Imaging of extranodal extension: why is it important in head and neck cancer?

Abstract

The assessment of extranodal extension by imaging (iENE) is important in the management of patients with metastatic nodes from head and neck cancer (HNC). Over many decades iENE has guided surgical and radiotherapy planning using both early iENE, to detect small metastatic nodes that otherwise appear normal on imaging, and late iENE, to detect invasion into adjacent structures. Currently, new prognostic indicators of iENE are under a spotlight for treatment planning, prognosis and staging. This is aided by ongoing international interdisciplinary consensus building on iENE grading for prognosis and by the advantages of imaging over pathological assessment (pENE), which include coverage of all nodal groups in all patients irrespective of age, cancer stage or treatment. These advantages are especially relevant to those cancers treated primarily by (chemo)radiotherapy, such as viral-related human papillomavirus (HPV)-positive oropharyngeal squamous cell carcinomas and Epstein–Barr virus-positive nasopharyngeal carcinomas.

Evidence for the role of iENE as an indicator of poor prognosis is accumulating. Inclusion of iENE in the N category of the upcoming American Joint Committee on Cancer (AJCC) and the International Union Against Cancer (UICC) HNC staging guidelines is under consideration and has already been included in the latest AJCC ninth version of cancer staging guideline for nasopharyngeal carcinoma. New indications for iENE in treatment planning are under evaluation in areas such as the selection of HPV-positive oropharyngeal squamous cell carcinoma for treatment de-escalation. However, there are many gaps in our knowledge of iENE, which are a concern in multidisciplinary meetings because of their potential effect on decisions regarding disease management. New indications for iENE in the management of HNC patients are an exciting advance but more research is needed for it to reach its full potential.

Highlights

• •iENE identifies metastatic nodes and the extent guides surgical and radiotherapy planning.
• •iENE advantages over pENE include assessment of all nodal groups in all patients irrespective of cancer stage or treatment.
• •New roles for iENE in treatment planning and staging of viral-related cancers treated by (chemo)radiotherapy have emerged.
• •Important knowledge gaps need to be addressed, especially the impact of earlier grades iENE on management and staging.
• •International interdisciplinary consensus and education on iENE grading systems facilitates research and management.

Introduction

The detection of extranodal extension by imaging (iENE) is one of the imaging criteria used to diagnose metastatic nodes from head and neck cancer (HNC) and the extent of iENE is used to map invasion of adjacent tissues. Currently iENE is under a spotlight because of its new indications for treatment planning, prognosis and staging in cancers treated with (chemo)radiotherapy, where pathological assessment (pENE) after surgery is not available. This is related primarily to viral-related cancers, namely human papillomavirus (HPV)-positive oropharyngeal squamous cell carcinoma (OPSCC) and Epstein–Barr virus (EBV)-positive nasopharyngeal carcinoma (NPC). This article reviews the diagnostic performance of iENE compared with pENE, highlighting the advantages and disadvantages of both techniques for detecting ENE, and the current and potential roles of iENE for treatment planning and cancer staging. It also highlights gaps in our knowledge of iENE, which is a concern in multidisciplinary meetings because of the potential effect on decisions regarding disease management.

Imaging ENE: criteria and imaging modalities

The presence of iENE is a well-established criterion for diagnosing metastatic nodes derived from HNC and for mapping extension into adjacent tissues, to guide treatment planning. Over many decades, a range of iENE features on computed tomography (CT) and magnetic resonance imaging (MRI) have been described.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40 These features cover a wide spectrum from the earliest signs of thickening of the capsule and irregularity of the nodal margin to invasion of perinodal fat, invasion of adjacent nodes (known as matting, coalescent or conglomerate nodes), to the most advanced form of invasion into adjacent structures (muscle, skin, neurovascular structures or glands). Recently, an international group of oncologists, radiotherapists, surgeons and radiologists published a consensus standardising iENE into four grades: grade 0 (no iENE or irregular margins), grade 1 (perinodal fat invasion), grade 2 (nodal matting) and grade 3 (advanced invasion into adjacent structures) (Table 1).⁴¹ The spectrum of iENE on CT and MRI and grading are illustrated in Figure 1. Even more recently, a task force of radiologists from the USA, Europe and Asia (The AOSHNHR-ASHNR-ESHNR Joint Task Force Guide to iENE) have produced educational materials to improve agreement across centres and continents, for the unequivocal designation of iENE for N categorisation in cancer staging (available on the websites of these organisations). To avoid false-positive findings iENE criteria are not applied to metastatic nodes that have undergone fine needle aspiration cytology (FNAC)/biopsy, treatment or those that are infected.

Table 1.

Proposed international grading system for extranodal extension by imaging (iENE)⁴¹

Grade 0	Grade 1	Grade 2	Grade 3
None of the radiologic features of iENE	Clearly irregular or ill-defined nodal margins AND extension into and confined to perinodal fat	Clear invasion through two or more inseparable adjoining nodes, also known as conglomerate, matted or coalescent nodes, with/without grade 1 features	Clear invasion into adjacent structures, such as muscle, skin, neurovascular structures, or glands, with/without grade 1 or 2 features

iENE, imaging extranodal extension.

