November 2000, VOLUME 175

Recommend & Share

November 2000, Volume 175, Number 5

Technical innovation

Radiofrequency Ablation of Spinal Tumors
Temperature Distribution in the Spinal Canal

+ Affiliations:
1Department of Diagnostic Imaging, Brown University School of Medicine, Rhode Island Hospital, 593 Eddy St., Providence, RI 02903.

2Department of Radiology, Harvard Medical School, Beth Israel/Deaconess Hospital, 1 Deaconess Way, Boston, MA 02114.

Citation: American Journal of Roentgenology. 2000;175: 1263-1266. 10.2214/ajr.175.5.1751263

Next section

Radiofrequency ablation has been used in the musculoskeletal system for the treatment of lower back pain related to facet osteoarthritis or prior surgery (e.g., failed back syndrome) [1] and for the percutaneous treatment of osteoid osteomas [2]. The conventional monopolar radiofrequency technology used for these applications causes thermocoagulation necrosis no larger than 1.6 cm in diameter. Newer technology, however, most notably internally cooled radiofrequency electrodes, allows the creation of larger regions of thermocoagulation approaching 5 cm with a single electrode [3] and 7 cm with a cluster of three electrodes spaced 0.5 cm apart [4]. Because some metastatic tumors and osteoid osteomas will involve the vertebral body, and because radiofrequency heating at around 45°C has been shown to be cytotoxic to the spinal cord [5, 6] and peripheral nerves [7], the temperature effects of radiofrequency heating on the adjacent thecal sac contents must be considered before radiofrequency ablation is applied to spinal tumors. Therefore, to determine whether radiofrequency ablation can be used safely in the vertebral body, we applied radiofrequency to pig vertebral bodies while simultaneously measuring the temperature changes in the adjacent spinal canal. This was followed by an ex vivo experimental study of radiofrequency-generated heat transmission in bone and other issue. Additionally, as an initial clinical step, we treated one patient with vertebral body metastasis and one patient with an osteoid osteoma using radiofrequency ablation.

Materials and Methods
Previous sectionNext section
In Vivo Studies

Five adult pigs (weight range, 50-75 kg) were used for this study. Approval of the institutional subcommittee on animal research care was obtained before the study. Animals were intubated and anesthetized with halothane anesthesia and placed in the left lateral decubitus position. Cardiac and respiratory parameters were monitored throughout the procedure.

Using CT-guided (Xpress; Toshiba Medical Systems, Tokyo, Japan) helical acquisition (5-mm slice thickness, 7-mm collimation, 1:1 pitch, 120 or 140 kV, and 280-300 mA), a 14-gauge Ackermann bone biopsy sheath (Cook, Chicago, IL) was placed into a lumbar vertebral body (n = 5) or paraspinal muscle (n = 5). For lumbar vertebral bodies (L1-L3), a posterolateral approach through the pedicle was used (Fig. 1). A hole was cut into he vertebral body with an inner 15-gauge trephine cutting needle. The bone core was removed and a 17-gauge internally cooled radiofrequency electrode with a 1-cm active tip was placed through the 14-gauge Ackermann cannula. One to two additional holes were made through the lamina at the same level, and remote temperature sensors were placed 5-15 mm from the radiofrequency electrode. The tip of at least one temperature probe was positioned in the anterior spinal canal adjacent to the posterior longitudinal ligament so that the radiofrequency electrode was between 5 and 15 mm away from the tip of the radiofrequency electrode. The thermistor probe (TCA; Radionics, Burlington, MA) was positioned in the spinal canal at 5, 10, and 15 mm from the radiofrequency electrode.

View larger version (185K)
Fig. 1. Transverse CT scan of lumbar spine of anesthetized pig shows positions of radiofrequency electrode (arrow) in vertebral body and temperature sensor (arrowhead) in spinal canal.

A single 12-min radiofrequency application was performed at the maximum current that could be applied without observing impedance rises using a radiofrequency generator (RFG-3C; Radionics). During the ablation, 0°C saline solution perfused the electrode at 80 mL/min. Temperature measurements were recorded every 5 min during and after the treatment until they returned to baseline. Five measurements were obtained by applying radiofrequency using similar parameters in paraspinal muscle (maximum current without impedance rise) with a remote thermistorprobeat5-15mmfromtheradiofrequencyelectrode. The muscle measurements were reflective of the soft-tissue temperature distribution without intervening cortical bone or cerebrospinal fluid.

