Cervicothoracic Junction Tumors: Regional Challenges

Published on 02/04/2015 by admin

Filed under Neurosurgery

Last modified 22/04/2025

Print this page

This article have been viewed 3299 times

Chapter 112 Cervicothoracic Junction Tumors

Regional Challenges

Cristian Brotea, Michael P. Steinmetz

Incidence and Statistics

The upper thoracic vertebrae account for 15% of patients with tumors of the spine. The T1-4 region accounts for 10% of all spinal metastases.¹ Subaxial cervical spine tumors also are uncommon and constitute less than 10% of spinal metastases.² Due to the smaller canal size, tenuous blood supply at the cervicothoracic junction, along with unique biomechanical stresses that can lead to instability, the chance of neurologic involvement can be as high as 80%.³

The cervicothoracic junction is a transition zone between the more mobile cervical spine, with its lordotic alignment, and the much stiffer, kyphotic thoracic spine, stabilized ventrally by the rib cage. A variety of pathologic entities that can result in kyphotic deformity can be seen at the cervicothoracic junction, including tumor, infection, spondylosis, traumatic injury, and iatrogenic destabilization. Following surgery, the kyphotic deformity may worsen further as a result of failing to account for the unique biomechanics of the cervicothoracic junction.⁴ Cervicothoracic pathology can include either primary or metastatic oncologic disease. Several studies have looked specifically at tumors in this anatomic region.⁵^–⁷ One series of 19 patients with cervicothoracic tumors had 6 patients with primary pathology, including angiosarcoma,¹ chordoma,¹ lymphoma,² plasmacytoma,¹ and schwannoma.¹ The other 13 patients (68%) had metastatic disease, with metastases to the lung⁵ and prostate² being the most common. Other tumors seen were breast cancers, melanoma, and renal, ovarian, and colon cancers. Another series with 32 patients described 19 who had lung cancer and vertebral body invasion (Pancoast tumors), 11 with metastasis, 1 with chondrosarcoma, and 1 with myeloma.⁶ An and Vaccaro⁷ reported a group of 15 patients with cervicothoracic pathology, including 8 metastatic lesions (2 breast, 2 lung, 2 prostate, 1 thyroid, and 1 adenocarcinoma of unknown origin) and the remainder primary benign (aneurysmal bone cyst, giant cell tumor) and malignant tumors (myeloma, lymphoma, neurofibrosarcoma, and hemangioendothelioma).

Nonsurgical and Surgical Treatment

Spinal metastases, especially at the cervicothoracic junction, present unique treatment considerations. The goal of treatment is to improve or maintain neurologic function, achieve spinal stability, and attain local tumor control. The traditional methods of dealing with this pathology include radiation therapy, chemotherapy, and hormonal treatment, as well as surgery. Bilsky et al.⁸ expanded on the NOMS pneumonic (neurologic, oncologic considerations, mechanical instability, and systemic disease and comorbidities), developing an algorithmic approach to decision making.

Neurologic and mechanical instability components can be detected in a patient’s clinical presentation. For example, epidural spinal cord compression due to metastases can be inferred from radiographic presentation as well as the patient’s signs and symptoms. A red flag for a physician is the complaint of neck and back pain mostly at night or in the early morning, resolving during the day, that is nonmechanical pain. Diurnal biologic pain is mainly due to hormonal secretion and inflammatory mediators excreted by the tumor. This is one of the earliest symptoms of bony metastasis and may correlate with tumor invasion of the vertebral body. While difficult to distinguish from tumor-related pain, pain due to mechanical instability can be a large component of the patient’s complaints. It is related to movement and can correlate directly with bone and ligamentous destruction. This type of pain is described as deep and agonizing and occurs when loading the spine, as when walking or carrying objects. Flexion and extension can elicit pain in the mechanically unstable cervical and thoracic spine and occasionally can be evident on radiographic imaging. Whereas cervical instability is more evident with the patient erect, lying down can cause pain in patients with thoracic instability as the thoracic kyphotic deformity straightens. Radiculopathy may be present if instability results in foraminal compromise or if tumor has extended into the foramen.

Radiation, Chemotherapy, and Stereotactic Radiosurgery Preoperatively and Postoperatively

The ultimate goal of treatment in patients with cervicothoracic tumors is preservation or recovery of neurologic and functional status and treatment of pain. Although radiation is less invasive, has fewer complications, and traditionally has been found to be at least equivalent to surgery for the treatment of metastases, patients with epidural spinal cord compression have been shown to respond well to surgical intervention.⁹ Initially, basic decompressive procedures, such as laminectomy without instrumentation, were used to resect epidural disease, although these procedures did create iatrogenic instability. Patients receiving radiation therapy as a primary treatment had 79% maintenance of ambulatory status and a 42% rate of neurologic recovery. By comparison, patients undergoing surgical intervention as mentioned above had worse odds of improving their preoperative neurologic status, with 33% chance of recovery and up to 50% of the patients worse off neurologically.^10,¹¹

In 1978, Gilbert and colleagues compared radiation therapy directly with laminectomy alone. They showed that patients with radiosensitive tumors, including breast cancers, myeloma, and lymphoma, had significantly better functional outcomes when compared with less radiosensitive tumors, including lung, colon, and renal cell carcinoma, regardless of the tumor treatment. Gilbert et al. helped steer primary treatment in certain metastatic diseases to radiation therapy.¹²

In the last 20 years, neurologic outcomes, specifically ambulation, have been improving with surgical treatment. Fewer than 5% of patients undergoing surgical treatment have worse neurologic outcomes after surgery, a change that can be attributed mainly to technical advances and improvement in surgical approaches and instrumentation.

