Introduction

Metastatic spread to the vertebral column is the most common malignant disease of the skeletal system [1]. Symptoms may be the consequence of a pathologic fracture secondary to vertebral destruction, with development of spinal instability and compression of adjacent neurological structures. In patients with spinal metastases, aggressive hemangiomas or multiple myeloma a pathologic vertebral compression fracture is a frequent cause of debilitating back pain, resulting in deteriorated quality of life, physical function and psychosocial performance [2]. Even survival time for patients may be significantly reduced following fracture or spinal cord compression [3]. Chemotherapy, hormonal therapy and radiation therapy have proven to be effective in halting the tumorous process and reversing the neurological compromise; however, severe side effects, ineffectiveness and delayed onset of the effect are major drawbacks of these aforementioned treatment options [4–6]. Furthermore, these modalities can not provide immediate stabilization to an unstable vertebral segment. Surgery enables to restore spinal canal support and therefore facilitate nursing, improve neurological function and support pain control [7], but major surgery performed on severely ill patients carries a high risk for complications and therefore is usually not recommended in patients whose expected survival is limited [8].

Percutaneous vertebroplasty (PV) is a minimally invasive, radiologically guided procedure in which bone cement [polymethylmethacrylate (PMMA)] is injected into structurally weakened vertebrae. PV has proven to be an effective treatment option in osteoporotic vertebral compression fractures [9–11]. Since its introduction during the 1980s by French researchers, PV has been progressively developed and adopted to treat spinal tumoral osteolysis, making it possible to provide biomechanical stability and pain relief [12–17].

Patient selection

Treatment of tumoral osteolysis to the spine is complex and challenging and requires systemic and local therapies. Because PV is only aimed at treating the pain and consolidating the weight-bearing bone, other specific tumor treatment is required for tumor management. Therefore, the decision to perform PV should be made by an interdisciplinary team consisting of interventional radiologists, radiation oncologists, spine surgeons and oncologists to ensure appropriate adjuvant therapy and follow-up.

A detailed clinical history and examination, with specific emphasis on the neurological signs and symptoms, should be performed to confirm the underlying vertebral fracture as the cause of debilitating back pain and rule out other causes like,for example, degenerative spondylosis or radiculopathy. This should be correlated with the imaging studies, including magnetic resonance (MR) imaging, computed tomography (CT) and technetium 99 m pertechnetate bone scintigraphy [18, 19]. Whenever doubt persists, the lesion should be sampled for biopsy during the PV procedure. In metastatic disease, fractures might be present at multiple levels of the spine, not all of which require treatment with PV.

Indications [20, 21]

  • excruciating pain with adverse effects (constipation, urinary retention and/or confusion) to opioid treatment or opioid tolerance developed in patients with formerly controlled pain [22]

  • extensive osteolysis due to malignant infiltration with or without fracture of the affected vertebral body [23]

  • intended posterior surgical procedure for stabilization, in which reinforcement of the affected vertebral body or pedicle is requested

Contraindications [20, 21]

Absolute

  • patient well improving on appropriate analgesic medication

  • asymptomatic vertebral fracture and low risk for biomechanical instability and collapse

  • apparent systemic infection, osteomyelitis, discitis

  • local infection at the puncture site

  • uncorrectable coagulopathy

  • known allergy to any of the components used for PV

  • diffuse non-focal back pain

Relative

  • asymptomatic displacement of a fracture fragment producing spinal canal narrowing

  • radiculopathy

  • extension of the tumor into the spinal canal with or without cord compression

  • collapse of the posterior vertebral body wall (increased risk of PMMA leakage)

  • vertebra plana resulting in difficulties in needle placement

  • cardiorespiratory compromise such that safe sedation or anaesthesia cannot be accomplished

  • lack of monitoring facilities and surgical back-up

Technique

During a preprocedural consultation with the patient the procedure, intended benefits and possible complications should be discussed and balanced. A patient informed consent is mandatory.

Periprocedural application of antibiotics is mandatory in immunocompromised patients. Antibiotics are usually administered via a venous access 30 min ahead of the beginning of the procedure. However, there is no clear consensus in the literature [24, 25]. During PV vital signs of the patient are monitored and strict asepsis is maintained.

