Navigation of Tumor and Metastatic Lesions in the Thoracolumbar Spine

Introduction of Technology

Spine oncology is a broad field encompassing a wide array of pathologies. The vast majority are spinal metastases arising from the vertebral body, most commonly in the thoracic (70%) or lumbar spine (30%), secondary to breast, prostate, lung, and renal cell cancers. The patients with these pathologies are often extremely frail and benefit from more minimally invasive approaches. Over the past 5 years, advances in spinal navigation in robotics have helped facilitate the development of these approaches. Additionally, they are increasingly being applied for the resection of primary vertebral column malignancies. Here, we review the application of both technologies in the field of spinal oncology.

Present Practice

For both primary vertebral column malignancies and spinal metastases with spinal cord compression, surgical intervention is considered the standard of care. In isolated primary malignancies, the indication for surgery is curative resection, which has been shown to improve local control and disease-specific survival. For spinal metastases, the indications for surgery are (1) neural element decompression and (2) fixation of an unstable tumor-affected segment.

Spinal metastases make up the core of the practices of most spinal oncologists and decision-making can be guided by either the N eurologic, O ncologic, M echanical, and S ystemic (NOMS) or L ocation, M echanical instability, N eurology, O ncology, and P atient fitness, prognosis and response to prior therapy (LMNOP) frameworks. In both systems, the epidural disease burden can be graded using the Epidural Spinal Cord Compression System ( Table 5.1 ) and mechanical instability can be graded using the Spinal Instability Neoplastic Score (SINS; Table 5.2 ).

Table 5.1

Epidural spinal cord compression system for the classification of spinal cord impingement in patients with metastatic vertebral column disease.

No Cord Compression Cord Compression
Grade Definition Grade Definition
0 Bone-only disease 2 Epidural disease with cord compression, but only partial effacement of CSF space
1a Epidural disease without deformation of thecal sac 3 Epidural disease with cord compression and complete effacement of CSF space
1b Epidural disease with thecal sac deformation but no cord abutment
1c Epidural disease with thecal sac deformation and cord abutment without compression

CSF , Cerebrospinal fluid.

Table 5.2

The Spinal Instability Neoplastic Score (SINS) for classifying vertebral column instability in the context of metastatic spine disease.

Component Score
Subluxation/translation 4
Kyphosis/scoliosis only 2
Normal 0
Bone quality
Lytic 2
Mixed lytic/blastic 1
Blastic 0
Junctional: O–C2, C7–T2, T11–L1, L5–S1 3
Mobile: C3–C6, L2–L4 2
Semirigid: T3–T10 1
Rigid: S2–S5 0
Yes, mechanical 3
Yes, oncologic only 2
No 1
Posterolateral involvement
Bilateral 3
Unilateral 1
No involvement 0
Vertebral body involvement
>50% height loss 3
0%–50% height loss 2
No collapse, but >50% body involvement 1
None of the above 0

Categorization Score
Unstable 13–18
Potentially unstable 7–12
Stable 1–6

For these patients, advances in radiation treatment modalities have made it such that extensive tumor resection is not necessary to achieve durable long-term local control. As a result, with the exception of patients with de novo deformity requiring anterior column reconstruction, minimally invasive approaches such as separation surgery via a mini-open approach are generally preferred due to their lower associated morbidity. Such approaches employ either transcutaneous or transfascial placement of instrumentation. In both cases, the lack of direct visualization means that navigation of some sort must be used for instrumentation placement. Conventionally this has been done under the guidance of two-dimensional fluoroscopy; however, many surgeons are increasingly using stereotactic navigation systems (e.g., BrainLab, StealthStation) and robotic assisted instrumentation. Both strategies have the potential to reduce the overall radiation exposure to the surgeon and surgical staff. Additionally, nearly a dozen published series have suggested that operative and clinical outcomes are superior for patients with spinal metastases treated using minimally invasive, navigation-assisted versus open techniques. As indicated in two recent systematic reviews, minimally invasive techniques appear to reduce intraoperative blood loss, hospital length of stay, and complication rates relative to open surgery. Equally as important, the pain-related and neurological outcomes appear to be equivalent.

