Primary Site and Regional Lymphatics

Larynx

The larynx is divided into three regions: the supraglottis, glottis, and subglottis. The supraglottic larynx lies above the level where the mucosa of the upper surface of the true vocal cords turns upward to form the lateral wall of the ventricle. It consists of the false vocal cords, arytenoids, aryepiglottic folds, and infrahyoid and suprahyoid epiglottis. The glottic region by definition includes the true vocal cords and extends 0.5 cm inferiorly, and the subglottic region extends from there to the superior aspect of the trachea (Fig. 33-1).

Figure 33-1 Anatomy of the larynx and its subdivisions.

Primary glottic cancers are about three times more common than supraglottic carcinomas, whereas subglottic carcinomas are extremely rare (about 2% of larynx cancers). At diagnosis, nearly two thirds of patients with laryngeal cancer have their disease confined to the laryngeal structures and less than 10% present with distant metastases. The patterns of spread, presentations, and management of cancers arising from these three subsites are very different.

Glottis

Most glottic lesions arise on the free margin and upper surface of the anterior two thirds of the true cord. These tumors tend to grow slowly and remain confined to the mucosa of the true cords. Eventually they will spread superiorly or inferiorly onto the mucosa of the supraglottic or subglottic structures. Some tumors may grow outward, in an exophytic fashion, leading to voice changes and subsequently respiratory obstruction. Invasion into the intrinsic musculature and joints of the larynx decreases vocal cord mobility and ultimately causes vocal cord fixation. Noteworthy is that tumor bulk alone can cause impaired mobility. Less commonly, infiltration of the recurrent laryngeal nerve may also cause impairment of mobility.

Infiltrative tumors will breach into the paraglottic and pre-epiglottic spaces. These spaces lie between the external frame of the larynx (thyroid-cricoid cartilages and hyoid bone) and the internal components of the larynx (epiglottis, muscles, and ligaments) and are filled with fat and hence offer little resistance to infiltration. From these spaces, infiltrative tumors may break through ligaments above or below the thyroid cartilage into the neck or may invade the cartilage directly.

The mucosa of the true vocal cords has a sparse lymphatic supply. Thus, glottic carcinomas have a low propensity for lymphatic spread. The incidence of lymphadenopathy at diagnosis is approximately 5% for T1 and T2 lesions and approximately 20% for T3 and T4 tumors.³⁸ The frequency of occult nodal involvement is also low. Byers and colleagues³⁹ found microscopic nodal involvement in 9 of 57 patients who underwent elective nodal dissection for T3 or T4 vocal cord lesions; most frequently involved nodes were the upper jugular (level II), midjugular (level III), and paratracheal groups (level VI).

Supraglottis

There are some variations of growth patterns within subsites of the supraglottic larynx. Suprahyoid epiglottic tumors sometimes grow in an exophytic pattern off the tip of the epiglottis, causing no symptoms until they become quite large. At other times they spread inferiorly, infiltrating and eroding the epiglottis. These lesions can eventually autoamputate the tip of the epiglottis. Complete regression of these tumors after RT may leave only a small epiglottic stump. Such an anatomic aberration increases the risk for aspiration.

Infrahyoid epiglottic lesions tend to spread anteriorly or circumferentially They can invade directly through the base of the epiglottis into the pre-epiglottic space and spread onto the lingual epiglottis and subsequently invade the vallecula, pharyngoepiglottic folds, and tongue base. At other times they spread onto the aryepiglottic folds, usually spreading from their midline position in both directions to involve both folds, producing a horseshoe-like appearance. They then spread either inferiorly onto the false vocal cords or over the aryepiglottic folds onto the piriform sinuses.

Lesions originating on the aryepiglottic folds can spread in all directions, that is, anteromedially onto the epiglottis, posteriorly to the arytenoids, inferiorly to the false cords, or laterally into the piriform sinus. Similar to lesions of the false vocal cords, they have variable growth patterns but often are infiltrative with easy access to the paraglottic space. They can frequently involve the cricoarytenoid joints or the musculature of the larynx, impairing mobility.

The primary difference between supraglottic cancers and true glottic cancers is the likelihood of developing cervical nodal metastases. At diagnosis, 55% of patients with supraglottic cancers have clinically involved lymph nodes. Lymphatic vessels in the supraglottic larynx collect in channels that pass through the piriform sinuses to drain to nodes along the jugular chain, particularly the upper (level II) and midjugular (level III) lymph nodes. Lee and associates⁴⁰ reported on the data of a subgroup of patients with intermediate-stage disease who underwent supraglottic laryngectomy with neck dissections. One third of patients had palpable nodes on presentation, and nearly an additional third of patients had pathologic nodal involvement.

Subglottis

Subglottic tumors spread inferiorly to the trachea, extend through the cricothyroid membrane into the neck, or directly invade the cricoid cartilage. Superiorly, they almost always involve the undersurface of the true glottis; and from a staging perspective, it is sometimes difficult to differentiate between a tumor truly originating from the subglottis and a glottic tumor with significant subglottic extension.

Hypopharynx

The hypopharynx is the inferior portion of the three divisions of the pharynx (Fig. 33-2). It extends from the hyoid bone superiorly to the cricoid inferiorly. Valleculae, pharyngoepiglottic fold, and lateral projections of aryepiglottic folds are considered the superior border separating the hypopharynx from oropharynx. Inferiorly, the hypopharynx ends at the cervical esophageal inlet. The hypopharynx is subdivided into three components: the pharyngeal walls, the piriform sinus, and the postcricoid pharynx. The hypopharyngeal walls are a continuation of the lateral and posterior oropharyngeal walls. The pair of piriform sinuses is created by the invagination of the larynx into the hypopharynx. They are conical (more truly pear shaped, hence the derivation of its name). Each sinus (or recess) consists of three walls. The medial wall is essentially the lateral aspect of the larynx; superiorly, it becomes the aryepiglottic fold. The lateral wall is a continuation of the lateral wall of the oropharynx. Anteriorly, the medial and lateral walls converge to form the narrow anterior wall. Superior is the base or vestibule formed by the rim of the three walls. Inferiorly, the three walls merge to form the apex.

