Parotidectomy for benign tumors is undergoing constant evolution. The potential for recurrence and malignant transformation of pleomorphic adenomas creates complexities that have forced head and neck surgeons to undertake more comprehensive parotid surgery with facial nerve dissection. This approach carries inherent morbidities, including facial nerve injury, Frey syndrome, and facial asymmetry, that have to be addressed. Extracapsular dissection is compared with conventional superficial parotidectomy; surgical histologic findings are discussed as well as outcome data. More novel approaches are discussed as well. This article provides a systematic approach to benign parotid tumor surgery.
Key points
- •
Facial nerve anatomy and variable risk of injury in different areas of the parotid gland.
- •
Pleomorphic adenoma’s histologic characteristic leading to microscopic positive margins in most if not all cases, and its significance.
- •
Extracapsular dissection (limited parotid surgery) compares favorably with superficial parotidectomy in terms of outcomes but requires extensive experience to avoid morbidity.
- •
There are a great number of surgical approaches and incisions to access the parotid, and with greater experience less invasive approaches are appropriate.
- •
Recurrent pleomorphic adenomas are very difficult to cure, and the treatment plan needs to take into account the nature and extent of the primary surgery, and the role of adjuvant radiation therapy.
- •
Adjuvant radiation therapy plays a role in treatment of recurrent pleomorphic adenoma.
Introduction
Surgical treatment of benign parotid tumors in the early twentieth century was, as it is today, shaped by the significant risk to the facial nerve, and a lack of clear understanding of tumor biology. Surgery for benign tumors such as pleomorphic adenoma (PA) focused on intracapsular enucleation, in which the tumor capsule is opened and the contents removed. Joseph McFarland from University of Pennsylvania in the 1940s is credited with recognizing a high rate of recurrence after parotidectomy. Recurrence was observed in up to 45% of patients treated by intracapsular enucleation.
As the reports for recurrence began to mount in the midcentury, and more in-depth pathologic studies were undertaken, the technique and philosophy of parotidectomy for benign parotid lesions were refined. The resection of the tumor capsule and a margin of surrounding healthy tissue was advocated, as well as complete facial nerve dissection in an anterograde or retrograde direction. This approach evolved into standard and obligatory dissection of the facial nerve and its branches and in most cases removal of the superficial parotid gland, and less frequently the totality of the gland, which is the philosophy of care in a great many centers around the world presently. Nevertheless, great controversy exists as to the appropriate extent of surgical treatment of benign disease.
Clinicians who support complete facial nerve dissection with superficial parotidectomy point to increased safety for the facial nerve and a decreased rate of recurrence in the long term. Other surgeons think that there is less risk and morbidity when meticulous dissection is done outside the tumor capsule without preidentification or exposure of the main trunk of the facial nerve (extracapsular dissection). These surgeons also claim that dissection of the facial nerve increases the risk of intraoperative nerve damage, and causes scarring in the area of the nerve, which makes revision surgery much more difficult and risky.
Just as in other fields of surgery, salivary gland surgery is moving toward minimally invasive techniques that reduce the length of incision and surgical dissection, thereby potentially decreasing the risk of short-term and long-term complications. However dealing with parotid tumors presents certain inherent complexities: (1) facial nerve anatomy is widely variable and unpredictable, and (2) fine-needle aspiration (FNA) diagnostic pathology carries a high false-negative rate (4%–7%). These two factors demand that parotid surgeons have a high degree of expertise so that during the operation they can make decisions as to the appropriate extent and type of procedure, and how to address or avoid facial nerve injury. There is evidence to support both surgical philosophies in addressing benign parotid tumors, which means a nuanced approach must be considered.
Considerations in parotid tumors
PAs represent the most challenging benign tumors to address because of their predominance, histologic characteristics, high recurrence rate, and potential for malignant transformation. The treatment that is appropriate for this type of tumor could be effectively used for other benign parotid tumors as well.
Facial Nerve Anatomy and Dysfunction
The facial nerve has widely variable branching anatomy that is akin to a tree; no two facial nerves are alike. During parotidectomy surgery, the facial nerve branches often abut the tumor. Most studies have indicated that risk factors for transient facial nerve dysfunction include size of tumor, inflammatory condition, patient’s age, malignancy, and the type of surgical procedure. However, the location of the tumor within the parotid is of utmost importance. Laws of real estate govern parotid tumor’s involvement of the nerves as well; the areas that have more substance and allow a separation between the tumor and the branches of the facial nerve, such as the parotid tail or the posterior inferior parotid, are less likely to have nerve involvement. Areas of less substance, including superiorly where the superior division branching occurs or anteriorly over the masseter muscle, show a greater involvement with the nerve, and thus a higher incidence of temporary facial nerve dysfunction (see Table 2 ). This relationship holds true for the deep lobe of the parotid as well, in that tumors in the upper portion of the deep lobe tend to have closer association with the facial nerve, and require more manipulation of the nerve (circumferential dissection and transposition) and potential for injury. Gaillerad and colleagues suggested that manipulation of the facial nerve as it passed adjacent to the tumor was the cause of temporary facial nerve dysfunction (TFND) in most the cases. Cannon and colleagues concluded that the length of facial nerve that is dissected and exposed correlated with TFND. Dulguerov and colleagues reviewed the pathophysiology of TFND and reported nerve stretching as the cause of TFND.
