Distal Junctional Failure After Fusion Stopping at L5 in Patients With Adult Spinal Deformity: Incidence, Risk Factors, and Radiographic Criteria
Article information
Abstract
Objective
To identify risk factors and establish radiographic criteria for distal junctional failure (DJF) in patients with adult spinal deformity (ASD), who underwent fusion surgery stopping at L5.
Methods
This retrospective study was undertaken from January 2016 to December 2020. Patients with ASD who underwent fusion surgery (≥5 levels) stopping at L5 were analyzed. DJF was defined as symptomatic adjacent segment pathology at the lumbosacral junction necessitating consideration for revision surgery. Demographic data and radiographic measurements were compared between the DJF and non-DJF groups. Receiver operating characteristic curve analysis was performed to identify the radiographic cutoff value for DJF.
Results
Among 76 patients, 16 (21.1%) experienced DJF. DJF was associated with older age, antidepressant/anxiolytic medication, longer level of fusions, and worse preoperative sagittal alignment. Antidepressant/anxiolytic medication (odds ratio, 5.60) and preoperative pelvic incidence (PI)–lumbar lordosis (LL) mismatch>40° (odds ratio, 5.87) were independent risk factors for DJF. Without both factors, the incidence of DJF has been greatly reduced (9.1%). Two radiographic criteria were determined for DJF: last distal junctional angle (DJA)>-5° and Δ last DJA–post DJA>5°. When both criteria were met, the sensitivity and specificity of the DJF were 93.3% and 91.7%, respectively.
Conclusion
Use of antidepressant/anxiolytic medication and preoperative PI–LL mismatch >40° were independent risk factors for DJF. DJF could be diagnosed using postoperative changes in the DJA. If both criteria were met, DJF could be strongly suggested.
INTRODUCTION
Among the mechanical complications after long fusion surgery in patients with adult spinal deformity (ASD), junctional problems are the most significant [1,2]. Particularly, distal junctional problems have a greater clinical impact, such as increased sagittal decompensation or reoperation rates, than proximal counterparts, including proximal junctional kyphosis/failure (PJK/PJF) (Fig. 1) [3,4]. Nevertheless, research on distal junctional problems has not been widely reported, in contrast to research conducted on PJK/PJF. Previous studies on distal junctional problems have primarily focused on pediatric populations with conditions such as adolescent idiopathic scoliosis or Scheuermann’s kyphosis and have mainly investigated kyphosis occurring at the thoracolumbar junction or upper lumbar level, creating a gap in the understanding of junctional problems in patients with ASD, who usually have instrumentation at the lumbosacral junction.
Reports on distal junctional problems in patients with ASD include mainly comparative studies on surgical outcomes when instrumentation is stopped at L5 or S1, with inconsistent results regarding complications and revision surgery rates [4-6]. Instrumentation stopping at L5 has been reported to be associated with a high frequency of subsequent disc degeneration at the L5–S1 disc (60%–70%). However, it is unclear whether this was a natural degeneration or true pathologic process because there was no relevant control group. Moreover, confounding factors for distal junctional failure (DJF) were not adequately adjusted for [7,8]. The clinical impact of subsequent disc degeneration was also unclear [5,7,8].
Recently, McDonnell et al. [9] reported that more than 40% of patients with ASD suffered from DJF in the postoperative period. They concluded that pedicle subtraction osteotomy (PSO), poor correction of lumbar lordosis (LL), and sagittal vertical axis (SVA) were independent risk factors for DJF. However, they included all mechanical problems occurring at the distal end, even in surgeries involving sacropelvic fixation, as DJF. In cases of PJK/PJF, which can be considered counterparts to distal junctional kyphosis (DJK)/DJF, a significant proportion of failures (over 30%) occur at the noninstrumented mobile segment (such as the junctional disc, ligament, or uninstrumented adjacent vertebrae), in addition to failures that occur at the UIV [10,11]. It is inappropriate to discuss DJF in cases of sacropelvic fixation because there is no adjacent mobile segment present in such cases. Therefore, we investigated DJF in patients with ASD, who underwent long fusion surgery stopping at L5, exclusively.
Nowadays, even if the L5–S1 disc is relatively healthy, instrumentation extending to the pelvis has become a common practice in ASD surgeries, because of concerns regarding the subsequent degeneration at the L5–S1 disc. However, pelvic fixation is associated with substantial complications such as fixation failure, painful implant, and increased risk of PJK [12-16]. Furthermore, it could further restrict the activities of daily living of patients due to the longer construct, compared with preservation of the lumbosacral junction [17-19].
