Neck dissections: radical to conservative

Gasparo Aselli, in 1622, first described the lacteal vessels indicating the presence of a lymphatic system. Paolo Mascagni, in 1787, and Sappey, in 1875, published atlases on lymphatics. Most of the later knowledge was derived from autopsy or post-surgical studies on nodes. The standard anatomical depictions of nodes and their drainage were based on Rouviere's descriptions. Surgical removal of nodes has paralleled the anatomical understanding. From these in vitro autopsy or post surgical studies on lymph drainage, we have progressed to in vivo sentinel node evaluation.

In the early 1800s, complete removal of disease was considered impossible once cancer had spread to 'cervical glands', although Warren in 1847 described an operation for removal of metastatic neck nodes. Butlin advised removal of nodes through a Köcher incision. A systematic operative procedure for removal of cervical lymphatics and nodes, and the first description of radical neck dissection (RND) based on anatomical principles, was described by George Crile Sr., in 1906 [1]. Hayes Martin (the father of modern head and neck cancer surgery) and his associates standardized the technique of RND [2]. Both of these surgeons followed Halsted's principles which were popular at that time. The RND is still considered by some to be the 'gold standard' treatment for neck nodes. However, modifications of the procedure are accepted and are being performed with increasing frequency. Research into breast cancer has contributed a great deal to the understanding of cancer as a systemic disease in contrast to the earlier Halstedian principles that cancer followed an anatomical step-wise progress. Suarez proposed that muscles, vessels and nerves could be preserved without adversely affecting regional control and described 'functional neck dissection' [3]. With increasing knowledge of tumor biology, the next stage in the surgery of metastatic neck disease was the promotion of the functional neck dissection by Bocca [4]. A comparable recurrence rate compared with that of RND was also claimed [5]. Later studies have shown that there is no compromise on oncological effectiveness even though there is deviation from the previous standard en bloc resection.

Regional lymphatic drains from the scalp and skin of head and neck region, the mucosa of the upper aero-digestive tract, salivary glands, and the thyroid gland to specific regional lymph nodal groups. In addition, tumor dissemination via regional lymphatics to these lymph node groups occurs in a predictable and sequential fashion. Hence, specific regional lymph node groups should be appropriately addressed in treatment planning for a given primary tumor site. The anatomy of the nodal involvement and the biology of the disease have to be considered in planning treatment.

At this stage it is essential that we understand the division of neck nodes into 'nodal groups' described as 'levels' rather than 'anatomical regional groups'. This serves the dual purpose of understanding the natural anatomical spread of disease and the principles of various neck dissections. The nomenclature has been standardized for uniformity in documentation and cross-communication. Retrospective studies have documented the patterns of spread of cancers from various primary sites in the head and neck area to cervical nodes [6]. The terminology for categorizing the lymph node groups was originally described at Memorial Sloan-Kettering Cancer Centre, New York, and is widely accepted. It divides the hemi-neck into five regions. A sixth region was added later [7]. The nodal levels are diagrammatically shown in figure 1 and a brief description of neck nodal levels [8] is described in Table 1. The regional lymph node staging [9] is described in Table 2. Thus, all regional nodes are not at risk of metastases initially from any primary site in the absence of grossly palpable nodes. For example, in oral tumors, if neck is clinically negative, level IV and V nodes are generally not at risk of harboring micro-metastases. A similar situation applies to other organ and node metastases. The first nodal basin varies for each site and is described in the Table 3. Skip metastases in the absence of first echelon nodes being involved are unusual. On the other hand, when clinically palpable nodes are present, comprehensive clearance of all regional nodal groups is warranted (see also discussion under selective dissection for node positive neck). Several well documented studies have confirmed this to be the case [7].

The risk of nodal metastases is dependent on various factors related to the primary tumor. These include the site, size (large or small), T stage, location of the primary tumor (within an organ such as the vocal cord compared with supraglottis) and histomorphology of the primary tumor. The risk of metastases increases as one progresses from the anterior to posterior part of the upper aero-digestive tract; from lip (10%) progressing along the tongue (25%), gum (30%), floor of mouth (40%), oropharynx (55%) to hypopharynx (65%). Endophytic tumors, poorly differentiated tumors, and tumors with a greater thickness (tongue and floor of mouth) are more likely to have metastases [10].

