Of these associations, exposure to ionizing radiation may be regarded as the one established risk factor for salivary gland tumors, both benign and malignant. Studies have demonstrated not only a strong association ⁶ but also a temporal latency (10 to 25 years) and a dose-response relationship with the exposure, especially for mucoepidermoid carcinomas.^4,7,8 The risk is greater for malignant than benign salivary gland tumors.^4,⁸ These observations are based on several independent long-term cohort studies of victims of atomic bomb exposure^6,^7,⁹ and patients receiving therapeutic radiation for benign⁸ and malignant indications.⁴ Radiation-induced malignant salivary gland tumors may occur with a higher frequency in the minor salivary glands.

Salivary gland tumors are regarded as rare, with a reported overall incidence in the Western world of approximately 2.5 cases to 3.0 cases per 100,000 per year.¹⁰ Malignant salivary gland tumors account for more than 0.5% of all malignancies and 3% to 5% of all head and neck cancers.¹⁰ They comprise approximately 11% of all oropharyngeal cancers.¹¹ Most patients with malignant salivary gland tumors are in the sixth or seventh decade of life.¹²

The incidence patterns of salivary gland cancers were recently reported in a Surveillance, Epidemiology, and End Results (SEER) population-based study that offers the advantage of eliminating the inherent biases of clinical series.¹³ Almost 6,400 major salivary gland carcinomas were estimated to have been diagnosed from 1992 to 2006. The most common nonsquamous salivary gland cancers were mucoepidermoid carcinomas, followed by adenoid cystic and acinic carcinomas with comparable incidence rates. Males had a 51% higher incidence rate compared with females. The parotid gland accounted for 80% of all major salivary gland malignancies,¹³ but tumors of the parotid are in general less likely to be malignant compared with the other major salivary glands and the palate of the oral cavity.¹⁰ Except for adenoid cystic carcinomas, which were equally seen in the parotid and submandibular glands, other histologies occurred mainly in the parotid gland.

Among males, the most common nonsquamous salivary gland cancers were mucoepidermoid carcinoma and adenocarcinoma not otherwise specified (NOS).¹³ Among females, the most common nonsquamous salivary gland cancers were mucoepidermoid, acinic cell, and adenoid cystic carcinomas. Squamous cell carcinoma, adenocarcinoma NOS, and salivary duct carcinoma occurred with more than a twofold higher incidence among males. In contrast, females had a higher incidence of acinic cell and adenoid cystic carcinomas.

Acinic cell carcinomas were more likely to be diagnosed at younger ages compared with other cell types.¹³ Incidence ratios of salivary gland cancers were significantly lower among African Americans, Asians, and Pacific Islanders compared with whites. However, incidences of mucoepidermoid carcinoma and adenoid cystic carcinoma occurred equally among all races.

Prevention and Early Detection

The recognition that radiation exposure is a risk factor for benign and malignant salivary gland tumors (especially mucoepidermoid carcinomas) argues that long-term surveillance is likely a prudent recommendation but with unproven benefit. Patients with a known history of accidental, diagnostic, or therapeutic radiation head and neck exposure should be counseled about the potential risk of late manifestation of malignant tumors, especially if their anticipated life span may be 10 to 25 years. It is not clear if younger patients receiving therapeutic radiation are at an increased risk,^4,⁸ necessitating more vigilant observation. Issues of cost-effectiveness of various surveillance investigations and the frequency of follow-up have not been established.

Because a dose-response relationship for developing salivary gland tumors from therapeutic radiation has been described (especially for mucoepidermoid carcinomas), this raises the intriguing question of whether this risk may be mitigated with modern radiation treatment planning techniques. Modern radiotherapy techniques have the advantage that the dose and volume of normal tissue irradiated may be reduced. This includes the use of proton radiation with its favorable Bragg peak physical beam characteristic. Details of this issue are further discussed in the section considering the role of protons in the management of salivary gland tumors. At this time this possibility remains largely speculative because the risk of developing salivary gland tumors has been demonstrated even with low doses (i.e., doses less than 280 cGy to the parotid glands).⁸ Thus the judicious use of diagnostic and therapeutic radiation and the avoidance of any unnecessary radiation exposure are currently the most effective preventive strategies that can be recommended.

Biologic Characteristics and Molecular Biology

Significant progress has been made in the understanding of the genetics and molecular biology of salivary gland malignancies. These considerations may be organized based on biologic characteristics of the tumor cell and of the tumor microenvironment. Although many biomarkers have been evaluated and shown to have differential expression compared to normal cells, highlighted here are several that have important potential therapeutic implications especially when shown to have independent prognostic significance.

Salivary Gland Genetic and Molecular Biology

Chromosomal Rearrangements

In recent years, investigations have demonstrated recurrent, nonrandom, and hallmark chromosomal rearrangements, especially translocations that characterize both benign and malignant salivary gland tumors.¹⁴ These translocations result in fusion oncogenes that affect the apoptotic threshold, cell cycle regulation, tumor angiogenesis, and growth independence. These gene arrangements have been a more common theme among leukemias, lymphomas, sarcomas, and recently with various epithelial malignancies, including thyroid and breast carcinomas. Characterizing these recurring translocations and their clinical significance is an active area of investigation in hopes of identifying future prognostic and therapeutic targets. Several noteworthy observations have been made to date.

In 2003, the recurrent and hallmark t(11;19)(q21;p13) translocation was described by several groups, demonstrating it to be a key pathogenic event that underlies the development of mucoepidermoid carcinomas.^15,¹⁶ This translocation has been shown to result in the cyclic adenosine monophosphate–responsive element binding protein (CREB) regulated transcription coactivator 1 (CRTC1)/mastermind-like 2 (Drosophila) (MAML2) fusion transcript that results in a transcription factor acting on the Notch and the CREB pathways.¹⁷

Okabe and colleagues have demonstrated that this fusion transcript was associated with a longer disease-free (DFS) and overall survival (OS) on univariate and multivariate analysis.¹⁸ Others have reported on its favorable and independent prognostic impact on survival.^19,²⁰ Despite this translocation correlating with low-grade mucoepidermoid carcinoma, in multivariate analysis for OS this translocation was found to be independent of tumor grading. Thus this translocation can potentially identify a favorable cohort of high-grade mucoepidermoid carcinomas.²¹ Lower rates of local and distant relapses have been reported in fusion-positive mucoepidermoid carcinomas.²⁰ The prognostic relevance of this translocation offers not only the promise of future risk-stratified treatment approaches but also that insights into the development of novel therapeutic targets may be gained from further characterization of this fusion gene. For example, unfavorable fusion-positive mucoepidermoid carcinomas are believed to have acquired further somatic mutations conferring increased invasiveness, such as deletion in the CDKN2A (formerly p16) gene.²²

Persson and colleagues ²³ have also demonstrated a recurrent and hallmark t(6 : 9) translocation in adenoid cystic carcinomas resulting in the MYB-NFIB fusion oncoprotein. These investigators suggest that this may also be a key oncogenic event in the pathogenesis of adenoid cystic carcinoma.²³

Epidermal Growth Factor Receptor (EGFR) and Human Epidermal Growth Factor Receptor 2 (ERBB2)

EGFR and ERBB2 (formerly and often referred to as HER2/neu) are members of the epidermal growth factor receptor transmembrane receptor family and when activated transduce mitogenic signals. Several studies have reported high immunohistochemical expression levels and that it may be more dominantly expressed in non–adenoid cystic carcinomas.²⁴^–²⁶ However, the clinical significance of EGFR protein overexpression remains controversial, with multivariate analyses demonstrating both an independent effect²⁷ and no independent association with poor survival.²⁶

Similarly, ERBB2 overexpression has been reported for malignant salivary gland tumors. These include strong overexpression in mucoepidermoid carcinomas ²⁸ and in salivary duct carcinomas^29,³⁰ but rarely seen for adenoid cystic carcinomas.³⁰ Univariate^31,³² and multivariate independent prognostic significance has been reported for ERBB2 overexpression.^33,³⁴ Agulnik and colleagues²⁴ suggest that greater clinical activity with the dual EGFR and ERBB2 inhibitor lapatinib may come in identifying tumors with both high EGFR and ERBB2 expression and that ERBB2 may be especially important. Identifying subgroups of patients more dependent on ERBB2 signaling may be important in further defining the therapeutic benefit with ERBB2 inhibition.

