The treatment approach varies with disease stage and site(s) of disease in the head and neck [13]. Approximately 1/3 of patients with HNSCC present with early stage disease. Depending on tumor location and institutional preferences, they are treated surgically or with radiotherapy, and approximately 80% are cured. Patients with locoregional advanced disease, which is surgically unresectable, and patients in whom definitive treatment is administered with an attempt at organ preservation (e.g., oropharyngeal and laryngeal carcinomas) undergo treatment with concurrent chemoradiotherapy [14, 15]. Cisplatinum is the drug with the most randomized clinical trial data to support its use as a drug enhancing the effects of radiotherapy in this setting. The larynx-preservation paradigm is supported by the results of a study that randomized 547 patients with stage III–IV supraglottic and glottic larynx cancer into three treatment arms: concurrent chemoradiotherapy, induction chemotherapy consisting of cisplatinum and 5-fluorouracil (5-FU) followed by radiotherapy, or radiotherapy alone. After a median follow-up of 3.5 years, the rates of locoregional control were 78, 61, and 56% for the three treatment arms, respectively. The fraction of patients who maintained an intact larynx at 2 years (and thus the ability to speak and swallow after the end of therapy) was also better with the concurrent regimen (88, 75, and 70% for the three treatment arms, respectively) [15]. Overall survival rates were similar in all three groups. The utility of concurrent high-dose cisplatinum for other subsites was established in a randomized study of 295 patients with unresectable head and neck cancer [16]. Concurrent chemoradiotherapy is therefore now widely applied as the definitive treatment of choice for locoregional advanced HNSCC. If residual disease is detected after the end of therapy or during follow-up, salvage surgery (e.g., laryngectomy) may be offered.

The management of the neck when using an organ-­preservation approach has remained somewhat controversial. Complete response rates in irradiated cervical lymph nodes vary between 59 and 83% and to some degree are related to nodal size, dose of radiotherapy, as well as the time point when response is determined: complete response rates are almost 100% in N1 disease, higher in N2 than in N3 disease, and better when the largest metastatic node is <3 cm [17]. In N2–N3 disease, residual cancer in neck nodes has been reported in 16–39% of patients achieving a clinical complete response (no overt residual neck mass) [17–19]. Over the past decade, locoregional control rates with concurrent chemoradiotherapy have improved. It is now widely accepted that routine neck dissection is not appropriate in this setting. Instead, [¹⁸F]FDG PET/CT has emerged as a gatekeeper for the management of the neck in this setting [13] (see following discussion).

Newer biologic therapies in HNSCC include angiogenesis inhibitors and drugs targeting the EGFR. EGFR is overexpressed and/or activated in the majority of HNSCC relative to normal tissue, and high expression is associated with poor disease control [20, 21]. While these data suggest HNSCC as an ideal malignancy for treatment with EGFR inhibitors, the selection of appropriate patients remains challenging. In studies of anti-EGFR agents in patients with advanced HNSCC, the objective response rates have been approximately 10%, depending on the specific agent [22]. Cetuximab, a chimeric IgG1 antibody against the extracellular domain of EGFR, is the most widely studied agent and has been tested in a number of clinical trials in patients with advanced disease. Cetuximab also enhances the efficacy of radiotherapy. In a randomized study of 420 patients [23], the addition of cetuximab to radiotherapy improved locoregional tumor control and overall survival without increasing mucositis and dysphagia when compared to radiotherapy alone. The corresponding median progression-free survival was 24 months versus 15 months, and the median overall survival was 49 months versus 29 months. This study showed that concurrent radiotherapy plus cetuximab is a viable treatment option in locoregional advanced HNSCC. Whether this alternate regimen improves outcome vis-à-vis the standard approach with cisplatinum based chemotherapy and concurrent radiotherapy is under investigation. Other ongoing trials are investigating the role of adjuvant cetuximab in patients with recurrent or metastatic HNSCC undergoing platinum-based chemotherapy. Preliminary data show a survival benefit over chemotherapy alone (reviewed in [24]). Antiangiogenic ­therapy has mainly been studied using Bevacizumab, a recombinant humanized, monoclonal IgG antibody against VEGF-A, usually in combination with chemotherapy or in combination with cetuximab. Preliminary data suggest that partial responses or disease stabilization can be achieved in >75% of patients [25]. Other drugs under study include various tyrosine kinase inhibitors. Recent improved understanding of relevant signaling pathways may open up new avenues for targeted drug therapies in HNSCC. For instance, downregulation of the transforming growth factor-β (TFG-β) receptor is frequently found, and the PI3K/PTEN/Akt/m-TOR pathway (important for increased proliferation, evasion of ­apoptosis, and increased survival) is frequently upregulated. A number of drugs interfering with these tumor pathways are under study [24].