Figure 1.

Contrast-enhanced T1-weighted MRI (top row) (A-E) and CT images (bottom row) (F-J) of metastatic nodes (open arrows) showing the spectrum of iENE and the grading system (grades 0-3) proposed by Henson et al.⁴¹ (A) and (F) without iENE (grade 0 iENE); (B) and (G) with irregularity of the nodal margin without perinodal fat iENE (grade 0 iENE); (C) and (H) with perinodal fat iENE (arrow head) (grade 1 iENE); (D) and (I) with matting iENE (also known as coalescent or conglomerate nodes) (grade 2 iENE); (E) and (J) with advanced iENE into adjacent structures: muscle (arrow head), skin (solid arrow), and obliteration of the internal jugular vein (normal internal jugular vein on the opposite side, curved arrow, for comparison) (grade 3 iENE).

The choice between CT and MRI for nodal assessment, and hence iENE assessment, is dependent on the modality chosen to map the primary tumour and on the available resources and expertise. For both imaging modalities, contrast-enhanced images and multiplanar imaging are preferred. Multiplanar imaging includes volume acquisition with multiplanar reconstruction to search for iENE in all directions. The best imaging modality for iENE is contentious, but in general, MRI is preferred to CT because of the excellent contrast afforded by the multiple sequences, including T2- and T1-weighted postcontrast sequences with fat suppression. MRI aids in the detection of early iENE and in the distinction between nodes abutting or invading adjacent nodes and structures. Features of iENE on ultrasound and [¹⁸F]-fluorodeoxyglucose (FDG) positron emission tomography (PET) images are reported less frequently. Ultrasound^3,10 relies on similar morphological features to those described for CT and MRI, whereas a high maximum standardized uptake value on FDG PET predicts pENE.^16,42,43

Imaging ENE: diagnostic performance

Diagnostic performance compared with pathology

For more than three decades, studies have compared the diagnostic performance of iENE with pENE obtained from neck dissection specimens following surgery.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40 The diagnostic performances of iENE from 25 studies that reported sensitivity and specificity using CT, MRI, or CT/MRI, based on the presence of one or more features to designate iENE or pENE positivity, are summarised in Table 2.^1,2,4,6,11, 12, 13, 14^,16, 17, 18, 19, 20, 21, 22, 23^,26, 27, 28, 29^,34,36, 37, 38^,44 It is difficult to draw conclusions because the reported performance of iENE shows wide variations for both sensitivity (CT, 27% to 100%, median = 63%; MRI, 40% to 78%, median = 66%) and specificity (CT, 53% to 98%, median = 81%; MRI, 83% to 100%, median = 94%). Four meta-analyses^43,45, 46, 47 narrowed the reported variation of iENE performance, showing CT and MRI sensitivities ranging from 63% to 77% and 60% to 85%, respectively, and specificities from 77% to 85% and 78% to 96%, respectively (Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2025.105519). Two studies compared iENE positivity (any degree of iENE) with microscopic (≤2 mm) versus macroscopic (>2 mm) pENE and showed a significant reduction in iENE sensitivity for detecting microscopic pENE,^32,36 but without a reduction in specificity.³²

Table 2.

Diagnostic performance of iENE on CT and/or MRI, compared with pENE in patients with head and neck SCC

Year	Authors, Journal	HPV/P16 Status	Sensitivity (%)		Specificity (%)
			CT	MRI	CT	MRI
1991	Carvalho et al. Clin. Radiol.1	—	63	—	60	—
2000	Hao et al. Otol. Head Neck Surg.2	—	—	73	—	95
2004	King et al. Eur. J. Radiol.4	—	65	78	93	86
2009	Souter et al. J. Laryngol. Otol.6	—	66/80a	—	91/90a	—
2013	Chai et al. JAMA Otolaryngol. Head Neck Surg.11	Negative and positive	49/65a	—	84/54a	—
2013	Lodder et al. Oral Oncol.12	—	—	60	—	93
2013	Url et al. Eur. J. Radiol.13	—	71/76a	—	91/91a	—
2014	Prabhu et al. Inter. J. Radi. Onco. Bio. Phy.14	Negative	44	—	98	—
2015	Lee et al. Ann. Surg. Oncol.16	—	73	—	92	—
2015	Maxwell et al. Laryngoscope17	Positive	55/47a	—	70/85a	—
2016	Liu et al. Oral Oncol.18	—	64	—	72	—
2017	Carlton et al. Neuroradiol. J.19	Negative and positive	57/66a	—	81/76a	—
2017	Geltzeiler et al. Oral Oncol.20	Positive	64	—	68	—
2017	Moreno et al. J. Laryngol. Otol.21	—	—	43	—	100
2017	Sharma et al. Indian J. Otol. Head Neck Surg.22	—	100	—	58	—
2018	Almulla et al. Int. J. Radiat. Oncol. Biol. Phys.23	Negative	61	40	95	97
2018	Patel MR et al. ORL J. Otorhinolaryngol. Relat. Spec.26	Positive	88/100a	—	53/63a	—
2019	Lee et al. Korean J. Radiol.27	Positive	62		78
2019	Noor et al. J. Med. Imaging Radiat. Oncol.28	Positive	57/61a	—	73/67a	—
2020	Ariji et al. Oral Radiol.29	Negative	55	—	71	—
2021	Kowalchuk et al. Oral Oncol.44	Positive	54	—	71	—
2023	Nemmour et al. Eur. Arch. Otorhinolaryngol.34	Negative and positive	62	72	94	83
2024	Duguet-Armad et al. Laryngoscope36	Negative	27	—	96	—
2024	Hancioglu et al. J. Comput. Assist. Tomogr.37	Negative	39	—	82	—
2025	Liao et al. Head Neck38	Negative	68		90

