Acta Neurochir (Wien) (1998) 140: 737±744

Acta Neurochirurgica > Springer-Verlag 1998 Printed in Austria

Thoracoscopic Approaches to the Thoracic Spine M. Visocchi, R. Masferrer, V. K. H. Sonntag, and C. A. Dickman Division of Neurological Surgery, Barrow Neurological Institute, Mercy Healthcare Arizona, Phoenix, Arizona

Summary Microsurgical approaches for the treatment of pathology located in the ventral thoracic spine using video-assisted thoracic surgery (VATS) allow neurosurgeons to access the disc spaces, vertebral bodies, paravertebral soft tissues, spinal cord, spinal nerves, and sympathetic chain with minimally invasive surgery. This has been associated with substantial clinical bene®ts including reduced postoperative pain, lower complication rates and shorter recovery times when compared with standard thoracotomy techniques. This article describes the experience at our institution with VATS for discectomy (20 cases), corpectomy and spinal reconstruction (8 cases), thoracic sympathectomy (3 cases), and nerve sheath tumor removal (1 case). The technique can be mastered but requires surgeons to learn the new psychomotor skills needed to perform endoscopic spine surgery. The learning curve is steep. Special training in instructional seminars, surgical skill laboratories, and clinical preceptorships is needed before this surgical approach can be used clinically to treat spinal pathology. VATS has signi®cant advantages compared to standard thoracotomy, including reduced incisional pain and avoidance of the postthoracotomy pain syndrome. If intercostal neuralgia develops postoperatively, it is milder and usually transient compared to the pain associated with standard thoracotomy. Better cosmetic outcomes, earlier mobilization, and faster recovery are added bene®ts. The surgical techniques are relatively new for neurosurgeons and require dedicated practice to master them. Once the surgical skills are perfected, VATS is feasible for spinal pathology and can be performed safely and e¨ectively. Keywords: Corpectomy; discectomy fusion; spinal instrumentation; spinal reconstruction; sympathectomy; thoracoscopy.

Introduction The development of video thoracoscopy, or videoassisted thoracic surgery (VATS), potentially represents one of the most signi®cant developments in thoracic surgery in the past decade. In just 6 years, VATS has been accepted as an important clinical tool in the developing ®eld of ``minimally invasive surgery.'' Microsurgical techniques can be combined with this new approach for the neurosurgical treatment of pathology of the ventral thoracic spine. The disc spaces, vertebral bodies, paraspinal tissues, spinal

cord, thoracic roots, and sympathetic chain can be easily reached by this approach. This ease of access has been associated with the clinical bene®ts of minimal incisional surgery, including reduced postoperative pain, reduced complication rates, and faster recovery times. This article describes our experience with VATS for discectomy, corpectomy with spinal reconstruction, thoracic sympathectomy, vertebral biopsy, and thoracic schwannoma resection. Patients and Methods The materials included in this study come from a retrospective review of the medical records of 32 patients operated consecutively at our institution between 1994 and 1997. The group includes 31 patients who presented with herniated thoracic discs (20 cases), vertebral body lesions (8 cases), sympathetic dysfunction (3 cases), and thoracic nerve sheath tumor resection (Table 1). Seventeen of these patients have been reported elsewhere [5]. All the patients had complete neurological examinations and chest radiographs after admission. All the patients, except the three with sympathetic dysfunction, had neuroimaging studies of the thoracic spine with computerized tomography (CT), CT myelography, and/ or magnetic resonance (MR) imaging. The primary indications for surgery were 1) spinal cord compression, 2) intractable radicular pain from nerve root compression, 3) progressive spinal deformity with spinal canal compromise, 4) reconstruction and ®xation of the unstable thoracic spine with bone grafting and instrumentation, 5) sympathetic hyperactivity, and 6) tumor removal. Pre- and postoperatively, patients with myelopathy were graded using the Frankel scale [9]. All surgical specimens were sent for neuropathological analysis. All patients underwent postoperative chest radiographs. Plain spinal radiographs as well as CT and/or MR imaging were obtained from all patients, except the three who underwent sympathectomies.

