40

Print Preview

Chapter 40

Salivary Glands

Chris H. J. Terhaard

The salivary glands consist of the three large, paired major glands—parotid, submandibular, and sublingual (Fig. 40.1)—and many smaller, minor glands located throughout the upper aerodigestive tract. Salivary gland malignancies make up only approximately 0.4% of all cancers, and account for less than 5% of the annual incidence of head and neck malignancies in the United States. The international variation in the incidence is between 0.4 and 2.6/100 per year (72), with a mean of approximately 1/100. No significant change in incidence has been shown in the past decades in the United States and Sweden (72,97).

Anatomy

Major Salivary Glands

Parotid Gland

The parotid gland is located superficial to and partly behind the ramus of the mandible and covers the masseter muscle. Superficially, it overlaps the posterior part of the muscle and largely fills the space between the ramus of the mandible and the anterior border of the sternocleidomastoid muscle. One or more isthmi that wrap around the branches of the facial nerve connect the superficial and deep lobes of the gland. The nerve enters the deep surface of the gland as a single trunk, passing posterolateral to the styloid process. It usually leaves the gland as five or more branches, emerging at the anterior, upper, and lower borders of the gland. The facial nerve runs superficial to the main blood vessels that traverse the gland but is interwoven within the glandular tissue and its ducts. Thus, removal of all or part of the parotid gland demands meticulous dissection if the nerve is to be spared.

The parotid gland contains an extensive lymphatic capillary plexus, many aggregates of lymphocytic cells, and numerous intraglandular lymph nodes in the superficial lobe. Lymphatics drain from more lateral areas on the face, including parts of the eyelids, diagonally downward and posteriorly toward the parotid gland, as do the lymphatics from the frontal region of the scalp. Associated with the gland, both superficially and more deeply, are parotid nodes. These drain downward along the retromandibular vein to empty in part into the superficial lymphatics and nodes along the outer surface of the sternocleidomastoid muscle and in part into upper nodes of the deep cervical chain. Lymphatics from the parietal region of the scalp drain partly to the parotid nodes in front of the ear and partly to the retroauricular nodes in back of the ear, which, in turn, drain into upper deep cervical nodes (46) (Fig. 40.2).

Submandibular Gland

The submandibular gland largely fills the triangle between the two bellies of the digastric and the lower border of the mandible and extends upward deep to the mandible. It lies partly on the lower surface of the mylohyoid and partly behind the muscle against the lateral surface of the muscle of the tongue, the hypoglossus. The submandibular gland has a larger superficial part, or body, and a smaller deep process. The inferior surface is adjacent to the submandibular lymph nodes, and the deep process of the submandibular gland lies between the mylohyoid laterally and the hyoglossus medially, and between the lingual nerve above and the hypoglossal nerve below (7). Bimanual palpation with one finger in the floor of the mouth and one under the edge of the mandible facilitates clinical detection of masses in this gland.

A rich lymphatic capillary network lies in the interstitial spaces of the gland. From the lateral and superior portions of the gland, lymph flows to the prevascular or preglandular submandibular lymph nodes. The posterior portion of the gland gives rise to one or two lymphatic trunks, which follow the facial artery and go directly to the anterior subdigastric nodes of the internal jugular chain (40,43). The nodes overlying the submandibular gland, followed by the subdigastric and high midjugular lymph nodes, are those involved in nodal metastases.

Sublingual Gland

This smallest of the three major salivary glands, along with many minor salivary glands, lies between the mucous membrane of the floor of the mouth above and the mylohyoid muscle below, the mandible laterally, and the genioglossus muscles of the tongue medially (Fig. 40.1). This is a rare site for malignant neoplasms; they are difficult to distinguish from cancer of the floor of the mouth accounting for fewer than 2% of all reported cases of salivary gland tumors (72,93). The sublingual gland drains either to the submandibular lymph nodes or more posteriorly into the deep internal jugular chain between the digastric and omohyoid muscles. Rarely, the lymphatics of the sublingual gland drain into a submental node or supraomohyoid jugular node (40).

Minor Salivary Glands

Minor salivary glands are widely distributed in the upper aero-digestive tract, palate, buccal mucosa, base of tongue, pharynx, trachea, cheek, lip, gingiva, floor of mouth, tonsil, paranasal sin-uses, nasal cavity, and nasopharynx.

Epidemiology

Seventy percent of all salivary tumors arise in the parotid gland, 8% in the submandibular gland, and 22% in the minor salivary glands (93). The proportion of malignant tumors increases from parotid (25%) to submandibular (43%) to minor salivary glands (65%) (57,93). There is a preponderance of benign tumors in women; malignant tumors exhibit an equal sex distribution. Patients with benign tumors are younger (mean age, 46 years) compared with those with malignant tumors (mean age, 54 years), with a trend to an older age for submandibular and minor salivary gland locations (93). Two percent to 3% of salivary neoplasms occur in children, in whom half of the tumors are malignant (21). The majority of cancers are located in

the parotid gland, with mucoepidermoid cancers predominating (88).

Etiologic factors are not clearly defined. Nutrition may be a factor because Eskimos in the Arctic, who have low intake of vitamins A and C, have a high incidence (56). Cigarette smoking and alcohol consumption is in general not related with salivary gland cancer (68), although cigarette consumption more than 80 pack-years may contribute to salivary gland cancer (98). Irradiation can also be a cause, as evidenced by the increased incidence in survivors of the atomic bombs dropped on Hiroshima and Nagasaki, and in those irradiated to the head and neck for benign conditions during childhood (67,83,84,111). Saku et al. (83) studied salivary gland tumors in atomic bomb survivors of Hiroshima and Nakasaki. Two-thirds of all cases were parotid, and the remainder was equally distributed between submandibular and minor salivary glands. Mucoepidermoid cancer and Warthin's tumor (benign) were particularly elevated compared with nonexposed persons, and disproportionately high at high radiation doses. Modan et al. (67) found a clear dose-response effect in a matched control study of patients who had low-dose head–neck irradiation in childhood; there was a 2.6-fold increase of benign tumors and a 4.5-fold increase of cancer. The majority of these tumors are mucoepidermoid cancers (84,111); however, less than 1% of salivary gland tumors may be caused by former irradiation (8).

