The presence of cervical lymph node metastasis reduces survival of patients with squamous carcinoma of the upper aerodigestive tract by up to 50%. Since Crile's landmark publication in 1906, the radical neck dissection (RND) has remained the criterion standard for excision of cervical nodal metastases resulting from head and neck cancer.  Hayes Martin further popularized use of the RND in the 1950s.  The procedure uses, in Crile's words, "a 'block dissection' of the regional lymphatic system . . . on exactly the same lines as the Halstead operation for cancer of the breast. Such a dissection is indicated whether the glands are or are not palpable."
The RND remained the mainstay of cervical lymphadenectomy procedures for the treatment of most patients with regionally metastatic head and neck cancer until the 1990s, when the head and neck community began to embrace less morbid approaches. This philosophical shift in the treatment paradigm occurred in the absence of substantive prospective investigational research evidence that demonstrated equivalent oncologic efficacy to the RND. Nonetheless, the life-altering morbidity associated with the RND became a driving force for change and critical re-evaluation of the rationale for the surgical management of head and neck cancer that was irreversible.
In 1952, Martin began to address the morbidity associated with the RND. Since then, the focus of criticism against the RND has addressed the related morbidity, causing other surgeons, including Jesse and Ballantyne, to search for cervical lymphadenectomy procedures that could provide oncologic cure with less morbidity.  This article reviews the rationale for and technique of the modified radical neck dissection (MRND) that developed from these efforts.
An image depicting a neck dissection incision can be seen below.
The neck is the part of the body that separates the head from the torso. The Latin-derived term cervical means "of the neck." The neck supports the weight of the head and is highly flexible, allowing the head to turn and flex in different directions.
The midline in front of the neck has a prominence of the thyroid cartilage termed the laryngeal prominence, or the so-called "Adam's apple. Between the Adam’s apple and the chin, the hyoid bone can be felt; below the thyroid cartilage, a further ring that can be felt in the midline is the cricoid cartilage. Between the cricoid cartilage and the suprasternal notch, the trachea and isthmus of the thyroid gland can be felt.
The quadrangular area is on the side of the neck and is bounded superiorly by the lower border of the body of the mandible and the mastoid process, inferiorly by the clavicle, anteriorly by a midline in front of the neck, and posteriorly by the trapezius muscle.
Muscles in the front of the neck are the suprahyoid and infrahyoid muscles and the anterior vertebral muscles. The suprahyoid muscles are the digastrics, stylohyoid, mylohyoid, and geniohyoid. The infrahyoid muscles are the sternohyoid, sternothyroid, thyrohyoid, and omohyoid.
Morbidity Associated With Radical Neck Dissection
Spinal accessory nerve
An extensive body of literature documents morbidity following RND. In 1961, Nahum et al discovered that patients who had undergone radical neck dissection (RND) commonly experienced shoulder discomfort with limitation of shoulder abduction.  Similarly, in 1952, Ewing and Martin evaluated 100 patients who had undergone RND.  Of these patients, 42 experienced shoulder discomfort, and 60 demonstrated shoulder stiffness and decreased range of motion.
Subsequently, Short et al compared 12 patients who underwent RND with 23 who underwent neck dissection that spared the spinal accessory nerve (SAN).  On average, the patients who underwent RND demonstrated greater shoulder-related disability. Other researchers have similarly demonstrated the disability resulting from neck dissection. Most recently, Terrell and colleagues found that neck dissections sparing the spinal accessory nerve (SAN) were associated with less shoulder and neck pain and less need for pain medications than dissections not sparing the SAN. 
Injury to the SAN results in dysfunction of the trapezius muscle. The trapezius elevates, rotates, and retracts the scapula. When the SAN is transected, the shoulder droops as the scapula is translated laterally and rotated downward. Patients present with an asymmetric neckline, a drooping shoulder, winging of the scapula, and weakness of forward elevation.