In general, surgical resection of epidural metastases, if tolerated well from a medical standpoint, should be followed by radiation therapy, or possibly systemic chemotherapy or hormonal therapy. This combined approach can be used to secure local tumor control.⁹ Newer techniques, including image-guided intensity-modulated radiation therapy (IGIMRT) and the cyberknife, improve the chances for tumor control in the cases of radiation therapy–resistant tumors, especially postoperatively, as opposed to conventional external-beam radiation therapy (EBRT).^13,¹⁴ At Memorial Sloan-Kettering Cancer Center, the typical tumor board protocol currently dictates that a patient with a radiosensitive tumor, regardless of the degree of spinal cord compression or myelopathy, be irradiated with standard EBRT or with single-fraction IGIMRT or chemotherapy for noncompressive masses. Patients with radioresistant tumors and a low degree of compression can undergo EBRT or IGIMRT 24-Gy single fractions, whereas the patients with higher-grade spinal compression are offered surgical decompression and fixation followed by IGIMRT.⁸ Despite diligent treatment, the chance of recurrence in some of the surgical series remains high, approaching 96% at 4 years.¹⁵

Superior sulcus non–small-cell lung carcinomas, which are extremely difficult to treat because of their aggressiveness and invasion of adjacent spine, brachial plexus, and subclavian vessels, have been treated traditionally with radiation and surgery, yielding a 30% 5-year survival rate. More recent data suggest that induction chemoradiation (two cycles of cisplatin and etoposide with radiation) and surgical resection (thoracotomy) followed by two more cycles of chemotherapy can yield higher survival rates, of 44% at 5 years.¹⁶

Surgical Treatment

Surgical treatment of tumors of the cervicothoracic junction often is challenging due to the presence of major vascular and soft tissue structures requiring exquisite preoperative planning and approach. The cervicothoracic junction is unique biomechanically. This uniqueness must be accounted for when considering surgery in this region. Preoperative imaging helps define the neurovascular involvement of the tumor, guiding the direction of the surgical treatment. In the cervical spine, for example, MRI can help identify epidural spinal cord compression and vertebral artery involvement. Preoperative angiography and balloon occlusion tests may be necessary if tumor dissection or occlusion of a single vertebral artery is likely or necessary.¹⁷ For ventral surgical approaches to the subaxial cervical spine and cervicothoracic junction, the recurrent and superior laryngeal nerve functions must be evaluated. Preoperative laryngoscopy can be used to assess bilateral vocal cord dysfunction; if present, the approach should be from the ipsilateral side. Superficial laryngeal nerve function must be evaluated by swallowing studies.

Although radiation therapy remains the treatment of choice for many patients with metastatic disease, ¹⁷ primary malignant tumors often are treated surgically for oncologic considerations, because they can be resistant to chemotherapy and radiation therapy. En bloc resection, which offers a higher probability of cure and local disease control, is preferred, but it typically is more technically demanding than an intralesional resection, especially if it extends beyond the vertebral body or dorsal elements. It may occasionally entail sacrificing the adjoining vertebral artery or an exiting nerve root.

Ideally, most hypervascular spinal tumors should undergo arterial embolization prior to surgical resection to minimize the risk of intraoperative and postoperative bleeding and enhance the surgeon’s ability to safely decompress the spinal cord and maximize tumor removal.¹⁸^–²⁰ The need for preoperative embolization is based on tumor histology, as seen in renal cell carcinoma, sarcomas (e.g., angiosarcoma and leiomyosarcoma), follicular and papillary thyroid carcinoma, hepatocellular carcinoma, germ cell tumors, and neuroendocrine tumors such as paragangliomas. The arterial blood supply must be conducive to such an attempt, because the feeder vessels must be large and accessible for microcatheterization. Some other hypervascular tumors, such as melanoma and multiple myeloma, are fed by a finer capillary network, which may not be as accessible.¹⁸ On MRI, tumor hypervascularity may be identified as bright contrast enhancement and flow voids representing blood vessels. Hyperintensity also may be present on both T1- and T2-weighted images due to extracellular methemoglobin or hypointensity from the breakdown products of methemoglobin. Prabhu et al.¹⁸ have recommended preoperative angiography for the following circumstances: if the tumors are of known hypervascularity, regardless of MRI findings; if the MRI suggests hypervascularity regardless of histology; or if the primary tumor origin is unknown. Cervical tumors need special consideration, because there is an increased risk of cerebral or brainstem infarction with preoperative embolization.²¹ Two percent of patients experience neurologic complications, and 4% to 10% can experience local or systemic complications following embolization.¹⁸ Therefore, only in special circumstances is embolization used for cervical malignancies.

Historical Surgical Approaches

Ventral surgical approaches to the thoracic spine initially were developed to treat Pott disease, and similar principles are used today to treat primary and metastatic tumors of the spine, with the cervicothoracic junction being the most challenging. Menard ²² was the first to describe an approach to this pathology; he resected a portion of the rib and transverse process to gain access to a relatively small portion of the vertebral body. Capener²³ took this approach a step further by removing a longer segment of the rib, resulting in a more generous exposure and allowing ventrolateral decompression of the spinal cord.

In 1957, Nanson ²⁴ described an oblique approach at the cervicothoracic junction that used a supraclavicular incision to gain access to the upper thoracic sympathetic ganglia. The exposure of the vertebral bodies, however, was very limited. Cauchoix and Binet²⁵ described a more aggressive approach to the cervicothoracic junction that involved performing a median sternotomy. This approach was essentially abandoned due to the 40% mortality rate in patients with Pott disease²⁶ and a 33% mortality rate in patients with cervicothoracic pathology.³ In 1984, Sundarsan et al.²⁷ described a ventral approach, which involves a T-shaped incision with the transverse portion just proximal to the clavicles and a vertical component in the midline over the sternum (Fig. 112-1). Flaps are raised superiorly and inferiorly, and the external jugular vein and medial supraclavicular nerve are divided. The medial one third of the clavicle is excised, as well as a rectangular block of bone from the sternum (Fig. 112-2). Under the manubrium sterni the subclavian vein is dissected free, and the thymus may be removed for further exposure. The laryngeal nerve must be identified as the dissection continues deep between the carotid sheath laterally and the esophagus medially. The resected clavicle can be used as structural bone graft at the corpectomy site without further instrumentation.²⁷

FIGURE 112-1 The T-shaped incision for the transsternal exposure of the ventral cervicothoracic junction and the upper thoracic spine.

