| BackgroundPercutaneous vertebroplasty (PVP) was first introduced by Galibert et al. in 1987 for the treatment of vertebral hemangioma. Since then it has become widely used for patients with painful osteoporotic vertebral compression fractures (OVCF), spinal metastases, aggressive hemangioma and multiple myeloma. In PVP, bone cement (polymethyl methacrylate, PMMA) is injected into the vertebral cavity at the defect site to fill the void and stabilize the anterior column. Although PVP is a relatively safe procedure that provides quick pain relief and rapid recovery, the technique has associated complications. One of its most common complications is bone cement leakage.Bone cement can leak into the spinal canal or the intervertebral foramina, causing neurological complications such as paraplegia and root compression, although compression of the spinal cord or nerve root can be asymptomatic. Bone cement can also leak into the pulmonary artery, causing pulmonary embolism that can lead to death. The reported incidence of cement leakage varies from 5% to more than 80%. Early recognition and prompt treatment of bone cement leakage are essential to prevent a devastating outcome. A previous study reported that fracture severity and the viscosity of PMMA bone cement are general, strong and independent risk factors for cement leakage. That study found no explicit association between leakage and other potential risk factors, including surgical approach, volume of bone cement, fracture cause and sex. In addition, different leakage types might be associated with different risk factors. To provide doctors with information to guide their treatment choices, we conducted this retrospective study of 485 consecutive patients who underwent PVP at our institution in southern China over a 10-year period to determine the incidence and possible risk factors for bone cement leakage.PurposeTo provide doctors with information to guide their treatment choices, and determine the incidence and possible risk factors for bone cement leakage.Methods1. Inclusion Criteria and Exclusion CriteriaInclusion Criteria:Between August 2003 and August 2013, patients with OVCF, spinal metastases, hemangioma and underwent PVP in our institution.Patients with OVCF had to meet the following four inclusion criteria:(1) back pain lasting at least 2 months with no relief of symptoms with conservative treatment; (2) one or more OVCFs confirmed by radiography; (3) radiological examination showing that OVCF was related to back pain; and (4) the presence of bone marrow edema on T2-weighted short-tau inversion recovery sequences in the corresponding collapsed vertebral body on magnetic resonance imaging (MRI).Patients with spinal metastases had to meet three criteria:(1) severe back pain and radiological examination showing multiple osteolytic lesions; (2) MRI showing signal change and destruction of the vertebrae or vertebral arch; and (3) primary lesions such as lung cancer or breast cancer.Exclusion Criteria:Patients who underwent intervertebral fusion and instrumentation at the PVP level were excluded to control confounding factors.2.1maging Examination and Criteria:To correlate clinical symptoms with imaging findings, patients had plain anteroposterior and lateral radiographs of the spine; preoperative MRI was performed to confirm the diagnosis and determine the levels of surgery.(1) Fracture morphology classification:Fracture morphology was classified into three types:wedge, biconcave and crush, according to the classification of Genant et al..(2) Fracture severity classification:Based on the percentage of vertebral body collapse, fracture severity was characterized as mild (20-25% collapse), moderate (26-40%), or severe (>40%).(3) For purposes of statistical analysis, we added fourth categories for fracture morphology and fracture severity, "nearly normal" and "very mild" (body collapse <20%), respectively.3.Operation MethodsAll operations were performed by experienced spinal surgeons. Patients were asked to assume a prone position for the procedure. Biopsy needle insertion and bone cement injection were performed with the guidance of a C-arm X-ray machine (SIEMENS, Germany). After administration of local anesthesia or general anesthesia, one or two 11- or 13-gauge needles (KINETIC, China) were unilaterally or bilaterally inserted into the anterior third of the vertebral body transpedicularly. Under normal conditions, the unilateral approach was preferred; a bilateral approach was frequently chosen when the distribution of cement was unsatisfactory or asymmetrical. After placement of the needles into the anterior third of the vertebral body, bone cement (PMMA, Stryker Corporation, USA) was quickly injected into the vertebral body under X-ray guidance. The cement was injected during its "toothpaste-like" phase,2 min after mixing, to minimize the risk of extravasation. They were asked to return after three months, six months, and one year to monitor for longterm complications.4.How to identify the bone cement leakageAll enrolled patients underwent plain anteroposterior and lateral radiography of the spine the day after the procedure. Bone cement leakage was defined as the presence of any extravertebral cement.Using the postoperative X-ray images, we defined sixtypes of leakage according to a previous study, with some modification:leakage I) into the venous plexus beside the vertebrae, II) into the paravertebral soft tissue, III) into the spinal canal, IV) into the intervertebral foramina, V) into the intervertebral space and VI) into the needle channel.5.Potential risk factors included(1) Basic clinical imformation about patients:age, gender, diagnosis, type of fracture and fracture severity.