The goal of treatment of cervical nodal metastases is regional control of disease. Micrometastases and minimal gross metastases may be controlled by radiotherapy but surgery remains the mainstay of treatment as it provides comprehensive clearance of all grossly enlarged nodes and offers accurate histological information regarding the presence of micrometastases. While the indications for comprehensive surgical clearance of regional nodes in patients with clinically metastatic disease are obvious, the indications for elective nodal dissection (END) in patients with N₀ neck status are more challenging and are discussed separately. The present discussion is aimed at answering why, when, and how to treat neck nodes in patients with primary squamous head and neck cancers?

The term "neck dissection" refers to a surgical procedure in which the fibro-fatty soft tissue content of the neck is excised to remove the lymph nodes that are contained therein [11]. Over the years, a number of variations and modifications of RND have been described. In addition, different terminologies have been used to represent the same surgery as well as different surgeries. The classification by the American Academy of Head and Neck Surgery, and which is also endorsed by American Society of Head and Neck Surgery, has become accepted widely [7]. The principles are as follows:

The first three procedures are also grouped together under the name of "comprehensive neck dissections". A modified classification is outlined in Table 4. An example of extended neck dissection includes removal of parapharyngeal or buccinator or superior mediastinal nodes. It could also involve removal of non-lymphatic structures such as the hypoglossal or vagus nerves, paraspinal muscles or carotid artery.

Other terminologies used are "elective" and "therapeutic" neck dissections. If a neck dissection is carried out when there is no evidence of neck disease it is termed an "elective" neck dissection (END). Some authors use the word "prophylactic" instead of "elective" to denote the same procedure. However, the word "prophylactic" is not recommended as it gives a false impression. If the neck dissection is undertaken for metastatic disease in the neck it is called a "therapeutic" neck dissection. An END is usually, but not always, a SND. Hence END and SND are not synonymous terms

Palpable cervical nodes in patients with carcinoma of the upper aero-digestive tract are usually assumed to be due to metastases. Further evaluation with FNAC is needed only in patients where the nodes do not appear to be due to metastatic diseases as assessed clinically. At the other extreme, computerized tomography (CT) scan or magnetic resonance imaging (MRI) or a Doppler study may be required depending on what vital structure appears to be involved by nodal disease, or when there is a doubt about the ability to surgically remove the disease.

The presence of nodal metastases reduces survival by 50% [12, 13]. The prognosis of patients with cervical metastases is affected by the number of metastases, level of metastases, tumor burden, presence of extra capsular spread, nodal resectability and previous treatment with surgery or irradiation [12]. With the exception of patients with nasopharyngeal carcinomas, clinically apparent cervical lymph node metastases should be treated with a neck dissection [14].

A classical RND has been regarded as the gold standard for surgical treatment of clinically apparent metastatic cervical lymph node disease. When regional metastases are clinically palpable, comprehensive clearance of all the regional nodes which are at risk is mandatory. The underlying principle is that once metastases have occurred to one node, than there is likely to be sub-clinical metastases in other nodes. However, the operation has a significant morbidity which includes shoulder syndrome (shoulder weakness, deformity, pain, and adhesive capsulitis), anesthesia over the cutaneous distribution of the cervical plexus, and cosmetic deformity due to loss of the sternocleidomastoid muscle. The procedure involves removal of all ipsilateral cervical lymph node groups extending from the body of the mandible superiorly, to the clavicle inferiorly, anteriorly from the lateral border of sternohyoid, hyoid, contralateral anterior belly of digastric muscle belly, to the anterior border of trapezius posteriorly. Nodal groups from level I through to V are included in this dissection. Important extra-nodal structures such as the spinal accessory nerve, internal jugular vein and sternocleidomastoid muscle are sacrificed. Post- auricular, periparotid, suboccipital, perifacial, buccinator, retropharyngeal and paratracheal nodes are not removed in this nodal dissection.