Surgical Technique All the instruments used during VATS are spinal instruments modi®ed for this purpose. The tips of the instruments need to reach

738

M. Visocchi et al.

Table 1. Thoracoscopic Spine Surgery Mean age (range) Thoracoscopic discectomy

Males/females Symptoms

53 (14±79) 6/14

Thoracoscopic corpectomy 61 (38±77) 6/2 Sympathectomy

39 (23±41) 1/2

Schwannoma

52 (0)

0/1

Levels accessed

Myelopathy (n ˆ 7) Radiculopathy (n ˆ 4) both (n ˆ 9) Myelopathy (n ˆ 2) Radiculopathy (n ˆ 1) both (n ˆ 5) Hyperhidrosis (n ˆ 2) RSD (n ˆ 1) Dyspnea

Mean blood loss Mean surgical time (range) (cc) (range) (min)

T3±T4 to T10±T11 460 (124±1500)

225 (160±330)

T3±T4 to T12±L1

857 (200±1500)

541 (180±740)

T1 to T4

100 (50±150)

160 (90±210)

T2±T3

200

240

rsd Re¯ex sympathetic dystrophy.

the epidural space percutaneously through the thoracoscopy portals and therefore have longer shafts than standard neurosurgical instruments. The tips of the tools were angled and etched with markings at 1-cm intervals to judge the depth and position of the tools during the dissection. As in any other endoscopic procedure, the instruments are guided under direct endoscopic visualization; the operative images are projected on a television screen. Positioning All patients underwent a standardized thoracoscopic surgical opening. Patients were placed on the operating table in a lateral decubitus position with an axillary roll under the dependant axilla. The lower extremities were ¯exed, and the knees and other bony prominences were padded with pillows or foam padding. A double-lumen endotracheal tube was used to de¯ate one lung temporarily during the operation to facilitate exposure of the vertebral column and paravertebral soft tissues. A right or left-sided approach was used, depending on the location of the pathology. When possible, a rightsided approach was preferred in the middle and upper thoracic spine because more working space is available over the spinal surface behind the azygos vein compared with that behind the aorta. The left side was preferred for the lower thoracic spine because at this level the liver causes the right hemidiaphragm to rise and partially obstruct the working space. A team approach was used, combining a cardiothoracic surgeon and a neurosurgeon both with experience in VATS. Intraoperative Studies Intraoperative somatosensory evoked potentials (SEPs) were recorded to monitor spinal cord function in all patients. The peroneal or tibial nerves were stimulated with 300 ms pulses at a rate of 6.3/sec with an initial current of 2 to 15 mA (approximately two times the motor threshold) for 500 to 4000 repetitions. Recording ®lters were set at 30 to 1500 Hz. Recording electrodes were placed over the cervical region and contralateral hemispheres (Xomed Treace, Nicolet Viking IV, Madison WI). Incisions and Endoscopy set-up A potential thoracotomy incision was marked on the skin, and the scapula on the ipsilateral side was outlined in case it became necessary to convert the thoracoscopic procedure into an open thoracotomy. Thoracoscopy portals were inserted in the intercostal

Fig. 1. Thoracoscopic spine surgery is performed with the patient in a lateral decubitus position. Three or four narrow, ¯exible portals are inserted through the intercostal spaces into the chest after the ipsilateral lung has been de¯ated. The view of the surgical procedure is transmitted from the endoscope to video monitors in the surgical suite. With permission from Barrow Neurological Institute

spaces positioned in the anterior, middle, and posterior axillary lines. Three or four separate skin incisions approximately 15-mm long were marked on the skin in the preselected areas (Fig. 1). The skin was then incised with a No. 10 scalpel blade, and the intercostal muscles and parietal pleura were penetrated with a tissue clamp similar to the techniques used for the placement of a chest tube. The portals were inserted with a blunt-tipped trocar to guide them into the chest (Fig. 2). A 0 - or 30 -angled, 1-cm diameter, rigid rod-lens endoscope was used for illumination, magni®cation, and visualization of the contents of thoracic cavity and mediastinum. The insertion of subsequent portals needed for lung retraction, suction, irrigation, and dissection instruments was visualized directly with the endoscope. At this point, all patients were rotated anteriorly to allow the atelectatic lung to fall away from the surface of the spine.