Advances in intraoperative navigation have also allowed the application of nonsurgical interventions to the spinal metastasis population, such as radiofrequency ablation, and have enhanced the ability to plan and execute osteotomies for primary malignancies. In both primary and metastatic lesions of the vertebral column, intraoperative navigation systems may be useful for assessing the degree of tumor extraction, an especially important benefit for primary lesions given the association of greater tumor excision with superior local control.

Guidelines for Spinal Navigation

The common platforms used for cranial and spinal navigation mean that many of the techniques developed by trainees during cranial cases can be applied when learning how to apply navigation to spine oncology cases. Recently, the International Society for Computer Assisted Orthopaedic Surgery published a set of guidelines for how navigation can best be applied for spine surgery. The core principles highlighted in the guidelines include (1) the importance of verifying navigation registration pre-instrumentation using known anatomic landmarks ( Fig. 5.1A–D ), (2) maintaining a clear line of sight between the infrared camera of the navigation system and the patient-fixed reference frame, (3) continuous verification of registration accuracy by comparing the image location with the known landmarks, and (4) minimizing changes in the locoregional alignment prior to the completion of navigation-assisted maneuvers, as alignment changes can result in significant registration inaccuracies. Along these lines, the group similarly indicated that the contraindications to navigation-assisted surgery include excessive spinal mobility, the inability to place a stable and rigid tracker, and poor preoperative image quality, the latter of which limits the accuracy of the navigation system. Within the realm of tumors, significant disruption of the locoregional landmarks by oncologic disease may preclude the ability to employ spinal navigation effectively.

Fig. 5.1

Illustration of the use of intraoperative navigation to verify positioning during en bloc spondylectomy.

(A) The navigated wand is placed in the field over the planned instrumentation or osteotomy site, and (B) the wand position is correlated with the projected position on the intraoperative navigation. Shown here is the wand being used to verify the position of the tumor in a patient undergoing a single-level piecemeal spondylectomy. The projected image is from an intraoperative cone-beam CT scan acquired following the posterior exposure. (C and D) Illustrate the use of the wand to monitor the extent of vertebral column resection.

The group similarly acknowledged the existence of a learning curve, a finding corroborated by others. As it pertains to the adoption of this program for spine oncology, our opinion is that navigation should be integrated into the surgical workflow in a staged approach. It should begin with application to more straightforward procedures, such as the placement of pedicle screws in normal, non-dysplastic vertebrae, and gradually move toward more complex maneuvers, culminating in its application to the guidance of osteotomy cuts. In all cases, it is our opinion that the surgeon should be acquainted with the execution of the procedure without navigation prior to attempting it under navigation guidance.

Applications of Spinal Navigation to Tumors and Present Outcomes

In spine tumor surgery, the main applications of navigation technology are instrumentation placement and tumor resection. The evidence for the former is relatively limited in the spine tumor population, as the fundamentals of pedicle screw instrumentation are highly similar for patients with oncologic and degenerative disease. Within the degenerative literature, however, several series suggest that navigation-assisted instrumentation provides superior instrumentation accuracy. The recent series include those of Jing et al., Budu et al., and Yang et al. In their series of 60 patients undergoing thoracolumbar instrumentation using either O-arm (n = 191) or fluoroscopy-assisted placement (n = 150), Jing et al. found that screws placed under navigation were significantly more likely to have clinically acceptable placement. Budu et al. similarly showed in their series of 831 screws (296 navigated using BrainLab and Orbic 3D, Siemens) that screws placed under navigation were half as likely to be misplaced. Additionally, they found that the use of navigation reduced the variability in misplacement rates across surgeons. Navigation was associated with improved accuracy in those who had greater inaccuracy using fluoroscopic assistance, suggesting that it may elevate the performance of surgeons who have subpar results using fluoroscopic assistance. These results reflect the findings of an earlier systematic review by Gelalis et al. In a review of 26 prospective trials of 6617 screws, the authors found that navigation assistance was associated with superior instrumentation accuracy. They also found that, whereas free-hand placed screws tended to deviate medially (into the canal), those placed under CT-guided navigation more commonly perforated the lateral pedicle wall. Consequently, ceteris paribus, navigation-assisted placement may decrease the neurological risk profile associated with thoracic screw placement by biasing the screws laterally, away from the spinal canal and cord.