The current indications for classical RND [15] are:

On the other hand, when appropriate indications exist, a function-preserving comprehensive neck dissection, sparing one or more vital structures, should be considered as long as it does not compromise the performance of a satisfactory clearance of metastatic disease. Preservation of the accessory nerve alone significantly reduces the morbidity of neck dissection. Thus, if the spinal accessory nerve is not involved by metastatic disease, it should be routinely preserved even in patients with clinically palpable metastatic nodes as the accessory nerve is rarely invaded by metastatic disease, and its preservation does not affect local recurrence of disease or overall patient survival. The cervical plexus contributes in part to the motor supply of trapezius through the accessory nerve [16‐18]. However, it has also been pointed out that such a contribution may be insignificant and inconsistent [19]. Spinal accessory nerve dysfunction occurs even after selective dissection and is usually attributed to stretching of the nerve due to retraction during clearance of level IIb lymph nodes and/or ischemia [20, 21]. Recent anatomical findings have shown that, functionally, the most important descending part of the trapezius muscle is innervated by a fine single branch arising from the spinal accessory nerve in the posterior triangle of the neck. Preservation of this branch may help to prevent more morbidity in patients undergoing modified radical neck dissections [22]. Accessory nerve sparing should not result in less than optimum nodal disease clearance and the nerve should be sacrificed if warranted in order to achieve this. Preservation of the sternocleidomastoid muscle, or internal jugular vein, is not recommended in patients with palpable nodes from an upper aero-digestive tract primary tumor. However, the same can be preserved, in addition to the accessory nerve, in nodal metastases from a well differentiated thyroid carcinoma. In addition, the sternomastoid can be preserved for cosmesis in young individuals and in those with minimal nodal disease. When bilateral neck dissection is undertaken simultaneously, it is important to preserve at least one internal jugular vein, preferably on the less involved side. It has also been realized that neck dissection is inadequate when employed as a sole modality in disease with adverse histological features such as perineural invasion and extracapsular spread. This has led to the usage of combination therapy of surgery with irradiation in selected patients. The common indications for RND, MRND and structures removed or spared are listed in Table 5. The overall results comparing RND and MRND [10, 23‐31] irrespective of the primary tumor are listed in Table 6. From the above discussion it is apparent that spinal accessory nerve preservation lessens the morbidity to a great extent and preservation of sternocleidomastoid muscle and internal jugular vein is of lesser importance. However, this issue becomes of significance where bilateral neck dissections are required.

The outlook for patients with "fixed" nodes is poor [12, 32]. The poor prognosis is probably due to an incomplete resection and there being extracapsular spread of disease. The term "fixed node" is vague in the sense it conveys different meaning to different people unless it is known to what structure is the node "fixed". True fixation occurs in 23 to 50% of patients [32]. Tumors fixed to mandible, larynx, sternocleidomastoid, prevertebral muscles and mastoid process can be resected with a margin of normal tissue. Invasion of skin can similarly be addressed by a wide margin of resection with resurfacing then carried out. Adjuvant irradiation can then be employed for microscopic disease. However, the 5-year survivals are poor and are reported to be around 15% [32]. Irradiation followed by salvage surgery did not improve survival and the concept of preoperative irradiation 'to make an inoperable tumor' "operable" is questionable [32]. Invasion of the carotid sheath does not preclude surgery and can be present in 5% of patients [32, 33]. Carotid artery invasion is also associated with a poor prognosis. When the external carotid artery alone is involved, it can be sacrificed. When the common or internal carotid is involved, the options are either no treatment, debulking with adjuvant therapy, mixed beam therapy or resection with or without revascularization. Patients should undergo cerebral blood flow studies prior to surgery and elective carotid artery resection can be undertaken. Earlier, carotid resection carried a mortality of 12% and hemiplegia rate of 33% [34]. Recently 22% 2 year disease-free survivals, with acceptable morbidity, have been described [35]. A median disease-specific survival of 12 months, with 24% of patients dying from distant metastases has been reported. The high-risk of complications, loss of quality of life and mortality, must be balanced against the natural history of the disease if it is left untreated [36]. When cervical spine or brachial plexus is involved, sacrifice of these with their attendant morbidities should be weighed against the local control of disease that could be achieved and the benefits that it will offer to patients.

Bilateral neck disease at initial presentation is a bad prognostic sign with a 5-year survival of only 5% then being likely [13]. Survival is not affected by laterality but by nodes larger than 6 cm in size [37]. Therefore, the presence of bilateral mobile nodes should not preclude treatment. At least one internal jugular vein should be preserved or reconstructed with a saphenous graft [38] or polytetrafluoroethelyene graft. Simultaneous ligation of both internal jugular veins can cause venous congestion, edema of the head and neck, raised intracranial pressure and the syndrome of inappropriate secretion of anti-diuretic hormone. Salvage surgery following a previous neck dissection also carries a dismal outlook, with a long-term survival of less than 5% [39].

