Percutaneous vertebroplasty and kyphoplasty for painful vertebral body fractures in cancer patients

ESTRUCTIVE vertebral lesions are a common sourceof morbidity in patients with metastatic disease andmultiple myeloma. Approximately 30% of patients

with various neoplastic conditions develop symptomaticspinal metastases during the course of their illness,4,26 andpain is the presenting complaint in the majority of cases.12,27

Nonoperative treatments include analgesic medicationand radiotherapy. In cases of certain tumor types, hormonetherapy, cytotoxic drugs, and bisphosphonates are increas-ingly used.21 None of these modalities is uniformly effec-tive in relieving pain or improving ambulatory status.Surgical management generally involves vertebrectomy,reconstruction with a cage or PMMA bone cement, and

stabilization with pedicle screws.7 Surgical proceduresrequire a significant postoperative recovery period andhave associated morbidity and mortality in patients whooften have limited life expectancy. In addition, surgery israrely used to treat patients with multifocal spinal disease.

Developed in France in the late 1980s, minimally inva-sive vertebroplasty involves the percutaneous injection ofPMMA into a fractured VB. Although this does not reex-pand a collapsed vertebra, reinforcing and stabilizing thefracture seems to alleviate pain. The procedure was firstused to treat aggressive vertebral hemangiomas9 and waslater applied to other lesions that weaken the VB, includingosteolytic metastases5,16,24 and osteoporotic VB collapse.1,2,

10,13,16,25 Although the European experience with vertebro-plasty in the setting of spinal metastases and myeloma is more extensive,5,6,14,16,18,24 the indications for treatmentamong most North American series are currently heavilyweighted toward osteoporotic bone disease.1,2,10,13

Procedural complications are relatively rare; however,

J. Neurosurg: Spine / Volume 98 / January, 2003

J Neurosurg (Spine 1) 98:21–30, 2003

Percutaneous vertebroplasty and kyphoplasty for painfulvertebral body fractures in cancer patients

DARYL R. FOURNEY, M.D., F.R.C.S.(C), DONALD F. SCHOMER, M.D., REMI NADER, M.D.,JENNIFER CHLAN-FOURNEY, PH.D., DIMA SUKI, PH.D., KAMRAN AHRAR, M.D., LAURENCE D. RHINES, M.D., AND ZIYA L. GOKASLAN, M.D.

Departments of Neurosurgery and Radiology, The University of Texas M. D. Anderson CancerCenter, Houston, Texas; and Department of Neurosurgery, Johns Hopkins University, Baltimore, Maryland

Object. The current North American experience with minimally invasive vertebro- and kyphoplasty is largelylimited to the treatment of benign osteoporotic compression fractures. The objective of this study was to assess thesafety and efficacy of these procedures for painful vertebral body (VB) fractures in cancer patients.

Methods. The authors reviewed a consecutive group of cancer patients (21 with myeloma and 35 with other pri-mary malignancies) undergoing vertebro- and kyphoplasty at their institution. Ninety-seven (65 vertebro- and 32kyphoplasty) procedures were performed in 56 patients during 58 treatment sessions. The mean patient age was 62years (� 13 years [standard deviation]) and the median duration of symptoms was 3.2 months. All patients sufferedintractable spinal pain secondary to VB fractures.

Patients noted marked or complete pain relief after 49 procedures (84%), and no change after five procedures (9%);early postoperative Visual Analog Scale (VAS) pain scores were unavailable in four patients (7%). No patient wasworse after treatment. Reductions in VAS pain scores remained significant up to 1 year (p = 0.02, Wilcoxon signed-rank test). Analgesic consumption was reduced at 1 month (p = 0.03, Wilcoxon signed-rank test). Median follow-uplength was 4.5 months (range 1 day–19.7 months). Asymptomatic cement leakage occurred during vertebroplasty atsix (9.2%) of 65 levels; no cement extravasation was seen during kyphoplasty. There were no deaths or complicationsrelated to the procedures. The mean percentage of restored VB height by kyphoplasty was 42 � 21%.

Conclusions. Percutaneous vertebro- and kyphoplasty provided significant pain relief in a high percentage ofpatients, and this appeared durable over time. The absence of cement leakage–related complications may reflect theuse of 1) high-viscosity cement; 2) kyphoplasty in selected cases; and 3) relatively small 3volume injection. Preciseindications for these techniques are evolving; however, they are safe and feasible in well-selected patients withrefractory spinal pain due to myeloma bone disease or metastases.

KEY WORDS • fracture • bone cement • vertebroplasty • kyphoplasty • myeloma •metastasis

D

21

Abbreviations used in this paper: AP = anteroposterior; CT =computerized tomography; IBT = inflatable bone tamp; MR = mag-netic resonance; PMMA = polymethylmethacrylate; VAS = VisualAnalog Scale; VB = vertebral body.

most are related to leakage of PMMA through cortical de-fects, with epidural compression of the neural elements.20

Percutaneous balloon kyphoplasty, a recent modifica-tion of vertebroplasty, involves inflation of a balloon with-in a collapsed VB to restore height and reduce kyphoticdeformity, prior to stabilization with PMMA.11,15,23 Therisk of cement extravasation is theoretically reduced be-cause inflation of the balloon creates a void within the VBinto which cement can be injected under relatively lowpressure.