This improved accuracy offered by navigation may be even greater in deformity operations, where patients may have hypoplastic or obliquely oriented pedicles that are difficult to cannulate without navigation assistance. As deformity is not uncommon in tumor patients, similar benefits may be observed within the oncology population. Additionally, as metastatic lesions frequently disrupt the local bony anatomy, intraoperative navigation may enable surgeons to identify more easily compromised bone and select more optimal screw trajectories. De la Garza Ramos et al. recently provided evidence to support the superiority of navigation-assisted pedicle screw instrumentation in patients with spinal metastases. In a series of 62 patients instrumented with 547 total screws (249 navigated), they found that the utilization of navigation led to a significantly lower rate of pedicle breach compared to free-hand instrumentation.

By comparison, the support for navigation in facilitating tumor resection is unique to the oncology population. For the resection of primary tumors, several groups have described good results. Most of the early descriptions were for benign pathologies. One of the first descriptions was reported by Moore and McLain, who described their experience using the Viewpoint Stereotactic Navigation System (Z-kat) for the resection of an osteoid osteoma and osteoblastoma of the cervicothoracic junction. Rajasekaran et al. also used navigation to treat benign lesions. They described the application of Iso-C 3D (Seimens) fluoroscopy-guided resection to four osteoid osteomas of the mobile spine. All the lesions arose from the posterior elements and were successfully treated with piecemeal resection, leading to durable recurrence-free survival in all patients at an average of 2 years. Nagashima et al. subsequently described the application of navigation to an osteoid osteoma of the C2 pedicle. Using the StealthStation (Medtronic) navigation system and a preoperative CT, they achieved piecemeal resection with minimal morbidity; the postoperative outcomes were not reported.

A contemporaneous description of the application of intraoperative navigation to primary vertebral column malignancies was reported by Dasenbrock et al. The authors employed intraoperative navigation for partial sacrectomy and en bloc resection of three sacral chordomas. The group utilized a preoperative CT scan and the BrainLab (BrainLab) system to guide the osteotomy cuts for the partial sacrectomy. The imaging reference frame was affixed to the L5 spinous process and the superior articulating processes of S1 were used as locoregional landmarks during point-to-point mapping for registration of the system. To guide the cuts and monitor the trajectory of the high-speed drill, a tracking probe was attached to the drill. This facilitated tracking in the sagittal and axial planes in real time. The authors reported en bloc excision with negative margins and the patient remained disease free at the 44-month follow-up. The same group also applied intraoperative, image-guided navigation to resect a recurrent cervical chordoma via an endoscopic, transcervical approach with good results.

Yang et al. subsequently employed a similar technique in their series of 26 patients undergoing computer navigation-aided resection of sacral chordoma. They reported achieving marginal (R1) or negative (R0) margins in 22 cases (85%); the overall recurrence rate was only 8% at a mean of 39 months. Jeys et al. also reported good results in 23 patients treated for primary tumors of the pelvis or sacrum. Ninety-one percent of patients had clean bone and soft tissue margins and local recurrence was seen in only 13% of cases. Perhaps more importantly, the authors found the navigation system to reflect the real-time position of their surgical instruments with respect to the patient’s anatomy. The average registration error was less than 1 mm; although no comparisons were made to surgeons navigating by anatomy alone, it is likely that the navigation significantly improved surgeon accuracy. Examinations by the same group have found navigated procedures to be associated with reduced operative time, intraoperative blood loss, and iatrogenic neurological injury, suggesting that the use of navigation may also reduce the morbidity of surgical intervention.