Workers in various occupations experience an increased risk of salivary gland cancer (47). For women employed as hairdressers or working in beauty shops, a significant elevated risk was observed in a study by Swanson and Burns (98). A correlation of incidence with ultraviolet exposure remains controversial (97,98).

Women with salivary gland cancer may have a 2.5 elevated breast cancer risk (50), probably confined to women with salivary gland cancer before age 35 (97). After treatment for salivary gland cancer, an increased risk for subsequent oropharyngeal, thyroid, and lung cancer is noted (97).

Natural History

Local invasion is the initial route of spread of malignant tumors of the salivary glands, depending on location and histologic type. For parotid tumors, this may result in fixation to structures in around 20% of cases (75). Skin invasion is more often seen in parotid tumors (10%), compared with submandibular tumors (3%) (101).

Approximately 25% of patients with a malignant parotid salivary gland tumor present with facial palsy from cranial nerve invasion (31,75,93,100,105).

A detailed study of the Dutch Head and Neck Oncology Group (NWHHT) concerning patients with a salivary gland malignancy found an overall incidence of clinically positive nodes of 14% and clinically occult, pathologically positive nodes in an additional 11% of patients (101). This percentage depends on the number of neck dissections performed, the tumor location, histology, and T-stage. The number of elective neck dissections performed varies between the tumor locations. Stennert et al. (94) performed a neck dissection in all malignant parotid tumors and found 53% unilateral positive nodes and 0% contralateral nodes. In selected patients in other studies, the percentage positive nodes varied between 20% (115) and 38% (100). Lymph

node involvement for parotid malignancies, combining clinical and pathologic information, is around 25% (10,62,79,100,105). Resection of submandibular tumors is combined with a (partial) neck dissection in most cases. Pathologic neck nodes may be seen in up to 42% of cases (100). Salivary gland tumors arising in the oral cavity produce an incidence of cervical node metastases of less than 10% (57,65,74,100). Nasopharyngeal salivary gland tumors have a high risk of occult metastases (50%) (85). The risk of positive findings in the neck may be based on a combination of T-stage and histology (79,100). The highest risk is seen for squamous cell, undifferentiated cancer, and salivary duct cancer (79,100). There is an intermediate risk for mucoepidermoid cancer and a low risk for acinic cell, adenoid cystic carcinoma, and carcinoma ex pleomorph adenoma (100). A 15% risk is found for T1 tumors, 26% for T2, and 33% for T3–4 (100). An example of a rating scale to estimate the risk of positive neck nodes, based on tumor location, T-stage, and histologic type, is shown in Table 40.1.

Distant metastases overall are encountered in 3% of patients at presentation and in 33% after 10 years (101). They are fairly common with adenoid cystic, salivary duct, squamous cell, and undifferentiated carcinomas; in the case of adenoid cystic carcinomas, they may occur quite late in the course of the disease, without recurrence of the primary tumor (42,52,74,92,101). Distant metastases are primarily to lung, bone, and occasionally to the liver (101). Reported incidence of distant metastases in patients with adenoid cystic carcinoma after 10 years of follow-up is around 40% (64,92,101). Five years after diagnosis of distant metastases of adenoid cystic carcinoma, more than one third of the patients are still alive; 10% are alive after 10 years. An update of the survival data of the NWHHT study after diagnosis of distant metastases of salivary gland cancer is shown in Figure 40.3.

Clinical Presentation

Three of four parotid masses are benign (93). Patients most often have a painless, rapidly enlarging mass, often present for years before a sudden change in its indolent growth pattern prompts the patient to seek medical attention. Duration of clinical symptoms before diagnosis may last more than 10 years (93,101). For malignant tumors, the median duration of clinical symptoms generally is shorter (3 to 6 months) (9,101) compared to that of benign tumors, although for some minor salivary gland tumors, median periods of 2 year have been reported (65,85).

Pain is more frequently associated with malignant disease (93). Although as many as one third of parotid cancers may have facial nerve involvement, only 10% to 20% of patients complain of pain (75,93,101). Pain may appear with involvement of deeper structures (masseter, temporal, and pterygoid muscles). Rarely, tumors of the parotid may involve the base of skull and cause intractable pain and paralysis of various cranial nerves.

The signs and symptoms associated with tumors of the minor salivary glands vary because of their diverse locations. The distribution of presenting sites for 492 cases of minor salivary gland tumors, 88% of which were malignant, is shown in Table 40.2 (93). Most are intraoral, and a painless lump is the most common presenting symptom. For tumors arising in the nasal cavity or sinuses, facial pain is the most common presenting symptom, followed by nasal obstruction. Laryngeal primary tumors most frequently cause hoarseness or voice change.

Clinical features suggesting a malignant salivary gland tumor are rapid growth rate, pain, facial nerve palsy, childhood occurrence, skin involvement, and cervical adenopathy.

