Radiologic Criteria of Retropharyngeal Lymph Node Metastasis in Nasopharyngeal Carcinoma Treated with Radiation Therapy

Introduction

Nasopharyngeal carcinoma (NPC) often involves regional lymph nodes. Retropharyngeal lymph nodes (RLNs) represent first-echelon lymph nodes typically involved in NPC (1–3). RLNs are classified in medial and lateral groups. The prognostic importance and role of metastatic lateral RLNs in published staging systems are ambiguous (4–8). Researchers in studies reported that lateral RLN metastases indicated a poor prognostic outcome in NPC patients and proposed that metastatic RLNs should be classified as N1 lesions (7,8). However, issues about the radiologic criteria of metastatic lateral RLNs have not yet been thoroughly resolved.

Investigators in previous studies (2,6,9–11) either have used the radiologic criteria of cervical node metastases or have used the upper limit of benign RLNs or the lower limit of malignant RLNs that were based on data from smaller series of patients in whom surgery or follow-up imaging was performed as the diagnostic criteria of RLN involvement. In magnetic resonance (MR) imaging studies by King et al (2) and Ma et al (9), the researchers focused on the size of benign RLNs in nondiseased patients and found that normal RLNs ranged from 1.5 to 2.5 mm or 1.5 to 4.5 mm in minimal axial diameter; as such, a minimum axial diameter of 4 mm or larger or 5 mm was recommended as the size criterion for metastatic RLNs in these two reports, respectively. However, large nodes can be reactive and not metastatic, and small nodes can contain metastases (12,13); thus, one should be wary of using the lower limit for malignant nodes or the upper limit for benign nodes as diagnostic criteria. Given these uncertainties, it remains important to optimally define the size criteria of RLN involvement according to data collected from head and neck cancer patients.

We are unaware of existing data in regard to the response of RLNs to radiation therapy (RT) or the clinical importance of MR imaging follow-up of RLNs after RT in NPC patients. The current study concentrated on the radiologic response of RLNs in NPC patients treated with definitive RT. Determination of the nature of the RLNs was based on the results of MR imaging follow-up. The objective of this study was to determine the radiologic criteria of metastatic RLNs in patients with NPC.

Materials and Methods

Patients

Institutional review board approval was obtained for the study. All examinations were performed after written informed consent was obtained from patients or their next of kin. From July 1, 2003, to March 31, 2005, 303 ethnic Chinese patients (228 male patients [mean age, 46 years; range, 13–73 years], 75 women [mean age, 44 years; range, 21–68 years]) with newly diagnosed, untreated, and nondisseminated NPC were enrolled in our study. Patients with additional known head and neck cancers or acute inflammation were excluded from this study. All patients underwent a pretreatment evaluation that included MR imaging of the neck and nasopharynx, chest radiography, abdominal ultrasonography, and a whole-body bone scan (T3 disease, metastasis of the area below the inferior margin of cricoid cartilage [C6], or lymph node size of ≥30 mm). When a metastatic lesion was suggested, confirmatory findings of contrast material–enhanced computed tomographic or MR imaging studies were used for corroboration. The medical records and findings of imaging studies were reviewed, and in all patients, staging was performed according to the 2002 American Joint Committee on Cancer staging system (14). T category classification was as follows: T1 lesions, 86 patients; T2a lesions, nine patients; T2b lesions, 48 patients; T3 lesions, 116 patients; and T4 lesions, 44 patients. N category classification was as follows: N0 lesions, 109 patients; N1 lesions, 118 patients; N2 lesions, 60 patients; N3a lesions, nine patients; and N3b lesions, seven patients. The characteristics of these patients are presented in Table 1.

Table 1 Characteristics of 303 Patients with NPC

Note.—AJCC = American Joint Committee on Cancer, WHO = World Health Organization.

^*Numbers in parentheses are percentages.

Imaging Protocol

All patients underwent MR imaging with a 1.5-T system (Signa CV/i; GE Healthcare, Milwaukee, Wis) employing spin-echo technique. The region from the suprasellar cistern to the inferior margin of the sternal end of the clavicle was examined with a head and neck–combined coil. T1-weighted images in the axial, coronal, and sagittal planes (repetition time msec/echo time msec, 575/17), and T2-weighted images in the axial plane (2850/102) were obtained before injection of the contrast material. After intravenous injection of gadopentetate dimeglumine (Magnevist; Bayer Schering Pharma, Berlin, Germany) at a dose of 0.1 mmol per kilogram of body weight, T1-weighted fat-suppressed axial, coronal, and sagittal sequences were performed sequentially, with parameters similar to those used at imaging before injection of gadopentetate dimeglumine. The section thicknesses and intersection gaps were, respectively, 5 mm and 1 mm for the axial plane and 6 mm and 1 mm for the coronal and sagittal planes.

