Nasopharyngeal carcinoma is an uncommon tumor in the United States, where the incidence is approximately 0.2 to 0.5 cases per 100,000 people.² In comparison, the incidence is considerably greater in southern China and Hong Kong (25 to 50 cases per 100,000 people) and among the Inuit residing in Alaska and Greenland (15 to 20 cases per 100,000 people). The incidence of nasopharyngeal cancer is also higher in other parts of Southeast Asia (Taiwan, Vietnam, and Thailand), the Philippines, and Malaysia and in some Mediterranean and North African populations (8 to 12 per 100,000 people). These areas of increased incidence of nasopharyngeal cancer are considered endemic areas, and the incidence outside endemic areas is much lower and more often associated with tobacco use.

Several geographic-specific etiologic factors have been implicated in the incidence variations throughout the world. The high consumption of salted fish in southern China and Hong Kong has been implicated as a possible etiologic factor for nasopharyngeal carcinoma in these areas.^3,⁴ It has been suggested that various macromolecular lignins associated with these foods in southern China (and perhaps other areas such as Alaska) may activate EBV,⁵ which has also been identified as a probable etiologic factor in nasopharyngeal carcinoma.^6,⁷ Studies by Bouvier and colleagues⁵ involved the fractionation of harissa, a homemade spice mixture used in Tunisia on various foods, including salted fish. Harissa was separated into various macromolecular fractions by column chromatography. The lignin-containing complexes extracted from harissa induced the EBV promoter in Raji cells.⁵ The possibility that these environmental factors are important in the etiology of nasopharyngeal carcinomas is further supported by the finding that incidence rates decrease for successive generations of people who originally emigrated from southern China to California.⁸

Other potential environmental etiologic factors that have been associated with nasopharyngeal carcinoma include alcohol consumption and exposure to dust, fumes, formaldehyde, and cigarette smoke.^9,¹⁰ Cigarette smoke and alcohol have long been associated with many other head and neck carcinomas, but their association with nasopharyngeal carcinoma has been controversial. Some studies have suggested that alcohol consumption and cigarette usage were not associated with nasopharyngeal carcinoma.⁹^–¹¹ Nam and associates¹⁰ conducted a case-control study using a National Mortality Follow-Back Survey based on death certificates and found that cigarette smoking and alcohol consumption are independent statistically significant risk factors for nasopharyngeal carcinoma. The increased risk of nasopharyngeal carcinoma with heavy smoking (adjusted for alcohol consumption) was threefold, and an excess risk of 80% was demonstrated for heavy alcohol consumption (adjusted for cigarette usage).

Prevention and Early Detection

The nasopharynx cannot be visualized externally, and tumors of this area often present after they have eroded into a vital structure and produced a presenting symptom. The association of EBV and nasopharyngeal cancer has led some investigators to hypothesize that serologic screening (EBV titer) may be useful in certain high-risk populations to identify groups of patients who might benefit from frequent nasopharyngeal examinations.¹² In a study from China of 338,868 patients who underwent serologic screening for EBV titer, 9367 persons had immunoglobulin (Ig) A antibodies to EBV. Of these 9367 patients, 306 were positive for IgA to EBV early antigen. Nasopharyngeal cancer was detected in 113 of the 9367 patients (1.2%) and in 63 of the 306 (20.5%) who were positive for IgA to EBV early antigen. Most of the tumors (>85%) were early-stage lesions. Serologic EBV IgA screening is currently being used in endemic areas.

In addition to EBV serologic screening as a means of early detection and possible prevention of advanced-stage disease, several environmental and genetic predisposing factors are being explored as possible markers that could identify groups of patients at high risk for nasopharyngeal cancer.¹³ Real-time polymerase chain reaction (PCR) techniques show promise as screening tools for nasopharyngeal cancer.¹⁴ EBV DNA detection, in particular detection of EBV nuclear antigen, eliminates many false-positive results and improves the sensitivity and specificity of EBV IgA serologic screening.¹⁵ A retrospective analysis of blood samples from patients with disease relapse from previous clinical trials indicates that EBV nuclear antigen screening by PCR would have led to earlier detection of distant failure but, because of the quality of current imaging, would not have significantly affected detection of local recurrences.¹⁶ Real-time PCR screening for EBV latent membrane protein 1 (LMP1) in nasopharyngeal swabs has also been shown to be a promising screening tool for high-risk populations, with a sensitivity of 87% and specificity of 98%.¹⁷ These techniques are likely to replace serum IgA screening in endemic areas and have a promising role in screening for relapse in EBV-positive patients.¹⁸ The further study of other environmental and genetic markers may lead to the identification of patient populations that could benefit from screening or chemoprevention.¹³

Chemoprevention has been tested extensively in patients with a prior diagnosis of head or neck malignant tumors because second primary tumors in the respiratory and digestive tracts are common in that population.¹⁹^–²² Hong and others^19,²⁰ conducted a prevention trial for patients with head and neck carcinoma after curative surgery or radiotherapy, or both. Patients were randomly assigned to receive 1 year of 13-cis-retinoic acid (isotretinoin) as a chemoprevention agent (see Khuri and associates¹³ for a discussion of the biologic factors) or placebo. The treated group had significantly fewer second primary tumors (4%) than the placebo group (24%) (p = .005); however, there was no difference in rates of survival or incidence of relapse between the two groups. In addition, the chemopreventive effect appeared to abate after 3 years; the incidence of second primary tumors was similar in the two groups after 3 years. Acute toxicity was significant.¹⁹

The results of the trial by Hong and colleagues ¹⁹ were not confirmed by a controlled trial of 316 patients by Bolla and associates,²¹ in which patients were randomly assigned to 2 years of etretinate (a retinoic acid similar to isotretinoin) at a maintenance dosage of 25 mg/day. Subsequently, a large intergroup effort was mounted, and 1302 patients with previously treated head and neck malignant tumors were randomly assigned to receive 3 years of isotretinoin or placebo.²² Because of unacceptable toxicity seen with higher doses of isotretinoin (100 to 200 mg/m² [≈150 to 400 mg total]) in prior studies, the dosage was decreased to 30 mg/day in this trial. No difference was seen in the occurrence of second primary tumors (4.6% in both arms), which was the primary endpoint of the study.²³ A decrease in local recurrence was seen, however, and 13-cis-retinoic acid remains a topic of investigation.

The issue of whether retinoids or other chemopreventive agents should be used in the setting of premalignant lesions of the oral cavity has been considered, and many trials have suggested that retinoids can induce considerable responses in oral leukoplakia.^24,²⁵ Toxicity has been a limiting issue in these trials, however. It will be important to determine whether these chemoprevention strategies, developed for leukoplakia, can be extrapolated to prevention of nasopharyngeal cancer in subgroups of patients at high risk for such malignant tumors.

Blockade of cyclooxygenase-2 (COX-2) has also shown some promise as a chemopreventive strategy, and COX-2 inhibitors were recently the topic of investigation for various malignant tumors, including head and neck cancers.²⁶ A phase II study at Fox Chase Cancer Center evaluated a 3-month course of celecoxib in patients with biopsy-proven dysplastic or hyperplastic leukoplakia, with results pending (NCT00101335). In another pilot randomized study of celecoxib in oral premalignant lesions conducted by the M.D. Anderson Cancer Center (MDACC), doses of 100 mg and 200 mg twice daily were shown to be ineffective in controlling these lesions.²⁷ Most recent COX-2 inhibitor cancer prevention trials have been closed, except in very high risk settings such as familial polyposis, because of the increased rate of cardiovascular events seen in a few of the celecoxib and rofecoxib prevention trials.

Pathology

Malignant tumors of the nasopharyngeal area are generally carcinomas (90%), with lymphomas making up approximately 5% of lesions.²⁸ Histologic classification of carcinomas continues to evolve as we find out more about the biologic characteristics of these tumors. The most current pathologic classification of nasopharyngeal carcinomas by the World Health Organization (WHO) includes three major categories: keratinizing squamous cell carcinoma (formerly, WHO type I), nonkeratinizing squamous cell carcinoma (including differentiated, or former WHO type II lesions, and undifferentiated, or former WHO type III lesions), as well as recently described basaloid squamous cell carcinoma.²⁹

On histologic testing, keratinizing squamous cell carcinomas appear to be well differentiated with intercellular bridges. These lesions make up approximately 20% of carcinomas.^29,³⁰ Differentiated nonkeratinizing carcinomas, representing 30% to 40% of nasopharyngeal carcinomas, lack well-defined squamous cell characteristics but continue to show a “pavement stone pattern” characteristic of squamous cell histology.²⁹ Undifferentiated nonkeratinizing carcinomas, comprising 40% to 50% of carcinomas, also formerly known as lymphoepitheliomas, are characterized by a lymphoplasmacytic infiltrate.²⁹ On cytologic testing, the cells appear uniform, with round to oval nuclei and prominent nucleoli. The frequency of the different histologic types varies with geographic area. In the United States, the majority of nasopharyngeal carcinomas have keratinizing squamous cell histologic findings, whereas in Asia, the undifferentiated histologic type is by far the most common (Table 32-1).