c-KIT/CD117

The c-KIT protein is a membrane tyrosine kinase receptor that when activated through binding to the ligands stem cell factor or mast cell growth factor provides signals for cell survival, proliferation, and differentiation. Its overexpression has been particularly noted in adenoid cystic carcinomas ³⁵ and has been suggested as one way in which adenoid cystic carcinomas may be distinguished from polymorphous low-grade adenocarcinomas,³⁶ although contradictory findings have also been reported.³⁷

Despite its overexpression, it is clear that it is the nature of any point mutations in the protein that impacts on its response to inhibition with imatinib, a small molecule tyrosine kinase inhibitor that competitively inhibits the activation of the c-KIT receptor and several other structurally similar receptor tyrosine kinases. Although several investigators have not demonstrated the presence of mutations, recent studies using more sensitive polymerase chain reaction techniques confirm the presence of multiple mutations in the c-KIT gene in adenoid cystic carcinomas.³⁸ Although both gain of function (activating function of the receptor) and loss of function c-KIT mutations have been described for other malignancies, the therapeutic implications of these recently described mutations in the management of adenoid cystic carcinomas remain to be determined. Although the independent prognostic significance for c-KIT overexpression has not been evaluated, it has remained an attractive therapeutic target owing to the clinical success of imatinib.

Salivary Gland Tumor Microenvironment

Neoneurogenesis

The study of perineural invasion is an important biologic consideration for many cancers, especially for malignant salivary gland tumors such as adenoid cystic carcinomas. The molecular determinants of this mechanism of spread have only begun to receive attention in recent years. Observations in other cancer sites currently suggest that perineural invasion is the result of a mutual neurotropic interaction with cancer cells characterized by the release of various paracrine growth factors.³⁹ As with neoangiogenesis, it is clear that the ingrowth of nerve endings into a tumor may be stimulated by the tumor release of various neurotrophins such as brain-derived neurotrophic factor (BDNF),⁴⁰ nerve growth factor (NGF), and its receptor tyrosine kinase A (TrkA).⁴¹ Neurotropin staining has been demonstrated to be present in adenoid cystic carcinomas. NGF and TrkA both correlated with the presence of perineural invasion.⁴¹ These tumor-innervating nerve cells may release neurotransmitters that function as proliferative and promigratory signals for the tumor cells. Furthermore, nerve fibers are used as routes for tumor cell dissemination. For pancreatic carcinoma, detailed studies of serial histologic sections have demonstrated that tumor cells can progress along branching nerve fascicles in a continuous fashion.⁴²

Hypoxia

In contrast to other solid malignancies of the head and neck, the role of tumor hypoxia has not been as extensively evaluated. This is an important consideration for malignant salivary gland tumors given their relative resistance to radiotherapy and the improved results seen with neutron radiation that is less oxygen dependent.

Preliminary investigations using methodologies such as immunohistochemical staining for the oxygen-dependent binding of 2-nitroimidazoles suggest that salivary gland malignancies may be well-oxygenated tumors.⁴³ This observation would not support the hypothesis that an underlying mechanism for not only the radioresistance but also the high rates of distant metastasis characterizing malignant salivary gland tumors is the presence of tumor hypoxia. In a study of 12 patients receiving surgery, the hypoxia marker pimonidazole was administered preoperatively. Of the 8 patients yielding sufficient tumor specimen for analysis, immunohistochemical staining was not seen with pimonidazole and for the hypoxia-inducible factor (HIF)-1a and the HIF-1a regulated carbonic anhydrase (CA) IX and glucose transporters (GLUT) 1 and 3. Given that salivary gland malignancies are rare and of diverse histologies, further evaluation is needed before more generalized conclusions can be made.

Pathology and Pathways of Spread

Pathology

The classification of salivary gland tumors is primarily based on morphologic patterns, with two classification systems often recognized. This includes the Armed Forces Institute of Pathology (AFIP) system; and, in 2005, the World Health Organization (WHO), which further developed the classification scheme for salivary cancers originally set forth by Foote and Frazell ⁴⁴ (Table 35-1). The WHO classification includes the following categories: adenomas, carcinomas, nonepithelial tumors, malignant lymphomas, secondary tumors, unclassified tumors, and tumor-like lesions. In contrast, the AFIP classification includes benign epithelial neoplasms, malignant epithelial neoplasms, mesenchymal neoplasms, malignant lymphomas, metastatic tumors, and non-neoplastic tumor-like conditions. Speight and Barrett¹⁰ have reported a comprehensive summary of salivary gland histopathology. Given the histologic diversity, this chapter will focus on histopathologies that impact on the pattern of spread and its clinical manifestations.

TABLE 35-1 Classification of Malignant Salivary Gland Tumors

NOS, not otherwise specified.

From Barnes L, Eveson J, Reichart PA, Sidransky D, editors: World Health Organization Classification of Tumors: Pathology and Genetics of Tumors of the Head and Neck. Lyon, 2005, IARC Press.

Important prognostic factors for disease relapse and survival are summarized in Table 35-2. Practically, histologies may be categorized as high-grade at a greater risk of nodal and distant metastases or low-grade with a lower risk of nodal metastasis. The former includes high-grade mucoepidermoid carcinomas, high-grade adenocarcinoma NOS, carcinoma ex-pleomorphic adenoma, salivary duct carcinomas, and squamous cell carcinomas. Low-grade histologies include low-grade mucoepidermoid carcinomas, adenoid cystic carcinomas, acinic cell carcinomas, and polymorphous low-grade adenocarcinomas. An intermediate group has also been described consisting of intermediate mucoepidermoid carcinomas.

TABLE 35-2 Summary of Prognostic Factors

Endpoint	Prognostic Factors
Local relapse	T3 and T4 category, unresectability/skull base invasion
Neck metastasis	Male sex, T3 and T4 category, primary site involving the pharynx, high-grade histology, especially mucoepidermoid and adenocarcinoma NOS
Distant metastasis	Clinical perineural invasion, skull base invasion, nodal metastasis
Survival	Stage IV, nodal metastasis

NOS, not otherwise specified.

For mucoepidermoid carcinomas, the histologic grading has been shown to be of prognostic significance. Several three-tiered grading schemes have been shown to be reproducible and predictive of the patient’s outcome by defining low-, intermediate-, and high-grade tumors using five histopathologic features yielding a numerical score for differentiation.⁴⁵^–⁴⁷ High-grade mucoepidermoid carcinoma has a greater proportion of epidermoid cells than mucus-producing cells, with a solid tumor appearance often mistaken for squamous cell carcinoma. High-grade disease also increases the risk of nodal metastasis sufficient to warrant elective management (e.g., a neck dissection). Despite this, the risk of nodal metastasis with low and especially intermediate grades can be as high as 24% to 30% in some series. High-grade mucoepidermoid carcinomas demonstrated nodal metastases in 56% of patients. In the modified Healey grading schema, lymph node involvement was 0% (low grade), 22% (intermediate grade), and 72% (high grade). T category also is associated with an increased risk of nodal metastasis for both major and minor salivary gland mucoepidermoid carcinomas.⁴⁸ T1 high-grade disease may be at low risk for nodal metastasis in major salivary glands, that is, anatomic sites not involving the mucosa.⁴⁸

Pathways of Spread

When considering all histologies together, the risk of lymph node metastasis is increased with T3 and T4 disease, involvement of a pharyngeal site, and high-grade mucoepidermoid carcinomas and adenocarcinoma NOS.⁴⁹ Table 35-3 shows similar observations with T category, anatomic site, and histology.⁵⁰ For minor salivary glands, the density of the lymphatics in the anatomic site has a significant influence on the risk of nodal metastases. In general, cancers arising within the oropharynx or nasopharynx have about a 60% incidence of lymph node metastasis compared with 5% to 10% for hard palate and paranasal sinus sites. Minor salivary gland cancers arising within the tongue and floor of mouth have an approximately 40% incidence of lymph node metastasis; those arising within the gingiva have a 20% incidence of lymph node metastasis; tumors of the nasal cavity, buccal mucosa, and lip have a 15% incidence or less.