Diagnostic Work-Up and Staging

Major Salivary Glands

The diagnostic work-up of major salivary gland tumors includes a careful history and physical examination, with particular

attention to signs of local fixation or regional adenopathy. Computed tomography (CT) scans are useful in evaluating the extent of lesions involving the parotid gland, especially the deep lobe. Magnetic resonance imaging (MRI) is superior to other modalities, especially when malignancy is suspected (Fig. 40.4). T1-weighted images are excellent to assess the margins, deep extent, and patterns of infiltration because the (fatty) background of the gland is hyperintensive. In general, benign tumors are hyperintensive, and malignant tumors are intermediate or low intensivity at T2-weighted MR images (73,114). The MRI has a sensitivity of 87% and a specificity of 94% (73). Perineural invasion of adenoid cystic carcinoma may be evaluated with both CT (foraminal enlargement) and MRI (fat-suppressed T1-weighted images). Such findings may well change the surgical approach and treatment regimen (15,114).

Fine-needle aspiration in the diagnosis of parotid and submandibular salivary gland tumors is a reliable procedure. The sensitivity for malignancy varies between 80% and 90%; the specificity is more than 90% (20,73,76,95). The negative predictive value for malignancy, however, is around 70% to 75% (20,73). False-negative findings may be seen as result of lack of representative material or a cyst. In these cases, ultrasound fine-needle aspiration is advised (76). The relative low negative predictive value of fine-needle aspiration will be improved if MRI and fine-needle aspiration are combined (74). Fine-needle aspiration has been quite accurate in the diagnosis of benign salivary gland tumors (20,95).

The sixth edition of the manual of the American Joint Committee on Cancer (2) and the sixth edition of the classification system of the International Union Against Cancer (90) are identical for major salivary glands (Table 40.3). They are based on size, extension, and nodal involvement. Relative survival rates for major salivary gland cancer according to stage are shown in Figure 40.5.

Minor Salivary Glands

Various radiographic studies may be used, including plain films, to ascertain bone erosion in advanced lesions, and CT and MRI scans may be used to evaluate depth and contiguous involvement. The definitive diagnostic procedure is an excisional biopsy, particularly if malignancy is clinically expected. Unplanned incisional biopsies should be avoided, and fine-needle biopsies are impractical because of the polymorphism of most malignant salivary gland tumors.

A formal staging system has not been developed for minor gland tumors. The same staging system for minor salivary glands as for squamous cell carcinoma in sites other than the parotid or submandibular glands may be used. The American Joint Committee on Cancer and International Union Against Cancer classification and stage regrouping system has been reported to be a major long-term outcome predictor in minor salivary gland carcinoma (104).

Pathologic Classification

The histologic classification of salivary gland neoplasia is very demanding for the head and neck pathologist. In 1991 the World Health Organization classification for salivary gland tumors was expanded. Various types of carcinomas were distinguished based on recognition, prognosis, and treatment, discussed more in detail by Seifert and Sobin (87) (Table 40.4). Classification may be difficult, as shown in a re-evaluation of 101 intraoral salivary gland tumors by experienced pathologists; major disagreement was seen in 8 and there was minor disagreement in 33 (103).

Neoplastically transformed myoepithelial cells play a role in the development of monomorphic and pleomorphic adenomas, adenoid cystic carcinomas, mucoepidermoid carcinomas, and the rare myoepithelomas (7). The reserve cell system of the intercalated and excretory duct is thought to be the site of origin of most neoplasms (7).

Salivary gland carcinomas may be graded as low and high malignant, particularly for mucoepidermoid tumors. However, there is a disparity for grading, even among experienced pathologists (13). Low-grade mucoepidermoid, polymorphous low-grade adenocarcinoma (PLGA), epithelial-myoepithelial, and acinic cell carcinomas comprise a group of low-to-moderate malignancy; high-grade mucoepidermoid, malignant mixed, adenoid cystic, squamous, undifferentiated, and salivary duct carcinomas represent more high-grade malignancies (105). The percentage of the histologic subtypes varies from series to series, and from the localization of the tumor (Fig. 40.6). In parotid tumors in children and adults, the most common malignant subtype is the mucoepidermoid (10,21,37,88,91,108). Acinic cell cancer derives from cells of the terminal ducts and intercalated ducts. Grading for acinic cell cancer is controversial (45). Most tumors (86%) are located in the parotid gland (Fig 40.6) (45). Adenoid cystic carcinoma is most common in minor salivary glands (Fig. 40.7) (57,74,101,104), followed by

the submandibular gland (Fig. 40.6) (11,96,101,108). Perineural invasion is common in adenoid cystic carcinoma (107). The adenoid cystic variety has a tubular pattern that has been associated with the best prognosis, a cribriform pattern with an intermediate prognosis, and a solid pattern with the worst prognosis (28). The PLGA and salivary duct carcinomas are the most common new subtypes of the World Health Organization 1991 classification. PLGA is a solid, ovoid, nonencapsulated mass with a highly variable growth pattern (Fig. 40.8). Most are located in the palate, and the prognosis generally is good (17,29). Salivary duct carcinoma resembles ductal breast cancer morphologically (Fig. 40.8). They derive from excretory duct cells. They are usually located in the parotid gland and are highly aggressive (42,48,52). There is also a low-grade subtype (23).

Prognostic Factors

A number of prognostic variables have been studied in the management of salivary gland cancer. In these studies, multivariate analyses have been performed considering locoregional control, distant metastases, and survival. Results of studies with sufficient number of patients and follow-up have been summarized in Table 40.5. Local control and overall survival is influenced by site, favoring tumors of the oral cavity (101,102). T- and N-stages are independent variables for locoregional control, distant metastases, and survival, regardless of site (10,37,57,58,62,65,66,74,75,92,101,102,105). In the NWHHT study, histologic type was an independent prognostic factor for distant metastases (101).

