Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Head and neck cancers commonly metastasize to cervical nodes. The frequency of nodal metastases from head and neck squamous cell carcinoma (HNSCC) is approximately 50% although this is variable, being dependent on factors such as the primary tumour site and stage. Nodal metastases cause morbidity and mortality by invading vital structures in the neck and by increasing the risk of more distant tumour spread. Nodal status is a very important prognostic indicator in HNSCC as metastatic cervical nodes reduce patient survival by 30–50% (Layland et al. 2005; Tankere et al. 2000) and multiple nodes, extranodal spread and lower cervical metastases are additional adverse prognostic factors. The treatment options of surgery and/or (chemo) radiotherapy are influenced by nodal status. Given the importance of nodal disease on patient outcomes, prompt detection of nodal metastases at initial presentation and after treatment is essential to prevent their rapid progression to more advanced and unsalvageable disease. Clinical palpation, although a valuable skill, is inaccurate for nodal detection and characterisation (Merritt et al. 1997; Johnson 1990) and therefore imaging has an essential role at every step of the patient pathway from initial diagnosis through to post therapeutic surveillance.

This chapter will review the normal nodal distribution and lymphatic drainage pathways in the head and neck region and the imaging features that are employed for distinguishing metastatic nodes from benign nodes using both anatomical and functional imaging criteria. Nodal staging and imaging strategies with respect to clinical and subclinical nodal disease, as well as prediction of therapeutic response and post therapeutic evaluation, are outlined. Emphasis throughout this chapter is on HNSCC although other head and neck carcinomas as well as lymphoma will be discussed briefly.

2 Nodal Group Classification and Pathways of Lymphatic Drainage

Early classifications of cervical nodal distribution were purely anatomical in that individual nodes or nodal groups were named according to the structures they were in close proximity to, such as submandibular nodes, jugulodigastric nodes, etc. While this classification remains useful today for describing specific nodes, over the years a levels-based system has emerged with the objectives of being simple, reproducible and relevant in the era of selective surgical and radiotherapy treatments as well as cross-sectional imaging. In this system, the majority of cervical nodes are assigned to one of six levels (I–VI) whose boundaries are readily identifiable on imaging. A further seventh level (VII) is sometimes used also which refers to nodes in the superior mediastinum (Som et al. 1999). Several neck nodal groups are located outside these levels and are still referred to anatomically such as the retropharyngeal, facial, parotid and periparotid, occipital and peri auricular nodal groups. A more detailed account of the levels-based system can be found in “Clinical and Endoscopic Examination of the Head and Neck” and by reference to the article by Som et al. (1999), but a brief summary is shown in Fig. 1 and illustrated by axial MRI images (Figs. 2a–f and 3a, b).

Fig. 1
figure 1

Schematic illustration of the imaging-based level system of nodal distribution in the right side of the neck. Solid lines indicate boundaries of levels and dotted lines indicate level subdivisions. Facial, parotid, post auricular, occipital, retropharyngeal and superficial external jugular nodes (latter two groups are not shown) reside outside these levels and referred to by anatomical names. Level VII nodes are in the superior mediastinum (not shown). (d anterior belly of the digastric muscle, ca carotid artery (common and internal), cr cricoid cartilage, hy hyoid bone, ijv internal jugular vein, smg submandibular gland, scm sternocleidomastoid muscle)

Full size image
Fig. 2
figure 2

Distribution of cervical nodes using the imaging-based level system on axial T1 W post contrast MRI. (CCA common carotid artery, ICA internal carotid artery, IJV internal jugular vein, scm; sternocleidomastoid muscle) a Level IA (submental); nodes below the mylohyoid muscle, between the anterior bellies of the digastric muscles and hyoid bone. b Level IB (submandibular); nodes lateral to the anterior belly of digastric muscle and anterior to the posterior margin of the submandibular gland (the latter is denoted by a dashed line). Level II (upper internal jugular); nodes around the upper internal jugular vein. The lower limit is the inferior margin of the body of the hyoid bone. IIA nodes are anterior, medial or lateral, or posterior but inseparable from the vein. c Level IIA and Level IIB (upper internal jugular); nodes posterior to the upper internal jugular vein are IIA if inseparable from the vein and IIB if separable from the vein. d Level III (Mid jugular); nodes around the mid-internal jugular vein between the inferior margin of the body of the hyoid bone and inferior margin of the cricoid cartilage, lateral to the common carotid artery and anterior to the posterior margin of the sternocleidomastoid muscle. Level VA (upper posterior triangle); nodes posterior to posterior margin of the sternocleidomastoid muscle, anterior to the trapezius muscle and above the lower border of the cricoid cartilage (dashed line denotes the boundary between level III and VA). e Level IV (lower internal jugular); nodes around the lower internal jugular vein between the level of the cricoid cartilage and clavicle, lateral to the common carotid artery, and anterior to a line extending from the posterior margin of the sternocleidomastoid muscle to the posterolateral margin of the scalenus anterior muscle (denoted by a dashed line).Level VB (lower posterior triangle) nodes lie posterior to this line, anterior to the trapezius muscle and inferior to the cricoid cartilage. f Level VI (central compartment or visceral); nodes medial to the carotid arteries and extending between the hyoid bone and the manubrium. These include nodes along the thyroid gland, prelaryngeal and paratracheal nodes

Full size image
Fig. 3
figure 3

Additional nodes not covered by the level-based system (Facial, post auricular and occipital nodes as well as level VII (superior mediastinal) nodes are not shown) (ICA internal carotid artery, IJV internal jugular vein). a Retropharyngeal nodes, medial to the internal carotid artery and posterolateral to the pharynx. b Parotid and periparotid nodes

Full size image

Lymphatic drainage in the head and neck tends to be site-specific and follow predictable pathways down the neck. The first draining node or nodal basin for a given site is termed the first echelon or sentinel node(s) and subsequent nodal basins are termed the second and third echelons, etc. The typical lymphatic drainage for different primary cancer sites in the head and neck is listed in Table 1. Awareness of this lymphatic drainage has relevance for nodal staging, treatment planning and to search for occult primary cancers. Midline structures such as the tongue, or sites with a particularly enriched lymphatic network, such as the pyriform sinus, frequently drain bilaterally. Furthermore, lymphatic drainage can vary between individuals and can be disrupted by previous infections, neck surgery and radiotherapy.

Table 1 Primary head and neck cancers—main routes of nodal drainage
Full size table

3 Imaging Modalities

This section will provide a brief overview of the main modalities used for neck node evaluation, as well as of some of the newer functional imaging techniques which are presently in the domain of clinical research. Computed tomography (CT), magnetic resonance imaging (MRI), ultrasound (US) and (18)F-fluorodeoxyglucose positron emission tomography (FDG-PET) are the four main imaging modalities in routine clinical use for assessment of cervical nodal metastases. There is considerable variation in the reported accuracies of these modalities, which is unsurprising given the heterogeneity of study designs in terms of patient selection, imaging protocols, radiological diagnostic criteria and histopathological evaluation (Castelijns and van den Brekel 2002; van den Brekel et al. 1996). However, the choice of imaging technique is influenced also by other factors, which include the optimum modality for evaluating the primary site, available resources and radiological expertise. The limitations of conventional imaging for lymph node assessment have provided the impetus for research into new functional imaging techniques. These include techniques such as sentinel node lymphoscintigraphy and techniques that can be added to MRI such as diffusion-weighted imaging (DWI), proton magnetic resonance spectroscopy (MRS), dynamic contrast-enhanced MRI (DCE-MRI) and ultrasmall superparamagnetic iron oxide (USPIO) MRI, or added to CT such as dynamic contrast-enhanced CT. Some of these new techniques will be briefly mentioned at the end of this section.