CT, computed tomography; HPV, human papillomavirus; iENE, imaging extranodal extension; MRI, magnetic resonance imaging; pENE, pathological extranodal extension.

^aStudy with two observers, the performance of each observer is shown and analysed separately.

It is even more difficult to draw conclusions regarding the performance of individual iENE features (capsule/nodal margin abnormalities, perinodal fat invasion, matting and adjacent structure invasion),^{5,10,12,14,15,19,20,24,26,30,32,}35, 36, 37^,39 as comparisons are rarely made with similar features on pENE (i.e. ‘like-for-like’). Instead, individual iENE features tend to be compared with pENE positivity (presence of any evidence of pENE); therefore, unsurprisingly, advanced iENE has the lowest sensitivity (least frequent pattern), although specificity remains relatively high across all grades of iENE.

When comparing the performance of CT and MRI, two of the four meta-analyses noted above^43,45, 46, 47 found that MRI was significantly more sensitive than CT,^43,45 while no difference was found for specificity.^43,45, 46, 47 However, few studies have directly compared iENE on both imaging modalities in the same patient and for the same node. One such study by our group in the pre-HPV era⁴ showed higher sensitivity, but lower specificity, for MRI compared with CT when both modalities were carried out in the same patients and with node-by-node radiological–pathological correlation, but the differences were not statistically significant.

Factors contributing to the wide variation in the diagnostic performance of iENE

Many factors have contributed to the wide range in performance of iENE for detecting pENE. Studies have used differing criteria to indicate iENE. The studies listed in Table 2 used one or more established criteria for iENE positivity [irregular nodal margin, perinodal fat invasion, nodal matting or advanced invasion (muscle, skin, neurovascular or glandular structures)], but the number/combinations of criteria used in each study varied. Interobserver disagreements are another contributor to variability in the reported performance for designating positive or negative iENE results by CT^{4,6,13,15,17,19,23,}26, 27, 28^,36,37,48 and MRI,^4,23,27,48 which extends also to the selection of nodes for assessment.³⁵ Assessment is subjective, requiring experience to detect subtle iENE and to make decisions in equivocal cases. In general, high intra-rater and inter-rater concordance for iENE positivity can be achieved by expert raters in both HPV-negative²³ and HPV-positive SCC,⁴⁸ although lower levels of agreement have been reported, especially in HPV-positive disease.⁴⁰ For individual iENE criteria agreements are the lowest for nodal border irregularity (median kappa coefficient = 0.36).^{19,30,35,37,39} The assessment of indistinct or irregular margins on images is also hampered by artifacts, such as pulsation from adjacent vessels.⁴ There is also considerable disagreement about nodal matting (median kappa coefficient = 0.50)^{19,30,35,39,48} with regard to variability in both the number of nodes (two or three) and the features that distinguish matted nodes from ‘touching’ nodes. Better agreement has been reported for perinodal fat invasion (median kappa coefficient = 0.615)^{15,26,30,35,37,39} and advanced structure invasion (median kappa coefficient = 0.62).^19,37,39,48 Other factors contributing to discrepancies in iENE performance between studies include variations in (i) sample size, (ii) imaging protocols, (iii) nodal selection, (iv) time between imaging and surgery,²³ and (v) the meticulousness with which radiological–pathological correlation is performed.

