Introduction
In recent years, the assessment of patient well-being has evolved significantly, with a growing emphasis on patient-centered outcomes, especially health-related quality of life (HRQOL). Most studies have reported pronounced enhancements in HRQOL outcomes among older adults undergoing lumbar arthrodesis surgery [1,2]. However, significant challenges persist in elderly populations, as bone mass reductions—reaching approximately 50% in women and 30% in men—can compromise the efficacy of primary pedicle implant fixation and escalate the risk of premature arthrodesis loosening [3]. This can lead to complications such as pseudoarthrosis, persistent pain, and the necessity for reoperation, significantly impacting patient well-being [4]. Despite surgical advancements, the relationship between screw loosening (SL) after lumbar arthrodesis surgery and its clinical implications remains poorly understood. Although certain studies have found no obvious difference in postoperative outcomes between patients with and without SL [5–7], others highlight its potential to exacerbate symptoms and the need for revision surgeries [8,9].
Our preliminary findings from a multicenter randomized controlled study demonstrated that titanium-coated polyetheretherketone (PEEK) cages achieved superior bone fusion compared to conventional PEEK cages after posterior lumbar interbody fusion (PLIF); however, this variance did not translate into meaningful differences in clinical outcomes [10]. Accordingly, our subsequent analysis explored this association, hypothesizing that SL might negatively influence HRQOL. The current study conducts a retrospective analysis to investigate the impact of SL following PLIF on HRQOL and explores associated risk factors through propensity score-matched analysis.
Materials and Methods
Institutional review board approval
This multicenter randomized controlled trial was approved by the Institutional Review Board of Hamamatsu University School of Medicine (research approval no., 15-276) and registered with the University Hospital Medical Information Network (UMIN) clinical trial registry (UMIN000022618). All participants provided informed consent prior to enrollment.
Participants
This study retrospectively analyzed data from a randomized trial conducted at three university hospitals and their affiliated institutions, between 2016 and 2018. We included 138 patients (75 men, 63 women; mean age, 67.0 years) who underwent single-level PLIF or transforaminal lumbar interbody fusion (TLIF) for conditions including degenerative spondylolisthesis, lumbar spinal canal stenosis, radiculopathy, disc herniation, or isthmic spondylolisthesis (Fig. 1). Patients aged 20 years or older with lumbar degenerative diseases who provided informed consent were eligible for inclusion. Those with a history of lumbar spine radiation therapy, bone tumors, metabolic bone diseases, or previous spinal surgeries were excluded from the study. Patients were randomly assigned in an open-label trial to receive either a titanium-coated PEEK cage (n=67) or a conventional PEEK cage (n=76). Postoperatively, all patients used a soft lumbar corset for at least 3 months and engaged in standardized physical therapy, including exercises back strengthening and walking exercises. Importantly, all patients were monitored for a minimum of 1 year, and the majority provided complete datasets suitable for analysis.
Radiographic, clinical, osteoporotic evaluation, and grouping
Patient characteristics and surgical details, including the type of surgery and level, were retrieved from the medical records. SL was defined radiologically as a continuous radiolucent zone exceeding 1 mm in width with a thin sclerotic border, as identified on axial computed tomography (CT) slices (Fig. 2) [11]. Cage subsidence was defined as the downward displacement of the cage from its initial postoperative position [12]. Intervertebral osseous union was defined radiographically as bony continuity between the upper and lower vertebrae, with coronal and sagittal CT slices in the central cage region classified as grade I. Following established criteria from previous studies, each CT slice was scored 1 point, yielding a maximum total score of 2 [13]. SL, cage subsidence, and intervertebral osseous union were independently assessed by four blinded physicians using three-dimensional CT scans at 1, 2, 4, 6, and 12 months postoperatively. Patients were categorized into two groups based on CT findings at 12 months following PLIF or TLIF: those with SL (SL group) and those without loosening (NL group) at 12 months.
The femoral neck bone mineral density (BMD) was determined preoperatively utilizing dual-energy X-ray absorptiometry. Procollagen type I amino-terminal propeptide and tartrate-resistant acid phosphatase 5b (TRACP-5b) levels were evaluated as indicators of bone metabolism. Whole spine standing radiographs were obtained preoperatively and at 2 and 12 months postoperatively. The following spinal parameters were measured: thoracic kyphosis (T4–T12), lumbar lordosis (LL), sagittal vertical axis (SVA), pelvic tilt (PT), sacral slope, pelvic incidence (PI), the difference between PI and LL (PI–LL), and Cobb angles.