The objective of this study was to assess the safety andefficacy of vertebro- and kyphoplasty in the treatment ofpainful VB fractures in cancer patients. To our knowledge,this report represents the largest series for both of theseprocedures in the cancer setting.

Clinical Material and Methods

We performed a retrospective review of all patients whounderwent percutaneous vertebro- or kyphoplasty at TheUniversity of Texas M. D. Anderson Cancer Center be-tween October 2000 and February 2002. Five patients inwhom there was no diagnosis of cancer (four with osteo-porotic compression fractures and one with a T-6 heman-gioma) were excluded.

Patient Population

The study population included 56 patients, 31 men(55%) and 25 women (45%), whose median age was 64years (range 30–82 years).

Demographic data, duration of symptoms, indicationsfor treatment, primary tumor site, location of disease in thespine, and treatment history (spinal surgery, vertebroplasty,kyphoplasty, spinal radiotherapy, and chemotherapy) wererecorded. A multidisciplinary team of physicians consistingof a neurosurgical or orthopedic spine surgeon and a radia-tion and/or medical oncologist evaluated all patients. Inalmost every case, pain management and rehabilitationmedicine specialists were also involved. Neurological sta-tus was evaluated using the classification system describedby Frankel, et al.8 A VAS22 was used to measure pain sta-tus. We also evaluated the use of pain medication, as out-lined in Table 1. All patients underwent plain radiographyand MR imaging of the spine. In some cases, CT scanningand radionucleotide bone scanning were also performed.

Details of the procedures, postoperative course, lengthof stay, and complications were reviewed. Vertebroplastyand kyphoplasty were performed by radiologists (D.F.S.or K.A.); however, patient evaluations (within 24 hourspostoperatively, at 1 month after surgery, at 3-month inter-vals for the 1st year, and approximately every 6 monthsthereafter) were performed by the spine surgeon. All pa-tients underwent posttreatment plain radiography. Addi-tional imaging studies were performed if treatment failedto relieve pain, for relapse of pain, or if there was a changein neurological status. The length of follow-up review wascalculated from the date of procedure to the most recentclinic visit or death.

Treatment Indications

A treatment algorithm for VB fractures in cancer pa-tients is provided in Fig. 1.

A diagnosis of cancer and disabling pain secondary topathological thoracic or lumbar VB fractures was estab-lished in all patients. Severe pain also limited the ambu-latory capacity of some patients. Conservative therapyconsisting of analgesic medication, bed rest, and in somecases external brace therapy had failed in all patients.Many were also considered poor surgical candidates be-cause of comorbid medical illnesses or multilevel spinaldisease. Contraindications included epidural compressionof the neural elements; failure to localize symptomaticlevel(s); pain that was predominantly radicular in nature;and significant medical contraindications such as uncor-rected coagulopathy, local infection at the planned injec-tion site, or intolerance to being positioned prone. In-formed consent was obtained from all patients.

Several relative factors determined whether vertebro- orkyphoplasty was performed at a given symptomatic level.Kyphoplasty was favored in the presence of 1) kyphosisthat was deemed to contribute significantly to morbidity(that is, a deformity � 20˚); and 2) disruption of the posterior vertebral cortex, where more controlled deliveryof bone cement was desired. In patients with significantvertebral collapse (vertebra plana), kyphoplasty was pre-ferred to restore height; however, vertebroplasty was per-formed when the collapse was too severe to permit in-sertion of the balloon device. Finally, vertebroplasty was

D. R. Fourney, et al.

22 J. Neurosurg: Spine / Volume 98 / January, 2003

TABLE 1Classes of analgesic medications used in patients

who underwent vertebro- and/or kyphoplasty*

Category Medication

1 none2 acetaminophen, nonsteroidal antiinflammatory medication3 codeine, hydrocodone, oxycodone, propoxyphene hydrochloride4 morphine SR/IR, fentanyl TD, oxycodone SR/IR5 intravenous narcotic agent

*IR = intermediate release; SR = slow release; TD = transdermal.

FIG. 1. Treatment algorithm for painful thoracic or lumbar VBfractures in cancer patients. See text for details. RT = radiotherapy.

performed if the patient could not tolerate general anes-thesia or the relatively longer procedure time required forkyphoplasty.

Vertebroplasty Technique

Vertebroplasty was performed under biplane fluoros-copy in the neurointerventional angiography suite, involv-ing the same sterile protocols as used in operating rooms.At the discretion of the anesthesiologist, the procedurewas performed after induction of general anesthesia orafter administration of a local anesthetic combined withintravenous narcotic or sedative drugs. Patients were posi-tioned prone with horizontal rolls placed under the chestand pelvis.

