Palliative care of spinal metastases represents a

Transcription

1 ORIGINAL ARTICLE Single-stage Posterolateral Vertebrectomy for the Management of Metastatic Disease of the Thoracic and Lumbar Spine A Prospective Study of an Evolving Surgical Technique John Street, PhD, Charles Fisher, MD, Joseph Sparkes, MD, Michael Boyd, MD, Brian Kwon, PhD, Scott Paquette, MD, and Marcel Dvorak, MD Abstract: Appropriate surgical management of spinal metastases combines maximal neural decompression with simultaneous immediate spinal column stabilization in the context of a paliative operation undertaken to improve patients quality of life. We have used a single-stage posterolateral vertebrectomy (SPLV) for disease of the lumbar spine, combined with bilateral costotransversectomies in the thoracic spine, for these challenging cases. In this prospective cohort study of 96 consecutive patients with metastatic disease of the spinal column for we describe our surgical technique in detail, we examine our learning curve in its use and we analyze the long-term surgical and quality of life results in 42 patients who underwent SPLV. The mean and maximum operative blood loss was significantly lower for the SPLV group when compared with combined approaches. All patients either remained neurologically stable or had improved with surgery. Both the mean and the range visual analog scale scores were significantly improved after the SPLV. The SPLV was the only surgical approach to demonstrate a statistically significant improvement in Eastern Cooperative Oncology Group scores at 3 months after the surgery. Seventyfive percent of patients were alive at 6 months and 50% of patients survived for more than 12 months after the surgery. Eleven patients had a major complication (26%) with 9 (21%) patients required early reoperation, 7 of them for wound failure. Our data demonstrates that the SPLV represents a technically achievable improvement in surgical approach to spinal metastases when key parameters are examined. On the basis of these results, we recommend that the SPLV should be considered in all cases where resection of thoracic or lumbar spinal metastatic disease and reconstruction is contemplated. Key Words: prospective, metastatic disease, posterolateral surgery, technique (J Spinal Disord Tech 2007;20: ) Received for publication June 13, 2006; accepted January 3, From the Vancouver Hospital and Health Services Center, Spine Program, Vancouver, BC, Canada. Reprints: Dr John Street, PhD, Vancouver Hospital and Health Services Center, Spine Program, D Heather Street, Vancouver, BC V5Z 3J5, Canada ( hrb_street@yahoo.com). Copyright r 2007 by Lippincott Williams & Wilkins Palliative care of spinal metastases represents a significant and growing disease burden with the traditional therapeutic modalities of chemotherapy and radiotherapy now being complemented to an everincreasing degree by surgery. 1 6 Indications for surgery include high-grade epidural spinal cord compression, spinal instability, or tumor progression after the radiation therapy. The goals of surgical management are neural decompression, attainment of spinal stability, and local tumor control. 6 8 Because most metastatic disease involves the vertebral bodies and pedicles, most of the patients require augmentation of both anterior and posterior spinal columns. Frequently, epidural compression of the spinal cord is circumferential. Additionally, instability in the thoracic and lumbar spine often involves a compression or burst fracture, with extension of the tumor into the posterior elements. 1,4,6 These patterns of tumor involvement have traditionally been addressed through combined anterior (transthoracic or retroperitoneal)/posterior approaches, performed as 1-staged or 2-staged procedures. Anterior approaches offer good exposure to the vertebral body, allowing decompression of the central neural elements. However, they offer little or no access to the posterior elements and can involve significant morbidity especially if access is required across the thoracic cavity Posterior approaches allow for limited decompression and segmental instrumentation but do not allow reconstruction of the load bearing anterior column Combined approaches provide access to anterior and posterior elements but also suffer because of associated devastating intraoperative complications and the potential to cause, or exacerbate preexisting comorbidities, principally of the respiratory system. They may also be associated with prolonged anesthetic times, increased intraoperative blood loss, increased wound infection rate, and myriad postoperative complications Selection of the most appropriate surgical tactic depends on several factors but most importantly on the tumor location and size, patient health and associated comorbidities, and surgeon expertise and experience. J Spinal Disord Tech Volume 20, Number 7, October

2 Street et al J Spinal Disord Tech Volume 20, Number 7, October 2007 The rationale for surgery should be based not only on biomechanical considerations but also on the expected goals of therapy and the longevity of the patient. Elderly, debilitated patients with impaired immune function, poor nutritional status, and low bone marrow reserve are at a much higher risk of mortality and morbidity, regardless of surgical approach. 10,11 Our clinical experience has found that combined (anterior/posterior) single-stage or 2-stage surgery is most often unnecessary. Attainment of surgical goals requires that sufficient access be gained to the diseased area to perform safe and adequate resection of affected tissue, allowing spinal decompression and immediate instrumented stabilization. All of this must be achievable with the minimum of technical difficulty and in an expedited and safe fashion. Exactly, how this is to be achieved depends on many factors. Surgical strategies for the management of spinal metastases are still in evolution and we report our experience on a departure from the traditional combined anterior/posterior approach to a single-stage transpedicular posterolateral approach for metastases of the lumbar spine. In contrast to the previous reports of similar techniques 16,17 we use a bilateral costotransversectomy with or without nerve root sacrifice to achieve simultaneous tumor decompression and