Anatomic landmarks and structures differ according to the vertebral level to be treated.

In the cervical spine, a right anterolateral approach is used. The carotid-jugular complex has to be displaced gently downward and laterally and separated from the trachea and oesophagus to expose an entry site for the needle.

Depending on the site of the neoplastic lesion in the thoracic and lumbar spine, three different approaches are feasible:

  • the classic transpedicular route, which can be performed either by a unipedicular or bipedicular approach;

  • the posterolateral approach, especially in the lumbar spine when the lesion involves the pedicles;

  • the intercostovertebral approach, especially used in the thoracic spine, which is more favorable when the pedicles are too small or destroyed by the tumor. It has to be taken into account, that this approach bears a higher risk of pneumothorax and paraspinal bleeding.

The PV can be performed either using biplane fluoroscopy guidance, dual guidance including CT and fluoroscopy or CT-fluoroscopy alone.

Biplane fluoroscopy guidance

The appropriate radiographic projection for the transpedicular approach is a straight anteroposterior view with 5–10° angulation, in which the pedicle appears oval. Using AP and lateral views, the needle is forced through the upper and lateral aspect of the pedicle because of the reduced risk of harming the spinal cord and nerve roots. Using lateral fluoroscopy, the tip of the needle is positioned in the anterior third of the vertebral body or in the osteolytic lesion, with the shaft of the needle aligned parallel to the endplates of the affected vertebral body. Using this technique, the final endpoint of the needle-tip is within the ipsilateral half of the vertebral body, therefore requiring a bipedicular approach for optimal filling of the vertebrae. The use of a beveled needle supports precise placement, since rotating the beveled tip allows for distinct steering and therefore the tip can be positioned anteromedial in the vertebral body. Usually, this technique allows sufficient filling of the vertebral body making a bipedicular approach unnecessary.

Dual guidance

The advantage in combining CT and fluoroscopy is the precise needle placement, which is particularly important in the upper thoracic spine, tumor cases and other difficult cases. This dual guidance technique reduces complications, and increases the comfort and the confidence of the interventional radiologist. It allows for visualization in three dimensions, with exact differentiation of anatomic structures at risk. Fluoroscopy is provided by placing a mobile C-arm in front of the CT-gantry. When the position of the needle-tip is considered satisfactory, the imaging mode is switched to C-arm fluoroscopy for real-time visualization of cement application in an AP and lateral view.

CT-fluoroscopy guidance

When the CT scanner is equipped with online CT-fluoroscopy, it can be used for monitoring and guiding the whole procedure. A prerequisite is that the table can be moved by the interventional radiologist using a joystick from inside the room, maintaining strict asepsis. The entry point at the skin and advancing the needle through the vertebral body can be visualized at all times. Therefore this technique allows for safe needle placement without harming critical structures (Fig. 1). The cement application can be monitored online, and by moving the table back and forth the whole affected vertebral body can be covered easily. This allows for reliable detection of any cement leakage, especially into the spinal canal. Particularly in tumor cases, CT-fluoroscopy guidance facilitates the correct positioning of the needle in the osteolysis and permits to reliably assess potential bulging of the posterior vertebral wall when fractured.

Fig. 1
figure 1

A 64-year-old female patient with two painful osteolytic metastases affecting the 12th thoracic vertebra (displayed) and the 3rd lumbar vertebra. Needle placement using online CT-fluoroscopy guidance is demonstrated. The tip of the needle in positioned within the centre of the osteolytic metastasis (arrow). Using CT-fluoroscopy the cement can be visualized while flowing through the hollow space of the needle as well as its dissemination within the vertebral body

Full size image

Cement application

After the needle (10–15 gauge) or in case of multiple affected vertebral bodies preferably up to three monolateral needles are carefully positioned, the cement is prepared. There are two options. To avoid contamination and the inclusion of air bubbles, the use of a closed mixing device is advocated. This system allows for homogenous mixing of the components and therefore increases its strength. If not available, the cement can also be mixed by hand in a sterile bowl. After 30–60 s of continuous mixing, the initially very fluid cement starts to get thick and pasty. Especially if only one vertebral body is supposed to get treated, it is advisable to wait additional 60–90 s before injecting the cement into the vertebral body. If the cement is administered in this pasty polymerisation phase, the risk of extravertebral leakage as well as venous intravasation is reduced.