Figure 33-2 Anatomic relationship of the hypopharynx to the pharyngeal axis.

The postcricoid pharynx begins superiorly at the level of the arytenoids and ends inferiorly at the esophagus. The posterior larynx makes its anterior border, and posteriorly it again is a continuation of the posterior pharyngeal wall.

Hypopharyngeal cancers also commonly present as nodal metastases. Lindberg⁴¹ reported a 75% incidence of nodal metastases in patients presenting to the M.D. Anderson Cancer Center (MDACC) with hypopharyngeal tumors. Level II and III nodes were most frequently involved, and bilateral lymphadenopathy was seen in 15% of patients. These tumors also have access to deep jugular and retropharyngeal lymph nodes.

Piriform Sinus Tumors

Piriform sinus tumors account for 70% of cancers originating in the hypopharynx. Tumors originating from the medial wall of the piriform sinus behave similarly to supraglottic tumors arising from the aryepiglottic folds, and it is often difficult to know the true origin of some of these lesions. Posteriorly, there is little to impede tumor extension into the postcricoid space or crossing from the ipsilateral arytenoid to the contralateral arytenoid. Lateral wall tumors also have few barriers to growth. They can extend medially along the mucosa to involve the posterior hypopharyngeal wall or anteromedially to involve the other two walls of the piriform sinus. Inferiorly, they have quick access to the apex and usually extend deep to the sinus submucosally to involve other adjacent structures, including the thyroid and cricoid cartilages, or directly into the thyroid gland. It is not uncommon to see advanced tumors from this location that have direct soft tissue extension into the neck. These tumors can also extend into the cervical esophagus, primarily through submucosal routes, making a true definition of disease extent extremely difficult. They can also behave similarly to more classic esophageal tumors with extensive spread along lymphatic spaces and with skip lesions.

Hypopharyngeal Wall Tumors

Tumors originating from the hypopharyngeal walls are less common. Because early presentations are infrequent, it is unusual to see a lesion confined to and clearly originating from the lateral or posterior hypopharyngeal wall. Hypopharyngeal wall tumors frequently spread along the walls to involve either the posterior or lateral oropharyngeal walls. Posterior wall tumors can also invade the prevertebral tissues or invade even more deeply to the bones of the cervical spine.

Postcricoid Tumors

Postcricoid tumors are the least common hypopharyngeal tumors. They almost always present in advanced stages. They can extend laterally into one or both piriform sinuses or directly invade laryngeal structures.

Distant Metastases

The incidence of distant metastases from cancers of the head and neck and specifically laryngeal and hypopharyngeal cancers is low and is generally thought to be less than 10%. An often-referenced study by Crile in 1906⁴² reported an incidence of 1% distant metastases in 4500 patients with epidermoid cancers of the head and neck. In subsets of patients with head and neck squamous cell cancers, patients with glottic tumors usually have the lowest rates of distant metastases whereas patients with hypopharyngeal carcinomas often have the highest rates.

Merino and colleagues⁴³ analyzed the incidence of distant disease in over 5000 patients with head and neck squamous cell carcinomas treated from 1948 through 1967. Among patients with control of disease above the clavicle, the incidence of distant failure was 1%, 13%, and 23% for patients with carcinomas of the true vocal cords, supraglottic larynx, and hypopharynx, respectively. Similarly, Marks and associates⁴⁴ found that 23% of patients with piriform sinus cancers developed distant metastases, with a higher rate among patients who underwent total laryngectomies as part of their therapy. The incidence of metastases is associated with stage of disease, because the clinical incidence of metastases is approximately 20% in patients with stage IV disease.

The sites of presentation of distant metastases are similar for cancers of the larynx and hypopharynx.⁴³ The lungs are the most common site and are the first site of presentation in nearly 60% of patients. Bones are the next most common site, because 20% of patients with distant disease develop osseous metastases. Liver metastases are common in autopsy studies, but clinical liver metastases develop in only 10% of patients with hematogenous spread of disease from the larynx and hypopharynx.⁴⁵ Spread to mediastinal lymph nodes, the brain, or other organs is very uncommon.

Laryngeal Cancer

Clinical evaluation of laryngeal cancer includes indirect laryngoscopy with a mirror that is frequently supplemented by fiberoptic endoscopy (Table 33-1). Similar to other tumors in the head and neck, the examiner assesses the tumor size, morphology, infiltration (defect, distortion) of adjacent structures, and vocal cord mobility. It is important to palpate the base of the tongue to determine direct invasion from the supraglottic larynx and to look for indirect signs of pre-epiglottic space invasion such as fullness of the vallecula or ulceration of the infrahyoid epiglottis. A direct laryngoscopy is often the final step in evaluation. It is needed to further outline the disease extent and, in particular, to obtain biopsy specimens for tissue diagnosis.

TABLE 33-1 Evaluation of Patients with Suspected Laryngeal or Hypopharyngeal Cancer

* In patients with more advanced disease, these tests may augment the workup for metastatic disease; PET and barium studies may help better define the inferior extent of advanced hypopharyngeal cancers.

Examination of the neck is important to detect lymphadenopathy or direct tumor extension manifesting as tenderness of the thyroid cartilage or the presence of a subcutaneous mass. Firm fullness over the thyroid notch is suggestive for pre-epiglottic space involvement.

Radiologic studies are indicated when there is the suggestion of deep infiltration. CT (Fig. 33-3) or MRI is useful for the assessment of the pre-epiglottic space, tongue base, paraglottic region, and subglottis (which is sometimes difficult to evaluate from above with the mirror or fiberoptic scope). These images may help differentiate direct primary tumor extension to the soft tissue of the neck from nodal involvement. Anterior commissure lesions may have subtle thyroid cartilage invasion only detectable by CT.

Figure 33-3 CT of laryngeal cancers. A, Cancer of the supraglottis invades the pre-epiglottic space (left). B, Cancer of the epiglottis with direct extension into the base of the tongue. C, Cancer of the true vocal cord with subglottic extension (left). D, T4 laryngeal cancer, with destruction of the cartilage and invasion of the anterior strap muscles.