However, the wide range in the diagnostic performance of iENE can be attributed not only to the shortcomings in radiological assessment but also to those related to the pathological assessment. There is a lack of consensus on a standard for reporting pENE, which varies between microscopic and macroscopic (often subjective) and between grading systems, such as those proposed by Lewis et al., and Yamada et al.^49,50 Interobserver agreement for pENE between pathologists is reported to be low.^51,52 Moreover, pathological assessment has several important disadvantages compared with imaging. First, pathological data are limited to a select group of patients with operable disease, often excluding those with advanced forms of pENE. Second, the extent of the surgical specimen and the number of nodes examined are variable and may be incomplete. Third, the meticulousness of the assessment varies and is limited by the sample preparation, which includes a single plane. A retrospective review of the samples within and across centres is cumbersome and time-consuming; therefore, research tends to be based on retrospective reviews of written reports. The reliance on retrospective assessments of pathologic reports means that very few radiological–pathological correlations are performed on a node-by-node or even nodal group-by-nodal group basis, often relying instead on hemi-neck or bilateral neck data. On the other hand, imaging has the advantage of being able to be performed in all patients, irrespective of age, comorbidity, the stage of cancer or treatment. Moreover, all nodes in the head and neck are included in the assessment, scans can be reconstructed into multiple planes and a retrospective review of the images is easier to facilitate. The relative advantages and disadvantages of radiological and pathological ENE assessment are summarised in Table 3.

Table 3.

Advantages and disadvantages of radiological and pathological assessments of ENE

	Radiological assessment	Pathological assessment
1. Patient selection	Performed in all patients, irrespective of age, coexisting morbidity, cancer stage or treatment	Performed in surgical candidates with operable disease
2. Nodal coverage	All nodes/nodal groups	Variability in the amount of tissue available and the number of nodes sampled
3. Technique	Variability in imaging modalities and protocols, including slice thickness	Variability in slide preparation for histological assessment, including slice thickness and orientation
4. Selection of nodes for ENE assessment	Variability in the selection of nodes: all nodes, largest node or node with the most extensive iENE	Variability in the selection of nodes: all nodes, largest node or node with the most extensive pENE
5. Analysis	Multiplanar reconstruction of any node	Confined to the plane and node/amount of the node in the prepared sample
6. Criteria/grades of ENE	Variability in the selection criteria: single criterion, two or more criteria, combinations of criteria, likelihood scores and observer impression. Criteria diluted by non-iENE criteria such as necrosis	Variability in the selection criteria: microscopic versus macroscopic, direct measurements or grading one criterion, two or more criteria and combinations of criteria
7. Observer variation	Intra- and inter-observer variation	Intra- and inter-observer variation
8. Review of data source	Review of images is readily available within and across centres	Review of histology samples may not be readily available, especially across centres and for large sample sizes; therefore, studies frequently rely on retrospective reviews of reports
9. Radiological–pathological correlation	1.Wide variations: node-by-node; nodal-group-by-nodal-group, unilateral-neck-by-unilateral-neck, neck-by-neck 2.Delay between imaging and surgery (>8 weeks) may lead to the progression of ENE

ENE, extranodal extension; iENE, imaging extranodal extension; pENE, pathological extranodal extension.

Imaging ENE: treatment planning

Background of iENE in treatment planning

For the purpose of treatment planning, it is usually desirable to select iENE features with high sensitivity (fewer false-negative results). This approach reduces the risk of patients missing out on optimal treatment by, for example, identifying suspicious nodes for FNAC and enabling the avoidance of triple therapy or inappropriate treatment de-escalation. These aspects are discussed in the following sections, but there are many unanswered questions regarding the effects of iENE on treatment and management, especially regarding early and mid-iENE grades, for which research is lacking.

Surgical planning

Early iENE

iENE is a criterion for diagnosing a metastatic node. In practice, most metastatic nodes with iENE on CT or MRI already have other features that identify them as malignant. Notably, there is a strong link between the presence of necrosis and pENE,^{8,12,15,30,35} and the incidence of pENE tends to increase with increasing nodal size.^11,12,14,28,53, 54, 55 However, pENE arises in metastatic nodes as small as 6 mm,²⁸ and 18% to 22.8% of metastatic nodes of <1 cm.^55,56 Early iENE in small metastatic nodes where iENE is the only criterion to identify the node as being metastatic may therefore influence FNAC/surgical planning. This may make the difference between an N0 and N+ neck (e.g. wait-and-see versus surgery approach) or the extent of surgery (e.g. selective versus modified radical neck dissection; unilateral versus bilateral neck dissection). Of note, FNAC is a low-risk procedure, in which even imaging features of impending iENE (irregularity in the nodal contour/margin), currently designated as grade 0 in the grading system because of concerns with subjectivity,⁴¹ are nevertheless useful to increase sensitivity.

Advanced iENE

At the other end of the spectrum, mapping the extent of advanced iENE is crucial for surgical planning and operability. Common sites of invasion in advanced iENE are muscle, skin, neurovascular and glandular structures, but depending on the location of the node, a range of other structures are also at risk, which also may influence the surgical approach and operability. The structures affected by pENE and the surgical implications are briefly summarised in Table 4. A more in-depth review has been reported by Hiyama et al.⁵⁷ Nodes with iENE may also merge with local spread from the primary tumour, especially those tumours arising in the hypopharynx and oral cavity. It is worth remembering that nodes, regardless of the presence or absence of iENE, may be unresectable because of their location. Such nodes tend to include those located medial to the carotid vessels (a notable exception is resectable primary hypopharyngeal cancer) and those located superiorly, either junctional or retropharyngeal nodes, which constitute a high surgical complexity and potential morbidity to remove, such that resection is only performed in highly specialised centres.^58,59

Table 4.