3.1 CT and MRI

CT or MRI is the most widely used modality which reflects the fact that these modalities have acceptable and comparable accuracies for nodal malignancy, and because these can simultaneously assess the primary tumour and the neck nodes. A detailed account of the relative merits of CT and MRI is beyond the scope of this chapter although in general, MRI is excellent for soft tissue characterisation and does not involve ionising radiation, whereas CT is quicker to perform hence less susceptible to motion artifacts which make it especially suitable for dyspnoeic or claustrophobic patients. In practice whatever modality is used to stage the primary tumour is used also to stage the neck. Both modalities require intravenous contrast medium administration for optimum nodal characterisation.

3.2 US and US-Guided Fine-Needle Aspiration Cytology

US is the most practically useful imaging modality for cervical lymph nodes, although it is unsuitable in some deeply seated nodal groups such as those in the retropharyngeal space. US is quick, inexpensive, well tolerated, does not require intravenous contrast media and can display specific imaging features of discriminatory value more readily than CT or MRI, such as the nodal hilum and vascularity. A further major benefit of US is that it can be used to guide fine-needle aspiration for cytology (US-FNAC), which with sufficient experience can be performed successfully in nodes as small as 4 mm in diameter. The superiority of US-FNAC compared to US alone has been shown in several studies (Atula et al. 1997; Stuckensen et al. 2000; de Bondt et al. 2007). For example, one study documented 93% accuracy for US-FNAC compared to US alone (75%), CT (78%) and MRI (82%) (van den Brekel et al. 1993). Errors on US-FNAC are usually caused by sampling errors or cytological misinterpretation leading to false negative results. Because of its high accuracy, US-FNAC is well suited for investigating equivocal nodal findings in head and neck cancer patients, especially with the aim of distinguishing between patients with a N0 and a N+ neck, as this distinction has a major impact on clinical management. It is also performed in patients presenting with unexplained neck lymphadenopathy or suspected thyroid cancer where US is the initial imaging technique for assessing the primary site.

3.3 FDG-PET Imaging

Most head and neck cancers, including primary tumours as well as metastatic nodes, have increased glycolytic activity that can be detected using (18)F-fluorodeoxyglucose-PET (FDG-PET) imaging. Studies indicate that FDG-PET is superior to anatomic imaging techniques for the detection of nodal metastases in HNSCC (Schoder and Yeung 2004; Adams et al. 1998; Ng et al. 2006). However, FDG-PET is limited by its poorer spatial resolution and lack of anatomic detail. Combined PET/CT overcomes these limitations by permitting direct co-registration of functional and anatomic data, which translates into a higher accuracy for tissue characterisation than either PET or CT alone (Syed et al. 2005; Schwartz et al. 2005). In a recent study comparing imaging modalities for preoperative staging in HNSCC, FDG-PET/CT detected nodal metastases with 93% accuracy on a level-by-level basis, compared to 86% for patients evaluated by either CT or MRI (Roh et al. 2007). However, false negative and positive results occur using FDG-PET/CT thus, at present, there is no general consensus that using FDG-PET/CT for routine staging significantly improves clinical management (Ng et al. 2006; Hyde et al. 2003; Brouwer et al. 2004; Krabbe et al. 2008)

3.4 Diffusion-Weighted Imaging

DWI is the most developed of all the new functional MRI techniques. It evaluates free water motion (diffusion), which is quantified using apparent diffusion coefficients (ADCs) calculated from rate of signal loss on DWI images acquired with increasing diffusion encoding gradients (b-factors). DWI is non-invasive, relatively quick and can be performed with a large field of view to feasibly image most nodal groups in the neck. Nodal metastases show restricted diffusion and hence lower ADC values compared to normal nodes (Abdel Razek et al. 2006; de Bondt et al. 2009a; Holzapfel et al. 2009; Vandecaveye et al. 2009). However, to date there are few studies comparing DWI with conventional imaging, although one such study has shown that, for nodes between 4 and 9 mm, DWI achieved appreciably higher sensitivity than conventional turbo spin-echo MRI (76% vs. 7%) while specificity was only slightly lower (94% vs. 99.5%) (Vandecaveye et al. 2009).

3.5 Magnetic Resonance Spectroscopy

Proton (1H) MRS measures tissue metabolites, which may then be used for tissue characterisation. Most research to date has focused on two metabolites, choline (Cho) and creatine (Cr), that can be detected in many tissues as spectral peaks at 3.22 and 3.03 ppm. Preliminary studies in HNSCC document higher Cho/Cr ratios in both primary tumour and metastatic nodes compared to normal muscle (Mukherji et al. 1997; Bisdas et al. 2007; King et al. 2004a), and in metastatic nodes compared to reactive nodes (Bisdas et al. 2007). Furthermore, MRS has been shown to be of utility for differentiating cervical lymphadenopathies (King et al. 2005). Nevertheless, at present, MRS has important technical limitations that hamper a role in nodal staging. These include constraints on the minimum voxel size (~1 cm3), contamination by surrounding tissues, and prolonged acquisition times which make MRS both susceptible to motion artifacts and unfeasible for assessing the entire neck.

3.6 Dynamic Contrast-Enhanced (DCE) MRI

DCE-MRI uses low molecular weight paramagnetic contrast agents such as gadolinium to assess the microvasculature and extravascular extracellular space of nodal cancers. In the head and neck region rapid serial imaging is performed using T1 W gradient recalled echo sequences as the intravenous contrast passes through the node, and the resultant data are analysed using time-intensity plots or pharmacokinetic models. Early work suggests that there may be a difference in the contrast enhancement of metastatic and normal nodes, with the former showing a delayed and lower peak enhancement, and slower washout phase (Fischbein et al. 2003).