TABLE 32-1 Irradiation Alone Results of Series in Three Diverse Locations

Anatomy and Pathways of Spread

The nasopharynx is a musculofascial tube that connects the nasal cavity and oropharynx. The anatomic borders of the nasopharynx are (1) the anterior border, consisting of the posterior nasal apertures and nasal septum; (2) the posterior border, the pharyngeal mucosa; (3) the superior border, the pharyngeal mucosa and body of the sphenoid; and (4) the inferior border, the oropharynx. The lateral wall of the nasopharynx contains the pharyngeal opening of the auditory tube. The medial cartilaginous extension of the auditory tube forms a protrusion from the lateral wall of the nasopharynx at the superior and posterior aspects of the opening of the auditory tube. This protrusion, with its overlying mucosa, creates the torus tubarius. Just posterior to the torus tubarius lies the pharyngeal fossa, or the fossa of Rosenmüller, which is formed by the junction of the lateral and posterior walls of the nasopharynx (Fig. 32-1A to D).

Figure 32-1 Anatomic relationships of the structures of the nasopharynx and adjacent structures. A, Coronal section through the center of the nasopharynx, with view of lateral structures. B, Sagittal section through the nasopharynx lateral to midline, with a view of the lateral relationships of the nasopharynx (inset). C, Sagittal section through the nasopharynx, with a view of the anterior and adjacent structures. D, Sensory innervation of the nasopharynx. Roman numerals refer to cranial nerves. a, artery; m, muscle; n, nerve; v, vein; V₁, ophthalmic division of trigeminal nerve; V₂, maxillary division of trigeminal nerve; V₃, mandibular division of trigeminal nerve.

A to C, Modified from Clemente CD: Anatomy: A Regional Atlas of the Human Body. Philadelphia, Lea & Febiger, 1975. D, Modified from Pernkopf E: In Ferner H (ed): Atlas of Topographical and Applied Human Anatomy, 2nd ed. Baltimore, Urban & Schwarzenberg, 1980.

The superior wall of the nasopharynx warrants review. Extension of tumors superiorly often leads to invasion of structures, the symptoms of which are typical of the initial clinical presentation of patients. The sphenoid bone lies superior to the mucosa and pharyngobasilar fascia of the nasopharynx. Several cranial nerves exit the base of skull in this area. Critical foramina housed in the sphenoid include the superior orbital fissure (cranial nerves III, IV, V [ophthalmic division], and VI), the foramen rotundum (cranial nerve V [maxillary division]), the foramen ovale (cranial nerve V [mandibular division]), and the foramen spinosum (recurrent branch of cranial nerve V [mandibular division]). The foramen lacerum is formed by the junction of the sphenoid and temporal bones. The internal carotid artery passes over the superior opening of the foramen lacerum but does not traverse through the foramen. The nerve of the pterygoid canal and the meningeal branch of the ascending pharyngeal artery course through the foramen lacerum. Deep to the mucosa of the superior aspect of the lateral wall of the nasopharynx lies the superior pharyngeal constrictor muscle. Beyond this muscle lie cranial nerves IX, X, and XI, which exit the base of the skull from the jugular foramen in the temporal bone, and cranial nerve XII, which exits the base of the skull through the hypoglossal canal of the temporal bone.

The sensory innervation is shown in Figure 32-1D. The maxillary division of the trigeminal nerve supplies the upper nasopharynx, the posterior part of the nasal cavity, and most of the palate and upper gums of the oral cavity. The general sensory branches of the lingual and pharyngeal branches of the glossopharyngeal nerve supply the sensory innervation of the lower part of the nasopharynx, the posterior third of the tongue, and part of the soft palate and oropharynx.

Understanding the anatomic relationships of the nasopharynx and adjacent structures is important to correctly diagnose patients with nasopharyngeal carcinoma. The importance of superior extension into the sphenoid bone of the base of the skull is exemplified by the fact that cranial nerves can be involved at presentation. The frequency of involvement of cranial nerves has been described in two large series.31,32 Cranial nerves V and VI are the most commonly involved. These nerves traverse the sphenoid bone and can be involved when tumors erode superiorly through bone. Tumors of the nasopharynx can also gain access to the cranial nerves of the base of the skull by eroding superiorly through the foramen lacerum, which is bordered superiorly by the internal carotid artery (Fig. 32-1B).

Lateral extension of nasopharyngeal carcinomas can lead to erosion of the medial opening of the auditory tube and the medial pterygoid plate and can involve cranial nerves IX, X, XI, and XII (Fig. 32-1B and C). Lateral extension can lead to involvement of the carotid artery and internal jugular vein.

Lymph node involvement is common; 65% to 80% of patients present with clinically involved cervical neck nodes. The level VA and level II lymph nodes are commonly involved. The recently published frequency of involvement of these areas in two large series 31,32 is presented later in the Clinical Manifestations section.

Biologic Characteristics and Molecular Biology

The potential etiologic link between EBV and nasopharyngeal carcinoma was first described more than 30 years ago ⁶; however, recent advances in molecular biology have shed further light on this association. Earlier work focused on the serologic response to EBV and showed elevated levels of IgA and IgG antibodies to the viral capsid antigen and a replication protein called early antigen in patients with nasopharyngeal carcinoma. Other studies have demonstrated the presence of EBV DNA and encoded proteins directly in the nasopharyngeal carcinoma tissue samples.³³ Latent infection of cells with EBV, initially studied in B lymphocytes, is associated with regulated expression of several viral genes, including latent membrane proteins (LMPs) 1, 2A, and 2B, six EBV nuclear antigens (EBNAs) 1, 2, 3A to C, and LP, and two small noncoding nuclear RNAs (EBERs).³⁴ Some of the molecular mechanisms by which expression of these genes can lead to transformation have been elucidated. For example, LMP1 induces expression of epidermal growth factor receptor (EGFR) and may therefore influence cellular growth. A restricted pattern of these EBV antigens is present in nasopharyngeal cancer cells, which may in part explain why immunotherapy directed against these antigens has been more successful in other EBV-associated diseases.³⁵ Allogeneic cytotoxic T cells to various EBV antigens have been developed and show some promise for treating locally recurrent nasopharyngeal carcinoma.³⁶

Several studies have suggested that serologic responses to EBV may be more specific for the nonkeratinizing carcinoma (former WHO types II and III).³⁷ This, however, was not confirmed by another group.³⁸ Using the Southern blot technique to detect the viral DNA, Raab-Traub and others³⁹ showed the presence of EBV in all former histologic variants of nasopharyngeal carcinoma, but the EBV copy number was lowest in keratinizing squamous cell carcinomas. The presence of EBV in keratinizing squamous cell carcinoma has been consistently detected by other groups.⁴⁰^–⁴³ The differences in the rate of EBV detection by different investigators may stem from variability in the sensitivity of different assays. Alternatively, it may reflect differences in the sampled populations. The detection rate of EBV infection in keratinizing squamous cell nasopharyngeal carcinoma is highest in endemic areas.

Understanding the molecular basis of EBV-induced carcinogenesis and the role of the immune system response in this process will likely have a direct bearing on clinical practice. An association between decreasing antibody titers and response to therapy suggested that antibody titers could be used to assess a patient’s prognosis.⁴⁴ Quantification of EBV DNA levels in pretreatment and postradiotherapy plasma samples from patients in Hong Kong has been shown to correlate with outcome,⁴⁵ and, therefore, could prove to be a useful tool in risk stratification and planning of therapeutic interventions in high-risk areas.

In addition to environmental factors, the potential association of various genetic factors and nasopharyngeal cancer is a topic of current investigation. Earlier work implicated a disease susceptibility gene linked to the human leukocyte antigen (HLA) region, which has been associated with the increased incidence of nasopharyngeal carcinoma in southern China.⁴⁶ Simons and associates⁴⁷ originally described an association in Chinese patients with nasopharyngeal carcinoma and the HLA-A2 antigen with a deficit of the second antigen at the second locus (B locus). Later, Simons and co-workers⁴⁸ reported on 110 Chinese patients in Singapore with nasopharyngeal carcinoma and hypothesized that the new B-locus antigen (Sin 2) may be associated with the tumor. It remains unclear whether specific HLA alleles may influence development of nasopharyngeal carcinoma directly, possibly by affecting the immune reaction to EBV, or whether, and perhaps more likely, the genetic susceptibility to nasopharyngeal carcinoma is encoded by a locus in close linkage disequilibrium. Various other genetic and epigenetic alterations have been reported in nasopharyngeal carcinoma. Deletions of the short arm of chromosome 3 (3p25, 3p14)²⁹ and of chromosome 9 (9p21-22)⁴⁹ are the most common cytogenetic changes. Candidate genes in these and other locations of interest are being identified. These findings may lead to a greater understanding of the genetic basis of nasopharyngeal carcinomas and the interplay of environmental factors with these genetic factors.