spinal stability in the thoracic spine. We anticipated that this approach would allow the surgical goals to be safely and effectively attained, while decreasing the surgical and postoperative morbidities associated with the combined procedures. Several authors have described their experience with similar techniques However, in most cases these have been retrospective studies of relatively small numbers of patients. The reported surgical techniques have differed in a number of ways and the periods of follow-up are generally insufficient to allow meaningful interpretation of the benefits of surgery. This reflects the fact that the patients are suffering from a terminal illness making any study, but most especially any sort of prospective study, a major challenge. Herein, we describe our surgical technique of the single-stage posterolateral vertebrectomy (SPLV) in detail. We examine our learning curve in its use and for the first time we analyze the long-term surgical results in a prospective cohort series. disease, and importantly the surgeon s familiarity with the approach and reconstructive options. Because most metastatic disease involves the vertebral bodies and pedicles, most of the patients require augmentation of both anterior and posterior spinal columns. Frequently, epidural compression of the spinal cord is circumferential (Fig. 1). The following is a detailed description of the procedure as we have used. A Mayfield pin skull fixation device is first attached (to prevent the pressure phenomena associated with prolonged prone position) and the patient is then positioned prone on the most appropriate table system for the level involved (Maquee table with Wilson frame for thoracic lesions or Jackson table with chest and iliac crest supports for lumbar levels). Initial lateral and posteroanterior fluoroscopy is used to, as best possible, localize the operative level and then the fluoroscopy is draped into the sterile field to help with pedicle instrumentation later. The proposed incision is infiltrated with at least 20 ml of 0.25% lidocaine with 1% adrenaline. We have found that this markedly reduces initial hemorrhage at the outset of the operation, a useful adjunct in these cases as cell saver systems cannot SURGICAL STRATEGY The SPLV is theoretically attractive because it allows access to both the anterior and posterior elements of the vertebrae via a single posterior incision and provides access for 360-degree resection/stabilization/reconstruction of the diseased segment. The SPLV therefore allows, via a single approach, the surgeon to perform circumferential decompression and fusion, providing immediate stability and avoiding a thoracotomy or retroperitoneal approach. The decision to use the posterolateral approach is based on tumor location, type of reconstruction required, patient comorbidities, extent of FIGURE 1. Rendition illustrating the circumferential nature of epidural spinal cord compression and the involvement of the vertebral bodies and pedicles with metastatic disease. Hence, most of the patients require augmentation of both anterior and posterior spinal columns. 510 r 2007 Lippincott Williams & Wilkins

3 J Spinal Disord Tech Volume 20, Number 7, October 2007 Single-stage Posterolateral Vertebrectomy routinely be used. A midline incision is made at least 2 segments above and below the level to be fused. A subperiosteal dissection is carried out using cautery to the tips of the transverse processes. The rib heads, dorsalmost 1 to 1.5 cm of rib and costotransverse joints of the involved vertebra are exposed. If the posterior elements are involved with tumor, care must be taken not to transmit pressure to the spinal cord during this initial exposure. Posterior element and adjacent soft tissue tumor is then piecemeal resected to the level of the lamina. Segmental pedicle screw fixation is then achieved using the morphologic parameters determined from the preoperative computerized tomography scan. We have found that adequate pedicle instrumentation has been possible in all cases in this series. If pedicle anatomy precludes instrumentation, it may be necessary to use a hook or similar construct. We then insert a single temporary rod, placed on the side of least tumor bulk, before proceeding with neural decompression. This helps in cases in which correction of kyphosis is attempted or where the exposure and vertebrectomy is likely to render the level unstable. This unilateral rod also facilitates distraction across the involved level during exposure. We prefer to complete segmental instrumentation before laminectomy, so that instruments and screws are not later passed to and fro above the exposed neural elements. The posterior bone work is initiated by removing the spinous processes with a rongeur. None of the locally excised bone is used for the grafting purposes. We recommend using an AM-8 burr on the Midas Rex drill (Fort Worth, TX) to thin the laminae exposing the ligamentum flavum, dura, or epidural tumor. Residual bone can be carefully removed with a 2-mm Kerrison rongeur. This laminectomy includes the bone overlying the disc spaces adjacent to the involved vertebral body segment and a normal dural plane adjacent to the epidural tumor (Fig. 2). The lateral vertebral body is exposed and extratumoral plane is created following resection of the costotransverse joint and rib head. This maneuver functions to expose the proposed vertebral body for vertebrectomy, while also dynamizing the level to facilitate deformity correction and thus cord decompression. The costotransversectomy allows better exposure of the involved vertebra and an improved working angle for circumferential decompression than the strictly transpedicular approach. Care is taken to preserve the parietal pleura and one or more thoracic nerve roots may be sacrificed as required. Thoracic nerve roots are ligated, if they are enveloped by tumor, to maximize the epidural tumor resection. The nerve roots are dissected free of tumor before ligation with vascular clips and suture ligatures. Nerve roots have not been ligated in the lumbar spine or when a major radicular feeding artery to the spine has been identified on preoperative angiogram, for example, the artery of Adamkowitz, usually between T8 and T10. The intercostal vessels should be identified