3.7 Lymphoscintigraphy for Sentinel Node Localisation

Lymphoscintigraphy can be used for sentinel node localisation. It is already an established technique in malignant melanoma and breast cancer and currently is under investigation in HNSCC. Sentinel node imaging typically entails injecting Tc-99 m radiolabelled colloid into tissues around the primary tumour and mapping the subsequent lymphatic drainage including first echelon nodes using preoperative planar scintigraphy or single photon emission computed tomography (SPECT) as well as intraoperative hand held gamma detectors. The localised sentinel node is subsequently excised and examined for cancer, which is used to predict whether there is likely cancer in downstream nodes. In recent years, there has been significant interest in the role of sentinel node biopsy (SNB) for staging the clinical N0 neck in HNSCC. Accumulating evidence from pilot studies of SNB in oral/oropharyngeal HNSCC suggest this technique can achieve over 90% sensitivity and specificity for predicting overall neck nodal status. On the basis of these results, the risk of occult metastases after a negative SNB may be under 5%, hence SNB in these patients could eliminate the need for prophylactic neck dissections (Alvarez Amezaga et al. 2007; Alkureishi et al. 2010). The indications for SNB have still to be established. In this regard it may be unsuitable for floor of mouth tumours due their close proximity to the lymphatic basin which hinders sentinel nodal detection (Alkureishi et al. 2010). As a modification of the sentinel node technique, pilot studies have evaluated using US-guided FNAC instead of excision biopsy for radiolocalised sentinel nodes in N0 HNSCC (Nieuwenhuis et al. 2000, 2002; Hoft et al. 2002). However, the results thus far have been discouraging as this approach achieves poor sensitivities for predicting a positive nodal status and offers no practical diagnostic gains compared to conventional US-FNAC (Nieuwenhuis et al. 2000; van den Brekel et al. 1991).

4 Imaging Criteria for Malignant Nodes

The main imaging modalities used for initial nodal evaluation, namely CT, MRI and US, utilise anatomic criteria to differentiate malignant from benign nodes, although some centres supplement this with functional data from FDG-PET. Imaging features more suggestive of normal or reactive nodes include small size, oval shape, presence of central hilum, hilar vascularity, homogeneous cortex (Figs. 4, 5, 6) and minimal FDG uptake. By comparison, imaging features suggestive of metastatic nodes include nodal enlargement, rounded shape, absent hilum, peripheral or other non hilar vascularity, heterogeneous parenchyma which includes necrosis or cystic change, border irregularity suggestive of extracapsular tumour spread (ECS) (Figs. 4, 5, 6) and avid FDG uptake. Each criterion is discussed in more detail below.

Fig. 4
figure 4

a Grey scale US of a normal lymph node (long arrow) showing the peripheral cortex (long arrow) central echogenic hilum (short arrow) with hilar vascularity. b Grey scale US (top) with corresponding grey scale and power Doppler US (below) of a metastatic node from SCC. The node shows enlargement, long:short axis ratio <2, absent hilum, absent hilar vascularity and a peripheral disorganised vascular pattern

Full size image
Fig. 5
figure 5

a Axial CT post contrast showing normal jugulodigastric lymph nodes bilaterally (long arrows). The hilum, when seen, appears as a focal depression of the parenchyma (short arrows). A nodal vessel is seen emerging from the nodal hilum on the right (arrowhead). b Axial CT pre and post contrast (left and right) in a patient with HNSCC showing enlarged rounded right level IIA metastatic nodes. The more anterior node (short arrow) appears solid, whereas the posterior node (long arrow) is necrotic

Full size image
Fig. 6
figure 6

a Axial T1 W MRI showing normal bilateral submandibular lymph nodes (arrows) with identifiable hila (arrowheads). b Axial T1 W post contrast MRI (left) and T2 W fat saturated MRI (right) showing metastatic left level IIA nodes (arrows) from nasopharyngeal carcinoma. The posterior node (short arrow) appears solid, whereas the anterior node (long arrow) is necrotic and has an ill-defined border suggestive of early extracapsular spread

Full size image

4.1 Size

Normal cervical lymph nodes range in size from a few millimeters to just over 2.5 cm in longitudinal diameter (Figs. 4, 5, 6). Nodal size declines with age and varies according to location. Nodes in the upper internal jugular chain, notably the jugulodigastric region, are usually the largest due to mild hyperplasia as a result of chronic exposure to repeated infections in the oral cavity and pharynx, while nodes towards the skull base and thoracic inlet tend to be smaller.

Nodal enlargement is an important feature of malignant infiltration although no threshold of nodal size is both highly sensitive and highly specific, which reflects the fact that moderately enlarged nodes in the neck are frequently reactive while small tumour deposits may not alter nodal dimensions. Over the years, numerous studies have investigated optimal size cut-offs for nodal metastases in HNSCC. One of the most influential is a CT-based study by van den Brekel et al. that documented short axis diameter thresholds of 11 mm for level II and 10 mm elsewhere as the most accurate overall, achieving 82% sensitivity and specificity. In addition, these thresholds could be lowered to 8–10 mm for groups of 3 or more nodes draining a primary tumour (van den Brekel et al. 1990). These criteria are still widely used today, albeit with minor modifications such as a lower threshold of 5 mm short axis diameter for retropharyngeal nodes (King et al. 2000).

Nevertheless, it should be highlighted that optimum size criteria are arbitrary and may be altered depending on whether a higher sensitivity or higher specificity is desired. One early study documented that lowering the short axis diameter cut-off from 15 to 5 mm increased the sensitivity of CT from 56 to 98% but lowered the specificity from 84 to 13% (Curtin et al. 1998). In addition, the patient population may influence optimum thresholds. Van den Brekel et al. performed another study of size criteria in HNSCC, this time using US, and found that thresholds achieving the highest global accuracy were 9 mm for level II and 8 mm for other levels (74% sensitivity and 78% specificity), whereas for a subgroup of patients without palpable nodes, optimal thresholds were lower, namely 7 mm for level II and 6 mm for other levels (80% sensitivity and 59% specificity) (van den Brekel et al. 1998a).

4.2 Shape

Normal or reactive nodes tend to be oval or reniform in shape and have a ratio of the longitudinal to transverse axis diameter (L/T) >2, while nodal metastases are more rounded resulting in a lower L/T ratio (Figs. 4, 5, 6). In clinical practice, a L/T ratio <2 is widely adopted as a criterion of malignancy. In one large study, nodal shape achieved 95% accuracy although it should be noted that the majority of these nodes were >1 cm (Steinkamp et al. 1995). Furthermore, normal nodes in the submandibular and parotid regions tend to be more rounded. The presence of eccentric cortical widening has been described in malignant nodes including in the neck (Vassallo et al. 1992), although this feature may also occasionally be found in normal or reactive nodes, especially those in the submandibular region.

4.3 Hilum

Lymph nodes contain a hilum composed of arteries, veins, lymphatics and sometimes fat (Figs. 4, 5, 6). This is visualised optimally by US due to the presence of numerous reflective interfaces within the lymphatic sinuses. Using grey scale US, the hilum appears as an echogenic linear structure traversing into the centre of the node, and is detected in ~66% and >90% of normal nodes <5 and >5 mm, respectively (Ying et al. 2001). Conventional CT and MRI are generally less capable of depicting hila, although with careful attention a hilum is often identifiable, especially when there is fatty metaplasia. Loss of the nodal hilum can occur in malignant nodes due to direct invasion or compression by tumour. The reported sensitivity of this feature on conventional US varies widely between 44 and 95% (Vassallo et al. 1992; Ahuja et al. 2001a; Ahuja and Ying 2002; Ying et al. 1998). False negative results using this criterion are not surprising given that the majority of microscopic tumour deposits are located within the subcapsular sinuses, as afferent lymphatic channels enter the nodes peripherally (van den Brekel et al. 1996).