(Redrawn with permission from Benzel EC, editor: Surgical exposure of the spine: an extensile approach, Rolling Meadows, IL, 1995, American Association of Neurological Surgeons, p 161.)

FIGURE 112-2 Resection of the manubrium and the medial part of the clavicle.

This approach was slightly modified by Birch et al.,²⁸ who described an osseomuscular flap containing the medial one third of the clavicle and manubrium in one piece attached to the sternocleidomastoid muscle and later reattached with wires and plate-screw fixation. There were four clavicle nonunions and one transient recurrent laryngeal nerve palsy in the 17 patients in this study. By leaving the sternocleidomastoid muscle attached, the risk of pulmonary compromise was lowered in patients with respiratory problems. Kurz and Herkowitz²⁹ modified this approach by only removing the medial clavicle without resecting the manubrium. Midline-crossing vascular structures limit exposure caudally to approximately T3 (Fig. 112-3). The lateral extracavitary approach to traumatic lesions of the thoracic and lumbar spine has been described for approaches as high as T3 but has limited utility with cervicothoracic junction tumors.³⁰

FIGURE 112-3 A, Ventral cervicothoracic exposure is complicated by midline crossing vascular structures, as depicted. B, This limits ventral exposure to approximately T3, even if the innominate vein is ligated.

(Reprinted with permission from Benzel EC, editor: Biomechanics of spine stabilization, ed 2, New York, 2001, Thieme.)

Ventral “Trapdoor” Approach

The “trapdoor” exposure to the cervicothoracic junction has the advantage of leaving the sternoclavicular junction as well as the clavicle itself intact. Originally described by Nazzaro et al.,³¹ it is a combination of the ventrolateral cervical approach, median sternotomy, and ventrolateral thoracotomy. This approach is ideal for patients with ventral pathology at the cervicothoracic junction down to the T3-4 level who need decompression and subsequent reconstruction. Patients undergoing the procedure must have their neurologic status closely monitored with motor and somatosensory-evoked potentials and intravascular volume status with the help of a central line or Swan-Ganz catheter. Lines should not be placed in the right internal jugular or subclavian vein, which will be in the field,³² assuming a right-sided approach is utilized.

The incision is made along the ventral border of the sternocleidomastoid muscle on the right side, to the sternal notch and down to the level of the fourth intercostal space, lying over the midline of the sternum and then curving laterally over the fourth rib. Proximally, the dissection is taken deep in the usual Smith-Robinson technique ³³ by incising the platysma muscle and staying in the avascular plane between the carotid sheath laterally and the tracheoesophageal viscera medially. It may be necessary to transect or ligate the omohyoid muscle and the middle thyroid vessels in order to obtain a wider exposure. Once the prevertebral fascia is reached overlying the ventral cervical spine, the longus colli muscles are visualized. Distally, a plane is identified between the manubrium and the underlying vascular structures. Once the right lung is deflated, the sternum is transected longitudinally to the level of the fourth rib, and the incision is connected to the laterally made thoracotomy dissection. A chest spreader is used to obtain deep access. The mammary artery is identified; this artery typically requires ligation.

Next, the vagus nerve is identified by opening the carotid sheath, and the innominate artery is dissected free to the level of the right carotid artery. The right recurrent laryngeal nerve is identified by tracing it from the right vagus nerve as it loops under the subclavian artery. The nerve is followed into the tracheoesophageal groove. This nerve, along with the aforementioned arteries, must be marked with vessel loops and protected. The cervicothoracic junction down to the level of T1 can be identified with further blunt dissection. The T3 and T4 levels are found behind the great vessels at that level, as is the junction of the mediastinal, apical, and dorsal chest wall pleurae. If the dissection is taken even further distally to T5, the azygos vein may be encountered joining the superior vena cava; it may be necessary to dissect and ligate it. Once the exposure is deemed adequate, the segmental arteries must be identified and ligated. The level should be confirmed with intraoperative fluoroscopy. In order to perform a vertebrectomy, the adjacent discs cephalad and caudad must be identified and removed. Using a high-speed diamond-tipped bur or an ultrasonic aspirator, the vertebral body may be removed to the level of the posterior longitudinal ligament (PLL). The PLL also can be removed to gain access to the ventral aspect of the thecal sac. Once the pathology is excised, the reconstruction can be performed with the help of newly developed titanium expandable cages or polymethylmethacrylate in patients with a shorter life expectancy. The column reconstruction can be augmented with a ventral plate and screws.

Risks with this approach include recurrent laryngeal nerve palsy as well as vascular damage. A dural laceration with this type of approach can be very difficult to manage. A piece of muscle can be applied over the dura with Gelfoam and fibrin glue layers. Postoperative lumbar cerebrospinal fluid (CSF) drainage may be necessary for 4 to 5 days postoperatively if CSF is visualized during surgery and a watertight closure cannot be achieved. It may be necessary to remove the chest tubes early, because intracavitary negative pressure may contribute to a continued leak. Another dreaded complication is iatrogenic injury to the esophagus, which may occur as a result of sharp deep dissection or the bur inadvertently making contact with it as it protrudes next to the retractors. Often the injury is not immediately recognized; it can present as postoperative infection, sepsis, mediastinitis, or fistula. If it is identified at the time of initial injury, the tear must be primarily repaired, the wound drained, and a nasogastric tube placed under direct visualization until it heals. Afterward, a barium swallow study should be performed to confirm that the tear is sealed. More likely, however, the injury is suggested postoperatively by the presence of crepitus in the neck or mediastinal air on chest radiograph. An esophagogram or CT scan usually can make the diagnosis definitively. At this point, reexploration has to be considered for primary repair, nasogastric drainage, and antibiotics.

