Risk Factors for Bone Cement Displacement After Percutaneous Kyphoplasty in Osteoporotic Vertebral Fractures: A Retrospective Analysis

Introduction

Osteoporosis (OP) is a common disease in elderly people, which can lead to bone loss and an increased risk of fractures, and the most common type of fractures associated with OP is osteoporotic vertebral fractures (OVFs) [1]. OVFs can lead to unbearable back pain, kyphosis, bedridden conditions, and further osteoporosis, which will seriously affect the patient’s quality of life. Unfortunately, the prevalence of OVFs remains high, contributing to a persistently severe disease burden. For example, in urban China in 2017, the incidence of OVFs among individuals aged 50 years or older was reported to be 152.13 per 100 000 person-years [2].

Vertebroplasty was represented by percutaneous kyphoplasty (PKP) and percutaneous vertebroplasty (PVP), and the initial experience with vertebroplasty was the use of PVP by French doctor Galibert in the 1980s for the treatment of vertebral angioma [3–5]. Vertebroplasty has been increasingly employed in the management of OVFs over the past 2 decades. However, the application of vertebroplasty is accompanied by some postoperative complications, such as vertebral refracture, cement leakage, bleeding, transient hypotension, shock, and pulmonary embolism [6]. Among them, bone cement displacement (BCD) is a complication that receives relatively little attention but can lead to serious consequences, such as exacerbated kyphosis, unrelieved pain, and even neurologic damage [7,8]. Patients with severe BCD may have to undergo open revision surgery, with serious trauma and high difficulty, to take out bone cement and rebuild spinal stability.

BCD was first introduced as a case report in 2003 [9]. Gao [8,10], Qi [11], and Chang [7] systematically studied the risk factors for postoperative BCD, and thought that high restoration of Cobb angle, cement leakage, the intravertebral vacuum cleft (IVC) sign, and low distribution score might be related to BCD. Zhong [12], Ma [13], and Guo [14] found that vertebroplasty-bone cement bridging screw system combined therapy could be an effective preventive measure to prevent the occurrence of BCD. However, the risk factors of BCD after vertebroplasty are unclear, and there have been few studies on the risk factors of BCD, especially those focusing on BCD after PKP. Therefore, the present study reviewed the records of patients who underwent PKP at our center to clarify the risk factors for BCD following PKP.

Material and Methods

STUDY SUBJECTS:

The study was approved by the Medical Ethics Committee of Peking University People’s Hospital (No. 2024PHB007-001) and complied with the ethical standards of the Declaration of Helsinki. Informed consent was waived for this retrospective study.

The inclusion criteria were as follows: (1) fragility fracture, (2) preoperative thoracic back pain or low back pain, with or without lower-extremity symptoms, (3) magnetic resonance imaging (MRI) showing vertebral edema, low signal on T1-weighted images, low or mixed signal on T2-weighted images, and fat-suppressed sequences showing high signal, and (4) patients with single-segment vertebral fracture undergoing single-entry PKP.

The exclusion criteria were as follows: (1) patients with spinal metastases, (2) multi-segmental vertebral fracture, (3) burst fracture, (4) previous open spine surgery, and (5) incomplete clinical data.

The diagnosis of BCD was made when lateral X-ray film showed that the bone cement was obviously displaced to the front of vertebral body or computed tomography (CT) showed a displacement greater than 2 mm [11,10]. Two experienced spinal surgeons independently reviewed the films, and any discrepancies in the evaluation were resolved after discussion by another doctor with higher seniority. Typical cases of BCD following PKP are shown in Figures 1 and 2.

The inpatient and outpatient follow-up data of 463 patients who underwent single-entry PKP from June 2016 to August 2022 at the Department of Spine Surgery, Peking University People’s Hospital, were individually reviewed. Ultimately, after 6–42 months of follow-up, a total of 181 patients were included, with a median 7.00 months follow-up, a mean age of 73.67±8.43 years, and 76.8% were females.

SURGICAL PROCEDURE:

All patients underwent single-entry PKP performed by senior surgeons with over 10 years of experience in our center. The procedure was carried out with the patient in a prone position following routine disinfection. Under fluoroscopy guidance, the surgeon identified the body surface position corresponding to the pedicle of the fractured vertebral body and administered local anesthesia using 2% lidocaine. After successful anesthesia, a 0.5-cm skin incision was made at this position. A puncture needle was then inserted with a cannula into the entry point of the pedicle and driven into the pedicle while adjusting the direction under fluoroscopy. Subsequently, the hollow-core needle was removed and replaced with a solid needle, which was then advanced into the vertebral body. A larger cannula was utilized to expand and shape the pedicle channel. Following this, an expansion balloon was placed into the fractured vertebral body by the surgeon, through which contrast agent was injected using a syringe equipped with a pressure gauge. The position and expansion of the balloon were observed under fluoroscopy while monitoring pressure levels. The contrast agent was injected at a volume of approximately 1.5~3.0 ml as the pressure gradually increased. Following decompression, the balloon was removed, and semisolid polymethyl methacrylate (PMMA) bone cement (Mendec^® Spine, Tecres S.p.a., Italy) was injected into the vertebral body. Finally, the cannula was removed, and the wound was covered with sterile dressing prior to transfer to the ward upon normalization of vital signs. The patient was instructed to wear braces for at least 1 month after the operation.

ASSESSMENT:

The following data were collected: age, sex, body mass index (BMI), underlying disease, history of fracture at other sites, injected bone cement volume, and postoperative anti-osteoporosis treatment, fractured vertebral segment, preoperative IVC, postoperative distance between the bone cement and anterior edge of the vertebral body, sagittal position of the cement filling, contact between the bone cement and the endplate, distance between the bone cement and the vertebral endplates, collapse rate of the vertebral body, recovery rate of the vertebral anterior margin height, restoration rate of the Cobb angle, bone cement distribution score, and bone cement leakage.

The calculations and definitions used were:

The sagittal position of the cement filling, the contact between the bone cement and the endplate, distance between the bone cement and the vertebral endplates, bone cement distribution score, preoperative fractured vertebral body height, postoperative height of the fractured vertebral anterior margin, and postoperative Cobb angle were also measured based on the lateral spinal X-ray film taken on the first day after surgery.

STATISTICAL ANALYSIS:

The patients were categorized into the following 2 groups: the bone cement displacement group and the bone cement non-displacement group. (1) SPSS (version 26, IBM) was used to compare the data between the 2 groups, and the level of significance was set at P<0.05. 1) The data were divided into continuous and categorical variables. Data normality was assessed using the Shapiro-Wilk test. The comparison of data between the 2 groups was performed by t test and chi-squared test. The independent samples t test was performed to compare the normally distributed continuous variables between the 2 groups, and the Mann-Whitney U test was performed to compare the non-normally distributed continuous variables. For categorical variables, the chi-square test was used to compare the 2 groups. 2) The above univariate analyses were used to determine the possible risk factors for BCD after PKP. Binary logistic regression analysis was performed to identify independent risk factors for cement displacement. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was used to evaluate the discriminative ability. 3) Using R 4.3.3 and RStudio (Version 2023.12.1, Posit Software), classification results of logistic regression were represented by confusion matrix as a heatmap and the predictive ability was assessed by logistic regression analysis.

Results

GENERAL:

A total of 181 patients with complete clinical data were included in this study (76.8% females, mean age 73.67±8.43 years). Of these patients, 12 (6.6%) experienced BCD after PKP and 169 (93.4%) did not experience BCD. All patients underwent single-entry PKP. The majority of vertebral fractures (n=181) occurred in the thoracolumbar segment (148, 82%, Figure 5). The time from surgery to the occurrence of BCD ranged from 1 to 42 months, with a median time of 7.00 (2.00, 17.50) months.

UNIVARIATE ANALYSIS:

The results of univariate analysis (Table 1) showed that BCD after PKP was not related to sex, age, underlying disease, history of fracture at other sites, fractured vertebral segment, preoperative vertebral height, collapse rate of the vertebral body, preoperative vertebral anterior margin height, pre- or postoperative Cobb angle, volume of cement, or postoperative anti-osteoporosis treatment (P>0.05).

BMI, preoperative IVC, postoperative vertebral anterior margin height, recovery rate of vertebral anterior margin height, restoration rate of the Cobb angle, sagittal position of cement filling, noncontact between bone cement and endplates, distance between bone cement and vertebral endplates, postoperative bone cement distribution score, and cement leakage were correlated with BCD following PKP (P<0.05).

BINARY LOGISTIC REGRESSION ANALYSIS:

The independent risk factors (Table 2) associated with cement displacement after PKP were BMI (OR=1.301, 95% CI 1.002~1.689), the restoration rate of the Cobb angle (OR=1.036, 95% CI 1.007~1.065), the distance between bone cement and vertebral endplates (OR=16.473, 95% CI 1.440~188.395), and anterior bone cement leakage (OR=29.200, 95% CI 3.552~240.027).