As an extension of conservation surgery, some authorities advocate SND for metastatic neck nodes [40, 41]. These have been only applied in cases without evidence of massive lymphadenopathy, node fixation or gross extra-capsular spread. In addition, when a suspicious node is encountered in the lower limit of the dissection, a formal RND or its modifications can be undertaken. Although the primary site in these cases varies, it has been reportedly performed in low grade buccal or buccoalveolar lesions by an experienced surgeon. SND, when used in combination with postoperative irradiation therapy, has been shown to achieve comparable results to those reported for RND [42]. However, as a routine, a SND is not recommended for patients whose neck nodes are positive for tumor.

The role of neck dissection after definitive chemoradiotherapy for squamous cell head and neck cancer is not clearly defined [43]. After chemoradiotherapy (5 Fluorouracil and cisplatin), clinical parameters do not identify those patients with residual neck node disease or those at risk for regional failure, suggesting that neck dissection should be considered for all patients with N2-N3 disease [43]. A retrospective non- randomized study showed that the 5-year survival rate was significantly higher for those who had postoperative radiotherapy (38.9%) compared with patients who had pre-operative radiotherapy (9.1%) and those having surgery alone (0%). They were unable to evaluate the role of definitive chemotherapy and/or radiotherapy and salvage surgery as the results were inconsistent and the available data was limited [44].

Surgical resection is an essential component of aggressive chemoradiation protocols to ensure tumor control at the primary site and also in the neck. Neck dissections in patients with initial node positive disease reveal residual nodal metastases in 22% [45]. A German multicenter randomized trial showed that accelerated chemoradiation (Mitomycin and 5 Fluorouracil) to 70.6 Gy was more effective than accelerated radiation to 77.6 Gy alone. Selective lymph node dissection of residual neck masses after completion of hyperfractionated accelerated radio-(chemo-)therapy is likely to contribute to loco-regional tumor control in patients with advanced head and neck cancers [46]. On the contrary, another study did not recommend neck dissection for patients who respond completely after chemoradiation but did recommend it for patients with either no or partial radiographic responses [47]. Another study showed improved neck control with neck dissection for patients with clinically residual disease or N3 neck cancer but no significant impact on the outcome of patients with N2 stage disease who were rendered clinically disease-free with intensive concurrent chemoradiation [48]. There is also a suggestion that neck dissection followed by chemoradiotherapy has a higher, and statistically significant, disease-specific survival rate compared to chemoradiation alone [49]. Consensus regarding the use of neck dissections for complete responders and incomplete responders has yet to be achieved and the data are controversial. A possible benefit from neck dissection after a complete response of the primary tumor after combined chemoradiation or definitive external radiation for advanced squamous cell carcinoma of the head and neck may only be anticipated in patients with persisting sub-clinical neck disease who have no other sites of disease. Some clinicians have even argued that the salvage rate for clinically detectable residual neck disease does not justify neck dissection being performed. Randomized trials addressing these questions, and a trial addressing the accuracy of new imaging modalities such as post chemotherapy and post radiation positron emission tomography scanning, across multiple institutions would be appropriate [50, 51].

Pathological evaluation of the neck dissection specimens is important. A RND or MRND specimen should comprise of at least 10 nodes on pathological examination. Apart from the number of involved nodes it is important to identify adverse prognostic factors resulting in a predisposition to local recurrence and a reduced survival. Perineural invasion, extracapsular spread and vascular invasion indicate a bad prognosis [52, 53]. Desmoplastic stromal reaction was associated with a seven fold increase in the risk of recurrent disease. Tumor has been identified up to 12 cm along the perineural space in patients with perineural invasion, and should be considered as a positive tumor margin [2]. Patients with poor prognostic factors listed above should receive adjuvant therapy such as irradiation, with or without chemotherapy, as required.

Neck dissection is the standard therapy for patients with metastatic neck nodes. RND has now been replaced with MRND for selected cases of patients with node positive neck disease. The improvements in understanding the biology of cancer and techniques in surgery have led to this change with its associated reduction in morbidity. Imaging studies to predict histological nodal metastases have so far not been successful. Evaluation is underway to replace MRND with SND for early nodal disease. SNB may replace SND for patients with N0 disease in future but certain issues still need to be addressed. Finally, whether endoscopic neck dissections will replace the existing procedures remains to be evaluated.