Summary of results
Nasopharyngeal cancer (NPC) has a high propensity of cervical node metastasis. The results of this meta-analysis based on 13 clinical trials using MRI for diagnosis and staging for NPC revealed that the most commonly involved cervical lymph node regions include lateral retropharyngeal nodes and level II nodes with an overall probability of 69.4% & 70.4% respectively for metastasis. These first echelon nodal groups are followed by levels III, VA, and IV, with probabilities of 44.9%, 26.7%, 11.2%, respectively. Certain cervical lymph node groups, including level I, level VI, parotid and supraclavicular nodes have a very low risk for metastasis. An important finding was that lymphatic spread in cervical nodal chain from NPC primary follows an orderly fashion. There was a very low risk of 0.5% in skip nodal metastasis [
14].
These findings are important for the management of NPC, particularly in defining proper treatment fields for definitive radiation therapy using conformal technology such as intensity-modulated radiation therapy (IMRT). As the subclinical involvement of cervical lymph nodes cannot be reliably detected by image studies including CT, MRI, and/or PET/CT, proper selection and delineation of clinical target volume for elective irradiation represents a major challenge. Bilateral cervical lymph node metastases usually occur in the early phase of disease development. Therefore, irradiation of the entire cervical lymphatic draining region has been a common practice in radiotherapy of NPC, including stage I disease,[
3,
28,
29] with radiation portals encompassing all levels of cervical lymph nodes from IB to V, including the supraclavicular region [
30]. Despite improved outcomes in terms of locoregional control and disease-free survival rates with IMRT,[
31,
32] such a treatment strategy might represent over-treatment using the current diagnostic and therapeutic technology. NPC patients who are cured of their disease may suffer from the long-term complications from treatment,[
5] including xerostomia, neck fibrosis, telangiectasia, thyroid dysfunction, brachial plexopathy and second malignancies, which can significantly impact on function, quality of life or life expectancy. While some of these side effects have been minimized with the advent of conformal radiotherapy, they cannot be fully prevented, especially if the nodal clinical target volumes are adjacent to the critical structures.
An effective strategy in reducing treatment-induced morbidity is to minimize the field for elective radiation in the uninvolved neck region. In a recently reported study by Lin
et al, exclusion of level Ib lymph nodes and supraclavicular region for elective treatment in IMRT for locoregionally advanced NPC did not reduce the probability of regional control rate as compared to historic controls [
33]. Furthermore, in a recently published study of more than 400 NPC patients with N0 disease who were treated with definitive dose of radiation to the primary and upper neck fields (levels II, III, and Va) only, recurrence out of the radiation field at the level IV neck region and supraclavicular area occurred in only one patient [
34]. A similar retrospective review of 924 NPC patients with N0 disease compared the inferior border of radiotherapy either at the cricoid cartilage or below the cricoid cartilage revealed no statistically significant difference between the two groups [
14]. The use of MRI in the diagnosis and staging and/or more advanced treatment strategies including IMRT and concurrent chemoradiation therapy might play substantial roles for the aforementioned findings. However, the optimal strategy of selection and delineation of the sub clinical regional disease in clinical target volume (CTV) in the treatment of NPC has not been well addressed. Knowledge on regional lymph node drainage in NPC diagnosed and staged in the modern era particular with MRI is limited, and the current available data are usually not complete with inconsistent results. More systemic and comprehensive understanding of the patterns of cervical nodal involvement in NPC is clearly necessary for proper design of clinical trials using conformal radiation techniques and will provide practice-changing clinical evidence.
Although our results represented the most comprehensive and conclusive data for the pattern and probability of cervical lymph node spread in NPC, a number of issues related to the design and analyses need to be addressed. As radiation therapy is the only curative treatment currently, and surgery including neck dissection has a limited role in the primary treatment for NPC, adenopathy is universally diagnosed by imaging studies. Histological diagnosis for cervical node metastasis is rarely performed. The radiology diagnosis of cervical lymph adenopathy is largely based on size and morphology criteria [
35] derived from surgical series. Prior to the use of MRI for staging and diagnosis, NPC was usually evaluated using contrast enhanced CT to assess the extent of disease in both primary and neck regions [
36]. Most of the available data on neck node involvement and its treatment are based on CT imaging. However, the sensitivity and specificity rates of enhanced CT for the diagnosis of cervical lymph adenopathy are approximately 14-60% and 78-92% respectively, compared to 29-80% and 82-92% respectively for MRI [
35]. The benefit of MRI over CT in evaluating cervical lymph nodes for NPC has also been recently reported [
37]. Additionally, a meta-analysis has shown that the accuracy of MRI appears superior to PET-CT in the evaluation of cervical lymphadenopathy [
38]. Accordingly, the updated 7
th edition of the TNM Classification of Malignant Tumours [
39] proposes that MRI should be considered the standard imaging modality for the diagnosis and staging of NPC.
In addition, a locally advanced tumor in the nasopharynx obviously would have a higher probability of cervical lymph node metastasis as compared to early stage disease. The knowledge of the exact NPC stage of each study subject could provide valuable insight as to how nodal metastasis changes with staging of the tumor. Unfortunately, this information was not readily available for such analyses.
One of the greatest challenges in performing this meta-analysis was the weighting of the individual studies. The studies were highly heterogeneous in their study design from prospective to retrospective (predominant). Many of them were not designed to study regional nodal metastasis as the main end point but were included due to the paucity of such data in the MRI era. As such, the usual criteria for weighting studies in systematic reviews could not be applied to this study [
40,
41].