739

Thoracoscopic Approaches to the Thoracic Spine

a a

b

Fig. 2. (a) A 15-mm incision is made in the skin, and the soft tissue is dissected to provide a tunnel over the superior surface of the ribs into the chest, similar to the way a chest tube is inserted. (b) A ¯exible portal is inserted and the trocar is removed. With permission from Barrow Neurological Institute Localization The apical (®rst) ribs are identi®ed and subsequent ribs are counted caudally to identify the appropriate vertebral level. The appropriate level is then con®rmed by placing a long tool or a needle through the portal into the disc space and obtaining plain radiographs or ¯uoroscopic images. Lateral views are seldom adequate for localization, but anteroposterior views are routinely used to con®rm the appropriate level intraoperatively. Dissection The parietal pleura over the level of the spinal pathology is incised and mobilized to expose the spine. This step is performed with electrocautery, grasping forceps, and periosteal elevators. If vascular control is needed, the segmental arteries and veins are identi®ed at the midportion of the vertebral body halfway between each of the disc spaces. The vessels are mobilized and ligated. Hemoclips are applied and the vessels are transected. If the spinal canal is to be accessed to perform a decompression, the rib head and proximal 2 to 3 cm of rib adjacent to the area of pathology are removed to expose the pedicles of the a¨ected vertebra. The rib head and proximal rib can be resected with Cobb periosteal elevators, curved curettes, and Kerrison rongeurs. Rib cutters and drills also can be used. The pedicles are key landmarks for identifying the lateral margin of the spinal canal. The ipsilateral pedicle is removed after resection of the rib using either Kerrison rongeurs or the Midas Rex8 (Midas Rex8 Pneumatic Tools, Inc., Fort Worth, TX) drill to expose the dura. After the dura has been exposed and identi®ed, disc or vertebral pathology a¨ecting the spinal canal can be approached. Discectomy The general exposure described above was used for the 20 patients with thoracic discs. If the disc herniation was at the level of the disc space, only the superior part of the pedicle was removed. If the disc had migrated caudally, the entire pedicle was resected. After the pedicle was resected and the dura mater was identi®ed in the spinal canal, the annulus ®brosis was incised with a Cobb peri-

b Fig. 3. (a) The rib head, proximal rib, and pedicle are removed to expose the dura adjacent to the herniated thoracic disc. (b) A cavity is made in the dorsal edge of the disc space and in the vertebral bodies to provide a working space or a cavity. Microsurgical tools are then used to deliver the herniated disc material away from the spinal cord into the cavity. Only the tips of the tools are used on the edges of the compressive pathology, which avoids placing any tools within the compromised epidural space. With permission from Barrow Neurological Institute

osteal elevator. A cavity was created in the dorsal disc space and vertebral body to create a working space to avoid placing tools into the compromised epidural space. The disc material was then curetted into the cavity and removed with pituitary rongeurs. Calci®ed disc material, osteophytes, and vertebral end plates were removed with curettes, microdissectors, drills, or disc rongeurs (Fig. 3). The depth of decompression was assessed by direct visualization with the thoracoscope, intraoperative radiographs, or ¯uoroscopic images. Corpectomy For the eight patients undergoing corpectomies, the pedicles and rib heads were removed to expose the dura. The disc above and below the corpectomy were incised to de®ne the margins of the corpectomy. A large cavity was created in the center of the vertebral body with osteotomes, drills, curettes, and/or rongeurs. The cavity in the a¨ected vertebral body was expanded until the posterior cortex was reached. The posterior cortex of the vertebral body and the posterior longitudinal ligament were removed in a piecemeal fashion with angled curettes and Kerrison rongeurs. Any material

