Revision surgery for posterior stabilized thoracolumbar fracture using mini‐open anterior approach and expandable cage

Abstract

Objective:  To evaluate the surgical techniques and outcomes of revision surgery for compromised posterior stabilization or insufficient neurological decompression using anterior mini‐open approach and expandable cage.

Methods:  From August 2005 to June 2008, a total of 235 patients were operated on in our center for thoracolumbar fractures with dorsal transpedicular stabilization. Twenty‐six of these patients underwent revision surgery, the main reasons being back pain and stagnant neurological recovery. The surgical procedure comprised a single‐level thoracolumbar corpectomy and/or canal clearance, followed by an expandable cage reconstruction. The average interval between primary and revision surgery was 5 months (range, 3–11 months). A transthoracic (n= 11) or transthoracic transdiaphragmatic (n= 15) mini‐open approach was conducted using a table‐mounted retractor.

Results:  The operating time averaged 105 min (range, 95–135 min) for the transthoracic approach and 152 min (range, 120–190 min) for the transthoracic plus transdiaphragmatic approach. The overall mean blood loss was 780 ml (range, 550–1700 ml). Over time, the pre‐operative neurological deficit improved in 6/7 patients by at least one Frankel/American Spinal Injury Association (ASIA) grade. On a visual analogue scale (VAS) from 0 to 10, the mean local thoracolumbar back pain was relieved significantly from 6.8 before operation to 3.8 at 3 months, 2.4 at 6 months, and 1.5 at 12 months postoperatively. None of the patients developed intercostal neuralgia or post‐thoracotomy pain syndromes.

Conclusion:  For patients with compromised stabilization or insufficient neurological decompression after primary dorsal transpedicular stabilization for thoracolumbar fracture, anterior revision surgery can produce good results. The mini‐open anterior approach for corpectomy in the thoracolumbar spine is safe, reliable, and economical. The expandable cage is an excellent alternative for anterior reconstruction.

Introduction

Over the past two decades, there has been a tendency for more and more spinal surgeons to choose dorsal stabilization with an internal fixation system as the preferred treatment modality for thoracolumbar fractures ¹ , ² . These surgeons believe that they can fulfill three targets of treatment: decompression, repositioning and stabilization, with just one surgical approach. However, as Slosar et al. have concluded, it is the biomechanical behavior of the injured and stabilized spine that must determine the treatment ³ . Clinically, there is disagreement on the necessity for anterior or combined anterior and posterior stabilization. For certain types of fracture where there is a risk of operative failure due to the anterior load‐bearing column being compromised, posterior stabilization is not enough ⁴ , ⁵ . In these cases, additional anterior reconstruction is necessary.

In the present retrospectively followed series of cases, the authors have evaluated the surgical techniques and outcomes of 26 patients who underwent revision surgery (using a mini‐open approach and expandable cage) because of compromised index posterior stabilization or insufficient neurological decompression after primary surgery using a dorsal approach only.

Materials and methods

From August 2005 to June 2008, a total of 235 patients were operated on for thoracolumbar fractures with dorsal transpedicular stabilization in our center. They incuded 183 men and 52 women, with a mean age of 37 years (19–64 years). During follow‐up, 19 patients were diagnosed as having symptomatic compromised posterior stabilization on the basis of their main complaints and radiologic results. Another seven patients were transferred from other clinics for the same reasons. These 26 patients (22 men, 4 women; mean age of 33 years with a range of 24–45 years) presented with back pain or stagnant neurological recovery. In all these patients only one level was involved and they all underwent one‐level corpectomy with expandable cage reconstruction via a mini‐open anterior approach (Table 1). The mean interval between primary and revision surgery averaged 5 months (range, 3–11 months).

Table 1.

