Anatomic Variability of the Accessory Nerve: Implications for Dissection of Level IIB

Abstract

Objectives:

During neck dissection, level IIB lymphadenectomy necessitates manipulation of the spinal accessory nerve that might be avoided and might cause postoperative disability. Current literature does not describe the effect of variation in the spinal accessory nerve in the upper neck. We sought to measure the effect of the dimensions of level IIB on nodal yield in level IIB and on patient reported neck symptoms.

Methods:

We measured the boundaries of level IIB of 150 patients undergoing neck dissection. Level II was dissected and separated into levels IIA and IIB intraoperatively. Patient-reported symptoms were assessed in 50 patients using the Neck Dissection Impairment Inventory. We computed descriptive statistics, and sought to identify correlation with the number and proportion of level IIB nodes and the number of metastatic nodes. Level IIB dimensions were analyzed as predictors of postoperative symptoms.

Results:

We measured 184 sides, with 37.7% of level II nodes found in level IIB. Mean accessory nerve length across level II was 2.5 cm. Every additional 1 cm in accessory nerve length was associated with two more level IIB nodes. At all accessory nerve lengths, meaningful numbers of nodes were present in level IIB. Accessory nerve length and other factors did not correlate with NDII scores.

Conclusions:

Longer lengths of accessory nerve across level IIB correlated with greater nodal yield. However, data did not point to an accessory nerve length cutoff below which level IIB dissection could be avoided. In addition, the dimensions of level IIB did not correlate with postoperative neck symptoms.

INTRODUCTION

For over a century, the surgical management of cervical lymph nodes has been an integral part of the management of patients with head and neck cancer. The nodes in the upper neck are among the most likely to be involved in metastasis in many head and neck cancers. The Rouvière classification,¹ based on study of lymphatic flow, described two chains of lymph nodes in the upper neck draining the head and neck: the jugular chain following the internal jugular vein and the spinal accessory chain taking an oblique course. The two chains overlap near the jugular foramen.² More recent terminology defines six nodal levels on each side of the neck. Level II comprises the lymph nodes deep to sternocleidomastoid extending from the skull base to the level of the hyoid.³ The spinal accessory nerve (SAN) crosses level II and divides it into level IIA inferomedially and level IIB superolaterally. Level II contains the overlapping pathways of both the jugular and the spinal accessory chains.

The morbidity of neck dissection derives in meaningful part from the position of the SAN within level II.⁴ Early descriptions of the procedure described removal of the spinal accessory nerve in the upper neck to achieve rapid lymphadenectomy while avoiding dissemination of cancer through transgression of lymphatic channels.^5–8 The nerve supplies the trapezius muscle, including the superior portion which elevates and controls the scapula. After radical neck dissection, patients often have long-term disability related to the shoulder and neck.⁹ The SAN must be exposed and dissected when it is preserved. Neck dissections which involve manipulation of the long segment of the SAN across level 5 are associated with both postoperative symptoms and neuromuscular deficits.^10,11

With data confirming the efficacy of less-than radical neck dissections for many cases, the routine resection of the spinal accessory nerve, and then the dissection of level V, gradually ceased.^2,12,13 As previous work has shown electrophysiologic changes in the SAN during level II dissection,^14,15 a relevant question is whether operations with less manipulation of the SAN in level II lead to better functional outcomes without compromising oncologic outcomes.¹⁶

Several studies have attempted to define the probability of occult disease in level IIB, attempting to evaluate the risk of leaving level IIB undissected.^17–19 Many of these studies predict level IIB nodal metastases by the primary tumor site and the volume and site of neck disease. Although different subsites of the upper aerodigestive tract might link to different lymphatic nodal chains as defined by Rouvière, the anatomic overlap of those chains in level II means that the upper aerodigestive tract subsite of a primary cancer may not be a good predictor of disease in level IIB.

There is significant variability in surgical anatomy, and the relationship between anatomic differences and neck dissection outcomes has not yet been characterized. Paleri et al. identified this variation as a potential confounding factor in many studies of level IIB surgery.²⁰ Our goal was to identify variability in the relative size of level IIB and IIA, the distribution of nodes and metastatic nodes in these levels, and to identify whether the relative size of level IIB correlates with postoperative neck and shoulder symptoms.