Assessment of HRQOL and pain
Clinical and neurological symptoms were graded preoperatively and at 1, 2, 4, 6, 9, and 12 months postoperatively using the Oswestry Disability Index (ODI) scores [14] and Visual Analog Scale (VAS). ODI scores were used to assess the degree of disability associated with low back pain. A 100-mm VAS was used to quantify lower back pain intensity, with 0 denoting no pain and 100 representing the worst possible pain.
Statistical analysis
A preliminary analysis was conducted to compare the baseline, surgical, and osteoporotic data between the unmatched SL and NL groups. Given the considerable baseline variations between the two groups, propensity score matching was performed. Logistic regression analysis incorporating age, sex, and body mass index (BMI) to calculate the propensity score. The quality of fit was assessed employing C-statistics. In a subsequent analysis, the matched SL and NL groups were compared across baseline characteristics, surgical and osteoporotic data, whole-spine radiographic parameters, pain scores, and HRQOL outcomes.
Categorical variables are presented as absolute numbers and percentages and analyzed using either the chi-square test or Fisher’s exact test, as appropriate. The Shapiro-Wilk test was applied to ascertain the normality of the continuous variables. For continuous variables with normal distributions, means±standard deviations were reported and analyzed using unpaired t-tests. Nonnormally distributed data were reported as medians with interquartile ranges and analyzed using the Mann-Whitney U test. The interobserver reliability for the presence of SL was calculated by determining the intraclass correlation coefficient (ICC) and kappa coefficient. Statistical analyses were performed using the IBM SPSS software ver. 23.0 (IBM Corp., Armonk, NY, USA), with a p-value <0.05 deemed statistically significant.
Results
Baseline, surgical, and osteoporotic data between the SL and NL groups
The interobserver reliability for identifying SL was robust, as reflected by an ICC of 0.812 and a kappa coefficient of 0.811. PLIF was performed in 29 (21%) patients, and 109 (77%) in the SL and NL groups underwent PLIF surgery, respectively (Fig. 1). Among the 138 patients, no instances of screw backout were observed. Significant differences were observed between the groups in age, presence of lumbar herniation, femoral neck BMD, and serum TRACP-5b levels (p<0.05) (Table 1). A subgroup analysis revealed no statistically significant differences in the incidence of SL across the nine participating institutions (p=0.063).
Following propensity score matching, 22 patients were selected from each group (Fig. 1). The C-statistic for goodness of fit was 0.78 (95% confidence interval, 0.69–0.87; p<0.001) in the propensity score model. After the propensity score-matching analysis, no significant differences were observed between the SL and NL groups in terms of age, lumbar herniation status, femoral neck BMD, or serum TRACP5b levels (Table 2).
Whole spine radiographic assessment between the SL and NL groups
Table 3 details the pre- and postoperative whole-spine radiographic parameters for the SL and NL groups. The SL group exhibited significantly higher mean preoperative SVA, PT, SS, PI, and PI–LL values compared to the NL group (p=0.037, p=0.001, p=0.016, p=0.033, and p=0.001, respectively).
After propensity-score matched analysis, the two groups exhibited no significant difference regarding SVA, PI, and PI–LL pre- and postoperatively. However, the mean pre- and postoperative PT values remained significantly higher in the SL group than in the NL group preoperatively and at 2 and 12 months postoperatively (p=0.010, p=0.001, and p=0.002, respectively) (Table 4).
Change in ODI scores and VAS values between the SL and NL groups
Fig. 3A illustrates the variation in ODI scores between the unmatched groups. The SL and NL groups demonstrated significant differences in the alterations in ODI scores at 2, 4, 6, 9, and 12 months postoperatively (p=0.003, p=0.001, p<0.001, p<0.001, and p<0.001, respectively). Preoperative VAS scores for low back pain were comparable between the two groups (Fig. 3B). The SL group reported significantly higher VAS scores for low back pain at 2, 4, 6, and 9 months postoperatively than the NL group (p=0.020, p=0.039, p=0.033, and p<0.001, respectively).
Fig. 3C depicts the variation in ODI scores between the groups after the propensity score-matched analysis. At 9 months postoperatively, the change in the ODI scores differed significantly between the SL and NL groups (p=0.025). Fig. 3D illustrates the variation in the VAS scores of low back pain between the groups following propensity score matching. The change in the VAS scores for low back pain at 9 months postoperatively significantly differed between the SL and NL groups (p=0.023, respectively).