Clinical Manifestations, Patient Evaluation, and Staging

As mentioned, nasopharyngeal carcinoma is rare in the United States and, therefore, is not often suspected as a possible cause of a patient’s early symptoms. In addition, the list of possible early symptoms of nasopharyngeal carcinomas includes many symptoms that could have more common causes. In a series of 378 patients from the MDACC,³² the presenting symptoms included a neck mass in 41%; hearing loss, ear drainage, or otalgia in 27%; nasal bleeding or obstruction in 21%; cranial nerve deficits in 8%; and other nonspecific symptoms in 8%. Investigators from Washington University noted that the typical presentation involved multiple symptoms.³¹ In this series of 143 patients, presenting symptoms included otitis in 43%, throat pain in 39%, nasal obstruction in 37%, a neck mass in 35%, nasal bleeding in 29%, cranial nerve involvement in 24%, and trismus or other symptoms in 5%.

Similar to these data, most series have shown that neck masses, obstructive ear symptoms, and cranial nerve deficits are common presenting symptoms for nasopharyngeal carcinoma. In the Washington University series,³¹ 66% of patients were found to have ipsilateral neck masses and 28% had contralateral neck masses on examination, although only 35% of patients had reported a neck mass as the reason for seeking medical advice. Of the patients with clinically involved neck disease, 60% had enlarged ipsilateral level II lymph nodes and 32% had enlarged ipsilateral level V lymph nodes. The most frequently involved cranial nerves in the series were cranial nerve VI in 15% of patients, V in 7.7%, and VIII, X, and XII in 5.6% of patients each.

In the series from the MDACC,³² the level VA lymph nodes were most commonly enlarged (54% of patients), followed by the level II nodes (49% of patients). The level III and lower level VA/upper level VB groups were involved in 24% and 22% of patients, respectively. The level IV lymph nodes, level VB lymph nodes, and supraclavicular lymph nodes were involved in 10%, 13%, and 10% of patients, respectively. Similar to the series from Washington University, cranial nerve VI was the most frequently involved nerve (6% of cases).

Patient Evaluation

The initial evaluation of patients with nasopharyngeal carcinoma should include a history and physical examination, with special attention to the level of nodal involvement (if any) (Fig. 32-2). It is helpful to diagram or digitally photograph the clinically involved lymph nodes in the neck because it can be helpful when the radiotherapy boost is considered.

Figure 32-2 Diagnostic algorithm for nasopharyngeal carcinoma.

Patients should undergo flexible fiberoptic nasopharyngoscopy to determine the involvement of mucosal surfaces. A diagram or digital photograph of these findings should also be made. Biopsy of the primary lesion should be attempted first. If this procedure gives indeterminate results, a needle biopsy of a potentially involved lymph node should be taken.

CT and MRI of the head and neck are useful in the evaluation of both erosion of tumor into the bony structures of the base of the skull and retropharyngeal and cervical lymphadenopathy. Although the same information is often provided by CT and MRI, some investigations have suggested that MRI may be more useful in delineating soft tissue invasion outside the nasopharynx and the extent of retropharyngeal lymph node involvement,⁵⁰ whereas CT may be most useful in delineating skull base erosion.^51,52,53 Given the level of anatomic information required for the current T staging system, MRI is becoming the imaging study of choice in the evaluation of nasopharyngeal tumors.

Positron emission tomography (PET) can be useful for staging ^54,⁵⁵ as well as for post-therapy management where other imaging is unclear. PET/CT, which provides additional anatomic information, is more commonly used now and may be helpful for radiotherapy planning.

The completion of the diagnostic evaluation involves acquiring routine complete blood counts, a chemistry panel, and a chest radiograph (see Fig. 32-2). Further evaluation of possible metastases should be done on the basis of the clinical presentation of the patient.

Staging

After a pathologic assessment has been made, the workup or evaluation leads to a clinical stage (Table 32-2). Clinical staging has been a matter of controversy, and the prognostic significance of the various TNM staging criteria remains under investigation. Several staging systems have been in use throughout the world. The most commonly used systems in North America and Europe are those of the American Joint Committee on Cancer (AJCC)⁵⁶ and the International Union Against Cancer (Union Internationale Contre le Cancer [UICC]),⁵⁷ which have become essentially analogous. In Asia, where nasopharyngeal carcinoma is endemic, the Ho⁵⁸ classification system was initially developed in Hong Kong. In addition, in 1992, Chinese physicians adopted an independent system similar to the Ho system.⁵⁹ All four systems, shown in Table 32-2, have limitations and continue to evolve, with researchers often drawing on each others’ experience. Updated versions of the Chinese system and the AJCC system were released in 2008^60,⁶¹ and 2010,⁶² respectively. The AJCC/UICC remains the most commonly used system in the English literature.

TABLE 32-2 Comparison of Staging Systems

A well-recognized advantage of the Ho classification system lies in its approach to lymph node metastasis, namely, use of lymph node location in N-stage assignment and a more even distribution of nodal disease in stage grouping ^58,⁶³ (see Table 32-2). Nodal involvement appears in stage II, and supraclavicular nodal metastasis, which is known to carry a worse prognosis, is represented in a separate stage IV category (see Table 32-2). The 1992 AJCC system used lymph node size and laterality in N-stage assignment, both of which carry some prognostic significance. However, only some patients with N1 disease were included in stage III, and most patients with nodal involvement were grouped into stage IV, together with patients with metastatic disease. Modifications to the AJCC classification implemented in 1997⁶⁴ and maintained in 2002⁶⁵ and 2010⁶² have incorporated lymph node location into N-category assignment, designating involvement of the supraclavicular fossa, as originally defined by the Ho system, as N3 disease, and have placed all N3M0 lesions in a separate stage IVA grouping. In addition, stage groupings were changed to provide a more even stage distribution. These changes improved risk stratification of the 1997 AJCC classification in comparison to both the prior version of the AJCC system and the 1978 Ho system.^66,67

Important changes to the most recent AJCC system targeted T-category assignment. Invasion into the soft tissues of the nasopharynx was used in the 1997 and 2002 AJCC classification to segregate T1 and T2 lesions, but studies have shown that this carries no prognostic significance.⁶⁸^–⁷⁰ The parapharyngeal involvement used to segregate T2a and T2b lesions does,^71,72,73 however, and therefore the new 2010 AJCC system segregates lesions with parapharyngeal involvement into the T2 subgroup, whereas all tumors either confined to the nasopharynx or extending into surrounding subsites but without parapharyngeal involvement are now in the T1 subgroup. Furthermore, cranial nerve involvement carries a significantly worse prognosis than base-of-skull involvement,^31,^69,⁷⁴^–⁷⁶ but both were aspects of the T4 subgroup in the 1997 and 2002 AJCC systems and of the T3 subgroup in the Ho system. In the 2003 and 2010 versions, akin to the 2008 Chinese classification system, base-of-skull involvement has been downstaged to T3. Furthermore, data are emerging that extensive cranial nerve involvement, orbit involvement, and intracranial extension are associated with worse outcomes within the T4 subgroup.^67,76 Therefore, some of the proposed modifications to the AJCC system include a separation of the T4 subgroup into two categories. T4a would include tumors with involvement of the masticator space (infratemporal fossae), asymptomatic radiographic cranial nerve involvement, and involvement of the hypopharynx. T4b would include tumors with intracranial extension, orbital involvement, and symptomatic cranial nerve palsies. However, more data need to be collected to validate this proposed modification.

With respect to N-category classification, the seventh edition of the AJCC staging manual for the first time includes disease that has spread to the retropharyngeal lymph nodes in staging, placing it in the N1 category. On MRI evaluation, these nodes are involved in 83% of patients with nasopharyngeal carcinoma compared with 74% involvement of the level II to IV nodes, and are considered to be the first echelon of nodal spread.⁷⁷ Therefore, their inclusion into the staging system is expected to improve the prognostic accuracy.^78,⁷⁹

The evolution of technology and, specifically, CT and MR imaging, has allowed for more precise evaluation of tumor invasion into surrounding structures and lymph node involvement, and will undoubtedly lead to improvement in staging accuracy and risk stratification.

Primary Therapy

Traditionally, the nasopharyngeal area has not been easy to examine without fiberoptic technology and has been difficult to approach surgically. Surgical exposure of the area and resection of tumors with adequate tumor margins have long been challenging.⁸⁰ For these reasons, primary surgical intervention fell out of favor in the 1950s; primary treatment has generally consisted of radiotherapy alone and, more recently, radiotherapy plus concurrent chemotherapy.