METHODS

This study was approved by the University of Iowa institutional review board and written informed consent was obtained from all patients. Eligible patients were those having neck dissection involving level II but not level V for malignancy at any site. Patients whose neck surgeries required resection of the accessory nerve, upper sternocleidomastoid muscle, or posterior belly of digastric were excluded. Participating patients underwent surgery under the direct supervision of the first author using his routine technique,²¹ including concurrent resection of the primary site as indicated.

During surgery, the anterior border of the sternocleidomastoid was elevated off the underlying superficial layer of the deep cervical fascia and the accessory nerve was identified. The elevation of the sternocleidomastoid was completed with the nerve in view. The nerve was dissected from the sternocleidomastoid up to the posterior belly of digastric. We kept the level II specimen intact during dissection by passing the IIB component inferomedially underneath the accessory nerve. Level II neck specimens were divided on the back table into level IIA and IIB components by the operating surgeon, and these were sent separately to pathology. After the neck dissection was complete, the borders of level IIB were measured. A single retractor was placed on the sternocleidomastoid muscle to permit measurement of distances between three points: accessory nerve at the inferior border of the posterior belly of digastric; entry of accessory nerve into sternocleidomastoid, and junction of sternocleidomastoid and posterior belly of digastric (Fig. 1).

Fig. 1.

Level IIB dimension measurements in a right neck. A = accessory nerve; B = sternocleidomastoid; C = posterior belly of digastric.

The Neck Dissection Impairment Inventory (NDII) is a 10-item instrument validated for the purpose of assessing patient-reported symptom burden after neck dissection.²² We assessed NDII in a sub-cohort of the patients recruited for this study. These patients were contacted 2–6 months after surgery in person or by phone for survey completion. The composite NDII score was used as an outcome measure.

Measurements and clinical features were recorded in a REDCap database. The pathology service employed routine accessioning techniques to identify and report the number of nodes present in each level IIA and IIB specimen and the number of nodes with metastatic deposits.

The area of the level IIB triangle described by the measured anatomic lengths was computed using Heron’s formula. Univariate linear regression was used to estimate the relationship between IIB measurements and the number of IIB nodes. Logistic regression analysis was used to analyze the distribution of nodes into level IIA and IIB by modeling the location of each node as a Bernoulli trial as a function of the measured anatomic and clinical factors. We also analyzed the presence of at least one metastatic node in level IIB by univariable logistic regression using clinical and anatomic factors as independent variables. Analysis of NDII scores was performed per patient rather than per side, with the shorter of the level IIB dimensions used in cases of bilateral surgery. A priori sample size calculation for the patient-reported outcome analysis was performed based on α = 0.05, β = 0.2, minimal clinically important difference of 15, and standard deviation 16 based on Gallagher et al.²³ Statistical analysis was performed using SAS 9.4 (SAS Institute, Cary, NC).

RESULTS

We made anatomic measurements in 150 patients, encompassing 184 nerves. Characteristics of the patients are shown in Table I. Subjects’ mean age was 60.8 years; mean body mass index (BMI) was 29.1 m²/kg, and mean height was 67.9 inches.

TABLE I.

Patient Characteristics.

Variable	Category	N (%)
Sex	Male	99 (66.0)
	Female	51 (34.0)
Primary site	Oral cavity	71 (47.3)
	Larynx	19 (12.7)
	Thyroid	24 (16.0)
	Oropharynx	13 (8.7)
	Hypopharynx	7 (4.7)
	Salivary	11 (7.3)
	Skin	4 (2.7)
	Unknown	1 (0.7)
Clinical T class	X	26 (17.3)
	1	20 (13.3)
	2	37 (24.7)
	3	30 (20.0)
	4	37 (24.7)
Clinical N class	0	57 (38.0)
	1	49 (32.7)
	2a	6 (4.0)
	2b	23 (15.3)
	2c	14 (9.3)
	3	1 (0.7)
History of radiotherapy	Yes	27 (18.0)
	No	123 (82.0)

Node dissection findings are shown in Table II. Across all specimens, 37.7% of the nodes (863 of 2288) were found in level IIB. There were 27 positive level IIB nodes (3.1%), and 117 of 1425 level IIA nodes were positive for metastasis (8.2%). One or more metastatic nodes were present in 65 of 184 (35.3%) sides, with 62 positive level IIA specimens and 12 positive level IIB specimens; 9 (4.9%) had metastatic nodes at both IIA and IIB and 3 sides (1.6%) had metastatic nodes at level IIB but not level IIA.