TABLE 35-3 Incidence of Lymph Node Metastases at Initial Diagnosis by Site, Grade, T Category, Size, and Facial Nerve Paralysis

Adenoid cystic carcinomas at the primary site have a predilection for perineural invasion as a unique pattern of local disease extension. Adenoid cystic carcinomas occur most commonly in the palate and frequently spread by perineural invasion. This may be seen in more than 50% of cases. Spread centrally within and peripherally along the nerve are common routes in which this tumor spreads. Growth along the nerve has been shown to have “skip” areas of involvement and noninvolvement, so that a negative nerve margin does not guarantee a final negative margin. However, a recent study of serial sections for pancreatic carcinoma suggests that perineural invasion may, in fact, be contiguous along the branching nerve fascicles and suggest that it may be the branching nature of the nerve and pathologic sampling of the sections evaluated that contribute to the uncertainty in margin assessment.⁵¹

The risk of distant metastases is increased with the presence of lymph node metastasis, skull base involvement, and high-grade histology.⁵² Common distant metastatic sites include lung, liver, and brain and are also associated with the histologic grade.

Clinical Manifestations, Patient Evaluation, and Staging

Clinical Manifestations

Most patients presenting with salivary gland tumors are asymptomatic and present typically with a solitary, painless mass. Low-grade tumors tend to be slow growing and rarely invade local structures. Symptoms are largely related to the anatomic location from which it originates. Symptoms may include local pain with involvement of the parotid gland and facial paralysis with involvement of the seventh cranial (facial) nerve. Local obstructive symptoms and possibly drainage from the ipsilateral ear can result from extension to the external auditory canal, especially with deeper parapharyngeal space involvement. Deeper parapharyngeal space involvement can also result in symptoms of trismus because of involvement of the pterygoid muscles. Involvement of the skull base can result in other cranial nerve deficits, including symptoms of dysphagia. Minor salivary gland malignant tumors commonly involve the palate, where they can cause local symptoms, but patients again typically are asymptomatic or have a vague sensation of a mass. Salivary gland tumors may also present with mass effect symptoms in the submandibular gland, maxillary sinus, oropharynx, and larynx. Tumor involvement of functionally sensitive sites, such as the larynx, is more likely to lead to symptoms and an earlier disease presentation.

Perineural invasion may result in symptoms of pain and numbness but more commonly presents as an asymptomatic mass with an indolent history of growth. Clinical symptoms are more likely to be associated with radiologic evidence of perineural invasion (Fig. 35-1). Extension through the skull base with intracranial growth can also occur and result in mass effect symptoms. Because adenoid cystic carcinomas commonly occur in the palate, spread along the palatine nerves through the greater and lesser palatine foramen is an important route of local spread.

Figure 35-1 A and B, Preoperative T1-weighted postcontrast coronal MR images demonstrating an adenoid cystic carcinoma of the left submandibular gland with clinical perineural invasion (PNI) extending along the thickened mandibular branch of the fifth cranial nerve (between arrowheads) between the medial pterygoid (MP) and the lateral pterygoid (LP) muscles to involve the foramen ovale (short arrow). There is also abnormal signal intensity and tissue in the region of the mandibular foramen along the left ascending ramus of the mandible and in the adjacent medial pterygoid muscle (between long arrows).

Patient Evaluation

As for any patient with a suspected or documented malignancy, proper evaluation begins with a detailed history and physical examination (Table 35-4). Presence of a new mass in the face, neck, or mouth should be investigated for associated symptoms based on the surrounding anatomic structures that may be affected, thus helping to localize the disease extension. This should also include symptoms that may relate to perineural spread of the cancer. Physical examination should include a full oral cavity inspection and bimanual palpation of structures that are or may be involved. Minor salivary gland tumors typically appear as a submucosal mass and commonly involve the palate. The proximity of the mass and any changes to the greater and lesser palatine foramen should be noted. Flexible fiberoptic nasopharyngolaryngoscopy should be performed to assess for the location and extent of any mucosal lesions in the pharynx or to assess the function of critical cranial nerves that may be suspected to be involved because of a mass at the skull base.

TABLE 35-4 Diagnostic Algorithm for Malignant Salivary Gland Tumors

General

History

Physical examination

Radiographic Studies

Chest radiography or CT

CT or MRI of the head and neck with contrast enhancement

PET/CT for high-grade aggressive histologies

Laboratory Studies

Liver function tests (lactate dehydrogenase, alkaline phosphatase, aspartate aminotransferase)

Fine-needle aspiration or ultrasound-guided core needle biopsy of suspicious lesions may be useful to distinguish malignant from benign processes. In one retrospective study of 879 patients, fine-needle aspiration demonstrated a sensitivity of 83% and specificity of 99%, with overall accuracy of 93%.⁵³ However, other studies have demonstrated false-negative rates as high as 33% and 43% for adenoid cystic carcinoma and low-grade mucoepidermoid carcinoma, respectively.⁵⁴

Both CT and MRI are necessary to assess the extent of disease. Bony destruction and regional lymph node metastases are better visualized with CT, whereas MRI provides information regarding soft tissue infiltration, intracranial spread, and perineural invasion.⁵⁵ Although the role of PET/CT for salivary gland cancers has yet to be established, early studies have demonstrated an accuracy of more than 90% for detection of the primary tumor and an increased ability in identifying unrecognized local, nodal, and distant metastases.^56,⁵⁷

Staging of Major and Minor Salivary Gland Malignancies

Unlike minor salivary gland tumors, which are staged according to the anatomic site of origin, major salivary gland cancers have their own staging system. In general, tumors of the major salivary glands are staged according to the tumor size and whether there is any extraparenchymal extension (Table 35-5). Extraparenchymal extension is described as either clinical or macroscopic evidence of invasion of the soft tissues. Microscopic extension does not constitute extraparenchymal extension. For parotid tumors the prognostic significance of seventh cranial nerve involvement is also considered. Although tumor grade does affect outcome, the current (7th edition) American Joint Committee on Cancer staging system does not incorporate tumor grade as a component.

TABLE 35-5 Primary Tumor Staging Criteria

Tx	Primary tumor cannot be assessed
T0	No evidence of primary tumor
T1	Tumor 2 cm or less in greatest dimension without extraparenchymal extension*
T2	Tumor ≥2 cm but <4 cm in greatest dimension without extraparenchymal extension*
T3	Tumor ≥4 cm and/or tumor having extraparenchymal extension*
T4a	Tumor invades skin, mandible, ear canal, and/or facial nerve
T4b	Tumor invades base of skull and/or pterygoid plates and/or encases the carotid artery

* Extraparenchymal extension is defined as the presence of any clinical or macroscopic evidence of invasion of the adjacent soft tissues. Microscopic evidence alone does not constitute extraparenchymal extension for classification purposes.

Nodal category and group staging for major salivary gland tumors follows other head and neck cancer sites and is not described here.⁵⁸

Primary Therapy

The treatment approach for salivary gland tumors continues to be based on a surgical resectability paradigm. The therapeutic dilemma occurs when there is significant involvement of a functional region in the head and neck involved in speech or swallowing or where unresectable disease occurs. Therapeutic advancements have largely been limited in recent years to the development of novel organ-preserving surgical approaches (Figs. 35-2 and 35-3). Radiotherapy advancements include the development of more conformal radiotherapy techniques and the advent of particle beam therapy. Advancements in systemic agents have included the translation of molecular therapeutics that offer a potentially more favorable therapeutic ratio.

Figure 35-2 Transoral robotic resection (TORS) for an adenoid cystic carcinoma of the base of the tongue. A, The patient was swallowing a full oral diet in the early postoperative setting and did not require a mandibulotomy. B, Very good swallow function can be seen with no pooling of secretions in the vallecula with further wound healing. C, In contrast, a second patient was treated with a traditional mandibulotomy and a radial forearm free flap reconstruction in the left base of the tongue demonstrating more limited swallow function with pooling of secretions in the vallecula.

Figure 35-3 Transoral robotic surgery (TORS) of a parapharyngeal space pleomorphic adenoma. This demonstrates the robotic wristed instruments performing the early incision and dissection through the left lateral oropharyngeal wall adjacent to the base of the tongue (BOT) and soft palate (SP). Excellent visualization can be demonstrated through the magnified optics.