The needle entry site over the pedicle was localized inthe AP plane. The skin entry site, the underlying subcuta-neous tissue, and the periosteum were infiltrated with a50/50 mixture of 1% lidocaine and 0.25% bupivacaine. A

13-gauge needle (Osteo-Site bone biopsy needle; Cook,Inc., Bloomington, IN) was introduced through a smalldermatotomy and advanced to the posterior aspect of eachpedicle along its superolateral cortex. The bevel of theneedle was directed so that the tip pointed laterally, toavoid penetration of the spinal canal.16 The needle wasdirected anteriorly, medially, and caudally through thepedicle to reach a point within the anterior third of the VB,near the midline in the sagittal plane (Fig. 2). Within theVB, the tip was directed medially—the optimal orienta-tion for PMMA delivery to the VB. When the diagnosiswas uncertain, a 16-gauge needle (Franseen; Cook, Inc.)was introduced coaxially to obtain biopsy samples.17

We combined 40 ml of the PMMA powder (Simplex P;Stryker-Howmedica-Osteonics, Rutherford, NJ) with 6 gof sterile barium sulfate powder (Bryan; Woburn, MA) foropacification and 1 g of powdered tobramycin (Nebcin;Eli Lilly, Indianapolis, IN) for antibiotic prophylaxis priorto adding the 10 ml of liquid monomer. The preparationwas mixed until a doughy, cohesive consistency (similarto toothpaste) was obtained.

Multiple 1-ml syringes were filled with the cement mix-ture and injected into the interstices of the VB under con-stant fluoroscopic control. The objective was to fill theanterior two thirds of the VB as visualized on the lateralprojection (Figs. 3 and 4). If PMMA filled less than 50%of the VB, contralateral cannulation and injection was per-formed. Total injection volumes ranged from 2 to 8 ml.

When extravasation of cement beyond the confines ofthe VB was visualized, the procedure was usually termi-nated and a CT scan was obtained. In some cases, we wereable to prevent further leakage by adjusting the needle

J. Neurosurg: Spine / Volume 98 / January, 2003

Vertebroplasty and kyphoplasty in cancer patients

23

FIG. 2. Artist’s rendering of the ideal needle trajectory for ver-tebroplasty or kyphoplasty in AP (A) and lateral (B) views.

FIG. 3. Imaging studies demonstrating L-2 disease in a patient with multiple myeloma undergoing vertebroplasty.Left: Preoperative sagittal MR image revealing osteolytic changes at L-2 (arrow). Anteroposterior (upper) and lateral(lower) fluoroscopic images obtained during the procedure, demonstrating needle trajectory (center vertical column) andfinal PMMA casting (right vertical column).

position or by performing the injection through the con-tralateral pedicle.

At the completion of vertebroplasty, the needle was with-drawn, the puncture site closed with steri strips, and a ster-ile dressing applied. Patients were kept in bed for a mini-mum of 1 hour to allow the cement to polymerize fully.

Kyphoplasty Technique

Kyphoplasty was always performed with the patient in

the prone position after induction of general anesthesia;biplane fluoroscopic guidance was used in all cases. Bi-lateral access to the VB was uniformly obtained. The pro-cedure is depicted, step-by-step, in Fig. 5. Thirteen-gaugeneedles were advanced through the pedicles as for verte-broplasty (Fig. 2); however, the endpoint was within theposterior one third of the VB. A series of instruments wereused to create two working channels (Fig. 5). A hand-mounted drill was used to create bilateral channels within

D. R. Fourney, et al.

24 J. Neurosurg: Spine / Volume 98 / January, 2003

FIG. 4. Imaging studies revealing L-2 and L-3 disease in a patient with lung cancer metastases. The PMMA was pur-posefully injected only on the right side at L-2 to avoid cement leakage from osteolytic disruption of the left posteriorVB cortex (see upper right). Left and Center: The final cement casting is shown on AP (left) and lateral (center) fluo-roscopic images. Upper and Lower Right: Axial CT scans (L-2 [upper]; L-3 [lower]) also demonstrating final casting.

FIG. 5. Artist’s rendering of the kyphoplasty technique. Using a bilateral transpedicular approach, bone biopsy nee-dles are directed into the posterior third of the VB. Guide pins (K-wires) are used to exchange the biopsy needles for bluntcannulated obturators (1). Working cannulas (2) are then advanced, and the obturators and K-wires are removed. A hand-mounted drill (3) creates bilateral channels within the anterior aspect of the VB for placement of the IBTs (4). Ballooninflation allows restoration of VB height. Inset: The IBTs are removed and the osseous void is filled with PMMA dis-placed from bone cement cannulae (5).

the anterior aspect of the VB for placement of the IBTs(Kyphon, Inc., Sunnyvale, CA).

The IBT is a high-pressure balloon designed to reducethe VB back to its original height by creating a cavity thatis subsequently filled with the PMMA (Fig. 6). The IBTsare available in lengths of 15 and 20 mm (maximum vol-ume of 4 and 6 ml, respectively) and are selected accord-ing to the AP diameter of the VB. The IBT is ideally cen-tered between the collapsed endplates in the anterior twothirds of the VB. There are two radiopaque markers thatidentify the balloon for accurate placement (Fig. 7).Balloon inflation was performed under strict lateral fluo-roscopic control, and the inflation pressure was monitoredvia an in-line pressure gauge. Inflation endpoints were: 1)fracture reduction; 2) contact of the balloon with any cor-tical surfaces; or 3) attaining the maximum inflation pres-sure of 220 psi.

The PMMA preparation was the same as that for verte-broplasty. Specialized bone cement cannulas were filledwith the PMMA preparation. The IBTs were deflated andexchanged for the cement cannulas. A stylet, which acts asa plunger, displaces the cement into the VB. Filling wasstopped once the void left by the IBT was filled and thePMMA was observed to extend out into the trabecularspaces. The volume applied was typically 1 to 2 ml greaterthan the final inflation volume, because of interdigita-tion of PMMA into the surrounding cancellous bone. Thepuncture wounds were closed with steri strips, steriledressings were applied, and the patient was placed in thesupine position for the duration of the recovery.