Single-Modality Therapy

Although histologic presentations of nasopharyngeal tumors vary throughout the world, with more undifferentiated tumors found in southern China and Hong Kong, the 10-year survival rates for patients treated with radiation therapy alone in the United States (MDACC),³² Denmark,⁸¹ and Hong Kong⁸² are similar, at 34%, 37%, and 43%, respectively (see Table 32-1). The rate from the Hong Kong study may be somewhat inflated because many patients had early-stage disease (node-negative disease in 39%) and few patients with the keratinizing histology (0.3%) were present in the series.⁸² However, elevation of EBV titers has been correlated with a poor prognosis,³⁴ and higher titers would be expected in Hong Kong, an endemic area. The North American and European populations have a greater percentage of patients with keratinizing (WHO type I) histologic findings than with nonkeratinizing histologic findings (WHO types II and III), however, and the prognosis for the former has been poorer than for the latter.⁶⁸ Therefore comparison of groups from various parts of the world is difficult, and our understanding of the role of EBV and histologic classifications is incomplete. Still, some general statements can be made.

The MDACC series elucidated the long-term outcome in a large population of patients treated with radiotherapy alone.³² Using the 1992 AJCC staging system (see Table 32-2), researchers found that advanced T category, squamous histologic typing, and cranial nerve deficits were predictive of poor prognosis for local control in univariate and multivariate analyses. Five-year local control rates for T1, T2, T3, and T4 categories were 93%, 79%, 68%, and 53%, respectively. Similar results were observed in other, smaller series.^31,⁸³^–⁸⁶

The MDACC results illustrate the good tumor control that can be achieved with radiotherapy alone for early-stage lesions but demonstrate the need to explore multimodality options for more advanced lesions. Significant improvements were achieved with a combined chemoradiation approach and the evolution of radiotherapy techniques. Altered fractionation schemes were initially explored, but the improved therapeutic ratio associated with the use of intensity-modulated radiotherapy (IMRT) has made this technique the radiotherapy of choice.

Combined-Modality Therapy

Irradiation Plus Chemotherapy

The randomized trial of Sanchiz and associates ⁸⁷ served as an introduction for using chemotherapy in nasopharyngeal cancer. However, only a small number of patients with nasopharyngeal cancer were included.

Brizel and co-workers ⁸⁸ reported a small randomized trial comparing twice-daily irradiation (1.25 to 75 Gy) to twice-daily irradiation (1.25 to 70 Gy) with concurrent cisplatin and 5-fluorouracil (5-FU) treatment. This trial included 121 patients with locally advanced (T3 and T4) squamous cell carcinomas of all head and neck sites (including some nasopharyngeal carcinomas). With a median follow-up of 41 months, the locoregional control rate at 3 years was 70% for the combined-modality arm versus 44% for the irradiation alone arm (p = .01). The overall survival (OS) at 3 years was 55% for the combined-modality arm versus 34% for the irradiation alone arm (p = .07). Mucositis appeared to be more prevalent in the combined-modality arm; 45% of patients required feeding tubes compared with 28% in the irradiation alone arm. However, the severe late complications of soft tissue necrosis or osteoradionecrosis were uncommon, with similar numbers in both arms (three patients in the irradiation group and five in the combined-modality group).

Several groups have conducted phase III trials specific to nasopharyngeal primary tumors (instead of a wide array of head and neck cancers). The trial design compared chemotherapy plus irradiation versus irradiation alone * (Table 32-3).

TABLE 32-3 Major Concurrent Chemotherapy and Radiotherapy Randomized Trials

The intergroup study 0099 showed a dramatic improvement in survival rates with combined-modality treatment.¹ The trial was closed at a planned interim analysis (October, 1995) after review of the preliminary results by the Data Safety and Monitoring Committee for the Southwest Oncology Group. The results are shown in Table 32-3. Patients treated with standard irradiation received 70 Gy to gross disease in daily fractions of 1.8 to 2 Gy. Patients enrolled in the combined-modality arm received 100 mg/m² of cisplatin on days 1, 22, and 43 of irradiation, and cisplatin (80 mg/m²) and 5-FU (1000 mg/m²) on days 1 to 4 every 3 weeks for three courses after irradiation. A marked statistically significant improvement in both disease-free survival (DFS) and OS in the combined-modality arm was documented at both initial analysis and in the final update of the trial published in 2001.⁹³ The 5-year OS for patients with the keratinizing squamous cell carcinoma histologic type (WHO type I) was 37%; the differentiated nonkeratinizing histologic type (WHO type II), 55%; and the undifferentiated nonkeratinizing histologic type (WHO type III), 60%.

Although the first reported results from a chemoradiation trial conducted in an endemic area were disappointing,⁹¹ the recently published final update⁹⁴ as well as four additional independent trials conducted in Taiwan,⁸⁹ Hong Kong,^92,98 and Singapore⁹⁶ all confirmed an advantage to cisplatin-based chemoradiation (see Table 32-3). A randomized trial from Hong Kong of concurrent weekly cisplatin chemotherapy (40 mg/m²), without adjuvant or neoadjuvant therapy, versus irradiation demonstrated a significant improvement in progression-free survival (PFS) at a median follow-up of 2.71 years⁹¹ and in OS at a median follow-up of 5.5 years (hazard ratio [HR], 0.51; p = .013)⁹⁴ for a subset of patients with T3 to T4 disease. A trial conducted by Lin and colleagues⁸⁹ in Taiwan randomized 284 patients to irradiation alone versus irradiation with two cycles of concurrent cisplatin (20 mg/m²/day) and 5-FU (400 mg/m²/day) by continuous infusion given over a course of 4 days. In this trial, combined-modality treatment was associated with a significant survival advantage. The differences in the degree of benefit derived from the combined-modality treatment in these two studies may stem from the different staging systems used to define entry criteria (AJCC system in the Lin study vs. Ho system in the Chan et al study⁹¹), different histologic makeup of the tumors (73.2% differentiated nonkeratinizing carcinoma in the Lin study vs. 93.7% undifferentiated carcinoma in the Chan study), or different chemotherapy and/or radiotherapy schedules. The role for concurrent chemotherapy in the management of nasopharyngeal carcinoma has been further supported by two other trials comparing concurrent and adjuvant chemotherapy with irradiation alone^92,96 (see Table 32-3).

Through the collective experience with cisplatin-based chemotherapy in these as well as other cancer trials, it became apparent that a weekly cisplatin regimen such as that used in the Chan study is better tolerated than the intergroup regimen. Ninety-five percent of the patients assigned to the concurrent chemoradiation arm with weekly cisplatin were compliant with the regimen, whereas only 65% of the patients in the combined treatment arm received all three cycles.

Many additional trials have been performed exploring combination chemotherapy and radiotherapy for nasopharyngeal carcinoma.59,90,95,99 So far, no benefit was shown to neoadjuvant or adjuvant chemotherapy in any of the randomized trials. Likewise, a recent meta-analysis using combined individual data from eight randomized trials enrolling 1753 patients showed no OS benefit with either neoadjuvant or adjuvant chemotherapy, although a meaningful treatment effect was seen for concurrent chemoradiation.¹⁰⁰ However, the question remains whether addition of induction and/or adjuvant chemotherapy to concurrent chemotherapy may provide an additional advantage. Two ongoing trials are aimed at answering that question. A trial by the Nasopharyngeal Cancer Study Group of Hong Kong is designed to compare induction versus adjuvant chemotherapy with cisplatin and 5-FU in combination with concurrent chemoradiation (NCT00379262). A multicenter phase III trial opened by the Taiwan National Health Research Institutes is designed to compare a cisplatin-based multiagent induction regimen (mitomycin, epirubicin, cisplatin, 5-FU, and leucovorin) followed by concurrent chemoradiation with chemoradiation alone (NCT00201396). Both studies are expected to be completed by 2013.

In addition to evaluating the best sequencing for chemotherapy and irradiation, investigators are also examining the role for other chemotherapeutic agents in treatment of nasopharyngeal carcinoma. A recent study explored the efficacy of weekly carboplatin in the setting of concurrent chemoradiation for patients with locally advanced disease.¹⁰¹ At a median time of 26 months, OS and DFS were similar in the cisplatin and carboplatin groups, with less toxicity noted in the latter group. Therefore treatment with carboplatin is reasonable when use of cisplatin is precluded by patient comorbidities.

Setting the precedent, several landmark randomized controlled phase III studies have demonstrated that taxanes in the neoadjuvant setting greatly improve outcomes in other head and neck carcinomas.^102,¹⁰³ Phase II studies of patients with locally advanced nasopharyngeal cancer have already shown favorable results with paclitaxel, both in the neoadjuvant setting followed by concurrent therapy with platins or as a part of a concurrent regimen.¹⁰⁴^–¹⁰⁶ To further address this question, a phase III trial was opened in 2009 by the international Groupe Oncologie Radiotherapie Tete et Cou (GORTEC), randomizing patients with nasopharyngeal carcinoma to receive induction chemotherapy with docetaxel, cisplatin, and 5-FU followed by concurrent chemoradiation with cisplatin or cisplatin-based concurrent chemoradiation alone (NCT00828386).