Role of Surgery

The surgical treatment of salivary malignancies can be divided into categories based on the gland involved. The most common site for salivary malignancy is the parotid gland. The submandibular gland and the minor salivary glands have a lower absolute number of cancers; however, the risk of a neoplasm being malignant is higher in submandibular and minor salivary glands than it is in the parotid gland.⁵⁹

Parotid Gland Malignancy

The mainstay of surgical management of parotid malignancies involves a parotidectomy, and the extent of the surgery is dictated by the tumor extent. The gland itself is divided into two separate components by the seventh cranial nerve, and tumors can exist and arise in either or both parts of the gland. Although there is no embryologic boundary between the two components of the gland, most tumors arise from the superficial lobe.

Surgery involves an incision in the preauricular crease extending postauricularly and then into the upper neck (see web-only Fig. 35-1 on the Expert Consult website). This incision, known as the modified Blair incision, gives access to the seventh cranial nerve as it comes out of the stylomastoid foramen. One of the keys to the procedure is to first identify the nerve, trace it out along its main branches, and then manage the tumor once its relationship to the nerve has been clarified. In cases in which the nerve is compromised by the tumor, or in which the tumor abuts the stylomastoid foramen, it may be necessary to identify the nerve before it comes out of the mastoid bone. This can be accomplished by drilling out the mastoid bone and identifying the nerve before it comes out of the foramen.

Web-Only Figure 35-1 A modified Blair incision used for surgical access for a left parotid salivary gland tumor. The incision includes a preauricular incision that transitions under the ear into a postauricular and upper neck incision.

Once the appropriate branches of the nerve have been identified, the tumor is removed. Great care is taken to avoid entry into the tumor, and attempts are made to avoid residual positive margins, because this is a known poor prognostic indicator.^60,⁶¹ Because most tumors involve only the superficial lobe of the parotid, dissecting the tumor carefully off the seventh cranial nerve is appropriate surgical care. Tumors that involve the deep lobe of the parotid gland or that involve the parapharyngeal space just medial to the parotid gland may require extended procedures to perform complete resections. These procedures are often possible through the same parotid incision but occasionally require extended neck incisions or mandible-splitting transoral procedures. Excision of soft tissue, skin, muscle, and neurovascular structures in the region is appropriate if they are involved with tumor.

All attempts should be made to preserve the seventh cranial nerve during parotid surgery. Complete paralysis of this nerve will not only cause a significant cosmetic deficit but also has the potential to affect many aspects of a patient’s outcome. Corneal abrasion, nasal congestion, poor oral competence, and slurred speech are some of the common complaints after this event. Most surgeons would agree that tumor proximity to the nerve is not adequate justification for its sacrifice. However, there are circumstances under which most surgeons would agree that sacrifice of all or a part of the nerve is reasonable. Complete or partial paralysis of the face preoperatively often heralds an intraoperative finding of direct invasion of the seventh cranial nerve by tumor. If the tumor extends into the nerve, or grossly encases the nerve, then sacrifice of the nerve is warranted.

Submandibular Gland Malignancy

Although the absolute number of malignancies of the submandibular gland is far less than the parotid gland, the chance of a lesion being malignant is much higher, nearing 50%.⁶² For this reason, a high level of vigilance must be in place when managing masses in the submandibular region. Most commonly, patients present with painless masses in the region. Less commonly, painful masses or neck adenopathy may occur. Any of these presentations warrants a workup of the area, including possibly CT or MRI and fine-needle aspiration biopsy.

As with parotid cancers, salivary gland tumors of the submandibular gland are treated by surgery. Definitive treatment often includes a combination of therapies but almost always begins with surgical excision. The surgical approach should include, at the minimum, a complete submandibular gland excision with clear margins. There is no role for enucleation of the tumor, because this has been clearly associated with a higher rate of local recurrence.⁶³ In patients who have undergone subtotal surgery, only to be diagnosed with cancer postoperatively, one should strongly consider reoperation to obtain negative margins before considering further therapy.

Sublingual and Minor Salivary Malignancy

Malignancies of the sublingual gland and minor salivary glands are rare. Minor salivary glands are present throughout the upper aerodigestive tract, and salivary gland tumors can occur at any of the sites. The oral cavity and oropharynx are the most common sites of disease, but the nasal cavity, paranasal sinuses, hypopharynx, larynx, nasopharynx, and parapharyngeal space are all at risk for these cancers.⁶⁴

As is the case with other salivary cancers, surgical therapy with negative margins is the mainstay of therapy. Depending on the site of the cancer, this may include a variety of surgical approaches. The challenge for the management of these malignant salivary gland tumors is the functional impact that surgery can have on speech and swallow function. Unlike the case for squamous carcinomas, the role of nonsurgical therapy has not been well established.

Organ-preserving surgical approaches that do not use transcervical exposure techniques are valuable in limiting the swallowing complications resulting from surgery. These include various transoral approaches, including transoral laser microsurgery (TLM) and transoral robotic surgery (TORS)^65,⁶⁶ (see Figs. 35-2 and 35-3). TORS has several advantages, including the improved visualization of an en bloc tumor resection in three dimensions avoiding sight-line limitations with magnified optics. Its application has included function-preserving resection of salivary gland tumors of the base of the tongue⁶⁷ and in the parapharyngeal space⁶⁵ (see Fig. 35-3).

Management of the Neck

Patients who present with clinically palpable or radiologically evident disease in the neck lymph nodes require management of this disease. This generally involves a planned neck dissection at the time of the primary surgery with or without postoperative radiation.⁶⁸ What remains more controversial is the role of elective neck treatment in patients with no clinicoradiologic evidence of disease in the neck.

Some studies have demonstrated that all high-grade malignancies (regardless of histology) would benefit from elective management of the neck.⁶⁹ Other studies have suggested elective management of the neck based on histologic type and/or size. This includes elective neck surgery for all squamous cell carcinomas, adenocarcinomas, high-grade mucoepidermoid carcinomas, undifferentiated carcinomas, and all T2 or higher T-category tumors.^48,⁷⁰ Still, other groups have suggested either nodal sampling of the upper neck lymph nodes at the time of the primary surgery or including the neck in the primary radiotherapy field for certain histologies.^71,72 What does appear clear is that high-grade, high-stage cancers benefit from aggressive multimodality therapy, including treatment of the primary site and the lymphatic drainage pathway for the tumors.

Role of Radiotherapy

Postoperative Radiotherapy

At present no randomized studies have been conducted to establish the value of postoperative radiotherapy. Tables 35-6, 35-7, and 35-8 summarize the treatment results from modern series comparing surgery with surgery and postoperative, adjuvant irradiation for malignant major salivary gland, submandibular gland, and minor salivary gland cancers, respectively. These demonstrate that the addition of postoperative radiation therapy, when indicated (Table 35-9), appears to be associated with improved local control rates and survival in patients with malignant salivary gland cancers.

TABLE 35-6 Results of Surgery Alone and Combined Surgery and Postoperative External Irradiation for Malignant Salivary Gland Tumors

TABLE 35-7 Results of Surgery and Surgery with Postoperative, Adjuvant Radiation Therapy for Malignant Tumors of the Submandibular Gland

TABLE 35-8 Results of Surgery and Surgery with Postoperative, Adjuvant Radiation Therapy for Malignant Minor Salivary Gland Tumors

TABLE 35-9 Indications for Postoperative, Adjuvant Radiotherapy for Malignant Salivary Gland Tumors

Close or microscopically positive surgical margins or gross residual disease

High-grade cancer

Involvement of skin, bone, nerve (gross invasion or extensive perineural involvement) and tumor extension beyond the capsule of the gland with periglandular and soft tissue invasion

Lymph node metastases

Large tumors (T3 or T4 disease) requiring radical resection

Tumor spillage during operation

Recurrent cancer (including benign histologies such as recurrent pleomorphic adenoma)

In general, the use of postoperative radiotherapy doses typically to 60 Gy appears to be sufficient for perineural involvement of unnamed nerves.⁵⁰ Skull base relapses may be reduced from 15% to 5% with elective skull base irradiation, especially for T4 disease.⁸³ With perineural invasion of a major named nerve, elective skull base irradiation has also been shown to be effective.⁸¹ Patients with skull base relapses had a poorer survival when compared with those who did not experience a relapse in a retrospective review of 140 patients.⁸³

Doses between 50 and 60 Gy have also been used for elective irradiation of the neck with relapses less than 5%.⁸⁴ Terhaard and colleagues⁵⁰ observed a trend to a lower rate of neck control when elective radiotherapy doses less than 46 Gy were used, compared with 46 Gy or more (p = .07). Again, patients with neck relapses have been associated with a poorer survival compared with those patients without neck relapses. When a higher burden of microscopic disease is present based on the presence of a positive margin or perineural invasion of a major named nerve, doses less than 56 Gy are insufficient and doses more than 60 Gy may be associated with improved local control rates.⁸⁵

With surgery and postoperative radiotherapy to the doses described, long-term locoregional control rates on the order of 80% to 90% may be expected for T1-3 lesions with N0-1 disease. With T4 lesions or multiple nodes, locoregional control rates are significantly lower, on the order of 60% to 70%.