Lateral Parascapular Extrapleural Approach

A straight ventral approach occasionally should be supplemented by a dorsal approach to allow for supplemental instrumentation or decompression. In 1991, Fessler et al.³⁴ described a lateral parascapular approach to the cervicothoracic junction that allowed access to the anterior column of the spine and simultaneous placement of dorsal instrumentation. The patient is intubated with a double-lumen endotracheal tube and placed prone on chest rolls. (We often do not use a double-lumen tube.) The incision is begun midline over the spinous processes of the subaxial cervical spine and directed distally in a curved fashion to the ipsilateral scapula. The trapezius and rhomboid muscles are identified under the deep fascia and elevated subperiosteally toward the medial border of the scapula as a musculocutaneous flap, with the interspinous ligaments and a cuff of muscles left intact for approximation later. The intact paraspinal muscles (the spinalis thoracis and longissimus thoracis) are elevated and retracted dorsally and medially. The ileocostalis group of the erector spinae musculature can be elevated and retracted either medially or laterally. These steps provide the surgeon with an excellent exposure of the upper dorsal rib cage and dorsal vertebral elements (Fig. 112-4).

FIGURE 112-4 Fascial incisions. A T-shaped (A) or a paramedian straight (B) fascial incision may be used to transgress the thoracodorsal fascia. C, Incision of the thoracodorsal fascia gains access to the lateral aspect of the erector spinae muscle in the thoracic region. D, Retraction of the thoracic erector spinae muscles medially begins to expose the underlying ribs. E, In the thoracic region, the medial 6 to 10 cm of the ribs are resected. F, Further subperiosteal dissection along the undersurface of the ribs, pedicles, and vertebral bodies gains access to the lateral aspect of the spinal column. G, The appropriate pedicle is removed, and visualization of the lateral aspect of the dural sac, as well as its relationship between the ventrally situated compressive pathology, is attained. H, A trough is made just ventral to the dorsal aspect of the vertebral body with a high-speed bur and/or curved curet. I, The trough is thinned to facilitate the ventral displacement of the remaining shelf of bone away from the dural sac via a reversed-angle curet. J, A rib or other suitable graft is impacted into slots fashioned in the caudal aspect of the rostral vertebral body and rostral aspect of the caudal vertebral body.

(Reprinted with permission from Benzel EC, editor: Surgical exposure of the spine: an extensile approach, Rolling Meadows, IL, 1995, American Association of Neurological Surgeons.)

At this point, the ipsilateral lung is collapsed to provide better exposure. The neurovascular bundles are stripped subperiosteally off the ribs and the costrotransverse and radiate ligaments excised sharply. The nerves and arteries are ligated and transected. Next, the thoracic cage is opened dorsally by removing ribs from their costrotransverse and costovertebral articulations laterally to the dorsal bend. Most often, only one rib is resected, but more than one may be removed if necessary. The ligated nerves are followed medially to the vertebral foramen. The sympathetic chain is identified on the lateral vertebral surface, and once the rami communicantes are transected and the segmental arteries controlled, it can be displaced ventrolaterally in a subperiosteal dissection uncovering the underlying vertebral body, pedicle, and foramina. The en bloc tumor resection/corpectomy can now be performed. If possible, the anterior and posterior longitudinal ligaments should be left intact. The resected ribs can be used as structural autograft. The nerve roots should be ligated proximal to the dorsal root ganglion to prevent formation of neuroma and postoperative neuralgia. Dorsal instrumentation also can be placed from this same approach. Complications include pneumonia, wound problems, Horner syndrome or sympathectomy, intercostal neuralgia, chest wall and medial arm hypalgesia, vascular spinal cord injury, C8 and T1 radiculopathy, CSF leak, and pneumothorax.

Dorsal Approach

Patients with cervicothoracic junction tumors that involve the dorsal elements without fixed deformity and with adequate cervical lordosis, may be approached dorsally. Surgeons must be familiar with this approach when performing multistage reconstructions that may involve the previously described ventral approach supplemented by dorsal instrumentation. Preoperative imaging including 3-foot scoliosis views is required to evaluate for any significant kyphosis or any coronal plane deformity throughout the entire spine. Flexion and extension views also may be useful to determine whether intraoperative deformity correction will be possible or if osteotomies and soft tissue releases are necessary. The choice and number of approaches are influenced by all of these factors, with some deformities necessitating combined two- or three-stage approaches. CT imaging is valuable for looking at bone quality; the specific cervicothoracic anatomy, including the size of the subaxial cervical lateral masses; C7-T3 pedicle widths; and angulations needed for instrumentation.

The patient usually is placed prone on the operating table with the head position well controlled with a three-point skull fixation device. Neck position should be checked with intraoperative fluoroscopy to prevent inadvertent fusion in an unbalanced position sagitally or coronally. Neurologic monitoring, including motor and somatosensory evoked potentials baseline values, should be obtained prior to positioning and rechecked afterward. With stenosis due to epidural pathology, cervical hyperextension can result in traumatic spinal cord injury. For these reasons, it may be necessary to perform intubation fiberoptically with the head and neck in neutral position. The arms must be carefully tucked at the sides without applying undue traction and inducing cervical plexopathy. The incision is made midline over the spinous processes at the cervicothoracic junction, exposing the lamina and lateral masses by subperiostally elevating the paraspinal muscles. In the thoracic spine, the transverse processes should be visualized to help with the placement of instrumentation and exposure of an adequate fusion bed. When only a partial vertebrectomy or hemivertebrectomy has to be achieved from a dorsal approach, the corresponding roots must be identified and ligated, following which an oblique osteotomy may be performed to isolate the pathology. Instrumentation has to be placed on the contralateral side to stabilize the segment prior to the excision.