Impact on surgery of advanced iENE to map invasion into adjacent structures

Structures	Resectable	Unresectable (operability may vary between centres)
Muscle	•Sternocleidomastoid (degree of invasion for surgical planning) •Other muscles: masseter, medial pterygoid, mylohyoid, strap, digastric - posterior and anterior bellies	•Prevertebral •Paravertebral (early invasion may be resectable) •Anterior scalene (proximity to brachial plexus and phrenic nerve)
Skin, platysma and deep facial planes	•Skin and platysma (flap reconstruction)	•Deep fascial planes (prevertebral is usually unresectable, paravertebral more controversial)
Vessels	•Carotid arteries: external and major branches •Internal jugular vein (obliteration or tumour thrombus, other factors that may be considered include loss of surrounding fat planes, compression and deformity; IJV reconstruction required if bilateral)	•Carotid arteries: common, bifurcation and internal (tumour contact >270°, other factors may be considered, including the length of craniocaudal contact) •Vertebral (resectability guided by site and length of craniocaudal contact) •Subclavian and brachiocephalic arteries (and veins)
Nerves	•Cranial nerves: lingual (trigeminal V3 branch), marginal mandibular (facial VII branch), vagus (X) and recurrent laryngeal branch (tracheo-oesophageal groove), spinal accessory (XI) and hypoglossal (XII) •Phrenic and sympathetic (Neural involvement inferred from iENE at expected location of the nerve, secondary muscle denervation, perineural spread)	•Brachial plexus •Cranial nerves, phrenic and sympathetic [Nerve involvement may also be unresectable; for example, perineural spread proximally (e.g. hypoglossal and lingual); close relationship to the carotid sheath, between carotid vessels and IJV and posterior to the IJV involving the deep fascia; unacceptable morbidity from nerve palsy]
Bone	•Mandible (periosteal/segmental mandibulectomy ± reconstruction) •Hyoid (major morbidity to swallowing function)	•Cervical spine
Aerodigestive tract	•Floor of the mouth •Larynx and/or pharynx (laryngectomy and/or pharyngectomy)	•Oesophagus and trachea (reconstruction may be possible depending on the cancer type, degree of invasion and mediastinal extension)
Glands	•Parotid (especially from junctional node) •Submandibular (submandibulectomy, part of standard neck dissection) •Thyroid

ENE, extranodal extension; iENE, imaging extranodal extension; IJV, internal jugular vein.

Radiotherapy and chemoradiotherapy planning

Like its influence on surgical options, when iENE is the only criterion to identify the node as being metastatic, it may influence the need for FNAC, change the extent of the radiotherapy field or dose, or potentially up-classify the N category, leading to a change in treatment from radiotherapy to chemoradiotherapy. The presence of iENE influences the radiation dose applied to nodal groups (a higher dose for metastatic nodes with iENE versus those without iENE). Contouring metastatic nodes for radiotherapy planning is also influenced by the proximity of the iENE to structures that are more vulnerable to radiation injury, such as the salivary glands and brachial plexus.

Avoidance of triple therapy (surgery followed by chemoradiotherapy)

Many factors are considered when deciding between a surgical and nonsurgical treatment option, including the factors related to iENE described above. In addition, the preoperative detection of iENE can be used to avoid surgery in favour of proceeding straight to primary treatment with chemoradiotherapy. This is based on data from studies, including a landmark study by Bernier et al. in 2005,⁶⁰ which showed that the postsurgical discovery of pENE in neck dissection specimens (and/or positive primary tumour margins) is an adverse feature indicating risks of recurrence and poor survival, and requires post-operative chemoradiotherapy. This results in patients receiving triple therapy (surgery, chemotherapy and radiotherapy), which is associated with adverse effects of grade ≥3 in up to 77% of patients.⁶¹