740

M. Visocchi et al.

a

b

c

d

Fig. 4. Thoracoscopic corpectomy. (a) The proximal ribs are removed adjacent to the pathology, and the segmental vessels are ligated. (b) The pedicles are removed for direct visualization of the dura, and then a large cavity is created in the vertebral body. (c) The posterior longitudinal ligament and pathology compressing the spinal cord are removed with microsurgical tools. The material is moved away from the spinal cord into the cavity in the vertebral body. (d) A bone graft is placed after decorticating the end plates of the adjacent normal vertebrae. With permission from Barrow Neurological Institute

encroaching into the spinal canal can be removed safely by dissecting it away from the spinal cord into the cavity created in the vertebral body. After decompression of the spinal canal, the end plates of the vertebral bodies above and below the corpectomy level were denuded of the overlying cartilage and partially decorticated with angled curettes or a drill. The bone graft (humerus allograft or iliac crest strut autograft) was then inserted into the corpectomy defect (Fig. 4). Instrumentation was performed with a plate and screw technique (Z-plate, Sofamor Danek, Memphis, TN). The plate, screws, and bolts were inserted through the chest wall portals in the appropriate coaxial trajectories under ¯uoroscopic and endoscopic guidance. Sympathectomy One-stage bilateral thoracic sympathectomies (T2±T4 ganglionectomies) were performed in three patients (Fig. 5). The parietal pleura overlying the sympathetic trunk at the T3±T4 interspace was tented using a grasping forceps and then incised with endoscopic scissors. The pleural dissection was carried in a cephalad direction to the stellate ganglion overlying the ®rst rib. After the pleural incision was completed, the sympathetic chain was removed using sharp dissection. The sympathetic chain is usually super®cial to the vessels on the surface of the spine and often can be removed without sacri®cing any vessels. Occasionally, however, it may be necessary to ligate vessels to expose and remove the sympathetic chain. The sym-

Fig. 5. Thoracoscopic sympathectomies are simple to perform. The sympathetic chain directly overlays the rib heads just beneath the parietal pleura. With permission from Barrow Neurological Institute

741

Thoracoscopic Approaches to the Thoracic Spine pathetic trunk was transected below the level of the T4 ganglion and then mobilized from caudal to cephalad. The rami to each ganglion were cauterized and divided until the inferior border of the stellate ganglion was reached. The sympathetic trunk was transected at the level of T1, preserving the stellate ganglion. Nerve Sheath Tumor Resection A large cystic schwannoma arising from the widened T2±T3 neural foramen and extending into the apex of the right hemithorax was resected in one patient. Using the standard thoracoscopic exposure previously described, the tumor was identi®ed and a needle was inserted into the cystic component of the tumor aspirating 20 ml of xanthochromic ¯uid. After this internal decompression was completed, the mass was mobilized. The pleura over the tumor was incised and opened widely. The vascular supply to the tumor was identi®ed, coagulated, and transected. The tumor capsule was then dissected from the chest wall and mobilized away from the spine. The head of the rib and the pedicle of the vertebra caudal to the enlarged foramen were removed, and the epidural space and proximal nerve root were identi®ed. The nerve root was mobilized and ligated with an endoscopic loop ligature. The nerve root was transected distal to the ligature, and the entire tumor was removed. Closure At the end of the thoracoscopic surgical procedures, extensive irrigation was used to remove all tissue debris from the thoracic cavity. The lung and mediastinal contents were visualized with the endoscope to rule out areas of hemorrhage or air leaks. One or two chest tubes were inserted into the thoracic cavity through the endoscopic portals. One of the chest tubes was placed at the apex to facilitate lung rein¯ation and the other posteroinferiorly to drain ¯uid. The chest tubes were anchored to the skin with heavy silk sutures and connected to suction with a pressure of ÿ20 cm H2 O.

