Discussion

The clinical presentation in our series was consistent with what is found in the literature.^{Reference Singh, Gupta and Singla1,Reference Riffat, Dwivedi, Palme, Fish and Jani3,Reference Kuet, Kasbekar, Masterson and Jani10} The majority of our patients were asymptomatic at the time of diagnosis, which is a strong indicator that parapharyngeal space lesions are underdiagnosed. In terms of clinical investigation, MRI and FNAC were sufficient for an adequate clinical diagnosis and to determine the preferential surgical approach. Magnetic resonance imaging is the preferred imaging as it provides sufficient information concerning the tumour origin, nature, size and its relation to nearby vital structures. Moreover, it is well documented that MRI is superior to CT in diagnosis and assessment of pleomorphic adenoma of the parapharyngeal space.^{Reference Som, Sacher, Stollman, Biller and Lawson11,Reference Basaran, Polat, Unsaler, Ulusan, Aslan and Hafiz12} Therefore, we can suppose that MRI is sufficient to suggest the diagnosis of pleomorphic adenoma pre-operatively. However, we believe that FNAC can still provide a significant input, despite limited access to the parapharyngeal space. It largely relies on the experience of the cytologist and the surgeon or radiologist taking the sample.^{Reference Matsuki, Miura, Tada, Masubuchi, Fushimi and Kanno13–Reference Arnason, Hart, Taylor, Trites, Nasser and Bullock15}

Understanding the anatomy of parapharyngeal space is crucial for the correct diagnosis and, most importantly, to decide how to approach the parapharyngeal space for the tumour excision. The majority of parapharyngeal space tumours are of salivary origin and are located in the pre-styloid compartment.^{Reference Riffat, Dwivedi, Palme, Fish and Jani3,Reference Chen, Sun, Tang and Hu4,Reference Kuet, Kasbekar, Masterson and Jani10} There are several factors determining the surgical approach to a parapharyngeal space tumour (e.g. its size, location and histopathology). Nevertheless, the decision is always subjective and depends on the surgeon's preference and comfort level.^{Reference Boyce, Curry, Luginbuhl and Cognetti16} An attempt to standardise surgical approach decision-making was made by Kanzaki and Nameki before the advent of transoral robotic surgery and relying on the pre-operative imaging and the division of the parapharyngeal space into six compartments.^{Reference Kanzaki and Nameki17}

The transoral approach was first described by Harry Ehrlich in 1950 and is the most debated surgical approach in the literature.^{Reference Markou, Blioskas, Krommydas, Psillas and Karkos18,Reference Ehrlich19} It is associated with minimal access and poor vision of the operating field,^{Reference De Virgilio, Costantino, Mercante, Di Maio, Iocca and Spriano20} and it was only used during the first years of our study period, prior to the use of transoral robotic surgery in our centre. Transoral robotic surgery is a relatively new approach, with the first documented use in 2007.^{Reference O'Malley and Weinstein21} The Da Vinci® robotic system has the advantages of a magnified three-dimensional visualisation, tremor-reduction technology and wristed instruments with seven degrees of freedom.^{Reference Boyce, Curry, Luginbuhl and Cognetti16} Some studies suggest these advances have reduced the risk for tumour rupture, incomplete excision and uncontrollable bleeding intra-operatively.^{Reference De Virgilio, Costantino, Mercante, Di Maio, Iocca and Spriano20} For centres where transoral robotic surgery is not available, an endoscope-assisted transoral approach seems to provide satisfactory access.^{Reference Liu, Yu and Zhen22} The pre-operative investigation is crucial before making the decision of a transoral approach. The evaluation of the mouth opening is the most important aspect to anticipate the quality of both exposure and working space. Pre-operative imaging is paramount to rule out carotid artery encasement and bony erosion of the skull base. The main contraindications for transoral approaches are trismus, macroglossia and maxilla-mandibular defects.^{Reference Maglione, Guida, Pavone, Longo, Aversa and Villano23,Reference Arora, Kotecha, Acharya, Garas, Darzi and Davies24} Other studies advised against transoral robotic surgery for tumours bigger than 6 cm.^{Reference Maglione, Guida, Pavone, Longo, Aversa and Villano23} Another advantage of transoral robotic surgery is to limit the need for a transpalatal approach. The entrance door to the parapharyngeal space is made by a linear mucosal incision around the tonsillar fossa, preserving the palatoglossus, palatopharyngeus and superior constrictor muscles. Dividing the pterygoid muscles can improve the mouth opening and transoral exposure. There is usually no need for a mucosal reconstruction as a primary closure can be achieved.