In this study, we evaluated the patients with ASD who underwent long fusion surgery stopping at L5 and compared the parameters between the DJF and non-DJF groups. We aimed to identify the independent risk factors and establish the radiographic criteria for DJF.
MATERIALS AND METHODS
This study was approved by the Institutional Review Board of Kyung Hee University Hospital at Gangdong (KHNMC 2023-03-001). The requirement for informed consent was waived because this study was retrospective cohort study on chart review and involved minimal risks to the subjects. From January 2016 to December 2020, the medical records of consecutive adult (aged >18 years) patients with ASD, who underwent long fusion surgery stopping at L5 were reviewed retrospectively. Patients with a fusion of 5 or more segments (upper instrumented vertebra at or above L1) were included. The minimum follow-up period was 2 years after surgery. Patients who underwent grade ≥3 osteotomy (e.g., PSO or vertebral column resection) were excluded. Patients who underwent surgery for posttuberculous kyphosis, infectious etiology, or scoliosis associated with idiopathic or syndromic etiology were also excluded. Patients with a radiographic evidence suggestive of the lumbosacral transitional vertebra, or an incomplete preoperative radiographic study (e.g., absent whole-spine standing radiographs) were also excluded. Finally, patients with significant deformities involving the L5–S1 disc (oblique takeoff) were excluded. We defined the significant oblique takeoff as the intercrestal line subtended by a horizontal plane >15° [20,21].
Patients with persistent (≥3 months) back or lower extremity pain, disability, or neurological deficits (e.g., lower extremity weakness) were considered candidates for surgical treatment. All patients underwent a staged operation that was a posterior approach for instrumentation and facetectomy and/or laminectomy in the first stage. Thereafter, an oblique retroperitoneal approach followed by oblique lateral interbody fusion and posterior rod assembly in the second stage. L5 laminar hooks were applied in most cases (80.3%) to augment L5 screws. All patients wore a brace for 3 months after surgery. Moreover, rehabilitation therapy was provided concurrently in cases of neurological deficits (e.g., ankle weakness) or prolonged difficulties with walking.
DJF was defined as adjacent segment pathology (i.e., disc degeneration, instability, stenosis, facet degeneration, fracture, and deformity) at the lumbosacral junction accompanied by clinical deterioration necessitating consideration for revision surgery. We did not define DJF based on the junctional angle because no consistent angle criterion applies to patients with ASD. This definition was referenced form the definition of PJF, determined by Yagi et al. [11], which could be considered as the counterpart of DJF. Finally, a total of 76 patients were enrolled and divided into the DJF and non-DJF groups.
Demographic data, including age, sex, body mass index (BMI), comorbidities, American Society of Anesthesiologists (ASA) physical status classification, medication, bone mineral density (measured using dual-energy x-ray absorptiometry), follow-up period, history of prior lumbar fusion including L5, and the number of fused vertebrae, were evaluated. Operative data were evaluated in terms of surgical time, surgical bleeding, and the length of hospital stay. The radiographic measurements were performed by 2 neurosurgeons who did not participate in the surgery. Spinopelvic parameters, including pelvic incidence (PI), LL, PI–LL mismatch, thoracolumbar angle (TL), pelvic tilt (PT), the SVA, and coronal curve characteristics were measured from whole-spine standing radiographs preoperatively, postoperatively (within 1 month), and at the last follow-up. The LL was defined as the angle between the L1 superior endplate and S1 superior endplate. We measured L4–S1 lordosis as the angle between the L4 superior endplate and S1 superior endplate. The distal junctional angle (DJA) was defined as the angle between the L5 inferior endplate and S1 superior endplate on standing radiographs. In the DJF group, the last radiographic parameter was measured at the time of DJF occurrence. Detailed radiographic characteristics of the L5–S1 discs were assessed in terms of Weiner classification, Pfirrmann grade, modified Pfirrmann grade, disc height, disc angle, and segmental motion (disc angle between flexion and extension). The L5–S1 disc angle was measured in both the supine and standing positions, and the change in angle according to the position was also measured. Negative values represented lordosis, whereas positive values represented kyphosis.
We used IBM SPSS Statistics ver. 25.0 (IBM Co., Armonk, NY, USA) for statistical analysis. The independent t-test and Mann-Whitney U-test were employed to compare the means of the parametric and nonparametric variables, respectively. The normality of the data was determined using the Shapiro-Wilk and Kolmogorov-Smirnov tests. The chi-square and Fisher exact tests were used for dichotomous data analyses. To identify the predictive factors for DJF, we performed multivariable binary logistic regression analysis for potential predictors with p-values of <0.05 in univariable analysis. Odds ratios (ORs), 95% confidence intervals (95% CIs), and p-values were calculated. Receiver operating characteristic (ROC) curve analysis, including the area under the curve (AUC), was performed to identify the radiographic cutoff value for the occurrence of DJF. A p-value of <0.05 was considered statistically significant.