For the administration of cement into the vertebral body a dedicated screw-like injection set, provided by several manufactures, is usually employed. The advantage to use a dedicated set instead of a 2-cc Luer lock syringe is that the cement can be administered with continuous flow and minimal effort for the treating interventional radiologist. Furthermore, if a leak is noticed the pressure can be stopped and reversed immediately. The injection of cement is observed under continuous lateral fluoroscopic or online CT-fluoroscopy control to allow for instant detection of epidural leakage. If a leakage is detected, it is very important to stop the procedure, reverse the pressure and to wait for up to 60 s. This time allows the cement to harden and probably seal the leak. If the leak then persists the needle either has to be repositioned or the bevel direction should be modified. In cases in which these measures are not effective, the procedure should be abandoned. To complete the filling of the affected vertebral body the contralateral approach could be used. In order the avoid cement leakage through the puncture canal, the initial needle should remain in place.

The procedure is completed when the osteolytic lesion within the vertebral body is completely or partially filled (Fig. 2). It has to be taken into account that the cavity of the needle contains an additional 1–2 cc of cement. While the stylet of the needle is pushed forward, this cement is also injected into the vertebral body. This should be done under fluoroscopic control to avoid leakage. Especially in tumor vertebroplasty, the use of online CT fluoroscopy is favorable, since not only cement leakage can be detected but also bulging of soft tissue tumor components into the spinal canal can be depicted. The re-insertion of the stylet reduces the risk of a cement antenna in the paravertebral soft tissue. If the stylet is not completely repositioned, the interventional radiologist should wait until the cement is completely hardened. Then the cement antenna within the needle can be broken off and carefully removed together with the needle.

Fig. 2
figure 2

Same patient as in Fig. 1. Two osteolytic metastases involving the 12th thoracic vertebra (a) and the 3rd lumbar vertebra (b). The posterior wall of the 3rd lumbar vertebra is destroyed. The post-procedural CT scans [axial (c, d), coronal (e) and sagittal (f)] show partial filling of the lesion in the 12th thoracic vertebra and complete filling of the metastasis in the 3rd lumbar vertebral body. Initial leakage of cement in an epidural vein (d, arrow) with no significant narrowing of the spinal canal. In the coronal view a small cement leakage (e, arrow) in a paravertebral vein can be appreciated

Full size image

The time between mixing and hardening of the cement is approximately 8–10 min (room temperature, 20°C). Some manufactures provide cement with longer setting times. Therefore, in case of multiple affected vertebral bodies, preferably up to three vertebral levels can be treated in a single session. Concerning the cardiotoxicity of PMMA (especially when spilled into the venous circulation), no consistent data are published in the literature. The deteriorating baseline mean arterial blood pressure during PV is, according to the literature, most likely associated with the increase of pressure within the vertebral bodies rather than with the use of PMMA [26, 27].

The pain relief in vertebral destructions due to malignant tumors is not directly proportional to the percentage of lesion filling [28]. Therefore the volume of cement as well as the number of injections performed to obtain complete lesion filling should be limited, especially when extensive cortical destruction is present. Consequently it should decrease the risk of leaks of cement, particularly epidural, foraminal and venous leaks. Discal and paravertebral leaks of PMMA seem to have no clinical importance for the patients. Depending on the size of the osteolytic lesion smaller volumes (1.5–3 cc) are usually associated with good clinical results.

Results

The precise mechanism of action of PV is uncertain. While in healthy vertebrae, burst fracture occurs only under high impact loading, patients with vertebral metastases may experience burst fractures under normal physiologic loading conditions [29, 30]. Factors such as bone density and tumor volume as well as tumor location and shape have been shown to be important in assessing burst fracture risk [30, 31]. The strengthening effect of the PMMA application is thought to provide stability and prevent fracture or further collapse of the affected vertebrae. Concerning pain relief it is discussed that the vascular, chemical, and thermal forces associated with the inflammatory reaction to the heat of polymerization of PMMA probably have a more pronounced effect than the mechanical forces and may account for the clinical improvements of the patients [32].

