study period, 1,121 patients underwent thoracolumbar fusion surgery. The 428 of these patients were excluded due to tumor, infection, prior spine surgery, adjacent segment decompression, or concomitant scoliosis. During the follow-up period, 92 patients developed ASD; an incidence of 13.28% (92/693). Of these 92 patients, ASD occurred at <1 year after surgery in 34 patients (4.9%) and at ≥1 year after surgery in 58 (8.4%), and these patients constituted the 2 study groups, that is, the early and late-onset groups, respectively (Fig. 3).

Demographic characteristics (sex, age, and bone mineral density) were not significantly different in the 2 groups (p>0.05). Patients in the early and late groups were diagnosed with ASD at 200.35±94.60 and 1,323.31±466.76 days postoperatively, respectively, and this difference was statistically significant (p<0.005). Fusion length was significantly greater in the early group (2.38±1.18 vs. 1.74±0.9, p=0.0045). Distributions of fusion lengths also differed significantly in the 2 groups (p=0.011). However, types of fusion techniques used were not associated with time to ASD onset (p=0.991) (Table 1).

Pfirrmann and facet degeneration grades at adjacent segments were determined using preoperative MR images. However, no significant intergroup difference was found (p=0.09, p=0.09 respectively). Accordingly, preoperative degenerative changes were not correlated with time to ASD onset.

Sagittal alignment and pelvic parameters were measured in the 92 study subjects. Postoperative whole spine lateral X-rays showed SVA in the early group was 50.14±46.34, which is slightly higher than normal, but that SVA in the late group was 26.76±36.53, which is within the normal range, and these values were significantly different (p=0.009). LL measured after fusion surgery was 28.67±14.46° in the early group and 37.83± 10.57° in the late group. Both angles were inadequate for normal lumbar lordosis, and the early group had a significantly smaller lumbar lordotic angle than the late group (p≤0.005). The SS in the early onset group was significantly lower than in the late-onset group (20.76±7.86 vs. 30.57±8.39, p<0.005), but PT in the early group was greater than in the late group. (28.31±11.94 vs. 23.19±10.36, p=0.03) However, other pelvic parameters including PI and PI minus LL mismatch did not reveal a significant difference between the 2 groups (Table 2).

Using the clinically meaningful risk factors, a multivariate logistic regression model was constructed (Table 3). The factor with the highest odds ratio (OR) was fusion length (OR, 1.61; 95% confidence interval [CI], 1.02-2.51). LL showed a negative correlation (OR, 0.95; 95% CI, 0.91-0.99)

Fusion is a well-established surgical technique for managing many types of spinal disease. Recent advances in medical devices and biologics have substantially increased the number of fusion surgeries performed worldwide. However, fusion itself causes tremendous stress to facet joints and discs of adjacent segments¹⁶⁾, increases mechanical stress at adjacent segments, and irreversibly changes spinal biomechanics²⁾.

Radiographic degeneration at adjacent segments is a common finding but does not necessarily correlate with clinical symptoms. According to related reports, the incidence of asymptomatic radiographic ASD ranges from as little as 8% to 100%, whereas that of reported symptomatic ASD ranges from 5.2% to 18.5%^13,25,27), and only a small proportion of these cases require a second surgery⁷⁾. In the present study, ASD was defined as the presence of radiologic change at adjacent segments regardless of symptoms, and we found that the overall incidence of ASD was 13.28% and that 34 cases (4.9%) occurred at <1 year and 58 cases (8.4%) at ≥1 year. Other studies have reported similar findings. Wang et al.³⁴⁾ reported a 6.3% incidence of postoperative ASD in 15 patients with degenerative lumbar disorders, which included lumbar disc herniation, lumbar spinal stenosis, degenerative lumbar spondylolisthesis.