In addition to the facial nerve, the greater auricular nerve (GAN), auriculotemporal nerve, and sympathetic fibers also innervate the parotid and adjacent tissue. The GAN courses behind the sternocleidomastoid muscle (SCM) and travels anteriorly and superiorly toward the parotid and the ear. Over the SCM it divides into anterior, posterior, and deep branches. The anterior branches innervate the skin over the parotid, whereas the deep branch enters the gland. The posterior branching system gives sensation to the earlobe and the skin beneath the cartilaginous auricle. GAN sacrifice is routinely done at many centers when performing parotidectomy. Ryan and Fee showed the following in patients who had GAN sacrificed during parotidectomy: 47% of the patients had anesthesia, 58% had paresthesia, and 26% had neither anesthesia nor paresthesia. None of their patients reported this deficit to be interfering with daily activities, which is echoed by other investigators. Although preservation of the GAN, specifically the posterior branch, does not prevent some degree of sensory loss, especially in the anterior distribution, it does minimize it. In addition, it can minimize other unwanted morbidity, including dysesthesia, discomfort on cold exposure, and traumatic neuroma.
The auriculotemporal nerve arises from deep in the parapharyngeal space and passes posterior to the temporomandibular joint to innervate the skin of the tragus and the temple. Along its course it gives secretomotor branches to the parotid gland as it passes posteromedial to the mandibular condyle. These nerve fibers are responsible for the development of Frey’s syndrome. The sympathetic input into the gland comes from the fibers that travel on the superficial temporal artery and then branch out to innervate the gland.
Histology
One-fifth of the parenchyma of the parotid gland lies deep to the facial nerve (deep lobe). Because most of the gland is superficial to the nerve, 90% of parotid tumors occur in the superficial lobe and 80% occur in the lower part of the gland.
Parotid gland tumors represent 3% of tumors occurring in the head and neck area, and 80% of tumors of the salivary glands. Eighty percent of parotid gland tumors are benign. Of these, the most common type is PA, which accounts for 65%, followed by cystadenoma lymphomatosum or Warthin tumor, which accounts for about 25%. Far less common, basal cell adenomas and oncocytomas are some of the other benign tumors. The remaining 20% of parotid tumors are cancers with varying degrees of aggressiveness and behavioral patterns. PA tumors occur in persons of all ages, with the highest incidence in the fourth to sixth decades. PA is most often diagnosed when the tumor is located in the superficial lobe, small (<4 cm), and mobile.
Most small cancers (T1/T2) of the parotid are clinically silent and appear demarcated on examination or even imaging. FNA has a high sensitivity and specificity for PA, and has a notable false-negative rate (4%–7%). However, FNA along with the imaging studies help guide clinicians as to whether the tumor is aggressive and requires more than limited surgery. Some clinicians advocate making a decision as to the aggressiveness of the tumor based on surgical findings and thus deciding to proceed with more involved surgery. However, this approach may be disappointing as well, because small cancers may have a normal appearance and feel soft, like benign tumors.
Zbären and colleagues reviewed the histology of PAs and found that the capsule can be either thick or thin, and at times is absent in certain areas. Breaches in the capsule can allow the tumor to grow into the surrounding tissue and appear as fingerlike projections or pseudopods. The capsule can be invaded and penetrated by tumor as well, again giving rise to pseudopods and, rarely, satellite lesions. In superficial parotid PAs, studies have shown there to be no satellite lesions in the deep lobe. Deep lobe tumors seem to differ in that they have thicker capsules.
Recurrent (or residual) pleomorphic adenomas
Recurrent (or residual) PAs (RPAs) are almost always multifocal. Theories as to the increase of recurrence include capsular rupture, satellite lesions, pseudopods, and grossly positive margins. Enucleation procedure or incisional biopsy for suspected lymph nodes also leads to recurrence. Initial multicentricity of primary PAs seems to be an extremely rare phenomenon. Biological and genetic factors may be at play as well, but are not well understood. PA has a low rate of recurrence when treated by a nonenucleation procedure (1%–4%). In the patient population that later develops RPA, the mean age at presentation is lower (33–35 years) than for those who remain free of disease (45–50 years). Recurrences are typically observed 7 to 10 years after initial treatment, except in cases of enucleation of the PA, which recur after a much shorter interval.