The main risks for all transoral approaches are neurovascular injury, tumour spillage or implantation, incomplete excision, and surgical site infection.^{Reference Markou, Blioskas, Krommydas, Psillas and Karkos18,Reference Chan, Tsang, Eisele and Richmon25} In our series, two patients out of six from the transoral robotic surgery group presented with intra-operative capsular rupture. Both patients had undergone transoral incisional biopsies prior to the surgery, which was likely to induce a capsular fragility. A history of biopsy should be considered as an argument in favour of a transoral approach, in order to avoid seeding along the transcervical incision and to obtain a safe mucosal margin intra-orally. Moreover, follow-up imaging did not show any evidence of residual or recurrent disease. Some authors claim that tumour spillage during a transoral approach allows for a thorough washing and prevents tumour dissemination in the neck.^{Reference Sethi, Dale, Vidhyadharan, Krishnan, Foreman and Hodge26}

The transparotid-transcervical approach was the most common in our series and is preferred by many for the good access and visualisation it provides.^{Reference Cohen, Burkey and Netterville27–Reference Zhi, Ren, Zhou, Wen and Zhang29} The keystone for this surgery is to preserve the facial nerve and limit the retraction injury. In order to improve the exposure, the authors recommend a nasotracheal intubation for a complete jaw closure during the surgery and an extension of the classical Blair's incision in the submandibular area to provide a better exposure of the major blood vessels and cranial nerves. The posterior belly of digastric muscle, stylohyoid muscle and stylomandibular ligament are commonly divided. The submandibular gland is retracted anteriorly or excised, and the mandible is protracted to identify the major neurovascular structures of the post-styloid space.^{Reference Olsen8,Reference Vallabhaneni, Mandakulutur, Vallabhaneni, Prabha and Banavara30} A parotidectomy is performed, as completely as possible posteriorly and in the deep lobe. Some studies advocate for preserving the superficial lobe of the parotid gland whenever feasible, with the aim of reducing the risk of Frey's syndrome and for a better cosmetic outcome.^{Reference van Hees, van Weert, Witte and René Leemans31} The main indication for this approach is a mass reaching the skull base and/or merging laterally within the deep lobe of parotid gland with ill-defined contour on radiological investigations.^{Reference Chen, Sun, Tang and Hu4,Reference Horowitz, Ben-Ari, Wasserzug, Weizman, Yehuda and Fliss5}

The transcervical approach is the most frequently used for excision of parapharyngeal space mass of all origins, most notably those arising from the post-styloid compartment.^{Reference Markou, Blioskas, Krommydas, Psillas and Karkos18} The parapharyngeal space is accessed through a submandibular incision at the level of the hyoid bone, which can be further extended to the submental area for mandibulotomy with lip splitting. Many surgical steps are similar to those of the transparotid-transcervical approach (i.e. retraction of the submandibular gland, division of the posterior belly of the digastric muscle, the stylohyoid muscle and stylomandibular ligament and protraction of the mandible). The facial nerve is not identified. The mylohyoid muscle can be transected for further exposure of parapharyngeal space.^{Reference Basaran, Polat, Unsaler, Ulusan, Aslan and Hafiz12} However, none of our cases underwent an exclusively transcervical approach because all tumours originated from the deep lobe of the parotid gland, which justified a parotidectomy in our opinion. Some studies suggest the transcervical approach should be avoided for masses bigger than 4 cm because the mandible is a barrier for tumour manipulation and extraction.^{Reference Riffat, Dwivedi, Palme, Fish and Jani3,Reference Bozza, Vigili, Ruscito, Marzetti and Marzetti6,Reference Markou, Blioskas, Krommydas, Psillas and Karkos18,Reference Cohen, Burkey and Netterville27,Reference Papadogeorgakis, Petsinis, Goutzanis, Kostakis and Alexandridis32} Furthermore, it is not recommended for tumours with a long vertical dimension and with radiological suspicion of cranial foramen invasion as they are associated with a high risk of internal carotid artery laceration during blunt dissection.^{Reference Basaran, Polat, Unsaler, Ulusan, Aslan and Hafiz12} Overall, many authors support the transcervical approach because it can provide good control of the lesion while avoiding dissection of the facial nerve, but it is mainly reserved for tumours with limited attachment to the deep lobe of the parotid.^{Reference Horowitz, Ben-Ari, Wasserzug, Weizman, Yehuda and Fliss5,Reference Malone, Agrawal and Schuller33}