Molecular therapies are also under investigation. Although one trial showed a benefit to the use of cetuximab in other head and neck cancers,¹⁰⁷ it remains to be seen whether these data will be confirmed for squamous cell carcinomas of the head and neck in general¹⁰⁸ or for nasopharyngeal carcinoma in particular.

Altered versus Standard Fractionation with or without Chemotherapy

In the last several decades, alongside trials exploring the role for chemotherapy in the treatment of nasopharyngeal carcinoma, other studies aimed to evaluate altered fractionation regimens. Hyperfractionation, accelerated fractionation, and a combination thereof were explored, either alone or in combination with concurrent chemotherapy. As in the trials focusing on chemotherapy, these were based on promising data from other head and neck sites.

A large randomized trial by Sanchiz and colleagues ⁸⁷ addressing the role of twice-daily versus once-daily irradiation for head and neck cancer included tumors of the nasopharynx. In this study, 892 patients (852 evaluable patients) with advanced head and neck cancers (T3-4N0-3M0 by UICC staging), including 92 patients with nasopharyngeal cancer, were randomly assigned to once-daily irradiation (group A), twice-daily irradiation (group B), and once-daily irradiation with 5-FU (group C). No differences were seen in median duration of response or OS between groups B and C; however, when either group B or C was compared individually with group A, there were significant improvements in both measures.

An accelerated fractionation regimen using six 2-Gy fractions per week to a total dose of 66 to 68 Gy was compared with a conventional fractionation regimen to the same total dose in a large randomized Danish study of 1476 patients with head and neck cancer (435 patients with pharyngeal tumors, including nasopharyngeal tumors).¹⁰⁹ In this trial, accelerated fractionation resulted in significantly improved locoregional control and disease-specific survival (DSS) but not OS.

The same fractionation scheme, either with or without concurrent chemotherapy, was evaluated in a trial from the Hong Kong Nasopharyngeal Study Group of 189 patients with advanced nasopharyngeal cancer.⁹⁷ Patients were randomly assigned to one of four arms: (1) conventional fractionation alone, (2) accelerated fractionation alone, (3) conventional fractionation with concurrent cisplatin and 5-FU, and (4) accelerated fractionation with concurrent cisplatin and 5-FU. At 2.9 years of follow-up, accelerated fractionation with concurrent chemotherapy (arm 4) achieved the most substantial tumor control, with PFS of 97% versus 70% (arm 1), 63% (arm 2), and 74% (arm 3), without an improvement in OS. A significant increase in acute and late toxicity in the combined chemotherapy–accelerated fractionation arm was also noted.

Based on these and other experiences with altered fractionation schemes in nasopharyngeal carcinoma, as well as other head and neck cancers, it became apparent that both the total dose and overall treatment time were important determinants of outcome.¹¹⁰ With the advent of IMRT, dose escalation (delivery of adequate doses to the gross and subclinical disease) has become possible without concomitant changes in the toxicity profile, significantly improving the therapeutic ratio of concurrent chemoradiation and rendering the above fractionation schemes obsolete.

Intensity-Modulated Radiotherapy with or without Chemotherapy

The proximity of numerous critical structures within the nasopharynx often makes it challenging to provide an adequate dose to an irregularly shaped tumor while still keeping the dose to surrounding tissues within acceptable limits. Therefore the capacity of IMRT to provide highly conformal and precise coverage with sharp dose gradients has been widely embraced in the head and neck cancer community. In addition, the ability to assign different doses to a desired spatial distribution when using IMRT has allowed for simultaneous delivery of different fractionation schemes to different portions of the target. Multiple institutions both in the United States and the endemic regions have reported improved clinical outcomes since adoption of IMRT techniques for treatment of nasopharyngeal carcinoma (Table 32-4).

TABLE 32-4 Nasopharyngeal Carcinoma: Results of IMRT Alone or IMRT Plus Chemotherapy*

University of California San Francisco (UCSF) researchers were the first to report their experience with IMRT in 67 patients with nasopharyngeal cancer, including 45 patients with stage III to IV disease, treated between 1995 and 2000.¹¹¹ Patients were prescribed 59.4 Gy in 1.8-Gy fractions to the planning target volume (PTV) encompassing microscopic and gross disease and typically received 65 to 69.96 Gy to the gross disease and involved lymph nodes in 2.12- to 2.25-Gy fractions. Twenty-six patients also had an intracavitary brachytherapy boost to a total dose of 5 to 7 Gy at the discretion of the treating physician. Patients with advanced disease received concurrent and adjuvant chemotherapy. The reported 4-year estimates of local PFS, locoregional PFS, and OS were 97%, 98%, and 88%, respectively.¹¹¹ Long-term xerostomia was minimal at 2 years, with 66% of the patients having no xerostomia and no grade 2 or higher late xerostomia.

Locoregional control rates above 90% have since been noted in other single-institution prospective ¹¹²^–¹¹⁴ and retrospective¹¹⁵ series, as well as in a recently published multi-institutional Radiation Therapy Oncology Group (RTOG) 0225 study.¹¹⁶ Similar to the UCSF regimen, the RTOG 0225 trial used a simultaneous integration of the boost, with 2.12 Gy per fraction delivered to the gross disease with a minimal 5-mm margin (PTV₇₀), while 1.8 Gy was delivered to the subclinical PTV (PTV_59.4). Therefore, in 33 fractions, the PTV₇₀ received a total dose of 69.96 Gy and the PTV_59.4 received 59.4 Gy. A total dose of 50.4 Gy in 28 fractions was delivered to the supraclavicular nodes. After a median follow-up of 2.6 years, the estimated 2-year local PFS, locoregional PFS, and OS rates were 92.6%, 89.3%, and 80.2%, respectively. Importantly, while providing similar or better locoregional tumor control in comparison with conventional irradiation techniques, IMRT has resulted in significantly lower rates of late complications. Only nine patients (13.5%) had grade 2 xerostomia at 1 year from the start of IMRT, two patients had grade 3 xerostomia, and none had grade 4 xerostomia. The improved toxicity profile of IMRT compared with conformal radiotherapy in the treatment of nasopharyngeal carcinoma has been since confirmed in two prospective randomized trials.^117,118

Locally Recurrent or Persistent Disease

In contrast to cancers of other head and neck subsites, recurrent nasopharyngeal carcinoma can undergo successful salvage in a substantial number of patients. Therefore aggressive management is usually advocated, and a number of different approaches, including surgery, brachytherapy, and external beam irradiation (EBRT), can be successfully used. Chemotherapy is generally used in combination with local treatment for all patients with advanced disease.

Surgical Management

Surgical resection is the preferred treatment for persistent or recurrent lymph node involvement in the neck. However, surgical management of recurrent nasopharyngeal cancer at the primary site is hampered by the same factors that make surgery at this site difficult as a primary treatment, such as difficulties in obtaining adequate exposure and obtaining adequate surgical margins.80,119,120,121 Salvage nasopharyngectomy has been achieved by several different approaches, the most common of which are the transpalatal, transmandibular, and transmaxillary approaches. Recently, minimally invasive techniques, such as a transnasal approach and an endoscopic approach, have also been employed successfully in a select group of patients. The choice of surgical approach is dictated by the tumor size and its location. Generally, surgical treatment of recurrent nasopharyngeal carcinoma remains controversial, but it may offer a reasonable disease-free period in certain well-selected cases. Several recent surgical series reported locoregional control and OS of 40% to 72% and 30% to 54%, respectively.^{122–124,125}

Reirradiation

Reirradiation remains the cornerstone of treatment for recurrent nasopharyngeal cancer within the nasopharynx. Early experiences with reirradiation were often associated with a high risk of late effects, including trismus, stricture, dysphagia requiring gastrostomy, hearing loss, unilateral blindness, cranial neuropathy, temporal lobe necrosis, and endocrinopathies. However, the evolution of radiotherapy techniques that allow for more precise dose delivery, and imaging techniques that allow for earlier detection and more precise tumor delineation during treatment planning, have significantly improved the therapeutic ratio.

In general, earlier stage at recurrence and longer time to recurrence are associated with better outcomes.126,127,128 Higher doses (≥60 Gy) are more effective than lower doses¹²⁸ and can be accomplished with EBRT alone or in combination with radiosurgery, or a brachytherapy boost (intracavitary or interstitial treatment).^{126,128–130,131} IMRT is particularly well suited to this task, and reirradiation results with this method are promising.

