15.3.1 Vertebroplasty

Early uptake of VP was rapid and driven by positive early data from observational studies and meta-analyses. However, two RCTs having a total of approximately 200 patients across both studies published in the New England Journal of Medicine (NEJM) found that VP offered no significant benefit over NSM. Since that time, these articles have been widely discredited and downgraded in their level of evidence category, but they have provided incentive for additional investigation of the efficacy of VP. Several large randomized trials have been released in recent years, focusing on rigorous inclusion criteria and minimizing methodological limitations. The key findings of the main prospective VP trials are summarized in ▶Table 15.1.

The 2007 VERTOS trial was the first multicenter prospective RCT to compare VP to medical management for pain relief in osteoporotic VCFs. ⁸ Inclusion criteria included severe back pain persisting despite medical therapy for at least 6 weeks, fracture aged less than 6 months, focal tenderness on physical examination, and bone marrow edema on MRI. In total, 34 patients were randomized to receive VP (n = 18) or conservative management (n = 16). The primary outcome measures were back pain intensity (as measured by the VAS) and analgesic use at 1 day and 2 weeks. Analgesic requirement was quantified by assigning ordinal variables to different analgesic types: 0 (no analgesia required), 1 (paracetamol/acetaminophen), 2 (nonsteroidal anti-inflammatory medications) or 3 (opioids).

VP resulted in significant pain relief at 1 day postprocedure, with reduction in the baseline mean VAS from 7.1 to 4.7 (difference between groups −2.4 in favor of VP). While this effect was not maintained at the 2-week endpoint, 88% of patients from the conservative group had crossed over to the VP group. Analgesic use was reduced in the VP group (–1.4; 95% confidence interval [CI] −2.1 to −0.8). Secondary outcomes of disability—defined by Roland–Morris Disability Questionnaire (RMDQ)—and quality of life (QOL; Quality of Life Questionnaire of the European Foundation for Osteoporosis [QUALEFFO]) were also significantly improved in the VP group.

The main limitations of the VERTOS trial were its small size and lack of blinding. No long-term follow-up was possible, as crossover was permitted after 2 weeks and 14 out of 16 patients in the conservative group requested crossover to VP.

In 2009, two RCTs comparing VP to a sham procedure for osteoporotic VCFs were published in the NEJM. ^{9 , 10} The results of both trials contrasted with earlier observational and meta-analysis data, and called into question the efficacy of VP. The Investigational Vertebroplasty Safety and Efficacy Trial (INVEST) screened 1,812 patients to randomize 131 to receive either VP (n = 68) or a sham procedure (n = 63). ⁹ Due to low initial recruitment, the proposed sample size was reduced from 250 to 130 patients, and the inclusion criteria were broadened. Inclusion criteria were as follows: age older than 50 years, pain intensity of ≥3/10 on numerical rating scale (NRS), and fracture age less than 1 year. Undetermined fracture ages were confirmed using MRI or radionuclide bone scan. The sham procedure involved the injection of local anesthetic onto the periosteum of the pedicle, combined with placing pressure on the patient’s back and opening a container of methacrylate monomer to replicate PMMA odor. At 1-month follow-up, there was no difference between groups in back pain NRS (p = 0.19) or disability (measured by RMDQ score; p = 0.06). No long-term follow-up was feasible due to crossover; by 3 months, 27 patients (43%) in the control group had crossed to the VP arm.