Internal jugular vein
Although some clinicians believe that simultaneous bilateral internal jugular vein (IJV) ligation is of minimal risk if the operative time is less than 5-6 hours, the literature is filled with reports of serious complications secondary to bilateral RND. Increased intracranial pressure may result, and reports of blindness, laryngeal edema, stroke, and death exist within the medical literature. In a review by Dulguerov et al of bilateral RND, the reported mortality rate ranged from 0-3% for staged RND to 10-14% for simultaneous RND.  The morbidity associated with bilateral IJV resection has led to staged procedures or recommendations for IJV reconstruction in which bilateral IJV resection is necessary.
Preservation of the IJV may also facilitate microvascular surgical reconstruction.
Sternocleidomastoid muscle resection
Concerns regarding sternocleidomastoid muscle (SCM) resection focus mainly on the cosmetic deformity that results from loss of muscle mass and the subsequent change in normal contours of the anterior neck. The SCM can also serve a protective function for the carotid artery when flap necrosis or fistula formation occurs. In general, however, loss of the SCM results in the least morbidity of the 3 nonlymphatic structures sacrificed during RND.
Events Leading to Modified Neck Dissection
Increasing awareness of the morbidity associated with radical neck dissection (RND) led head and neck surgeons to explore modifications of the classic procedure. Suarez initiated a change in the surgical approach to cervical lymph node metastases by illustrating in anatomic studies that cervical lymphatics are contained within well-defined fascial compartments that partition them from the muscular, vascular, and neural structures of the neck. He proposed that these nonlymphatic structures could be preserved during neck dissection for limited disease without adversely affecting regional control. Suarez, therefore, defined the anatomic basis and surgical criteria for the functional neck dissection (FND).
Bocca subsequently popularized the FND in the English-language literature and heralded this procedure as oncologically equivalent to the RND in the node-positive neck.  The FND was a comprehensive neck dissection, removing all cervical lymph node levels included in the RND but preserving the internal jugular vein (IJV), sternocleidomastoid muscle (SCM), and spinal accessory nerve (SAN).
At MD Anderson, Byers reviewed 1372 neck dissections, and at Memorial Sloan-Kettering, Shah reviewed 1119 RNDs for squamous carcinoma of the upper aerodigestive tract. [10, 11, 12] They demonstrated that nodal metastatic disease predictably occurs in certain regions of the neck based on the site of the primary tumor. These 2 studies offered valuable insight about patterns of nodal metastases and provided a rationale for the modified neck dissection and the selective neck dissection. (Their findings are reviewed in the section Supporting Evidence.)
During this period, the neck dissection literature lacked any standardization in the nomenclature describing each surgical procedure, resulting in an initiative by the Committee for Head and Neck Surgery and Oncology of the American Academy of Otolaryngology/Head and Neck Surgery to standardize the diverse nomenclature. This classification has become widely accepted and has also been endorsed by the American Head and Neck Society.
Radical neck dissection (RND) refers to the removal of all ipsilateral cervical lymph node groups extending from the inferior border of the mandible superiorly to the clavicle inferiorly and from the lateral border of the sternohyoid muscle, hyoid bone, and contralateral anterior belly of the digastric muscle anteriorly to the anterior border of the trapezius muscle posteriorly.
Modified radical neck dissection (MRND) is defined as the excision of all lymph nodes routinely removed in a radical neck dissection with preservation of one or more nonlymphatic structures (SAN, IJV, SCM). The preserved nonlymphatic structures should be specifically mentioned (eg, modified radical neck dissection with preservation of the IJV and SCM).
A selective neck dissection refers to any neck dissection in which one or more lymph node groups removed in RND is preserved.
The most widely accepted terminology for categorizing the lymph node groups in the neck was originally described by head and neck surgeons at Memorial Sloan-Kettering Cancer Center, dividing the neck into 5 regions. A sixth region was later added to characterize the lymph nodes in the anterior neck (see Table 1). Levels I, II, and V have also been divided into sublevels, and the anatomic and radiographic boundaries of each level and sublevel have been clarified.
Table 1. Lymph Node Groups of the Neck (Open Table in a new window)
|Level||Lymph Node Group|
|I||Submental and submandibular nodes|
|II||Upper jugular nodes|
|III||Middle jugular nodes|
|IV||Lower jugular nodes|
|V||Posterior triangle nodes|
|VI||Anterior compartment lymph nodes|
Supporting Evidence for the Modified Radical Neck Dissection
No prospective studies compare modified radical neck dissection (MRND) with radical neck dissection (RND), and few studies exist that have compared the outcomes following RND with outcomes following MRND. Buckley et al identified 9 retrospective studies that were deemed suitable for analysis.  The mean combined recurrence rate following RND was 13.6% (95% confidence interval, 12.0-15.2%), whereas the mean combined recurrence rate following MRND was 6.9% (95% confidence interval, 5.4-8.4%).