Definitive Radiotherapy

Definitive radiotherapy is used primarily for patients with locally unresectable salivary gland tumors and is discussed in the next section.

Locally Advanced Disease and Palliation

Definitive Radiotherapy: Conventional Fractionation

With locally unresectable salivary gland malignancies, the use of definitive irradiation on the order of 70 Gy with conventionally fractionated schedules typically results in locoregional control rates of 20% to 30% or less.86,87,88 These results are also limited by the inherent biases seen with retrospective institutional reviews but are significant in their independent verification of a consistently poor outcome for unresectable salivary gland malignancies when treated with photon radiotherapy alone. These results are also consistent with the standard arm in a randomized trial of neutron versus photon radiotherapy for unresectable salivary gland malignancies.⁸⁹

Altered Fractionation

Although the Radiation Therapy Oncology Group (RTOG) 9003 study demonstrated that altered fractionation improves locoregional control in other sites of the head and neck,⁹⁰ the role of this modality in salivary gland tumors is unclear. Wang and Goodman⁹¹ reported results using high-dose, accelerated, hyperfractionated photon beam therapy in patients with inoperable and unresectable major and minor salivary gland cancers. All patients were treated with 1.6 Gy per fraction, twice a day, combined with various boost techniques to obtain a total dose of 65 to 70 Gy. Local control and survival rates were promising, with a 5-year 100% local control and 65% OS for parotid gland lesions and a 5-year 78% local control and 93% OS for minor salivary gland lesions. Late complications were minimal. However, the follow-up was relatively short and, thus far, no update has been reported. Although this suggests a potential role for altered fractionation, it should be regarded as investigational until further data are reported.

Neutrons/Heavy Charged-Particle Therapy

High linear energy transfer (LET) radiation from neutrons and heavy charged particles increases the probability of direct damage to DNA and is characterized by a reduced oxygen enhancement factor. Compared with low LET radiation from standard photon therapy, damage from high LET radiation is less reparable and less dependent on cell cycle state. These characteristics make high LET potentially suitable for malignant salivary gland tumors, which are thought to be radioresistant owing to a low growth fraction and a long doubling time.

The only randomized trial comparing neutron radiation to conventional photon irradiation was conducted by the RTOG and the Medical Research Council (MRC) for patients with locally unresectable primary and recurrent malignant salivary gland tumors.⁸⁹ The study closed early because of ethical concerns resulting from interim analysis performed on 25 patients. At 2 years, this analysis demonstrated a significantly better local control with neutron versus photon irradiation (67% vs. 17%, p <.005) and a trend toward improved OS (62% vs. 25%, p = 0.1). The local control rate of 67% was consistent with prior single-institution reports associated with the use of fast neutrons.^92,93,94 A median survival of 2.97 years with neutrons versus 1.23 years with photons was initially reported. With longer follow-up, the OS curves showed no difference. The majority of failures on the neutron study arm were systemic. In contrast, the major relapse pattern in the photon study arm was predominantly locoregional failures. The incidence of severe or life-threatening toxicity was greater in patients treated with neutrons, with nine patients treated with neutrons having at least one “severe or greater” complication, compared with four patients in the photon arm of the study. This finding of a relatively high rate of severe complications from neutrons has been observed in another comparative series.⁹⁵

The overall lack of survival difference from the randomized trial, and the high rate of severe complications coupled with the lack of institutions with available treatment capabilities, has limited the broad use of fast neutron therapy. However, mature uncontrolled, single-institution studies continue to demonstrate impressive results, with lower rates of toxicity. Douglas and colleagues ⁹⁶ published a retrospective study of 148 patients with major salivary gland tumors treated at the University of Washington with a 5-year actuarial locoregional control rate of 100% for tumors less than 4 cm and only a 6% grade 3 or 4 complication rate. An updated larger series of 279 patients from the same institution demonstrated a 6-year actuarial locoregional control rate of 59% with a grade 3 or 4 toxicity rate of 10%.⁵² Neutrons have also been used in the postoperative setting for microscopic residual disease with a 100% 5-year actuarial locoregional control rate reported.⁹⁷

In part, the increased complication rates are the result of the use of neutrons at the skull base and the adjacent central nervous system. The sensitivity of the central nervous system structures to high LET radiation often precludes safe delivery of therapeutic doses of high LET radiation to tumors with skull base involvement. Despite this, it is interesting to observe that neutron radiation does not appear to be associated with an increased risk of facial nerve palsy as a treatment complication.^93,⁹⁸ Complicating the risk of central nervous system injury is the recognition that skull base involvement is associated with a poorer prognosis with lower rates of local control and survival in a group of 159 patients with unresectable adenoid cystic carcinomas.⁹⁷

To address this technical limitation of neutrons, investigators from the University of Washington have been investigating the role of boosting the skull base disease with stereotactic radiosurgery after reduced dose neutron beam radiotherapy at the superior portion of the tumor. In 34 patients with skull base disease treated with a Gamma Knife boost (see later discussion), the 40-month actuarial local control was 82%, compared with a historical control rate of 39% in patients with skull base disease treated with neutrons alone (p = .04).⁹⁹

Heavy charged-particle radiation may also hold promise because it combines the biologic qualities of high LET with the rapid Bragg peak dose fall-off. Limited data with carbon ion radiation suggest similar tumor control as neutron beam therapy, with fewer late effects.¹⁰⁰

In summary, for patients with locally advanced, unresectable disease of the salivary glands, neutrons offer superior locoregional control compared with standard photon irradiation, but at a potential risk of increased severe toxicity. For postoperative radiation therapy it is not clear that the therapeutic ratio is improved upon given the potential for increased late normal tissue toxicities. Techniques to improve the therapeutic ratio, such as those seen with stereotactic boost or with heavy charged particle radiation therapy, are promising and are the subject of ongoing investigations.

Stereotactic Radiosurgery

Whether the use of hypofractionated doses of radiotherapy in the form of stereotactic radiosurgery (SRS) or fractionated stereotactic radiotherapy (SRT) can overcome the radioresistance associated with conventionally fractionated photon therapy has not been clearly established. There have been two main indications that have been the subject of clinical investigations.

The first is the study of SRS at the skull base as a treatment strategy to more safely escalate the effective biologic radiotherapy dose delivered. Douglas and colleagues ⁹⁹ used a Gamma Knife radiosurgery boost as a strategy to reduce the relative biologic neutron dose to the skull base and minimize the risk of central nervous system injury but to improve on the local control rates. The median prescribed Gamma Knife dose was 12 Gy to the 50% isodose line, resulting in a median neutron dose of 19.2 nGy to the isocenter. The median number of isocenters was 17. The median target volume treated was 12.4 cm³ (range, 1.9 to 28.9) with a median total volume treated of 18.3 cm³ (range, 5.9 to 53.9). This resulted in a median dose of 11.98 nGy to the adjacent tip of the temporal lobe. In total, 34 patients were treated, and with a median follow-up of 20.5 months the actuarial 24- and 40-month local control rates were 82% and 82%, compared with 81% and 39% in historical patients not receiving SRS boost. A total of four failures in the SRS group were described with two in-field and two out-of-field relapses. Noteworthy complications included three cases of radionecrosis with one of three patients with symptomatic headaches responsive to corticosteroids.