Subsequently, Nasser et al. described their use of O-arm (Medtronic) assisted navigation for the biopsy or resection of 41 spinal column tumors of the mobile spine and sacrum. Seven patients with metastatic lesions and 13 with primary tumors underwent extensive bony resection with the assistance of navigation. Although no non-navigated group was included for comparison, the authors concluded that navigation assistance was highly useful and could improve the precision of resection margins and instrumentation placement.

More recently, Bosma et al. published results suggesting that intraoperative navigation may improve a surgeon’s ability to achieve negative margins during the resection of primary tumors of the sacrum and pelvis. Patients who underwent navigation-assisted surgery were four times more likely to have negative margins than were historic controls who underwent resection without navigation assistance. Similar comparisons have not been conducted for lesions of the mobile spine; however, Ando et al. reported their experience of 18 patients with primary spine tumors (two sarcomas) and saw no incidence of recurrence at a median of 24.5 months. Additional experience comprised solely of patients with malignant neoplasms is required though.

Surgical Treatment for Spinal Pathology

Owing to the overall rarity of primary vertebral column tumors, the practicing surgeon is more likely to employ navigation-based techniques in a patient with metastatic disease. Nevertheless, we illustrate here the integration of intraoperative navigation into the treatment of a primary lesion.

Clinical Case 1

History/Clinical Presentation

A 65-year-old male was referred to the clinic of the senior author from an outside hospital for the management of biopsy-proven sacral chordoma. The patient had begun to notice right-sided buttock pain roughly 5 months prior to presentation that was progressive and proved refractory to medical management. The patient also endorsed progressive bowel and bladder symptoms characterized by constipation and mild urinary retention. He also endorsed perianal numbness but denied radicular pain or weakness. On examination he was found to be neurologically non-focal with 5/5 strength in all major muscle groups of the bilateral upper and lower extremities; his sensation was intact to light touch and pinprick throughout all extremities and his reflexes were normoactive at the knee and ankle. Romberg and gait testing revealed no abnormalities.

Preoperative Imaging

Imaging conducted at an outside hospital revealed a large sacral mass that was hypointense on T1- and hyperintense on T2-weighted MRI sequences ( Fig. 5.2A–F ). The lesion had a large 5 × 6 × 7 cm presacral component encasing the bilateral S1–S3 roots. Repeat imaging with contrast-enhanced sequences revealed enhancing lesions of the bilateral posterior superior iliac spine consistent with skip metastases.

Fig. 5.2

Preoperative imaging demonstrating a large (8.2 × 10.4 × 7 cm) T1-hypointense, T2-hyperintense sacral mass invading the bilateral S1–S4 foramina. A biopsy of the lesion revealed chords of physaliphorous cells with a myxoid matrix, consistent with classical chordoma. The cells stained positively for Brachyury, S100, and pankeratin, and stained negative for TTF-1, Pax8, CK7, and RCC. (A and B) Precontrast T1 imaging shows a hypointense sacral mass with a large presacral component. (C and D) Postcontrast T1-imaging shows minimal enhancement. (E and F) T2-weighted imaging shows a large presacral mass isointense to surrounding adipose tissue and hyperintense to nonaffected bone.

Surgical Plan

Consistent with the current literature, it was felt that the patient would benefit from multimodal management combining stereotactic radiation and en bloc surgical resection. The decision was made to employ neoadjuvant radiation, which comprised five fractions of 800 cGy delivered to the sacrum and bilateral medial pelvis.