Analgesic effect

Clear improvement is usually defined as complete pain relief with no necessity for analgesic medication or enough of a decrease in pain that the dose of analgesic drugs can be reduced by at least 50%. Also, the replacement of narcotic drugs with non-narcotic drugs is considered to reflect clear clinical improvement.

In vertebrae with metastases and debilitating pain, pain palliation can be achieved in 50–97% of the patients (Table 1, [12, 22, 28, 33–36]). These data are similar to those reported for surgical treatment [37].

Table 1 Efficacy of PV for pain reduction in patients with osteolytic metastatic tumors
Full size table

Cotten et al. [28] described partial or complete pain relief in 36 of 37 patients within 6–72 h of PMMA injection, independent from the percentage of cement fill in the vertebral body. Depending on the size of the osteolytic lesion, smaller volumes (1.5–3.5 cc) are usually sufficient to provide good results [38]. In a retrospective analysis on 37 patients suffering from various primary tumors with metastatic spread to the spine, Weill et al. [22] described clear to moderate improvement in 35 patients. At first month follow-up, there was no recurrence in pain in those patients who initially benefit from the procedure. At 3 months follow-up, nine of 14 patients presented with stable pain palliation. In those five patients with a variable degree of recurrence of pain, additional vertebrae involved by metastases adjacent to the formerly treated vertebral body were detected by MR imaging. This suggests that recurrence of pain is most likely due to the new lesions instead of failure of the initial procedure. This has been well described by Hodler et al. [39].

Alvarez et al. [12] demonstrated an immediate analgesic efficacy in 90% of 21 treated patients, with a persistent pain relief in 67% of the patients over prolonged periods of time. Another important finding was that 69% of non-ambulatory patients became ambulatory after the PV procedure due to the substantial relief of pain.

Biomechanical stabilization

To date, the mechanical properties of the metastatic spine and the mechanisms of collapse are not fully elucidated. Moreover, the correlation between vertebral body collapse and the location and extent of the metastatic tumor is not fully understood. Taneichi et al. [23] evaluated 100 thoracic and lumbar vertebrae (53 patients) with osteolytic lesions, determined risk factors for vertebral collapse and estimated the probability of collapse under various states of metastatic vertebral involvement. The most important risk factor leading to vertebral collapse in the thoracic region was involvement of the costovertebral joint. Tumor size within the vertebral body was the second most important risk factor. Interestingly, destruction of the costovertebral joint more strongly induced vertebral body collapse than the size of metastatic tumor within the thoracic vertebral body. Under the condition that the metastatic lesion was confined to the vertebral body, impending collapse existed when the vertebral body involvement was 50–60% in the thoracic spine and 35–40% in the thoracolumbar and lumbar spine. This indicates that the thoracic vertebra has greater tolerance to collapse than the thoracolumbar and lumbar vertebrae as long as the costovertebral joint remains intact. The impact of metastatic pedicle involvement on vertebral collapse was much greater in the thoracolumbar and lumbar spine than in the thoracic spine. The involved vertebral body with pedicle destruction may be more vulnerable to axial overload than the vertebrae with intact pedicles. This might be attributed to the fact that posterior load bearing structures (e.g., facet joints) can no longer support the involved vertebral body from the posterior aspect once disconnection between the vertebral body and the posterior elements occurs as a result of pedicle destruction.

Tschirhart et al. [40] published a study aimed at determining the effect of cement location and volume of cement injected during percutaneous vertebroplasty on improving vertebral stability in a metastatically-compromised spinal motion segment using a parametric poroelastic finite element model. Sixteen scenarios (different tumor locations within the vertebral body, different cement locations within the vertebrae) were investigated pre- and post-vertebroplasty using a serrated spherical representation of tumor tissue and various geometric representations of PMMA.