RESULTS
1. Comparison of DJF and Non-DJF Groups
Among the 76 patients, 16 (21.1%) experienced DJF and were recommended revision surgery. The median time to DJF occurrence was 2.4 years (range, 0.5–4.5 years). Eventually, 8 patients underwent fusion extension to the pelvis, 7 refused revision surgery, and 1 was unable to undergo revision surgery due to a poor general condition (cerebral infarction). The most common type of DJF was junctional kyphosis with endplate sclerosis or disruption (n=6), followed by progression of sagittal imbalance (n=5). Other types of DJF included disc space collapse with foraminal stenosis (n=3) and junctional compression fractures (L5 or S1; n=2).
The DJF group was significantly older than the non-DJF group (73.4±5.0 years vs. 69.7±6.4 years, p=0.034) (Table 1). The proportion of patients with a history of taking antidepressant/anxiolytic medication was higher in patients with DJF than in those without (37.5% vs. 10.0%, p=0.015). Other demographic parameters, including sex, BMI, Charlson comorbidity index, ASA classification, antiplatelet medication, osteoporosis, follow-up period, and the number of interbody fusion, did not differ between the groups, except the number of fused vertebrae (7.1±1.1 vs. 6.3±1.3, p=0.037). Surgical details such as surgical time (284.8±85.1 minutes vs. 293.1±54.8 minutes, p=0.636), surgical bleeding (931.3±262.6 mL vs. 928.3±242.2 mL, p=0.967), and the length of hospital stay (20.9±8.3 days vs. 19.5±7.8 days, p=0.566) did not differ between the 2 groups.
The perioperative spinopelvic parameters are summarized in Table 2. PI and coronal curve magnitude did not differ between the groups. Most preoperative radiographic parameters of sagittal alignment were significantly worse in the DJF group than in the non-DJF group in terms of LL (-2.4°±28.0° vs. -22.2°±17.5°, p=0.015), PI–LL mismatch (51.7°±29.1° vs. 29.8°±20.0°, p=0.011), PT (39.7°±11.7° vs. 24.0°±9.2°, p=0.041), and SVA (169.6±107.1 mm vs. 92.6±80.6 mm, p=0.002). Patients with DJF also showed markedly worse sagittal alignment at the last follow-up than the non-DJF group, in terms of LL (-22.4°±19.4° vs. -43.2°±13.6°, p<0.001), L4–S1 lordosis (-1.3°±19.0° vs. -21.7°±11.9°, p<0.001), and SVA (191.7±90.4 mm vs. 53.9±51.6 mm, p<0.001).
We assessed the detailed preoperative radiographic characteristics of the L5–S1 disc space (Table 3). The degree of disc degeneration in terms of the Weiner classification, Pfirrmann grade, and modified Pfirrmann grade did not differ between the groups. Other radiographic parameters, such as disc height, disc angle, and segmental motion between flexion and extension, did not differ between the groups. However, the mean change in the disc angle between the supine and standing postures was significantly greater in the DJF group than in the non-DJF group (3.3°±8.1° vs. 0.1°±5.5°, p=0.009) (Fig. 2).
Perioperative changes in the DJA significantly differed between the groups (Table 4). The DJA at the last follow-up (last DJA) was significantly greater (kyphotic) in the DJF group than in the non-DJF group (6.0°±14.3° vs. -9.3°±7.9°, p=0.001). The amount of changes in the last DJA from preoperative DJA (pre DJA, 16.1°±19.2° vs. 4.3°±6.3°, p=0.029) and from postoperative DJA (post DJA, 10.7°±11.8° vs. -0.9°±4.8°, p=0.001) were also significantly greater in the DJF group than in the non-DJF group (Fig. 3).
2. Identification of Risk Factors for DJF
To identify preoperative risk factors for DJF, we conducted multivariable logistic regression analysis in terms of potential predictors, including age, antidepressant/anxiolytic medication, number of fused vertebrae, LL, TL, PT, SVA, PI–LL mismatch, and the difference in disc angle between supine and standing postures. Consequently, antidepressant/anxiolytic medication (OR, 4.96; 95% CI, 1.09–22.62; p=0.039) and preoperative PI–LL mismatch (OR, 1.04; 95% CI, 1.01–1.07; p=0.009) were identified as risk factors for DJF.