Mandibulotomy is essentially indicated for malignant tumours, recurrent neoplasms, large benign tumours and highly vascular neoplasms with a need for vascular control.^{Reference Basaran, Polat, Unsaler, Ulusan, Aslan and Hafiz12} There are three ways of conducting the osteotomy: through the body (lateral mandibulotomy), and midline and paramedian approaches. Following the osteotomy, the mandible is swung and the floor of the mouth is stretched. An incision in the floor of the mouth is made 1 cm medial to the gingiva across the mucosa, soft tissues and muscles of mouth floor and extended up to the anterior tonsillar pillar. It carries a possible risk of inferior alveolar nerve anaesthesia, malocclusion, teeth loss, malunion or non-union of the mandible, and in some cases it might require a tracheostomy.^{Reference Vallabhaneni, Mandakulutur, Vallabhaneni, Prabha and Banavara30,Reference Patel34} No mandibulotomy was performed in our series, although it is reported in 2.0 to 20.5 per cent of cases in the literature.^{Reference Olsen8,Reference Cohen, Burkey and Netterville27,Reference Malone, Agrawal and Schuller33,Reference Hughes, Olsen and McCaffrey35,Reference Miller, Wanamaker, Lavertu and Wood36} From our standpoint, mandibulotomy is now unnecessary for the vast majority of pleomorphic adenomas of the parapharyngeal space. Additionally, no combined approach was necessary in our series for a complete tumour manipulation and excision. As a result, the morbidity and complications of the mandibular split were avoided, and the length of hospitalisation was limited.

The histopathological subtypes of pleomorphic adenoma are divided into three groups based on their epithelial element: mucoid, myxoid or chondroid.^{Reference Seifert, Brocheriou, Cardesa and Eveson37} The most common is the myxoid subtype, which is documented to have a higher rate of recurrence and is associated with incomplete and thin capsules.^{Reference Dulguerov, Todic, Pusztaszeri and Alotaibi38} However, there are a number of pathological differences between a pleomorphic adenoma originating from the superficial or deep lobe of the parotid gland. Deep lobe pleomorphic adenomas tend to be bigger, have a thicker capsule and there is less invasion of the capsule by the tumour.^{Reference Harney, Murphy, Hone, Toner and Timon39} The rate of recurrence documented in the literature ranges between 0 and 10.5 per cent.^{Reference Horowitz, Ben-Ari, Wasserzug, Weizman, Yehuda and Fliss5,Reference Basaran, Polat, Unsaler, Ulusan, Aslan and Hafiz12,Reference Cohen, Burkey and Netterville27,Reference van Hees, van Weert, Witte and René Leemans31,Reference Shahab, Heliwell and Jones40,Reference Lim, Park, Kang, Kim, Choi and Kim41} Our rate of complete surgical excision and the absence of recurrence during the follow up seem to validate our management. Indeed, a concession can be made on the surgical exposure in order to avoid the morbidity of a systematic transcervical approach or a mandibulotomy.

The complications were dominated by facial nerve palsy in our series because of the prolonged retraction of the nerve during tumour dissection in the parapharyngeal space. This occurs from 5.2 to 30.8 per cent of cases in the literature. Permanent facial nerve palsy has been reported after up to 7.7 per cent of procedures.^{Reference Ijichi and Murakami2,Reference Chen, Sun, Tang and Hu4–Reference Bozza, Vigili, Ruscito, Marzetti and Marzetti6,Reference Cohen, Burkey and Netterville27,Reference van Hees, van Weert, Witte and René Leemans31}

The limits of our study are its retrospective design and the small number of patients, which prevented statistical comparison between groups and a short follow-up period. Indeed, these tumours evolve slowly and late recurrences can still occur.

Our decision tree for management of pleomorphic adenoma of the parapharyngeal space was based on clinical examination and radiological findings (Figure 3). The transparotid-transcervical approach was preferred when close contact to the facial nerve was suspected. It should be preferred to the transcervical approach for salivary tumours. Transoral robotic surgery is a good option for selected, well-defined masses in the parapharyngeal space with oropharyngeal bulge clinically and radiologically. The classic transoral approach should be avoided if transoral robotic surgery is available because it is associated with significant risk of neurovascular injury with limited capability for intra-operative management. Indications for mandibulotomy must remain exceptional.

Fig. 3. Proposal of decision-tree for management of pleomorphic adenoma of the parapharyngeal space (PPS). TORS = transoral robotic surgery.