Biomedical Research

Journal Banner

Patterns of retropharyngeal lymph node metastasis in nasopharyngeal carcinoma

Qifeng Hua1, Jianjun Zheng1, Bibo Hu1, Ming Shu2*, Li Shen1 and Jiebo Chen3

1Department of Radiology, Ningbo No.2 Hospital, Ningbo 315010, PR China

2Department of the Third Surgery, Ningbo No.2 Hospital, Ningbo 315010, PR China

3Department of Radiation Therapy, Ningbo No.2 Hospital, Ningbo 315010, PR China

*Corresponding Author:
Ming Shu
Department of the Third Surgery, PR China

Accepted on May 24, 2017

Visit for more related articles at Biomedical Research


This study aimed to investigate the patterns of retropharyngeal lymph node (RLN) metastasis in patients with nasopharyngeal carcinoma (NPC). Sixty-two patients with NPC underwent nasopharyngeal magnetic resonance imaging. Magnetic resonance images were reviewed collectively by the research team. Nodal involvement was classified according to the Radiation Therapy Oncology Group nodal classification guidelines. Locations of the involved RLNs were recorded, and the relationships of RLN metastasis with primary tumor extension and clinical stage were analyzed. Among 62 patients with NPC, 47 (75.8%) exhibited level IIb nodal metastasis, and 39 (62.9%) had RLN metastases. Overall, 60 metastatic RLNs were detected. The center, cranial, and caudal borders of metastatic RLNs respectively included four (6.7%), 17 (28.3%) and three (5.0%) lymph nodes in the occipital bone; 45 (75.0%), 40 (66.7%), and 33 (55.0%) lymph nodes in the first cervical vertebra (C1); 10 (16.7%), three (5.0%), and 21 (35.0%) lymph nodes in the second cervical vertebra (C2); and zero (0.0%), zero (0.0%), and three (5.0%) lymph nodes in the third cervical vertebra (C3). The incidence of RLN involvement did not increase with the incidence of primary tumor invasion to the parapharyngeal space, prevertebral muscle, oropharynx, nasal cavity, skull base, paranasal sinuses, and cavernous sinus. The incidence of nodal metastasis among patients with NPC was highest at the cervical lymph node level IIb, followed by the RLNs. There was a linear decrease in the incidence of metastatic lateral RLNs from levels C1 to C3.


Nasopharyngeal carcinoma, Retropharyngeal lymph node, Metastasis, Pattern.


Nasopharyngeal carcinoma (NPC) is one of the most common types of malignant tumors, and, as the nasopharynx is rich in lymphatic drainage, 60-90% of affected patients present with lymph node metastasis prior to treatment. Treatment efficacy depends on tumor T stage, N stage, histological type, adjacent invasion, tumor size, and selected treatment programs [1]. Currently, radiotherapy is the mainstay of treatment for NPC, and the overarching principles of lymph node metastasis have been a significant factor in radiation field design. A study by Kato et al. indicated that magnetic resonance imaging (MRI) exhibited a significantly better ability than computed tomography (CT) in detecting retropharyngeal lymph node (RLN) metastasis [2]. Soft tissue is more readily visible on MRI; hence, MRI can present lymph node metastasis and NPC clearly and completely. MRI has been widely applied for the diagnosis of NPC, largely because it decreases radiation exposure. In 2008, Liao [3] reported a significant difference between CT and MRI in distinguishing invasion of the base of the skull (CT 31. 0%, MRI 52. 6%) in 420 patients diagnosed with NPC. Zhan et al. [4] found that the positive rates of invasion were 48.3% and 33.3% using MR and CT, respectively. Therefore, MRI has more advantages in distinguishing invasion of the base of the skull, and it can be used to accurately estimate invaded range and accurately sketch the tumor target.

In this study, the principles of RLN metastasis in NPC were analyzed using MRI. According to the literature, the cervical lymph nodes and RLNs are the most common sites of metastasis. Although pharyngeal lymph node metastasis is common in NPC, its prognostic value is not clear, and the current staging system does not specify whether RLNs should be subject to T or N staging. Therefore, studies of the principles of RLN metastasis are important not only when determining regions that should be targeted with radiotherapy in cases of NPC, but also for clarifying their prognostic value in staging systems.

Materials and Methods

Clinical data

Sixty-two patients with pathologically confirmed NPC underwent initial treatment at our hospital between November 2009 and April 2010. The patients had a median age of 49 years, and a male: female ratio of 3.4:1 (48 men, 14 women). Before treatment, all patients underwent nasopharyngeal MRI, chest radiography or CT, and abdominal B-ultrasonography. Patients with stage T3-4 disease, lymph node diameter >4 cm, or supraclavicular lymph node metastasis also underwent whole-body bone scans. This study was conducted in accordance with the declaration of Helsinki. This study was conducted with approval from the Ethics Committee of Ningbo No.2 Hospital. Written informed consent was obtained from all participants.