However, 7 of these studies were published before the development of a standardized classification for neck dissections, and modified neck dissections at that time included both the MRND and the selective neck dissection. In addition, these studies differed in their indications for neck dissection, use of preoperative or postoperative radiotherapy, and follow-up interval. Treatment selection bias probably also affected the results of these retrospective studies because MRND may have been used to treat less-advanced nodal metastatic disease. Hence, comparing the outcomes of RND and MRND in these series is of limited value.
A closer look at the individual findings of Byers and Shah, however, can provide clinicians with applicable information that supports the use of the MRND in properly selected patients.
Byers reviewed 182 functional neck dissections that did not receive postoperative radiotherapy. Of 124 necks with pathologically negative findings (pN0), the regional recurrence rate was 2%. In 31 pathologically staged N1 necks (pN1), no recurrences were recorded. However, 26% of the 27 neck dissections that had multiple involved lymph nodes with extracapsular spread developed recurrence. Postoperative radiotherapy reduced the recurrence rate to 7% in 43 additional dissected neck sides with extracapsular spread (see Table 2).
Table 2. Functional Neck Dissection* (Open Table in a new window)
|Number of Neck Dissections||Recurrence Rate|
Multiple nodes or extracapsular spread
|Surgery + postoperative radiotherapy||43||7%|
|*Adapted from Byers.|
Patients who underwent a selective neck dissection or MRND that preserved the SAN had a recurrence rate in the dissected neck of only 7.4%. Of these neck dissections, 36% had metastatic nodes greater than 3 cm in size or multiple metastatic lymph nodes. When the jugulodigastric nodes in the upper anterolateral neck were pathologically involved and the nerve was preserved, the recurrence rate decreased from 17% to 4.7% with the use of postoperative radiotherapy. Resection of the IJV and SCM with preservation of the SAN was performed on 94 patients. Of the patients who underwent surgery alone, 73% had pathologic jugulodigastric node involvement with an 8% recurrence rate in the dissected neck. Patients who underwent surgery and postoperative radiotherapy, 100% of whom had jugulodigastric nodal involvement, demonstrated a 0% recurrence rate (see Table 3).
Table 3. Neck Dissections Preserving SAN* (Open Table in a new window)
|Patient Category||Recurrence Rate|
Jugulodigastric node involvement
Surgery + postoperative radiotherapy
MRND (SAN preserved)
Surgery + postoperative radiotherapy
|*Adapted from Byers.|
Shah retrospectively reviewed 1081 patients who underwent 1119 RND procedures with an average harvest of 39 lymph nodes. Of 343 elective neck dissections, 113 had pathologically documented nodal metastases. Level V metastases were present in 2% of patients with oral cavity tumors, in 7% of patients with oropharyngeal cancer, in 0% of the necks of patients with hypopharyngeal cancer, and in 7% of those with laryngeal carcinoma (see Table 4).
Table 4. Level V Metastasis in RND Patients* (Open Table in a new window)
|Location of Primary||
Elective RNDs, Number
Therapeutic RNDs, Number
|*Adapted from Shah.|
Shah also considered oral cavity lesions by lymph node level and found that only 1 patient out of 65 (1.5%) with a clinically negative neck (cN0) and pathologically proven nodal metastases had level V involvement, while 8 of 152 patients (5.3%) with clinically positive neck (cN+) findings and pathologically proven metastases had level V involvement. All of the patients with level V involvement had nodal involvement at other levels. Patients with oral primary tumors of the floor of mouth or gingiva developed level V metastases, while patients with carcinoma of the tongue, retromolar trigone, or buccal mucosa did not.