Planning was then performed for en bloc resection of the lesion. Given the presence of the bilateral foci involving the PSIS = posterior superior iliac spine, the planned margins necessarily included the bilateral medial hemipelvis. The superior osteotomy was planned through the L5/S1 disk space and the lateral osteotomies were planned to include the medial 7 cm of the bilateral hemipelvis. As this would require complete disruption of the pelvic ring, the decision was made to include instrumentation, extending from L3 down to the pelvis. To improve the torsional strength of the planned construct, the decision was made to include a femoral strut allograft compressed between the two hemipelves. Fibular strut allograft was also planned for the construct to help facilitate fusion across the large defect that would result from the en bloc sacrohemipelvectomy.

Intraoperative Imaging

The surgery was performed in two stages, given the extent of the dissection planned. The first stage involved an anterior transperitoneal approach. A midline laparotomy incision was performed and the bowel contents were swept laterally to expose the posterior peritoneum, which was divided to expose the presacral component of the lesion. Using a combination of sharp and electrocautery dissection, the iliac vessels were mobilized bilaterally, exposing the L5/S1 disk space. A L5/S1 anterior hemidiskectomy was performed, forming the superior border of the surgical specimen. A rectus flap was formed with the assistance of plastic surgery to assist with closure following the posterior portion. A silastic sheath was then placed ( Fig. 5.3A ) and the incision closed; the patient was neurologically intact upon awakening.

Fig. 5.3

Intraoperative imaging showing (A) the anterior approach following placement of the silastic sheath. (B) Delivery of the tumor specimen left (C) a large sacral defect. (D) Due to the post-resection instability, a quadruple rod construct with fibular and femoral strut allograft was employed. (E) Examination of the resected specimen shows the clean edges facilitated by intraoperative navigation.

The second stage of the operation employed a posterior midline approach from L2 to the tip of the coccyx, with dissection extending over the bilateral hemipelvis to expose the PSIS and medial iliac crest. After the spinous processes were exposed, a reference frame was affixed to the L2 spinous process by way of a spinous process clamp. Registration of the patient anatomy to a preoperatively acquired CT image was then performed using the StealthStation (Medtronic) system. The acquired sequences in the preoperative CT included 0.75-mm and 3-mm slices; the 0.75-mm sequences with bone-windowing was employed for intraoperative navigation due to the higher special resolution. The L4 spinous process and bilateral L4/5 facet joints were used as image registration points. After registering the reference frame to the preoperative image, registration accuracy was confirmed by touching the L5 spinous process and the S1 bilateral laminae and superior articulating processes.

Fenestrated screws were then placed bilaterally in the L3–L5 pedicles, the positions of which were confirmed using an intraoperative CT system (O-Arm, Medtronic, Minneapolis, MN). The gluteal muscles were detached from the medial iliac crest bilaterally, preserving a cuff of healthy muscle over the PSIS to allow for negative margins. With the sacrum and hemipelves exposed, an L5 laminectomy was then performed using an ultrasonic cutting device affixed with a clamp registered to the StealthStation system. After removing the laminae, the L5 roots and thecal sac were identified and the thecal sac was doubly ligated just below the level of the L5 roots. The decision was then made to perform a three-column osteotomy through the posterior half of the L5 vertebral body to avoid compromise of the S1 tumor capsule. The osteotomy was made using the same registered ultrasonic cutting device and was connected to the L5/S1 anterior hemidiskectomy from the first stage, completing the dissociation of the sacrum from the mobile spine. The ultrasonic cutting device was again employed to form osteotomies bilaterally through the medial hemipelves, extending from the iliac crest to the sciatic notch, roughly 7 cm lateral to the sacroiliac joints. When executing the osteotomies, care was taken to start at least 5 mm outside of the proposed tumor margin, so as to avoid an intracapsular dissection. In our experience, this is essential for chordoma, which has been demonstrated to have micro-skip metastases at a mean distance of 3 mm from the tumor in up to 43% of patients. The L4 and L5 roots, and the rectum were then meticulously dissected off of the anterior face of the specimen. The specimen was delivered (see Fig. 5.3B ), leaving a large defect for reconstruction (see Fig. 5.3C ).