MRI examination

A SIEMENS 1.0 T device (Siemens, Munich, Germany) and head and neck coil were used for dual gradient MRI as follows: for cross-sectional and sagittal T1-weighted imaging, the repetition time (TR) was 400-500 ms and the echo time (TE) was 10-15 ms; for cross-sectional T2-weighted imaging, the TR and TE were 4000-5000 ms and 80-100 ms, respectively. Imaging was repeated after an intravenous injection of gadopentetic acid contrast agent (Gd-DTPA; 0.1 mmol/kg); specifically, cross-sectional and coronal T1-weighted fast spoiled gradient echo fat suppression sequence scans (TR: 150-250 ms, TE: 2-10 ms) were performed using the following settings: cross-sectional thickness: 6 mm; interlayer spacing: 1 mm; sagittal and coronal plane thickness: 4 mm; and interlayer spacing: 1 mm.

Image reading

All MRI scans were read collectively by two head and neck radiologists. The evaluation criteria for lymph node metastasis were as follows: 1) smallest neck lymph node diameter ≥ 10 mm; 2) minimum lateral group RLN diameter ≥ 5 mm; 3) any visible RLN in the outer group; 4) lymph nodes distributed in a cluster; and 5) presence of central necrosis or invasion beyond the lymph node capsule. The cranial and caudal borders and center position (reference: cervical spine) were recorded. Cervical lymph node positions were determined using the methods suggested by Roy et al. [5].

Statistical analysis

All data were entered into the SPSS 11.5 software (SPSS, Inc., Chicago, IL, USA), and the chi-square and Fisher’s exact tests were used to analyze the relationships among RLN metastasis, primary tumor, surrounding tissue involvement, and clinical staging. A P value<0.05 was considered statistically significant.


Overall situation of lymph node metastasis

Of the 62 evaluated patients, 39 (39/62, 62.9%) had RLN metastases, 47 (47/62, 75.8%) had level IIb lymph node metastases, and 54 (54/62, 87.1%) had some type of lymph node metastasis. Four patients (4/62, 6.5%) had only RLN metastasis, 15 (15/62, 24.2%) had only cervical lymph node metastasis, and 36 (36/62, 58.1%) had both RLN and cervical lymph node metastases. The rate of RLN metastasis was significantly lower than that of cervical lymph node metastasis (62.9% vs. 82.3%, P=0.028).

RLN metastasis

Of the 39 patients with RLN metastasis, 22 (22/39, 56.4%) had unilateral RLN metastasis, and 17 (17/39, 43.6%) had bilateral RLN metastasis. Sixty metastatic RLNs were located in the outer group; no lymph node enlargement was visible in the inner group. The average minimum and maximum transverse diameters and longitudinal length of the metastatic RLNs were 1.5 ± 0.6 cm (range: 0.5-3.6 cm), 1.6 ± 0.7 cm (0.8-4.8 cm), and 2.5 ± 1.4 cm (0.9-6.8 cm), respectively. The center and cranial and caudal borders (reference: cervical spine) in the 60 patients are shown in Table 1. Overall, the RLN metastasis rate gradually reduced from the first cervical vertebra (C1) to the third cervical vertebra (C3) level (Fisher’s exact test, P=3.17 × 10-4, P<0.05).

Locations on MR Central location of lymph node Cranial border of lymph node Caudal border of lymph node
Occipital bone 4 (6.7%) 17 (28.3%) 3 (5.0%)
The first cervical vertebra 45 (75.0%) 40 (66.7%) 33 (55.0%)
The second cervical vertebra 10 (16.7%) 3 (5.0%) 21 (35.0%)
The third cervical vertebra 0 (0%) 0 (0%) 3 (5.0%)

Table 1. Location of 60 metastatic RLN.

Of 39 patients with 60 RLNs, 20 (20/39, 51.3%) had one RLN, 17 (17/39, 43.6%) had two RLNs, and two (2/39, 5.1%) had three RLNs. A total of 18 RLNs (18/60, 30%) had central necrosis. The diameters of necrotic RLN ranged from 0.5 cm to 3.6 cm. No necrosis was found in RLNs with diameters<0.5 cm.

Relationship between lymph node metastasis and tumor invasion site

Data describing the relationship between lymph node metastasis and tumor invasion site are presented in Table 2. Notably, RLN metastasis did not correlate with invasion of the pharyngeal space, nasal cavity, oropharynx, palate sail muscle, head longus, pterygoid and wing muscle, skull base, pterygopalatine fossa, infratemporal fossa, paranasal sinuses, eye orbital, cavernous sinus, or other sites. Additionally, we did not observe a relationship between RLN metastasis and staging, which might be attributable to the small number of cases (Table 3).