TABLE II.

Nodal Yields From Level IIA and IIB.

Cancer Site	Number of Nerves (%)	Mean Positive/Total Level IIA Nodes (%)	Mean Positive/Total Level IIB Nodes (%)
Oral cavity	80 (43.5)	0.5/9.0 (5.6)	0.1/4.8 (2.1)
Larynx	32 (17.4)	0.2/5.7 (3.5)	0/4.6 (0.0)
Thyroid	29 (15.8)	1.0/8.9 (11.2)	0.1/3.0 (3.3)
Oropharynx	13 (7.1)	0.8/4.8 (16.7)	0.1/4.2 (2.4)
Hypopharynx	13 (7.1)	1.2/4.9 (24.5)	0/5.5 (0.0)
Salivary	11 (6.0)	0.9/9.8 (9.2)	1.4/8.5 (16.5)
Skin	5 (2.7)	1.00/6.6 (15.2)	0.20/4.2 (4.8)
Unknown	1 (0.5)	1.0/2.0 (50.0)	0/3.0 (0.0)
Total	184	0.6/7.7 (7.8)	0.1/4.7 (2.1)

A summary of anatomic measurements appears in Table III. Height was a statistically significant predictor of SAN length but this relationship was clinically insignificant: regression predicted 0.13 cm longer accessory nerve for every additional 10 cm in height, p = 0.013. There were also statistically significant differences in accessory nerve length by sex, with an estimated length 0.39 cm shorter in women (p = 0.0004). Interestingly, there was a negative correlation between body mass index and the number of level IIB nodes, although at 0.02 fewer mean nodes for every 1 point increase in BMI (p = 0.007) this was not clinically meaningful.

TABLE III.

Dimensions of Level IIB.

Dimension	Mean (95% Confidence Interval)	Minimum	Interquartile Range	Maximum
Accessory nerve (cm)	2.5 (2.4–2.6)	1.0	2.0–3.0	5.5
Sternocleidomastoid (cm)	2.3 (2.2–2.4)	0.8	1.8–2.7	4.2
Posterior belly digastric (cm)	1.1 (1.1–1.2)	0.4	0.9–1.3	2.7
Level IIB area (cm2)	1.3 (1.1–1.4)	0.2	0.7–1.6	4.6

Table IV shows the results of logistic regression analysis relating the dimensions of level IIB with odds of a level II node being located in level IIB. The association was strongest for accessory nerve compared to sternocleidomastoid, posterior belly of digastric, and calculated area; it was statistically significant but the residual variation in the data, as measured by the c-statistic, was substantial. The regression model incorporating accessory nerve length predicted the proportion of level II nodes found in level IIB, showing for example that a 1.5 cm accessory nerve would predict 26% of level II nodes being located in level IIB, whereas a 3.5 cm accessory nerve would predict 48% of nodes being located in level IIB. Linear regression analysis on the number of IIB nodes resulted in a prediction of two more level IIB nodes for every additional 1 cm length of accessory nerve, as shown in Figure 2. Neither height nor sex predicted the number of level IIB nodes.

TABLE IV.

Univariate Analysis of Odds of a Level II Node Being Located in Level IIB.

Variable	Odds Ratio	95% Confidence Interval	c-Statistic	P
Accessory	1.59	1.41–1.80	0.591	<0.0001
Sternocleidomastoid	1.49	1.33–1.68	0.584	<0.0001
Post belly digastric	1.63	1.31–2.03	0.550	<0.0001
IIB Area	1.42	1.27–1.59	0.573	<0.0001

Fig. 2.