Subsequently, investigators have questioned the need for dissection of sublevel IIB, IV, and the apex of level V (sublevel VA) in some clinical situations. These results suggest that a RND is frequently not indicated for the treatment of many patients with head and neck cancer, and a modified or selective neck dissection that minimizes traumatic manipulation of the SAN may be more appropriate, particularly if the neck is clinically negative.
Bocca et al reported the outcomes of 843 FNDs in 1984.  Overall, he encountered an 8.1% recurrence rate in the neck 5 years following surgery. The neck recurrence rate for elective FND was 2.3% versus 30.4% for therapeutic FND.
Khafif et al compared the outcomes of patients who underwent RND with the outcomes of those who underwent MRND in 1990.  They were unable to demonstrate a statistical difference in the neck recurrence rate overall or in patients with N2 or N3 disease. The 3-year survival rate in the MRND group was 74% versus 63% in the RND group. However, patients receiving radiotherapy combined with surgery were not evaluated separately from patients who underwent surgery, and patients who underwent delayed neck dissections were also grouped with patients treated with immediate neck dissections.
These results compare favorably with the experience with RND reported by Strong, who found a 6.7% recurrence rate in necks with pathologically negative findings, a 36.5% recurrence rate when nodal metastases were present within one level, and a 71.3% recurrence rate when multiple lymph node levels were involved. 
In the absence of a prospective randomized comparison of these 2 surgical procedures, the results discussed above support the oncologic safety of the MRND in a number of situations in which the RND was previously routinely performed. However, the reader is cautioned that assessment of survival statistics in different studies is complex, and many factors must be considered, including the relative distribution of primary sites, T stages, extent of nodal disease, management of the primary site and the contralateral neck, and the performance of various types of neck dissections by surgeons with different backgrounds and levels of expertise.
Indications for Modified Radical Neck Dissection
No formalized indications have been created for modified radical neck dissection (MRND), and indications vary widely from one clinician to another. In general, MRND is indicated whenever preservation of the spinal accessory nerve (SAN), internal jugular vein (IJV), or sternocleidomastoid muscle (SCM) is possible without compromising oncologic efficacy, and when a selective neck dissection would not be considered an adequate oncologic procedure. An en bloc approach should be pursued, and nodal metastases encapsulated by reactive connective tissue must be resected atraumatically.
For patients who are clinically staged N0 or N1, a selective neck dissection or MRND would be appropriate. In patients with N2 disease, MRND is reasonable if any of the aforementioned nonlymphatic structures can be safely preserved. Many patients with N3 disease (nodal metastases > 6 cm) require RND, but MRND can be considered when dissection is feasible.
MRND is contraindicated whenever preservation of the nonlymphatic structures of the neck would compromise complete resection of the cervical metastatic disease. Because outcome data are limited, the precise role of MRND remains undefined, especially in N2 and N3 nodal disease. Although a growing body of evidence supports the performance of selective neck dissection in carefully selected patients with clinically positive nodal disease, no prospective randomized clinical trials have been conducted. Consequently, the indications for MRND and selective neck dissection in these patients also remains poorly defined.
Modified Radical Neck Dissection: Operative Technique
A skin incision is made that optimizes exposure of the neck. The author prefers to use an apron flap design that extends from the mastoid tip to the mandibular symphysis (see the first image below). Alternatively, a hockey stick incision can be made. Incisions that result in a trifurcation are less desirable because of the potential for distal flap necrosis and carotid artery exposure. If bilateral neck dissections are planned, the incision extends from one mastoid tip to the other to create a single apron flap (see the second image below).
The subplatysmal flap is elevated superiorly to the mandibular border and inferiorly to the supraclavicular region. Dissection in a plane that separates the fascia from the underlying platysma facilitates an en bloc approach to the lymphatic structures within an envelope of fascia. The marginal mandibular nerve can be preserved by elevating the submandibular gland fascia as part of the flap or by elevating the flap deep to the common facial vein after dividing it. Flap elevation proceeds posteriorly immediately deep to the subcutaneous adipose tissue back to the anterior border of the trapezius muscle. Care is taken during flap elevation to remain superficial, preserving the greater auricular nerve and the SAN (see the image below).