Ex Vivo Studies

Additional ex vivo studies were performed to compare radiofrequency-produced heat transmission through cortical bone, cancellous vertebral bone, liver, and an agar phantom. Fresh tissue was excised from pig cadavers. Cubes of cancellous bone (2.5 cm) and 2.5 cm2 of 5-mm-thick cortical bone were embedded in a 5% agar, 1% sodium chloride phantom. Radiofrequency was applied for 6 min at tip temperatures of 94 ± 2°C using noncooled monopolar electrodes of 2-cm tip exposure. Temperatures were monitored 10 mm from the electrode tip. The radiofrequency electrode was placed directly into the center of the cancellous bone, liver, or agar. For cortical bone the radiofrequency electrode was placed within 3 mm of the bone, with two temperature sensors placed 10 mm from the electrode on both sides of the bone (Fig. 2).

View larger version (21K)
Fig. 2. Diagram shows ex vivo experimental configuration used to measure heat transmission differences of radiofrequency in cancellous and cortical bone at similar distances, using remote temperature sensors. RF = radiofrequency, Temp. = temperature.

Patient Studies

A 54-year-old woman with a history of metastatic hemangiopericytoma presented with mid to lower back pain. CT imaging revealed a focal osteolytic lesion in the L2 vertebral body. The imaging characteristics and histology from biopsy were consistent with metastasis (Fig. 3A,3B). The patient refused external beam radiation, and the patient's oncologist did not believe chemotherapy would be of significant benefit. Because of the location and intact posterior vertebral body cortex, radiofrequency ablation was planned to treat the lesion and to prevent future posterior extension. After informed written consent was obtained and the patient was given local anesthesia with 1% lidocaine and IV conscious sedation with midazolam 2 mg and fentanyl 150 μg, a 14-gauge Ackermann bone biopsy needle was placed into the tumor with CT guidance. A 3-cm active Cool-tip radiofrequency electrode (Radionics) was placed through the Ackermann cannula into the lesion (Fig. 3A,3B). A 12-min radiofrequency lesion was created with a maximum output of 1400 mA using a Cosman Coagulator radiofrequency generator (Radionics). The radiofrequency electrode was perfused with chilled saline at a flow rate of 80 mL/min to keep the electrode tip between 10° and 15°C.

View larger version (165K)
Fig. 3A. 54-year-old woman with metastatic hemangiopericytoma. Transverse CT scan of lumbar spine shows lytic metastasis (arrow) in anterior L2 vertebral body. Note intact posterior vertebral body cortex.

View larger version (199K)
Fig. 3B. 54-year-old woman with metastatic hemangiopericytoma. Axial CT scan shows 3-cm active-tip radiofrequency electrode in L2 vertebral body. Electrode is positioned to optimize thermocoagulation of metastasis.

A 14-year-old boy presented with a 9-month history of back pain and scoliosis. Physical examination was remarkable for a thoracic levoscoliosis, and an osteoid osteoma was suspected. CT of the thoracic spine revealed an area of rounded low attenuation that measured approximately 1 cm with a 5-mm center of high attenuation. Surrounding sclerosis extended into the pedicle laterally and into the vertebral body and right lamina (Fig. 4A,4B,4C). CT findings were consistent with a diagnosis of an osteoid osteoma. Percutaneous thermocoagulation using radiofrequency ablation was discussed and agreed on after informed written consent by the patient's parent. The patient was placed prone and given general endotracheal anesthesia. The area over the right T11 pedicle was prepared and anesthetized with 1% buffered lidocaine. A 14-gauge Ackermann needle was used to obtain a path to the nidus, and a core of bone was removed for pathologic analysis. A 5-mm active-tip monopolar radiofrequency electrode was then inserted in the nidus (Fig. 4A,4B,4C). The osteoid osteoma was treated at 90°C for 6 min with a radiofrequency generator (Radionics). After being observed in the recovery room for 2 hr without any neurologic deficits, the patient was discharged from the hospital.

View larger version (155K)
Fig. 4A. 14-year-old boy with spinal osteoid osteoma. Transverse CT scan of T11 vertebral body shows typically calcified nidus (arrow) and surrounding sclerosis in right T11 pedicle, which is consistent with osteoid osteoma.

View larger version (164K)
Fig. 4B. 14-year-old boy with spinal osteoid osteoma. Transverse CT scan with patient prone shows radiofrequency electrode within nidus (arrow). Nidus was treated for 6 min at 90°C.

View larger version (193K)
Fig. 4C. 14-year-old boy with spinal osteoid osteoma. Transverse CT scan of treated region obtained 8 months after B shows sclerosis and filling in of treated nidus.