Key limitations of the INVEST were that the trial was under-powered and suffered from prominent selection bias (screened 1,812 patients to enroll 131), the crossover of patients in the INVEST trial was far greater for those patients crossing over from sham to VP (51%) as compared to the VP patients crossing over to sham (13%), the inclusion of fractures up to 12 months old, and the lack of inclusion requirements for physical examination or advanced imaging with MRI or radionuclide bone scan. The clinical and imaging diagnostic criteria for inclusion were very different from those of most RCTs, with patients having a pain score of 3 or more on the VAS being eligible for inclusion. There was no description of a clinical examination used to determine if the pain came from the VCF itself or from another issue. There was also criticism that the INVEST trial was not a true sham, with 63% of the sham patients correctly guessing their treatment, and with the injection performed using a paraspinal injection of local anesthetic that has been used to successfully palliate patients’ pain from VCFs for up to 8 weeks. ¹¹ Despite all of these limiting factors, if the same response rate for the 131 patients had been carried out to the originally intended 250 patients, VP would have been found to be significantly better than sham treatment at a p-value of less than 0.01. Additionally, if only one patient had reported a different response (i.e., a favorable response in the VP group or an unfavorable response in the sham group), VP would have been found to be significantly better than sham with a p-value of less than 0.04.

A second multicenter sham-controlled RCT on VP for osteoporotic VCFs was published in the NEJM in 2009, by Buchbinder et al. ¹⁰ This trial included patients with back pain of less than 12 months’ duration and fracture confirmed on MRI with bone marrow edema or fracture line. A total of 78 patients were enrolled from four recruiting centers, and randomized to VP (n = 38) or sham procedure (n = 40) groups. The sham procedure did not involve the injection of anesthetic; a needle was inserted onto the lamina, with the sharp stylet replaced by a blunt stylet. To further simulate VP, the vertebral body was lightly tapped, and PMMA was mixed in the room but not injected. No significant difference between groups in pain scores was observed at 1 week, 3 months, or 6 months. There was also no difference in disability and quality-of-life (QOL) scores.

As with INVEST, Buchbinder and colleagues did not require a physical examination component and there was no description of a clinical examination used to ascertain VCF-related pain. The Buchbinder trial assessed “overall pain” rather than spine-related pain, undermining the validity of the measurement in this population replete with potentially comorbid painful conditions. Both subacute and chronic fractures (up to 12 months old) were included, with only 32% of patients having fractures less than 6 weeks old. Similar to the INVEST, the Buchbinder trial experienced difficulties with enrolment, taking 4.5 years to enroll only 78 patients, and making this trial subject to selection bias. Additionally, 68% of patients in the study were recruited at one of the four centers, with two of the remaining centers recruiting only five patients. This may have caused outcomes to be weighted to the treatment effect at a single center. This single center tended to inject only small amounts of PMMA with the mean volume of cement being 2.8 mL. A later review by Boszczyk et al concluded that the data strongly indicate that the treatment arm included patients who were not treated in a reasonably effective manner. ¹²

In 2012, a meta-analysis was published that included prospective randomized and nonrandomized trials comparing VP to NSM or sham therapy for osteoporotic VCFs. ¹³ Nine trials were analyzed, including INVEST, Buchbinder et al, and VERTOS II, with a total of 886 patients. No difference in pain relief was found between VP and sham procedure groups due to the reliance on the previously discussed VP versus sham trials as the only two trials of this type. When compared to the NSM used to treat patients with painful VCFs, however, VP was found to be superior to NSM at all time points studied, in both pain relief and QOL measures.

Several authors expressed dissatisfaction with the findings of the 2009 trials, raising concerns about the inclusion and exclusion criteria, the use of small volumes of PMMA cement, the selection bias, the precarious statistical calculations, the high rate of crossover, the low initial pain scores, the inclusion of worker compensation patients, the use of a sham that is a known active treatment, the absence of an appropriate physical examination component, and the lack of long-term follow-up. ^{14 – 16} In response, in 2013 Comstock et al published a study that followed up the INVEST cohort over 12 months to determine long-term outcomes. ¹⁷ At 1 year, there was a modest pain reduction in the VP arm, although no differences in disability measures were found. The potential for the INVEST sham group to have acted as an “active control,” thus confounding results, was also raised. ^{18 , 19} The same year, a meta-analysis by Anderson et al was published that analyzed both NEJM sham studies and downgraded them to level II data based on flawed inclusion criteria (in both studies) and a subsequent high crossover rate (in the Comstock et al study). This downgrade was based on the Cochrane Risk of Bias table and Levels of Evidence for Primary Research as adopted by the North America Spine Society. ²⁰