Given the potential shift in the locoregional anatomy, the image registration was confirmed using the bilateral iliac crests. After ensuring accurate registration, pelvic screws were placed with the assistance of the intraoperative navigation system. Vertebroplasty was performed through the L3–L5 screws bilaterally to reinforce them, and rods were contoured to the placed screws. To reinforce the sacral defect, a 20-cm femoral strut graft was placed and attached to the quadruple rod construct via crosslink; the hemipelves were collapsed onto the femoral strut graft to maximize the torsional strength of the patient’s pelvic ring. A fibular strut allograft was placed to define the final construct (see Fig. 5.3D ) and the rectus flap formed during stage 1 was completed to help eliminate dead space in the sacrectomy defect. Examination of the delivered specimen showed well-defined osteotomies (see Fig. 5.3E ), which were confirmed on postoperative histology to have produced negative margins.

Postoperative Imaging

Postoperative imaging at the 6-month visit demonstrated a residual seroma from the surgical procedure. However, no evidence of tumor recurrence was noted on MRI ( Fig. 5.4A–D ), nor were signs of instrumentation failure noted on the postoperative CT image.

Fig. 5.4

Postoperative imaging demonstrating no evidence of recurrence. (A) Sagittal and (B) axial precontrast T1 images and (C) noncontrast T2-weighted images show no evidence of lesion recurrence. (D) Postcontrast T1-weighted imaging shows no evidence of an enhancing lesion. Significant metal artifact is noted.

Postoperative Recovery and Follow-Up

The patient tolerated the initial procedure well with only mild neurological deficit, as was expected based upon the decision to sacrifice the sacral roots. His main deficits were 0/5 strength in dorsiflexion and plantarflexion with accompanying sensory loss in the S1–C0 dermatomes. He also required periodic self-catheterization for a neurogenic bladder secondary to sacrifice of the sacral roots.

Unfortunately, the patient did experience a postoperative ileus as well as a deep wound infection with mixed anaerobes and gram-negative species requiring washouts and surgical revision. Despite this, the patient continues to make a good recovery and shows no evidence of recurrence as of his 24-month follow-up.

Clinical Case 2

History/Clinical Presentation

A 64-year-old male with a known history of widely disseminated renal cell carcinoma presented to the clinic of the senior author for the evaluation of severe mechanical back pain and progressive deficits of the bilateral lower extremity. The patient had previously undergone conventional external beam radiation of his spine lesions and had been evaluated for percutaneous cementoplasty. Given his past history of unsuccessful vertebroplasty and his progressive neurological decline secondary to epidural disease, it was felt that he might benefit from surgical evaluation.

On examination, the patient was grossly 4+/5 in his bilateral lower extremities, but endorsed subjective weakness throughout. He similarly endorsed paresthesias throughout his bilateral lower extremity, although no focal sensory deficits were noted in the light touch or pinprick modalities. His gait was mildly myelopathic, although he was able to tandem walk with assistance. He endorsed mild pain with palpation of the lumbar spine and moderate to severe pain with axial loading.

Preoperative Imaging

CT imaging demonstrated a previously cemented T12 lytic lesion with anterior column height loss and de novo deformity at the thoracolumbar junction ( Fig. 5.5A and B ). Multifocal thoracolumbar disease was noted, recommending against aggressive surgical realignment. MRI demonstrated a T1-hypointense, mildly T2-hypointense mass of the T12 ( Fig. 5.5C and D ) with mild abutment of the left-sided anterolateral cord.

Sep 9, 2023 | Posted by in NEUROSURGERY | Comments Off on Navigation of Tumor and Metastatic Lesions in the Thoracolumbar Spine

Full access? Get Clinical Tree

Get Clinical Tree app for offline access