Previous sectionNext section
In Vivo

The maximum current achievable in the vertebral body without observing impedance rises was 220 ± 35 mA. This was markedly reduced compared with the 326 ± 42 mA applied in muscle (p < 0.01). The maximum temperature observed at the epidural space was 44°C. Temperatures generated were consistently lower within the vertebral body than those in paraspinal muscle for similar treatments. For example, at 10 min the maximum temperatures observed in bone were 48°, 41°, and 39°C at a distance from the radiofrequency electrode of 5, 10 and 15 mm, respectively, compared with a maximum of 84°, 62°, and 58°C, in paraspinal muscle (p < 0.01, all comparisons).

Ex Vivo

Ex vivo experiments confirmed decreased heat transmission at a 10-mm distance from the electrode through cancellous bone (13.4 ± 4.5°C) compared with that of either liver (20.0 ± 3.4°C) or an agar phantom (18.5 ± 3.1°C) (p<0.05, both comparisons). An insulating effect was documented for cortical bone as temperatures in the bone cortex with the electrode were increased (25.7 ± 7.0°C > baseline) compared with temperatures at an equal distance but on the other side of the cortical bone (11.2 ± 2.0°C > baseline; p<0.01).


The patient with lumbar metastasis did not experience any radiating pain during the procedure and was neurologically intact on sensorimoter physical examination after the procedure. Treatment of the vertebral metastasis was successful. The patient's pain remarkably improved, and the patient remains asymptomatic 13 months after treatment. The patient has since developed additional metastatic deposits in the sacrum that have been treated with radiofrequency ablation.

In the patient with the T11 pedicle osteoid osteoma, follow-up 6 months after the procedure found the patient to be pain-free and without any lower extremity weakness or sensory loss. A mild thoracic scoliosis persisted. A follow-up CT scan 8 months after the procedure showed interval healing of the osteoid osteoma nidus (Fig. 4A,4B,4C).

Previous sectionNext section

This study shows that radiofrequency thermal ablation can potentially be safely performed in the vertebral body without cytotoxic temperature elevations in the spinal canal. Despite the use of internally cooled electrodes at maximum output, injurious elevations of temperatures in the epidural space did not occur. Ex vivo studies confirmed decreased heat transmission in cancellous bone and an insulative effect on cortical bone. This insulating quality of bone has not been shown in earlier experiments with non-internally cooled radiofrequency electrodes that were used in the liver [4]. An additional factor that may account for the differences in heat distribution observed are local heat sinks from the rich epidural venous plexus and cerebrospinal fluid pulsations. Perfusion-mediated tissue cooling has been shown to negatively influence the extent of coagulation that can be produced in in vivo liver, and reduction in blood flow by mechanical or pharmacologic means can increase the diameter of coagulation necrosis [8].

Clearly, in cases in which preserved cancellous or cortical bone is between the lesion and the spine, a margin of safety will be provided. This was shown in our ex vivo study, because cortical and cancellous bone provided diminished heat transmission compared with that of soft tissue. This was also shown in the measurements of paraspinal muscle, which again showed much greater heat transmission to surrounding tissue than to the vertebral body and spinal canal.

In patients with extensive osteolysis with no intact cortex between tumor and spinal cord or nerve roots, radiofrequency may not be an option because of the potential of thermal injury to adjacent neural tissue. Theoretically, if a cerebrospinal fluid space was present between the tumor and neural tissue, radiofrequency could be applied without unwanted neurotoxicity.

In the patient with osteoid osteoma, the lesion clearly abutted the thecal sac, yet no neurotoxicity was evident, and the treatment was a clinical success. Internally cooled electrodes that increase the effective radius of thermocoagulation may pose significant neurotoxicity if used in this area, but this has not been proven. Froese et al. [6] used radiofrequency energy to heat the spinal cord of mice and determined that the effective dose for injuring 50% for heating for 1 hr was 43.1°C and for heating at 45°C was 10.8 min. In the pig model used in our study, the temperature in the epidural space reached 44°C. However, we did not perform histologic or neurologic testing to determine what effects our treatment had in these animals.

A remote temperature sensor may offer a prudent margin of safety because the procedure could then be terminated if deleterious temperature rises are observed adjacent to nervous tissue. Osteoid osteomas need only be treated for 6 min [2]. Large tumor ablations are performed for different durations on the basis of the electrode technology used. Internally cooled electrodes provide a much larger diameter of coagulation necrosis, and the treatment durations depend on the regional tissue constituents and blood perfusion. In previous studies on the liver [3, 4], the maximum effect is reached by 12 min. However, in less perfused areas such as a bone metastasis, the treatment time is gauged by the impedance rises during the treatment. If the impedance continues to increase despite lowering the radiofrequency current, then continued therapy will not increase the diameter of coagulation necrosis. This time is unique to every tumor.