Standard imaging studies for the staging and follow-up of head and neck cancer include contrast-enhanced CT, MRI, and ¹⁸F-FDG PET/CT. CT or MRI is essential for characterizing the primary tumor and in defining its extent. Bone involvement is better detected by CT and early bone marrow disease better with MRI. MRI can better define the interface between normal muscle and tumor. It is essential in staging nasopharyngeal carcinoma (NPC) and also has particular advantages for detecting intracranial tumor spread, perineural spread, and for differentiating entrapped secretions in paranasal sinuses from adjacent tumor.

Lymph node staging is done equally well with either CT or MRI. Traditionally, both imaging tests have relied on the assessment of lymph node size in order to identify metastatic disease. In general, lymph nodes ≥1.5 cm in levels I and highest level II, as well as lymph nodes ≥1.0 cm in all other locations, are considered abnormal. This approach is suboptimal because metastasis can be present in normal size lymph nodes and enlarged nodes may not contain tumor. Depending on the chosen cutoff for lymph node size, one can achieve high sensitivity at the expense of specificity or vice versa [26]. Secondary criteria are presence of grouped lymph nodes (usually >3 nodes in one location), heterogeneous nodal enhancement, stranding of surrounding fat indicating extracapsular spread, and demonstration of central necrosis (the most reliable criterion for nodal metastasis). Special MRI sequences, such as diffusion-weighted imaging (DWI), or newer MR contrast agents (including ultrasmall superparamagnetic iron oxide particles or more recently various nanoparticles) may improve the characterization of nodes. However, they are not yet routinely employed, and all available data are preliminary.

The vast majority of head and neck cancers show intense [¹⁸F]FDG uptake; however, uptake in salivary gland tumors may be variable (high [¹⁸F]FDG uptake in some benign tumors, low uptake in some malignant tumors). Combined PET/CT is a more accurate than PET alone and has beneficial effects on patient management [27, 28]. We perform [¹⁸F]FDG PET/CT imaging from the skull base to the floor of the pelvis for appropriate N and M staging in patients with head and neck cancer. Both distant metastases and potential synchronous primaries in the aerodigestive tract are hypermetabolic and therefore easily detectable. In order to save cost and potentially radiation dose, some investigators have proposed a limited field of view scan (e.g., ending at the level of the diaphragm) for patients with head and neck cancer. We, however, believe that a different approach should be taken: rather than using [¹⁸F]FDG PET indiscriminately in all patients, the test should only be used in subsets of patients where imaging findings are most likely to alter management, and in these cases it remains important to image the entire torso. For instance, it is essential in surgical patients with locally advanced disease. In addition, we use [¹⁸F]FDG PET/CT in all patients who are candidates for definitive chemoradiotherapy for the purpose of staging and radiotherapy planning; these scans are performed in the radiotherapy position and mask. PET/CT is also used for response assessment in patients with advanced disease undergoing chemotherapy or enrolled on experimental protocols.

Interpretation of head and neck PET/CT requires special expertise in head and neck anatomy and an understanding of physiologic [¹⁸F]FDG uptake in normal tissues and awareness of the many potential normal variants. Scan interpretation is primarily based on visual assessment, although SUV numbers may be helpful in estimating biologic aggressiveness and for response assessment. SUV is related to lesion size and may be underestimated in small primary tumors. Physiologic [¹⁸F]FDG uptake is nearly always noted in lymphoid tissues (tonsils) composing “Waldeyer’s ring,” including the pharyngeal tonsil (upper nasopharynx), symmetrical uptake in the palatine tonsils (unless the patients is status posttonsillectomy or suffers from unilateral tonsillitis or tonsillar carcinoma), and the lingual tonsil at the base of tongue. Variable uptake is noted in the tongue and floor of mouth muscles and is usually more intense if the patient was moving the tongue or swallowing during the uptake phase. Asymmetrical uptake in normal tongue can be noted after hemiglossectomy or even resection of smaller unilateral tongue lesions, leading to increased muscle activity and [¹⁸F]FDG uptake in the contralateral tongue. This is also noted after unilateral denervation, e.g., after hemimandibulectomy or due to hypoglossal nerve paralysis, where fatty atrophy and lack of [¹⁸F]FDG uptake in the denervated hemitongue is associated with relatively intense uptake in the contralateral normal tongue. Asymmetrical uptake, due to spasm, can sometimes be observed in pterygoid or temporalis muscles. Uptake along scalene muscles is frequently seen, either symmetrical or asymmetrical, and is easily characterized as normal variant on fusion images. Retained mucous in paranasal sinuses shows no [¹⁸F]FDG uptake unless there is severe inflammation, infection, or tumor. Similar to MRI (but not CT) [¹⁸F]FDG PET/CT differentiates well between tumor and entrapped secretions due to sinus outflow obstruction. While mucous retention or small polyps are fairly common in the paranasal sinuses, fluid retention in mastoid air cells should prompt careful assessment of the oropharynx to identify possible reasons for obstruction of the Eustachian tube.