and ligated in a stepwise fashion during this part of the FIGURE 2. The laminectomy includes the bone overlying the disc spaces adjacent to the involved vertebral body segment and a normal dural plane adjacent to the epidural tumor. Pedicle screw instrumentation has already been achieved. exposure. Thus the pedicles and lateral margins of the vertebrectomy are exposed. Bilateral facetectomies and complete pedicle resection to the base of the vertebral body are accomplished with the drill and curettes (Fig. 3). After bone removal, the ligamentum flavum and epidural tumor are resected starting at the interface between the tumor and dura. Bipolar cautery used on a low setting at this interface may help define the proper plane for dissection. Having dissected the epidural tumor from the posterior and lateral dura, the disc spaces adjacent to the diseased vertebral body are carefully curetted to expose normal endplates. In the thoracic spine, it may be necessary to resect a portion of the pedicle caudal to the involved vertebral body to provide exposure of the caudal disc space. A cavity is created in the vertebral body by piecemeal resection of tumor using curettes and pituitary rongeurs. Using a downgoing curette (BD zero from the Karling curette set), the vertebral body is carefully mobilized from the ventral surface of the posterior longitudinal ligament (PLL). Once exposed, resection of the intact PLL helps provide a gross resection of tumor at the anterior dura. In the thoracic spine, the plane between the dura and PLL may be difficult. Curettage or blunt dissection of the PLL may put excessive traction on the spinal dura and should be avoided. Once the anterolateral plane between the dura and PLL has been identified, the PLL can often be dissected along the anterior dura. No significant epidural bleeding has been encountered using this maneuver. Piecemeal resection of the vertebral body then is completed. The drill may be used to create a larger cavity or to resect infiltrated bone (Fig. 4). Once the vertebrectomy cavity has been created, a right angle clamp is used to create starting holes in the vertebral body at the proper depth for placement of the r 2007 Lippincott Williams & Wilkins 511

4 Street et al J Spinal Disord Tech Volume 20, Number 7, October 2007 FIGURE 3. The pedicles and lateral margins of the vertebrectomy are exposed. Bilateral facetectomies and complete pedicle resection to the base of the vertebral body are accomplished with the drill and curettes. The dotted line indicates the proposed degree of vertebrectomy. Steinmann pins, kirshner-wires, or AO small fragment (3.5 mm) bone screws. Alternatively, a fashioned length of intercostal tube can be used to contain the cement in the intervertebral space (Fig. 5). The pins are bent at a 90- degree angle and driven into cranial and caudal vertebral bodies using a needle driver with a gentle rotational movement. The screws can be inserted using the articulated screwdriver from a Pelvic Trauma or Reconstruction instrument set (Figs. 6A, B). Once radiographic confirmation shows good pin/ screw placement, polymethylmethacrylate (PMMA) containing 2 g of tobramycin is placed into the defect covering the pins. The cement is placed into the vertebrectomy defect once it achieves a consistency conformable with a wet glove. It is essential to keep the defect as dry as possible just at the time of cement application. The PMMA expands slightly just before it hardens, so it should not directly abut the anterior dura. The PMMA should be compressed against the vertebral endplates, under constant saline irrigation with a flat broad (Penfield 3) dissector to prevent gaps from forming at the bone-cement interface. Similarly, the dissector should be used under direct visualization to keep the FIGURE 4. Circumferential decompression of the neural elements is achieved. This may require the sacrifice of a thoracic nerve root as illustrated. The posterior rib, rib head, and costotransverse joint are resected to facilitate wide resection. The anterior vertebral wall is left intact to contain the anterior column cement reconstruction. dorsal surface of the cement ventral to the level of the posterior vertebral wall. Once the cement has completely cured and is solid, it should be carefully tested to ensure that it is satisfactorily fixed and does not dislodge from the intervertebral space. Now with the anterior column reconstructed, the posterior segmental instrumentation is completed, the vertebrectomy level is compressed and rod-to-rod cross connectors are applied as deemed necessary (Fig. 7). Care must be taken to then examine the exit foraminae to ensure that the resultant shortening and compression has not compromised the exiting nerve roots. Since the publication of a randomized controlled trial in 2005, we now routinely irrigate the wound with 1 L of dilute betadine as described by the authors. MATERIALS AND METHODS The Vancouver Combined Neurosurgical and Orthopedic Spine Program (CNOSP), with its extensive referral base from the adjacent British Columbia Cancer Agency, has been instrumental in the evolving management of patients presenting with metastatic disease of the spinal column. This multidisciplinary process has developed to the point where, if reasonable, surgical consultation is sought before radiation or chemotherapy is 512 r 2007 Lippincott Williams & Wilkins