The other major indication for SRS or SRT has been for palliative indications. The rationale for the use of an SRT approach is twofold. First, the stereotactic approach affords greater confidence in the accuracy of the target lesion to be irradiated, allowing a smaller margin of uncertainty and thus reducing the volume of normal tissues that receive both unnecessary entry and exit radiation. The second is the potential for greater therapeutic benefit that may be achieved with higher doses delivered over a single fraction or with just a few fractions. To date, the reported experiences are limited and largely amount to small case and heterogeneous cohorts of patients with head and neck cancers, including some recurrent salivary gland malignancies.¹⁰¹^–¹⁰³ These observations limit the ability to draw any definitive generalized conclusions about the risks and benefits of SRS and SRT at this time, including the optimal prescribed dose. The range in prescribed dose includes single fractions of 11 to 18 Gy and fractionated schedules including 30 Gy in six fractions. Effective palliation of pain with durable responses including freedom from local tumor progression (ranging from 10 to 35 months) has been reported.¹⁰³ In general, the major serious but infrequent risk associated with this palliative treatment approach includes the risk of optic, facial, and trigeminal nerve injury. At this time the use of this treatment approach should be limited to treatment on a clinical protocol or institutions with a broad stereotactic experience at the skull base for other malignancies in which the risks are well defined.

Proton Radiotherapy

There is very limited published experience with the use of proton therapy for salivary gland tumors. The Paul Scherrer Institute in Switzerland published a comparative dosimetry study of intensity modulated photon and proton therapy in the treatment of head and neck tumors.¹⁰⁴ The primary end point was the risk of secondary cancers based on the risk estimated using the organ equivalent dose model. For proton plans, both the scatter and the neutron dose contributions were accounted for. Their modeling demonstrated that the risk of secondary malignancies may be reduced with the use of intensity modulated proton therapy (IMPT) compared with IMRT with photons. The risk was particularly reduced by reducing the number of proton fields used.

Pommier and colleagues ¹⁰⁵ at the Francis H. Burr Proton Center in Boston have treated 23 patients with adenoid cystic carcinomas of the skull base with a combination of proton and photon radiotherapy. Only 2 of 23 patients had a complete gross resection, but with residual positive margins as the indications for treatment. The mean total dose to the primary tumor was 75.9 cobalt-gray equivalent. With a median follow-up of 64 months, the 5-year actuarial local control rate was 93%. The 5-year DFS and OS were 56% and 77%, respectively. Multivariate analysis demonstrated that tumor involvement of the sphenoidal sinus and the clivus and the presence of vision change at presentation were significant adverse risk factors for OS. No grade 4 or 5 ocular toxicities were reported. Significant grade 3 neurologic toxicities were observed in 10 patients, and 2 patients developed grade 5 toxicity. These findings suggest that with radiotherapy dose escalation, which may potentially be safer because of the Bragg peak characteristics of the proton beam, local control rates may be improved beyond the historic 20% to 30% local control rates.⁸⁶

Chemoradiation

The data for concurrent chemoradiation is scant. In the adjuvant context, a single case control study compared 12 patients with a variety of histologies treated with chemoradiation with 12 matched controls treated with radiation alone.¹⁰⁶ This analysis showed a significant survival and local control advantage with chemoradiation, with a 3-year OS and DFS of 83% and 77% compared with 44% and 52% for radiotherapy alone. However, in addition to the small number of patients analyzed, several other key caveats must be noted. Performance status and comorbidity of the two groups were not reported, IMRT was used more frequently in the chemoradiation group, and a variety of histologies were treated. Thus, adjuvant chemoradiation must be considered experimental. However, in the absence of level I data to support its efficacy, it is commonly used for patients with aggressive disease, positive margins, extracapsular extension, and other factors associated with a high risk of recurrence. This practice is largely based on the experience of chemoradiation effects on normal tissues when used typically for squamous cell carcinoma of the head and neck. Hence, its use should be limited to the dose ranges and concurrent chemotherapeutics where the toxicity spectrum and risks have been well established if it is being considered.

Much of the available literature on chemoradiation for salivary gland tumors involves small, single-institution series. One case report describes definitive chemoradiotherapy concurrent with intra-arterial cisplatin and docetaxel in two patients.¹⁰⁷ A retrospective study of 13 patients treated with cisplatin chemoradiation demonstrated complete response to initial therapy in all patients.¹⁰⁸ A single patient developed delayed local failure at 39 months (but achieved successful surgical salvage), and a second patient developed distant metastases; 5-year OS was 83%. Finally, a retrospective analysis of 14 patients reported a 3-year locoregional control rate of 51.6% and a 3-year OS of 35.7% with 5-fluorouracil and hydroxyurea chemoradiation after a median radiation treatment interval of 48 months.¹⁰⁹ Although also experimental, definitive chemoradiation is reasonable for patients who are not candidates for surgery, either based on anatomic constraints or medical inoperability.

Palliation: Role of Chemotherapy

Data on the efficacy of chemotherapy for salivary gland malignancies is limited to case reports, retrospective reviews, and small phase II studies. An expanded version of this topic is included on the Expert Consult website.

Once metastatic, salivary gland malignancies are not curable. Nonetheless, their course can frequently be indolent, especially for adenoid cystic carcinoma with metastatic disease limited to the lung. Median survival with metastatic disease with adenoid cystic carcinoma is approximately 3 years,¹¹⁰ with some patients living substantially longer. Results with systemic agents, both chemotherapeutic and molecular targeted agents, have been disappointing. Therefore careful consideration must be given to administration of supportive care, with systemic therapy reserved for rapid or symptomatic progression.

Chemotherapeutic agents are of limited efficacy in salivary gland malignancies, with particularly poor responses in adenoid cystic histology. A variety of single-agent chemotherapies have been studied in the phase II setting. Limited responses have been observed with cisplatin, vinorelbine, mitoxantrone, epirubicin, methotrexate, and paclitaxel. For example, the Eastern Cancer Oncology Group (ECOG) 1394 trial evaluated paclitaxel, given at 200 mg/m² every 3 weeks.¹¹¹ No responses were seen in 14 patients with adenoid cystic carcinoma, whereas a 26% response rate was seen in patients with mucoepidermoid or adenocarcinoma histologies.

The best studied combination regimen is CAP (cyclophosphamide, doxorubicin [Adriamycin], and cisplatin).¹¹² Although the CAP regimen has demonstrated good response rates across histologies, there is no clear evidence for superior efficacy as compared with single-agent chemotherapy. Only a single randomized study compared single-agent therapy with a platinum-based doublet. Airoldi and associates¹¹³ compared vinorelbine (given at 30 mg/m² on days 1 and 8 every 3 weeks) to cisplatin plus vinorelbine (with cisplatin administered at 80 mg/m² on day 1 and vinorelbine at 25 mg/m² on days 1 and 8). This study showed a doubling of response rate from 20% to 44% and a strong trend toward improved median survival (8.5 vs. 11 months, p = .058) with use of the doublet. Of particular note was the 44.4% response rate in the typically chemoresistant adenoid cystic carcinoma. Many oncologists consider cisplatin plus vinorelbine as the standard of care when cytotoxic chemotherapy is chosen, with paclitaxel as a reasonable consideration in patients with non–adenoid cystic carcinoma.

Malignant salivary gland tumors commonly overexpress a variety of molecular targets. Given the success of targeted agents in other cancers overexpressing these markers ¹¹⁴^–¹¹⁷ and the relative tolerability of the agents targeting these markers, investigators have sought to apply these agents to the malignant salivary gland tumors¹¹⁸^–¹³¹ (see web-only Table 35-1 on the Expert Consult website).