De-escalation of treatment regimes in patients with HPV-positive SCC

HPV-positive OPSCC has a better prognosis than HPV-negative OPSCC,^62,63 related mainly to better locoregional control,64, 65, 66 but currently there is insufficient clinical trial phase III data to alter therapy based on HPV status67, 68, 69 and the treatment of patients with HPV-positive OPSCC is similar to those with HPV-negative OPSCC. Nevertheless, data from phase II trials is promising70, 71, 72, 73 and therapeutic approaches that de-escalate treatment to reduce chemoradiotherapy toxicity⁷⁴ are under scrutiny, especially in patients with early-stage HPV-positive OPSCC, including those with early nodal disease without ENE. One approach to reduce toxicity is to treat with radiotherapy alone or with a reduction in the radiation dose.^63,71,75 Another approach is to use transoral robotic surgery for resection of early primary tumours combined with neck dissection for nodal disease⁷⁶ that is limited in size and extent, with or without adjuvant radiotherapy. In these two approaches, de-escalation of treatment may not be appropriate if iENE is detected. However, many questions remain regarding the effect on the outcome of the presence/absence of ENE and the grade of ENE detected by either imaging or pathological assessment. Clinical trials that include ENE are ongoing (NCT02215265), and a recent trial showed a favourable outcome in patients with HPV-positive OPSCC with limited nodal disease (N1-2 disease) in whom the pENE was ≤1 mm.⁷² Moreover, resources that allow close post-treatment imaging surveillance could also influence treatment choices in the future, especially for patients with early grades of iENE. Gaps in our knowledge of the impact of iENE on outcomes, especially for earlier grades of iENE, currently cause concern in iENE discussions in multidisciplinary meetings that may affect management or post-treatment imaging strategies; therefore, more research is needed.

Imaging ENE: prognosis and staging

Imaging ENE: pathological basis for using iENE for prognosis

Interest in the prognostic value of iENE is based on evidence that has accumulated over many decades from pathological studies, especially in patients with HPV-negative SCC, showing that pENE is associated with poor outcomes. A meta-analysis by Dunne et al.⁷⁷ concluded there was a ∼50% decrease in 5-year overall survival (OS) in patients with pENE compared with those without pENE, and a meta-analysis by Mermod et al.⁷⁸ found that this poor outcome was related to distant metastases, rather than nodal recurrence. The eighth edition of the AJCC/UICC guidelines for staging HPV-negative disease and other HNCs (salivary gland carcinoma, unknown primary and cutaneous SCC) includes pENE in the N category, irrespective of micro- or macroscopic disease, based on the size and number of nodes with pENE (N2a ENE in ipsilateral nodes <3 cm; N3b: ENE nodes >3 cm or multiple nodes). pENE was not incorporated into the staging classifications for HPV-positive OPSCC because the evidence at that time was equivocal,^78,79 although evidence is now mounting for the association of pENE with decreased OS and distant recurrence, supported by a meta-analysis by Benchetrit et al.⁸⁰

The stratification of HNC patients with pENE into high-risk groups is also related to the extent of the pENE.⁸¹ Advanced forms of macroscopic pENE, particularly extension into adjacent structures and complete loss of nodal architecture (soft tissue metastasis), are associated with the worst outcome^49,82,83 while microscopic pENE has little effect.83, 84, 85 The effect of macroscopic grades between these two extremes is less clear, and attempts to introduce grading systems, such as those proposed by Lewis et al.⁴⁹ and Yamada et al.,⁵⁰ have had limited success. Even the distinction between macroscopic and microscopic pENE can be imprecise, with proposed thresholds for HPV-negative SCC of 1.7 mm,⁸⁴ 1.9 mm,⁸⁵ 2 mm86, 87, 88 and 5 mm.⁸⁷

Imaging ENE: background for prognosis and staging

The limitations of radiological–pathological correlation, described earlier, contribute to the poor reputation of imaging for detecting pENE and, hence, to the slow adoption of iENE for prognosis. However, imaging clearly has advantages over pathology, especially as it is the only method for assessing ENE in the many patients with HNC who do not undergo surgery. It is, therefore, timely for the prognostic significance of iENE to be considered in its own right, rather than as a surrogate for pENE, and research in this area is well underway.

For staging using iENE, it is desirable to use grades with low subjectivity, as findings must be unequivocal, and grades with high specificity for N categorisation to prevent dilution of the performance of the cancer staging guidelines.⁸⁹ This contrasts with treatment planning, which favours iENE grades with high sensitivity and where equivocal findings may still contribute to patient management decisions. These requisites tend to favour the advanced grade (grade 3; invasion into adjacent structures), which is also the grade most closely related to clinical ENE (involvement of skin or nerves, or soft tissue fixation) in the eighth edition of the AJCC/UICC cancer staging guidelines for staging HPV-negative SCC. Moreover, iENE should not only be significantly associated with recurrence or survival, but its addition to the N category should improve the prognostic performance of the staging classification. Research into the prognostic value of iENE has been most active in NPC and OPSCC, and iENE has already been introduced into the N categorisation in the ninth version of the AJCC cancer staging guidelines for NPC and is likely to be introduced for other HNCs, including HPV-positive OPSCC, in the near future. Research into the prognostic value of iENE in the three major groups of HNCs, namely HPV-negative SCC, HPV-positive OPSCC and NPC, is briefly summarised in the following sections.