Image Assessment

Two radiologists (L.Z.L., Y.Z.L, with 6 and 8 years of experience in NPC MR imaging, respectively, at the time of the study) separately evaluated the MR images. Any disagreements were resolved in consensus. The RLN assessment included the medial and lateral groups. RLNs were identified by separating them from the primary tumor. The lymph nodes that were contiguous with the primary tumor were clearly identified by a contrast-enhancing rim or a difference in signal intensity compared with that of the primary tumor. The minimal and maximal axial (perpendicular to the course of the internal jugular vein) diameters and the longitudinal diameter of each visible RLN were measured. The minimal axial diameter corresponded to the widest diameter of the lymph node in the axial plane that was perpendicular to the maximal axial diameter. The longitudinal diameter paralleled the course of the internal jugular vein (15). Axial diameter measurements were made on T2-weighted axial images, and longitudinal diameter measurements were made on unenhanced T1-weighted coronal images (9,16).

A criterion for the diagnosis of nodal central necrosis on MR images was a focal area of high signal intensity on T2-weighted images and a focal area of low signal intensity on T1-weighted images, with or without a surrounding rim of enhancement (17). A criterion for groups of RLNs was that there were two or more lymph nodes in the ipsilateral retropharygeal space.

Treatment

All patients were treated with definitive-intent RT. Most of the patients (n = 186, 61.4%) were treated with conventional techniques, but 102 (33.7%) patients were treated with intensity-modulated RT, and 15 (5.0%) patients were treated with three-dimensional conformal RT. Details regarding the RT technique performed at the cancer center of Sun Yat-sen University have been reported previously (18–20). In our cancer center, all RLNs that were visible on MR images were incorporated into the gross tumor volume when we drew the target volume, and they received the same high dose of 68–70 Gy.

A total of 183 (60.4%) patients with local or regional advanced disease (classified as T3 or T4 lesions or N2 or N3 lesions) received neoadjuvant, concomitant, or adjuvant chemotherapy. When possible, salvage treatments (including afterloading, surgery, and chemotherapy) were provided in the event of a documented disease recurrence or when the disease persisted.

Follow-up and Assessment of RLNs

During RT (at 46–50 Gy) and 14–30 days after the completion of RT, the tumor response was assessed by using MR imaging and flexible nasopharyngoscopy. Residual primary tumors, cervical lymph nodes, or RLNs were noted, and close follow-up was performed through further clinical study and MR imaging. All patients were regularly followed up for at least 30 months. The maximum follow-up in these patients was 51 months, and the mean follow-up was 42 months.

The nature of the RLNs was judged jointly by the NPC research team (G.Y.Z., W.H.W., with 6 and 12 years of experience in radiation oncology, respectively; L.Z.L., Y.Z.L., two radiologists with experience as noted before; Y.M.D., with 15 years of experience in medical oncology at the time of the study), and diagnosis was based on the findings at follow-up MR imaging. When discrepant findings were reported, the images were reviewed by the team, and the final conclusion was based on consensus. An RLN was considered to be positive for malignant involvement (hereafter described as positive) if it resolved after the patient completed RT or showed stability in size after RT but progressed during the subsequent MR imaging follow-up. An RLN was considered to be negative for benign process (hereafter described as negative) if it showed stability in size after the completion of RT and the patient remained disease-free during the subsequent follow-up (21) (Fig 1).

Figure 1a:

Download as PowerPointOpen in Image Viewer

Figure 1b:

Download as PowerPointOpen in Image Viewer

Figure 1c:

Download as PowerPointOpen in Image Viewer

Figure 1d:

Download as PowerPointOpen in Image Viewer

Statistical Analysis

Computer software (SAS, version 9.0, 2002; SAS Institute, Cary, NC) was used to perform the statistical analysis. Generalized estimating equation analysis that was based on a binary logistic regression model was used to assess the likelihood of metastasis in relation to the minimal and maximal axial diameters of RLNs. We used the receiver operating characteristic (ROC) curve and the area under the curve (AUC) to evaluate the effectiveness of different size criteria. The curves represent the relationship between the sensitivity and specificity: Lines that have the largest slope near the y-axis have the best sensitivity at a high specificity. The overall accuracy is represented by the AUC: The larger the area, the better the test. The sensitivity, specificity, and accuracy were calculated by using the standard definitions (22).

The independent-samples t test was used to compare the positive and negative RLNs in regard to the minimal axial diameter, maximal axial diameter, and longitudinal diameter. A two-tailed P value of less than .05 was considered to indicate a significant difference.