Hypopharyngeal Cancer

Evaluation of hypopharyngeal tumors is similar to that of larynx cancers. Because these lesions are typically at an advanced stage, it is not uncommon to palpate disease in the neck by direct extension. The thyroid click may be lost owing to anterior displacement of the larynx by a posteriorly located (particularly postcricoid) lesion. Phonation may provide better visualization of the piriform sinus on indirect mirror and fiberoptic examination. If unsuccessful, a Valsalva maneuver may open the sinus. Deep infiltrative lesions in the apex may be hard to see but are suspected by either pooling of secretions or arytenoid edema. Assessment of laryngeal mobility is important in medial wall lesions because they can invade directly into the laryngeal framework. In addition to CT (Fig. 33-4), combined ¹⁸F-fluorodeoxyglucose-labeled positron emission tomography and CT (FDG-PET/CT) may be helpful in defining tumor extent, particularly the inferior border where subtle changes in the inferior hypopharynx and superior cervical esophagus may be hard to identify solely on anatomic images.

Figure 33-4 CT of hypopharyngeal cancers. A, A small (<1-cm) lesion arises off the posterior hypopharyngeal wall (T1) with large (N3) neck metastases. The primary tumor was managed with irradiation only, and the neck was treated with combined irradiation and surgery. B, An exophytic lesion was confined to the walls of the left piriform sinus (T2) and was treated with hyperfractionated RT. C, An infiltrative lesion off the medial wall of the piriform sinus invaded and fixed the hemilarynx (T3). This patient was treated within the framework of a study of induction chemotherapy and irradiation. D, An advanced lesion of the piriform sinus extended directly into the soft tissues of the neck. This lesion required a resection with a free jejunal autograft repair followed by postoperative RT.

Larynx

Tumors are staged using the American Joint Committee for Cancer (AJCC) TNM system²⁷ (Table 33-2). The system is currently in its seventh iteration, most recently revised in 2010. The system classifies/categorizes larynx cancer primary tumors (T) by sites of extension and cord mobility but not size. For glottic tumors, stage T1 is disease limited to the vocal cord(s). The AJCC staging system further divides T1 glottic tumors into T1a for tumors confined to one true vocal cord and T1b for tumors involving both vocal cords. In general, this subdivision is more useful for communication and may influence the treatment decision (i.e., limited surgery vs. RT) but does not have a strong impact on prognosis. T2 glottic tumors have either supraglottic or subglottic extension and/or impaired mobility. Although not defined by the AJCC, T2 glottic lesions can be subdivided into T2a for normal mobility and T2b for impaired mobility. This subdivision is not uniformly used, however, because there is no general agreement that treatment outcomes differ among the substages.

TABLE 33-2 American Joint Committee on Cancer Staging of Laryngeal and Hypopharyngeal Cancers

From Edge SB, Byrd DR, Compton C, et al, editors: AJCC Cancer Staging Handbook, ed 7. New York, 2010, Springer, pp 33-57. Used with permission of the American Joint Committee on Cancer.

Two significant revisions were made to the sixth edition of the staging system.⁶¹ Historically, the definition of T3 glottic disease has been fixation of the vocal cord. The sixth edition expanded on this definition and included paraglottic space invasion or minor thyroid cartilage invasion. Previous editions had defined T4 disease as invasion through the thyroid cartilage, but a common misinterpretation was to include any thyroid cartilage invasion into the T4 category. The new definition was made to clarify this point. The significance of this distinction is in the era of larynx preservation (see later); many still think that the presentation of invasion through the cartilage into the extralaryngeal tissues precludes a nonsurgical approach, but patients presenting with minimal invasion into the inner cortex may have disease suitable for larynx preservation.

The second modification is the subdivision of T4 disease. For head and neck cancers in general, the sixth edition had subdivided stage T4 into resectable (T4a) and unresectable (T4b) categories. Thus glottic cancer T4a disease has been defined as tumor that invades through the thyroid cartilage and/or invades tissues beyond the larynx (e.g., trachea, soft tissues of neck including deep extrinsic muscle of the tongue, strap muscles, thyroid, or esophagus), and T4b is tumor invading prevertebral space, encasing carotid artery, or invading mediastinal structures. Subsequently, the only change made in the seventh edition was descriptive. Stage T4a now describes moderately advanced local disease, and T4b describes very advanced local disease.

The staging of supraglottic cancers requires the division of the supraglottis into five subsites: the suprahyoid epiglottis, infrahyoid epiglottis, aryepiglottic folds, arytenoids, and ventricular bands (false cords). T1 disease is disease confined to one subsite (with normal vocal cord mobility). T2 disease invades more than one adjacent subsite or region outside the supraglottis (mucosa of base of tongue, vallecula, or medial wall of the piriform sinus). T3 disease is similar to glottic carcinoma, although it also includes pre-epiglottic space invasion or postcricoid involvement. Last, in the newest edition, the definition of T4 supraglottic cancer is the same as that for glottic cancer with subdivisions into moderately advanced and very advanced (T4a and T4b) categories.

Hypopharynx

T category for hypopharyngeal tumors is also based on sites of involvement and larynx motion (an indirect measurement of disease extension). The fifth edition of this staging classification had a significant modification to also include the size of the tumor in staging.⁶² This change addressed a major deficiency in prior systems, because the older systems tended to understage hypopharynx tumors relative to other head and neck mucosal cancers. The hypopharyngeal T category still does not reflect morphology, which remains an important criterion for selection of local therapies, because exophytic tumors in this site are usually selected more frequently than infiltrative lesions for organ-preserving therapies. The sixth edition made one additional modification. As described for laryngeal cancers, resectable T4a and unresectable T4b categories were defined. T4a hypopharyngeal disease can invade thyroid or cricoid cartilages, hyoid bone, thyroid gland esophagus, or central compartment, while T4b disease invades prevertebral fascia, encases carotid artery, or involves mediastinal structures. The newest edition now defines T4a and T4b as moderately advanced and very advanced.

Regional Classification/Categorizaton

The nodal (N) category is uniform for all head and neck cancer, including laryngeal and hypopharyngeal tumors. No changes were made in the seventh edition. The details of T and N categories for laryngeal (glottic and supraglottic) and hypopharyngeal cancers are summarized in Table 33-2.