Imaging ENE: HPV-negative SCC prognosis and staging

In the eighth edition of the AJCC/UICC cancer staging guidelines, the assessment of ENE in patients with HPV-negative SCC treated with (chemo)radiotherapy, rather than surgery, is reliant on clinical examination, with iENE playing only a supportive role. This situation proves problematic when advanced iENE is detected but is not found on clinical examination. Hence, there is a desire to include iENE in future staging classifications. Seven prognostic iENE studies^18,32,36,90, 91, 92, 93 have been reported on CT or MRI, with the number of patients ranging from 18 to 406 (median, 90) and using a range of iENE criteria or combinations of criteria. They showed associations between iENE and poor OS^18,32,36,90, 91, 92, 93 or disease-free survival (DFS)^18,90,91,93 and links with locoregional recurrence^18,91,93 and the development of distant metastases.⁹² One study found that using iENE positivity (one or more of the following: ill-defined margins, coalescent or advanced to muscle/skin/carotid vessels) as a criterion to up-classify the N category, improved prognostic performance of the eighth edition of the AJCC/UICC cancer staging guidelines.⁹¹

Imaging ENE: HPV-positive SCC prognosis and staging

In the eighth edition of the AJCC/UICC cancer staging guidelines, ENE assessment is not included for patients with HPV-positive OPSCC, either pathologically, clinically or by imaging. As most patients with this cancer are treated with (chemo)radiotherapy, there is a push to include iENE in the next versions of the staging classifications. Thirteen prognostic iENE studies in HPV-positive SCC with significant findings^18,27,44,48,92, 93, 94, 95, 96, 97, 98, 99, 100 have been reported on CT or MRI, with the number of patients ranging from 22 to 517 (median, 233) and using a range of iENE criteria or combinations of criteria. These studies showed associations between iENE and poor OS,^{18,48,92,95,97,99,100} DFS/disease-specific survival^{48,93,94,97,99,100} and links with locoregional recurrence^18,93 and the development of distant metastases.^92,94,97,100 A meta-analysis by Benchetrit et al.⁸⁰ confirmed that iENE is associated with reduced OS and distant recurrence, but not with locoregional recurrence. They also found that iENE had greater prognostic value than pENE. Two studies found that iENE positivity (ill-defined margins, coalescent or advanced to muscle/skin/carotid vessels) improved the performance of the eighth edition of the AJCC/UICC cancer staging guidelines.^48,100 One of these studies showed that iENE could be used to reclassify N1 to N2 and N2 to N3,⁴⁸ and the other showed an increase in the concordance index ≥0.05.¹⁰⁰

Imaging ENE: unknown/mixed HPV-status SCC prognosis and staging

Four studies of SCC with unknown or mixed HPV status confirmed the association of iENE with OS.101, 102, 103, 104 Three of these studies found an association with locoregional recurrence.^101,102,104

Imaging ENE: NPC prognosis and staging

Evidence for the prognostic role of iENE is strong for NPC, which is the first HNC for which iENE is included in the cancer staging classification guidelines. Advanced iENE with invasion into muscle/skin/neurovascular structures is now in the N3 category of the ninth version of the AJCC cancer staging guidelines. It is applicable only to cervical nodes and not to retropharyngeal nodes, in which invasion into the prevertebral muscles is common and the merging of the primary tumour and these metastatic nodes hampers the assessment of iENE.

NPC is treated with (chemo)radiotherapy, and in endemic areas, EBV-related NPC has many similarities with HPV-related OPSCC, with both cancers having a propensity for nodal spread with ENE and the development of distant metastases but good locoregional control. It has long been recognised that iENE is an important indicator of poor prognosis in NPC. A study by Mao et al. in 2008¹⁰⁵ reported the prognostic value of iENE positivity, and in the 2008 Chinese staging classification guidelines,¹⁰⁶ iENE was included in the N2 category based on the presence of any iENE feature to indicate iENE positivity. A study by Ai et al. from our group in 2019¹⁰⁷ examined the influence of the extent of iENE on outcomes in 546 patients with NPC. The results showed that an advanced grade of iENE (muscle, skin, neurovascular or glandular structures) was prognostic, and the addition of advanced iENE improved the performance of the eighth edition of the AJCC/UICC cancer staging guidelines by reclassifying N1 or N2 nodes with advanced iENE to N3. Since that time, 12 more prognostic iENE studies have been published, bringing the total to 14 MRI studies^105,107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119 involving 61-8334 patients (median, 1349) with long-term follow up. Most of these studies analysed outcomes in all patients (i.e. those without and with nodal metastases and/or iENE) using a range of iENE criteria or combinations of criteria, including matting, irregular margins or perinodal fat infiltration in most studies and advanced iENE into adjacent structures in all studies. These studies, including two large recent studies in 2024 by Du et al. (8334 patients)¹¹⁸ and Pan et al. (4914 patients),¹¹⁹ showed associations between iENE and poor OS^105,107, 108, 109^,111,113, 114, 115, 116, 117, 118, 119 and DFS^{98,105,109,115,116,118} and links to locoregional recurrence,^{107,108,111,116,119} with even stronger links to the development of distant metastases.^105,107,108,110, 111, 112, 113, 114, 115, 116^,119 Importantly, when advanced iENE is added to the N3 category, it consistently improves the performance of the eighth edition of the AJCC/UICC cancer staging guidelines.107, 108, 109^{,111,115,116,118,119} The strong link between advanced iENE and distant metastases also has the potential to select high-risk patients for closer post-treatment surveillance. Matting also shows promise as a prognostic factor,^108,109,115 but the evidence is not as strong as that for advanced ENE. This may be because matting iENE covers a wide spectrum from matting of two adjacent nodes, to extensive matting of multiple nodes with complete loss of nodal architecture (soft tissue metastasis). Further research is needed before matting is considered for inclusion in the N category for NPC.