Vertebral bulge and vertebral axial displacement were used as a measure of stability. Vertebral bulge was defined as the maximum radial bulge of the vertebral body under load as a burst fracture predictor irrespective of endplate failure, and vertebral axial displacement represented the maximum axial displacement of the vertebral body under load as a predictor for burst fracture following endplate failure.

The results indicated that vertebral bulge and axial displacement decreased with the addition of cement in all scenarios and therefore PV decreased the risk of burst fracture initiation. However, the magnitude of the increase in stability depended on the location and geometry of the cement injected. Burst fracture risk appeared to be minimized when cement was injected near the posterior wall of the vertebral body.

The cement volume required to restore vertebral stability to that of a baseline intact model was calculated. To significantly decrease vertebral bulge, a 16–17% fill of the vertebral body was required for the posterolateral and anterolateral PMMA deposition, whereas stabilization of the axial displacement required up to a 32% fill of the metastatically involved vertebrae. This is especially noteworthy since other authors have indicated that cement leakage becomes imminent at approximately 20% cement volume of the involved vertebral body [16]. Therefore the authors concluded that it may not be necessary to exceed a volume of 20% of PMMA, especially since the vertebral bulge rather than the axial displacement has shown clinically to best predict burst fracture in the metastatic spine [29]. Further results of this study indicated that augmentation of metastatic vertebrae was beneficial for stabilization even when the PMMA was asymmetrically located and focused on areas of bone destruction.

These results are supported by Ahn et al. [13]. They experimentally measured the load-induced spinal canal narrowing to determine the biomechanical stability of metastatically involved vertebrae following percutaneous vertebroplasty. Specimens with cement located in a posterior location showed reduced load-induced spinal canal narrowing of 39% following PV. Morever, Ahn and co-workers found that a reduction in load-induced spinal canal narrowing did not occur with PMMA located in the anterior portion of the vertebral body.

Combination with tumor ablation

The location of cement injection relative to the tumor tissue is critical in attaining maximum vertebral stability following PV. Vertebrae with tumors located in the posterior region are at a higher risk for the occurrence of burst fractures due to their proximity to the posterior vertebral body wall [31]. Tschirhart et al. [40] described that stabilization effects of PV on vertebrae with posterior tumors are not as positive as for tumors in other locations.

Using current methods for PMMA injection, there is a high risk of cement leakage into the spinal canal when the cement is injected in the posterior region of the vertebral body. However, the above mentioned biomechanical studies indicate that cement injection in this region tends to provide improved biomechanical stability to the vertebral body [13]. Cement leakage, particularly into the spinal canal, is still considered as the primary concern associated with PV. However, as the effects of cement location and volume become better understood, less cement injection may be required to provide adequate stability and at the same time reduces the risk of leakage. With online CT-fluoroscopy guidance and its excellent overview and spatial resolution, it is feasible to perform ablation on tumors located adjacent to the posterior wall of the vertebra and it becomes feasible to inject the cement reliably into the region of the posterior vertebral body wall with minimal likelihood of leakage (Fig. 3). Therefore, tumor ablation could decrease the tumor volume prior to PV, providing the space necessary for cement injection and allowing for a quasi-shell geometry of the cement to encapsulate the tumor with minimal risk for leakage. This is of special interest for stabilizing vertebrae with metastases in the posterior portion of the vertebral body. Recently published studies have shown improvement in vertebral stability and reduced PMMA leakage when tumor ablation was performed prior to PV [41–43].

Fig. 3
figure 3

A 71-year-old male patient suffering from a painful and destabilizing osteolytic renal cell cancer metastasis of the 5th lumbar vertebra (a, d). In order to avoid radiculopathy as a result of cement contact with the adjacent nerve root and heating of the nerve tissue during tumor ablation (c, long arrow RFA electrode within the metastasis) and the polymerization phase of the PMMA, a spinal needle was positioned in close proximity of the neuroforamina and saline was injected slowly to cool the nerve root (f, short arrow). Post-procedural axial (b) and sagittal (e) scans show partial filling of the osteolytic metastasis. The patient developed no neurological symptoms but significant pain relieve

Full size image

Complications

Complications are classified into minor and major adverse events. Minor adverse reactions are defined as unexpected or undesirable clinical occurrences that require no immediate or delayed surgical intervention [44]. A major adverse event is defined as the occurrence of an unexpected or undesirable clinical event, which requires surgical intervention or results in death or significant disability. Published data have placed the complication rates in osteoporotic fractures treated with PV at <1%, while the rates in metastatic vertebrae with fractures may be as high as 10% [14, 45].

