Combined transnasal and transoral endoscopic approaches to the craniovertebral junction

Objectives: To describe and evaluate a new technique of a combined endoscope-assisted transnasal and transoral approach to decompress the craniovertebral junction. Materials and Methods: A retrospective cohort of patients requiring an anterior decompression at the craniovertebral junction over a 12-month period was studied. Eleven patients were identi ﬁ ed and included in the study. Eight of the patients had an endoscopic approach [endonasal (2), endooral (2), and combined (4)]. Four of the 8 patients in the endoscopic group had a prior open transoral procedure at other institutions. These 8 patients were compared with a contemporary group of 3 patients who had an open, transoral–transpalatal approach. Charts, radiographic images, and pathologic diagnosis were reviewed. We evaluated the following issues: airway obstruction, dysphagia, velopharyngeal insuf ﬁ ciency (VPI), length of hospital stay (LOS), adequate decompression, and the need for revision surgery. Results: Adequate anterior decompression was achieved in all the patients. The endoscopic cohort had a reduced LOS ( P = 0.014), reduced need for prolonged intubation/tracheotomy ( P =0.024) and a trend toward reduced VPI ( P = 0.061) when compared with the open surgery group. None of the patients required a revision surgery. Conclusion: Proper choice of endoscopic transnasal, transoral, or combined approaches allows anterior decompression at the craniovertebral junction, while avoiding the need to split the palate. A combined transnasal–transoral approach appears to reduce procedure-related morbidity compared with open, transoral, and transpalatal surgeries.


INTRODUCTION
Several surgical approaches provide anterior access to the craniocervical junction and the upper cervical spine, including transoral, [1,2] high transcervical, [3,4] and endoscopic transnasaltransoral approaches. [5][6][7][8][9][10][11] Th e standard open transoral approach has gained wide acceptance by spine surgeons to treat ventral spinal cord compression at the C1-C2 level. However, to approach lesions of the craniovertebral junction, splitt ing of the transoral palate is oft en required for adequate exposure. Mummaneni et al highlighted a surgical technique variation to avoid the palate split by using simple retraction of the soft palate with a red rubber catheter passed transnasally and secured to the uvula. [12] Despite this modifi cation, in some circumstances, invasive approaches (ie, splitt ing the soft palate, resecting the hard palate, glossotomy, or midline mandibulotomy) are still required to provide surgical access to the craniovertebral junction. Such approaches are oft en used for decompression of lesions located high above the level of the palate. In addition, these invasive open approaches may also be needed in patients with atypical oral anatomy, or severe trismus (inability to distract the jaw open).
Palatal splitt ing has been reported to increase patient morbidity, especially velopharyngeal insuffi ciency (VPI), dysphonia, and dysphagia. [13] VPI occurs when there is incomplete closure of the nasopharynx with resultant escape of air and food into the nose during speech and swallow. Whereas dysphagia oft en resolves within 12 months following surgery, VPI oft en persists for a long term. We have recently referred 2 patients for pharyngoplasty aft er 1 year of persistent, signifi cant VPI following an open transoral approach.
In order to avoid splitt ing the soft palate (and glossotomy/ mandibulotomy, etc), we have used endoscopic transnasaltransoral techniques to decompress the craniovertebral junction in patients with challenging anatomic features.

MATERIALS AND METHODS
A retrospective chart review was performed to review all the patients who underwent surgery of the craniovertebral junction during a 12-month period at our hospital (UCSF). We included only those with lesions located between the clivus and the body of C2. Eleven patients were identifi ed and included. Th e medical records were reviewed for demographics and disease-specifi c information, including age, sex, diagnosis, surgical approach, length of hospital stay (LOS) aft er surgery, and surgery-related complications. Th e median age was 54 years (18-64 years). Eight of the patients had an endoscopic approach [endonasal (2), endooral (2), and combined endoscopic transnasaltransoral (4) approaches]. Th ese 8 patients were compared with a contemporary group of 3 patients who had an open, transoraltranspalatal approach for lesions of the craniovertebral junction.
Diagnoses for the endoscopic group included infection (2), tumor (2), rheumatoid arthritis (1), and basilar impression (3). Four of the endoscopic patients had prior transoral surgery at another hospital in the past. All the 3 patients having an open approach had rheumatoid arthritis. None of these 3 had prior C1-C2 surgery.
Early and late postoperative complications were recorded, including documented VPI, dysphagia, need for insertion of percutaneous gastric feeding tube, and airway complications defi ned as need for endotracheal intubation longer than 24 h, or a tracheotomy as a result of the surgery. Preoperative and postoperative images [computed tomography (CT) and/or magnetic resonance imaging (MRI)] were carefully reviewed for evaluation of the adequacy of resection or decompression.
Th e data were stored in an excel spreadsheet and transferred to SAS (SAS Institute Inc., Cary, NC, USA) for data analysis by the UCSF Department of Biostatistics, using Fisher's exact test and the Mann-Whitney U test where appropriate.