Tumor involved sites Number of positive RLN (%)
No violation Violation χ2 value P value
Parapharyngeal space 18/32 (56.3) 21/30 (70.0) 1.255 0.263
Sail palate muscle 22/33 (66.7) 18/29 (62.1) 0.143 0.706
Oropharynx 37/58 (63.8) 2/4 (50.0) 0.000 0.986
Nasal cavity 32/50 (64.0) 8/12 (66.7) 0.000 1.000
Prevertebral muscle 26/41 (63.4) 13/21 (61.9) 0.014 0.907
Wing extrafusal 35/54 (64.8) 4/8 (50.0) 0.174 0.676
Pterygoid an flap 33/51 (64.7) 7/11 (63.6) 0.000 1.000
Rupture hole 32/50 (64.0) 7/12 (58.3) 0.133 0.715
Slope 27/43 (62.8) 12/19 (63.2) 0.001 0.978
Foramen ovale 33/52 (63.5) 6/10 (60.0) 0.000 1.000
Temporal bone petrous apex 30/48 (62.5) 9/14 (64.3) 0.015 0.903
Pterygopalatine fossa 37/58 (63.8) 2/4 (50.0) 0.000 0.986
Sphenoid wing 34/52 (65.4) 6/10 (60.0) 0.000 1.000
Sphenoid 33/51 (64.7) 6/11 (54.5) 0.400 0.527
Ethmoid 38/59 (64.4) 2/3 (66.7) 0.000 1.000
Maxillary 37/59 (62.7) 2/3 (66.7) 0.000 1.000
Orbital apex 38/60 (63.3) 1/2 (50.0) 0.000 1.000
Cavernous sinus 35/54 (64.8) 4/8 (50.0) 0.174 0.676
Temporal fossa 37/59 (62.7) 2/3 (66.7) 0.000 1.000

Table 2. Relationship between RLN metastasis and primary NPC involved scope.

RLN Number of patients (%)
  T1 T2 T3 T4
Negative 5/13 (38.5) 8/24 (33.3) 4/13 (30.8) 5/12 (41.7)
Positive 8/13 (61.5 16/24 (66.7) 9/13 (69.2) 7/12 (58.3)
Negative 2/2 (100.0) 7/16 (43.8) 8/26 (30.8) 6/18 (33.3)
Negative 0/2 (0.00) 9/16 (56.2) 18/26 (69.2) 12/18 (66.7)

Table 3. Relationship between RLN metastasis and staging.


Radiotherapy is the treatment of choice for NPC, and it is even preferred over surgery. Therefore, it is difficult to perform pathology via lymph node dissection, and the reported 5-year post-radiotherapy survival rate among patients with NPC is as high as 83.3% [6-8]. Due to the application of intensity-modulated radiotherapy, the 5-year survival rate has improved to 78-83.3% [9-10]. Herein, all cases with long-term follow-up may be used to assess the diagnosis of lymphatic metastasis, and relevant results may validate the accuracy of lymphatic metastasis by excluding the history of other potential tumors. Therefore, it is difficult to define the relationship between pathological diagnosis and imaging performance of lymph node metastasis. Previous studies have found that RLNs shrink with age; specifically, Ng et al. [11] and Ogura et al. [12] reported that on MRI scans, the minimum RLN transverse diameters among patients aged 6-19 years, 20-38 years, and 42-74 years were 6.4 ± 1.4 mm, 4.2 ± 1.1 mm, and 3.2 ± 1.0 mm, respectively. The minimum transverse diameters of normal RLNs on MRI scans in our study were ≤ 4.5 mm. Therefore, the present study used a minimum RLN diameter ≥ 5 mm as a diagnostic criterion for metastasis. Although some authors have used central necrosis as a diagnostic criterion for RLN metastasis, we seldom observed this phenomenon in lymph nodes with small diameters. The present study suggests that these standards cannot improve the sensitivity of RLN diagnosis. In this study, 62.9% patients had RLN metastases, consistent with the rate of 63.6% reported in the literature. RLNs include the outer and inner groups; the former is located outside the pharyngeal constrictor muscle and vertebral anterior muscle and inside the internal carotid artery (Figure 1). The latter is located near the body midline, close to the posterior pharyngeal wall. In this study, all assessed RLNs were located in the outer group. Because the RLNs in the inner group were not normally visible, any visible RLNs in this group on MRI were considered metastatic; however, a previous article reported that NPC with RLN metastases in the inner group was very rare.