Cervical laminectomy alone as a decompressive procedure is no longer the standard of care. In biomechanical studies, torsional stiffness decreased dramatically, a 2.5% increase in dorsal strain was noted when more than 50% of the facet had been removed at the C5-6 level, and an increase of more than 25% in dorsal strain was seen with a 75% to 100% facetectomy when compared with an intact specimen.³⁵ The decision to use a postoperative orthosis is complex and based on a number of factors, including patient comorbidities such as smoking, diabetes, poor nutritional preoperative status, bone density and, consequently, screw fixation, and, finally, the amount of deformity correction and the iatrogenic stresses placed on the construct.

Cervicothoracic Junction Challenges: Biomechanical Considerations

Five biomechanical studies have looked specifically at fixation at the cervicothoracic junction.³⁶^–⁴⁰ Albert et al.³⁶ compared a lateral mass plating system, a dorsal rod-hook system, and a ventral plate system, showing that all three systems stabilize a dorsal two-column injury similarly. The dorsal fixation systems were found to be stiffer than the others. Chapman³⁷ found that a lateral mass screw fixation system combined with thoracic laminar hooks can provide adequate stabilization at the cervicothoracic junction.

The degree of instability, specifically the number of columns involved, plays an important role in the surgical treatment algorithm for pathology in this region. A biomechanical cadaveric study looking specifically at dorsal stabilization at the cervicothoracic junction offered a comparison of three different fixation devices for stability with dorsal two- and three-column injuries. Kreshak et al.³⁸ evaluated three frequently used dorsal cervical fixation devices, including dorsal rod-screw systems and a dorsal plate-screw system. In extension, there was a significant reduction in construct stiffness with three-column injuries when compared with the fixation of the two-column injuries and the intact spine. Similar results were found with flexion testing. In lateral testing, all the fixation systems were found to be significantly stiffer than the intact spine. The dorsal plate-screw system was found to be weaker in maintaining stability than the screw-rod systems. There are limitations to the plate-screw system, including the need for more surgical precision when placing the screws, difficulty in accounting for variability of the interfacet distances, and difficulty in lining up the predrilled holes in the dorsal plates with the lateral masses. It has been postulated that with a dorsal two-column injury the anterior longitudinal ligament as well as the ventral aspect of the intervertebral disc continue to provide resistance to extension, explaining the findings in this model. Dorsal fixation alone, although it provides a tension band effect in flexion, was not enough to provide stability in the three-column injury model.³⁸

The data also suggest that for a two-column injury at the cervicothoracic junction, there is a trend toward increasing stiffness with prolonged dorsal thoracic fixation, with no differences recorded between the intact state and the dorsal fixation model from C5-T1, with the authors recommending fusion from C6-T2 in order to save cervical fixation points.³⁹ With three-column injuries, dorsal fixation alone results in excessive flexibility in flexion/extension even with instrumentation to T3. Supplementation of such fixation with anterior column instrumentation leads to increased strength of the construct, without any differences in flexion-extension and lateral bending when compared with the intact model, even if the dorsal fixation was stopped at T1. In order to reduce flexibility with axial rotation, however, the instrumentation would have to be extended to T3. For a three-column injury with corpectomy, similar recommendations were made to those for the three-column model with ventral and dorsal instrumentation.³⁹

Another recent study looked at four different dorsal cervicothoracic rod-and-screw constructs, specifically at varying rod diameters and rod connector types, and tested them in flexion bending and axial rotation. There were no significant differences between a dual-diameter rod (3.5 and 5.5 mm) and a solid domino connector extending between two separate rods of the same diameters. A flexible domino connector was found to have similar stiffness but lower ultimate and yield force.⁴⁰

The length of dorsal spinal instrumentation following decompression is a very important consideration that must be based on the patient’s age, the underlying pathology, the length of the laminectomy, and the quality of the bone stock.⁴¹ General indications for spinal fusion, as postulated by White and Panjabi, include (1) restoring clinical stability to a spine with compromised structural integrity, (2) maintaining alignment after deformity correction, (3) preventing formation of a new deformity, and (4) alleviation of pain.⁴² The location of the decompression, especially at the cervicothoracic junction, must be carefully considered, because there is a higher incidence of postoperative kyphotic deformity. The normal sagittal weight-bearing axis lies dorsal to the C2-7 vertebral bodies, lowering the demand on the cervical paraspinal muscles to maintain alignment. Because laminectomy disrupts the posterior column and impairs the tension band construct dorsally, there is a tendency for the weight-bearing axis to shift ventrally. The ventral shift can lead to a kyphotic deformity, which puts the dorsal muscle groups at a significant mechanical disadvantage, requiring extra work to maintain an upright posture and leading to muscle fatigue, pain, and more kyphosis. This is even harder to achieve following dorsal cervical surgery. The use of a postoperative cervical orthosis, while it can help maintain cervical lordosis and alignment, also may lead to further paraspinal muscular atrophy. Pal and Sherk⁴³ have looked at each column individually at C6 and measured the axial loads. They found that 36% of the total load applied rostrally is transmitted through the two anterior columns, and most of the load is transmitted through the articular processes dorsally, again emphasizing the importance of integrity of the posterior column in load sharing.