Although MRI can show the extent of disease, multinodularity, and deep lobe involvement, on histologic analysis of surgical specimens many more nodules are often found than imaging studies and clinical examination reveal. MRI can also show the amount of residual parotid gland remaining and help determine the extent and appropriateness of further surgical intervention. The facial nerve is the main concern when deciding on the appropriate course of treatment of RPA. Trauma and scar tissue formation from the primary surgery can make it difficult to distinguish the nerve. The natural planes between salivary tissue, tumor, and nerve may be lost with surgery and thus recurrent tumors can be more adherent to the nerve. In addition, the normal course and anatomy of the nerve is distorted. Strategies to identify the facial nerve in RPA, therefore require significant preoperative planning. The authors typically identify the facial nerve in the vertical segment of the mastoid or peripherally in a retrograde fashion. If the tumor is located away from the area of the nerve, nerve stimulation and identification without exposure in the scar are also feasible. It is also advisable to resect the incisional scar from the primary surgery, en bloc with the specimen if feasible. Recurrences are more common after treatment of recurrent tumors compared with primary surgery.
The widespread distribution of recurrent disease makes curative treatment difficult and often impossible, carrying with it a high risk of facial nerve complications and further recurrence, therefore the initial surgical approach must be sound and performed with expertise. PAs that are not surgically removed have the potential to undergo malignant transformation. The most common form of malignancy secondary to PAs is carcinoma ex PA. The development of this malignancy has a direct relationship with duration of PA tumor presence, with rates being as high as 9.4% at 15 years. It is often unclear how long the tumor has been present at the time of diagnosis, to be able to assess the appropriate level of risk. Thackray and Lucas estimated that 25% of untreated PAs eventually undergo malignant transformation. The incidence of malignant transformation is even higher in recurrent PAs.
In a landmark article by Witt and colleagues, the significance of margins in parotid tumor surgery was assessed. Retrospective analysis of pathology specimens was done, comparing total parotidectomy, partial superficial parotidectomy (PSP), and extracapsular dissection (ECD). They concluded that:
- 1.
The major outcomes (capsular exposure, tumor–facial nerve interface, capsular rupture, recurrence, and permanent facial nerve dysfunction) from surgical treatment of mobile, superficial PA smaller than 4 cm are not significantly altered by surgical approach (total parotidectomy, PSP, or ECD).
- 2.
Greater parotid tissue sacrifice results in higher rates of transient facial nerve dysfunction and Frey’s syndrome.
- 3.
Focal capsular exposure is a near-universal finding for cases of small (<4 cm), mobile PA predominantly of the superficial lobe, regardless of the extent of parotid tissue sacrifice.
- 4.
Dissecting PA from the facial nerve results in positive margins because of an incomplete capsule or perforating pseudopodia. Fewer separations of pseudopodia from the main tumor occur with expertly performed contemporary parotid surgery because most of the PA has a margin of normal parotid tissue.
- 5.
Capsular rupture resulted in a significantly higher rate of recurrence but did not vary by extent of parotid tissue sacrifice for small PAs (excluding enucleation).
- 6.
Hypocellular tumors did not result in a higher rate of capsular rupture or recurrence, and tumor multicentricity was not found in the clinically uninvolved deep lobe.
- 7.
Recurrence now occurs in a small percentage of patients receiving the best care but happens most frequently because of the continued practice of enucleation.
Other studies have shown little difference in recurrence rates between patients who had negative margins and those who had microscopic positive margins.
Extracapsular Dissection Versus Superficial Parotidectomy
An ECD is defined as a partial parotid surgery that involves removal of the tumor with preservation of capsule without exposure of the main trunk of the facial nerve. The incision to access the parotid varies between different centers, but most use the modified Blair incision. Once the incision is made and skin flaps are elevated, the tumor is localized and assessed. If the tumor is mobile and does not show suspicious physical findings, then ECD is performed. Neuromonitoring is used in all cases, assisted by nerve stimulation during the procedure. At first, a 5-mA current is used for nerve stimulation, and as soon as a branch is identified in the proximity of the tumor the current is reduced to 2 mA and the branch is exposed. If further exposure of the branch is deemed necessary for removal of the tumor, the current is reduced even further to 1 mA. A 2-mm to 3-mm margin of normal tissue is obtained when permissible by the facial nerve. Theoretically, because of less exposure of the nerve and less manipulation, the incidence of nerve paresis should be fewer. Frey’s syndrome should also occur with less frequency because the extent of parotid tissue exposure and thus secretomotor nerve reveal is much less.