Most complications like infection, fracture of ribs, posterior vertebral elements or pedicles, allergic reaction as well as bleeding from the puncture site have a reported incidence below 1% and are considered as minor complications [46, 47].

A recently published paper described six (5.1%) local and two (1.7%) systemic complications in 117 patients treated for vertebral metastases with PV [48]. Local complications consisted of haematoma at the puncture site (resolved uneventful with no further treatment required) and radicular pain due to ipsilateral foraminal venous cement leakage. The symptoms resolved under appropriate medication with non-steroidal anti-inflammatory drugs and a single bolus of prednisolon. None of the patients required surgical debulking of the cement. Interestingly, two patients developed pulmonary embolism (PE) detected on their post-procedural chest radiographs and CT scans. One patient did not develop any pulmonary or haemodynamic signs or symptoms of PE, while the other patient (adenocarcinoma of the lung) developed ventilatory and haemodynamic symptoms of PE and died despite treatment with anticoagulants one week after PV. In total, the per-procedure and per-patient morbidity rates were 5.0% (eight of 159 procedures) and 6.8% (eight of 117 patients), while the single death recorded meant that the procedural and patient mortality rates were 0.6% (one of 159 procedures) and 0.9% (one of 117 patients), respectively.

Although leakage of cement is well tolerated in most cases [49], it is the main source of pulmonary and neurological complications. A transient neurological deficit is observed in 5% of patients with malignant aetiology. Symptoms respond well to nerve-root blocks, or oral medication; rarely do they require surgical decompression [22, 28]. While leakages of cement into the spinal canal only infrequently lead to neurological complications or even paraplegia, intraforaminal leakage is more harmful. Cotten et al. [28] found that leakage into the spinal canal was well tolerated in all their 15 patients while two of eight cases of foraminal leakage were associated with radiculopathy.

It has been reported that cement leakage is more common when PV is used for metastatic osteolytic tumours or myelomas of the spine than in osteoporotic fractures. However, Vasconcelos et al. [50] observed no major differences, although they noted venous leaks slightly more frequently in patients with metastatic lesions. When PV was performed in osteoporotic vertebral compression fractures, leakage into the disc space was more commonly observed. Mousavi et al. [51] reviewed post-procedural CT scans in patients with osteoporotic vertebral compression fractures and metastatic lesions of the spine and concluded that in osteoporotic vertebrae leakage occurred mainly into the disc, whereas in metastatic lesions it was found in various different locations.

McGraw et al. [52] found that intraosseous venography predicted the flow of PMMA during vertebroplasty in 83% of cases; however, this has not been confirmed by other authors, and the use of pre-procedural venography has largely been abandoned, except in hypervascular tumours [53, 54]. The viscosity of the cement has been shown to represent the most important factor for cement leakage. Heini et al. [55] described that the risk of cement intravasation is diminished if the flow of cement is directed in a medial direction within the vertebral body. Therefore they suggested using a side-opening cannula.

Conclusion

Percutaneous vertebroplasty in vertebrae with destructions due to metastatic spread is a minimally invasive, radiologically guided procedure that provides immediate and long-term pain relief. In addition, PV contributes to spinal stabilization and can be successfully combined with chemotherapy, radiation therapy, tumor ablation and posterior laminectomy. Therefore PV should be included in the therapeutic arsenal for patients with metastatic spread to the spine. Of course, further comparisons of this technique with other conventional techniques, such as open surgical fusion, vertebrectomy and radiation therapy are still needed to define precisely the role of PV in the treatment of metastatic disease. However, especially in patients who are not eligible for the aforementioned classical treatment options, PV is a powerful and promising adjunct in the armamentarium of therapeutic procedures in the palliative treatment of spinal metastases.