Oncogene expression has been evaluated in the search for additional prognostic factors. Expression of the oncoprotein p53 was found in an Italian study to be higher in malignant tumors than in benign tumors (35). Furthermore, tumors with moderate-to-high expression of p53 were more frequently associated with regional and distant metastasis and a lower disease-free and overall actuarial survival rate, compared with patients with no p53 expression. Univariate and multivariate analyses confirmed the independent prognostic value of p53 expression. Vascular endothelial growth factor significantly correlates with p53 expression and is an independent prognostic factor

for survival for salivary gland cancer (61). Overexpression of HER-2/neu was seen in approximately one third of mucoepidermoid carcinomas in a series of 50 parotid gland cancers studied at the University of Southern California (78) and in 20% in a series of 50 salivary duct carcinomas in a study from Germany (52). Overexpression was seen and appeared to be an independent marker of poor prognosis. It also has been similarly associated with poor prognosis in carcinomas of the breast, ovary, and endometrium. Another molecular feature studied in relationship to prognosis was the DNA content in adenoid cystic carcinomas. DNA aneuploidy is correlated with the solid type, and thus with poor prognosis (28). Franzen et al. (32) found a correlation of grade with aneuploidy as well as stage.

Major Salivary Glands

The survival of patients with submandibular cancers is inferior to that of parotid cancers according to a study by Spiro et al. (91). Extraglandular extension (10,37) and skin invasion (70,101,105) in parotid cancers results in decreased disease-free survival. More advanced age was found to be a negative prognostic factor for locoregional control in some studies (58,75,79) and for disease-free and overall survival in most studies (10,58,62,75,91,105,106). Impairment of function of the facial nerve is a known prognostic factor, not only influencing locoregional control (37,70,101), but also disease-free survival (36,62,70,105). Pain at presentation may be associated with reduced disease-free survival (105). Perineural invasion and pain are closely related: not pain, but perineural growth, in some studies, is an independent prognostic factor for distant metastases (101) or disease-free survival (37,44).

The importance of histologic subtype for major salivary gland cancer varies in published studies. In most studies, histologic types are subdivided into low and high grade. The main prognostic significance of grading relates to disease-free survival (10,62), although grade was not a prognostic factor in most studies. The best prognosis is seen for acinic cell and (low-grade) mucoepidermoid cancer (91,101), the worst for undifferentiated (70,101) and squamous cell cancer (91,101). At the Netherlands Cancer Institute, a prognostic score for patients with parotid carcinoma was developed and validated adequately with the NWHHT database (105,106). The preoperative prognostic score was based on a weighted combination of prognostic factors (age, pain, clinical T- and N-stages, skin invasion, and facial nerve dysfunction); histology and grading were not incorporated. Four subgroups were formed with markedly different prognoses. In the postoperative score, perineural invasion and positive surgical margins were also included. Positive or close surgical margins result in an increase in local recurrence rate(77,96,101,102).

Radiation therapy in addition to surgery improves locoregional control in patients with adverse prognostic factors (70,75,81,100,102). Improvement of survival has only been shown in two studies (11,70), and for stage III and IV major salivary glands in a matched-pair analysis (4).

Minor Salivary Glands

The poorest prognosis is associated with adenoid cystic carcinoma (9,65). Stage, base of skull involvement, and bone invasion are risk factors for locoregional recurrence and survival in minor salivary gland cancers (9,24,25,57,65,77). Locoregional control may be improved by adding postoperative radiotherapy (74).

General Management

The general management of salivary gland malignancies in most patients includes surgical excision followed by radiation therapy for unfavorable prognostic factors (Table 40.5) (3). Postoperative radiotherapy to enhance local control is recommended for T3–4 tumors, close or incomplete resection, bone involvement, perineural invasion, high-grade cancer, and

P.881

recurrent cancer (9,37,58,70,96,101). To date, adjuvant chemotherapy has not been considered efficacious. For advanced, inoperable, and recurrent salivary gland cancers, primary neutron therapy may lead to superior local control rates, compared with primary photon therapy, without evidence of improved survival rates (24,49,60). The use of conventional radiation therapy along with hyperthermia has been reported to have similar efficacy in this patient population (34).

Major Salivary Glands

Surgical technique depends on location and extent of primary disease and regional adenopathy. Preservation of the facial nerve, at least partially, followed by postoperative radiotherapy is the preferable treatment unless the facial nerve is involved by tumor (6). Aggressive surgery does not improve disease-free survival. A decrease in extended surgery, resulting in a decrease of sacrifice of the facial nerve, has been shown in the course of years (91). Cable facial nerve grafting with the greater auricular or sural nerve graft decreases the incidence of facial palsy postoperatively, especially if branches and not the main trunk are involved (14,56). Adjuvant postoperative radiotherapy has no negative effect on facial nerve function (14).

Surgical treatment includes neck dissection in cases of clinically positive nodes, followed by postoperative radiotherapy (100). The risk of occult nodal disease depends on T-stage and histologic type. As shown in the scoring system in Table 40.1, the decision to treat the neck for parotid tumors will be indicated by a score of at least 4 (100). When local prognostic factors indicate postoperative radiotherapy, no elective neck dissection has to be performed; the neck nodes will also be irradiated (30,94). Parotid tumors with facial nerve weakness are associated with

frequent occult neck nodes; elective treatment is also indicated (30). In most cases, elective neck dissection of level I-III combined with a local resection is performed for submandibular tumors. There is no indication for neck dissection for T1 acinic or T1 adenoid cystic tumors (Table 40.1) (96,100).