4.4 Vascular Pattern

The vascular architecture within lymph nodes can be assessed using power Doppler US, which in normal nodes typically extends into the hilum and branches outwards into the parenchyma (Figs. 4, 7). This hilar vascular pattern is more florid in reactive nodes. In contrast, metastatic nodes tend to display disorganised vascular patterns including peripheral cortical or subcortical vascularity, and focal areas of displaced or absent vascularity, with or without preservation of the hilar pattern (Ariji et al. 1998; Na et al. 1997a; Wu et al. 1998). Published studies using vascularity to differentiate metastatic from reactive nodes have reported sensitivities of 83–89% and specificities of 87–98% (Ariji et al. 1998; Wu et al. 1998). While vascular patterns are important, they are not always demonstrable in normal nodes, especially in those under 5 mm (Ying et al. 2001). Furthermore, abnormal vascular patterns are non-specific as they may occur in non-malignant lymphadenopathies (Ahuja et al. 2001b). Quantitative indices of vascular resistance including the resistive index (RI) and pulsatility index (PI) can be measured in intranodal vessels using spectral Doppler US. Several studies have shown that metastatic nodes have higher RI and PI than benign nodes (Na et al. 1997a; Wu et al. 1998; Ahuja et al. 2001b; Steinkamp et al. 1994; Dragoni et al. 1999). Measuring these indices is highly operator dependent and so currently there is no consensus on the optimal cut-offs for malignancy.

Fig. 7
figure 7

Grey scale US with power Doppler showing a prominent reactive submandibular lymph node with florid hilar vascularity (arrows)

Full size image

4.5 Parenchymal Heterogeneity and Necrosis

Normal parenchyma in the nodal cortex appears relatively uniform on conventional imaging whereas parenchymal heterogeneity is usually pathological (Figs. 4, 5, 6). Moreover, in the context of known HNSCC, nodal heterogeneity is specific for metastasis. These heterogeneous areas may be composed of variable amounts of solid tumour, deposits of keratin and necrotic tumour, which may or may not be cystic. Intranodal necrosis in HNSCC correlates with nodal size as it reflects tumour growth beyond its blood supply, and is identified in 56–63% of metastatic nodes >1.5 cm compared to 10–33% of nodes <1 cm (Friedman et al. 1993; Don et al. 1995). CT, MRI and US can all depict necrosis >3 mm in size; the accuracies of MRI and CT for detecting necrosis are similar (91–92%) and higher than that of US (84%) (King et al. 2004b). The lower accuracy of US may be explained partly by the fact that necrosis sometimes can appear isoechoic to normal nodal parenchyma. Metastatic nodes can be almost entirely cystic and thus may cause initial diagnostic confusion with benign cystic lesions, namely branchial cleft cysts, especially if solitary and located along the mid to upper jugular chain. Nevertheless, the precise location, appearance and different clinical presentations of head and neck cancers and benign cysts can usually discriminate these conditions. Metastatic nodal calcification is rare at presentation in HNSCC (Eisenkraft and Som 1999; Gormly and Glastonbury 2004) and the presence of calcification should raise the suspicion of nodal metastases from thyroid cancer. Nodal calcification may also be found in benign pathologies, especially granulomatous diseases, as well as other untreated and treated malignancies such as adenocarcinoma and lymphoma (Eisenkraft and Som 1999).

4.6 Border Irregularity

In the absence of infection or previous treatment, loss of the normally smooth nodal border is highly suspicious for ECS (Figs. 6 , 8, 9 ). The risk of ECS increases with nodal size, being present in approximately 50 and 75% of nodal metastases with a diameter of 2–3 cm and >3 cm, respectively (Snow et al. 1982; Collins 1987). In addition, ECS is reported to occur in 23% of metastatic nodes <1 cm, which accounts for 30% of all nodal ECS (Don et al. 1995; Collins 1987). ECS has major prognostic significance, which is discussed in a later section. The reported sensitivity and specificity of CT for ECS is 62–81% and 60–90% (Carvalho et al. 1991; Steinkamp et al. 1999; Xu et al. 1998; Souter et al. 2009) and of MRI is 74 and 72% (Steinkamp et al. 2002). Several early studies reported that MRI was inferior to CT for detecting ECS, although more recent studies indicate comparable accuracies (King et al. 2004c). Moreover, another MRI study documented that the presence of border irregularity on T2 W fat suppressed sequences achieved up to 87% sensitivity and 94% specificity for metastases, and encouragingly the accuracy remained high for nodes <10 mm in short axis diameter (up to 78% sensitivity and 97% specificity) (de Bondt et al. 2009b). Nevertheless, imaging detection of ECS is subjective and dependent on radiologists experience as it may be subtle or mistaken for perinodal inflammatory changes (Souter et al. 2009; Steinkamp et al. 2002; de Bondt et al. 2009b). It is noteworthy that metastatic nodes may appear paradoxically to have sharper, well-demarcated margins than benign nodes on grey scale US because of their reduced echogenicity, as this increases their conspicuity against the echogenic surrounding tissue (Shozushima et al. 1990).

Fig. 8
figure 8

Axial T1 W post contrast MRI in a patient with HNSCC showing metastatic left mid-upper internal jugular lymphadenopathy with gross extracapsular spread (small arrows). The enhancing nodal borders are ill defined and extend into the surrounding deep cervical fascia as well as invade the left sternocleidomastoid muscle and paravertebral muscles, both of which show abnormal enhancement. Tumour also contacts the left internal carotid artery and internal jugular vein. (ICA; internal carotid artery, IJV; internal jugular vein, scm; sternocleidomastoid muscle)

Full size image
Fig. 9
figure 9

Axial CT post contrast showing a heterogeneous right level II nodal mass with gross extracapsular spread. There is complete encasement of the right internal carotid artery (ICA), which is narrowed, as well as frank invasion of muscles and overlying skin (short arrows)

Full size image

4.7 FDG-PET Uptake

On FDG-PET, tissue FDG uptake can be assessed semi-quantitatively using standardised uptake values (SUVs). In normal nodes there is typically no or minimally increased SUV relative to background activity in the surrounding cervical tissue. By comparison, metastatic nodes typically show elevated SUVs, and the higher the SUV, the more likely a node is malignant. Despite these generalisations, at present there is no consensus in the literature regarding an appropriate SUV cut-off for malignancy, which partly reflects the overlap between uptake in benign and malignant nodes. False negative FDG-PET results are usually due to intrinsic low metabolic activity in small tumour deposits, and for this reason the utility of FDG-PET in patients with no palpable nodal disease is debated (Ng et al. 2006; Brouwer et al. 2004; Krabbe et al. 2008). Identification of metastatic nodes may be hampered also by low FDG uptake in markedly necrotic nodes, and when a node lies in close proximity to the primary tumour, as nodal uptake may be obscured by intense uptake at the primary site (King et al. 2008). False positive FDG-PET results can occur due to hypermetabolism in reactive nodes as well as other benign lymphadenopathies, recent biopsy and infection.