This study shows that a greater fusion length was strongly associated with early ASD (OR, 1.61; 95% CI, 1.02-2.51), concurs with previous reports. Song et al.³²⁾ reported that the prevalence of ASD is high when the fusion range is above 3 segments. Because stress at adjacent segment increases in proportion with prior fusion length, remnant mobile segments are easily damaged. Therefore, spine surgeons always need to bear in mind the possibility of ASD and consequent fusion extension. Fusion should be decided not only by considerations of current pathology but also of overall spinal column structure, and the proper fusion site should be determined based on careful medical history-taking and physical and radiologic evaluations³⁰⁾.

In the current study, degenerative alteration of intervertebral discs and facet joints of adjacent segments before surgery was not associated with time to ASD onset. It remains controversial whether degenerative changes of adjacent segment intervertebral discs and facet joints affect the progression of ASD. Gillet⁸⁾ reported that segments with degenerative changes before surgery did not play a significant role in terms of long-term results, and thus, performed fusion surgery only on segments related to current symptoms. On the contrary, there have been reports that degenerative changes in adjacent segments after surgery were accelerated after initial fusion and increased the prevalence of adjacent segment disease^3,6,9,35).

We measured preoperative segmental angles of upper and lower lumbar vertebrae adjacent to segments at which ASD occurred after surgery. The segmental angle was determined by obtaining the flexion, extension, and ROM, and there was no intergroup difference observed³⁸⁾. Initially, we believed that preoperative instability might cause ASD during the early postoperative period, but no clear association was observed.

PT importantly influences spinal alignment and ASD. According to a study conducted by Di Martino et al.⁵⁾, patients with a PT >20° and an SS <40° represent a high-risk group for clinically evident ASD and subsequent reintervention. Pelvic retroversion and hypolordosis are the 2 main mechanisms of compensation that create an unbalanced spine⁵⁾. Yamasaki et al.³⁷⁾ found greater PT results in reduced LL, hip and knee flexion contracture, and nerve or muscle disorders. In our study, postoperative PT and SS values in the early and late ASD groups were significantly different. Therefore, we conclude lower SS and higher PT are risk factors of early-onset ASD.

It has been consistently suggested that changes in sagittal plane anatomy are positively associated with the incidence of ASD^17,29). Kumar et al.¹⁴⁾ showed more significant degenerative changes at adjacent levels occurred when C7 plumb line alignment was abnormal after lumbar fusion surgery. In addition, many studies have concluded that the incidence of ASD is lower when the postoperative lumbar lordotic angle remains in the normal range. The present study confirmed that ASD occurs rapidly when the degree of sagittal imbalance is severe. When this plumb line is antepulsed, it means that pelvic compensation either did not occur (e.g., hip arthritis) or was insufficient to correct a grossly abnormal amount of antepulsion. The instantaneous axis of rotation of a structurally normal spine passes through the anterior third of lumbar disc spaces and the moment arm of the center of mass is balanced by the moment arm of spinal muscles. However, in the presence of antepulsion, the moment arm of the center of mass increases and increases loading of unfused motion segments. Another reason for antepulsion-associated disc degeneration concerns extensor muscle activity during attempts to maintain balance³³⁾. It has been shown that compressive loading of discs is highest during trunk extension exercises¹⁾.

The present study has several limitations. First, because of its cross-sectional design, the associations found between risk factors and ASD do not imply cause-and-effect relationships. Second, we compared differences between early and late-onset ASD groups, and had a group that did not develop ASD been added, more meaningful results might have been obtained. Third, due to the relatively low incidence of ASD, limited numbers of patients were enrolled. We suggest a multi-center study be conducted on symptomatic ASD.

The incidence of ASD after thoracolumbar fusion was found to be 13.28%. Early-onset ASD was associated with greater fusion length, inadequate acquisition of lumbar lordosis, lower SS, and higher PT. Therefore, these findings indicate that to reduce the occurrence of ASD, fusion length be minimized by integrating only necessary parts as determined by accurate physical examination, and that sagittal alignment be corrected as close to normal as possible.