Minor Salivary Glands

The treatment of minor salivary gland tumors varies with location but usually involves an attempt at adequate surgical excision first. Irradiation has been used in surgically inaccessible sites or combined with surgery because of locally aggressive tumor behavior and the occurrence of incomplete resection (9,38,74,101). For tumors arising in the palate, tongue, floor of the mouth, oral cavity, or oropharynx, surgical exposure is readily available, and resection usually can be accomplished with acceptable morbidity. Tumors arising in the posterior nasal cavity, nasopharynx, or sphenoid region, however, are relatively inaccessible and are mostly treated with radiation therapy (38). Elective neck treatment is usually not indicated (38,74,100), except for tumors of the floor of mouth, oral tongue, pharynx, and larynx (74,85,115). Surgery alone may be used to treat early-stage hard palate lesions without evidence of positive margins, perineural spread, or bone invasion; simple excision must be avoided (9). Patients with adenoid cystic carcinoma can have a long natural history with late recurrences (38), and consideration should be given to careful surgical reconstruction and rehabilitation because even patients who are not cured can live many years before dying of disease (Fig. 40.3) (101). Occasionally, a patient may present after simple excision (shelling out) of a lesion and the pathologic examination shows adenoid cystic carcinoma. If re-excision would cause significant functional or cosmetic sequelae, irradiation alone may be used (74). However, simple excision is not recommended as the initial management of these tumors because of the potential for a significant volume of residual disease.

Radiation Therapy Techniques

Pleomorphic Adenoma

The pleomorphic adenoma (benign mixed tumor) is histologically benign, occurs frequently in a relatively young population, and comprises 65% to 75% of all parotid epithelial tumors (57,91). Standard therapy has been conservative (superficial) parotidectomy, with recurrence rates of about 0% to 5% (112). Simple excision results in a high recurrence rate of around 25% as focal capsular exposure occurs in virtually all cases (112). In the past at some institutions, local excision and radiation therapy have been used to lower the frequency of facial nerve injury and Frey's syndrome (22). Dawson and Orr (22) reported results for 311 patients. They found a 2.5% recurrence rate at 10 years and an additional 5.5% by 20 years. None of the patients had malignant recurrences at 10 years, 0.5% had such recurrences at 15 years, and 3% had recurrences at 20 years. The later recurrences were more likely to show malignant transformation. The authors concluded that the primary treatment should be surgery because of the patient's young age, benign histology, and the remote possibility of subsequent radiation-induced malignancy. However, certain patients may be referred for radiation therapy (63). Indications for postoperative irradiation may include recurrent disease; microscopically positive margins after surgical resection; and large, deep-seated lesions that may not allow complete surgical excision with adequate margins or would require sacrificing the facial nerve (16,22,63,80). Radiotherapy may decrease the risk of a second recurrence in case of multinodular recurrence only, not for uninodular disease (81). The entire parotid area should be irradiated with a dose of 50 to 60 Gy in 5 to 6 weeks.

Parotid Gland

The volume of irradiation is determined by pathologic findings, such as perineural invasion of a major nerve. Typically, the entire ipsilateral parotid gland is delineated on the postoperative CT scan performed in a stabilization device (37,71). The delineation of the clinical target volume will be individualized based on the extent of the disease and surgery (37). The parapharyngeal space and the infratemporal fossa have to be covered adequately (71). The primary treatment volume includes the ipsilateral subdigastric nodal areas because the inferior pole of the parotid lies in this region (3). In general it is not necessary to treat the scar to full skin dose because only 1% of the patients have a scar failure (58). For very superficial localized tumors and in case of skin invasion, a bolus over the scar is required. In tumors with named perineural invasion (e.g., adenoid cystic carcinoma), it is important to cover the cranial nerve pathways from the parotid up to the base of the skull (39,96). Focal perineural invasion only is not an indication for routine inclusion of the nerve pathways (39). No clear relationship between dose and local control has been found. In general, a dose of at least 60 Gy postexcision is recommended (39,71,100), and at least 66 Gy (33 fractions) for positive margins (37,39).

The ipsilateral neck is treated after a neck dissection has been performed for positive nodes; level I-V should be included (100). There is no indication for bilateral elective neck treatment (94). The recommended postoperative dose for positive nodes is at least 60 Gy (30 fractions) (100). Elective irradiation of the neck should be considered for advanced T-stage, certain histologic subtypes (Table 40.1), facial nerve dysfunction at presentation, and recurrent disease. At least level Ib, II, and III should be included (5,30,100). A dose of around 46 to 50 Gy is recommended (3,37,100).

Three basic radiation therapy approaches are used, depending on available equipment: Conventional, three-dimensional conformal radiation therapy (3DCRT) planning procedure, and intensity-modulated radiation therapy (IMRT) planning. The first involves unilateral anterior and posterior wedged pair fields using 60Co or 4- to 6-MV photons (Fig. 40.9A). A slight inferior angulation of the beams avoids an exit dose through the contralateral eye. A simpler technique uses homolateral fields with 12- to 16-MeV electrons in combination with photons (37,41,113). Usually, 80% of the dose is delivered with electrons and 20% with 60Co or 4- to 6-MV photons to spare the opposite salivary gland, reduce mucositis, and decrease the skin reaction produced by electrons (Fig. 40.9B). Yaparpalvi et al. (113) compared nine conventional treatment techniques. Ipsilateral wedge pair technique with 6-MV photons, wedged anteroposterior and posteroanterior and lateral technique with 6-MV photons, and mixed beam using 6-MV photons and 16-MeV electrons (1:4 weighted) were most optimal, considering dose homogeneity within the target and dose to normal tissues. Electron beam (9 to 12 MeV) and tangential photon fields are effective conventional techniques for sparing the underlying spinal cord (from doses more than 45 Gy) and the opposite parotid gland in elective neck irradiation. Conventional techniques do not allow for tissue heterogeneity (air cavity, dense bones, and tissues); underdose and overdose may be seen.