Scatterplot of number of nodes found in each level IIB by length of accessory nerve, with shading representing the 95% regression confidence interval (CI).

Eighty operated sides were clinically N0, 53 had one clinical node ≤3 cm, 8 had one node >3 cm, and 43 had multiple positive nodes. Preoperatively, 64 sides were known to have clinically evident metastatic level II nodes. On univariate analysis, no patient or anatomic factor correlated with the presence of one or more metastatic node in level IIB. Of the 120 neck sides without known level II metastasis, dissection yielded metastatic nodes in five cases. Univariate logistic regression analyses evaluating predictors of level IIB metastasis did not identify predictive patient or anatomic factors. With both analyses, regression computation failed when the primary site variable was included.

Considering the overall number of level IIB nodes and the number of metastatic IIB nodes with respect to the measured dimensions, it did not appear that there was a reasonable length cutoff below which level IIB could be routinely left undissected.

Of the 50 patients to whom the NDII survey instrument was provided, 43 patients completed a postoperative survey, with five lost to follow up and unreachable by telephone and two deceased. Median time between surgery and survey completion was near 5 months. The median NDII standard score was 92.5 (interquartile range 80–97.5). We did not find significant relationships between NDII scores and the dimensions of level IIB, nor with age, body mass index, height, prior radiotherapy, or primary site. Figure 3 shows a plot of NDII scores by accessory nerve length, with regression predicting NDII scores lower by 4.7 points for each additional 1 cm of accessory nerve, p = 0.34. Note that the outlier observation with a high burden of neck symptoms was not strongly influential in this analysis.

Fig. 3.

Scatterplot of Neck Dissection Impairment Index (NDII) composite score by accessory nerve length, with shaded area representing the 95% regression confidence interval.

DISCUSSION

Oncologic surgery requires calculation of the risks and benefits of certain procedures. Taking wider margins and removing more structures might lower recurrence and mortality risks at costs related to functional and other important outcomes. In this spirit, abundant literature addresses the question of whether level IIB should be included with every dissection of level II or if it can safely be left in place. Some studies seek to describe the magnitude of the benefit by reporting the frequencies at which metastatic nodes are present in level IIB under a variety of circumstances. Other studies have assessed the risk by describing functional changes related to manipulation of the accessory nerve or other nearby structures.²⁴ The variation in level IIB surgical anatomy is a factor that has not been included in these studies.

Anatomic Findings

The variability in the length of the accessory nerve and the other dimensions of level IIB are not well described. The measurements reported in this study are helpful to surgeons in executing standard lymphadenectomy operations, informing the intraoperative identification of the accessory nerve. Treves and Keith’s Surgical Applied Anatomy describes the accessory nerve entering the sternocleidomastoid muscle “about 1 inch” from the mastoid tip.²⁵ According to Tubbs et al., other sources describe the distance as “2–3 fingerbreadths,” which may be up to 5 cm.²⁶ This study measured the corresponding mean distance along the sternocleidomastoid at 2.3 cm, with 2.7 cm representing the 75th percentile and 4.2 cm the maximum observed in 150 cases. These measurements relate to surgical exposure. During parotid surgery, the mastoid tip is exposed in the middle of the field, whereas that same landmark is the superior limit of exposure during most other neck dissections. Visualization of the level II region in a laboratory setting would involve a wider exposure of the upper neck than during standard surgery, and this might explain variation in measurements.

Abakay et al. in a cadaver study reported the variation in distance between the mastoid tip and the point of emergence of the accessory nerve at the posterior border of the sternocleidomastoid,²⁷ which interestingly showed an association with the patient’s height. The present study’s focus on anterior triangle measurements is more relevant to common neck dissections for upper aerodigestive tract cancers, and we did not find a significant association with height.

Distribution of Nodes Into Level IIA and IIB

The number of nodes identified in level IIB is consistent with prior findings. In the systematic review reported by Paleri et al.,²⁰ data from 12 studies were presented with a weighted average of 4.9 nodes identified per level IIB. Previous studies of level II metastasis and the value of level IIB dissection have not accounted for the relative sizes of level IIA and IIB. Our anatomic data confirm the intuitive notion that level II compartments with longer accessory nerves have a greater proportion of nodes found in level IIB versus level IIA. From the perspective of ensuring that an adequate number of nodes are removed during neck dissection, the findings of this study mean that a longer length of accessory nerve should prompt surgeons to remove the level IIB specimen.