Results Intraoperatively, no patient experienced a change in SEPs. All neuropathological analyses con®rmed preoperative diagnoses. All patients except one were discharged within the ®rst week (3 to 5 days in most cases). Discectomy The discectomy procedures were successful in removing the herniated discs in all cases. Seventeen of the decompressions were completed during a single operative approach. Three patients required a second surgical approach. One patient had a retained disc fragment (intradural extension of an ossi®ed disc). Intraoperatively, a soft disc fragment migrated within the spinal canal of another patient, and one operation was performed at the wrong level. One patient with severe rigid kyphotic deformities underwent a two-level VATS discectomy, anterior re-

lease, and interbody fusion followed by an open posterior thoracic instrumentation and fusion. Sensory and motor de®cits improved signi®cantly in 89% of the myelopathic patients who underwent VATS discectomies. One patient experienced transient neurological worsening from a Frankel grade C to Frankel grade B and then recovered to the preoperative grade (Frankel grade C). The thoracic radicular pain improved or resolved completely in 13 patients. Two patients had residual radicular pain. The length of surgery ranged from 160 to 330 minutes (mean, 225 min). Blood loss ranged from 124 ml to 1,500 ml (mean, 460 ml). Six patients had mild, minor complications that resolved completely: postoperative pleural e¨usions (n ˆ 3), pneumonia (n ˆ 1), atelectasis (n ˆ 1), pneumothorax (n ˆ 1), and transient postoperative intercostal neuralgia (n ˆ 1).

Corpectomy The thorascopic corpectomies were performed to treat vertebral osteomyelitis (n ˆ 3); metastatic tumor involving the vertebrae (n ˆ 1); vertebral body fractures (n ˆ 2); and transdural, large, densely calci®ed discs (n ˆ 2). Of these eight patients, three had partial corpectomies without reconstruction while ®ve required reconstructions. Of the ®ve latter patients, all had anterior bone grafting and two underwent bone grafting and anterior screw-plate instrumentation. In three patients, a second-stage procedure for posterior instrumentation and bone grafting was performed because of concurrent posterior thoracic instability or because osteoporotic bone precluded anterior instrumentation. In one case, the vertebral body was biopsied and antibiotic treatment followed. Myelopathies and/or radicular pain improved in all patients who underwent corpectomies. The length of surgery ranged from 180 minutes to 740 minutes (mean, 541 min). Blood loss ranged from 200 ml to 1500 ml (mean, 857 ml). One patient had a screw back out and required endoscopic removal of the screw. One patient had pleural e¨usion that resolved completely, and one patient had pneumonia that resolved with medical therapy. Two patients with transdural discs were treated prophylactically with a lumboperitoneal shunt. The mean follow up of these patients was 13.9 months (range, 8 to 28 months).

742

Sympathectomy Sympathectomies were performed for hyperhidrosis in two patients and for re¯ex sympathetic dystrophy (RSD) in one patient. These patients recovered completely from their preoperative symptoms. Two patients experienced transient mild intercostal neuralgia, and a transient compensatory hyperhidrosis occurred in one patient. The length of surgery ranged from 90 to 210 minutes (mean, 187 min). Blood loss ranged from 50 ml to 150 ml. The follow-up for these patients ranged from 4 to 6 months. Tumor In the patient with the intrathoracic schwannoma, the length of surgery was 240 minutes. Blood loss was 200 ml. She was discharged home 2 days after surgery and has had no complications. At a 12-month follow up examination, she had recovered completely and had no residual or recurrent tumor. Discussion Surgeons have ventured into the thoracic spine since the 1600s. The results of early thoracic spine surgery were fraught with complications and failure. With the advent of anesthesia and antibiotics, surgical outcomes associated with the thoracic region improved but complications still tend to have disastrous results. Several unique characteristics of the thoracic spine and thoracic spinal cord contribute to the high risk of complications after surgical manipulation. Despite the relatively small size of the spinal cord in this region, the ratio of the spinal cord to the spinal canal is larger than elsewhere in the spine. The blood supply in this region may be tenuous and variable, particularly between T4 and T9 known as the ``watershed zone.'' This watershed area is vulnerable because the direct blood supply from the radicular and intramedullary circulations is limited and there is a paucity of vascularization at the microcirculatory level. The use of laminectomies for the treatment of thoracic disc herniations has been associated with extremely poor outcomes and is to be avoided [15]. Transpedicular and lateral extracavitary approaches (LEA) were developed to circumvent the limitations associated with a standard dorsal (laminectomy) approach. Both thoracic discs and masses encroaching into the spinal canal can be reached by LEA, some-