Imaging ENE: future improvements in the performance of iENE and gaps in knowledge

Agreement between observers, both experts and nonexperts from different disciplines, is vital to facilitate consensus on iENE criteria for research and successful implementation into clinical practice and cancer staging guidelines. International consensus building on iENE grading systems⁴¹ and production of educational material are already a major step in the right direction, but there are hurdles to overcome. Grading systems and grade definitions may require refinements, including in controversial areas such as the definition of nodal matting, and despite the mounting research into iENE there are many gaps in our knowledge. These gaps open new avenues for future research, especially regarding the impact on treatment planning and prognostic significance of early and mid-grades of iENE. In the meantime, radiologists are often placed in an uncomfortable position when asked to make judgment calls about iENE in their reports that could alter patient management. Some of these knowledge gaps are listed below.

• •Capsular thickening, irregularity or indistinct margins without perinodal extension.

Currently, this feature is not included in the grading system because there are concerns about subjectivity, and based on pENE, the prognostic impact of this earliest sign is likely to be limited. However, this feature is used by radiologists as a criterion for diagnosis of small metastatic nodes (for example, as an indicator for FNAC) and so may require some consideration in future grading systems for treatment planning.

• •Perinodal fat iENE (grade 1)

Evidence is limited for the effect of this grade alone and whether the number or size of nodes with this grade contribute to prognosis. Perinodal fat extension is also a feature that may be more readily detected by MRI compared with CT. Furthermore, it is not known whether MRI can accurately distinguish between true perinodal pENE and inflammatory or stromal reactions and, if so, whether this affects treatment planning and prognosis.

• •Matting (conglomerate or coalescent nodes) iENE (grade 2)

This grade is the most challenging to define. Definitions vary regarding the number of nodes required to constitute matting and the features used to distinguish matted from adjacent ‘touching’ nodes, such as loss of the fat plane, deformation of the contour, loss of an acute angle between adjoining contours and disruption/loss of the intervening capsule, features which may also be influenced by the imaging modality. It is possible that the prognostic value for matting will not simply rely on positive or negative findings but rather on the extent of matting.

• •Advanced iENE (grade 3)

This grade applies to muscle, skin, neurovascular or glandular structures, but the definition of unequivocal invasion for staging may require future refinements. Further research also is needed to determine whether this grade should be (i) subdivided to reflect the differences in prognostic significance of different sites of iENE, (ii) expanded to include other structures such as glands, bone and mergence with the primary tumour, which may need to be cancer specific (i.e. mergence with the primary tumour in SCC tends to be included as grade 3 because it involves invasion of intervening muscle but is excluded from NPC).

Conclusions

The role of iENE in the assessment of metastatic nodes from HNC is expanding from diagnosis to prognosis. iENE continues to be widely used to plan surgical and radiotherapeutic treatment, using both early iENE to detect otherwise unsuspected small metastatic nodes and advanced iENE to map the extension into adjacent structures.

The adoption of iENE for prognosis and staging has been a slower process, although imaging is the only technique that includes all nodal groups in all patients, irrespective of their age, coexisting morbidity, cancer stage or treatment. This has been partly the result of concerns regarding the accuracy of iENE to detect pENE, lack of consensus on the qualitative assessment of iENE and a relative paucity of research in the past. However, the variations in reported accuracy of iENE for pENE detection are due to inherent limitations not only related to imaging assessment but also to pathological assessment. Internationally, interdisciplinary cooperation is now bearing fruit with respect to consensus building for iENE grading systems and the production of educational material to improve observer agreements. Evidence is mounting showing that iENE is an important predictor of poor survival in HPV-negative SCC, HPV-positive OPSCC and NPC, especially in relationship to the development of distant metastases. Advanced iENE is already incorporated into the N3 category of the ninth version of the AJCC cancer staging guideline for NPC, and iENE will likely be incorporated in the new AJCC/UICC cancer staging guidelines for other HNCs.

Despite these advances, future work is needed to refine the grading systems and precise definitions for each grade for both treatment planning and staging. Moreover, important gaps in knowledge need to be addressed including in areas causing concern and controversy in management, such as the influence of early and mid-grades of iENE in treatment planning for early-stage HPV-positive OPSCC. iENE has potential new benefits for HNC management, but more research is needed for it to reach its full potential.