Patients' demographic data

	No. of patients (%)
Sex
Male	22 (84.6)
Female	4 (15.4)
AO type of primary fracture
Type A2 or A3 fracture	17 (65.4)
Type B fracture	6 (23.0)
Type C fracture	3 (11.5)
Pathology level and approach
T10‐T11 Transthoracic approach	11 (42.3)
T12‐L2 Transthoracic and transdiaphragmatic approach	15 (57.7)
Symptoms
Back pain	19 (73.1)
Stagnant neurologic recovery	7 (26.9)

The SynFrame retractor

The SynFrame retractor is fixed aseptically by two adjustable arms onto the operating table, allowing a 360° surgical access from any point inside the ring ⁶ (Fig. 1). It assures permanent stability of the surgical field, enabling the surgeon to create access to the anterior spine through a small incision in the thorax or retroperitoneal area. Blades from 6–16 cm in length can be clicked onto the ring at any point along it and adjusted individually in all three planes. In the current series, one or two blades were installed medially, one cranially, one laterally, and one caudally. Additionally, a fiberoptic light source can be fixed onto the ring by clamps, providing perfect illumination of the entire operating field.

Figure 1.

View of the SynFrame retractor. The blades can be adjusted individually, allowing 360° surgical access from any point inside the ring. 1. Special clamp, which is clicked onto the ring and holds the retractor arms or the fiberoptic light source. 2. Hohmann retractor fixed to the ring. 3. Fiberoptic light source mounted on the ring and delivering perfect illumination into the depths of the wound.

Surgical technique

Since the involved levels in this series were from T₁₀ to L₂ a left‐sided approach was used, except in cases with neural decompression, where the approach was on the side where intrusion on the canal was most severe. The patients undergoing thoracotomy were intubated with a double‐lumen tube, and the lung on the side of approach was partially deflated by the anesthesiologist after rib removal.

A 6–8‐cm incision was made over the rib one level above the corpectomy level. Approximately 10 cm of the rib was removed and the ipsilateral lung partially deflated. The parietal pleura was then opened parallel to the ribs, and the lung moved cranially with a surgical towel. The SynFrame adjustable retractors were then used in the dissection procedure.

Where a transdiaphragmatic approach was adapted, the diaphragm was opened along the thoracic wall, leaving a 2‐cm rim laterally, with traction sutures every 2 cm to facilitate later closure. The outer fascia and muscular layer of the diaphragm were incised with a monopolar electrode in an anteroposterior direction up to the lateral convexity of the spine, which could be readily palpated. Great care was taken to stay a safe distance from the perivascular layers of the abdominal aorta. After bluntly dissecting the inner fascia of the diaphragm, the peritoneum was pushed away medially with cottonoids. Using longer SynFrame blades to retract the peritoneal sac medially, the anterolateral convexity of the spine was exposed.

After confirming the correct level by image intensifier and adapting the position of the SynFrame blades, the corpectomy and its reconstruction could then be performed.

Corpectomy and reconstruction

With the overlying segmental vessels of the involved vertebra dissected and ligated, the disc above and below the corpectomy level was removed and the bone graft bed prepared by clearing all the cartilage of the corresponding end plates. A subtotal corpectomy of the fractured vertebral body was performed, leaving the contralateral and anterior vertebral body walls in place. In the case of a neurological deficit, the posterior wall was also removed to decompress the dura mater.

In this series, an expandable titanium cage of appropriate diameter was used for corpectomy reconstruction. The cage was filled with, and surrounded by, cancellous bone from the vertebra that had been removed. The cage was fixed to the inserter and progressively adjusted in situ by counterclockwise rotation to the height of the corpectomy defect (Fig. 2). In order to engage it well with the vertebral endplates in accordance with their sagittal alignment, end pieces of the cage with appropriate angles were carefully selected.

Figure 2.

Using an inserter, the cage is progressively adjusted in situ by counterclockwise rotation to the height of the corpectomy defect.

Postoperative management

A drain was placed and closure performed in layers. The patients were nursed supine and log‐rolled for comfort. Chest or retroperitoneal suction drains were removed when X‐ray films showed that the lung had expanded and drainage had reduced to less than 100 ml of fluid over 24 h.

Postoperatively, bed rest and limited activity with the protection of a customized brace was recommended for at least 4–6 weeks. During follow‐up, plain lateral and anteroposterior radiographs were taken at different postoperative intervals. Neurological deficit was assessed by a modified Frankel grading system (Table 2). Pain was evaluated on a visual analogue scale (VAS) from 0 to 10 for thoracolumbar (TL) back pain and for pain in the anterior approach site at 3, 6, and 12 months postoperatively. The patients' functional recovery and reintegration were also recorded. No removal of hardware was necessary.

Table 2.