After outlining of the target volumes and critical normal tissues on the planning CT scan, a more conformal 3DCRT plan by the use of geometrically shaped beams of uniform intensity may be reached (71). More normal tissue may be spared with this technique (71). Probably the most conformal radiation technique is IMRT. It can produce convex dose distributions and steep dose gradients. Five- to seven-field inverse IMRT allows excellent coverage of the tumor with sparing of mandible,

cochlea, spinal cord, brain, and oropharynx (12,71), compared with conformal 3DCRT. Figure 40.10 shows a comparison of 3DCRT and IMRT planning for a postoperative radiotherapy plan for a parotid cancer treated with a dose of 66 Gy. The mean dose to the mastoid, meatus acusticus externus, and contralateral parotid gland was 53 and 43 Gy, 57 and 51 Gy, 1 and 9 Gy, for 3DCRT and IMRT, respectively. The maximum dose to the cochlea was 39 and 32 Gy, respectively.

Submandibular Gland

Except for small acinic cell and adenoid cystic cancer (Table 40.1), the neck nodes level I-IV (5) should be irradiated electively, following the indications outlined for parotid tumors; technical considerations are similar. Bilateral fields may be required for tumor extension toward the midline. If there is no gross residual tumor or perineural invasion, 50 Gy in 5 weeks should be adequate for microscopic disease. If there is named perineural invasion of a major nerve, a tumor dose of 60 to 66 Gy in 6 to 6.5 weeks is recommended, and the nerve path to the base of skull should be treated, preferably by 3DCRT or IMRT. For an adenoid cystic carcinoma of the submandibular gland with only focal perineural invasion, an attempt to encompass the base of the skull would require a significant change in the treatment volume and may not be warranted because of potential morbidity and the low rate of relapse at that site (38). An example of 3DCRT for a T2 adenoid cystic carcinoma of the submandibular gland is shown in Figure 40.11.

Minor Salivary Glands

The radiation therapy technique for treating minor salivary gland tumors depends on the area involved and is similar to the treatment for squamous cell carcinomas in these areas, with two significant exceptions. First, when a named branch of a cranial nerve is involved by adenoid cystic carcinoma, the nerve pathways to the base of the skull should be electively treated. When only focal perineural invasion of small unnamed nerves is present, treatment of the base of the skull depends on the site. Second, for tumors of the palate or paranasal sinuses, the base of the skull is included because of its proximity to the tumor bed. In case of an adenoid cystic carcinoma with perineural invasion, IMRT may reduce the high-dose volume, compared to conventional bilateral opposed fields; Figure 40.12 shows an example for a patient with a minor salivary gland cancer of the palate. IMRT is a useful strategy for irradiating minor salivary gland sites such as the ethmoid sinuses while sparing the optic pathways (19).

Also, because the incidence of lymph node metastases is usually lower than that for squamous cell carcinomas of similar size, the radiation therapy fields are rarely extended to cover such areas if there are no palpable lymph node metastases. Indications for treating the neck are a primary tumor that arose in the tongue, floor of the mouth, pharynx, or larynx (74), and the neck was not dissected, or after resection of metastatic neck lymphadenopathy.

For patients receiving postoperative irradiation after surgical resection, a dose of 60 Gy is given for negative margins and 66 Gy for microscopically positive margins. For gross residual disease after surgery or for lesions treated with irradiation

alone, a total dose of 70 Gy is recommended at 2 Gy per fraction.

Results of Therapy

Surgery Plus or Minus Postoperative Radiotherapy

Tables 40.6, 40.7 and 40.8 list local control rates and 5- and 10-year survival rates for several series reporting the surgical, irradiation, and combination treatment of carcinomas of the major and minor salivary glands. Little adverse effect of delay between surgery and radiotherapy may be predicted for what are, in general, slow-growing salivary gland cancers. In only two studies, one concerning submandibular cancer (96) and another for minor salivary gland (38), impaired locoregional control rates were seen for a delay of more than 6 weeks, which was not confirmed in the Dutch study (100). The prognosis for children with a malignant salivary gland cancer (mostly mucoepidermoid cancer of the parotid gland) is excellent, with a 10 year overall survival of more than 90% (88). Most are treated with surgery alone because of the possible risk on radiation-induced malignancies.

Long-term follow-up is recommended because failures may appear after 5 years, especially for minor salivary gland tumors (17,29,57,101). Recurrent tumors in general are more difficult to control than are primary ones, so high initial locoregional control rates should be the goal (101). Because of high rates of local failure of approximately 40% for parotid, 60% for submandibular, and 65% for minor salivary glandswith surgery alone in the past (93), many institutions have advocated postoperative irradiation especially to reduce the incidence of local failure. Local tumor control appears to be improved by the combination of surgery and irradiation, although randomized, controlled trials have not been performed. Evidence of a positive role of postoperative radiotherapy is based on retrospective studies and a matched-pair analysis. In the study by Armstrong et al. (4), postoperative radiotherapy significantly improved locoregional control (from 17% to 51% for stage III-IV), not for stage I and II major salivary gland cancer. Locoregional control for patients with positive nodes increased from 40% to

69%. In most studies, an imbalance in prognostic factors is seen comparing surgery alone with combined therapy, favoring surgery alone. Despite this imbalance, locoregional control with combined surgery and postoperative radiotherapy is superior to surgery alone for patients with negative prognostic factors, irrespective of site (33,66,100,101,102). In the nationwide Dutch study, the relative risk for surgery alone, compared with combined treatment, was 9.7 for local recurrence and 2.3 for regional recurrence (100). In a study from Denmark, the relative risk of no radiotherapyversus radiotherapy was 4.7 for locoregional control (102). Postoperative radiotherapy is particularly effective if there are close and microscopic positive resection margins, enhancing local control from around 50% to 80% to 95% (33,41,77,100,102). Comparable results are noted for T3-T4 tumors and pathologically confirmed bone and perineural invasion (66,100). However, for a T1 or T2 tumor that was completely resected with no bone or perineural invasion, surgery alone will result in more than 90% 10-year local control rate, and radiotherapy is not indicated (100).