WEB-ONLY TABLE 35-1 Summary Table of Clinical Trials of Molecular Targeted Agents for Malignant Salivary Gland Tumors

Because c-KIT is expressed by most adenoid cystic carcinomas, imatinib mesylate—an inhibitor of the c-KIT, BCR-ABL, and PDGFR tyrosine kinases—was studied. Although responses were noted in three case reports ^118,¹¹⁹ there were no responses in 30 patients with adenoid cystic carcinoma treated in two phase II studies.^120,¹²¹ c-KIT staining was assessed only with no mutational analysis performed. Stable disease was common in these studies, including stable disease longer than 6 months. However, the lack of a control group and the indolent behavior of many of these tumors leave unclear whether this stability was because of activity of imatinib or the underlying indolent nature of the tumors. Pending more effective therapies, many clinicians still consider imatinib for c-KIT-expressing adenoid cystic carcinomas because of its lack of toxicity relative to cytotoxic therapies. An ongoing phase II study is evaluating the more potent c-KIT inhibitor dasatinib.¹¹⁰

The anti-ERBB2 (formerly HER2/neu) antibody trastuzumab and the tyrosine kinase inhibitor lapatinib have demonstrated efficacy in ERBB2 overexpressing breast cancer. All histologies of malignant salivary gland tumors have been reported to overexpress ERBB2 at rates from 24% to 56%, depending on the histology, leading to a phase II study of trastuzumab with planned accrual of 50 patients.¹⁰⁶ A total of 137 tumors were screened for ERBB2, with 17% found to be overexpressed at 2+ or 3+.³⁰ This result was substantially lower than previously reported in smaller series, and the study was closed early with only 14 patients treated. A single patient with mucoepidermoid carcinoma histology achieved a long-lasting partial response.¹²³ It is relevant to note for future investigations of ERBB2 as a therapeutic target that ERBB2 overexpression in adenoid cystic carcinoma was rare at 4% but common in secretory duct cancers, with 21% in mucoepidermoid carcinoma, 83% in salivary duct tumors, and 60% in squamous histologies.³⁰ A Southwest Oncology group trial of trastuzumab also closed early owing to poor accrual. In this trial, 2 patients were enrolled and treated: one had stable disease for almost a year and the other progressed on therapy.¹²³

Lapatinib is an oral tyrosine kinase inhibitor that targets both ERBB2 and EGFR. A phase II study of this agent enrolled 29 patients with adenoid cystic carcinoma and 28 patients with non–adenoid cystic malignant salivary gland tumors.²⁴ Eighty-eight percent of the patients with adenoid cystic carcinoma and 97% of those patients with non–adenoid cystic carcinoma expressed EGFR and/or ERBB2. There were no responses in either group, but 47% of the patients with adenoid cystic carcinoma and 24% of the non–adenoid cystic carcinoma patients had stable disease lasting at least 6 months. Although the indolent nature of the disease itself likely contributed to some of this stability, biologic activity of lapatinib was likely greater than that represented by the response rate. To meet response criteria, a tumor must shrink 30%. Many patients classified as having stable disease experienced reductions in tumor volumes that failed to meet this threshold but nonetheless are suggestive of activity. Furthermore, in 36% of the patients with stable disease for more than 6 months, the disease was progressing before starting therapy, this suggests that although the therapy may not have been sufficiently active to shrink the cancer, it could at least stop its growth.

Similar results were found with the tyrosine kinase inhibitor gefitinib, which targets primarily the EGFR. At the time of presentation 21 patients were assessable for response, with no responses.¹²⁴ Thirteen of 14 patients with stable disease had adenoid cystic carcinoma, but the median duration of stable disease was only 13 weeks, again raising concerns that the disease was secondary to the indolent nature of the cancer rather than efficacy of therapy. Similarly, a phase II study of cetuximab failed to achieve any responses. However, 20 of 23 (87%) patients with adenoid cystic carcinoma achieved stable disease, with 12 of these 23 (52%) achieving stability lasting at least 6 months.¹²⁵ Of note, 11 patients with adenoid cystic carcinoma had disease that was actively progressing at the time of study entry; all of these patients achieved stable disease for at least 6 months.

Bortezomib inhibits the 26S proteosome, indirectly inhibiting NF-κB activity. A phase II study treated 25 patients with adenoid cystic carcinoma.¹²⁶ Again, no responses were seen. Patients with progressive disease were treated with the combination of bortezomib and doxil, based on preclinical data for synergy. Four patients were treated with this combination; 1 achieved a partial response and 2 had stable disease. A phase II study of the combination of bortezomib and doxil just closed secondary to poor accrual.¹²⁷

Data for the efficacy of the hormonal agent tamoxifen is limited to three case reports of long-term stable disease.^128,¹²⁹ Case reports also describe responses, including a complete response, to luteinizing hormone–releasing hormone analogues in patients with androgen-receptor positive adenocarcinoma.¹³⁰

Given the paucity of data, no regimen may be considered a standard of care for any histology of malignant salivary gland tumor. Thus, when therapy is required, the authors consider a clinical trial to be the preferred option. When no phase II trial is available, phase I studies with agents directed at relevant targets with low expected toxicity should be considered. In the absence of available trials and the need for therapy, it is reasonable to evaluate for molecular markers of interest (estrogen receptor, EGFR, ERBB2, androgen receptor, and c-KIT) and consider treating toward any marker that is present. Although response rates are low for targeted agents, multiple studies suggest the possibility of prolonged stable disease. For rapidly progressive disease, especially in the patient with good functional status, good renal function, and good bone marrow function, the combination of cisplatin plus vinorelbine may be considered. Additional clinical trials are desperately needed, but the rarity of these tumors will likely require novel cooperative strategies to promote successful accrual. Ideally, studies of targeted agents aimed at promoting stable disease would help to establish a clinically meaningful benefit through creative study designs, such as a randomized discontinuation design, but the rarity of these tumors may make such definitive design impossible.

Irradiation Techniques and tolerance

Target Volume Delineation

Elective neck treatment varies with the histology, grade, T category, anatomic site, and nodal category. Lloyd and colleagues ⁴⁹ analyzed the SEER database identifying 2667 cases of minor salivary gland cancers with known lymph node status from 1988 to 2004. Although there are inherent limitations with SEER database analyses, the results are consistent with smaller clinical series that, in turn, lack the size for a multivariate analysis. Analysis demonstrated that male sex, T3 to T4 category, primary site involving the pharynx, and high-grade adenocarcinoma or high-grade mucoepidermoid carcinomas were independent risk factors for the development of nodal metastasis for minor salivary glands.

Although these findings suggest that the biology of salivary gland malignancies do influence the risk of nodal metastases, they also suggest that lymphatic-rich anatomic sites may be relevant for minor salivary gland malignancies as seen with squamous carcinomas. For adenoid cystic carcinomas, Mendenhall and colleagues ⁸⁴ reported that this was a significant consideration in their institutional approach. They reported a 10-year neck control rate of 90% when the neck was observed versus 98% with elective irradiation. Other relevant but unestablished anatomic factors that have been well described for squamous carcinomas can also be considered during treatment planning for malignant salivary gland tumors. These include the presence of mucosal disease at the coronal midline increasing the risk for contralateral nodal metastases¹³² and the involvement of the posterior pharyngeal mucosa influencing the risk of retropharyngeal lymph node metastases.¹³³

Simulation/Field Arrangements

A CT is obtained in the treatment position. Patients are typically positioned supine, with arms down and the head and neck in the hyperextended position to move the orbits out of the radiation field. A thermoplastic face mask is used for immobilization. For cases with skull base or intracranial involvement, an MR image can be obtained and fused to the CT to assist in target delineation. The target volumes and normal critical structures are defined on the representative CT and/or MRI axial cuts.

Traditional standard techniques used to treat parotid tumors include wedged pair or mixed beam techniques, whereas intensity-modulated radiation therapy (IMRT) is now commonly used as well. It is the authors’ preference to use either a wedged pair technique or an IMRT technique (allows for comprehensive coverage of the parotid bed and cervical lymph nodes while avoiding the potential impact of electron beam dosimetry on the mandible). IMRT also allows for improved conformality at the skull base, especially when the surgical bed and scar extends superiorly above the plane of the middle and inner ear complex and may lie adjacent to the temporal lobe. Traditional radiotherapy techniques have often relied on patch fields typically with electrons to treat the course of the surgical scar if it was deemed to be at risk. This results in potential dosimetric uncertainties with a photon-electron match with the potential for a dosimetric hotspot that may lie in the temporal lobe.