Modified Frankel classification

Grade	Signs
A	No motor or sensory function
B	Preserved sensation only
C	Preserved motor function (nonfunctional)
D	Preserved motor function (functional)
D1	Motor function 3/5, or total sphincter paralysis
D2	Motor function 4/5, or neurogenic bladder or bowel
D3	Highly functional 4/5, and normal sphincters
E	Completely normal

Results

A left sided mini‐open transthoracic approach was chosen in 11 patients to access the level of T₁₀‐T₁₂ and a transthoracic plus transdiaphragmatic approach in 15 patients for intervention at the T₁₂‐L₂ level. No additional instrumentation was used in this series.

The operating time (OT) was recorded from time of incision to closure. For the transthoracic approach, the OT averaged 105 min (range, 95–135 min) and for the transthoracic plus transdiaphragmatic approach, 152 min (range, 120–190 min). In cases requiring spinal canal decompression, an average of an additional 50 min (range, 35–65 min) was needed. The average OT for inserting an expandable cage was 20 min (range, 15–35 min). The overall mean blood loss averaged 780 ml (range, 550–1700 ml).

No intra‐operative complications occurred; in particular, there were no vascular or visceral complications. None of the patients developed intercostal neuralgia or post‐thoracotomy pain syndromes. Neither postoperative wound infection nor deep venous thrombosis occurred. There were two cases of mild postoperative ileus which resolved spontaneously without any intervention. The average postoperative stay in hospital was 15 days (range, 10–24 days).

Postoperative X‐ray films showed good hardware position in all patients in the study. During an average of 19 months (range, 12–32 months) follow‐up, no hardware failure was detected. Fusion was achieved in all the cases requiring decompression and no loss of correction was observed in this group (Fig. 3).

Figure 3.

A 38‐year‐old male patient with L₁ burst fracture. (A) Preoperative sagittal CT image prior to reconstruction. (B) Five months after primary transpedicular fixation, this patient complained of back pain and a lateral X‐ray film indicated potential loosening of the hardware. (C) Three months after surgical anterior L₁ corpectomy and expandable cage reconstruction. An angular stable plate was added.

Seven patients with stagnant neurological recovery after primary surgery underwent anterior decompression and reconstruction. By the 12 month follow‐up, three patients had improved from Frankel C to D; two patient had improved by two grades from Frankel B to D; and one patient had recovered almost totally from Frankel D to E. One patient with a bladder function deficit had no obvious recovery postoperatively. In all, over time the pre‐operative neurological deficit improved in 6/7 patients by at least one Frankel/ASIA grade.

Mean local TL back pain at the fracture site was 6.8 (range, 5–9) before operation. Postoperatively, the pain score decreased to 3.8 (range, 2–5) at 3 months, 2.4 (range, 1–4) at 6 months, and 1.5 (range, 0–3) at 12 months, these scores being significantly different to the preoperative scores (P < 0.05). Mean pain at the site of the anterior approach was 2.8 (range, 1–6) at 3 months, 2.0 (range, 0–5) at 6 months, and 1.2 (range, 0–4) at 12 months. Fifteen of 19 (79%) of patients with back pain as the main preoperative complaint reported satisfactory pain relief and had unlimited function at the 12 month follow‐up. Ten of them returned to their previous level of employment and another five were able to work, but at a lighter job. Four (21%) patients in this series experienced slight or no improvement in their back pain compared with that before operation. They could only work part time or not at all. The persistence of pain was not considered to be related to the severity of the original injury, surgical techniques or fusion conditions.

Discussion

With modern design and appliances constructed of new material, transpedicular screw fixation by a dorsal approach offers a fast, stable and safe means of achieving stabilization and correcting malalignments ⁷ . In a prospective study of 133 patients with unstable thoracic and lumbar spinal fractures who underwent dorsal instrumentation, Oertel et al. demonstrated that successful repositioning, reliable fracture consolidation and neural decompression as well as good neurological recovery can be achieved via the dorsal approach in most cases ⁵ . However, with this approach, special attention should be paid to possible complications. First, in the absence of ventral load‐bearing reconstruction, subsequent postoperative loss of correction including residual kyphotic deformities, a high pseudoarthrosis rate and implant fatigue with implant failure can occur ⁸ . Second, although achieving decompression indirectly by ligamentotaxis ⁹ , this technique sometimes results in insufficient neurological decompression. In the current series, 19 of 235 patients (8%) who underwent index surgery with dorsal stabilization required subsequent revision surgery for anterior reconstruction with or without decompression due to compromised posterior stabilization or insufficient neurological decompression. Another nine patients referred from other hospitals underwent the same procedure for the same reasons. Their main complaints included back pain or stagnant recovery of neurological function.