T1 and T2 Glottic Tumors

The management of early glottic carcinomas is controversial and is often determined by the preferences and expertise of the attending physicians. Oncologists advocate either irradiation or voice-preserving partial laryngectomy. Large published series have for all practical purposes demonstrated equivalent control rates with surgery (excision, cordectomy, or hemilaryngectomy) and RT. There has not been a true comparative study between these two modalities, and it is highly unlikely that a prospective trial will be performed.

There are several difficulties in comparing the results of various treatment options. Nearly all series are retrospective single-institutional studies. It is thus difficult to ascertain if small differences in control rates between series are real. Mendenhall and colleagues⁶³ address this issue by defining the surgical procedure that would have been recommended in their irradiated patients. They found that the surgical treatment option would have required total laryngectomy, rather than partial voice-conserving procedures, in 10% of patients with T1 disease and 55% of those with T2 disease. However, the supracricoid laryngectomy, a procedure that removes an entire circumferential portion of the larynx and subsequently reconstitutes the larynx, has largely been developed during the past decade.⁶⁴ Many patients described by Mendenhall and colleagues in 1993 deemed suitable only for total laryngectomy may now be managed with this voice-preserving surgery. Olsen⁶⁵ states that it is very unusual for disease that does not fix the arytenoids to require total laryngectomy. Although this is possibly true from a technical standpoint, many larynx cancer patients are older, have long histories of tobacco exposure, and medically may not be suitable for partial laryngeal surgery other than excisions of small T1 lesions on the midcord.

In general, T1 tumors can be treated effectively with laser excision,⁶⁶ laryngofissure,⁶⁷ partial laryngectomy,^68,69 or irradiation. Table 33-3 summarizes the results of RT. Local control rates of RT range from 85% to 95%, and ultimate control rates after salvage surgery for recurrences are greater than 95%. Surgical salvage for the uncommon local relapse often requires a total laryngectomy, but an occasional patient can have a lesion amenable to a voice-conserving partial laryngectomy.^70–74

An important factor in selecting therapy for individual patients is the anticipated voice quality after therapy. Unfortunately, studies comparing voice quality between laser excision and RT have yielded conflicting conclusions. Hirano and associates⁷⁵ concluded in their analysis that hoarseness and incomplete glottic closure were more frequently seen in patients treated with laser. McGuirt and associates⁷⁶ had physicians, patients, and speech pathologists compare voice quality of patients who received RT or laser excision for early glottic cancer. They concluded that patients treated with limited excisions of tumors of the midcord had equivalent quality compared with those receiving RT. Using computerized-assisted voice analysis, Harrison and associates⁷⁷ demonstrated that most of their patients irradiated for early vocal cord cancer maintained excellent voices. Mittal and colleagues⁷⁸ reported a cost savings with radiation compared with hemilaryngectomy and thought this should be an important component in treatment decisions if other important criteria are equivalent. A recent meta-analysis performed by Higgins and associates⁷⁹ of over 7600 patients suggested a trend toward improved voice quality with RT as compared with transoral laser excision.

The outcome of RT for T2 cancers is, as expected, less favorable than that for T1 tumors. Local control rates range from 65% to 80%, and ultimate control rate is about 90% (Table 33-4). The wider range in the reported local control rates can, at least in part, be ascribed to a considerable heterogeneity of presentations within this stage, that is, varying from a superficial tumor spilling into the ventricle to a more bulky lesion transgressing the paraglottic space and impairing motion.

Oncologists have searched for prognostic determinants in this group of patients. Several reports showed that impairment in cord mobility adversely affects tumor control by RT (Table 33-5). Some physicians favor a hemilaryngectomy, if feasible, for T2b tumors.^68,69,80 However, Karim and colleagues⁸¹ did not find impaired cord mobility to result in a worse outcome and attributed this observation to prescribing higher radiation doses to patients with such a tumor feature. Similarly, a report from MDACC did not show a relationship between mobility and outcome but some patients were treated with altered fractionated schedules that were theorized to have a therapeutic advantage.⁸² An update of this report reinforced the lack of impact of mobility on prognosis, independent of fractionation.⁸³ This was contrary to the finding of Wang,⁸⁴ who found that twice-daily radiation improved local control of T2 tumors with normal mobility but not those with impaired mobility.

Several studies were also undertaken to assess the influence of impaired cord mobility on the success rate of salvage surgery for patients who developed recurrences. Schwaab and associates⁷⁴ noted difficulties in attempting salvage on a small cohort of patients presenting with T2b tumors (although, surprisingly, the initial control rate of T2b tumors in this series was better than that of T2a lesions). Wiggenraad and colleagues⁸⁵ reported ultimate control rates of 98% for T2a and 76% for T2b tumors (p <.05). Other groups, however, have reported equivalent ultimate local control rates for tumors with normal or impaired cord mobility.^74,83,86 Ang and Peters⁸⁷ suggested that the timeliness of diagnosis and the extent of surgery rather than the initial disease presentation might account for the difference in the salvage results.

Anterior commissure involvement was considered a potential prognostic factor because it may carry an increased risk for unsuspected thyroid cartilage invasion. However, this feature was not found to be a prognostic determinant in most recent series. Sessions and associates⁸⁸ found that anterior commissure⁸⁹ involvement was more frequently associated with larger tumors, and it was the extent of disease rather than commissure involvement per se that impacted on outcome. Le and associates⁹⁰ found, in a multivariate analysis, that patients with anterior commissure involvement had a poorer local control rate but concluded that these patients also had a larger tumor burden and sometimes unsuspected subglottic disease. Modern high-resolution imaging techniques can detect clinically unsuspected thyroid cartilage invasion and eliminate the clinical significance of anterior commissure involvement.

Glottic Carcinoma in Situ

Controversies of management even exist for carcinoma in situ of true vocal cords. Although not all lesions progress to invasive cancer, Hintz and colleagues⁹¹ reported that with a watchful waiting policy nearly two thirds of patients developed invasive cancer that was not always suitable to voice-preserving therapies. Ferlito and associates⁹² polled otolaryngologists and found practices of management ranging from postbiopsy observation, vocal cord stripping, cordectomy, open partial laryngectomy, to primary irradiation.