Figure 1. Schematics of RLN metastasis in outer group. A: T1- weighted MR images; B: T1-weighted fat-suppressed enhancement MR image.

King et al. [13] and Chong et al. [14] described the locations of RLNs in detail. Chong et al. [14] found that the majority of RLN metastases were located above the level of the second cervical vertebra, and King et al. [13] found that 63% of RLNs extended down to the level of the second cervical vertebra, 19% reached the second and third cervical levels, and 6% reached the level of the third cervical vertebra. Differences between the two studies were mainly attributable to differences in the methods used to record RLN positions. Wang et al. [15] described in detail the longitudinal center diameters and cranial and caudal borders in 618 patients with NPC and RLN metastases. Studies have shown that most metastatic lymph nodes were located in the first cervical vertebral level, regardless of the center or the cranial and caudal borders, and that the RLN metastasis rate gradually decreased from the first to the third cervical vertebral levels. Som et al. [16] first proposed defining RLNs as those located under the skull base to within a 2 cm range of the skull base, according to imaging. However, Wang et al. [15] and Kao et al. [17] observed that many RLN metastases in NPC cases exceeded this range on CT, and Ogura et al. [12] and additional MRI results from the present study indicated that the range of RLNs as defined by Som et al. [16,18-20] was too small and not suitable for application to NPC. The study group also observed multiple ipsilateral RLN metastases located between the base of the skull to the lower edge of the third cervical vertebra; accordingly, the RLN irradiation range in patients with NPC should extend from the base of the skull to the lower edge of the third cervical vertebral level.

Although previous research has shown correlations between the lymphatic drainage pathway in the head and neck and the primary tumor and the location and scope of tumor invasion, the risk of RLN metastasis was found to increase when the tumor extended beyond the pharyngobasilar fascia. Liu et al. [21,22] reported a significantly lower RLN metastasis rate among T1 patients, compared to T2-4 patients (48.3% vs. 71.0%), and observed that RLN metastasis increased significantly when the sail palate muscle, oropharynx, prevertebral muscle, pterygoid muscle, and styloid gap were involved. Wang et al. [15] also reported that the RLN metastasis rate increased significantly with parapharyngeal space involvement, but did not increase when the nasal cavity, oropharynx, head longus, skull bone, paranasal sinuses, eye orbital, cavernous sinus, or other sites were involved. However, in the present study, we did not observe a similar relationship, which might be attributable to the small number of cases (Table 3).

The effect of RLN metastasis on NPC prognosis was inconclusive. At present, both domestic and international staging systems do not clearly define RLN etiologies, which vary among tumor centers. Given the high incidence of NPC and higher NPC metastasis rates in Southeast Asia and the southeast coastal areas of China, in-depth studies regarding the prognostic significance of the RLN metastasis rate are required. Some authors observed a higher distant metastasis rate among patients with NPC and RLNs; however, in a CT-based study by Chua et al. [23-30], which used a transverse diameter ≥ 10 mm as a diagnostic criterion of RLNs, RLN metastasis was not found to have an effect on the prognosis of NPC. Among the 62 patients in this group, the vast majority of patients with RLN metastasis also had cervical lymph node metastasis (Figure 2); only four patients (6.5%) had RLN metastasis alone, and eight (12.9%) patients had no lymph node metastases. Therefore, a longer period is needed to accumulate more cases and better evaluate the impact of RLN metastasis on the prognosis of NPC during long-term follow-up.


Figure 2. A) T1 B-weighted MR image; B) Coronal fat-suppressed enhancement T2-weighted MR image showed bilateral posterior RLN metastasis.

In short, the highest rate of metastasis of NPC was observed in level IIb lymph nodes, followed by RLNs. The majority of RLN metastases were located in the outer group; metastasis in the inner group was rare. The rate of RLN metastasis decreased gradually from the first to the third cervical vertebral levels. RLN metastasis had no significant correlations with the sites of NPC involvement or staging. This finding is inconsistent with other reports from larger sample sizes. We believe that the transfer of lymphonodi retropharynici is not correlated with invaded areas in nasopharyngeal cancer and neoplasm staging. Moreover, except for specific regions in Ningbo located in the coastal city, it has certain characteristics. Despite its small sample size, this study findings support confirmation with a larger sample, which may improve radiotherapy treatment in patients with nasopharyngeal carcinoma lymph node metastasis.

Our study had some limitations. First, it was a retrospective follow-up study, and the majority of patients had no pathological diagnosis. Second, the sample size was small, and statistical errors were possible. Third, we must recognize the limitations of MRI and note that some cases had false-positive or false-negative clinical performances.


This study was supported by Leader and Top-Notch Talent Project of Ningbo city (2012-131).