5 J Spinal Disord Tech Volume 20, Number 7, October 2007 Single-stage Posterolateral Vertebrectomy FIGURE 5. A fashioned length of intercostal tube can be used to contain the cement in the intervertebral space. The cement is placed into the vertebrectomy defect once it achieves a consistency conformable with a wet glove. It is essential to keep the defect as dry as possible just at the time of cement application. applied. Thus both oncologists and surgeons can, together, optimize the staged care of the patient, thus maximizing symptom relief and minimizing adjuvant therapy-related surgical complications. Since 1999 all patients undergoing surgical intervention for metastatic spine disease have been prospectively enrolled in a comprehensive study program to determine: (a) if surgical intervention is conferring clinical benefit and (b) if any surgical approach or technique provides superior results. With specific reference to the SPLV, this prospective study has allowed us to assess its efficacy and safety in this challenging population of patients. With increasing familiarity with the posterolateral approach, it has become the preferred surgical strategy in all suitable cases for the reasons previously outlined. A prospective cohort study was performed after ethical approval had been received. All patients referred to our service for the assessment for surgical treatment of spine metastases were eligible for inclusion. Exclusion criteria were, previous history of spine surgery and if the patient had a primary spine tumor. Although patients were prospectively enrolled in this study, they were not randomized to the surgical approach used. This was a decision for the attending surgeon to make based on the parameters discussed earlier. Written consent to participate in this study was obtained from all patients. Data FIGURE 6. A and B, Posteroanterior and lateral radiographs of a study patient who underwent SPLV and PMMA reconstruction. The 3.5-mm screws can be inserted using the articulated screwdriver from a Pelvic Trauma or Reconstruction instrument set. The dissector was used under direct visualization to keep the dorsal surface of the cement ventral to the level of the posterior vertebral wall, as shown. were collected prospectively and entered on a computerized database by independent observers using standardized data collection forms and validated research FIGURE 7. Now with the anterior column reconstructed, the posterior segmental instrumentation is completed, the vertebrectomy level is compressed, and rod-to-rod cross connectors are applied as deemed necessary. r 2007 Lippincott Williams & Wilkins 513

6 Street et al J Spinal Disord Tech Volume 20, Number 7, October 2007 instruments [Eastern Cooperative Oncology Group (ECOG)]. Time-points for data collection were immediately before surgery and at 6, 12, 24, and 48 weeks after surgery. All patients were operated on by 1 or more of 5 fellowship trained spinal surgeons and the surgical approach for each case was that deemed most likely to allow the aims of the surgery to be realized. The decision for the need for surgical intervention and the most appropriate surgical approach was based on mechanical, neurologic, and patient specific criteria. The primary outcome measure was to determine the suitability of the SPLV for the treatment of metastases to the thoracic or lumbar spine based on achieving surgical goals and incidence of perioperative mortality and morbidity. Secondary outcome measures included ECOG at 3 months after the surgery, operative time, intraoperative blood loss, complications, and length of hospital stay. Estimated blood loss was recorded at the end of the surgery by an individual from the anesthesia department, who was blinded as to the surgical approach used. The ECOG score was calculated from patient data responses immediately before the surgery and at 3 months after the surgery. In the ECOG scale, 0 = fully active, able to carry on all presurgery performance without restriction; 1 = restricted in physically strenuous activity but ambulatory and able to carry out work in a sedentary nature; 2 = ambulatory and capable of all self-care but unable to carry out any work activities or up and about more than 50% of waking hours; 3 = capable of only limited self-care, confined to bed or chair more than 50% of waking hours; 4 = completely disabled. Cannot carry on any self-care. Totally confined to bed or chair; and 5 = dead. The average pain score was measured (0-10) using a patient-completed visual analog scale (VAS) and again was completed immediately before the surgery and at 3 months after the surgery. Data collected were then subjected to statistical analysis. Statistical analysis was performed using the Stata 7.0 statistical package. Comparison of parametric outcome data for each patient was performed with a paired t test. Comparison between the different groups was performed using an unpaired t test. Comparison of ordinal outcome data was performed with a Wilcoxon matched-pairs signed-ranks test. Detection of associations between variables was performed using analysis of variance for parametric variables; Cohen k coefficient for nominal data and Spearman rank correlation coefficient for ordinal data. RESULTS A total of 96 patients presented with spinal metastases within the study period. Eleven patients were ineligible for the study (3 refused to participate, 4 were unable to complete questionnaires owing to dementia or language difficulties, 1 had a simultaneous primary tumor of the spinal column, 2 had previous surgery for spinal metastases, and 1 patient originally scheduled for surgery underwent an open biopsy and refused further surgery). Nineteen patients had metastatic disease involving primarily the cervical spine and thus they were also excluded. Forty-two patients underwent simultaneous 360-degree tumor resection and immediate spinal column reconstruction through a single transpedicle/costotransverse approach and they are the focus of this report. The remaining 24 patients had thoracic or lumbar tumor decompression and spinal stabilization performed through either an anterior or posterior or combined anterior and posterior surgical approach as was deemed appropriate to achieve the surgical and therapeutic goals of neural decompression and spinal stabilization. This latter cohort of patients is used for comparison of the parameters measured to determine the efficacy of this single-stage vertebrectomy procedure to reliably achieve the stated surgical goals. The patients were not randomized before surgery but rather they had the surgical approach deemed most appropriate by the surgical team at the time. All tumors were high-grade metastatic malignancies. The histologic distribution of the primary tumor was as illustrated in Figure 8, with breast, kidney, and lung as the principle primary organs involved. For the SPLV group, there were 19 females and 23 males with an average age of years (range 19 to 76). Thirty-four patients had a thoracic SPLV and 8 patients had a lumbar. Of the 42 SPLV cases, 31 were for single level vertebrectomy. The distribution of the levels involved in the single level vertebrectomies is illustrated in Figure 9. The numbers of vertebral