The contents of the submental triangle (sublevel IA) are then elevated from the inferior border of the mandible and the opposite digastric muscle off of the mylohyoid muscle, leaving the overlying muscle fascia intact. Dissection in the proper plane allows for an en bloc elevation of the contents into the submandibular triangle (sublevel IB) and to the posterior border of the mylohyoid muscle. Retraction of the mylohyoid muscle anteriorly allows for identification of the submandibular duct, which is ligated and divided, and the lingual nerve, which supplies innervation to the submandibular gland. These contributions are divided inferior to the submandibular ganglion. The dissected contents of sublevels IA and IB are then elevated over the digastric muscle in continuity with the nondissected portion of the neck (see the images below).
At this time, intraoperative assessment is necessary to determine the proximity and/or fixation of lymph node metastases to the IJV, SAN, and SCM. The contents dissected from level I are elevated caudally to visualize the superior internal jugular vein. Retraction of the posterior belly of the digastric with a vein retractor may facilitate exposure. Sacrifice or preservation of the new lymphatic structures usually depends on the size and extent of lymph node metastases at level II, the upper jugular lymph nodes.
Identification of the SAN can be performed anterior or posterior to the SCM. However, if the possibility of spinal accessory preservation is in question, anterior identification will more quickly determine whether to sacrifice or preserve the nerve. Posterior to anterior identification also bisects the SCM, preventing an en bloc dissection if RND is necessary. Identification anterior to the SCM requires retraction of the anterior border of the SCM posteriorly in the region of the digastric muscle. Elevation of the fascia from the undersurface of the SCM using electrocauterization or blunt dissection with a hemostat parallel to the SAN can be used to easily isolate the nerve (see the image below).
Posterior to the SCM, the nerve invariably can be found approximately 1 cm cephalad to the greater auricular nerve as it wraps around the SCM (see the image below). The approach is more useful when the surgeon chooses to resect the SCM from the onset of surgery.
The relationship of nodal metastases to the IJV, SAN, and SCM is then evaluated. Inspection and palpation are used to assess for IJV thrombosis, encasement of the IJV or SAN, and extracapsular involvement of the SCM. Atraumatic dissection with a small hemostat or finger may facilitate the development of a "safe" plane of dissection. Such dissection should preserve the thick reactive fibrous tissue encapsulating most nodal metastases. The inability to develop a clean plane of dissection mandates sacrifice of the involved nonlymphatic structure. Furthermore, if evidence of extranodal fixation to the surrounding soft tissues of the neck (ie, deep cervical musculature) is found, performance of an RND (or extended RND) must be considered because of the advanced stage of regional metastatic disease present.
If the SAN can be preserved, dissection is then continued from its proximity to the IJV posterocaudally to the trapezius muscle, dividing the SCM (see the image below). If the SCM is going to be preserved, the SAN must be carefully dissected by identifying the nerve both anterior and posterior to the SCM. Stretching of the SAN must be minimized.
A posterior to anterior dissection is then performed beginning at the anterior border of the trapezius muscle. The posterior triangle contents are elevated in an en bloc fashion off the fascia of the deep cervical musculature, preserving the phrenic nerve and the brachial plexus, located deep to this fascia.
The SAN must then be freed from the soft tissues of the posterior triangle and can be carefully retracted away from the region of dissection with a vessel loop or nerve hook. Dissection is continued to the posterior border of the SCM. The posterior triangle contents posterosuperior to the nerve (sublevel VB) are rotated under the nerve inferiorly. Retraction of the SAN in the posterior triangle can be minimized if dissection of the apex of level V (sublevel VA) is not required. The posterior triangle contents are dissected from the posterior border and the undersurface of the SCM.
If the SCM is being resected, the muscle is bisected and elevated in continuity with the posterior triangle contents. The superior SCM and the supraspinal posterior triangle may be too bulky to rotate underneath the SAN, requiring removal in 2 pieces.
Careful dissection of the supraspinal portion of level II, also known as the submuscular triangle (sublevel IIB), is necessary to minimize trauma to the SAN and the IJV. The lymph node-bearing fibroadipose tissues in this region are also rotated under the SAN in the same manner that dissection of sublevel VA was performed, up to the lateral aspect of the IJV.