Primary RT is an effective treatment for carcinoma in situ of the larynx. However, conservative surgical approaches such as microexcision, laser ablation,^93,94 or vocal cord stripping⁹⁵ are considered sufficient for initial treatment. Repeated biopsies, strippings, or laser excisions may be counterproductive and lead to worsening of quality of voice. RT is usually reserved for recurrences or for diffuse lesions that are not suitable for limited surgery as the first therapy.

The local control rates reported in modern series of patients with carcinoma in situ treated with radiation ranges from 70% to 100% and are similar to rates described for invasive T1 disease (Table 33-6). It should be recognized that, owing to limited biopsy sampling, many radiation series may include patients with unrecognized invasive disease. For example, nearly one third of the patients in the series of Pene and Fletcher⁹⁶ diagnosed with carcinoma in situ had clinical T2 disease. Most patients developing recurrent laryngeal carcinoma in situ undergo salvage surgery, most often with total laryngectomy, although some series have reported salvage with voice-preserving surgery.^97,98 Although control rates after irradiation are similar to those reported for early invasive disease, recurrences in patients with carcinoma in situ take longer to manifest. Most series report median times to failure of greater than 2 years, but the majority of tumors recur within 5 years of treatment.^{91,96,98–100}

T1, T2, and Selected T3 Supraglottic Tumors

Similar to glottic cancer, the optimal management of supraglottic carcinomas is still being debated. Confounding the issue even more is the necessity to manage the neck lymphatics in addition to the primary tumor. As mentioned earlier, nearly one third of patients with early- to intermediate-stage supraglottic neoplasms present with palpable cervical lymphadenopathy41,101 and another third have subclinical nodal involvement detected by elective neck dissections.¹⁰²

Voice-preserving treatment options for early- and intermediate-stage tumors are supraglottic laryngectomy or primary RT.^{40,68,71,103–107} Local control rates in most surgical series range from 80% to 90%. Bocca and colleagues¹⁰⁵ from Milan reported a local control rate of 85% in one of the largest series of over 400 patients with predominantly T2 tumors. Of note is that postoperative RT is often given to many patients undergoing partial laryngectomy. Common indications are presence of multiple nodes, extracapsular spread of nodal disease, and close or positive surgical margins. For example, 50 of 60 (83%) patients who underwent supraglottic laryngectomy at MDACC from 1974 to 1987 had indications for postoperative RT.⁴⁰ The leading complication of supraglottic laryngectomy is aspiration, which can be morbid in heavy smokers (the majority of patients) with chronic pulmonary disease. Although preliminary data of endoscopic removal of early supraglottic lesions have been reported recently,^108,109 more experience is needed to determine its relative value.

Most surgical series selected middle-aged men who are medically fit for supraglottic laryngectomy. Hinerman and colleagues¹¹⁰ analyzed 274 patients who underwent primary RT at the University of Florida and retrospectively grouped these patients into three groups: (1) suitable for conservation surgery, (2) anatomically suitable but medically unsuitable, and (3) anatomically unsuitable. Using 1998 AJCC criteria, they reported that 45% of T1 patients, 36% of T2 patients, and 14% of T3 patients were suitable for laryngectomy. Only 14% of T1 patients were anatomically unsuitable, compared with 46% and 68% of T2 and T3 patients, respectively. Of note, results after RT for all their patients were dependent on T stage but were not dependent on suitability for conservation surgery. Similar to the situation in patients with glottic carcinomas, the development of the supracricoid laryngectomy would alter the balance of patients with T2 and T3 patients thought to be anatomically unsuitable for a conservative procedure. In a small series, Laccourreye and associates¹¹¹ reported on 19 patients treated with supracricoid laryngectomy. This cohort had disease in the infrahyoid epiglottis and was not suitable for more traditional conservative procedures.

In 1974, Bataini and colleagues¹¹² published an overall locoregional control rate of 76% for a relatively large series of patients with T1-2, N0-2 supraglottic carcinomas treated by RT alone. Subsequent series reported control rates of 84% to 100% for T1 tumors.^84,113,114 Table 33-7 summarizes the outcome of RT in T2 and favorable T3 lesions. Fletcher and Goepfert¹¹³ obtained a local control rate of 76% in this subset of patients by escalating the total dose of the conventional daily irradiation to 70 Gy. Application of an altered fractionation regimen during the past decade appears to have resulted in a trend toward further improvement in the local control.^84,102,114 In the Danish Head and Neck Cancer Group (DAHANCA) randomized fractionation trials, the benefit of accelerated treatment for head and neck cancer in general was most noticeable in the subgroup of patients with supraglottic cancers.¹¹⁵

The impact of clinically positive nodal disease on local control has been evaluated in several series. Wall and colleagues¹¹⁶ found the presence of lymphadenopathy was associated with a significantly lower local control rate. Similarly, Wang and associates¹¹⁷ describe a marked worsening of local control in node-positive patients. The 5-year actuarial local control rate obtained with accelerated fractionation was 86% in patients with no adenopathy as opposed to 46% in patients with N2-3 disease. In contrast, Freeman and colleagues¹¹⁸ did not find an association between local control and the presence of lymph nodes.

The potential impact of tumor volume, estimated by CT measurements, on the results of RT was addressed by Freeman and associates¹¹⁹ in patients with T3 supraglottic tumors. They found a control rate of 83% for tumors of less than 6 cm³ compared with 46% for tumors of greater than 6 cm³. This latter group may be more suitable for surgery, particularly if voice-conserving surgery is feasible. Kraas and associates¹²⁰ also studied tumor volume measurements, and although their threshold was 8 cm³, the concept of increasing tumor volume predicting worse outcome with RT was validated in a separate cohort.