bodies resected were 1 in 74%, 2 in 19%, and 3 in 7%. Of the 11 cases involving multilevel vertebrectomy, 7 involved 2 contiguous levels, 3 involved 3 levels (T2-4, T6-8, and T2-4), and 1 involved 2 noncontiguous vertebral bodies (T12 and L2). Preoperative and 3-month postoperative American Spinal Injuries Association (ASIA) scoring and neurologic deficit is outlined in Figure 10. Although all patients had high-grade epidural compression on presentation, 40% were neurologically normal and 33% had only minor motor or sensory deficits. All patients either remained neurologically stable or had improved. One patient who presented as ASIA B improved to a C, 3 ASIA C s improved to D, and 10 of 14 patients who were ASIA D were normal at the end of follow-up. All patients who presented with an isolated root deficit and the single patient who presented with a Cauda Equina picture had made complete recovery. The mean and maximum operative blood loss was significantly lower for the SPLV group when compared with all other groups, particularly when compared with the combined approach, which would have necessary to achieve the same surgical goals. This is illustrated in Figure 11. All forms of surgery for metastatic disease resulted in a significant decrease in subjective reporting of pain. Both the mean and the range VAS scores were 514 r 2007 Lippincott Williams & Wilkins

7 J Spinal Disord Tech Volume 20, Number 7, October 2007 Single-stage Posterolateral Vertebrectomy FIGURE 8. Bar graph illustrating the distribution of vertebral bodies involved in the 42 study patients who underwent SPLV for metastatic disease of the thoracic or lumbar spine during the study period. significantly improved after the SPLV. This is illustrated in Figure 12. The SPLV was the only surgical approach to demonstrate a statistically significant improvement in ECOG scores at 3 months after the surgery (Fig. 13). Figure 14 illustrates the number of patients alive over the follow-up period of the study. Seventy-five percent of patients were alive for 6 months and 50% of patients survived for more than 12 months after the surgery. Fourteen of the 42 study patients had received preoperative radiotherapy compared with 11 of the 43 patients who had undergone vertebrectomy through another surgical approach (either anterior, posterior or both). Within the 42 patients in the SPLV group, 14 (33.3%) had received preoperative radiotherapy. There were a total of 10 patients who had a major wound complication, of which 3 (30%) had had preoperative FIGURE 9. Pie chart representation of the primary tumor site involved in the 42 study patients who underwent SPLV for metastatic disease of the thoracic or lumbar spine during the study period. radiotherapy. In our study population, therefore, preoperative radiotherapy did not adversely affect the rate of complications in either population of patients. Complications Twenty patients in this series of 42 suffered 1 or more complications for an overall patient complication rate of 47%. The complications are outlined in Table 1. Eleven patients had a major complication (26%): 2 cases of multiorgan failure, 1 leading to the in hospital death of the subject, 5 (12%) cases of wound dehiscence all requiring reoperation, of which 3 had deep infection and 1 required free flap coverage. There were 2 other cases of deep wound infection without dehiscence, each requiring surgical debridement and 1 requiring local flap coverage. Nine (21%) patients required early reoperation, the 7 cases previously described and 2 others were: 1 for lumbar pedicle screw malposition causing postoperative radicular pain and the other for an epidural hematoma requiring evacuation. Two patients required delayed reoperation, 1 for recurrent tumor (T4 vertebrectomy, gastrointestinal adenocarcinoma primary, +preoperative radiotherapy) and the other for hardware failure (L5 vertebrectomy, malignant melanoma primary, preoperative radiotherapy). Fifteen of the 20 patients (75%) had multiple complications as shown in Table 1. Both recorded nerve root injuries recovered without any sequelae. DISCUSSION The results obtained from this study support the use of the SPLV in surgical management of metastatic disease of the thoracic and lumbar spine. We prospectively examined the outcome of 42 consecutive patients who underwent simultaneous 360-degree tumor resection and immediate spinal column reconstruction through a single transpedicle/costotransverse approach. We examined their results when compared with 24 patients who had thoracic or lumbar tumor decompression and spinal stabilization performed through either an anterior or posterior or combined anterior and posterior surgical approach as was deemed appropriate by the attending surgeon. The SPLV provides a safe, reliable, and reproducible alternative to combined anterior and r 2007 Lippincott Williams & Wilkins 515

8 Street et al J Spinal Disord Tech Volume 20, Number 7, October 2007 FIGURE 10. Preoperative and postoperative neurologic status of the 42 study patients who underwent SPLV for metastatic disease of the thoracic or lumbar spine during the study period. posterior surgeries, and may in fact be superior in achieving the stated surgical goals with less perioperative complications and improved patient health-related quality of life outcomes. This assertion could only be validated with a very large randomized controlled trial, where preoperative staging of the tumors and extensive stratification of the patients would allow for direct comparisons. The primary aim of this study, however, is to describe our modifications of this technique of posterolateral vertebrectomy. The patient demographics, primary tumor profile, distribution of the levels involved (primarily mid thoracic), and the numbers of vertebral bodies resected (75% single level) were as expected and similar to most previous reports on surgery for metastatic disease. 