Assessment of Restored Height and Kyphosis Correctionby Kyphoplasty

Assessments of restored VB height were performed asdescribed by Lieberman, et al.15 The vertical height (end-plate to endplate) at the center of the VB on lateral radio-graphs was measured before and after kyphoplasty. Thevertebra above the fracture was also measured as an es-timate of prefracture height. Calculations included the fol-lowing: height regained = posttreatment height � pre-treatment fractured height; height lost = estimatedprefracture height � pretreatment fractured height; per-centage of restored lost height = (height regained/heightlost) � 100.

Local kyphosis was assessed on the lateral radiographsby measuring the angle obtained by a line parallel to theinferior endplate of the fractured vertebra and that of thevertebra one level above, as shown in Fig. 8.

ResultsPatient Characteristics

The most frequent tumor type was multiple myeloma(21 of 56; Table 2). The median follow-up duration was4.5 months (range 1 day–19.7 months). The numbers ofpatients available for evaluation at each follow-up intervalwere 41 (73%) at 1 month; 37 (66%) at 3 months; 21(38%) at 6 months; and eight (14%) at 1 year. One patientdied during the follow-up period (2.5 months after verte-broplasty).

The median duration of spinal pain was 3.2 months

J. Neurosurg: Spine / Volume 98 / January, 2003

Vertebroplasty and kyphoplasty in cancer patients

25

FIG. 6. Kyphoplasty was performed at T-12 in a patient with multiple myeloma. Sequentially obtained AP (upper row)and lateral (lower row) fluoroscopic images demonstrating transpedicular placement of the working cannulas (left-sidedimages), inflation of the IBTs (center images), and filling of the bone voids with PMMA (right-sided images).

(range 1 week–26 months); the precise chronology of painwas difficult to determine in some patients. Neurologicaldeficits were limited to minor sensory complaints in twopatients; however, because of severe pain, 13 patients(23%) were ambulating with assistive devices such as acane or walker and two patients (4%) were essentiallywheelchair bound.

In many patients multiple levels of VB collapse or end-plate fracture were demonstrated on radiography and MRimaging. On a review of preprocedural neuroimaging re-ports, the median number of compressed VBs for the entirestudy population was two (range zero–nine). Symptomaticlevels were identified by correlating the clinical data withimaging findings (see Treatment Indications).

Several patients had risk factors for osteoporosis (ad-vanced age, postmenopausal status, chronic corticosteroiduse, and medically debilitated state), and it was often diffi-cult to determine the extent to which this was responsiblefor VB collapse compared with a purely osteolytic malig-nant process. Tumor enhancement with gadolinium on MRimaging studies, or what appeared to be osteolysis sec-ondary to multiple myeloma, was demonstrated in 42 pa-tients (75%). In three patients coaxial transpedicular biop-sy samples of vertebral lesions were acquired duringvertebroplasty, and in all of these cases metastatic diseasefrom a known primary tumor was confirmed.

Many patients had undergone previous therapy for spin-al disease (Table 2). Forty-seven patients (84%) receivedsome form of chemotherapy (hormonal or cytotoxic).Eighteen patients (32%) had undergone spinal radiothera-py. Seven patients (13%) had undergone invasive spinalprocedures: vertebroplasty at another institution in one andspinal operations in six. Surgery had been performed thelevel of a subsequent vertebroplasty in only one patient.

Ninety-seven (65 vertebroplasty and 32 kyphoplasty)procedures were performed in 56 patients during 58 treat-ment sessions. Thirty-four patients (61%) underwent ver-tebroplasty, 15 (27%) kyphoplasty, and seven (13%) un-derwent both procedures at separate levels. The meannumber of spinal levels treated per session was 1.7 (rangeone–five). The thoracolumbar junction was the most com-mon level treated (Fig. 9).

Although kyphoplasty was routinely performed after

induction of general anesthesia, 16 (46%) of 35 treatmentsessions for vertebroplasty were performed after adminis-tration of a local anesthetic.

A transpedicular approach to the VB was preferred;destruction of the pedicle necessitated an extrapedicularapproach during three vertebro- and two kyphoplasty pro-cedures. For kyphoplasty, bilateral vertebral body accesswas performed in all cases. For vertebroplasty, significantcross filling of the VB allowed a unilateral approach in 42(65%) of 65 procedures.

Postoperatively, most patients were transferred to thesame-day surgery unit. Patients were discharged on thesame day in 22 treatment sessions (38%), and within 24hours in 26 treatment sessions (45%). Four patients inwhom a treatment-related immediate analgesic effect didnot occur subsequently improved and were dischargedwithin 3 days. The remaining six treatment sessions werefor inpatients admitted prior to the procedures because ofsevere back pain. The median length of stay for the entirestudy population, and among each of the treatmentgroups, was 1 day.

D. R. Fourney, et al.

26 J. Neurosurg: Spine / Volume 98 / January, 2003

FIG. 7. Kyphoplasty was performed at T-6 in a patient with multiple myeloma. Left: Preoperative sagittal MR imagerevealing T-6 osteolytic compression fracture (white arrow). Center Left, Center Right, and Right: Sequential lateral flu-oroscopic images demonstrating the trajectory of the IBTs, inflation of the balloons, and filling of the osseous voids withPMMA. Two markers identify the balloon position for accurate placement (right [arrows]).