Comparative studies have been undertaken in an attempt to delineate therapy recommendations for different subsets of patients. Fein and colleagues¹⁰⁷ analyzed the University of Florida experience and concluded that RT yielded a local control rate equivalent to that obtained by conservative surgery but was associated with a lower complication rate. Robbins and colleagues¹⁰⁴ from MDACC compared the outcome of patients who underwent supraglottic laryngectomy with that of patients who received RT for tumors judged suitable for partial laryngectomy. They found a better tumor control rate in the surgically treated group, but the incidence of morbidity due to chronic aspiration was rather high. This finding contributed to the development of the current treatment policy at this institution. Briefly, supraglottic laryngectomy, with postoperative RT when indicated, is recommended for medically fit patients with bulky, infiltrative T2 and T3 tumors (e.g., invasion into the false cords or infrahyoid epiglottis). RT is selected for patients with either smaller or exophytic lesions (T2 or favorable T3) or patients with bulky lesions who are medically unfit for voice-conserving surgery.

T1 and T2 Hypopharyngeal Tumors

Although uncommon, these tumors can be treated with RT. Because of the high propensity for neck metastases, it is not unusual to see patients with small-volume primary disease suitable for RT but with large neck disease that requires multimodality treatment. In this situation, the options are to irradiate the primary tumor and neck nodes to high doses and dissect the residual neck mass about 6 weeks after RT or to perform a neck dissection followed by definitive RT to the primary tumor and with adjunctive RT to the neck. The latter approach is frequently recommended to patients who need to undergo dental extractions, requiring a delay in initiation of irradiation.¹²¹

Some results with primary RT for early-stage hypopharynx lesions are shown in Table 33-8. Compared with historical control data, local control rates seem to have improved with the use of hyperfractionated regimens. For example, analysis of MDACC data revealed actuarial local control rates of 86% for hyperfractionation (35 patients) and 63% for conventional fractionation (23 patients).¹²² Similarly, for stage T2 piriform sinus primary tumors, Amdur and colleagues¹²³ reported local control in 89% of patients treated with twice-a-day irradiation compared with 73% for those treated only once per day (p = .09). Although it is believed that hyperfractionation has improved the results, it should be realized that refinements in RT technique, including customized Cerrobend portal shaping, more generous coverage of the posterior route of spread, and the use of 6-MV x rays with smaller penumbra, may all have contributed to the improvement of outcome. In addition, it is difficult to discern whether the rather subjective patient selection criteria have changed over the years and what impact this has on the results. Superficial and exophytic tumors, particularly in the vestibule, tend to be selected for primary RT, whereas infiltrative tumors of similar stage, especially in the apex, are chosen for multimodality treatment.

Verrucous Carcinoma

Verrucous carcinoma is generally a well-differentiated, slow-growing, wartlike lesion. This tumor has a reputation for its relative radioresistance and a tendency for converting to a highly anaplastic neoplasm after RT.³³ However, no firm data are available to support this belief. A 5-year control rate of 59% was achieved in a series of 43 patients treated with RT at the Princess Margaret Hospital.¹²⁴ The majority of these patients had early-stage disease, although in over half the disease was stage T2. The surgical salvage rate for radiation recurrences was high, and only 1 patient died of the disease. Therefore it is reasonable to treat patients with early verrucous carcinoma with radiation if the surgical option is a total laryngectomy. However, those with well-lateralized early tumors may be better suited for conservative surgery.

Sarcomatoid Carcinoma

Sarcomatoid carcinoma of the larynx has a similar reputation for radioresistance and for dedifferentiation after RT. In a review of 28 irradiated patients with T1-2 glottic sarcomatoid carcinoma, we found a 5-year actuarial control rate of 89%, which is similar to that obtained for ordinary squamous cell carcinoma.¹²⁵ Although the control rate for 7 patients with T2 tumor was only 57%, all patients with recurrences could undergo salvage surgery and only 1 patient died of the disease.

Neuroendocrine Carcinoma

Neuroendocrine carcinomas of the head and neck are extremely uncommon. They can be subgrouped into three categories. The most common is small cell undifferentiated carcinoma. Moderately differentiated and well-differentiated tumors have also been described and are sometimes referred to as atypical and typical carcinoids, respectively. The latter is extremely rare. The behavior of these three subtypes is similar to their more common counterpart in the lung.

In a report¹²⁶ describing the results of patients presenting with nonsinonasal neuroendocrine cancer, we found that the majority of patients (65%) presented with laryngeal or hypopharyngeal disease. RT was very effective for locoregional control, but most patients succumb to distant metastases. The addition of chemotherapy appeared to decrease the risk of distant failure and translated into improved survival. Our current policy is to recommend neoadjuvant chemotherapy, typically with cisplatin and etoposide followed by RT to the primary tumor and lymphatics. Concurrent chemoradiation is offered to patients who have initial large primary tumors or have a suboptimal response to chemotherapy alone.

T1-2N0 Glottic Carcinomas

T1-2N0 glottic carcinomas are treated with a pair of small, lateral, opposed photon fields that encompass only the larynx proper (Fig. 33-5). In patients with a short neck, a 5- to 10-degree inferior tilt may be necessary to avoid irradiation through the shoulders. Other techniques include anterior wedged pairs or a four-field technique as described by Wang.⁸⁴ For T1 lesions the field is centered on the true vocal cord and extends at or above the top of the thyroid notch superiorly to the lower edge of the cricoid cartilage inferiorly and from the anterior margin of the vertebral bodies posteriorly to approximately 1-cm falloff anteriorly. The superior and inferior borders are adjusted for T2 lesions that extend out of the glottis. When there is subglottic disease, at least one tracheal ring must be included in the fields. Typical field sizes range from 25 cm² (5 × 5 cm) to 36 cm² (6 × 6 cm). Harwood and associates¹⁰¹ reported better results with larger fields but acknowledged that at the time they used a free setup that likely accounted for some cases of geographic misses.

Figure 33-5 A, T1 glottic cancer. B, Simulation radiograph of a 5 × 5-cm field covering the glottis only for treatment of an early glottic tumor. C, Field for a T1 glottic cancer is outlined on the patient’s skin.