IMRT may also allow for additional sparing of structures such as the cochlea and temporal lobes. Historically, hearing deficits, soft tissue necrosis, and bone necrosis have been observed in approximately 20% of patients treated with a traditional beam arrangement, especially with wedge-pair fields.¹³⁴ By using these techniques to treat the tumor bed with or without the ipsilateral neck, mean doses to the contralateral salivary glands can be effectively limited to less than 10% of the prescription dose, thereby preventing any symptoms of xerostomia during and after treatment.

Dose/Fractionation and Pathologic Considerations

Frequent pathologic indications reported for the use of postoperative radiotherapy include the presence of a positive if not a close surgical margin, perineural invasion, high-grade histology, extraglandular extension, bone invasion, tumor size, T stage, and the presence of nodal metastases. It is important to recognize that these pathologic indications have not been prospectively evaluated to determine what the absolute risk level is and the incremental risk reduction with adjuvant radiotherapy.

The typical postoperative dose to the surgical bed that has been reported has typically ranged from 60 to 66 Gy with a conventional daily fractionated schedule. For a positive or close surgical margin, localization of the area of concern based on preoperative imaging and intraoperative findings for a boost dose of radiation typically of 6 Gy to a total dose of 66 Gy is an important consideration. Other risk-stratified radiotherapy dose recommendations have not been clearly established and thus it is unclear if areas involved with extraglandular extension (i.e., T3 and T4) and nodal metastases necessitate a dose higher than 60 Gy.

For perineural invasion, an increased risk of locoregional relapse has been correlated especially for adenoid cystic carcinomas. Recent detailed histologic studies of serial sections suggest that the mechanism in which perineural invasion may be contributing to relapse is the result of the direct contiguous spread of cancer along the branches of the nerves.⁵¹ Therefore a judicious clinical target volume surrounding the primary tumor bed should be created to reflect the potential subclinical extent of cancer. The use of postoperative radiotherapy appears to be effective in reducing the risk of local relapse caused by perineural invasion (10-year local control rates of tumors with perineural invasion treated with surgery vs. surgery and postoperative radiotherapy were 60% vs. 88%, p = .01).⁵⁰

For more extensive perineural invasion resulting in clinically evident (symptomatic or radiologic) disease there is a further increased risk of locoregional relapse, especially with adenoid cystic carcinomas.⁷⁹ This observation is consistent with other cancer sites in which the size of the involved nerve appears to be increase the risk of relapse.¹³⁵ With involvement of major named nerves, the increased risk of relapse can be caused by the spread to the skull base and also with an increased risk of distant relapses. Thus elective irradiation along the course of the nerve is recommended.

For elective nodal irradiation the anatomic draining lymphatics should be encompassed. The usual dose has ranged from 50 to 54 Gy with a conventional daily fractionated schedule.

Gross, unresectable disease has typically been treated to 70 Gy, again with a conventional daily fractionated schedule. The optimal dose and fractionation schedule has not been clearly defined especially for unresectable gross disease. This is important to consider in light of evidence suggesting that radiotherapy dose escalation with photons ⁹⁹ and protons¹⁰⁵ may offer additional improvements in local control rates. Unknown at this time is whether there may be incremental gains in locoregional disease control rates with the practice of a simultaneous in-field boost (“dose painting”) that is now possible with the use of IMRT.

Treatment Side Effects

The side effects associated with radiation therapy are dependent on several factors: the site and volume treated, the dose fractionation regimen and total dose delivered, whether treatment is delivered definitively or adjuvantly, the type of radiation used (low LET vs. high LET), and whether concomitant chemotherapy is utilized.

In the treatment of salivary gland tumors, acute-phase reactions (within the first 90 days of completion of radiation therapy) are generally limited to rapidly proliferating tissues, such as the skin, and the oral and pharyngeal mucosa. Skin reactions can range from a mild erythema to moist desquamation, which would require aggressive skin care and antibiotic treatment. Xerostomia and dysgeusia may also occur during radiation therapy, but the risk of long-term xerostomia can be minimized with appropriate field arrangements and treatment planning, by excluding the contralateral parotid and submandibular gland. Acute mucositis is commonly observed if the oral cavity or pharyngeal mucosa is in the field. The risk of developing severe mucositis is increased with approaches such as accelerated fractionation, concurrent chemotherapy, or high LET radiation. Attention to pain control, close monitoring for the development of oral candidiasis, and aggressive nutritional support (with the placement of a percutaneous gastrostomy polyethylene glycol tube, if needed) are required with severe mucositis.

Late complications (occurring >90 days after completion of treatment) are associated with slower proliferating tissues, with side effects limited to the particular region treated. In treating the base of skull, careful treatment planning is required to limit the dose delivered to critical structures, thereby minimizing the risks of side effects such as pituitary dysfunction, temporal lobe necrosis, optic neuropathy, keratitis, and hearing loss. With conventional 1.8 to 2.0 Gy fractionation, the dose to the spinal cord should be limited to 45 Gy (maximum), cochlea to 45 Gy (mean), lacrimal gland to 30 Gy (mean), optic nerves and chiasm to 54 Gy (maximum), temporal lobes to 65 Gy (maximum), and brain stem to 60 Gy (maximum). These doses need to be appropriately adjusted if nonstandard approaches (concurrent chemotherapy, altered fractionation, high LET radiation) are used. Every patient should undergo preirradiation dental evaluation, with any necessary extractions to be done before initiation of radiation therapy; and if dentition is to be preserved, custom fluoride trays should be utilized daily. If extractions or oral surgery needs to be done after radiation therapy, hyperbaric oxygen therapy is recommended to minimize the risk of developing osteoradionecrosis. Another common late side effect in patients receiving radiation therapy to the neck is hypothyroidism. In patients receiving unilateral radiotherapy, the risk of xerostomia can be minimized through conformal delivery (via three-dimensional conformal radiotherapy, IMRT, or proton therapy), limiting the dose to contralateral major salivary glands, with mean doses less than 10% of the prescription dose readily achievable.

Treatment Algorithm, Conclusions, Controversies, and Future Possibilities

In summary, salivary gland tumors are characterized by their rare incidence, diverse histology, and the potential for an indolent natural history. It is this heterogeneity that has hampered clinical progress in the management of salivary gland tumors. In response, progress has been made in further understanding the genetic and molecular characteristics of these tumors but have not to date yielded any significant therapeutic insights. However, the favorable survival associated with expression of the CRTC1-MAML2 fusion oncogene in mucoepidermoid carcinomas is an important advancement toward greater homogeneous classification of salivary gland tumors. Therapeutic progress has been made with technical advances in surgical techniques such as the transoral robotic technique and in radiation oncology with the maturation of the use of fast neutron radiotherapy. The increasing study of charged particle beam radiotherapy, both protons and heavy ion beams, offers the potential to further reduce toxicities with improved locoregional control rates. Although various molecular targeted therapeutics have been introduced into clinical trials, the ability to make a significant impact on the risk of distant metastasis has not yet been realized. However, the toxicities associated with these systemic approaches are improved and disease stabilization may be realized with several targeted agents.

In conclusion, the treatment options for patients with salivary gland tumors have significantly improved in recent years, offering the potential for more effective locoregional therapy that preserves function. The authors’ preferred treatment algorithm is seen in Table 35-10. Major survival gains will likely be realized only with the development of more effective systemic therapy.

TABLE 35-10 Treatment Algorithm for Malignant Major and Minor Salivary Gland Cancers

Clinical Situation	Standard Therapy	Proposed Clinical Trial
Complete resection, adjuvant therapy	Postoperative, adjuvant EBRT when indicated (see Table 35-9)	Intergroup trial of surgical resection + postoperative adjuvant EBRT vs. surgical resection + postoperative adjuvant EBRT with concomitant and maintenance chemotherapy
Locally advanced (primary or recurrent; unresectable or resected but residual)	1.High-dose conventional photon irradiation (consider altered fractionation)	1.Neoadjuvant chemotherapy followed by resection (if feasible) and EBRT with concomitant and maintenance chemotherapy
	2.Maximal surgical resection, conventional EBRT with concomitant chemotherapy	2.High-dose conventional or altered-fractionation EBRT (photons or protons) with concomitant and maintenance chemotherapy
	3.Neutron beam therapy ± SRS boost	3.Carbon ion therapy
Locally recurrent, prior irradiation	Low- to moderate-dose EBRT, consider SRS or SRT Palliative chemotherapy