1,7,9,16 21 The recent report from Wang et al 16 had fewer cases of primary breast cancer and had an equal distribution of thoracic and lumbar metastases, reflecting, we believe, the specialized nature of that institute and its reputation for dealing with rare and challenging cases. Our experience probably more reflects that of the other major centers. Our data demonstrate that all patients either remained neurologically stable or improved after the surgery. Despite the preoperative degree of cord compression, of the 37 patients with ASIA cord deficit, only 6 (16%) were ASIA B or C. One patient who presented as ASIA B improved to a C, 3 ASIA C s improved to D, and 10 of 14 patients who were ASIA D were normal at the end of the follow-up. All patients who presented with an isolated root deficit and the single patient who presented with Cauda Equina syndrome had made complete recovery. We believe that this maintenance of neurologic function 3 months after the surgery is a reflection of the adequacy of neural decompression achieved with the SPLV. Wang et al 16 have raised some concerns about this and suggested that the posterolateral approach is limited in not providing direct vision of the anterior dura. In fact, they reported on 5 patients (3.5%) who were at least 1 ASIA grade worse at 1 month after the surgery. This residual neurologic compromise is obviously multifactorial but in our experience, the inclusion of unilateral or bilateral costotransversectomies overcomes the difficulty of visualization and ensures safe circumferential decompression of the neural elements. We use this maneuver in all cases and had only 1 reoperation for tumor recurrence. This was in a patient with gastrointestinal adenocarcinoma who had received preoperative radiotherapy. In their series, Wang et al 16 reports on 16 patients (11%) who required repeated resection for symptomatic recurrence. It is not unreasonable to expect that the addition of the costotransversectomy may have improved the adequacy of decompression. The mean and maximum operative blood loss was significantly lower for the SPLV group when compared with the combined approach, which would have necessary to achieve the same surgical goals, in comparable cases. Blood loss during the surgery was very variable but this approach had both a lower average blood loss (1514 ml) and less dispersion (200 to 5000 ml) than the other surgical approaches. As all tumors were treated according to oncologic principles and approach selection was not based on an assessment of bleeding risk, it appears that the SPLV may be a drier approach with consequent benefits for the patient in terms of reduced risk of intraoperative adverse events, postoperative blood-borne FIGURE 11. Mean and range of measured operative blood loss in milliliters for all 66 patients undergoing decompressive/reconstructive surgery for metastatic disease of the thoracic or lumbar spine during the study period. 516 r 2007 Lippincott Williams & Wilkins

9 J Spinal Disord Tech Volume 20, Number 7, October 2007 Single-stage Posterolateral Vertebrectomy FIGURE 12. Mean and range preoperative and postoperative VAS pain scores for all 66 patients undergoing decompressive/reconstructive surgery for metastatic disease of the thoracic or lumbar spine during the study period. infection, and there may be a cost benefit associated with reduced use of blood bank resources. The SPLV was comparable with the other approaches in reducing the patients subjective reporting of pain. Both the mean and the range VAS scores were significantly improved after the SPLV. The SPLV was the only surgical approach to demonstrate a statistically significant improvement in ECOG scores at 3 months after the surgery. The ECOG scale assessment showed a significant improvement in SPLV patients reflecting their decreased pain, improved spinal stability, and ability to perform activities of daily living. This was not seen in the comparison surgical groups, suggesting that the SPLV is the procedure of choice for optimum clinical benefit and attainment of surgical goals. Although these comparisons are not being made on matched or randomized groups, they do highlight the ability of the SPLV to achieve the stated surgical goals in a safe and reproducible manner. The circumferential nature of tumor involvement and the need for both anterior and posterior column reconstruction presents a complex treatment decisionmaking process for the spinal surgeon. Anterior approaches generally require an access surgeon for exposure, and although they have the advantages of good visualization and the potential for excellent ventral decompression, the neural elements to be decompressed, however, are usually visualized late in the decompression process. Thus much time and effort has been spent before the most precise part of the operation is to be performed. As discussed earlier, posterior instrumentation is almost always required and thus requires a second, separate procedure. Thus the patient is exposed to all of the risks inherent in a thoracic or thoracoabdominal approach in addition to the further risks of a posterior procedure. 2,6,8,13 Posterolateral approaches allow excellent decompression, with early visualization of the tumor and neural structures. Although the extent of the achievable decompression has been the subject of some debate, we have shown that it is possible to perform a complete decompression across the midline to the contralateral pedicle with use of the costotransverse approach. With bilateral costotransversectomies, it is possible to perform a nearly complete spondylectomy. Because the chest and abdominal cavities are not opened, the risk of damage to the contents of these cavities is also minimized, as is the risk of pulmonary dysfunction. The lateral extracavitary approach has, however, been criticized for a number of reasons These include the potential for significant blood loss and the technically demanding nature of the procedure. In the upper thoracic spine, the scapula needs to be mobilized, and the reversal of thoracic kyphosis can make resection of tumor and vertebral bodies difficult. At T11-12, the diaphragm progressively limits the working FIGURE 13. Mean and range preoperative and postoperative ECOG scores for all 66 patients undergoing decompressive/reconstructive surgery for metastatic disease of the thoracic or lumbar spine during the study period. r 2007 Lippincott Williams & Wilkins 517