Inoue et al.⁴⁴ had a series of 36 patients who had surgical decompression of their spinal cord tumors via laminectomy, laminoplasty, or hemilaminectomy. Those patients who underwent C7 laminectomy were found to develop kyphosis at the cervicothoracic junction with a compensatory lordosis of the cervical spine. Those who underwent laminoplasty experienced significantly less deformity.⁴⁴ Steinmetz et al.,⁴⁵ who looked at a large series of 593 total cases, found that laminectomy alone across the cervicothoracic junction led to failure in 38% of the cases. Development of postlaminectomy cervical kyphosis also has been documented in children who have undergone multilevel laminectomy.⁴⁶ De Jonge et al.⁴⁷ retrospectively looked at 76 children who had undergone noninstrumented laminectomy or laminoplasty and/or radiation therapy for malignant spine tumors. Of those, 88% developed postlaminectomy/postradiation spinal deformity. The growing spine also can be adversely affected by radiation therapy, possibly leading to asymmetrical vertebral growth and kyphosis or scoliosis. The use of instrumentation should be strongly considered in pediatric cases requiring destabilization of the dorsal elements at the cervicothoracic junction. In adults, although laminectomy is indicated in patients with cervical lordosis or straightening, it is absolutely contraindicated in the presence of preoperative kyphosis.

Cervicothoracic Junction Tumors: Instrumentation

Dorsal spinal instrumentation added to any decompressive procedure at the cervicothoracic junction should be expected to provide postoperative stability, protecting neurovascular structures from trauma and offering an optimal bed for bony fusion, allowing only micromotion at the fusion sites. These expectations are also the reason why postoperative cervical orthoses are largely unnecessary. Increased fusion rates as a result of instrumentation have been shown in single-level and multilevel discectomies with ventral plating.^48,⁴⁹ There are numerous disadvantages to instrumentation, such as hardware failure, malpositioning, and stress shielding. Imaging artifact from instrumentation should be considered in oncologic cases. The latter factor can inhibit fusion and promote pseudarthrosis, which is why newer dynamically designed plates allow some settling and sharing of the load with the spacers, leading to enhanced fusion as predicted by Wolff’s law, which implies that bone will remodel itself and become stronger when subjected to an external compressive load.

Certain principles must be respected when considering instrumented spinal fusion at the cervicothoracic junction. In general, focal and gradual curves as well as the apical and neutral vertebrae in two orthogonal planes, sagittal and coronal, must be identified. The neutral vertebra usually is the one least angulated and typically is located between the curves. The apical vertebra has the highest degree of segmental angulation at its rostral and caudal ends.⁴¹ A long construct should not terminate at or near an apical vertebra⁴ in either vertical plane, because loads experienced at that point would be highest due to the long moment arm of the axial load in relation to the construct. The apical vertebra is the one associated with the greatest angle (α) between adjacent vertebrae of all vertebrae, in the curve (Fig. 112-5). The second principle that must be respected is that a construct should not end at a junctional level—in this case, at the cervicothoracic junction (Fig. 112-6). It is prone to angular and translational deformation due to the difference in rigidity of the ventrally stabilized thoracic spine by the rib cage and the flexible cervical spine. Other considerations include loss of motion segments, complications related to placement of extra hardware, and cost. The quality of bone and its density also play a role and should be a consideration.⁵⁰

FIGURE 112-5 An apical vertebra (α) occurs at the horizon or apex of a curve, in the sagittal (A) or coronal (B) plane. It is associated with adjacent disc interspaces that have the greatest segmental angulation of all interspaces in the curve, as depicted.

(Reprinted with permission from Benzel EC, editor: Biomechanics of spine stabilization, ed 2, New York, 2001, Thieme.)

FIGURE 112-6 A long implant probably should not terminate at the cervicothoracic junction because this may result in exaggeration of deformity at the terminus of the implant.

(Reprinted with permission from Benzel EC, editor: Biomechanics of spine stabilization, ed 2, New York, 2001, Thieme.)

Historically, dorsal hooks-rods-plate cervical compact Cotrel-Dubousset instrumentation has been described in the context of treatment of cervicothoracic junction pathologies. However, certain limitations with the instrumentation have impeded its continued use.⁵¹ As a hook and rod system, the instrumentation required has to be longer than the levels involved with pathology because of the “claw” configuration, which requires fixation two levels above and two levels below the pathology. It obviously also must be above the levels of the laminectomy, and the hooks have to avoid levels with spinal canal stenosis. CCD hook intrusion into the spinal canal has been shown in a cadaveric study to be, on average, 27% of the diameter of the spinal canal.⁵²

In the subaxial cervical spine, lateral mass screw fixation has been used often. C7 is a transitional vertebra that has characteristics common to both the cervical and the thoracic spine, with the lateral masses transitioning to the size and orientation of the thoracic transverse processes, thus offering a poorer fixation point. Panjabi et al.⁵³ have measured the transverse pedicle thickness and found that it increases from 5.1 mm at C3 to 6.6 mm at C7, which would make it amenable to transpedicular fixation. An et al.⁵⁴ measured the mediolateral and superoinferior outer pedicle diameter as 6.9 mm and 7.5 mm, respectively. The average inner diameter was 5.18 mm. The length of the pedicle was found to be 9.1 mm, with a medial angulation of 34 degrees. The recommended pedicle entry point for the C7 pedicle was 1 mm inferior to the mid-portion of the facet joint with 25 to 30 degrees of medial angulation and perpendicular to the dorsal arch.

Pedicle screw fixation, while very popular in the thoracic and lumbar spine with superior fixation characteristics, has only recently become more widely used in the cervical spine. Anatomic studies have shown the pedicle width of C3 to average 4.9 mm in men and 4.5 mm in women, with the minimum reported width of 3 mm. The C4 pedicle averages 4.7 mm in men and 4.6 mm in women, with a minimum reported width of 3.1 mm.⁵⁵ Because of this variability, measurements should be made on a case-by-case basis using CT imaging.