4.8 Functional MRI

On DWI, metastatic nodes show restricted diffusion and a low ADC (Fig. 10). Regions of necrosis, which are common in HSNCC, should be excluded from the ADC measurement as these reduce the discriminatory performance of DWI by increasing the ADC. In a recent study comparing SCC with benign nodes, DWI achieved 84% sensitivity and 94% specificity for metastases in nodes measuring >4 mm in short axis diameter (Vandecaveye et al. 2009). However, at present the optimum thresholds for discrimination of metastatic from benign nodes have still to be confirmed. On MRS, metastatic nodes have high Cho/Cr ratios (Fig. 11) but again optimum discriminatory thresholds have not been established.

Fig. 10
figure 10

Axial DWI-MRI diffusion image at b1000 (left) with corresponding ADC map (right) in a patient with HNSCC. Metastatic right level II nodes (short arrows) and a left level II nodal mass (long arrow) have a mean ADC of 0.94 and 1.04 × 10−3 mm2/s respectively. These values are comparatively low, signifying restricted diffusion, which is compatible with malignant infiltration

Full size image
Fig. 11
figure 11

Proton MRS of a metastatic periparotid node from HNSCC. a) Axial T1W post contrast MRI showing MRS voxel placement (red box). b) The corresponding MRS spectrum shows an elevated Cho peak and a high Cho/Cr ratio of 5.64, which is suggestive of malignancy. Lipid peaks are also demonstrated

Full size image

5 Micrometastases

It should be noted that cancerous deposits may be microscopic, for instance comprising small clusters of cells, hence it is not feasible nor realistic to expect any of the current imaging modalities to detect these. Radiologically, the term micrometastases generally refers to tumour deposits <3 mm, which are not detected reliably by present day imaging techniques. These have been found in 25% of neck dissection specimens from patients with no clinicoradiological evidence of nodal disease (clinical N0) (van den Brekel et al. 1996). While micrometastases usually occur alongside larger deposits, in 10–25% of clinical N0 but pathologically positive neck dissection specimens, they are the sole manifestation of nodal malignancy (van den Brekel et al. 1992, 1996). Their exact clinical significance is unclear some of these could progress to clinically apparent disease if left untreated.

6 Nodal Staging

Nodal staging of metastases from HNSCC is incorporated into the TNM system, which currently uses the 7th editions of the AJCC and UICC guidelines (Sobin et al. 2009; Edge et al. 2010). Staging evaluates nodal number (single or multiple), site (unilateral, contralateral or bilateral) and greatest dimension (using 3 and 6 cm thresholds). Of note, while diagnosis of a metastatic node entails measurement of the minimum axial diameter, staging entails measurement of the maximum diameter. In general, a single ipsilateral node ≤3 cm indicates N1 disease, any node >6 cm indicates N3; any other combination indicates N2 disease. N2 is further subdivided into N2a (single node >3 cm), N2b (multiple ipsilateral nodes) and N2c (bilateral or contralateral nodes). Nodal staging in HNSCC, nasopharyngeal carcinoma and thyroid cancer are summarised in Table 2. In HNSCC, the presence of metastatic nodes denotes a moderate or advanced overall stage III or IV.

Table 2 N staging of lymph node metastasis of head and neck cancers (AJCC/UICC 7th editions)
Full size table

7 Impact of Nodal Imaging on Patient Management

7.1 Detection of Metastatic Nodes

To understand the impact of imaging for identifying or excluding metastatic nodes, it is worth outlining the treatment strategies for the neck with (N+) and without (N0) clinicoradiological evidence of nodal metastases.

N+ neck: In patients with any evidence of metastatic neck nodal disease, definitive surgical treatment usually entails a radical or modified radical neck dissection on the involved side(s). Both of these operations remove all lymph nodes within levels I–V operated(s), while these differ only in terms of whether extra-lymphatic structures (spinal accessory nerve, sternocleidomastoid muscle, internal jugular vein) are also removed (Robbins et al. 2002). In recent years several workers have advocated the use of a selective neck dissection (SND), which preserves at least one of the neck levels I–V to treat early node positive disease (localised, small volume nodes that lack ECS) (Andersen et al. 2002; Schmitz et al. 2009; Traynor et al. 1996) while minimising surgical morbidity. Radiotherapy fields usually encompass both sides of the neck although current radiotherapy regimes, which include intensity modulated radiotherapy (IMRT), deliver higher doses to positive nodes and lower doses to normal appearing nodal groups, especially those that lie close to important structures such as the parotid glands, oesphagus and spinal cord.

N0 neck: Depending on the primary site, up to 40% of necks in HNSCC with no palpable or radiological nodal metastases harbour occult nodal metastases (van den Brekel et al. 1996). Clinical management of these patients varies between a “watchful waiting” policy and elective treatment using radiotherapy or neck dissection. One surgical approach in the N0 neck is to perform a prophylactic unilateral SND on the same side as the primary tumour in order to diagnose and remove potential sites of occult metastases, and occasionally bilateral SNDs when the tumour is located in the midline. However, the optimal management of this subgroup is controversial in terms of how to balance the increased risk of unsalvageable recurrences that would occur under a “watchful waiting” policy in the minority of patients who have disease against the morbidity of prophylactic neck treatments in the majority of patients who do not have disease. To address this dilemma, many clinicians choose to surgically treat the neck electively if the risk of occult nodal metastases in the individual exceeds 20% (Weiss et al. 1994; Pfister et al. 2000; Pillsbury and Clark 1997). Patients with early T stage tumours in the glottis, parotid gland and paranasal sinuses have a low risk of nodal metastasis (<10%), hence the neck is usually managed by watchful waiting. Conversely, patients with tumours that have an advanced T-stage, are deeply invading (e.g. >4 mm in oral tongue cancers), have a high grade or are at high-risk sites (including the oropharynx, hypopharynx and supraglottis) usually undergo prophylactic neck treatment by surgery or radiotherapy (Shear et al. 1976; Martinez-Gimeno et al. 1995).

Based on the treatment stratifications described above, nodal imaging has the most impact on patient management in the following situations:

  1. 1.

    Distinguishing between patients with an N0 and an N+ neck. In this situation, finding even one single metastatic node in an otherwise normal neck could change management from a conservative “watchful waiting” policy to surgery/radiotherapy, or change the surgical approach from a limited to a more extensive neck dissection. Conversely, proving that a single indeterminate node is reactive rather than metastatic could have the opposite effect on management.

  2. 2.

    Identification of metastatic nodes outside the usual surgical and radiotherapy treatment fields including the contralateral neck, retropharyngeal, parotid, suboccipital, buccal, facial and level VI nodes; nodes adjacent to sites which may be preferentially spared during radiotherapy, such as the salivary glands. In these situations, the radiotherapy or surgical fields may need to be extended or the modality of treatment changed from one to the other.

To this end, US-FNAC is an invaluable ancillary investigation for any accessible lymph nodes with equivocal malignant features on imaging that will alter treatment (Dirix et al. 2006; Bussels et al. 2006).

7.2 Extracapsular Tumour Spread

Extracapsular tumour invasion of adjacent structures should be carefully documented as this may affect nodal resectability (Table 2) (Figs. 6, 8, 9). Tumour invasion of the internal or common carotid artery usually precludes neck dissection as the initial treatment due to the high risk of severe surgical morbidity combined with the risks of generally poor prognosis of these patients due to advanced disease (Freeman 2005). Unfortunately, imaging has a limited accuracy for detecting carotid invasion although, as a general rule, tumour directly contacting the vessel under 180° and over 270° of the vessel circumference indicates that vascular invasion is probably absent and present, respectively (Steinkamp et al. 1999; Yousem et al. 1995; Yoo et al. 2000). In addition, some workers have documented that tumour fixation to the artery may be detected by palpation during US (Gritzmann et al. 1990; Mann et al. 1994).