Occurrences of cage subsidence and intervertebral osseous union between the SL and NL groups
Postoperative changes in the proportions of patients with cage subsidence and intervertebral osseous union between the SL and NL groups at 12 months postoperatively are depicted in Fig. 4. The proportion of patients with cage subsidence at 12 months postoperatively was significantly greater in the SL group than in the NL group (p=0.041). At 6 and 12 months postoperatively, the proportion of patients achieving intervertebral osseous union was significantly greater in the NL group than in the SL group (p=0.016 and p=0.011, respectively).
Discussion
This study examined the impact of SL after PLIF surgery through a propensity score-matched analysis adjusted for age, sex, and BMI. Despite postoperative improvements in global spinal alignment in both groups, pre- and postoperative radiographic assessments revealed significant differences in PT between patients with and without SL. This finding clarified that preoperative and postoperative pelvic retroversion was significantly related to SL. Regarding HRQOL assessments, the preoperative ODI scores did not significantly vary between patients with and without SL. However, at 9 months postoperatively, patients with SL exhibited markedly smaller gains in ODI scores than those without SL. Furthermore, nonunion predominantly occurred at 6 months postoperatively. These results highlight the importance of preventing SL following PLIF, suggesting the potential for personalized interventions and improved prognostic assessments. However, it remains uncertain whether SL directly contributes to poor HRQOL or merely serves as a surrogate marker for nonunion or sagittal imbalance.
The relationship between SL and HRQOL remains poorly understood. The ODI score was used to assess the postoperative enhancement in HRQOL [15]. Notably, some studies indicate that VAS scores and ODI outcomes do not significantly differ between patients with and without SL [5–7]. However, other studies suggest that SL may influence clinical outcomes. Banno et al. [8] reported that patients with SL demonstrated poorer improvement in ODI scores 2 years following adult spinal deformity surgery than those without SL. Ohba et al. [9] reported poor improvement in ODI scores in patients with SL 1 year postoperatively, and SL occurrence was associated with back pain and recurrence. In this preliminary analysis, patients with SL reported higher VAS values for low back pain between 2 and 9 months postoperatively compared to those without SL. Patients with SL exhibited poor improvement in ODI scores, which persisted for a duration ranging from 2 to 12 months postoperatively. This could be attributed to insufficient bone fusion, causing screw instability and mechanical irritation of adjacent nerves and tissues, contributing to pain [16]. Although these results were notable, the influence of SL could not be conclusively determined without adjusting for age. Propensity score-matched analysis subsequently confirmed that patients with SL exhibited poor improvement in ODI scores at 9 months postoperatively.
Patients with SL demonstrated higher rates of lateral subluxation (>8 mm), elevated postoperative thoracolumbar kyphosis, higher postoperative SVA, and sagittal imbalance (>50 mm) [7]. A larger PI necessitates a correspondingly greater LL to maintain sagittal alignment, which is crucial for reducing the load on the adjacent segment [17]. Therefore, fusion and fixation may elevate stress on the spinal muscles and bone, potentially leading to SL, particularly in cases where PI increased over time. In this preliminary analysis, the preoperative LL did not vary between the groups; however, the SL group displayed significantly higher preoperative SVA, PT, PI, PI, and PI–LL values, potentially increasing the risk of SL. Postoperatively, patients with SL exhibited markedly higher SVA, PT, and PI–LL values, reflecting increased bending torque on distal structures resulting from greater sagittal imbalance [18]. Propensity score-matched analysis revealed that patients with SL tended to experience greater pelvic retroversion preoperatively and postoperatively compared to those without SL, emphasizing the role of preoperative and postoperative pelvic compensation in influencing spinal axis inclination and sagittal imbalance. This finding emphasizes the significance of pelvic compensation as a potential predictor for identifying patients at risk of SL after PLIF. Pelvic retroversion, as a compensatory mechanism for sagittal imbalance, may augment mechanical loading on the lumbosacral junction and predispose to SL.
The incidence of SL at 1 year ranged from 7.0% to 26.8% and increased from 24.7% to 32.0% over a 2-year follow-up [6,19]. Previous studies have reported SL rates of 1%–15% in patients without osteoporosis, rising to 60% in those with osteoporosis [7,20]. In this study, the incidence of SL was 21% at 12 months postoperatively, which is comparatively higher than previously reported rates. This finding may be attributed to the inclusion of older patients with osteoporosis in the study cohort (mean age, 67 years). Among the 161 patients enrolled in this study, one patient with screw backout was identified; however, this patient was excluded from this study due to an adverse event (Fig. 1). Among the 138 patients, there were no instances of screw backout.