Normally, symmetrical [¹⁸F]FDG uptake is seen in vocal cords and other laryngeal muscles (mild uptake is usually noted even if the patient does not vocalize during the uptake phase). Asymmetrical vocal cord uptake suggests vocal cord paralysis, usually due to damage of the ipsilateral recurrent laryngeal nerve. Secondary CT signs are sometimes helpful in confirming suspected vocal cord paralysis, including a ­paramedian position of the paralyzed vocal cord, tilting of the thyroid cartilage, displacement of the ipsilateral arytenoid cartilage, dilatation of the ipsilateral pyriform sinus, and prominent laryngeal ventricle. Long-standing vocal cord paralysis may lead to atrophy of the posterior cricoarytenoid and thyroarytenoid muscles [29, 30]. In searching for possible reasons for vocal cord paralysis, attention should focus on the upper mediastinum as well as lower neck to identify possible tumors or metastasis. Other reasons for vocal cord paralysis include nerve damage due to fibrosis (e.g., after radiotherapy) and chemotherapy (vinca alkaloids can cause transient vocal cord paralysis [31]). Asymmetrical or unusual focal [¹⁸F]FDG uptake can be observed in patients who received instillation of foreign material (e.g., Radiesse) for attempted medialization of a paralyzed cord. In this case the [¹⁸F]FDG uptake occurs on the treated site, likely due to foreign body reaction and inflammation. The instilled dense material is easily identified on the CT images. Large salivary glands usually show symmetrical [¹⁸F]FDG uptake (submandibular gland  >  parotid gland). A common cause for unilateral focal [¹⁸F]FDG uptake in parotid glands are pleomorphic adenomas (the most common benign tumor in the parotid gland) or “Warthin’s tumors” (which are entrapped normal lymphoid tissue). The submandibular gland is usually removed as part of neck dissections. Asymmetrical uptake in the contralateral remaining normal gland should not be confused with lymph node metastases. Postsurgical muscle flaps can show (normal) very intense [¹⁸F]FDG uptake, in particular at the flap pedicle, presumably related to hyperemia. This should not be confused with tumor recurrence. Some of the aforementioned normal findings and normal variations are described in detail in several review articles [32, 33].

This condition, also known as carcinoma of unknown primary, accounts for 3–15% of all cancer diagnoses and for approximately 1–2% of head and neck cancers. For practical purposes this entity should be defined as the combination of no history of previous malignancy, no clinical or laboratory evidence for primary neoplasm, and a neck mass that is histologically or cytologically proven to be carcinoma. Occurrence of nodal metastases in neck levels I–III increases the likelihood for a primary HNSCC. However, in 5–40% of cases a primary malignancy is not identified during diagnostic evaluation and long-term follow-up. Provided a primary tumor cannot be identified, in many institutions these patients are treated with extensive radiotherapy fields that cover the entire pharyngeal mucosa, the larynx, and bilateral necks. This is obviously associated with considerable morbidity. Identification of a primary tumor will direct rational therapy to a single mucosal site and the ipsilateral neck (unless the primary tumor is in, or close to, the midline). Detection of the primary tumor, in general, also allows for less extensive treatment (e.g., smaller radiation fields) and may potentially improve locoregional control. Therefore, these patients undergo extensive work-up. In some cases, the “unknown primary” is already identified in a thorough clinical exam by an experienced head and neck surgeon or oncologist; all other patients should undergo panendoscopy of upper aerodigestive tract and [¹⁸F]FDG PET/CT.