The preoperative sagittal parameters, including LL, TL, PT, SVA, and PI–LL mismatch, were divided in a dichotomous fashion for brevity and clarity, based on an ROC curve analysis. The optimal cutoff values were determined as follows: LL=-15°, TL=20°, PT=30°, SVA=100 mm, PI–LL mismatch=40°. All of these values were statistically significant. As a result of the multivariable logistic regression analysis, antidepressant/anxiolytic medication (OR, 5.60; 95% CI, 1.26–24.81; p=0.023) and preoperative PI–LL mismatch >40° (OR, 5.87; 95% CI, 1.66–20.81; p=0.006) were identified as risk factors for DJF. When these 2 factors were present, the incidence of DJF was 71.4%. On the other hand, without both of these factors, the incidence of DJF was greatly reduced (9.1%).
3. Radiographic Criteria for DJF
We conducted a ROC curve analysis to identify the optimal radiographic cutoff values for the diagnosis of DJF using 2 parameters: (1) last DJA and (2) Δlast DJA–post DJA (Fig. 4). The AUC values for the last DJA and Δlast DJA–post DJA were 0.871 (p<0.001) and 0.908 (p<0.001), respectively. Consequently, the optimal cutoff values for DJF were determined to be -5° for the last DJA and 5° for Δlast DJA–post DJA. We calculated sensitivity and specificity for the diagnosis of DJF using the last DJA>-5° (criterion A) and Δlast DJA–post DJA>5° (criterion B). When both A and B were met, the sensitivity and specificity of the DJF were 93.3% and 91.7%, respectively.
DISCUSSION
In the present study, DJF occurred in 16 out of 76 patients (21.1%) with ASD, after long fusion surgery stopping at L5. The principal findings were as follows: First, patients with DJF were older (73.4 years vs. 69.7 years), took antidepressant/anxiolytic medication more frequently (37.5% vs. 10.0%), and underwent longer fusion (7.1 vs. 6.3), than did those without DJF. Regarding radiographic parameters, the DJF group had worse preoperative sagittal alignment, such as LL, PI–LL mismatch, PT, and SVA. Second, the use of antidepressant/anxiolytic medication and preoperative PI–LL mismatch (>40°) were identified as independent risk factors for DJF. When both these factors were absent, the incidence of DJF was less than 10%. Therefore, stopping at L5 can be considered to avoid the potential complications associated with pelvic fixation in patients without risk factors. Finally, DJF could be effectively diagnosed based on the postoperative changes in DJA. Therefore, we propose 2 radiographic criteria for DJF: last DJA>-5° and Δlast DJA–post DJA>5°. If both criteria are met, DJF should be suspected (93.3% sensitivity and 91.7% specificity).
In cases of ASD undergoing long fusion surgery, many studies have reported the advantages of performing sacropelvic fixation. However, we believe that in certain cases (e.g., healthy lordotic L5–S1 disc or almost collapsed disc space with little segmental motion), consideration of stopping at L5 remains still valuable. The benefits of sparing the L5–S1 segment include the following: (1) Sparing the motion segment can reduce the stiffness experienced by patients and be advantageous for activities of daily living [17,22,23]. It is known that the longer the instrumentation, the greater the stiffness perceived by the patient [18]. (2) Sacropelvic fixation increases stress and strain on the construct, which is a known risk factors for PJK/PJF [15,24]. Additionally, issues related to sacropelvic fixation, such as increased surgical time, blood loss, and the prominence of iliac screws, can arise [12,19]. Stopping at L5 can mitigate these issues. Therefore, in cases of patients with ASD, without significant segmental instability or nerve root compression, we consistently deliberate on the option of stopping the fusion at L5.
It is noteworthy that the risk of DJF increased about 6 times in patients with a history of taking antidepressant/anxiolytic medication. This finding is consistent with previous studies that highlighted that depression and anxiety, the most common affective disorders, have been reported as risk factors for poor outcomes after spinal surgery in terms of functional recovery and postoperative pain scale [25-27]. Toombs et al. [28] also reported that mood and anxiety disorders were independent risk factors for devicerelated complications in patients with ASD. They hypothesized that worse outcomes may be related to patient compliance, the effect of medications on overall health, and postoperative recovery. Therefore, it is suggested that poor functional recovery and defective rehabilitation may be associated with DJF. However, further research will be necessary to clarify the precise causal relationship.