At the Memorial Sloan-Kettering Cancer Center (MSKCC), 29 patients with recurrent nasopharyngeal carcinoma were treated with reirradiation between 1996 and 2008, in the era of modern techniques.¹³² Sixteen patients received EBRT alone to an average dose of 59.4 Gy and 13 received EBRT to 45 Gy followed by intracavitary brachytherapy to 20 Gy. Twenty-four patients, 14 in the EBRT alone group and 10 in the combined-modality group, underwent IMRT. Choice of radiotherapy technique was at the discretion of the treating physician but was often dictated by the size and location of the recurrence. Thus the combined-radiotherapy group was primarily composed of patients with rT1 or T2 disease (85%), while the EBRT group included patients with more advanced disease (31%, rT1 or T2; 69%, rT3 or T4); 93% of the patients received chemotherapy. Five-year local control and OS were 52% and 60%, respectively. Although no significant differences were seen in local control or survival rates between the combined-radiotherapy and EBRT groups, the combined-radiotherapy approach was associated with fewer late complications (8% rate of grade 3 or greater late effects compared with 73% rate in patients receiving EBRT alone).

Stereotactic radiosurgery (SRS), or stereotactic body radiotherapy (SBRT), treatment of recurrent head and neck lesions is another promising approach because the technique allows for rapid fall-off of the radiation dose outside of the tumor volume and nearby critical structures that surround this anatomic area. SRS has been used to treat recurrent nasopharyngeal lesions after prior irradiation, either alone or in combination with EBRT.¹³³^–¹³⁵ When used as the only local treatment modality, it is generally reserved for smaller lesions, and doses of 11 to 35 Gy are used.^133,¹³⁵^–¹³⁷ Recent reports indicate that fractionated SBRT in the reirradiation setting can provide satisfactory control with an improved late toxicity profile.¹³⁸

Concurrent Chemoradiation

Concurrent chemoradiation is being used increasingly for treatment of recurrent head and neck cancer after primary radiotherapy.¹³⁹^–¹⁴² In general, the late morbidity of these regimens is less severe than would be expected, given that reirradiation doses are comparable to the original radiotherapy doses.^141,¹⁴² These reirradiation trials have usually included few nasopharyngeal cases. Historically, reirradiation without chemotherapy, possibly including brachytherapy, has given better results for nasopharyngeal cases than for other head and neck sites.¹³⁹ However, given the superiority of chemoradiation over irradiation alone in the treatment of primary nasopharyngeal cancer, it is reasonable to extrapolate these results to recurrent disease. In fact, most of the recent salvage irradiation series have included cisplatin-based chemotherapy for advanced-stage disease (Table 32-5).

TABLE 32-5 Results of Select Locally Recurrent Nasopharyngeal Carcinoma Reirradiation Studies

Irradiation Techniques and Tolerance

Conventional Irradiation

Treatment Fields

The radiation fields are arranged to cover the possible pathways of spread that account for the previously noted clinical manifestations (Fig. 32-3A to D). Field setup is designed to provide appropriate radiation doses to gross disease and areas at risk for microscopic extension, and this remains the case whether treatment is based on classically defined beam arrangements or on modern techniques such as IMRT. The Intergroup study 0099¹ (see Table 32-3) provides a structure for general irradiation guidelines with opposed lateral fields. The borders of this field arrangement are described below but can be modified on the basis of individual tumor characteristics as discussed in the subsequent section:

1 Superiorly: The field margin should be at least 2 cm beyond tumor that is visible on the head CT scan and should include the base of the skull and sphenoid sinus.

2 Posteriorly: The field should allow 2 cm of margin beyond the mastoid process. The posterior margin may extend further to allow at least a 1.5-cm margin on enlarged nodes.

3 Anteriorly: The field should include the posterior third of the maxillary sinus, posterior ethmoid air cells, and nasal cavity. This anterior border can be modified to accommodate adequate margins (2 cm) for tumors with anterior extension while excluding as much of the oral cavity as possible.

4 Inferiorly: The lower field margin is at the thyroid notch (superior to the arytenoids) to allow sparing of the larynx by a central block on the anterior lower neck field, which is matched to these lateral fields.

Figure 32-3 A to C, An example of a conventional field arrangement for stage T2 nasopharyngeal cancer. Pituitary shielding technique was used. The initial fields were opposed lateral fields (A), followed by oblique fields (B and C), with an electron field (not shown) to treat the contralateral neck. D, Base of skull included in the treatment of a stage T3 nasopharyngeal cancer. H, hyoid bone; T, thyroid.

A to D, Courtesy the Mayo Foundation, Rochester, Minnesota.

The lower neck is generally treated with an anterior field with a tapered central larynx block that extends inferiorly to the cricoid. This lower neck field is matched to the lower edge of the lateral fields and, therefore, the larynx block also prevents the spinal cord from receiving an excessive dose due to possible overlap of the lateral fields and anterior field. The lower neck field should extend inferiorly to the level of the lower border of the clavicle at the sternoclavicular joint. If the neck disease dictates that opposed anterior to posterior (AP-PA) fields are required for the lower neck treatment, it is sometimes convenient to set the isocenter at the level of the match between the upper lateral fields and the lower AP-PA fields. Therefore, the lower portion of the field can be blocked (beam-splitting technique) while treatment is given to the upper lateral fields; similarly, the upper half of the field can be blocked during the treatment of the lower AP-PA fields. The point of beam splitting can be feathered (moved superiorly or inferiorly 0.5 to 1 cm) every 10 to 15 Gy, to decrease the potential consequences of unintentional field overlap or gap. Feathering is important if the distribution of disease precludes the use of a larynx block or other means of safeguarding the spinal cord from unintentional field overlap. Boosting gross disease in the lower neck may require oblique fields to cover the target volumes and respect spinal cord tolerance.

Irradiation Dose

With the intergroup study 0099 ¹ as a general guideline for doses delivered when conventional techniques are used, the primary head and neck fields should be treated to a dose of 40 Gy (1.8 to 2 Gy per day) before placement of the spinal cord block (see Fig. 32-3). The treatment then proceeds to 50 Gy, with the newly blocked posterior neck portion of the field treated with electron fields. Electron energies should be determined by CT simulation and treatment planning to allow adequate treatment of the posterior neck with minimal dose to the spinal cord. After 50 Gy, the fields are reduced to include a margin of 1.5 to 2 cm around the primary tumor and any enlarged lymph nodes (>1 cm on CT scan of the neck). This reduced field arrangement is treated to a dose of 66 Gy; consideration can be given for a second reduction around gross disease to a final dose of 70 Gy.

Field Modifications and Boost Techniques

In addition to the opposed lateral field arrangement described above, several other techniques have been used to optimize target dose delivery. For example, patients may initially be treated with opposed lateral fields to the primary site and upper neck, and then continue with a boost to the primary site with either oblique fields, a single central field, or a three-field technique. The use of oblique and smaller opposed lateral fields is generally accepted and has been shown to have acceptable tolerance ³² (see Fig. 32-3). Interestingly, the use of three-dimensional conformal techniques to optimize the final or boost portion of treatment has not been shown to lead to a decrease in complications.^143,144

Another modification is to shield the pituitary and the anterior part of the hypothalamus in tumors with no evidence of erosion of the base of the skull or sphenoid and no extension of disease to the nasal fossa and ethmoid sinuses.¹⁴⁵ This approach allows for a decreased number of endocrine and temporal lobe complications without a compromise in tumor control in a well-selected patient population.

Brachytherapy

Brachytherapy can be used to deliver a boost to T1 to T2 lesions after EBRT or alone in the setting of reirradiation. Intracavitary * and interstitial¹⁵³^–¹⁵⁵ techniques have been used. Multiple applicators have been developed for intracavitary insertions. The Rotterdam Nasopharynx Applicator, developed in the Netherlands, is most commonly used today in conjunction with a remote-controlled high-dose-rate afterloading device¹⁵⁶ (Fig. 32-4A and B). The device, made out of pliable silicone and shaped to conform to the nasopharynx, is guided into position through the oral cavity. After it is immobilized and dummy sources are inserted, orthogonal radiographs are obtained to define the anatomy for treatment planning (Fig. 32-4C and D). Dose is prescribed to a fixed anatomic point. In cases where brachytherapy is used as a boost for EBRT, 5 to 6 Gy in two fractions given 5 to 6 hours apart is generally given. Given the excellent results with IMRT in treatment of nasopharyngeal carcinoma, however, many institutions employ brachytherapy almost exclusively for reirradiation.

Figure 32-4 Use of brachytherapy to treat the nasopharynx. A and B, A patient with the Rotterdam Nasopharynx Applicator (RNA) in place. The afterloading catheters are connected to a remote-controlled high-dose-rate afterloading device (microSelectron HDR, Nucletron-Oldeft Nucletron) (A). Close-up view of the RNA secured with a plastic clamp placed over the external portion of the applicator just outside the nose (B). Lateral (C) and anterior-posterior (D) localization radiographs with the dummy sources inserted.