Radiofrequency ablation has been proven to be as effective as surgery in the treatment of osteoid osteomas, with similar rates of success and recurrence but with the advantages of shorter hospitalizations and fewer complications [2]. However, radiofrequency ablation has been used mostly on lesions in the bones of the lower extremities such as the femur and tibia. It has been less commonly used on vertebral lesions because of the potential risk of adjacent neural tissue heating.

The clinical case that we present shows that a spinal osteoid osteoma can be treated successfully without neurotoxicity. A larger series of spinal osteoid osteomas treated with radiofrequency ablation is necessary to determine safety. Osteoid osteomas have the benefit of being small (usually <12 mm in diameter) and well contained within a thick rim of sclerotic bone. Therefore, conventional monopolar radiofrequency technology is all that is necessary to treat the small tumor region. Metastatic tumors are almost always too large to treat with conventional monopolar technology so the internally cooled radiofrequency electrode with a larger treatment diameter is necessary. Careful use of these newer electrodes is necessary in and around the spine because of the potential deleterious effects on adjacent nervous tissue. Continued work in this area will be important as this technique gains more widespread acceptance as a treatment option for malignant and benign bone spinal lesions. We hope that this innovative new approach will provide not only pain palliation but also local tumor control, thus avoiding additional therapy such as radiation or surgery.

Address correspondence to D. E. Dupuy.

Previous sectionNext section
1. Cho J, Park YG, Chung SS. Percutaneous radiofrequency lumbar facet rhizotomy in mechanical low back pain syndrome. Stereotact Funct Neurosurg 1997; 68:212-217 [Google Scholar]
2. Rosenthal DI, Hornicek FJ, Wolfe MW, Jennings LC, Gephart MC, Mankin HJ. Percutaneous radiofrequency coagulation of osteoid osteoma compared with operative treatment. J Bone Joint Surg 1998; 80:815-821 [Google Scholar]
3. Goldberg SN, Gazelle GS, Solbiati L, Rittman WJ, Mueller PR. Radiofrequency tissue ablation: increased lesion diameter with a perfusion electrode. Acad Radiol 1996; 3:636-644 [Google Scholar]
4. Goldberg SN, Solbiati L, Hahn PF, et al. Largevolume tissue ablation with radiofrequency by using a clustered, internally cooled electrode technique: laboratory and clinical experience in liver metastases. Radiology 1998; 209:371-379 [Google Scholar]
5. Yamada T, Tateishi A, Cho S, et al. The effects of hyperthermia on the spinal cord. Spine 1992; 17:1386-1391 [Google Scholar]
6. Froese G, Das RM, Dunscombe PB. The sensitivity of the thoracolumbar spinal cord of the mouse to hyperthermia. Radiat Res 1991; 125:173-180 [Google Scholar]
7. Letcher FS, Goldring S. The effect of radiofrequency current and heat on peripheral nerve action potential in the cat. J Neurosurg 1968; 29:42-47 [Google Scholar]
8. Goldberg SN, Hahn PF, Halpern EF, Gazelle GS. Radiofrequency tissue ablation: effect of pharmacologic modulation of blood flow on coagulation diameter. Radiology 1998; 209:761-767 [Google Scholar]

Recommended Articles

Radiofrequency Ablation of Spinal Tumors

Full Access, , ,
American Journal of Roentgenology. 2018;210:142-152. 10.2214/AJR.17.18205
Abstract | Full Text | PDF (1177 KB) | PDF Plus (1131 KB) 
Full Access, , , ,
American Journal of Roentgenology. 2010;195:W293-W298. 10.2214/AJR.10.4192
Abstract | Full Text | PDF (679 KB) | PDF Plus (736 KB) 
Full Access,
American Journal of Roentgenology. 2020;215:502-510. 10.2214/AJR.19.22521
Abstract | Full Text | PDF (851 KB) | PDF Plus (893 KB) 
Full Access,
American Journal of Roentgenology. 2018;211:856-860. 10.2214/AJR.17.19361
Abstract | Full Text | PDF (990 KB) | PDF Plus (355 KB) 
Full Access, , , ,
American Journal of Roentgenology. 2003;180:1075-1077. 10.2214/ajr.180.4.1801075
Citation | Full Text | PDF (3255 KB) | PDF Plus (3336 KB) 
Full Access, ,
American Journal of Roentgenology. 2017;209:713-721. 10.2214/AJR.17.18527
Abstract | Full Text | PDF (1048 KB) | PDF Plus (1002 KB)