Biomechanical factors, including screw placement and fusion levels, have been implicated in SL [21]; however, this study found no significant differences in the fusion levels between the groups. Osteoporosis, affecting 1/3 of women and 1/5 of men aged >50 years globally [22], is a significant risk factor for the occurrence of SL due to compromised bone quality. Aging impairs bone integrity, affecting screw stability and intervertebral osseous union, especially in older adults [23]. Poor bone quality, which is evident in decreased BMD and elevated TRACP-5b levels, reduces screw traction and insertion torque [24]. Elevated TRACP-5b levels, a bone metabolic marker of osteoclast activity, were associated with higher SL rates [20]. This preliminary analysis reaffirmed that older patients with osteoporosis were at elevated risk for SL after PLIF, aligning with a previous study [25]. Moreover, a lower femoral BMD was noted in the SL group, further underscoring the vital role of BMD and bone quality in SL occurrence [26]. A propensity score-matched analysis conducted to mitigate the influence of age, sex, and BMI between the groups revealed no significant difference in the lumbar herniation status, BMD, and TRACP-5b levels between the SL and NL groups.
This study discovered that SL was linked to poor improvement in HRQOL at 9 months postoperatively. The 9-month time point was emphasized because it was the only interval demonstrating statistically significant differences in both ODI and VAS after matching. Other time points demonstrated similar trends; however, these did not reach statistical significance. We hypothesize that the poor improvement in HRQOL was attributable to spinal alignment deterioration and intervertebral nonunion secondary to SL, rather than a direct effect of SL itself. Therefore, preventing SL is essential for optimizing HRQOL in the early postoperative period. According to Soshi et al. [27], the pull-out strength of pedicle screws is influenced by the severity of osteoporosis. Therefore, preventing SL requires enhancing the screw–bone interface, which is being investigated through preoperative administration of bone formation agents such as teriparatide [13], hydroxyapatite-coated screws [28], or filling the insertion holes with hydroxyapatite granules or bone cement to augment the initial fixation [29]. In addition to thorough guidance on lumbar corset use, employing modified screw trajectories and optimizing pedicle screw fixation—through appropriate selection of screw length and diameter—are promising methods for enhancing pedicle screw stability in spine surgery [30].
This study has certain limitations. First, this was a retrospective study with a relatively small sample size, and the follow-up period was restricted to 12 months postoperatively. Additionally, selection bias may have arisen from the exclusion of patients with incomplete datasets and adverse events, potentially limiting the generalizability of our findings. Second, SL was evaluated using CT imaging, which, although an optimal method for assessing bone conditions, entails radiation exposure. To mitigate this risk, this study employed a technique with 50% reduced radiation, and an iterative reconstruction method was employed for image generation. Although SL was defined based on radiolucency, we did not further quantify its severity or evaluate its relationship with clinical symptoms. Third, although propensity score matching was applied, TRACP-5b exhibited a postmatching p-value of 0.096, indicating a possible residual imbalance. Finally, the 12-month follow-up period precluded a detailed evaluation of long-term outcomes over 2 years, necessitating an extended follow-up for further accurate assessment.
Conclusions
After adjusting for age, sex, and BMI, pre- and postoperative pelvic retroversion were significantly linked to SL occurrence. Patients with SL exhibited poor improvement in ODI and VAS scores at 9 months postoperatively, with nonunion occurring primarily at 6 months postoperatively. Whether SL directly contributes to poor HRQOL or merely serves as a marker for other underlying factors such as intervertebral nonunion or sagittal imbalance remains unclear. We believe that potential strategies for preventing SL, including thorough guidance on lumbar corset use and the use of bone-forming agents or cement-augmented pedicle screws, could enhance early postoperative HRQOL and reduce the incidence of postoperative complications in patients exhibiting pelvic retroversion.
Key Points
In this study, the incidence of screw loosening (SL) occurrence was 21% at 12 months postoperatively, which is relatively high compared to previous reports.
Pre- and postoperative pelvic retroversion were significantly associated with SL occurrence after adjusting for age, sex, and body mass index.
Patients with SL exhibited poorer improvement in the Oswestry Disability Index and Visual Analog Scale scores at 9 months postoperatively, with nonunion predominantly occurring by 6 months.
Potential strategies for preventing SL, including lumbar corset use, bone-forming agents, and cement-augmented pedicle screws, may enhance early postoperative health-related quality of life and reduce postoperative complications in patients with pelvic retroversion.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