FIG. 8. Central VB height (double arrowheads) and local ky-phosis angles (curved arrows) were measured on lateral radio-graphs before and after kyphoplasty.

Pain and Functional Outcome

Improvement or complete pain relief was noted after 49procedures (84%), and no change occurred after 5 proce-dures (9%). Immediate postoperative VAS pain scores (ob-tained within 24 hours) were unavailable in four patients(7%). No patients were made worse by the procedures. Inmost patients, relief occurred almost immediately; howev-er, in some patients, improvements were delayed up to 36hours. Table 3 provides a summary of pain-related outcomeby treatment type. For the entire study population, themedian pre- and postoperative VAS scores were 7 and 2,respectively (p � 0.001, Wilcoxon signed-rank test). Painreduction remained significant at each follow-up intervalthrough 1 year (Fig. 10).

In the entire population, there was a significant decreasein the category of analgesic usage at 1 month (p = 0.03,Wilcoxon signed-rank test); however, the median preoper-ative score of 3 was unchanged (see Table 1 for descrip-tions). Analgesic data at 1 month are shown in Table 4.There was no significant change in the category of anal-gesic usage at any of the other follow-up intervals.

A subjective improvement in ambulatory capacity wasnoted in many patients; however, functional improvementby strict definitions of ambulatory status (Frankel grades)did not achieve statistical significance. Two (15%) of 13patients who were ambulating with assistive devices (acane or walker) preoperatively were able to walk un-assisted by 1 month. Two patients who were essentiallywheelchair bound because of severe back pain preopera-tively were both ambulating with assistive devices by 1month. Ambulatory status was not made worse in anypatient. No patients suffered neurological deficits result-ing from vertebroplasty or kyphoplasty.

Vertebral Body Height and Kyphosis Correction

The following values are presented as the mean � stan-

dard deviation. The mean vertebral height lost prior tokyphoplasty was 9.7 � 5.1 mm. The mean height regainedby the procedure was 4.5 � 3.6 mm. The differencesachieved statistical significance (p = 0.01, Student pairedt-test). The mean percentage of vertebral height lost thatwas restored by kyphoplasty was 42 � 21%.

The mean local kyphosis, measured from fluoroscopicimages obtained during kyphoplasty was 25.7 � 9.7˚). Af-ter the procedure, the mean kyphosis measured 20.5 � 8.7˚(p = 0.001, Student paired t-test). Mean improvement inlocal kyphosis was 4.1 � 3.72˚.

Complications and Relapse of Pain

There were no deaths within 30 days and no complica-tions related to the procedures. One patient was readmit-ted to the hospital 15 days after kyphoplasty for an exac-erbation of preexisting congestive heart failure. A patientwith metastatic esthesioneuroblasoma developed suddenparaplegia 13 days after an L-1 vertebroplasty secondaryto a progressive T-8 epidural metastasis.

Extrusion of PMMA beyond the confines of the VB wasobserved at six levels. All of these events occurred duringvertebroplasty (six [9.2%] of the 65 vertebroplasty proce-dures). In all cases, leakage was noted on fluoroscopy dur-ing PMMA injection and CT scanning was performedimmediately after the procedure. Cement extruded througha fractured endplate into the adjacent disc space in fivecases. Extravasation of PMMA into the anterior periverte-bral soft tissues occurred in one case, an L-3 vertebroplas-

J. Neurosurg: Spine / Volume 98 / January, 2003

Vertebroplasty and kyphoplasty in cancer patients

27

TABLE 2Demographic and clinical characteristics in 56 cancer patients

who underwent vertebro- and/or kyphoplasty*

No.of Cases (%)

Characteristic VP KP VP & KP Total

total no. of cases 34 (61) 15 (27) 7 (12) 56 (100)sex

male 19 (56) 8 (53) 4 (57) 31 (55)female 15 (44) 7 (47) 3 (43) 25 (45)

tumor typemultiple myeloma 11 (32) 6 (40) 4 (57) 21 (38)breast 8 (24) 1 (7) 0 9 (16)lung 3 (9) 2 (13) 1 (14) 6 (11)lymphoma 1 (3) 4 (27) 1 (14) 6 (11)prostate 5 (15) 0 0 5 (9)renal cell 2 (6) 0 0 2 (4)gastric 2 (6) 0 0 2 (4)other 2 (6) 2 (13) 1 (14) 5 (7)

previous treatmentVP or KP 0 0 1 (14) 1 (2)spinal op 2 (6) 4 (27) 0 6 (11)spinal radiotherapy 10 (29) 5 (33) 3 (43) 18 (32)chemotherapy 27 (79) 13 (87) 7 (100) 47 (84)

* KP = kyphoplasty; VP = vertebroplasty.

FIG. 9. Bar graph demonstrating the distribution frequency ofthe spinal levels treated by vertebro- or kyphoplasty.