Invasion of the prevertebral muscle or fascia tends to indicate unresectable disease to be within muscle. The presence of an intact fat plane between the tumour and the prevertebral fascia on imaging usually indicates that the fascia has not been breached, but the loss of this fat plane does not necessarily indicate that invasion is present.

Fortunately, tumoural invasion of the external carotid artery or its branches does not preclude surgery because these vessels can be sacrificed due to the rich collateral arterial supply in the head and neck region from the contralateral side. Similarly, the sternocleidomastoid muscle, internal jugular vein and skin can all be sacrificed if there is suspected involvement of these structures on imaging.

7.3 Identification of Patients at High Risk for Distant Metastases

Several nodal features at diagnosis are associated with a higher risk of developing distant metastases, including size >6 cm, >3 nodes involved, ECS and nodes in the lower neck/supraclavicular fossa (Alvi and Johnson 1997; de Bree et al. 2000; Leemans et al. 1993). These criteria may be used to decide whether further imaging, such as FDG-PET, is required to search for distant metastases centres recommend further imaging studies such as whole body PET-CT (Brouwer et al. 2006; Senft A et al. 2008).

8 Treatment Assessment

8.1 Prediction of Treatment Response to (Chemo)radiotherapy

There is currently great interest in identifying imaging markers with prognostic significance for (chemo)radiotherapy response which can be obtained either before treatment or early in a course of treatment, as these may facilitate better therapeutic stratification including dose escalations or cessation of (chemo)radiotherapy in favour of early conversion to surgery. Besides those markers already incorporated into the TNM staging (maximum nodal size, multiplicity and laterality), ECS is probably the most important as it indicates biologically aggressive disease as evidenced by increased risk of locoregional failure, distant metastases and reduced survival by as much as 50% (Myers et al. 2001; Larsen et al. 2009). At present most of the research regarding ECS is based on histology rather than imaging data. In the current edition of the TNM staging (Edge et al. 2010), ECS does not alter N stage per se although has been added as a descriptor (E+ or E– for clinical/radiologic evidence).

The prognostic significance of intranodal necrosis detected on imaging is unclear. A few studies using CT have documented that appreciable pre-treatment nodal necrosis is an independent risk factor for regional failure to (chemo)radiotherapy (Grabenbauer et al. 1998; Munck et al. 1991; Wang et al. 1996), which may be explained by the fact that tumour radiosensitivity is impaired by tissue hypoxia. Nevertheless, in clinical practice it is common for markedly necrotic nodes to demonstrate a complete response to (chemo)radiotherapy.

Several markers from functional imaging techniques obtained before treatment or early in the course of treatment show promise for prognostication. Early results suggest that local control is more likely to occur in nodes with higher vascularity as shown by a higher K trans on DCE-MRI (Kim et al. 2010). Nodes with a lower pre-treatment ADC and a higher early treatment rise in ADC on DWI (Fig. 12) (Kim et al. 2009; Vandecaveye et al. 2010) are also more likely to show tumour control. Using FDG-PET, primary tumours with a lower SUV are more likely to show an overall response in the neck (Schwartz et al. 2003; Wong et al. 2002). Furthermore, high pretreatment FDG uptake in nodal metastasis in HNSCC may be a risk factor for distant recurrence (Kubicek et al. 2010).

Fig. 12
figure 12

Axial DWI-MRI ADC map (left) and T1 W post contrast MRI in a patient with a left tonsillar SCC and metastatic IIA node pretreatment (upper row) and 2 weeks intra treatment with concomitant chemoradiotherapy (lower row). The ADC of the node increased from 1.1 × 10−3 mm2/s pretreatment to 1.5 × 10−3 mm2/s at 2 weeks intra treatment. This indicates increased diffusion, which favours a treatment response. There has been no locoregional recurrence as of 3 years post treatment

Full size image

8.2 Post Treatment Assessment

The majority of nodal metastases in HNSCC respond completely to (chemo)radiotherapy; in a minority of patients residual tumour persists, which may progress to surgically unsalvageable disease. Historically, regional control rates for early nodal disease treated surgically or by radiotherapy were broadly equivalent, whereas for advanced nodal disease, control rates from radiotherapy alone were poor and suboptimal compared to radiotherapy followed by a planned neck dissection. For this reason, it was common practice to perform a neck dissection in this cohort, irrespective of whether or not there was evidence of a complete clinic-radiological response to radiotherapy. While this aggressive strategy reduced the risk of regional failures, this produced unnecessary surgical morbidity as up to two thirds of planned neck dissections were pathologically negative (Ojiri et al. 2002). Fortunately, with the development of more efficacious chemoradiotherapeutic regimens, there has been a shift towards a more conservative approach in terms of only performing neck dissection after (chemo)radiotherapy if there is clinical evidence of residual disease. This shift in practice places an even greater burden on imaging modalities to distinguish patients with post treatment regional failure who will require early salvage surgery, from patients with regional control, who can be placed under close imaging surveillance instead.

MRI and CT both have a high negative predictive value (94–97%) for residual nodal disease if the nodal metastases have essentially disappeared (Ojiri et al. 2002; Liauw et al. 2006; Yeung et al. 2008; Lin et al. 2007); under these circumstances patients may be spared surgery and undergo imaging surveillance instead. However, sterilised nodal metastases frequently regress incompletely, to leave behind a variable amount of scar tissue (Fig. 13). CT-based studies have documented residual nodes ≤1.5 cm with no focal abnormality (Liauw et al. 2006), or nodes with >90% reduction in size (Labadie et al. 2000) to be associated with tumour control. Larger residual masses or those containing a focal abnormality are non-specific as these may represent residual cancer or a benign mass of inflammatory tissue and fibrotic scarring. The documented positive predictive value of a post treatment mass is low, ranging between 36 and 59% (Ojiri et al. 2002; Liauw et al. 2006; Inohara et al. 2009). Ojiri et al. documented that 57% of hemi necks with nodes >15 mm were pathologically negative on neck dissection (Ojiri et al. 2002), and Yao et al. documented that post treatment nodes >50 mm could be disease-free (Yao et al. 2004). Difficulties in post treatment nodal assessment are partly due to the fact that anatomic abnormalities associated with malignancy can resolve very slowly after successful treatment, especially in nodes that were predominantly necrotic or had extracapsular invasion into adjacent muscles before treatment. The performance of US is similar to CT and MRI in that the disappearance of nodes usually indicates response (Furukawa and Furukawa 2010; Yusa et al. 2000), whereas residual masses are non-specific. It has been shown that sonographic features associated with malignancy, namely abnormal vascularity, altered echogenicity, absent hilum and border irregularity can persist in the benign post treatment mass (Yusa et al. 2000). Furthermore, while US guided biopsy can be performed in these cases, this is prone to sampling errors due to the fact that residual cancer may be small and dispersed within larger areas of inflammation and scarring.