M. Visocchi et al.

what decreasing the risk of injury to the spinal cord. Critics of this approach are quick to point out that the ability to protect the spinal cord is still limited and that the technique is associated with high rates of blood loss, long operating times, and considerable postoperative pain. The standard transthoracic approach to the spine was ®rst used to manage thoracic disc disease in 1969, independently by Perot and Muno [13] and by Ransaho¨ and colleagues [14]. Its main advantage is the anterior exposure that eliminates manipulation of the spinal cord. With this approach, centrally located herniated discs or osteophytes are accessible. It o¨ers a superior exposure for multiple disc herniations, and interbody grafts can be placed easily [2, 3, 7, 15]. The approach also has disadvantages. A second surgeon is needed for the exposure, and there is a risk of injury to the great vessels or mediastinum. Patients also can develop a postthoracotomy syndrome with its painful consequences. During the early 1990s thoracoscopy was introduced as a diagnostic tool for the evaluation of pleural disease [2]. Today, many thoracic procedures previously performed via thoracotomy are routinely and preferentially performed using thoracoscopy. Several clinical studies have found signi®cant advantages to thoracoscopy when compared with thoracotomy for the treatment of thoracic pathology [8, 10]. Daniel Rosenthal (Frankfurt, Germany) and John Regan (United States) independently developed thoracoscopic spinal surgery in 1992. The ®rst published report on this subject by Mack, Regan, Bobechko, and Acu¨ appeared in the Annals of Thoracic Surgery in 1993 [11]. These authors described 10 patients with diverse spinal pathology who were treated e¨ectively by a thoracoscopic approach to the spine without major complications. There are no signi®cant di¨erences between VATS and LEA in terms of operative times, blood loss, postoperative pain, and length of hospitalization [5]. VATS, however, is superior to LEA in that it provides better visualization and access to the ventral spinal cord, therefore allowing a more complete decompression of midline pathology. When comparing VATS with standard thoracotomy, the endoscopic approach de®nitely fares better. VATS is associated with less blood loss, less chest tube drainage, fewer complications, less use of pain medication, and shorter lengths of hospitalization [5]. Intercostal pain and postthoracotomy

743

Thoracoscopic Approaches to the Thoracic Spine

syndrome are more common and severe after thoracotomy than after VATS. Biomechanically, the VATS approach, like any other anterior approach to the spine for reconstruction and/or instrumentation, signi®cantly increases the axial load-bearing capabilities of the spine. More than 80% of the compressive loads of the spine are borne by the anterior and middle columns [6]. However, in the case of a prior laminectomy, severe osteoporosis, or large kyphotic deformities, the anterior approach alone may be insu½cient to achieve reduction and ®xation. A posterior approach to the spine may be needed in conjunction with the VATS corpectomy, grafting and instrumentation. With the VATS approach, the neurosurgeon is able to access only the anterior and anterolateral aspects of the vertebrae and spinal canal. The posterior elements, the contralateral pedicle, or contralateral transverse process cannot be exposed. Althought VATS has not been routinely applied for intradural procedures, we have completely removed two intradural discs, closing the dural opening with ®brin glue and diverting the cerebrospinal ¯uid with a lumboperitoneal shunt. We are currently developing endoscopic techniques for dural suturing and closure in our laboratory. Regarding sympathectomies, the dorsal paramedian approach to the upper thoracic sympathetic trunk described by Cloward is the technique most commonly used by neurosurgeons for the treatment of hyperhidrosis and RSD. In comparison, thoracoscopy has a low complication rate when compared to other surgical methods [1, 4]. VATS allows direct visualization for the excision of the sympathetic trunk caudal to the stellate ganglion. Success rates between 87% and 100% have been reported with VATS for the treatment of palmar hyperhidrosis and RSD with a low morbidity rate and less postoperative pain [12]. Compared with other approaches, length of hospitalization is also shorter for VATS sympathectomies [12]. Although our small patient population limits our experience, the results thus far suggest that VATS sympathectomies are well-tolerated and cost-e¨ective alternatives to upper dorsal thoracic sympathectomy. Conclusion VATS has tremendous potential for improving patients' comfort, improving cosmetic results, and for shortening the length of hospitalization and recovery times. The morbidity is lower than that of tho-