Interestingly, the degree of L5–S1 disc degeneration evaluated using plain radiography and magnetic resonance imaging (MRI) was not associated with the occurrence of DJF. These results are inconsistent with the current recommendations that stopping at L5 is reasonable if the L5–S1 disc is healthy and there is no instability or major deformity [7,8,29]. One explanation for this discrepancy may be the paradox that the potential for future degeneration of healthy discs can be higher than the risk of further degeneration of an already degenerated disc. Another explanation is that the current grading systems measured using plain radiographs or MRI do not sufficiently reflect the true degenerative changes in the disc. In this study, postural change (Δsupine-standing) of disc angle was significantly greater in the DJF group than in the non-DJF group (3.3° vs. 0.1°). We suppose that, unlike the supine position without axial loading, a greater decrease in the lordotic angle in the standing posture may reflect the structural vulnerability of the L5–S1 disc in patients with DJF.
We defined DJF based on symptomatic failure at the adjacent segment, without considering the junctional angle because there were no reliable or consistent results for the angular criterion. This definition was referenced from the definition of PJF determined by Yagi et al. [11], which could be considered as the counterpart of DJF. In previous studies, the angular criterion for DJK was either DJA>10° or DJA increase >10° or both [2,30,31]. Another study defined DJK based on the disc angle as kyphotic change [32,33]. Moreover, most studies have evaluated pediatric populations in whom the lowest instrumented vertebra (LIV) is mostly located at the thoracolumbar junction (which normally has little lordosis or even kyphosis) [30-33]. Therefore, it is unreasonable to apply similar criteria in patients with ASD, in whom almost all LIV are located at the lumbosacral junction (which normally has significant lordosis). Based on the ROC curve results, DJF appears to occur with a smaller change in the junctional angle than PJK/PJF. Consequently, we propose new radiographic criteria for DJF (last DJA>-5°, Δlast DJA–post DJA>5°). Although DJF is strongly suggested when both criteria are met, clinicians should be aware of the possibility of DJF even when either of these criteria is met.
Notably, the DJA was measured between the inferior (not superior) endplate of L5 and the superior endplate of S1. Although most previous studies measured DJA as the angle between the superior endplate of the LIV and the inferior endplate of the LIV–1 [2,32,33], we found that it was difficult to discern the superior endplate of L5 in substantial cases for the following reasons: (1) The superior endplate of L5 overlapped with the iliac crest. (2) In contrast to pediatric patients, fusion with the interbody cage was frequently performed at the most distal segment (L4–5) in patients with ASD. Subsequent bony trabeculation, cage subsidence, or pseudarthrosis reduced the clarity of the superior endplate of L5. Therefore, we consider that the measurement of the DJA, as the angle between the inferior endplate of L5 and the superior endplate of S1, is more appropriate.
This study had some limitations. First, it was retrospective in nature and therefore is likely to have inherent challenges such as selection bias and confounding factors. Second, although this was the largest study conducted to date, the cohort was not large enough to perform logistic regression analysis without the risk of overfitting. We believe that this issue can be addressed by a larger sample size, which can be accomplished with future multicenter studies involving a larger number of patients. Lastly, due to severe pain and poor general condition, a substantial number of patients could not complete the detailed outcome measures. The lack of patient-reported outcome measures was another limitation.
Although these limitations were noteworthy, this study also had its own strengths. First, we noticed that applying the DJK angle criteria from prior studies on pediatric patients to patients with ASD was inappropriate because the anatomical characteristics of the distal junction were different. Second, to evaluate the true junctional failures at the noninstrumented mobile segment (such as the junctional disc, ligament, or uninstrumented adjacent vertebrae), we only included patients that underwent surgery stopping at L5, rather than sacropelvic fixation. Consequently, we think long fusion surgery stopping at L5 may be acceptable in patients with ASD, without risk factors. However, when the patient is taking antidepressant/anxiolytic medication or has preoperative PI–LL mismatch >40°, sacropelvic fixation should be considered to avoid postoperative DJF.
CONCLUSION
The incidence of DJF was 21.1% in patients with ASD, who underwent fusion surgery stopping at L5. The use of antidepressant/anxiolytic medication and preoperative PI–LL mismatch >40° were independent risk factors for DJF. Without both factors, the incidence of DJF was greatly reduced (9.1%). DJF can be diagnosed based on postoperative changes in the DJA. If both criteria (last DJA>-5°, Δlast DJA–post DJA>5°) are met, DJF should be strongly suspected.
Notes
Conflict of Interest
The authors have nothing to disclose.
Funding/Support
This study received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Acknowledgments
We would like to express our gratitude to Seon Hwa Lee, a medical statistician, for her valuable assistance with the statistical analysis.
Author Contribution
Conceptualization: DJJ, HYC; Data curation: SB, HYC; Formal analysis: HYC; Methodology: HYC; Project administration: DJJ, HYC; Visualization: HYC; Writing – original draft: SHD, SB; Writing – review & editing: SHD, HYC.