A to D, Reprinted from Lee N, Fu KK: Cancer of the nasopharynx. In Leibel SA, Phillips TL (eds): Leibel and Phillips Textbook of Radiation Oncology, 2nd ed. Philadelphia, Elsevier, 2004. Figures 27-16A and B and 27-17A and B.

Permanent interstitial implantation of either iodine-125 (¹²⁵I) seeds or gold-198 (¹⁹⁸Au) grains should be considered for localized lesions and can be accomplished through several approaches, depending on the location of disease.^153,^154,¹⁵⁷ For lesions in the low to mid posterior wall, transoral or transnasal approaches have been used, whereas a transpalatal approach is favored for lesions in the high posterior wall.^153,¹⁵⁴ Alternatively, a split-palate approach can be used.¹⁵⁷

Intensity-Modulated Radiation Therapy

In the 1970s and 1980s it was becoming clear that advancing technology would make it possible to optimize radiotherapy by deriving the set of optimal incident stationary and moving beams from a desired dose distribution.^158,¹⁵⁹ These techniques deliver a set of beamlets of varying weights, with the weights chosen to optimally conform the resulting distribution of dose to a target.^160,¹⁶¹ More generally, the optimization can be performed to optimize the biologic or physical objective functions, based on minimizing complications and maximizing tumor control probabilities.¹⁶² The technique of optimizing beam delivery with inverse planning has become known as intensity-modulated radiation therapy (IMRT) for historical reasons, although typically the beam intensity is not modulated.

Target movement places a potential limitation on the applicability of IMRT. Movement during beam delivery could lead to the unintentional overlap of beamlets or possibly to untreated portions of the target volume. For this reason, IMRT is currently used in situations where motion is minimal or can be controlled for with gating techniques. The head and neck is an area where target motion can be controlled reasonably well with immobilization devices, and for this reason much of the early work on IMRT centered on head and neck cancer. IMRT is particularly applicable to the nasopharynx, in that several critical structures are in close proximity to the target and the incidence of morbidity with conventional radiotherapy techniques is high.

Several institutions have shown a potential dosimetric improvement for IMRT over conventional techniques and three-dimensional conformal techniques in the setting of nasopharyngeal cancer.144,163,164,165,166 As discussed previously, several single-institution series as well as a phase II RTOG trial have shown the efficacy of IMRT in the treatment of nasopharyngeal cancer.¹¹⁶

Simulation and Treatment Planning

CT simulation is necessary for IMRT treatment planning. The patient’s head, neck, and shoulders are immobilized with a thermoplastic mask. The head should be hyperextended to allow for maximal separation between the primary nasopharyngeal tumor and lymphatics.¹¹¹ A limited CT scan or orthogonal radiographs are obtained to allow for isocenter localization, after which planning CT images are acquired, with contiguous slices of 3 mm or less in the region of the tumor and of 5 mm elsewhere. Some treatment planning systems require equal spacing of slices throughout the entire scan. Alternatively, the CT images can be obtained first and then a treatment isocenter placed within the treatment volume.

IMRT treatment planning requires precise spatial information on the full extent of gross disease and thorough knowledge of the potential sanctuaries of microscopic disease along the lymphatic routes of tumor spread, because the therapeutic dose will only be delivered to the designated areas. Therefore, MRI, CT, and/or PET imaging is instrumental in supplementing the information from clinical examination when delineating the tumor volume. It is recommended that fusion of diagnostic MRI and/or PET scans to the planning CT is performed to accurately define the gross tumor volume (GTV) and surrounding critical structures. Several clinical target volumes (CTVs) are delineated, CTV_{gross disease} and CTV_{subclinical disease}, which will receive different doses per fraction and total doses (Table 32-6). This practice is known as simultaneous integrated boost or dose painting. In addition, in the absence of low neck lymphadenopathy, a third dose schedule may be administered to the low neck, either with separate anterior-posterior fields matched to the IMRT plan (split-field technique) or with extension of the IMRT plan (extended whole-field technique). If the extended whole-field technique is used, then a third CTV_{low-risk subclinical disease} may be outlined. However, in the presence of any pathologic low neck lymph node, the uninvolved low neck lymph nodes are incorporated into the CTV_{subclinical disease.}

TABLE 32-6 IMRT Target Delineation for Nasopharyngeal Carcinoma

Primary Tumor and Upper Neck
CTV_{70 or gross disease} Dose: 2.12 Gy/Fr to 69.96 Gy	GTV: Primary tumor All (+) lymph nodes + 5- to 10-mm margin (1 mm is acceptable near the brainstem or spinal cord, optic nerve, and chiasm)

CTV_{59.4 or subclinical disease}
Dose: 1.8 Gy/Fr to 59.4 Gy
Can consider 54 Gy in N0 disease

GTV and entire nasopharynx

Posterior ethmoid sinuses

Bilateral cervical lymph nodes:

Level Ib-IV for N+ disease

Omit level Ib for N0 disease

Retropharyngeal lymph nodes

Based on MSKCC guidelines. Expansion is 3 to 5 mm for planning tumor volume, except near brainstem or spinal cord, where 1 mm is acceptable.

CTV, clinical target volume; Fr, fraction(s); GTV, gross tumor volume; Gy, Gray.

The primary tumor in the nasopharynx and any cervical or retropharyngeal lymph nodes larger than 1 cm or with evidence of necrosis on imaging are considered GTV_{gross disease}. CTV_{gross disease} is composed of GTV_{gross disease} with a 5- to 10-mm margin to account for possible microscopic disease (see Table 32-6). A notable exception is near the brain stem or spinal cord, where normal tissue tolerances necessitate that the margin be reduced, sometimes to as little as 1 mm. The CTV_{subclinical disease} includes all areas at risk for microscopic spread, including the entire nasopharynx, clivus, base of skull, pterygoid fossae, parapharyngeal space, sphenoid sinus, posterior one-third to one-half of the nasal cavity, posterior ethmoid sinuses, posterior one-third of the maxillary sinuses, and the following lymph node regions: bilateral upper deep jugular (junctional, parapharyngeal), level Ib, level II, level III, level V, and retropharyngeal nodes. In a patient with no clinically or radiographically detectable cervical lymph node involvement, no coverage of the submandibular region is required. The delineation of cervical lymph node stations on the planning CT scans is based on the published image-based consensus guidelines.^{167,168,169,170} The PTV is generally a 3- to 5-mm expansion of all the CTVs to account for potential setup errors and patient motion. Once again, limiting the margin to 1 mm around the CTV is allowed near the brainstem and spinal cord.

When the split-field technique is used to cover the low-neck lymph node regions (levels IV and V), the match is typically placed just above the arytenoids, so that a tapered midline block in the anterior field can be used to shield the larynx.

Dose-limiting critical normal structures that should be outlined on the planning CT include the brainstem, spinal cord, optic nerves and chiasm, parotid glands, pituitary, temporomandibular (TM) joints, middle and inner ears, oral cavity, submandibular glands, thyroid gland, brachial plexus, retinas, lenses, cochlea, and skin.

Treatment Planning and Delivery

Several commercial treatment planning systems are available.¹⁷¹^–¹⁷⁴ Inverse planning is generally used to generate a treatment plan. However, in the absence of sufficient experience with inverse planning, forward planning can be used with comparable clinical outcomes.¹⁷⁵ At the MSKCC, IMRT is delivered with the dynamic multileaf collimator system using a sliding-window approach and megavoltage 6-MV or greater x rays are used. Megavoltage beams with energy above 6 MV should not be used for cervical lymph node irradiation.

Figure 32-5B to D provides an example of an IMRT plan for a patient with nasopharyngeal cancer treated in the RTOG 0225 trial. Seven coplanar nonopposed fields were used with a 0.5-cm multileaf collimator using the sliding-window technique, optimized with a gradient inverse planning engine. For comparison, a conventional plan for the same patient is also illustrated (Fig. 32-5A to C). The conventional ports used for the comparison are shown in Figure 32-6. Considerable sparing of normal tissues, in particular the parotid glands, is achieved with IMRT, and there is better coverage of the base of the skull than can be achieved with lateral fields. Figure 32-7 shows the dose-volume histograms for this case, again comparing conventional technique with IMRT.

Figure 32-5 Dose colorwash comparison of coverage on a representative patient with stage T2bN0M0 nasopharyngeal cancer treated in the RTOG 0225 intensity-modulated radiotherapy (IMRT) trial. (For comparison, Figure 32-6 illustrates the conventional lateral ports used.) Computed tomography cuts are shown at the level of the pterygoid plates for conventional technique (A) and IMRT (B). The level of the uvula is also shown for conventional radiotherapy (C) and IMRT (D). The conventional plan is a composite of comprehensive head and neck fields with a boost to the primary lesion. Sparing of the parotid glands is the primary benefit of IMRT in this case. For the IMRT plan, seven coplanar nonopposed fields were used with a 0.5-cm multileaf collimator using the sliding-window technique, optimized with a gradient inverse planning algorithm.