There remained no large multicenter RCT comparing VP with medical management, until the 2010 publication of VERTOS II by Klazen and colleagues. ²¹ Addressing some previous concerns from the 2009 trials, it included fractures of less than 6 weeks’ duration, with pain severity of ≥5/10, focal tenderness on examination, and bone marrow edema on MRI. The 202 enrolled patients were randomized equally into VP and conservative management groups. At 1 month, VP resulted in significantly reduced back pain. The mean reduction in VAS score was 2.6 greater (95% CI: 1.74–3.37; p < 0.0001) in the VP arm than in the conservative arm, and this effect was durable at 1 year. VP also resulted in improved QOL (as measured by several standardized questionnaires) and significant (VAS reduction >3 points) pain relief was achieved earlier in VP (30 vs. 116 days; p < 0.0001).

Farrokhi et al followed with a trial comparing VP and medical management for osteoporotic VCFs in 2011. ²² Inclusion criteria included severe pain despite NSM for 4 weeks, fractures aged 4 weeks to 1 year, focal tenderness on examination, and bone marrow edema or fracture cleft on MRI. Eighty-two patients were randomized to receive VP (n = 40) or NSM. At 1 week, there was reduction in VAS (difference of −3.1; p < 0.001) and improvement in the QOL measures in the VP arm. Pain relief was durable to 6 months, while QOL outcomes sustained to 36 months. All VP patients were able to ambulate at 24 hours postprocedure, compared with 2% in the conservative group. VP also resulted in increased vertebral body height (mean 8 mm) and reduction in kyphosis (mean 8 degrees).

A further RCT comparing VP with medical management for osteoporotic VCFs followed in 2012, by Blasco and colleagues. ²³ Inclusion criteria were moderate pain (VAS ≥ 4/10), fractures aged ≤12 months, and edema on MRI or increased uptake on radionuclide bone scan. A total of 125 patients were enrolled and randomized to VP (n = 64) or NSM (n = 61) arms. VP resulted in greater VAS and lower requirement for rescue analgesia (5% of patients requiring rescue analgesia compared with 25% of the medical arm).

In 2016, the VAPOUR (Vertebroplasty for Acute Painful Osteoporotic Fractures) trial aimed to compare VP with a sham procedure while addressing some of the limitations of earlier trials. ^{24 , 25} An appropriately higher pain threshold was utilized (>7/10 compared with >3 in INVEST and no pain threshold in Buchbinder et al), all fractures were less than 6 weeks old, and all fractures were imaged with MRI or single photon emission computed tomography (SPECT). In total, 120 patients were randomized to receive VP (n = 61) or a sham procedure involving subcutaneous injection of local anesthetic (n = 59). At 2 weeks, VP resulted in significant pain reduction, with NRS scores decreasing to less than 4/10 in 44% of VP patients. This effect was sustained at 1 and 6 months. VP also led to improved QOL questionnaire scores, reduced functional disability, reduced analgesic requirements, and increased height of the vertebral body.

In a 2016 prospective trial, Yang et al randomized 135 patients aged ≥70 years to receive VP or conservative therapy. ²⁶ Early VP resulted in faster and greater pain relief and improved QOL, at 1 week, 1, 3, and 6 months, and 1 year (p < 0.0001). Surveys conducted at follow-up revealed that patients in the VP arm had greater overall satisfaction with their given treatment.

There are fewer high-quality studies available for VP in the treatment of neoplasm-related VCFs. A 2011 systematic review of 30 studies, with a total of 987 patients, found reductions in back pain ranging from 20.3 to 78.9% at 1 month following VP. At 6 months, pain reduction ranged between 47 and 87%. ²⁷ A 2016 systematic review performed by Health Quality Ontario included 78 studies, with a total of 2,545 VP patients with fractures due to spinal metastases, multiple myeloma, or hemangioma. ²⁸ They reported an overall rapid (within 48 hours) reduction in mean pain intensity scores following VP, along with parallel reduction in disability measures and opioid use.