Treatment results also may depend on histopathologic status. However, after review, histologic type may change, even among experienced pathologists. In general, the best prognosis is shown for acinic cell and mucoepidemoid cancer, with a 15% risk of distant metastases after 10 years and a 10-year locoregional control rate of around 85%. Ten-year overall survival is around 80% and 65%, respectively (11,45,93,101). In one of three patients, postoperative radiotherapy is indicated (45,101). Squamous cell and undifferentiated tumors have been associatedwith a 10-year overall survival of 35% or less, caused by a high risk of distant metastases (35% and 50%, respectively) and locoregional recurrence (77,93,101). Postoperative radiotherapy is indicated in all cases to improve locoregional control. The intermediate-risk group consists of adenoid cystic cancer and cancer ex pleomorphic adenoma. Distant failure after 10 years is around 35% (39,92,101). Although the risk of nodal recurrence is low (5% to 10%), local recurrence is diagnosed more often (20% to 30%). A precipitous decrease in relapse-free survival is noted among 5 (around 70%), 10 (around 50%),

P.886

and 15 years (around 45%) for patients with adenoid cystic carcinomas, which are well known for late recurrences (39,101). Significant improvement was reported in local control for adenoid cystic cancer with combined surgery and irradiation in several studies (33,39,81,89,96), regardless of site. Local tumor control rates with combined modality therapy for these tumors approach 85% to 90% at 10 years. Postoperative radiotherapy is also able to improve locoregional control rates with about 20% for high-grade tumors (62,81,96).

In the World Health Organization classification of 1991, among others, two new subtypes were described that are diagnosed relatively frequently. PLGA is situated solely in the palate.

Treatment consists of wide local excision. In a report by Castle et al. (17), treatment results of 164 tumors were analyzed, with 90% treated with surgery alone. Local control was 90%, with only few patients dying from PLGA. However, local failures may be seen even after long follow-up. In a series from Evans and Luna (29) of 40 patients with PLGA, local recurrence was seen in 43% of patients treated with surgery alone, mainly because of close and microscopic positive resection margins. No recurrence was seen in the nine patients treated with postoperative radiotherapy.

Salivary duct carcinoma is a very aggressive disease, and postoperative locoregional radiotherapy is indicated in all cases. Most patients die of disease, despite often successful locoregional combined therapy. Because of a high percentage of distant metastases, 5-year survival is only around 10% to 15% (42,52). The prognosis correlates with HER-2/neu receptor status; 3-year survival is 56% and 17% for (+)HER-2/neu and (+++)HER-2/neu, respectively (52).

Minor salivary gland tumors of the oral cavity have a more favorable prognosis than paranasal sinus tumors (maxillary and ethmoid sinus and nasal cavity) (38). Patients with hard palate lesions tend to be diagnosed when they have small asymptomatic lumps, which are easily detected on physical examination. On the other hand, paranasal sinus tumors usually do not cause symptoms until they are locally advanced. The surgical approach for these tumors is more difficult, with a greater chance for leaving behind residual disease, leading to high recurrence rates. A combined approach with surgery and postoperative irradiation is recommended.

Primary Radiotherapy

The poor results for salivary gland cancer with irradiation alone in several series have been attributed to the use of primary radiotherapy for patients with locally advanced lesions or distant metastases at presentation, who were essentially treated for palliation. Locoregional control rates after conventional photon or electron therapy are around 25% (49,60,66). For treatment with photons with curative intent, a clear dose-response relationship has been described (100). A dose of 66 to 70 Gy may result in 50% 5-year local control. Wang and Goodman (109) reported local control as high as 85% with accelerated hyperfractionated photon therapy. The follow-up was rather short, and the results have not been updated (109). The generally slow rate of regression of advanced salivary gland tumors have made them a logical target for alternative radiation therapy approaches, such as fast neutrons.

Neutron Therapy

Patients with inoperable primary or recurrent major or minor salivary glands were included in the RTOG-MRC randomized phase III clinical trial. Patients were randomized between 70 Gy for 7.5 weeks or 55 Gy for 4 weeks photon therapy and neutron

therapy. The study had to be stopped because of a statistically significant difference in 2-year locoregional control, after inclusion of only 32 patients. The 10-year locoregional control probability was 17% after photon therapy, and 56% after neutron therapy (60). However, survival was identical. Late morbidity was somewhat higher for neutron therapy. Douglas et al. (24) of the University of Washington have published results of 279 patients treated with neutrons. Almost all patients had evidence of gross residual disease. Major and minor salivary gland sites were equally distributed. Total dose, administered with neutrons, varied from 17.4 to 20.7 Gy. The 6-year locoregional control and cause-specific survival were 59% and 49%, respectively, conforming to the results of most studies. Locoregional control was only 19% for base of skull involvement and 67% for no involvement. Locoregional control was 72% for minor sites and 61% for major sites. The 6-year actuarial grade 3 and 4 toxicity was 10%. Less severe late morbidity may occur if neutron therapy is combined with photons. A study from Heidelberg for advanced, inoperable, recurrent, or incompletely resected adenoid cystic carcinoma compared results of treatment with neutrons, photons, or mixed beam (49). Severe late grade 3 and 4 toxicity was 19% with neutrons, compared to 10% with mixed beam and 4% with photon therapy. The 5-year local control was 75% for neutrons and 32% for mixed beams and photons; survival was identical.

In an effort to improve poor results for tumors invading the base of skull, several new techniques have been developed. A combination of neutron therapy with, after a 4-week split, a Gamma Knife stereotactic radiosurgical boost has been used for tumors invading the base of skull (26). Local control of eight patients treated with this technique looks promising; however, follow-up was only 2 years. Another option is a combination of photons (54 Gy) and carbon ions (18 Gy) radiotherapy (86). In a series of 16 patients with adenoid cystic cancer invading the base of skull, the 3-year local control was 65%, without late effects exceeding grade 2. Longer follow-up results of these new techniques are awaited.