At this point, the posterior triangle contents, with or without the SAN and SCM, have been elevated to the lateral aspect of the IJV. If the SCM is being resected, transection is performed below the mastoid tip and above the clavicle as in a RND. The contents of levels II, III, and IV have been elevated after division of the omohyoid muscle.
The IJV is once again evaluated. Additional nodal metastases are frequently present in the deep cervical chain immediately adjacent and posterior to the IJV and are also occasionally present deep and posterior to the IJV. These can usually be removed en bloc with the remainder of the dissection in a posteroanterior fashion, sharply incising the fascia of the jugular vein with a scalpel blade using a feather-light touch. Previous extensive blunt dissection along the posterior aspect of the IJV prevents elevation of the neck dissection over the IJV. If the IJV requires sacrifice due to metastatic nodal involvement or tumor thrombosis, the vein is ligated and divided superiorly and inferiorly following identification and preservation of the vagus nerve (see the image below).
Dissection is continued anteriorly, elevating the fascia and soft tissues up to the infrahyoid strap muscles and the hyoid-digastric junction (see the first image below). Preservation of a fascial layer superficial to the carotid artery is usually possible, and exposure of the carotid artery should be discouraged unless necessary. The neck dissection is then removed and oriented for lymph node level-specific histopathologic evaluation for pathologic staging purposes (see the second image below). Suction drains are strategically placed and a layered closure is performed. Antibacterial ointment is then applied to the incision line.
Postoperative complications following modified radical neck dissection (MRND) match those experienced with radical neck dissection (RND) and include hematoma, infection, skin flap necrosis, chyle fistula, marginal mandibular nerve injury, and carotid artery rupture. Patients should be counseled that loss of cutaneous sensation occurs in the distribution of the cervical plexus, including the skin of the neck and the periauricular region.
Sternocleidomastoid muscle (SCM) resection results in loss of normal contour in the anterior neck with resultant cosmetic deformity. Resection may also contribute to fibrosis and limitation of range of cervical motion.
Patients who undergo MRND with preservation of the spinal accessory nerve (SAN) may develop shoulder pain and dysfunction. Short et al found that 61% of patients who underwent MRND developed shoulder pain, whereas 100% of patients with RND had pain.  In addition, although patients who were treated solely with radiotherapy had normal shoulder abduction, those patients who underwent nerve-sparing neck dissections demonstrated decreased abduction. The RND patients had the greatest decrease in abduction.
Sobol et al performed electromyography 16 weeks following surgery.  Patients who underwent a MRND sparing the SAN demonstrated markedly abnormal electromyogram (EMG) findings 39% of the time, while only 30% had normal EMG findings. However, these patients tended to demonstrate improvement in their EMG results over time. Patients who underwent a supraomohyoid neck dissection had the least dysfunction. Of these patients, 22% had moderately abnormal EMG findings, while 56% had normal EMG findings.
Most recently, Terrell and colleagues found no significant difference in shoulder or neck pain between patients who underwent MRND and those who underwent RND, but patients treated by RND used pain medications more frequently.  When level V was not dissected, patients experienced less shoulder or neck pain, but the frequency of pain medication used was not significantly different.
In general, the preponderance of the research data supports the theory that dissection and skeletonization of the SAN are traumatic enough to cause pain and shoulder dysfunction.
Treatment includes routine postoperative shoulder physiotherapy following all neck dissections. Although early physiotherapy that is targeted at facilitating spinal accessory nerve recovery and increasing scapular muscle strength may reduce shoulder dysfunction, evidence in support of its effectiveness is lacking. Patients with persistent pain and dysfunction after 1 year of conservative treatment may be candidates for surgical reconstruction with the Eden-Lange procedure, in which the insertions of the levator scapulae, rhomboideus minor, and rhomboideus major are transferred.
Preservation of the IJV in MRND results in IJV patency rates of 86-87% (99% if patients with compression of the IJV by tumor recurrence or myocutaneous flaps are excluded). Radiotherapy can affect IJV patency. During surgery, trauma to the IJV should be minimized by atraumatic manipulation and avoiding IJV desiccation. Postoperatively, compression of the IJV should be minimized by avoiding tight dressings and tracheostomy tube ties. If simultaneous IJV ligation is performed, IJV reconstruction with a greater saphenous vein graft should be considered.