It is proper to use cobalt-60 gamma rays for treating glottic cancer, because higher-energy beams (6 MV or greater) may underdose the anterior commissure region, particularly in thin patients. Most radiation oncology centers nowadays no longer have cobalt-60 units and thus rely on linear accelerators. Local control rates of over 90% have been obtained for T1 lesions treated with 6-MV photons.153,154 Sombeck and colleagues¹⁵⁵ performed a dosimetric comparison between cobalt-60 and 6-MV beams and found a similar dose distribution except for the most anterior tissues (3 mm below the surface) where underdosing occurred with the 6-MV beam. Consequently, these authors cautioned against using the 6-MV beam for anterior commissure lesions. However, advocates of 6-MV photons for glottic cancer recommend either using tissue-equivalent bolus material for buildup or using a beam spoiler for thin patients or patients with anterior lesions.¹⁵⁴

A contour or CT plan is obtained to evaluate the dose distribution and determine the need for wedges (Fig. 33-6). Fifteen- or 30-degree wedges are often used and improve the dose distribution to the vocal cord, particularly for mid and posterior lesions. Treating without wedges results in a less homogeneous distribution, but this may be advantageous for anterior lesions when the built-in dose differential (5% to 10%) is desirable. Unilateral lesions can be treated by unequally weighting (e.g., 3 : 2) the fields, favoring the side of the lesion.

Figure 33-6 Dosimetry of parallel-opposed fields for early glottic carcinoma. Parallel-opposed, 5 × 5-cm fields were used with 30-degree wedges (heel anterior).

Radiation doses range from 60 to 70 Gy given in 2-Gy fractions. In general, a dose of 60 Gy is given to patients with no obvious clinical disease after stripping, 66 Gy to those with bulky T1, and 68 to 70 Gy to patients with T2 lesions. Most centers deliver 2 Gy per day, five fractions a week. Several studies evaluating the importance of dose per fraction principally for T1 disease show that 2 Gy per fraction yields better control rates than 1.8 Gy per treatment (Table 33-12). Some centers even advocate the use of larger fraction sizes.¹⁵⁶ Investigators from Osaka randomized patients to either 2 Gy or 2.25 Gy per fraction and demonstrated an improvement in control in patients receiving the higher dose per fraction.¹⁵⁷ Many groups now treat patients with T1 glottic disease with more hypofractionated regimens, and 63 Gy in 28 fractions is a very common regimen.

Based on biologic rationale, it has been attempted to improve the control rate of T2 glottic carcinoma by increasing the total dose through hyperfractionation.⁸² Two large retrospective studies have since been reported and both have suggested a trend (p >.05 but <.1) toward improved control with twice-daily treatments.^83,89 Mendenhall and colleagues⁸⁹ compared 182 patients treated with once-daily or twice-daily treatment and reported the following 5-year actuarial local control rates: T2a, 82% and 83%, respectively, and T2b, 71% and 69%, respectively. The authors contemplated that the advantages of hyperfractionation were mitigated by selection bias. Garden and colleagues⁸³ analyzed 230 patients, of whom 89 were treated with twice-daily fractionation. Local control rates were 79% and 67%, respectively, for patients treated with twice-daily or once-daily schedules. The authors hypothesized that although there may be a benefit to twice-daily treatment the data more strongly favored a higher daily dose as a more important treatment factor, because those patients treated with less than 2 Gy per day had control rates of 59% compared with an 80% control rate for those treated with more than 2 Gy per day.

The RTOG has completed accrual to a randomized trial designed to address this issue, and the results were presented in 2006. Hyperfractionation was associated with an increase in local control of approximately 9% compared with conventional RT at 5 years. Although this result was not statistically significant, it was noted that this small improvement was consistent with results seen in other hyperfractionation studies.¹⁵⁸

If hyperfractionation is not feasible, daily fractions of greater than 2 Gy per day result in superior outcomes to smaller daily doses. If patients can be treated occasionally twice daily, but not for the entire course, altered schedules such as concomitant boost or those of the DAHANCA are appropriate.

RT of early glottic cancers with small fields rarely induces severe complications. Fletcher and Goepfert¹¹³ described a 1% crude incidence of late edema in patients treated from 1962 through 1979. An analysis of late complications in 230 patients with T2 tumors revealed a 4% incidence of severe complications.⁸³ The complications were five cases of dysfunctional larynx and/or chondronecrosis requiring tracheotomy, one case of reversible laryngeal edema, and four cases of postoperative complications related to salvage.

Despite the absence of obvious complications, questions have arisen as to whether irradiation to the neck can lead to an increase in the incidence of stroke.¹⁵⁹ This has led several groups to assess the role of intensity-modulated radiation therapy (IMRT) in irradiating the larynx while sparing the carotid vessels. Two recent reports^160,161 have demonstrated that IMRT can reduce the dose to the carotid arteries (Fig. 33-7). A study by MDACC investigators also describes a preliminary experience using IMRT to treat 12 patients, and early data suggest local control rates equivalent to those expected for patients with stage I and II glottic cancer.¹⁶¹

Figure 33-7 Dosimetric plan of three-field IMRT for a patient with T1 cancer of the true vocal cord. A dose of 63 Gy was prescribed in 28 fractions. Note the 3000 cGy isodose line at the anteromedial edges of the carotid vessels.

T3N0 Glottic, T1-3 Supraglottic, T Any N Glottic, or Hypopharyngeal Carcinomas

T3N0 glottic and T1-3N0 supraglottic carcinomas are treated with parallel, opposed, lateral portals covering the primary tumor and the subdigastric (level II) and midjugular (level III) lymph nodes. A matching anterior portal (with superior midline block to prevent overlap on the spinal cord) encompassing the lower jugular (level IV) and supraclavicular lymph nodes is also used (Figs. 33-8 and 33-9). Hypopharyngeal cancers and larynx carcinomas with nodal involvement are treated with larger lateral portals covering the primary disease and the upper (level II) and midjugular (level III) nodes (under the insertion of the sternocleidomastoid muscle) and a matching anterior portal encompassing lower jugular (level IV) and supraclavicular lymph nodes. For hypopharyngeal tumors the inferior border must cover the entire piriform sinus including the apex (Figs. 33-10 and 33-11). Because of the posterior location of the majority of these tumors, the off-spinal cord fields extend slightly more posteriorly than for other head and neck primary tumors. This necessitates more frequent portal verification. It is important to encompass the retropharyngeal lymph nodes generously in patients with hypopharyngeal tumors. However, supraglottic lesions invading the medial wall of the piriform sinus only do not necessitate wide coverage of the retropharyngeal nodes.