Distribution of Metastatic Nodes

We found metastatic nodes in 12 of 184 (6.5%) sides. This is consistent with prior studies combining N0 and N+ necks, such as that by Hoyt et al.²⁸ Most studies of level IIB relate to the presence of clinically evident metastases in level IIA and to the tumor primary site.^{17–19,29–31} Despite those efforts, there is no consensus on when the dissection can be avoided. Myriad studies have found little benefit to dissecting a clinically negative level IIB based on the low likelihood of occult metastasis, but surgeons may be reluctant to leave it behind because of the irreparable downside of missing an occult metastasis. Our study unfortunately does not shed great light on the question. Given the small number of cases with level IIB metastasis, our study could not demonstrate the safety of skipping the dissection, nor did it show clear value of routine dissection.

Level IIB Anatomy and Postoperative Symptoms

Dziegielewski et al. reported the results of a randomized study that showed level IIB-sparing neck dissection to be associated with better postoperative NDII scores.³² No prior study has evaluated differences in patient reported neck symptoms based on the dimensions of structures in level II. Competing effects make it difficult to hypothesize even the directionality of these differences. A short accessory nerve length makes for limited access to the posterosuperior extent of level IIB, and dissection requires use of grasping instruments and cautery in proximity to the nerve. Conversely, if devascularization of the accessory nerve is the central cause of postoperative shoulder symptoms,³³ a longer nerve length predicts a longer devascularized segment and more severe symptoms. Our study showed that the length of the nerve did not account statistically for the variation in symptom burden. That the slope of the regression line in Figure 2 was not statistically different from 0 means dissection-related traction and devascularization were equally important factors. Surgeons should not make decisions about level IIB dissection out of concern that a short nerve would result in more severe postoperative impairment. Our study excluded patients having level V dissection, which has better-described adverse effects on nerve function and symptoms. Our findings imply that the level V dissection would have a greater effect on symptoms and objective shoulder outcomes than a decision about managing level IIB.

Limitations

Surgical exposure of the upper neck is different in cases ranging from parotidectomy to laryngectomy. Because of that, there may have been systematic differences in both the anatomic measurements and in the count of nodes in levels IIA and IIB. This study attempted to minimize that problem by including patients operated by a single surgeon, so therefore minimal variation in how the cases were done. Other surgeons have different routines and might measure different numbers. Given that the counts of IIB nodes and rate of finding metastatic IIB nodes were similar to those reported in other studies, it is reasonably likely that other surgeons would draw similar conclusions.

A limitation of the patient reported outcome analysis is that we did not explicitly account for outcomes related to the cervical plexus, injury to which might have affected NDII outcomes through its variable trapezius innervation.³⁴ The standard surgical plan employed in these cases included preservation of the cervical plexus with particular attention to branches which may join the distal accessory nerve to supply trapezius. Decisions about the cervical plexus were as standardized as possible across the study. Another limitation was the short follow up between surgery and the NDII questionnaire. Our study’s median interval of near 5 months was shorter than the 6 month interval used by Dziegielewski et al., after surgery. Other literature indicates that many patients’ neck symptoms have resolved by 6 months after surgery,³⁵ so our early assessment may actually overestimate the true long-term symptom burden. The study attempted to characterize the physical dimensions of level IIB by making a planar set of three measurements. However, level IIB is a three-dimensional volume that is not completely described by the three measurements we made. We included data collection on patients’ height and weight to account for that fact, as there is no natural and repeatable method of measuring the missing dimensions in the surgical setting.

CONCLUSION

Our findings establish the dimensions of the borders of neck level IIB and their variations. In patients in whom level IIB is larger, a higher proportion of nodes was found in level IIB. The data reported here do not allow surgeons to be secure in omitting level IIB dissection based on the measured length of the accessory nerve across level II. In addition, the length of the accessory nerve was not a significant predictor of postoperative neck symptoms.