In conclusion, neutron beam therapy seems to be the treatment of choice for unresectable, residual, or recurrent salivary gland tumors. Despite high locoregional control, survival is not improved and late toxicity is of concern.

Systemic Therapy

The rarity of these neoplasms and their localized nature provide limited opportunities for trials with chemotherapy. In a review by Lalami et al. (59), they stated that chemotherapy has to be considered as palliative treatment and should only be given for disease-related symptoms and rapidly progressive disease. Cisplatin as monotherapy shows a 20% response rate for locoregional disease and only 7% for distant failures, with a duration of 6 to 9 months. A combination of 5-fluorouacil, cyclophosphamide, cisplatin, and doxorubicin gives a response rate of 50% (59).

Carcinoma ex pleomorphic adenomas and salivary duct carcinomas express androgen receptors in a high frequency (69). There may be a possible role for antiandrogen therapy, combined with other treatment modalities. However, the efficacy of this treatment option for some salivary gland cancers still has to be proven.

Expression of vascular endothelial growth factor is seen frequently in salivary gland cancer and is related with poor prognosis. Overexpression of HER-2/neu also correlates with poor prognosis, and a great variety between histologic types has been demonstrated (59). In the future, the role of molecular-targeted therapy for these salivary gland cancers has to be established.

Sequelae of Treatment

The most notable complication of treatment of parotid malignancies is facial nerve paralysis, which is often caused by the initial or a repeated surgical procedure. However, various series have shown that facial nerve sacrifice is rarely necessary, unless the nerve is directly involved by tumor, particularly when postoperative irradiation is given (33,77,91). When facial nerve sacrifice is required, facial nerve grafting and postoperative radiation therapy achieve comparable facial nerve function compared with unirradiated graft despite more negative prognostic factors (14). Other postoperative sequelae, such as salivary fistulae and neuromas of the greater auricular nerve, are sometimes seen. Frey's syndrome (i.e., gustatory sweating) may occur in a few patients after parotid surgery, but it is rarely bothersome (58).

Partial xerostomia after irradiation of the parotid gland is frequently observed and may be permanent. Trismus may result from radiation-induced fibrosis of the temporomandibular joint or the masseter muscles. It usually occurs when there is extensive tumor infiltration of the masseter muscle and high doses are given. Data on dose-response relationship for radiation-induced hearing impairments are sparse. In a study by Chen et al. (18), with 21 patients treated for malignant parotid tumors, a significant hearing loss was noted after a cochlear dose of ≥60 Gy in 60%, and in no patient after a dose <60 Gy. Conductive hearing loss was caused by serous effusion in the middle ear and/or obstruction of the tuba Eustachius. In general, a dose as low as possible (<30 Gy) should be attempted (53).

Garden et al. (38) reported complications of irradiation in 51 of 160 patients receiving postoperative irradiation for minor salivary gland tumors. The most common complication was decreased hearing in 26 patients, 20 of whom had myringotomies or myringotomy tubes placed for serous otitis media. Bone necrosis or exposure was observed in several patients; however, this complication has been seen infrequently during the past decade with improved radiation therapy techniques and treatment of multiple, as opposed to single, fields per day. Complications to the eyes or optic pathways were most common in patients with paranasal sinus primary tumors. At least six cases of contralateral optic atrophy occurred. Other eye complications included dry eye syndrome, nasolacrimal duct obstruction, cataract, retinopathy, and perforated globe. To reduce the incidence of bilateral blindness, the dose to the optic chiasm and contralateral optic nerve is limited to 54 Gy. In patients with extensive tumor involvement of the orbit, it may be preferable to remove the eye surgically rather than to subject the entire orbit to high doses. Radiation-induced injury to the visual pathway is dose-dependent. None of the patients receiving a dose of less than 50 Gy develop optic neuropathy or chiasm injury, whereas the 10-year actuarial incidences of optic nerve chiasm injury is 5% and 30% for patients receiving 50 to 60 Gy and 61 to 78 Gy, respectively (54).

Radiotherapy of tumors of the pharynx, and less frequently the oral cavity, may result in permanent complaints of xerostomia. The mean dose to the parotid glands that relates to 1-year xerostomia may range from 26 Gy (27) to 39 Gy (82). This serious late complication may be significantly reduced by the use of IMRT (51,99). For those patients with a dose to both parotid glands that exceeds at least 39 Gy, amifostine administration during head and neck radiotherapy will reduce the severity and duration of xerostomia 2 years after radiotherapy (110), without compromising locoregional control.

Treatment of Recurrence

Retreatment usually involves additional surgery, if feasible, and postoperative irradiation in previously unirradiated patients (Fig. 40.13). In the retreatment of parotid neoplasms,

preserving facial nerve function and obtaining local control are more difficult than for the initial tumor. Therapy consisting of surgery with postoperative irradiation has demonstrated enhanced local control, and facial nerve sacrifice may be necessary less often if this combination is used. In certain histologic subtypes (e.g., adenoid cystic carcinoma), retreatment of locally recurrent disease yields prolonged survival (89). Aggressive local therapy for recurrent disease is indicated if the probability of long-term survival is high.

Chemotherapy also has been used for recurrent disease. Polychemotherapy for recurrent high-grade disease may result in around 45% response rate, with a median duration of 7.5 months (1). However, in view of its significant toxicity and modest response rates in a population that may have recurrent yet indolent progressive disease, trials of aggressive cytotoxic therapy are recommended only on carefully drafted protocols. In the future, molecular target agents may be tested in selected recurrent salivary gland cancers.