(2) Risk factors about the PVP procedure:Procedural characteristics included level of surgery, surgical approach (unilateral or bilateral), type of anesthesia (local or general) and bone cement volume.6.Statistic methodsAll statistical analysis were performed with the Statistical Packages for Social Sciences V20.0 (SPSS, Chicago, IL, USA). Differences were tested using univariate analysis for symmetrically distributed variables and the nonparametric Wilcoxon test for other variables. We used binary logistic analysis to provide a model for predicting cement leakage. Multinomial logistic analysis was performed using a stepwise approach to identify independent predictors for the occurrence of each type of cement leakage. P values <0.05 were considered significant.Results:1.Clinical im formationA total of 485 patients (112 men and 373 women) with 804 treated vertebrae were included in our study. The mean patient age was 68.1 years (range,19 to 94 years). Among the 804 vertebrae studied,378 (47.0%) were thoracic,424 (52.8%) were lumbar,1 (0.1%) was cervical and 1 (0.1%) was sacral. Because the sample size for the cervical and sacral spine was too small for analysis, we excluded these vertebrae. Therefore, a total of 802 vertebrae were finally included. The main diagnoses were vertebral compression fracture in 733 vertebrae (91.4%), spinal metastases in 50 vertebrae (6.2%) and hemangioma in 19 vertebrae (2.4%).The incidence of bone cement leakage was 58.2%(467/802). The most common type of leakage was type I, with 175 cases (37.5% of all cases). The other five types sorted by decreasing frequency were type V (101 cases,21.6%), type II (98 cases, 21.0%), type III (54 cases,11.6%), type VI (23 cases,4.9%) and type IV (16 cases, 3.4%).2.Statistical results(1) Univariate analysis & Nonparametric test resultsUnivariate analysis and nonparametric test results showed that four factors had a significant association with bone cement leakage after PVP. These factors included volume of bone cement (P<0.001), fracture severity (P<0.001), surgical approach (P<0.001) and gender (P=0.016).To help surgeons with surgical planning, we repeated univariate analysis to determine whether the volume of cement remained a risk factor for cement leakage at different levels of the spine. Interestingly, in both the thoracic and lumbar spine, vertebrae that had leakage had received a larger volume of cement injection than those without leakage (P<0.001 in both levels).Univariate analysis showed no significant difference in the volume of bone cement used for each fracture grade, indicating that the volume of bone cement and fracture severity were two independent risk factors.(2) Binary logistic analysis resultsBinary logistic analysis was used to further evaluate the potential associations between bone cement leakage and clinical and radiographic characteristics. Two variables were found to be statistically significant:volume of bone cement (P<0.001) and fracture severity (P<0.001).(3)Multinomial logistic regression analysis resultsWe performed multinomial logistic regression analysis to detect the risk factors for each of the six types of cement leakage (Table 6). Surgical approach (P<0.001), operated level (P=0.032) and volume of bone cement (P<0.001) were found to be strong and significant predictors of type I bone cement leakage. Surgical approach (P=0.001), volume of bone cement (P<0.001), gender (P=0.004) and fracture severity (P=0.005) were identified as strong risk factors for type II leakage. Volume of bone cement (P<0.001) and surgical approach (P=0.001) were risk factors for type III leakage; fracture severity also approached statistical significance (P=0.065) for this group. Volume of bone cement (P=0.017) and surgical approach (P=0.016) were strong predictors of type IV leakage. Surgical approach (P=0.001), fracture severity (P<0.001), volume of bone cement (P<0.001) and level of surgery (P=0.034) were risk factors for type V bone cement leakage. Only bone cement volume (P<0.001) was a risk factor for type VI bone cement leakage.ConclusionIn conclusion, higher fracture severity grade and larger volume of bone cement were the two strongest independent risk factors for bone cement leakage after PVP. Individualized treatment of patients undergoing PVP is essential to decrease the incidence of leakage. |
PDF Full Text Request