Increased sagittal vertical axis is associated with less effective control of acute pain following vertebroplasty

Patient enrolment

Between September 2012 and March 2015, consecutive patients aged over 70 years with a single-level acute stable OVCF at the thoracolumbar junction (T10 to L2) were prospectively recruited by a trained clinical research coordinator in the clinic. Acute OVCF was confirmed by MRI of the thoracic or lumbar spine which routinely includes a T2-weighted sagittal plane of the whole spine with fat suppression (1.5T Intera MRI scanner; Philips Medical Systems, Andover, Massachusetts). Osteoporosis was diagnosed by a T-score of ⩽ 2.5 on dual-energy X-ray absorptiometry (DXA; QDR - 4500A, Hologic; Waltham, Massachusetts).

To overcome any potential for confounding of the non-randomised study design, we excluded patients with known general contraindications for vertebroplasty. The known contraindications were as follows:^4,10 severe cardiopulmonary disease; uncorrectable coagulopathy; severe OVCF compressed over 75%; unstable fracture with the involvement of the posterior elements; spinal cord or nerve root compression causing radicular pain; and active sign of infection. The patients with previous vertebroplasty, pathologic fracture, workers’ compensation, or any diagnosed psychological problem, including depression, were also excluded.

12-week post-traumatic management protocol including vertebroplasty

At the time of diagnosis, a rigid thoracolumbosacral orthosis (TLSO) was applied, and a non-steroidal anti-inflammatory drug or opioid was administered. Tolerable weight-bearing was started immediately after bracing. The treatment was performed at an outpatient clinic, and only patients who could not walk due to severe back pain, were admitted to the hospital.

The patients were routinely followed at three weeks after the diagnosis at the outpatient clinic. The location and severity of back pain on the fracture site was re-evaluated. If the location of the back pain was confirmed as that of the fracture site and patients were not satisfied with pain improvement after three weeks of conservative treatment, vertebroplasty was suggested only as a possible treatment option for reducing acute pain. The risks and benefits were described to the patient who then decided whether or not to go ahead with vertebroplasty.

Percutaneous vertebroplasty was performed with the patient under local anaesthesia using the unilateral or bilateral transpedicular approach with a ten-gauge needle. Approximately 4 cc of polymethylmethacrylate was injected under fluoroscopic guidance.

Bisphosphonates (risedronate 35 mg) were recommended at the routine follow-up visit of six weeks after diagnosis. Bisphosphonate therapy was delayed in seven patients who were scheduled for invasive dental procedures. Teriparatide (20 ug, daily subcutaneous injection) was used in five patients. TLSO was maintained for 12 weeks.

A total of 342 patients finished the routine 12-week management course for a stable OVCF (Fig. 1). Written informed consent was obtained before enrolment, and the following assessments were conducted by a trained clinical research coordinator, who followed detailed testing protocols. Data collection and 12-week management course were routine procedures for all patients who visit our clinic and included patients were not subjected to extra procedures or questions.

Fig. 1

Flow chart showing patient enrolment and follow-up.

Assessment by interview

Sociodemographic characteristics were assessed in a patient interview. Characteristics included age, gender, and level of education. Height and weight for body mass index (BMI) were measured. A medical history was obtained from every patient based on the Charlson comorbidity index.¹¹ Previous vertebral fracture was evaluated based on the patient’s history and MRI findings. According to the previous article,¹² the cause of fracture was classified as follows: falling down; lifting heavy objects; and no specific trauma. The duration since time of pain onset was recorded.

Assessment of bone mineral density

Bone mineral density (BMD) was measured by DXA (QDR-4500A; Hologic), and the values of total lumbar and total hip were used for analysis.

Assessment of the radiological data

According to the previous reports, several validated radiological assessments were undertaken using the radiographs taken three weeks after the diagnosis. From the standing anteroposterior radiographs of the thoracolumbar spine, the severity of the body compression was assessed by anterior vertebral body compression percentage.^13-16 As a local sagittal alignment, the Cobb angle, formed by a line drawn parallel to the superior endplate of one vertebra above the fracture and a line drawn parallel to the inferior endplate of the vertebra one level below the fracture, was measured.¹³ Spinal canal encroachment was measured on MRI in relation to the cross-sectional spinal canal area at the level of injury relative to the cross section adjacent to the injured level.^13,17 From the standing whole spine radiograph, sagittal vertical axis (SVA) was measured.

Assessment of back pain

The location of the pain was recorded using a validated McGill Pain Map,^18,19 a homunculus showing the front and back of the human figure. Back pain intensity was measured with a VAS, ranging from 0 to 10, where 0 is no pain, and assessed as location-specific ratings (for upper and/or lower back). Pain assessment was carried out at diagnosis and at three, four, six, and 12 weeks after the diagnosis.

Data collection from chart review on complications from OVCF

According to Lee et al,²⁰ data on complications from OVCFs were collected and classified into three types: procedure-related; medical; and adjacent vertebral fracture. Cement leakage was identified and also classified into three types: leaks via basivertebral vein; leaks via segmental vein; and leaks through a cortical defect.²¹

Statistical analysis

Comparison of each independent variable between the two groups (vertebroplasty versus conservative treatment) was done by independent t-test for continuous variables and Pearson chi-squared test or Fisher’s exact test for categorical variable (Table I).