What should be mentioned is that, in contrast to the two‐stage procedure for combined anteroposterior treatment preferred by the members of the spine working group of the German Trauma Society (DGU) ¹⁰ , patients in this series were treated primarily only by dorsal stabilization, followed by a bed rest and immobilization for at least 4–6 weeks. During the first 3 months after surgery, they undertook limited activity. This is why revision surgery was conducted several months later, after compromised stabilization or stagnant neurological recovery had become evident.

In an attempt to avoid or diminish the incidence of complications such as intercostal neuralgia and post thoracotomy pain with the traditional wide open anterior approach to the thoracic and lumbar spine ¹¹ , ¹² , minimally invasive endoscopic approaches were introduced in the 1990s ¹³ . However, in a prospective study compared open thoracotomy with thoracoscopy, an equal incidence and intensity of persistent post thoracotomy pain was found ¹⁴ . Furthermore, because these techniques require a steep learning curve, prolonged anesthesiological monitoring and operation times, and considerable financial investment on an endoscopic set‐up and disposable instruments, they can hardly be recommended for common use. Recently, with the use of a new table‐mounted retractor system (SynFrame or other similar devices), mini‐open approaches to the anterior TL spine have been successfully introduced by several authors ⁶ , ¹⁵ , ¹⁶ . Such a mini‐open anterior approach combines the advantages of “pure” endoscopic approaches with those of an open procedure. These include a familiar direct three‐dimensional view of the anterior part of the spine, safer preparation of nerves and vascular structures, and facilitated corpectomy with or without spinal canal clearance as well as cage or graft insertion ¹⁵ , ¹⁷ . Though comparing surgical times of different techniques is of little value in different series with different surgeons, Kossmann et al. reported significantly shorter operating times for corpectomy and anterior reconstruction in the TL spine through his mini‐open approach compared with endoscopic approaches ¹⁵ . Similarly, Mühlbauer et al. reported encouraging initial experience with an open, minimally invasive, retroperitoneal approach in five corpectomy cases, using a different, but comparable, self‐retaining retractor ¹⁷ .

In order to overcome the shortcomings of autografts or allografts, various implants are available for vertebral defect replacement after corpectomy ¹⁸ , ¹⁹ . These implants are made from titanium, ceramic materials and carbon. They are cut to the required size and fitted into the spinal defect. Tightening of a previously inserted posterior internal fixator can provide a “press fit in situ”, as can additional anterior instrumentation ²⁰ . However, optimal placement of a nonexpandable spacer can be demanding and challenging. In an attempt to overcome the technical problems of nonexpandable cages, various expandable cages that can be adjusted in situ to the height of the corpectomy defect have been developed. These devices have been used successfully for various indications ²¹ . Although some studies have found no significant differences between the biomechanical properties of expandable and nonexpendable cages ²² , expandable cages can be introduced from various approach angles and distracted in situ via a unique cardanic mechanism, facilitating maneuvering during surgery. In addition, visual control of all critical anatomic landmarks remains feasible during cage insertion and expansion. Especially worthy of emphasis, fluoroscopy should be used to check that the cage fits well into the body defect with respect to its sagittal alignment. End plates of different angles can be chosen to accurately fit into the space between upper and lower endplates (Fig. 4). It is convenient and not time‐consuming to use such a cage. No related complications occurred in this series.

Figure 4.

Illustrative example of a case during surgery. (A) Inappropriate choice of end pieces, with poor fit into the upper and lower endplates. (B) Change to differently angled end pieces which fit well into the body defect and match the segmental lordosis.

For patients with compromised stabilization or insufficient neurological decompression after primary dorsal surgery for thoracolumbar fracture, anterior revision surgery can produce good results. The mini‐open anterior approach for corpectomy in the thoracolumbar spine is safe, reliable, and economical. The expandable cage is an excellent alternative for anterior reconstruction. In the current series, most patients attained a good or very good clinical outcome with further improvement in symptoms and function and a relatively high rate of returning to work.