RECEIVER OPERATING CHARACTERISTIC (ROC) CURVE AND CONFUSION MATRIX:

The ROC curve and confusion matrix were used to verify the predictive power of logistic regression for BCD after PKP in OVFs. The Youden’s index was calculated as “sensitivity – (1-specificity)” [18]. The optimal cutoff values were derived by identifying the index value at the point on the ROC curve that corresponded to the highest Youden’s index, which was essentially the coordinate closest to (0, 1.0) on the ROC curve (Figure 6).

The ROC curve analysis (Table 3) revealed that the area under the curve (AUC) of BMI was 0.687 (95% CI 0.510~0.864), the optimal cutoff value was 25.780 kg/m2, the sensitivity was 0.727, and the specificity was 0.707. The AUC for restoring the Cobb angle was 0.759 (95% CI 0.591~0.927), the optimal cutoff value was 11.095, the sensitivity was 0.818, and the specificity was 0.701. The AUC for the distance between the bone cement and vertebral endplates was 0.762 (95% CI 0.667~0.858), the optimal cutoff value was 0.175 cm, the sensitivity was 1, and the specificity was 0.539. The AUC for cement leakage was 0.689 (95% CI 0.540~0.839), the sensitivity was 0.727, and the specificity was 0.713. Furthermore, logistic regression analysis demonstrated remarkable predictive efficacy for BCD, with an AUC of 0.928, a sensitivity of 0.909, a specificity of 0.910 and a Youden’s index of 0.819.

The confusion matrix summarized the prediction results of the logistic regression, displayed as a heatmap of situation analysis (Figure 7). The accuracy, precision, recall, and F1 score were 0.967, 0.857, 0.545, and 0.666, respectively. Combined with the ROC curve and confusion matrix, logistic regression could effectively predict BCD.

Discussion

Since the initial case reported by Tsai [9] in 2003, there have been numerous case reports of postoperative BCD [11,19,20], yet few systematic studies have investigated its risk factors. In 2018, Nakamae [21] explored the risk factors for bone cement loosening for the first time, and it was not until 2022 that the first systematic studies of risk factors of BCD were performed. Due to the limited availability of recent research, with few studies focusing on PKP, it was imperative to conduct this systematic retrospective study to elucidate the incidence and risk factors for BCD after PKP.

In 2022, Gao [8] conducted a large sample study including 1538 patients and reported that the independent risk factors of BCD after vertebroplasty in OVFs included high restoration of Cobb angle, cement leakage, small degree of bone cement interweaving, and non-postoperative osteoporosis treatment, which was basically consistent with the conclusion of this study. In Gao’s study, patients with OVFs underwent vertebroplasty involving both PKP and PVP, and the overall incidence of BCD after vertebroplasty was 5%, including 6.02% after PKP and 1.95% after PVP, which was similar to the 6.6% incidence of BCD after PKP in our study. In 2023, Qi [11] and Gao [10] studied the risk factors of BCD after PKP and in patients with Kümmell’s disease, respectively, with the conclusions that the low distribution score, the intravertebral vacuum cleft (IVC) sign, and bone cement anterior leakage were independent risk factors for BCD after PKP, and that high restoration of the Cobb angle, uneven cement distribution, cement leakage, anterior cortex defect, and thoracolumbar fracture were risk factors for BCD in Kümmell’s disease. In 2024, Chang [7] reviewed the records of 203 patients and again found that preoperative IVC was an independent risk factor for BCD following PKP. The research conducted by Qi and Chang had significant similarities with our study, as investigations aimed to identify the risk factors associated with BCD following PKP, and reached similar conclusions, indicating that BCD after PKP was closely linked to IVC, bone cement leakage, and other contributing factors.

The restoration rate of the Cobb angle, which served as a critical metric for assessing spinal reconstruction following PKP, was associated with postoperative BCD and vertebral refracture [22,23]. It has been suggested that a high restoration rate of the Cobb angle might contribute to the rigidity of structural connections within the vertebral body, which could result in decreased flexibility and mobility [8], an imbalanced distribution of spinal stress, and a potential increase in stress at the bone cement interface [10], promoting BCD. Vertebroplasty-bone cement bridging screw system combined therapy was found prevent BCD because of its excellent ability to correct kyphosis and maintain Cobb angle [12–14]. Furthermore, Cao [24] demonstrated that the limited restoration rate of the Cobb angle following vertebroplasty might result in spinal stress alterations and overall sagittal imbalance, leading to vertebral body instability. Therefore, it is imperative to focus on segmental correction of kyphosis in patients with OVFs to restore the overall sagittal balance of the spine rather than solely pursuing recovery of vertebral height and correction of kyphosis in a single compressed vertebral body.