Several studies have specifically examined the use of distal subaxial cervical pedicle screw fixation and compared it with lateral mass fixation. The lateral mass anatomy and its particular relationship to nearby neurovascular structures have been precisely established by anatomic studies. Xu et al.⁵⁶ found that for C3-7, the average distance from the superficial, dorsal center of the lateral mass to the nerve root superiorly was 5.7 mm ± 1.5 mm. Inferiorly, the average distance was found to be 5.5 mm ± 0.8 mm. The average distance to the spinal sac from that same point was found to be 9.2 mm ± 1.4 mm. The average medial angle of the nerve root was 76.3 degrees. Closer to the cervicothoracic junction, at the C6 level, the transverse foramen was found to be located directly ventral to the dorsal midpoint of the lateral mass. These findings suggested that this point would be safe for initiation of screw insertion in the lateral masses.⁵⁶

Pedicle screw fixation in the subaxial spine has been found to be safer with direct palpation of the medial pedicle wall. Albert et al.⁵⁷ retrospectively looked at a group of 21 patients who had pedicle screws placed at C7 with the help of a small laminoforaminotomy at C7 and by palpating the pedicle with a right-angle nerve hook. There were no neurologic complications related to screw placement and no failure of fixation related to the use of the pedicles at 1-year follow-up.⁵⁷

In the thoracic spine, the transverse processes do not provide adequate fixation. The upper thoracic spine pedicles at T1 and T2 are larger than those found in the midthoracic spine.³⁷ On average, the width of the pedicles decreases caudally from 7.8 mm to 4.4 mm. The pedicle axis projects about 7 to 8 mm medial from the lateral edge of the superior facet and about 3 to 4 mm superior to the midaxis of the transverse process.⁶ An et al.⁵⁴ recommended an entry point for the T1 and T2 thoracic pedicles at the midpoint 1 mm below the facet joint, whereas Mazel et al.⁶ preferred 10 to 20 degrees of caudal angulation and 5 to 10 degrees of medial angulation. Mazel’s entry point for screw insertion at the upper thoracic level is located at the intersection of the transverse process axis and the midline of the inferior facet. The more caudal the fixation, the less caudal and medial is the angulation of the trajectories of the pedicle screws. Some have recommended upper thoracic pedicle screw fixation with the help of laminotomy and by directly palpating the external boundaries of the pedicle to reduce pedicle wall violation.⁵⁸ However, most authors avoid using this technique due to the added time and risks; instead, they use instruments at the level of the thoracic spinal canal.^6,⁵⁹ Kim et al.⁵⁹ performed a retrospective analysis of 394 patients who underwent dorsal instrumentation with a total of 3204 transpedicular screws placed using a freehand technique by two surgeons over a 10-year period. Their review showed a 6.2% violation of the pedicle cortex, with 1.7% violating the medial wall.⁵⁹ In this study, there were no neurologic, vascular, or any other soft tissue complications. The low rate of misplacement of transpedicular screws by freehand technique in the thoracic spine, however, should not give an inexperienced surgeon a false sense of security. It is still recommended to lower that risk by closely examining preoperative imaging studies and looking for any anatomic abnormalities such as unusually narrow pedicles.

Abumi et al.⁶⁰ experienced a 7% misplacement rate out of 669 cervical pedicle screws with conventional screw insertion techniques. Due to the high risks associated with transpedicular fixation at the cervicothoracic junction, including iatrogenic damage to the spinal cord, nerve roots, or vertebral artery,^55,⁶⁰ computer-assisted systems were developed to minimize those risks. Richter et al.⁶¹ dorsally instrumented 19 patients with computer assistance, placing 31 cervical pedicle screws and 10 high-thoracic pedicle screws. Postoperative CT scanning with multiplanar reconstruction for each screw did not demonstrate any misplacement greater than 2 mm or any screw-related injury to vascular, neurogenic, or bony structures. They strongly recommend computer-assisted instrumentation, especially at the C3-6 level. Certain disadvantages must be considered when using a computer-assisted instrumentation system at the cervicothoracic junction, however. Computerized systems are prone to errors and system crashes due to hardware, software, or human failure, and the surgeon must be well versed in the more traditional fixation methods. Difficulties can be encountered with registration of instrumented vertebrae where surface matching can be thrown off by small or indistinguishable bony landmarks, because most computer systems have been developed for thoracic and lumbar systems. According to Richter, “the use of a computerized assisted system for dorsal cervical spine surgery will make a good spine surgeon even better, but it can never make a bad spine surgeon a good one.”⁶¹

Other methods of dorsal fixation at the cervicothoracic junction that have been closely examined, in addition to screws, include traditional wires and hooks. With tumors of the dorsal elements at the junction it may be difficult to use spinous process wiring or laminar hooks, which can lead to spinal canal stenosis or iatrogenic spinal cord injury.⁶² Rhee et al.⁶² found no significant increase in normalized stiffness in any of the lateral mass constructs with wire augmentation that were examined. The C7 pedicle screw fixation constructs demonstrated significantly higher normalized stiffness than any of the other constructs, which included lateral mass screw fixation at C7, C6 and C7 lateral mass screws, and these constructs augmented with triple wiring. There was no predictable difference in stiffness between any of the constructs while in extension. Flexion, right and left lateral bending, and axial torsion showed superior results with the C7 pedicle construct when compared with the lateral mass screws. The addition of another lateral mass screw level at C6 did show increased stiffness comparable to the C7 pedicle screws.⁶² In another study, Ulrich et al.⁶³ showed that sublaminar wiring is inadequate, even with supplemental ventral plate fixation, in stabilizing torsional and translational loads. In an animal study, C7-T1 pedicle screw fixation was as effective as a ventral plate and dorsal wiring in stabilizing a three-column injury.⁶⁴ Dorsal wiring is adequate only in offering stability in flexion testing,⁶⁵ functioning as a supplemental tension band construct dorsally.