Fig. 13
figure 13

Axial T1 W post contrast MRI showing right nodal metastases pretreatment (left) and 6 months post chemoradiotherapy (right). An enlarged right level II node (arrow) has reduced significantly in size following treatment although there is a small amount of residual soft tissue remaining. This was believed to represent a post treatment scar and there has been no regional recurrence as of 3 years post treatment

Full size image

FDG-PET is now well established in the post treatment assessment of HNSCC (Fig. 14) and has become a part of routine imaging follow-up in some centres. The false positive and negative results from FDG-PET can be reduced by delaying the scan until at least several months after treatment. From the accumulating evidence which includes several review papers (Wong 2008; Isles et al. 2008; de Bree et al. 2009), the consensus appears to be that (1) FDG-PET/CT is more accurate than CT or MRI, (2) subjective assessment is often superior to SUV measurements, (3) FDG-PET/CT performed at least 10–12 weeks after treatment has a high negative predictive value (96%) although the positive predictive value is low (49%) (Isles et al. 2008).

Fig. 14
figure 14

Axial FDG-PET/CT at the oropharyngeal level in a patient with a right tonsillar SCC (short arrow) and a 1.8 cm metastatic right level II node (long arrow) before treatment (top row) and after completion of chemoradiation (bottom row). Pretreatment, both tumour sites show avid FDG uptake; tonsil SUVmax = 6.1, Level II node SUVmax = 4.3. Post treatment, neither site shows appreciable increased FDG uptake. The patient has had no recurrence as of 2 years post treatment. Mild physiological FDG uptake is present in lymphoid tissue in the tongue base (arrowheads)

Full size image

Functional MRI and CT techniques can be easily incorporated into post treatment assessment and of these DWI appears to be the most promising. Successful treatment leads to cell death and a progressive rise in ADC so that benign post treatment residual masses have significantly higher ADCs than residual cancers, even as early as 3–6 weeks post treatment (de Bree et al. 2009; Vandecaveye et al. 2011). It is difficult to establish a universal ADC threshold for residual cancer due to differences in DWI protocols between institutions although several studies suggest a potential discriminatory cut-off ~1.3 × 10−3 mm2/s (Vandecaveye et al. 2010; Abdel Razek et al. 2007; King et al. 2010) or a less than 20% rise in ADC values when compared to the pre-treatment DWI (Vandecaveye et al. 2011). One precaution when evaluating DWI is to exclude areas of frank necrosis from the ADC measurement as these may undergo early fibrotic organisation, which lowers the ADC and thus can simulate residual cancer. Using proton MRS, the persistence of a Cho peak after treatment may be a marker for residual cancer (King et al. 2010). Finally, preliminary studies suggest that residual cancers have greater vascularity than benign post treatment masses on DCE-MRI (Hoskin et al. 1999), although an optimum DCE parameter and threshold has not been established.

8.3 Post Treatment Surveillance

Even after employing anatomical imaging, functional imaging and imaging-guided biopsy, it may not be possible to distinguish between a nodal mass caused by residual cancer and one caused by benign post treatment inflammation and scarring. Currently, there is a shift towards less surgical intervention, even in those patients with residual nodal masses post chemo(radiation). Furthermore, some relapses occur at the tumour margin or from occult metastases, and these tumour recurrences may not manifest until much later in the post treatment period. Consequently, there is no substitute for close clinical and imaging surveillance.

The first post treatment baseline MRI or CT examination following chemo(radiotherapy) or surgery is usually performed between 4 and 8 weeks after treatment (Liauw et al. 2006; Lell et al. 2000; Som et al. 1993), although in those patients with a low risk of residual disease this may be postponed to 4–6 months. Patients at a greater risk of locoregional relapse, including those with advanced disease treated by chemoradiotherapy, should undergo regular follow-up imaging. Most regional relapses/recurrences occur within the first 3 years of treatment, with the majority being diagnosed in the first 2 years. Imaging follow-up is therefore usually performed every 4–6 months for 2–3 years (Lell et al. 2000; Som et al. 1993), being more frequent in the earlier post treatment period. MRI or CT is the main technique for surveillance although some centres now include FDG-PET, either routinely as a post treatment baseline or more selectively for patients with a suspicious residual node on conventional imaging. US-FNAC also has a role in post treatment surveillance and has been specifically investigated for the early detection of occult nodal metastases in patients with an N0 neck (van den Brekel et al. 1999) who have undergone surgery at the primary site without an elective neck dissection. A watchful waiting strategy involving regular US follow-up, including every 8–10 weeks in the first year, has been shown to detect nodal recurrences early, thereby increasing the chances of performing salvage surgery (van den Brekel et al. 1999).

9 Brief Overview of Non-HNSCC Lymphadenopathies

9.1 Lymphoma

Cervical disease in Hodgkin’s (HL) and non-Hodgkin’s lymphomas (NHL) is described in “Neck Lymphoma”. Neck HL nearly always presents as nodal involvement. In general about 50% of NHL in the head and neck shows neck node involvement (Chisin and Weber 1994; King et al. 2004d; Hanna et al. 1997), the incidence being greater in diffuse large cell lymphomas than Natural Killer/T cell lymphomas. The size criteria for the diagnosis of lymphomatous nodes are not well established in comparison to those for carcinoma. In general cut-offs between 1 and 1.5 cm are used although multiple smaller nodes are also considered suspicious (Cheson et al. 1999). Lymphomatous nodes show a spectrum of morphological imaging appearances. At one end of the spectrum, lymphomatous nodes have similar features to nodal metastases from HNSCC, including a round shape, loss of the hilum and in NHL necrosis, matting and extra capsular spread sometimes may be present (King et al. 2004d; Saito et al. 2001; Wang et al. 1999). At the other end of the spectrum, lymphomatous nodes have similar features to normal or reactive nodes, namely an oval shape, preserved hilum and hilar vascularity which may be florid. The hilum is frequently identified, even in markedly enlarged rounded nodes, although it may be compressed and eccentrically displaced (Fig. 15). Calcification prior to therapy is rare and described mainly in mediastinal rather than cervical nodes (Apter et al. 2002). DWI is especially promising for lymphoma detection as involved nodes have much lower ADC than SCC (Abdel Razek et al. 2006; Sumi et al. 2003; King et al. 2007) as well as normal nodes (Kwee et al. 2010). FDG-PET also shows promise in detecting disease in normal sized lymph nodes, which could influence radiotherapy planning. Both HL and NHL are FDG avid, and aggressive lymphomas tend to have higher SUV values than indolent lymphomas (Hicks et al. 2005). FDG-PET or FDG-PET/CT has been shown to be comparable or superior to morphological imaging techniques as well as gallium67 scintigraphy for detecting nodal lymphoma. In centres where PET/CT is available, it is increasingly becoming the main imaging investigation for staging and therapeutic monitoring (Collins 2006).