Bone cement leakage is a common complication subsequent to PKP, with anterior cement leakage being the most prevalent complication [9]. Not surprisingly, the incidence of anterior cement leakage was 9.5% in this study, surpassing that of any other form of leakage. Bone cement leakage was identified as a significant independent risk factor in BCD after PKP, particularly anterior cement leakage (OR=29.200, 95% CI 3.552~240.027), which is often the result of cortical defects [25]. When cortical defects occur, the resistance to forward movement of the bone cement weakens [13,26], especially in patients diagnosed with Kümmell’s disease.

The association between BMI and spinal stability following PKP remains inconclusive. Our study revealed a significant correlation between high BMI and postoperative BCD (OR 1.301, 95% CI 1.002–1.689), as determined by binary logistic regression analysis. Our findings were consistent with Dong [27], who conducted a retrospective study involving 346 patients and concluded that a high BMI was an independent risk factor for vertebral refracture after PKP. The possible reason was that individuals with a high BMI might experience increased pressure on the vertebral body, leading to a heightened susceptibility to vertebral compression fractures[28].

The distance between the bone cement and the endplates was found to be an independent risk factor for BCD in this study, which had not been previously reported. Although the relationship between the bone cement-endplate distance and the risk of BCD has not been extensively studied, Li [29] and Hou [30] noted that the greater the distance between the bone cement and the endplates is, the greater the risk of vertebral refracture. This was attributed to the concentration of stress on the fragile cancellous bone with osteoporosis around the cement when it was distant from the endplates [31]. After PKP, the distribution of bone cement was insufficient and tended to be lumpy, which resulted in a greater distance between the bone cement and endplates [29] and the stress further concentrated around bone cement. In addition, Sun [32] proposed that the proximity of bone cement to the upper or lower endplates could help alleviating pain and improve prognosis.

The discussion regarding the volume of bone cement injection was crucial for understanding both BCD and overall postoperative spinal stability. Some believed that smaller amounts of cement were just as satisfactory as placing a larger amount of cement. Wang [33] posited that when the injected dose of bone cement was adequate to ensure sufficient rigidity for the vertebral body, excessive injection did not reduce complication rates nor significantly enhance patients’ prognosis; rather, it might even elevate complications such as BCD. In this study, the displacement group and non-displacement group both received a small dosage of bone cement, with an average injection volume below 3 ml, and there was no significant difference in the volume of bone cement between 2 groups. Insufficient amounts of injected bone cement can result in a reduced contact area between the bone cement and surrounding cancellous bone, which can lead to excessive pressure at the interface between the 2 materials, complicating stable formation at their contact surface – a factor contributing to BCD. Currently, the optimal volume recommended for bone cement injections is 24% of vertebral body volume [34]: 6.0–8.0 ml for lumbar segments, 3.0–4.0 ml for thoracolumbar segments, and 2.5–3.0 ml for thoracic segments [35].

This study has several limitations. First, this was a single-center, retrospective study with a limited number of subjects. Multicenter, prospective research with a large sample size is needed to verify the conclusions of this study. Second, we only collected data from patients treated with PKP after OVFs. Over the past decade, our center has primarily concentrated on PKP for OVFs, with PVP being performed infrequently. Consequently, we face challenges in gathering data regarding PVP. Moving forward, we plan to undertake multicenter studies that will incorporate PVP data and analyze the risk factors associated with BCD following vertebroplasty. Thirdly, there were relatively few studies on risk factors for BCD, making meta-analysis difficult. The issue of BCD did not gain widespread recognition until after 2022, prior to which all studies on this topic were limited to case reports. Following 2022, systematic investigations into BCD began to emerge gradually, which were all included in this discussion. Despite the increasing attention, the relatively late initiation of research resulted in a limited number of studies focusing on the risk factors associated with BCD, rendering it technically unfeasible to conduct an effective meta-analysis at this stage. This underscores the significance of our study, which aimed to elucidate the risk factors for bone cement displacement through research conducted at our center and contributed towards future comprehensive meta-analyses based on a substantial body of literature. Finally, this study did not collect bone mineral density (BMD) data for patients. All subjects included in this study were patients who experienced vertebral fracture after minor trauma, equivalent to falling from a standing height or lower, and who could be clinically diagnosed with severe osteoporosis [36]. Therefore, we did not use dual-energy X-ray as a routine examination for patients for the purpose of scientific research, which can increase economic burden and pain during movement. Furthermore, there is no doubt that BMD is closely related to postoperative complications of vertebroplasty, and more research is needed to explore the relationship between BMD and BCD after vertebroplasty in the future.

Conclusions

Patients with a high BMI, a high restoration rate of the Cobb angle, a large distance between the bone cement and vertebral endplates, and bone cement leakage have an increased risk of bone cement displacement after PKP.