The effect of vertebroplasty on acute back pain (VAS) was calculated with the use of analysis-of-covariance (ANCOVA) models including adjustment for possible confounders (Tables II and III). Each treatment group (vertebroplasty group and conservative group) was divided into subgroups according to the independent variables (Table I), and ANCOVA analyses were also performed to compare the effect of treatment between each subgroup. Through subgroup analysis by the ANCOVA model, pre-operative prognostic factors associated with a less favourable outcome after vertebroplasty were identified among the independent variables, and multivariate logistic regression analysis was performed to confirm the relationship between pre-operative prognostic factors and a less favourable outcome of vertebroplasty. A less favorable outcome of OVCF is closely related to adjacent vertebral fracture. Adjacent vertebral fracture can occur regardless of treatment method, and act as a confounder for the association between pre-operative prognostic factors of vertebroplasty and a less favourable outcome. Therefore, multivariate logistic regression analysis was also performed to identify association between pre-operative prognostic factors and adjacent vertebral fracture which could act as a confounder. All statistical analyses were performed using the SPSS 17.0.0 statistics package (SPSS, IBM, Chicago, Illinois). A value of p < 0.05 was accepted as significant.

Baseline characteristics of the patients

Among 342 patients who finished the routine 12-week management course for a stable OVCF, 120 patients underwent vertebroplasty (35.1%) (Table I). There were no significant differences in the baseline characteristics of the patients between the two groups except for the number of medical comorbidities (p = 0.006) (Table I), which were higher in the vertebroplasty group. The intensity of pain at the time of diagnosis was not significantly different between the two groups (p = 0.161) (Table I).

Effect of vertebroplasty: overall treatment

Vertebroplasty was performed at a day between three and four weeks after the diagnosis, and there was a significant difference in the pain VAS score at three weeks after diagnosis (aka pre-operatively for the vertebroplasty group) between the two groups (Table II). The effect of vertebroplasty on acute back pain was analysed with ANCOVA models, including adjustment for the pain VAS score of three weeks after diagnosis. After vertebroplasty the effect of vertebroplasty was significant, measured at four and six weeks after diagnosis (one or three weeks after vertebroplasty) (Table II). However, the effect of vertebroplasty was not significant at 12 weeks after diagnosis (nine weeks after vertebroplasty) (p = 0.794) (Table II).

Effect of vertebroplasty: subgroup analysis

We performed subgroup analyses to identify pre-operative prognostic factors associated with a less favourable outcome after vertebroplasty in comparison with conservative treatment. At four or six weeks after the diagnosis (at one or three weeks after vertebroplasty), the effect of vertebroplasty still remained significant regardless of advanced age (> 75 years), high BMI (> 25 kg/m²), medical comorbidity, previous fracture, longer pain duration before diagnosis (more than two weeks), decreased bone mineral density (T-score < -3.0), increased anterior vertebral body compression ratio (> 40%) and increased canal encroachment (> 20%) (Supplementery table). The effect of vertebroplasty in all of these subgroups was not significant at 12 weeks after the diagnosis (Supplementery table). However, the effect of vertebroplasty was not significant in patients with increased SVA (> 5 cm) at all time points (Supplementery table) (p = 0.563 at four weeks, p = 0.329 at six weeks, and p = 0.505 at 12 weeks).

Multivariate logistic regression analysis on the relationship between increased SVA (> 5 cm) and a less favourable outcome of vertebroplasty

Multivariate logistic regression analysis was performed to confirm the relationship between increased SVA (> 5 cm) and a less favourable outcome of vertebroplasty. According to Ostelo et al,²² the minimal important change (MIC) for pain alleviation is considered to be two points in VAS score (0 to 10), therefore a less favourable outcome after vertebroplasty was defined as an improvement in pain at less than two points using the VAS. Vertebroplasty was performed between three and four weeks after the diagnosis, and there was a significant difference in pain VAS at three weeks after the diagnosis (pre-operative value for the vertebroplasty group) between the two groups (Table II). Therefore, multivariate adjustments were done for medical comorbidity and pain VAS score at three weeks after diagnosis according to the results of our analysis (Tables I and II). Even after multivariate adjustment, increased SVA was closely related to a less favourable outcomes after vertebroplasty (Table III).

Analysis of complications according to the treatment method

Complications of vertebroplasty were classified into three types:²⁰ procedure-related; medical; and adjacent vertebral fracture (Table IV).

The most common procedure-related complication was cement leakage. There were 29 cases (24.2%) of cement leakage in the vertebroplasty group, 11 leaks via the basivertebral vein, seven leaks via the segmental vein and 11 leaks through a cortical defect. However, there were no fatal procedure-related complications such as cement embolism or neurological deficit (Table IV). There were no differences in medical complications between the two groups (p = 0.873) (Table IV).

Adjacent vertebral fracture can occur regardless of treatment method. Adjacent fracture occurred in 5.8% (13 of 222 patients) of the conservative treatment group and in 10.8% (13 of 120 patients) of the vertebroplasty group. There was no statistical difference in occurrence of adjacent fracture between the two groups (p = 0.097). Logistic regression analyses were carried out to identify the association, if any, between increased SVA and risk of adjacent fracture which could act as a confounder for the association between increased SVA and a less favourable outcome after vertebroplasty. After multiple logistic regression analysis, increased SVA (> 5 cm) was not associated with increased risk of adjacent fracture in either the vertebroplasty group or the conservative treatment group (Table V).