TABLE 3Pain outcome after 58 treatment sessions involving

vertebro- and/or kyphoplasty*

Number of Treatment Sessions (%)

Pain Relief VP KP VP & KP Total

complete 8 (23) 1 (7) 3 (38) 12 (21)improved 22 (63) 11 (73) 4 (50) 37 (64)no change 3 (9) 1 (7) 1 (13) 5 (9)worse 0 0 0 0data unavailable 2 (6) 2 (13) 0 4 (7)total 35 (60) 15 (26) 8 (14) 58

* Results refer to an analysis of documented VAS pain scores within thefirst 24 hours. Multiple results during that period were averaged.

ty. There were no incidents of cement leakage into theepidural space or neural foramen. All cement leaks wereasymptomatic—there were no radicular complaints or neu-rological deficits as a result of cement extrusion.

There were no delayed complications as a result of ver-tebro- or kyphoplasty. Over the subsequent weeks ormonths several patients developed recurrent pain as a re-sult of osteolytic disease or vertebral collapse at other lev-els. Radiography and MR imaging did not reveal addi-tional compression or change in the pattern of the PMMA.

Two patients underwent repeated vertebro- or kypho-plasty for symptomatic osteolytic VB fractures at new lev-els—in both patients, multiple myeloma was the primarydiagnosis. Two patients, both with lymphoma and multiplesymptomatic levels of disease, eventually received spinalintrathecal morphine infusion pumps for persistent pain.

Two patients underwent subsequent spinal surgery. Thefirst was a 71-year-old woman with thoracolumbar oste-olytic disease and kyphosis secondary to multiple myelo-ma. Her initial analgesic response to T-11 kyphoplasty andT-12 vertebroplasty was good; however, the procedure lefther kyphosis largely unchanged, and within 2 months sheexperienced recurrent pain. She underwent T-11 and T-12vertebrectomy via a thoracoabdominal approach, and aseparately staged posterior instrumentation-assisted fu-

sion of T-6 through L-3 involving placement of pediclescrews and allograft. The second patient was an 82-year-old man with prostate cancer and an L-4 compressionfracture who underwent vertebroplasty with good initialsuccess. Within 2 months, his low-back pain returned andhe developed new radicular pain involving the left leg. Noneurological deficits were demonstrated. Plain radiogra-phy and CT scanning revealed the PMMA in good posi-tion and no evidence of extravasation or dislodgment.Magnetic resonance imaging demonstrated a combinationof foraminal stenosis and neural encroachment by pro-gressive epidural disease. The patient underwent two sep-arately staged procedures: an L-4 vertebrectomy via aretroperitoneal approach as well as posterior decompres-sion and L3–5 instrumentation-assisted fusion.

Five patients underwent spinal radiotherapy after verte-bro- or kyphoplasty without incident.

Discussion

The results obtained in this series indicate that, inselected cancer patients with VB fractures, vertebro- andkyphoplasty are well-tolerated procedures associated withearly clinical improvement of pain. For the entire studypopulation, there was a significant decrease in analgesic

D. R. Fourney, et al.

28 J. Neurosurg: Spine / Volume 98 / January, 2003

TABLE 4Relationship between pretreatment and 1-month posttreatment medication use*

1-Mo Postop CategoryPreop Total in

Category 1 2 3 4 5 Each Category

1 3 0 0 0 0 32 0 0 1 0 0 13 1 2 14 1 0 184 1 1 2 11 1 165 0 0 1 2 0 3

total 5 3 18 14 1 41

* The medication categories are described in Table 1.

FIG. 10. Bar graph showing pre- and postoperative median VAS scores in 56 patients who underwent vertebroplasty,kyphoplasty, both procedures, and all patients. A score of 10 indicates severe pain and a score of 0 indicates no pain. Allresults were significant (p � 0.05, Wilcoxon signed-rank test), except where indicated by an asterisk.

usage at 1 month, and the VAS pain scores reflected dur-able analgesic effects at each follow-up interval to 1 year.Lack of pain relief was mostly observed in a few patientswith advanced-stage disease. There were no early or de-layed symptomatic complications as a result of the proce-dures.

Although there may be much debate in the near futureregarding the relative risks, benefits, and costs of verte-bro- compared with kyphoplasty, the purpose of this studywas not to compare the two. Such an analysis based on an uncontrolled series with retrospectively acquired datawould be complex and misleading; selection criteria forthe two procedures differed. Rather, we emphasized paincontrol, neurological outcome, complications, and dura-bility of vertebro- and kyphoplasty in a unique patientpopulation. We agree that kyphoplasty may be differen-tiated from vertebroplasty on the basis of VB heightrestoration and, perhaps, risk of cement extravasation.15

Whether the additional complexity and cost of kyphoplas-ty are offset by these potential benefits, however, will onlybe borne out in long-term prospective randomized studies.

Vertebroplasty Procedures

To our knowledge, there are only two series in theEnglish-language literature in which investigators specifi-cally address the use of percutaneous vertebroplasty inpatients with osteolytic metastases and myeloma; both arefrom French institutions. Weill, et al.,24 reported a retro-spective series of 37 patients who underwent 52 vertebro-plasty procedures for spinal metastases. Twenty-four (73%)of 33 treatment sessions performed for pain relief resultedin “clear improvement.” Three patients had transient ra-diculopathy due to cement extrusion; surgical removal ofepidural cement in the neural foramen was required in onepatient. In the same year, Cotton, et al.,5 reported prospec-tively acquired data obtained in 37 patients who underwent40 vertebroplasty procedures for osteolytic metastases (29patients) and myeloma (eight patients). Partial or completepain relief was observed after 36 (97%) of 37 procedures.Extravasation of bone cement occurred in 29 of 40 injec-tions; most of these patients were asymptomatic, but tworequired decompressive surgery.