Most studies with ECD come from European centers where this procedure is advocated for small, mobile, superficial lobe tumors. McGurk reported on one of the largest series comparing ECD with superficial parotidectomy, in which his group used a 4-cm cutoff for consideration for this technique as well as intraoperative determination of mobility of the tumor to decide which procedure was most appropriate. Other studies show that the risk of facial nerve paresis after ECD of tumors 4 cm or greater was 21% compared with 4% for the smaller tumors. European centers favor the use of ultrasonography for tumor assessment and thus as part of their decision making in terms of which technique to use preoperatively, and at times intraoperatively.
However, these concepts are not unique to European centers. Donovan and Conley reported that 60% of the patients with PA surgery did not have clear margins on the capsule because of the facial nerve interface. None of these patients had a recurrence. These investigators challenged the popular belief that a wide margin of normal parotid tissue is necessary to prevent recurrence. The group at the Sydney Head and Neck Cancer Institute echoed these findings as well and advocated limited parotid surgery for PAs because of low morbidity and recurrence rates.
In terms of outcomes, ECD shows similar recurrence rates with decreased surgical morbidity. Mehta and Nathan reviewed the data and compared several of the larger studies, which overall showed similar recurrence rates and a much higher incidence of facial nerve paralysis or paresis and Frey’s syndrome in the patient group that underwent superficial parotidectomy ( Table 1 ).
| Study | No. of Patients | Recurrence Rates (%) | Facial Nerve Paresis or Paralysis (%) | Frey’s Syndrome (%) | |||
|---|---|---|---|---|---|---|---|
| ECD | SP | ECD | SP | ECD | SP | ||
| Albergotti et al, 2012 | 1882 | 1.5 | 2.4 | 8 | 20.4 | 4.5 | 26.1 |
| Barzan & Pin, 2012 | 349 | 2.3 | 12 | 1.3 | 6 | 1.3 | 44 |
| McGurk et al, 2003 | 630 | 1.7 | 1.8 | 10 | 32 | 5 | 32 |
The data from these large retrospective studies clearly favor ECD, but must be analyzed with caution, keeping in mind some key factors: the retrospective nature of these studies creates an issue with internal validity because it does not remove bias. In addition, inherent within the ECD technique is the intraoperative decision making as to the extent of surgery, and thus more challenging or worrisome tumors are removed using more extensive parotidectomy techniques. These two factors create an inherent bias that skews the data. The data also come from large tertiary referral centers with high surgical volume and great technical experience, making application of this technique most appropriate for the more experienced surgeons.
Types of Incisions for Superficial Parotid Tumors
The modified Blair incision, initially described by Blair in 1918 and later modified by Bailey in 1941, has been the standard incision that most head and neck surgeons are trained to use for parotidectomy, and is most often used to treat benign parotid tumors. This technique is excellent in terms of offering a wide exposure that makes finding the facial nerve and its branches, as well as removing the tumor, effective and safe. This technique is preferred in training programs because it allows excellent exposure and versatility to convert into more extensive surgery if needed (eg, neck dissection). The drawbacks include larger incision, greater area of dissection, and greater potential for asymmetry.
Appiani further modified the approach and described the use of a facelift incision to remove benign parotid tumors. By placing the incision behind the ear along the occipital hairline, the surgeon is able to reduce the visibility of the scar without compromising on the exposure needed to perform nerve identification and dissection, as well as tumor removal.
Both of these techniques were originally described without any reconstructive measures to prevent facial asymmetry and Frey’s syndrome. A variety of techniques for such a reconstruction can be used. Any of the approaches can be modified to incorporate reconstructive measures for best functional and cosmetic outcome.
Martí-Pagès and colleagues in Barcelona described an approach that included a more limited approach that involves an incision extending from the helical root, along the posterior aspect of the tragus, down under the earlobe, and up over the posterior aspect of conchal cartilage. There are no posterior occipital limbs extending to the hairline. We have been using this approach for the past 8 years with great success. We always combine this incision with a reconstructive technique to cover the exposed parotid gland, fill in the surgical defect, and create facial symmetry, and have termed it microparotidectomy (MIP) ( Figs. 1–12 ). As with the group in Barcelona, we find this approach excellent in allowing wide access to the entire parotid gland and full visibility of the facial nerve. We have also been able to preserve the posterior branches of the GAN, dissect and transpose facial nerve branches to access the deep lobe tumors, as well as access the SCM muscle for local parotid reconstruction. When necessary, we make modifications to this incision to allow for mastoidectomy for facial nerve exposure, extending the incision to the hairline for larger or recurrent tumors, and even extending an inferior limb to perform more extensive surgery, including neck dissection. MIP serves as the primary approach in our practice for treatment of benign parotid tumors.