Fig. 15
figure 15

Axial T1 W post contrast MRI showing a typical appearance of nodal lymphoma. There is gross lymphadenopathy with multiple enlarged rounded nodes, most of which appear solid although a right level II node demonstrates necrosis (arrow). Hila can be preserved even in grossly enlarged lymphomatous nodes (arrowheads)

Full size image

9.2 Other Head and Neck Carcinomas

  1. (1)

    Thyroid carcinoma: Thyroid cancer commonly spreads to cervical lymph nodes (see “Thyroid and Parathyroid Neoplasms”). Papillary carcinoma is the most common thyroid cancer and it has the highest incidence of nodal metastases (30–90%), followed by medullary carcinoma (50%), anaplastic carcinoma (40%) and follicular carcinoma (10%). Medullary carcinoma has a greater propensity to spread bilaterally and to the mediastinum (Machens et al. 2002). See Table 1 for routes of nodal spread and Table 2 for nodal staging.

    Ultrasound is the method of choice for imaging nodal metastases from thyroid cancer in the neck. In addition to the general features of metastatic nodes described for HNSCC, nodal metastases from papillary carcinoma may show micro- and macrocalcifications, increased echogenicity compared to muscle, and cystic change (Fig. 16) (Ahuja et al. 1995; Rosario et al. 2005). Calcifications may also be detected by CT, and cystic change may be seen by both CT and MRI; in the latter modality the cystic area may have a characteristic high T1 W signal intensity (Som et al. 1994; King 2008). The presence of unexplained nodes with calcification on CT or high T1-W signal cystic regions on MRI should prompt the search for a primary thyroid papillary carcinoma. Metastatic nodes in the central compartment tend to be small and more difficult to detect or characterise than those in the lateral compartment. Nonetheless, special effort should be made to try to evaluate these, especially those in the lower paratracheal/paraoesphageal regions, as failure to resect these may be responsible for so-called thyroid bed tumour recurrences. Nodal metastases from papillary carcinoma are frequently smaller than 5 mm, for which FNAC may be technically difficult to perform. Fortunately, detection of these small nodes is less of a concern for surgical planning because they can be treated successfully by radioactive iodine post operatively (Wang et al. 1999; Lell et al. 2000). Iodine scintigraphy is used following surgery to identify residual nodal metastases in the neck and nodal metastases in the mediastinum, but it will miss iodine negative nodes (this tendency is greater in the less well differentiated cancers). These iodine negative nodes may sometimes be detectable using (Hyde et al. 2003) F-FDG-PET/CT.

    Fig. 16
    figure 16

    Grey scale US showing nodal metastases from thyroid papillary carcinoma. The nodes are hyperechoic, with one showing an anechoic cystic component (arrows)

    Full size image
  2. (2)

    Salivary gland carcinoma: The majority of nodal metastases from salivary malignancies arise from primaries in the submandibular gland (42%), followed by parotid gland (25%) and oral cavity (9%) (Terhaard et al. 2004; Lloyd et al. 2010). Similar halves overall survival, independent of T stage. The frequency of nodal metastases is related to the underlying histology, grade and T stage; in this respect many of the subtypes have a high risk whereas adenoid cystic carcinoma uncommonly involves lymph nodes (Ferlito et al. 2001). Routes of spread are shown in Table 1. Conventional imaging criteria for nodal metastases in salivary cancer are similar to those used for HNSCC. In general, salivary cancers are relatively radioresistant, hence documented nodal metastases are treated by neck dissection (Stennert et al. 2003). The incidence of occult nodal metastasis in salivary cancer ranges widely, between 12 and 48% (Stennert et al. 2003; Armstrong et al. 1992; Regis De Brito Santos et al. 2001; Rodriguez-Cuevas et al. 1995), and not surprisingly the optimal management of clinical N0 disease is debated (Stennert et al. 2003; Regis De Brito Santos et al. 2001; Frankenthaler et al. 1993; Medina 1998; Korkmaz et al. 2002). In practice, parotid and submandibular nodes are usually resected along with the primary tumour specimen, and these may undergo intraoperative frozen section histological analysis to determine whether to proceed to a comprehensive neck dissection.

  3. (3)

    Nasopharyngeal carcinoma: Nasopharyngeal carcinoma commonly spreads to regional nodes and cervical lymphadenopathy may be the presenting feature (see also “Neoplasms of the Nasopharynx”). Nodal staging differs from that of HNSCC (Table 2) but the imaging criteria for nodal metastases are similar to those used for HNSCC. Lateral retropharyngeal nodes are common in this disease and are best identified by MRI (Fig. 3a). Routes of nodal spread are shown in Table 1. The presence of metastatic nodes in the lower neck or supraclavicular fossa increases the risk of distant metastases. With respect to post RT surveillance, it is common for benign nodes that have been spared from the radiation field (submental, paratracheal and suprasternal nodes) to become chronically enlarged due to altered lymphatic drainage.

  4. (4)

    Skin cancer: Skin cancer of the head and neck involves the superficial nodal groups more commonly than in HNSCC (Table 1). The appearance of metastatic nodes from squamous cell carcinoma of the skin are similar to those of other HNSCC whereas nodal metastases from malignant melanoma tend to show less necrosis and extracapsular spread (van den Brekel et al. 1998b), and may have a high signal intensity on T1 W MR imaging due the paramagnetic effects of melanin. For malignant melanoma, the risk of developing nodal metastases is related to lesion thickness, with melanomas <1 mm thickness having a rate of nodal metastases of <5% while >4 mm thickness increases the risk to about 35% (McMasters and Swetter 2003).

9.3 Non-malignant Lymphadenopathy

Conventional imaging criteria using size, morphology or FDG uptake to identify metastatic lymphadenopathy are based on studies that predominantly compare metastatic nodes with normal and reactive lymph nodes. However, these same imaging criteria are much less accurate for differentiating metastatic nodes from other nodal pathologies. As a consequence, biopsy is almost invariably required for diagnosis. A detailed account of benign cervical lymphadenopathies is beyond the scope of this chapter although tuberculous lymphadenitis merits special attention as it is one of commonest causes of persistent cervical lymphadenopathy in the developing world as well as in parts of the developed world. Tuberculous nodes may be discrete, matted or form confluent masses (King et al. 1999) and may show calcification. Other imaging findings include nodal enlargement, necrosis, abscess formation, peri-adenitis and increased FDG uptake; all of these overlap with nodal metastases on imaging (Moon et al. 1997). Tuberculous nodes can also fistulate to the skin, which is highly suggestive of this diagnosis. Several non-infectious cervical lymphadenopathies may overlap with malignancy on imaging including Castleman’s disease (Fig. 17), sarcoidosis, Rosai-Dorfman disease and Kikuchi-Fujimoto disease (Koyama et al. 2004; La Barge et al. 2008; Fulcher 1993; Na et al. 1997b).

Fig. 17
figure 17

Axial CT post contrast showing an enlarged homogeneous left level II lymph node. This was diagnosed as Castleman's disease

Full size image

10 Conclusion

Imaging of cervical nodal metastases has an important role in nodal detection, treatment planning and post treatment assessment. Imaging often requires a multimodality approach for the best results. A summary of the main points to be remembered when assessing nodal metastases are listed in Table 3.

Table 3 Summary of main points in the evaluation of cervical nodal metastases
Full size table