Outcomes and Complications Following Surgical Treatment

Ventral and dorsal approaches to the cervicothoracic junction carry specific postoperative complications, which have been discussed with the individual approaches. Recurrent laryngeal nerve injuries may be seen with low ventral cervical or cervicothoracic junction approaches, leading to unilateral vocal cord palsy with possible mild dysphagia and dysphonia. If persistent, treatment may be with medialization thyroplasty. When both recurrent largyngeal nerves are injured, vocal cord paralysis can occur, requiring permanent gastrostomy and tracheostomy.¹⁷ Higher approaches to the ventral cervical spine carry the risks of superior laryngeal and hypoglossal nerve injury because of their approximation to the superior thyroid artery.⁶⁶

Ventral cervicothoracic junction surgery and its complications also have been documented by Boockvar et al.⁶⁷ in a group of 14 patients, who were followed for a period of 21 months. Five patients (36%) had graft/plate failure, which needed revision, and one patient had a recurrent laryngeal nerve palsy. Surgical complications were associated with male sex, multiple levels of involvement, the use of allograft compared with autograft, and previous surgery. The authors concluded that supplemental fixation is needed when dealing with cervicothoracic junction pathology.

In An et al.’s ³ series of 15 patients with tumors, 12 presented with neurologic involvement, with 8 patients having incomplete cord syndromes and 4 having nerve root symptoms.³ Most of the patients improved neurologically by at least 1 or 2 Frankel grades. One patient with giant cell tumor had a recurrence that required a repeat ventral procedure. Three patients died because of tumor progression: breast metastasis at T1-2, myeloma at T1, and lung tumor at T2-3.

Bilksy et al.¹⁷ reported a series of 41 patients with subaxial cervical tumors who underwent spinal reconstruction: 12 patients from a ventral approach, 13 patients from a dorsal approach, and 16 patients from a combined approach. Thirty-three of the 41 tumors were metastatic, and 8 were primary (2 chordomas and 6 sarcomas). Twenty-five percent of the patients had surgery-related complications, including bilateral vocal cord paralysis after a right-sided trapdoor approach at the C6-7 level with the left side presumed present preoperatively, a unilateral recurrent laryngeal nerve palsy after thyroplasty, and one CSF leak. In this study, three patients experienced complications with the cervicothoracic junction instrumentation: two patients experienced pull-out of the dorsal plate screw instrumentation (with no ventral fixation performed), and the third experienced failure of the ventral graft following corpectomy of C6 and fusion with plate. No failures were noted in the patients who underwent cervicothoracic instrumentation with screws and rods. No wound issues arose following the preoperative irradiation, even though dissection sometimes was more problematic. Similar instrumentation failures were described by Lenoir et al.,⁶⁸ who looked at 30 patients who underwent surgery for destabilizing injuries at the cervicothoracic junction. One patient had mobilization of the cervical portion of a plate and another had a thoracic screw loosen. There was no direct fixation of the anterior column in these two cases.

Mazel et al.⁶ retrospectively reviewed 32 patients treated for cervicothoracic junction tumors who underwent dorsal fixation with cervical lateral mass fixation and thoracic transpedicular screw fixation. Nineteen of the patients with Pancoast tumors had en bloc resections and, therefore, total and partial vertebrectomies with destabilization of the ventral two columns. Thirteen of the patients had only dorsal palliative laminectomy and cervicothoracic fixation. Two of the 32 cases failed, with both of those patients having partial or total vertebrectomies, due to the limitation of fusion length. Mazel et al. concluded that in cases of total or partial vertebrectomies, dorsal instrumentation must extend three levels above and three levels below the site of pathology. Their group had positive results with the use of autologous fibula strut graft for multilevel corpectomies and femoral head bank bone for single-level vertebrectomy.

Steinmetz et al.⁴⁵ looked at outcomes following cervicothoracic region surgical fixation by reviewing 593 patients treated over a 5-year period at the Department of Neurosurgery at the Cleveland Clinic Foundation who were followed for a mean of 20 months. Treatment failed in 14 patients, with the failures statistically associated with laminectomy and multilevel ventral corpectomies, not supplemented by dorsal fixation, as well as histories of prior cervical surgery, surgery for correction of deformity (P = .033), and tobacco use (P = .019). Three patients underwent decompression and lateral mass fixation ending at, but not crossing, the cervicothoracic junction. Two patients underwent decompression and ventral plate fixation to, but not crossing, the junction. More specifically, three out of eight patients (37.5% of patients, P = .001) who underwent laminectomy that crossed the junction without supplemental fixation and/or fusion failed. Three of 18 patients (16.7%) who underwent two- or three-level corpectomy with fixation/fusion to T1 only also failed. Finally, three of the patients experienced treatment failure after dorsal instrumentation across the cervicothoracic junction. One of those patients, who had osteoporosis, had pedicle screw fixation ending at T1. Another patient, with Ehlers-Danlos syndrome, had undergone occiput-to-T4 instrumented fixation for correction of severe swan-neck deformity, and the tapered rods fractured. Finally, a patient who underwent dorsal fusion from C4-T2 for breast metastasis also experienced failure. Interestingly, no failures occurred in association with a combined ventral-dorsal fixation ending either at C7 or below. However, with the recommendation being made to consider extension of fixation to T1 or T2, a trend toward failure was noted in cases in which the dorsal fixation was terminated at C7. Similar findings were noted by Boockvar et al.,⁶⁷ who recognized that prior cervical surgery was associated with the failure of ventral surgery at the cervicothoracic junction.