The literature is replete with studies in patients with unknown primary. Unfortunately, most study populations are very heterogeneous, and the definition of unknown primary and methods of verification vary considerably. For this chapter, I have only considered studies in which patients had at least undergone clinical exam, CT or MRI, and office endoscopy prior to PET (Table 10.2). The percentage of primary tumors detected is quoted as 14–37% [34–45]. When reviewing the literature, it should be noted that the sensitivity of PET is oftentimes stated incorrectly: since all neck metastases must have an underlying primary tumor, the denominator for calculating sensitivity should be the total number of patients studied (not the much smaller number of patients in whom the primary is eventually detected by imaging and panendoscopy). Therefore, it appears overall less controversial and more meaningful to simply state the rate of primary tumors that were only detected by PET (Table 10.2). Depending on the interpretation criteria, some studies report a high number of false-positive findings. It is therefore necessary to strike a reasonable balance between the desire to identify the location of the unknown primary on PET and the danger of calling every little variant suspicious. Potential reasons for a false-negative PET studies include a small primary tumor (although lesions as small as 4 mm can be detected as long as there is intense [¹⁸F]FDG uptake); low metabolic activity (cystic/necrotic tumor or lymph nodes) or the primary tumor was accidentally removed at time of neck dissection.

In our institution, patients with unknown primary, negative clinical exam and office endoscopy first undergo [¹⁸F]FDG PET/CT, before panendoscopy; panendoscopy is defined as nasopharyngoscopy, direct laryngoscopy, and upper esophagoscopy (in the absence of specific symptoms or imaging findings, a bronchoscopy is no longer done routinely because the yield is exceedingly low). This approach may reduce the number of necessary panendoscopies and avoid false-positive [¹⁸F]FDG uptake after manipulation during endoscopy and biopsy (Fig. 10.5). Of note, other imaging studies are usually not needed, because it is extremely unlikely that they may identify a primary tumor that cannot be detected by PET. Statistically, most primary tumors originate in the ipsilateral palatine tonsil, base of tongue, pyriform sinus, nasopharynx (more commonly in East Asian populations), and larynx. Particular attention should therefore be focused during PET/CT interpretation to these sites. If the PET study is ­positive, biopsies should be obtained from the suggested location. If the PET is negative, panendoscopy with biopsy sampling from suspected locations (based on the location of the lymph node and statistical considerations) should be ­performed. Depending on the results of these biopsies, the patient may proceed to surgery or radiation therapy.

Most HNSCC are diagnosed clinically (Tables 10.3, 10.4, 10.5, and 10.6). The primary goal of imaging studies is therefore not tumor detection, but rather to define the extent of the primary tumor, in particular with regard to structures whose involvement may alter the surgical approach (e.g., depth of soft tissue infiltration, bone invasion, orbital invasion, skull base invasion, tumor “tracking” along nerves and blood vessels).

Head and neck cancers tend to spread locally with invasion of adjacent structures. For instance, many cancers of the oral tongue tend to extend into the floor of the mouth. Local staging of HNSCC is routinely done by either CT or MRI, not PET. However, occasionally [¹⁸F]FDG PET may visualize small submucosal primary tumors that are difficult to distinguish from adjacent normal tissues with anatomic imaging studies. In a study of 48 patients with HNSCC (mostly tumors of the oral cavity), [¹⁸F]FDG PET detected 35/40 ­primaries known or still present at that time (88%), whereas contrast-enhanced CT of the neck identified a primary tumor only in 18 of the 35 patients (51%) with known or present primary tumors. Of the 17 primaries not detected by CT, 11 were clearly visualized by [¹⁸F]FDG PET [46]. PET/CT readers should pay attention to secondary tumor signs. For instance, cancers at the ­anterior floor of the mouth can cause obstruction of Wharton’s duct, through which the submandibular gland empties into the oral cavity. Ductal dilatation and enlargement of the gland, sometimes associated with increased [¹⁸F]FDG uptake, may be noted. Perineural spread can occur in various locations in the head and neck but is particularly common with adenoid cystic carcinoma (submucosal cancers that are common in the hard palate). Therefore, PET/CT readers should also be familiar with the normal branching patterns of cranial nerves relevant for innervation of the head and neck, in particular the facial nerve and trigeminal branches V/2 and V/3.