Figure 32-6 Digitally reconstructed radiograph of the conventional ports used in the comparison of intensity-modulated radiotherapy (IMRT) with conventional therapy illustrated (see Figs. 32-5 and 32-7). The posterior neck was blocked and treated with electrons at 40 Gy, and the boost treated the primary tumor with margins using 15-MV photons.

Figure 32-7 Dose-volume histograms (DVHs) for the patient in Figures 32-5 and 32-6, comparing intensity-modulated radiotherapy (IMRT) with conventional technique. Substantial gains can be made with IMRT in terms of target coverage and normal tissue sparing. A, Target coverage. Differences in the CTV_59.4 (clinical target volume to be covered by 59.4 Gy) target coverage reflect the relatively high dose given to subclinical risk areas per RTOG 0225, with much of this target volume being intentionally prescribed a dose of 50.4 Gy in the conventional plan. Also, a conventional anterior low-neck field was used in the IMRT plan, with a midline block over the spinal cord and larynx in that region, which blocked a portion of the CTV_59.4, a portion that was not included in the optimization but was included in the DVH. The DVHs for central nervous system (CNS) structures (B) and other normal structures (C) are also shown. Greater relative sparing of CNS structures by IMRT would be seen for higher-stage patients.

Figure 32-8 illustrates the dosimetric coverage that can be obtained with modern IMRT techniques in a sample reirradiation case. This is a situation in which the critical need to spare normal structures makes IMRT essential.

Figure 32-8 Dose colorwash illustrating radiotherapy coverage near the base of the skull for a case of recurrent nasopharyngeal cancer, in which intensity-modulated radiotherapy (IMRT) was used to minimize reirradiation risk to normal structures. Colorwash ranges between 20 Gy and maximum dose. The tolerance of the brainstem and temporal lobes places constraints on the dosimetry of the base of the skull (A). The mandible and spinal cord tolerance constrains the neck coverage (B). The IMRT technique used was similar to that in another case (see Figs. 32-5 and 32-6) but with nine coplanar nonopposed fields.

Radiotherapy Dose

Based on the early experience with IMRT (see Table 32-4) and the consensus opinion used in and subsequently validated by the RTOG 0225, the prescribed dose to the PTV_{gross disease} is 69.96 Gy in 2.12-Gy fractions, the dose to the PTV_{subclinical disease} is 59.4 Gy in 1.8-Gy fractions, and the dose to the PTV_{low-risk subclinical disease} (if the extended whole-field technique is used) is 54.12 Gy in 1.64-Gy fractions. If the split-field technique is used, the anterior-posterior fields receive 50.4 Gy in 1.8-Gy fractions or 50 Gy in 25 fractions.

The following normal tissue dose constraints are currently used at the MSKCC:

Critical
Brainstem	D_max = 54 Gy or 60 Gy to 1% of the volume
Optic nerves	D_max = 54 Gy or 60 Gy to 1% of the volume
Optic chiasm	D_max = 54 Gy or 60 Gy to 1% of the volume
Spinal cord	D_max = 45 Gy or 50 Gy to 1% of the volume
Temporal lobes	D_max = 60 Gy or 65 Gy to 1% of the volume
Mandible and TM joint	D_max = 70 Gy or 75 Gy to 1 mL of the volume
Noncritical
Parotid glands	D_mean = 26 Gy (in at least one gland)
Submandibular glands	As low as possible
Oral cavity	D_mean = 40 Gy
Larynx	D_mean = 40 to 45 Gy or D_max = 70 Gy or = 105% of prescription
Cochlea	D_max = 50 Gy
Retinas (globes)	D_max = 45 Gy
Lenses	D_max = 5 Gy
Brachial plexus	D_max = 65 Gy and D05 D_max = 70 Gy

The aim of the treatment plan design and optimization process is to deliver the prescribed dose to the target while limiting the dose to normal structures. We recommend strict adherence to the dose constraints for the brainstem, optic structures, and spinal cord. Limiting the dose to other critical normal structures listed above is encouraged but should not compromise target coverage. In cases where all treatment planning goals cannot be met, clinical judgment must be used to determine the best compromise. In addition to the listed constraints, no more than 5% of the nontarget tissue can receive more than 70 Gy. The dose constraints for noncritical normal structures listed above are only suggestions and should be given the lowest priority in the treatment planning process.

Normal Tissue Tolerance and Complications

Because of the anatomic location and requirement for large treatment fields and high doses, conventional treatment of nasopharyngeal carcinoma was associated with complication rates of 31% to 66%32,84,176 (see Table 32-1). Severe late complications included xerostomia, neck fibrosis, trismus, thyroid and hypothalamic-pituitary dysfunction, soft tissue and bone necrosis, hearing loss, cranial nerve dysfunction, and transverse radiation myelitis.

The incidence and types of complications of irradiation for nasopharyngeal cancer at the MDACC between 1954 and 1992 are listed in Table 32-7. The actuarial frequency of grade 3 to 5 complications was 16%, 19%, and 29% at 5, 10, and 20 years, respectively.³² However, the actuarial frequency of grade 3 to 5 complications at 10 years decreased steadily over three consecutive time periods included in the study: 1954 to 1971, 14%; 1972 to 1982, 10%; and 1983 to 1992, 5%, respectively. Of nine severe spinal cord injuries, eight occurred in the early time period (1954 to 1971). On univariate analysis, the frequency of moderate-to-severe complications was dependent on treatment technique. The actuarial incidence of moderate-to-severe complications at 10 years was 58% for the single central field technique compared with 12% and 25% for the oblique fields and opposed lateral fields techniques, respectively (p <.001). It should be noted that patients treated with oblique and opposed lateral fields were treated in more recent periods, and the decreased frequency of complications may also be related to the use of custom blocking and CT treatment planning.

TABLE 32-7 Serious Complications of Radiotherapy Alone for Nasopharyngeal Carcinoma in 378 Patients 31,32

The use of conformal techniques for the final boost portion of radiotherapy could potentially decrease the risk of complications, but this has not been shown to be the case. Conformal techniques are associated with a 25% incidence of grade 3 to 4 morbidity, including hearing loss, trismus, dysphagia, chronic sinusitis, and cranial neuropathy.¹⁴³ This result has led to an interest in delivering the full course of therapy with intensity modulation.^{111,143,144,177}

Xerostomia is the most common toxicity associated with conventional therapy and may lead to dental caries. Strict dental hygiene and prophylactic fluoride treatment should be instituted to guard against dental decay. The use of IMRT allows for improved preservation and faster recovery of parotid gland function.111,114,117,118

Functional hearing loss was reported in up to 8% of patients treated with conventional technique, and at least 30% of those evaluated with audiograms showed a high-frequency sensorineural hearing loss.176,178,179 This radiation sequel may also be negatively affected by concurrent treatment with ototoxic chemotherapy. Dose-dependent cochlear damage is the likely mechanism of radiation-induced ototoxicity and can be minimized by keeping the mean dose to the cochlea below 48 to 50 Gy.^178,¹⁷⁹ Lowering the dose to the external auditory canal also reduces the incidence and severity of acute external otitis. Excluding the mastoid air cells reduces the incidence and severity of chronic serous otitis media. Unfortunately, given the location of the primary tumor in the nasopharynx, chronic eustachian tube dysfunction is unavoidable.

When treating lesions with base-of-skull involvement, it is important to consider the potential for optic neuropathy. A University of Florida review of 106 patients receiving treatment encompassing the optic nerves reported that patients who received more than 60 Gy and daily fractions of more than 1.9 Gy had a 47% incidence of optic nerve injury, compared with 11% for patients treated with daily fractions of less than 1.9 Gy.¹⁸⁰

Clinically significant endocrine dysfunction has been described in 5% to 10% of patients.^32,¹⁷⁶ Given the availability of pharmacologic treatment, one should have a low threshold for evaluating long-term survivors for the presence of hypothalamic-pituitary or thyroid endocrinopathies. Routine evaluation of survivors with serum sodium, cortisol, thyroid-stimulating hormone (TSH), follicle-stimulating hormone, luteinizing hormone, and testosterone levels may be considered.

Treatment Algorithm, Conclusions, and Future Possibilities

A preferred treatment algorithm for patients with nasopharyngeal carcinoma is illustrated in Figure 32-9. With early-stage disease, irradiation alone with once-daily fractionation generally yields excellent local control results. For locally advanced disease (nonmetastatic stage ≥T2 or ≥N1 in the current AJCC staging system), concomitant chemotherapy and radiotherapy results are encouraging, and entry of patients into further trials testing combined-modality treatment is indicated. The advancement in radiotherapy optimization and the integration of functional imaging with radiotherapy planning promise improvement in both tumor control and decreased morbidity of treatment.