10 Street et al J Spinal Disord Tech Volume 20, Number 7, October 2007 FIGURE 14. Survival curve for the 42 study patients who underwent SPLV for metastatic disease of the thoracic or lumbar spine during the study period. space. Also, with posterolateral approaches, the working distance can be extensive, making the surgery difficult. However, our data have shown that the SPLV achieves adequate and long lasting decompression, as illustrated by the maintained improvement in ASIA scores and the fact that only 1 patient required reoperation for symptomatic recurrence, with less blood loss than the combined anterior and posterior approach. Cybulski et al 26 have reported on 15 patients who underwent a modified costotransversectomy for the treatment of metastasis to the thoracic spine. Postoperative neurologic improvement was seen in 83% of the patients and there was only 1 case (7%) of wound infection requiring removal of the posterior hardware. Akeyson and McCutcheon 18 have reported on a single-stage posterior complete spondylectomy, reconstruction, and posterior instrumentation for metastatic spinal disease. They outline their reasons for choice of this approach: (1) coexistent medical disease precluding a transcavitatory approach, (2) preexisting pulmonary, mediastinal, or retroperitoneal disease; (3) significant tumor located dorsal to the pedicles; and (4) involvement of 2 or more levels or involvement at 2 discontinuous levels. They have used a bipedicular approach, with PMMA reconstruction and posterior instrumentation. All patients had improvement in reported pain scores and only 5% of patients had neurologic deterioration after the surgery. Their overall complication rate was 52%, with persistent cerebrospinal fluid leakage as the most common complication. Twelve percent of the patients were alive at 12 months. In our series, 20 major complications were prospectively recorded in 11 patients (26%): 2 cases of multiorgan failure, 1 leading to the in hospital death of the subject, and there were 5 (12%) cases of wound dehiscence all requiring reoperation, of which 3 had deep infection and onerequired free flap coverage. There were 2 other cases of deep wound infection without dehiscence, each requiring surgical debridement and 1 requiring local flap coverage. Twenty-one percent of patients required early reoperation, the 7 cases of wound failure as described and 2 others were: 1 for lumbar pedicle screw malposition and the other for an epidural hematoma. Two patients required delayed reoperation, 1 for recurrent tumor and the other for hardware failure. Of the patients with complications, 15 of the 20 patients had multiple complications. These data compare very favorably with the published literature. Wang et al 16 reported a 30-day major complication rate of 14.3%, and when the 11% reoperation rate is considered, this gives an overall major complication rate of 25.3%. Of all the major published TABLE 1. Clinical Data of all Patients With Complications Subject Number Primary Complication Secondary Complication Primary Tumor Age Pre Rx Level ASIA 13 Root palsy L2 Brachial plexopathy RCC 38 Rad L1, 2 E 34 Pneumonia UTI Lung 70 Rad T4, 5 D 36 DVT Urinary retention Lung 22 No L1 E 39 Durotomy Wound infection, respiratory failure, RCC 68 No T10 D renal failure 44 Died in hospital Multiorgan failure? 62 No T9 D 45 Delayed wound healing Multiple myeloma 30 No L4 E 47 Wound infection+flap Tear-duroplasty? 48 No T2, 3, 4 E 49 Screw malposition UTI Breast 51 No T8 C 50 Dehiscence Wound infection Breast 51 Both T11, 12 C 59 UTI? 73 No T6 C 61 Dehiscence Hyponatremia Lung 72 Rads T1 D 63 Dehiscence Wound infection, UTI Lung 53 Rads T7 C 69 Wound infection UTI, acute renal failure Breast 41 No T3, 4 D 70 Root palsy L3 Durotomy Melanoma 63 Chemo L4 D 74 Multiorgan failure Decubitus ulcer, UTI, pneumonia Lung 70 Both T6, 7 E 85 Recurrent tumor GI 39 Both T4 D 89 Hardware failure Wound infection Melanoma 29 Chemo L5 E 92 Epidural hematoma Single seizure Lung 54 Rads T8 C 93 Dehiscence+flap Wound infection Lung 55 No T2 D 95 Dehiscence Lung 58 No L3 L3 root Chemo indicates chemotherapy; DVT, deep vein thrombosis; Rads, radiotherapy; RCC, renal cell carcinoma; UTI, urinary tract infection. 518 r 2007 Lippincott Williams & Wilkins

11 J Spinal Disord Tech Volume 20, Number 7, October 2007 Single-stage Posterolateral Vertebrectomy series, the most difficult complication is that of wound failure, either septic or aseptic. We had a rate of 16.6%, whereas Wang reported 11.4%. Ghogawala et al 27 examined the effect on preoperative radiotherapy on the wound failure rate. They found that patients who underwent surgery within 7 days or radiotherapy had a 32% risk as compared with 12% for those who underwent surgery as the initial therapy. Compared with that series, we had no patients who underwent surgery within 7 days of radiotherapy and this may in part explain why we found no statistical difference in patients who underwent preoperative radiotherapy. Wise et al 28 examined the risk factors for complications, incidence of complications, and survival rates in 80 patients undergoing surgery for metastatic disease of the spine. In their series, 65 patients showed no change in Frankel grade and 19 improved one Frankel grade. Thirty-five complications occurred in 20 patients (25.0%); 10 patients (12.5%) had multiple complications accounting for 23 of the 35 postoperative problems (65.7%); and 60 patients had no surgical complications (75%). They suggested that the likelihood of a complication occurring was related to Harrington classifications [I, no neurologic involvement or minor sensory impairment; II, bony involvement without collapse or instability; III, major neurologic impairment (sensory of motor) without significant bony involvement; IV, vertebral collapse with mechanical pain or instability but without significant neurologic compromise; V, vertebral collapse/instability combined with major neurologic impairment], demonstrating significant neurologic deficits and the use of preoperative radiation therapy. In general, Harrington classifications with neurologic deficits and lower Frankel grades before and after the surgery were associated with an increased risk for complication. The investigators concluded that to minimize complications, patients must be carefully selected on the basis of expected length of survival, the use of radiation therapy, presence of neurologic deficit, and impending spinal instability or collapse caused by bone destruction. Wound failure is the most devastating of complications