Surgical technique for endoscopic craniovertebral junction decompression
Th e patients were positioned supine and were intubated orally and given general anesthesia. Neuromonitoring with somatosensory-evoked potentials was used throughout the procedure. Spinal traction was applied as needed to get the odontoid process into a more normal position. Flouroscopy and neuronavigation were used for surgical guidance.
Th e endoscopic transnasal approach (endonasal approach) consisted of a bilateral approach through the nostrils. In the expanded form, as described by Kassam et al, [5] a nasal septal fl ap was fi rst elevated for closure and then a corridor was prepared with a maxillary antrostomy, ethmoidectomy, middle turbinate resection on the right, posterior septectomy, and a wide sphenoidectomy. However, not all patients required sphenoidectomy or middle turbinectomy, and dissection was tailored to the individual's anatomy. Depending on the location of the lesion, the sphenoid fl oor and clival bone were drilled to access the craniocervical junction pathology. A midline incision was made with an extended needlepoint cautery through the posterior nasopharygeal mucosa down to the preveterbral fascia. Th e prevertebral muscles were dissected vertically in the midline and elevated laterally off the spine, which allowed exposure of the anterior tubercle of the atlas. Decompression was then performed using a drill, currett es, and/or Kerrison Rongeur.
Th e endosocopic transoral approach (endooral approach) was performed with soft palate retraction using 1 or 2 red rubber catheters tied to the uvula and pulled cranially through the nostrils. [12] Th e oral cavity and tongue were retracted open with a Spetzler-Sonntag oral retractor. Th e endoscope was guided under the retracted soft palate to visualize the posterior pharyngeal wall, and the pharyngeal incision was created and continued in the midline to the desired height to expose the C1-C2 area [ Figure 1a, b]. Th e soft tissue and bony structures causing ventral cord compression were resected in a similar fashion to a transnasal decompression, described above [ Figure  1c-e]. Th e endoscope allowed us to "look" cranially above the level of the soft palate to complete the decompression.
For a combined transnasal and transoral approach, the exposure was a combination of the above-mentioned steps in both routes. Th en the endoscope and surgical instruments were brought into the surgical fi eld alternatively through the nose and mouth, in order to maximize the exposure with less dissection. Decompression was straightforward because visualization was gained from 2 diff erent angles (from above and below the palate). Th e most favorable feature gained via combined transnasal and transoral approach is the ability to visualize laterally beyond the confi nes of the nasal cavity. Such lateral visualization is restricted by the nasal cavity/pterygoid plates in the transnasal-only approach. Th e addition of the transoral endoscopic approach increased the ability to reach out laterally beyond the confi nes of the transnasal approach.
CT-based image guidance navigation was typically used for the endoscopic cases. Aft er the pharyngeal incision was completed, surgical dissection was performed with 2 surgeons working Closure of the pharyngotomy was performed with absorbable sutures [ Figure 1f]. Tissue sealant and a transnasal merocel sponge were packed in the nose. A transoral feeding tube was then passed under endoscopic guidance. Postoperative CT scans or MRI were performed and adequate decompression was assured in every patient. If the clivus was resected as part of the dissection, a pedicled nasal septal fl ap was harvested and rotated over the clival defect for closure. Th is was held in place with an absorbable tissue sealant (DuraSeal, Covidien, Mansfi eld, MA, USA) and 2 transnasal merocel sponges.

DISCUSSION
Th is study presents our early experience with a combined endonasal-endooral approach. While the approaches reported earlier include purely endonasal [5,6] or endoscopic transcervical approach, [3] our approach uses a fl exible strategy with an   endonasal approach or an endooral approach or a combined endonasal-endooral approach. [12] Th e combination of an endoscopic transnasal and transoral route appears to be a pragmatic way to conserve the advantages of endoscopic visualization via diff erent corridors, while minimizing procedurerelated morbidity due to splitt ing of the soft palate. We found that the endooral approach was advantageous in providing access to lesions that extended too far inferiorly to be reached by a purely endonasal approach. [6] Moreover, in standard open transoral approaches with microscope visualization, the hard palate sometimes still obstructs visualization of the upper extent of the compressive lesion. Th e use of the endoscope overcame this obstacle with ease as it could be navigated to look around the palate.
Previously reported endoscopic transnasal odontoidectomy reports mentioned the most caudal limiting extent of the transnasal route to be the C1 rim (due to the position of the hard palate and the size of the nostril). [6] By combining the endonasal approach with a transoral endoscopic approach, we overcame this limitation and were able to reach lesions that extended into the mid-body of C2.
Appropriate utilization of the combined transnasal and transoral endoscopic approach allowed for a minimally invasive surgery with full exposure for anterior decompression at the craniovertebral junction while avoiding a split of the soft palate. Th e optimal choice of a transnasal, transoral, or combined endoscopic approach should be tailored according to each individual's anatomy. Physical examination of the patient will reveal anatomic factors, such as trismus that would prevent oral exposure of the pharynx. We have also found that a careful review of the preoperative CT or MRI scan to evaluate the relative location of the hard palate and the target for decompression allows us to pick the optimal choice for a surgical approach. A radiographic line drawn along the fl oor of the palate to the posterior pharynx (the nasopalatal line) serves as an excellent reference point to assess the lesion location. [14] Th e lesions can be categorized as types A (high above the nasopalatal NP line= nasopalatal line [ Figure 2], B (intermediate location above the NP line), and C (at the level of (or below) the NP line). For intermediately located lesions (Type B), either an endoscopic transnasal or an endoscopic transoral approach may be adopted for decompression. Such lesions may also be easily accessed using a standard, open transoral approach without a palate split as reported by several authors. [1,2,12,[15][16][17] Finally, for low lying lesions or lesions extending to the midbody of C2 (Type C, at or below the NP line), an endoscopic transoral approach may be used [ Figure 3].
We found that the endoscopic approach signifi cantly reduced Th e other issue was that this series was based on our referral patt erns at a major tertiary care center, which might be biased toward more complicated patients. Th is was evidenced by the fact that 4 of the 8 patients undergoing the purely endoscopic procedure had a prior open transoral procedure in the past at an outside hospital. Although revision surgeries might lead to a higher complication rate, our ability to perform adequate resections endoscopically aft er a prior open procedure further demonstrates the feasibility of the endoscopic approach.

CONCLUSION
Th e combined endonasal-endooral approach is a useful approach off ering a wide access to the anterior craniovertebral junction. Th is approach appears to reduce airway obstruction and LOS aft er surgery when compared with a standard open transoral/transpalatal approach.