Among North American centers, the published experi-ence with vertebroplasty is largely limited to the setting ofosteoporotic compression fractures.1,2,10,13 In these series,the largest number of vertebroplasty-treated patients withnonosteoporotic benign or malignant vertebral destructivelesions treated was eight.2

In the present study, in patients who underwent verte-bro- without kyphoplasty, outcomes were similar to thosein the literature. Complete pain relief or improvement oc-curred after 30 (86%) of 35 treatment sessions (Table 3).Median VAS scores were significantly reduced at eachfollow-up interval to 6 months (Fig. 9); however, results at1 year did not reach statistical significance because of thesmall numbers of patients.

The rate of cement leakage (9.2% in vertebroplasty pro-cedures) compares favorably with published rates.1,2,5,6,10,13,

14,24 Cement leakage is reported to occur during as many as73% of vertebroplasty procedures.5 Most cases are asymp-tomatic, including those in the present series.

Many authors advocate the use of vertebral phlebogra-

phy prior to cement injection to 1) identify potential sitesof epidural leakage and 2) evaluate for brisk filling ofanastomotic venous channels, which may increase the riskof pulmonary embolization with bone cement.1,5,16,24 Al-though vertebral venography demonstrates the periverte-bral venous drainage, it does not reliably predict the finalcasting of cement because the viscosity of the contrastmedia and the cement are not comparable.6 In addition,particularly for tumors with necrotic cavities, postphle-bography contrast residues may be difficult to wash outand may interfere with fluoroscopic control during cementinjection. In agreement with Deramond, et al.,6 we onlyperform prevertebroplasty phlebography in cases of verte-bral hemangiomas.

Three factors may have accounted for the relatively lowrate of cement leakage in this series. First, we chose a typeof PMMA that polymerizes rapidly. Insufficient polymer-ization has been implicated as a major risk factor for pul-monary embolization of acrylic cement, which in someseries has been fatal.19 We did not inject cement until ithad achieved a very cohesive, doughy texture. Second, theinjection volumes were relatively small (range 2–8 ml).Some authors have correlated complications with exces-sive PMMA injection,16 whereas others have found noassociation.1 On the other hand, the amount of lesion fill-ing has not been shown to correlate with the degree ofpain reduction.5 Additionally, in an ex vivo biomechanicalstudy involving osteoporotic cadaveric VBs with experi-mentally induced compression fractures, strength wasrestored to all regions of the spine when as little as 2 ml ofcement was injected.3 Thus, we found no justification inthe literature for large injection volumes. Third, kypho-plasty was performed in selected cases with cortical de-fects of the posterior vertebral wall. Extravasation ofcement is theoretically less likely to occur during kypho-plasty because injections are performed under relativelylow pressures into a preformed cavity within the bone.15

Kyphoplasty Procedures

In patients treated with kyphoplasty without vertebro-plasty, pain was completely relieved or improved after 12(80%) of 15 treatment sessions (Table 3). The VAS scoresindicated significant pain reduction at each follow-upinterval to 1 year (Fig. 10). Outcomes were similar in pa-tients who underwent an additional eight treatment ses-sions involving both kypho- and vertebroplasty.

Although the kyphoplasty device has attained the ap-proval of the Food and Drug Administration, we are awareof only two studies that document its indications, efficacy,and complication profile in humans.15,23 In both studies,the procedure was conducted only in patients with osteo-porotic compression fractures. The prospective series of30 patients by Lieberman, et al.,15 however, included sixpatients with multiple myeloma.

In the present series, there were no local complicationsresulting from kyphoplasty. Neither cement leakage norIBT rupture occurred during any of the procedures.

We found significant improvement in postkyphoplastyVB height and local kyphosis. The radiographic method,involving plain lateral x-ray films, for detecting heightrestoration and deformity correction was similar to pub-lished reports.15,23 Admittedly, this method does not ac-count for the spinal deformity as a whole. We agree with

J. Neurosurg: Spine / Volume 98 / January, 2003

Vertebroplasty and kyphoplasty in cancer patients

29

Lieberman, et al.,15 that pre- and postoperative 3-ft stand-ing lateral radiographs would best measure changes inkyphosis and VB height. In future prospective studies atour center we will incorporate such measures.

Conclusions

Current nonsurgical treatment options for malignantspinal disease include analgesic medication, radiotherapy,hormone therapy, cytotoxic drugs, embolization, and bis-phosphonates.21 Vertebroplasty and kyphoplasty representimportant additions to this therapeutic arsenal. A multi-disciplinary approach to patient selection and manage-ment is essential. Precise indications for each of thesetechniques are evolving; however, kyphoplasty has theadded advantage of addressing spinal deformity and ap-pears to be associated with fewer instances of bonecement extravasation. A full understanding of the risksand benefits of vertebro- and kyphoplasty will require ran-domized clinical trials.