and has been consistently reported in all series of surgeries in these challenging cases. We take many preoperative and perioperative steps to try to reduce the incidence of wound failure. All patients are medically optimized before the surgery; IV antibiotics are administered 2 hours before skin incision and continued until any drains are removed. At the end of the case, we have always irrigated with 3 L of saline containing antibiotic; however, since September 2005, we have irrigated with dilute betadine as described by Cheng et al. 29 Although these patients are not included in this current study, and the results are as yet anecdotal, we have had no cases of wound infection since using this dilute betadine solution. When deep infection has become established, we have had a low threshold for consulting our Plastic Surgery colleagues and the use of trapezius or latissimus dorsi transposition flaps. In our series, all patients who had such flaps went on to complete wound healing without the need to remove the hardware. To date, we have not yet needed to use a flap at the time of the index operation. Despite all these measures, wound failure is inevitable in many of these patients with disseminated metastatic disease. In summary, we present favorable long-term clinical outcomes of 42 patients undergoing SPLV for metastatic disease of the thoracic and lumbar spine. We describe the surgical technique in detail and in particular highlight the technical advantages of the costotransversectomy in allowing circumferential neural decompression. Our data demonstrate that with an acceptable degree of perioperative morbidity the surgical goals of enduring local disease control and spinal stability can be achieved. REFERENCES 1. Khan SN, Donthineni R. Surgical management of metastatic spine tumors. Orthop Clin North Am. 2006;37: Sundaresan N, Rothman A, Manhart K, et al. Surgery for solitary metastases of the spine: rationale and results of treatment. Spine. 2002;27: Olerud C, Jonsson B. Surgical palliation of symptomatic spinal metastases. Acta Orthop Scand. 1996;67: Ecker RD, Endo T, Wetjen NM, et al. Diagnosis and treatment of vertebral column metastases [Review]. Mayo Clinic Proc. 2005; 80: Simmons ED, Zheng Y. Vertebral tumors: surgical versus nonsurgical treatment [Review]. Clin Orthop Relat Res. 2006;443: Vrionis FD, Small J. Surgical management of metastatic spinal neoplasms [Review]. Neurosurg Focus. 2003;15:E Sundaresan N, Digiacinto GV, Hughes JE, et al. Treatment of neoplastic spinal cord compression: results of a prospective study. Neurosurgery. 1991;29: Bauer H, Tomita K, Kawahara N, et al. Surgical strategy for spinal metastases. Spine. 2002;27: Holman PJ, Suki D, McCutcheon I, et al. Surgical management of metastatic disease of the lumbar spine: experience with 139 patients. J Neurosurg Spine. 2005;2: Klekamp J, Samii H. Surgical results for spinal metastases. Acta Neurochir (Wien). 1998;140: Aebi M. Spinal metastasis in the elderly [Review]. Eur Spine J. 2003; 12(suppl 2):S202 S213. Epub September 23, Klimo P Jr, Dailey AT, Fessler RG. Posterior surgical approaches and outcomes in metastatic spine-disease. Neurosurg Clin North Am. 2004;15: Sundaresan N, Steinberger AA, Moore F, et al. Indications and results of combined anterior-posterior approaches for spine tumor surgery. J Neurosurg. 1996;85: Gokaslan ZL, York JE, Walsh GL, et al. Transthoracic vertebrectomy for metastatic spinal tumors. J Neurosurg. 1998;89: Jonsson B, Sjostrom L, Olerud C, et al. Outcome after limited posterior surgery for thoracic and lumbar spine metastases. Eur Spine J. 1996;5: Wang JC, Boland P, Mitra N, et al. Single-stage posterolateral transpedicular approach for resection of epidural metastatic spine tumors involving the vertebral body with circumferential reconstruction: results in 140 patients. Invited submission from the Joint Section Meeting on Disorders of the Spine and Peripheral Nerves, March J Neurosurg Spine. 2004;1: Bilsky MH, Boland P, Lis E, et al. Single-stage posterolateral transpedicle approach for spondylectomy, epidural decompression, and circumferential fusion of spinal metastases. Spine. 2000;25: Discussion Akeyson EW, McCutcheon IE. Single-stage posterior vertebrectomy and replacement combined with posterior instrumentation for spinal metastasis [Review]. J Neurosurg. 1996;85: r 2007 Lippincott Williams & Wilkins 519

12 Street et al J Spinal Disord Tech Volume 20, Number 7, October Wiggins GC, Mirza S, Bellabarba C, et al. Perioperative complications with costotransversectomy and anterior approaches to thoracic and thoracolumbar tumors. Neurosurg Focus. 2001;11:e Cahill DW, Kumar R. Palliative subtotal vertebrectomy with anterior and posterior reconstruction via a single posterior approach. J Neurosurg. 1999;90(1 suppl): Shaw B, Mansfield FL, Borges L. One-stage posterolateral decompression and stabilization for primary and metastatic vertebral tumors in the thoracic and lumbar spine. J Neurosurg. 1989; 70: Schmidt MH, Larson SJ, Maiman DJ. The lateral extracavitary approach to the thoracic and lumbar spine [Review]. Neurosurg Clin N Am. 2004;15: Lifshutz J, Lidar Z, Maiman D. Evolution of the lateral extracavitary approach to the spine. Neurosurg Focus. 2004;16: E Resnick DK, Benzel EC. Lateral extracavitary approach for thoracic and thoracolumbar spine trauma: operative complications [Review]. Neurosurgery. 1998;43: Discussion Graham AW III, Mac Millan M, Fessler RG. Lateral extracavitary approach to the thoracic and thoracolumbar spine. Orthopedics. 1997;20: Cybulski GR, Stone JL, Opesanmi O. Spinal cord decompression via a modified costotransversectomy approach combined with posterior instrumentation for management of metastatic neoplasms of the thoracic spine. Surg Neurol. 1991;35: Ghogawala Z, Mansfield FL, Borges LF. Spinal radiation before surgical decompression adversely affects outcomes of surgery for symptomatic metastatic spinal cord compression. Spine. 2001; Wise JJ, Fischgrund JS, Herkowitz HN, et al. Complication, survival rates, and risk factors of surgery for metastatic disease of the spine. Spine. 1999;24: Cheng MT, Chang MC, Wang ST, et al. Efficacy of dilute betadine solution irrigation in the prevention of postoperative infection of spinal surgery. Spine. 2005;30: r 2007 Lippincott Williams & Wilkins