Figure 33-8 This 55-year-old man had squamous cell carcinoma of the right true vocal cord. The entire cord was involved, and there was subglottic extension. The cord was fixed, and the patient’s disease was staged as T3N0M0. He received 70 Gy in 35 fractions with the shrinking fields shown and has been without disease for 30 months after therapy.

Figure 33-9 A 57-year-old woman presented with right otalgia and had a biopsy-proven squamous cell carcinoma of the epiglottis with extension of disease onto the aryepiglottic fold. The staging was T2N0M0, and she received 76.8 Gy at 1.2 Gy twice daily to the fields shown. She remains without evidence of disease.

Figure 33-10 This 74-year-old man had a biopsy-proven squamous cell carcinoma arising from the medial wall of the right piriform sinus. The lesion was 1 cm in diameter. The patient’s disease was staged T1N0M0. He received 66 Gy in 33 fractions. The initial fields included draining lymphatics and retropharyngeal nodes; an off-spinal cord reduction was made at 42 Gy, and the large fields continued to 50 Gy. A final reduction was made to the total dose. The patient remained without evidence of disease until his death 8 years later.

Figure 33-11 A 49-year-old woman with biopsy-proven stage T4N2b hypopharyngeal cancer is treated by irradiation. A, The initial treatment portals are right and left caudally with posteriorly oriented fields. B, After off-spinal cord reduction, the boost fields are parallel-opposed right anterior and left posterior oblique fields. C, A representative isodose distribution is shown.

The initial large field, on-cord RT can be accomplished with either cobalt-60 gamma rays or 6-MV photons depending on the neck diameter and the depth of the nodes. It is preferable to irradiate thick patients and those with deep-seated neck nodes with 6-MV beams. The off-cord and boost irradiations are administered with 6-MV beam with delivery of a supplemental dose to the posterior cervical nodes overlying the spinal cord with proper electron beams.

Treatment of head and neck cancer with IMRT has rapidly increased since the early 2000s. However, preliminary reports of IMRT have focused on treatment of nasopharynx and oropharynx cancer. In trials reporting more general head and neck results, the percentage of patients with laryngeal or hypopharyngeal cancers is extremely low.^162,163 However, particularly for hypopharyngeal cancers or patients with extensive nodal disease, the appeal of IMRT for potentially reducing xerostomia exists. Examples of selected target delineationand isodose distributions for patients with advanced supraglottic cancer and a patient with hypopharyngeal cancer are shown in Figures 33-12 to 33-14.

Figure 33-12 Target volumes for a patient treated with IMRT for T3N1 supraglottic cancer. Representative axial slices are shown at the level of the epiglottis (A) and arytenoids (B).

Figure 33-13 Dosimetric plan shown on an axial CT slice of a patient with T3N0 epiglottic cancer treated by IMRT. The bulky tumor can be seen at the level of the superior aspect of the hyoid. The tumor involves the epiglottis, fills the left pre-epiglottic space, and involves the left aryepiglottic fold and arytenoid. A dose of 70 Gy was prescribed to CTV1 (gross disease and margin). Because of concerns of larynx motion, the entire supraglottic larynx is included in CTV1. CTV2 and CTV3 are dosed to 63 and 56 Gy, respectively. CTV2 included level 3 nodes, and CTV3 included level 2 and 4 nodes. The patient was treated with concurrent cisplatin.

Figure 33-14 Isodose distributions on axial CT images of a patient with T3 carcinoma of the hypopharynx treated with concurrent chemotherapy and IMRT. Representative slices through the vestibule (A) and apex (B) of the piriform sinus are shown. The high-dose target volume for gross disease and the margin (CTV1) is transparent so the underlying tumor can be visualized. CTV1 was planned to 70 Gy and CTV2 (blue) was planned to 50 Gy. Treatment was planned with two sequential plans so that both targets would be fractionated at 2 Gy per fraction.

Patients with T1 disease usually receive daily 2-Gy fractions to 64 to 68 Gy. Patients with T2 to T3 lesions receive either daily 2-Gy fractions to 70 Gy, although preferably if they are to be treated with RT and no chemotherapy they should be treated with either hyperfractionated twice-daily RT with fraction size of 1.2 Gy to 76.8 Gy or, alternatively, are treated with an accelerated schedule such as concomitant boost or the DAHANCA schedule.¹¹⁵ The concomitant boost schedule¹⁶⁴ treats patients to their large fields with 1.8-Gy fractions to 54 Gy and delivers the boost dose with fraction size of 1.5 Gy per fraction during the last 2.5 weeks as a second daily fraction to bring the total dose to gross disease to 72 Gy delivered in 6 weeks. The DAHANCA regimen also treats patients in 6 weeks, but in 35 fractions. This is accomplished by delivering the boost as a second daily fraction once per week (for 5 weeks). Portal size is reduced to encompass the gross primary disease with 1- to 2-cm margins after reaching an elective dose of 50 Gy (2-Gy fractions) to 54 (1.8-Gy fractions) or 55.2 Gy (1.2-Gy fractions).

Because of larger radiation volume, RT of supraglottic and hypopharyngeal tumors induces a higher complication rate than treatment of early-stage glottic carcinomas. Mendenhall and associates¹¹⁴ have reported a 6% severe acute and late complication rate in 211 patients receiving primary RT for supraglottic carcinomas. In a series of 236 patients treated with hyperfractionation at MDACC, 21 patients developed grade 3 late complications; this includes 2 patients who died of massive hemoptysis 2 months post therapy and tracheal necrosis after a tracheotomy, respectively, and 2 patients who underwent laryngectomy for necrosis.¹⁶⁵ However, the 3-year actuarial complication rate has decreased from 14% to 8% since increasing the interfraction interval from 4 to 6 hours. We currently maintain the 6-hour interfraction interval for twice-daily irradiation.