References

1. Amar AP, Larsen DW, Esnaashari N, et al: Percutaneous trans-pedicular polymethylmethacrylate vertebroplasty for the treat-ment of spinal compression fractures. Neurosurgery 49:1105–1115, 2001

2. Barr JD, Barr MS, Lemley TJ, et al: Percutaneous vertebroplas-ty for pain relief and spinal stabilization. Spine 25:923–928,2000

3. Belkoff SM, Mathis JM, Jasper LE, et al: The biomechanics ofvertebroplasty. The effect of cement volume on mechanical be-havior. Spine 26:1537–1541, 2001

4. Cobb CA III, Leavens ME, Eckles N: Indications for nonoper-ative treatment of spinal cord comrpession due to breast cancer.J Neurosurg 47:653–658, 1977

5. Cotten A, Dewatre F, Cortet B, et al: Percutaneous vertebro-plasty for osteolytic metastases and myeloma: effects of thepercentage of lesion filling and the leakage of methyl methacry-late at clinical follow-up. Radiology 200:525–530, 1996

6. Deramond H, Depriester C, Galibert P, et al: Percutaneous ver-tebroplasty with polymethylmethacrylate. Technique, indica-tions, and results. Radiol Clin North Am 36:533–546, 1998

7. Fourney DR, Abi-Said D, Lang FF, et al: Use of pedicle screwfixation in the management of malignant spinal disease: experi-ence in 100 consecutive procedures. J Neurosurg (Spine 1) 94:25–37, 2001

8. Frankel HL, Hancock DO, Hyslop G, et al: The value of pos-tural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia. I. Paraplegia 7:179–192, 1969

9. Galibert P, Deramond H, Rosat P, et al: [Preliminary note on thetreatment of vertebral angioma by percutaneous acrylic verte-broplasty.] Neurochirurgie 33:166–168, 1987 (Fr)

10. Gangi A, Kastler BA, Dietemann JL: Percutaneous vertebro-plasty guided by a combination of CT and fluoroscopy. AJNR15:83–86, 1994

11. Garfin SR, Yuan HA, Reiley MA: New technologies in spine:kyphoplasty and vertebroplasty for the treatment of painful os-teoporotic compression fractures. Spine 26:1511–1515, 2001

12. Gokaslan ZL, York JE, Walsh GL, et al: Transthoracic ver-tebrectomy for metastatic spinal tumors. J Neurosurg 89:599–609, 1998

13. Jensen ME, Evans AJ, Mathis JM, et al: Percutaneous poly-methylmethacrylate vertebroplasty in the treatment of osteo-porotic vertebral body compression fractures: technical aspects.AJNR 18:1897–1904, 1997

14. Kaemmerlen P, Thiesse P, Jonas P, et al: Percutaneous injectionof orthopedic cement in metastatic vertebral lesions. N Engl JMed 321:121, 1989

15. Lieberman IH, Dudeney S, Reinhardt MK, et al: Initial outcomeand efficacy of “kyphoplasty” in the treatment of painful osteo-porotic vertebral compression fractures. Spine 26:1631–1638,2001

16. Martin JB, Jean B, Sugiu K, et al: Vertebroplasty: clinical expe-rience and follow-up results. Bone 25:S1–S15, 1999

17. Minart D, Vallee JN, Cormier E, et al: Percutaneous coaxialtranspedicular biopsy of vertebral body lesions during vertebro-plasty. Neuroradiology 43:409–412, 2001

18. Murphy KJ, Deramond H: Percutaneous vertebroplasty in be-nign and malignant disease. Neuroimaging Clin N Am 10:535–545, 2000

19. Padovani B, Kasriel O, Brunner P, et al: Pulmonary embolismcaused by acrylic cement: a rare complication of percutaneousvertebroplasty. AJNR 20:375–377, 1999

20. Ratliff J, Nguyen T, Heiss J: Root and spinal cord compressionfrom methylmethacrylate vertebroplasty. Spine 26:E300–E302,2001

21. Schachar NS: An update on the nonoperative treatment of pa-tients with metastatic bone disease. Clin Orthop 382:75–81,2001

22. Scott J, Huskisson EC: Graphic representation of pain. Pain 2:175–184, 1976

23. Theodorou DJ, Theodorou SJ, Duncan TD, et al: Percutaneousballoon kyphoplasty for the correction of spinal deformity inpainful vertebral body compression fractures. Clin Imaging26:1–5, 2002

24. Weill A, Chiras J, Simon J, et al: Spinal metastases: indicationsfor and results of percutaneous injection of acrylic surgical ce-ment. Radiology 199:241–247, 1996

25. Wenger M, Markwalder TM: Surgically controlled, transpe-dicular methyl methacrylate vertebroplasty with fluoroscopic guidance. Acta Neurochir 141:625–631, 1999

26. Wong DA, Fornasier VL, MacNab I: Spinal metastases: the ob-vious, the occult, and the impostors. Spine 15:1–4, 1990

27. York JE, Wildrick DM, Gokaslan ZL: Metastatic tumors, inBenzel EC, Stillerman CB (eds): The Thoracic Spine. St.Louis: Quality Medical Publishing, 1999, pp 392–411

Manuscript received May 3, 2002.Accepted in final form August 7, 2002.Address reprint requests to: Ziya L. Gokaslan, M.D., Department

of Neurosurgery, Neurosurgical Spine Program, Johns HopkinsUniversity, Meyer 7-109, 600 North Wolfe Street, Baltimore,Maryland 21287. email: [email protected].

D. R. Fourney, et al.

30 J. Neurosurg: Spine / Volume 98 / January, 2003