racotomy for providing full, direct access to the entire ventral surface of the thoracic spine and spinal cord. The techniques for this new neurosurgical approach to the thoracic spine require that the surgeon learn new psychomotor skills, which must be mastered with adequate instructional seminars, surgical skill laboratories, and clinical preceptorship programs. References 1. Adar R, Kurchin A, Zweig A, Mozes M (1977) Palmar hyperhidrosis and its surgical treatment: A report of 100 cases. Ann Surg 186: 34±41 2. Chin LS, Black KL, Ho¨ JT (1987) Multiple thoracic disc herniations. Case report. J Neurosurg 66: 290±292 3. Chou SN, Seljeskog EL (1973) Alternative surgical approaches to the thoracic spine. Clin Neurosurg 20: 306±321 4. Cloward RB (1969) Hyperhydrosis. J Neurosurg 30: 545±551 5. Dickman CA, Karahalios DG (1996) Thoracoscopic spinal surgery. Clin Neurosurg 43: 392±422 6. Dickman CA, Mican C (1996) Thoracoscopic approaches for the treatment of anterior thoracic spinal pathology. BNI Quarterly 12: 4±19 7. Ebersold MJ (1986) Surgical management of thoracic disk disease. Contemp Neurosurg 7: 1±6 8. Ferson PF, Landreneau RJ, Dowling RD, Hazelrigg SR, Ritter P, Nunchuck S, et al (1993) Comparison of open versus thoracoscopic lung biopsy for di¨use in®ltrative pulmonary disease. J Thorac Cardiovasc Surg 106: 194±199 9. Frankel HL, Hancock DO, Hyslop G, Melzak J, Michaelis LS, Unger GH, et al (1969) The value of postural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia I. Paraplegia 7: 179±192 10. Landreneau RJ, Mack MJ, Hazelrigg SR, Dowling RD, Acu¨ TE, Magee MJ, et al (1992) Video-assisted thoracic surgery: Basic technical concepts and intercostal approach strategies. Ann Thorac Surg 54: 800±807 11. Mack MJ, Regan JJ, Bobechko WP, Acu¨ TE (1993) Application of thoracoscopy for diseases of the spine. Ann Thorac Surg 56: 736±738 12. Malone PS, Cameron AE, Rennie JA (1986) Endoscopic thoracic sympathectomy in the treatment of upper limb hyperhidrosis. Ann R Coll Surg Engl 68: 93±94 13. Perot Pl, Jr., Munro DD (1969) Transthoracic removal of midline thoracic disc protrusions causing spinal cord compression. J Neurosurg 31: 452±458 14. Ransoho¨ J, Spencer F, Siew F, Gage L Jr (1969) Transthoracic removal of thoracic disc. Report of three cases. J Neurosurg 31: 459±461 15. Stillerman CB, Weiss MH (1996) Surgical management of thoracic disk herniation and spondylosis. In: Menezes AH, Sonntag VKH (eds.) Principles of spinal surgery. McGraw-Hill, New York, pp 581±601

Comment The paper reports the authors experience of video-assisted thoracic surgery in treating patients with a variety of pathologies including thoracic discs, vertebral body lesions, sympathetic dysfunction and a nerve sheath tumour.

744

M. Visocchi et al.: Thoracoscopic Approaches to the Thoracic Spine

The results are impressive and the authors demonstrate the e¨ectiveness of this technique in their hands. Very importantly, they emphasize the need for laboratory training, practice and perfection of the skills if this technique is to be used for spinal pathology. Their clear warning about the need to acquire competence before operating on patients should be heeded. G. Neil-Dwyer

Correspondence: Curtis A. Dickman, M.D., Neuroscience Publications, Barrow Neurological Institute, 350 West Thomas Road, Phoenix, AZ 85013-4496.