Results

Assessment of RLNs

Among our 303 patients with NPC, MR imaging revealed 523 RLNs in 265 (87.5%) patients. Of the 523 identifiable RLNs, the radiographic responses at 14–30 days after completion of RT were distributed as follows: complete resolution, 137 (26.2%) lymph nodes; partial resolution, 120 (22.9%) lymph nodes; and stability, 266 (50.9%) lymph nodes. Of the 120 RLNs with partial resolution, resolution occurred by 4 months in 73 (60.8%) lymph nodes, resolution occurred by 6 months in 21 (17.5%) lymph nodes, and resolution occurred by 8 months in 14 (11.7%) lymph nodes; progression during the follow-up period occurred in five (4.2%) lymph nodes. In the remaining seven (5.8%) lymph nodes (those with the shortest axial diameter of ≥1.2 cm), regression to their normal size occurred 1 year later. Of the 266 RLNs with stability, 259 (97.4%) lymph nodes were stable or had no evidence of progressive disease at further follow-up, and seven (2.6%) lymph nodes demonstrated progressive disease.

On the basis of the results at MR imaging follow-up, we found 264 (50.5%) positive RLNs of the 523 identifiable RLNs in 177 (58.4%) patients. The remaining 259 (49.5%) lymph nodes were classified as negative (Fig 2). The mean values of the minimal axial, maximal axial, and longitudinal diameters of the positive RLNs were all greater than those of the negative RLNs (all P < .001) (Table 2).

Figure 2:

Download as PowerPointOpen in Image Viewer

Table 2 Diameters of Positive and Negative RLNs

Note.—Data are the mean ± standard deviation, and numbers in parentheses are ranges.

Radiologic Criteria of Lateral RLN Metastases

Table 3 and Figure 3 show the number and percentage of metastatic lateral RLNs in relation to the minimal and maximal axial diameters, respectively. The smooth curves in Figure 3 were obtained by applying the logistic regression model to the data. Figure 3 shows that only RLNs with minimal axial diameters of 9 mm or larger or maximal axial diameters of 12 mm or larger correspond to 100% malignancy. The minimal axial diameter has a steeper curve, indicating that it is more accurate than is the maximal axial diameter in the prediction of a positive RLN.

Table 3 Nodal Diameter versus Percentage of Metastatic Lateral RLNs

Note.—Data are percentages, and numbers in parentheses are the number of positive RLNs/total RLNs of that size.

Figure 3:

Download as PowerPointOpen in Image Viewer

Table 4 and Figure 4 show the sensitivity and specificity for several cutoffs of different size criteria. The ROC of the minimal axial diameter showed greater effectiveness in the detection of metastases in RLNs (Fig 4). Of all the size criteria, the criterion that was based on a minimal axial diameter of 6 mm or larger was most accurate for the diagnosis of lateral RLN metastases (AUC = 0.875). The sensitivity, specificity, and accuracy of this criterion were 82.5% (217 of 263), 92.7% (240 of 259), and 87.5% (457 of 522), respectively. The positive predictive value of this criterion was 91.9% (217 of 236), and the negative predictive value was 83.9% (240 of 286).

Table 4 Sensitivity, Specificity, and Accuracy of Size Criteria

Note.—Data are percentages, and numbers in parentheses were used to calculate the percentages.

Figure 4:

Download as PowerPointOpen in Image Viewer

Of the 263 positive RLNs, 50 (19.0%) showed nodal necrosis, and all were revealed to be malignant at the posttreatment follow-up. Nodal necrosis had a specificity of 100% for diagnosis of RLN metastases. The minimal axial diameter of the necrotic RLNs ranged from 4 to 25 mm. There were only three metastatic, necrotic RLNs with a minimal axial diameter smaller than 6 mm. Nodal necrosis combined with the minimal axial diameter criterion led to a slight increase in the sensitivity at a stable specificity (sensitivity increased to 83.7% [220 of 263]).

Groups of two RLNs were seen in five (1.7%) of 303 patients; there were no patients with groups of three or more RLNs. On the basis of the results of follow-up MR imaging, all 10 RLNs with grouping (minimal axial diameter range, 5–11 mm) were classified as malignant lesions (Fig 5). Groups of RLNs also had a specificity of 100% for the diagnosis of RLN metastases. Of 523 identifiable RLNs, only one node pertained to the medial chains. This RLN vanished 1 month after the completion of RT and was also classified as a malignant lesion.

Figure 5a:

Download as PowerPointOpen in Image Viewer

Figure 5b:

Download as PowerPointOpen in Image Viewer

Discussion

Our study results demonstrated that the best size criterion for diagnosing RLN metastases was a minimal axial diameter of 6 mm or larger, with an accuracy of 87.5%. The positive predictive value of this criterion was 91.9% (217 of 236), and the negative predictive value was 83.9% (240 of 286). The maximal axial diameter of RLNs was used to assess metastasis in studies (6,23). However, our data suggested that the maximal axial diameter was not as useful as the minimal axial diameter in the prediction of tumor-positive RLNs; the more useful size criterion for the diagnosis of RLN metastases was the minimal axial diameter, in accordance with the results of a study in cervical nodes reported by Van den Brekel et al (15). The radiologic criterion of a minimal axial diameter of 5 mm or larger, which was often used to assess RLN involvement, was inferior to the criterion of the minimal axial diameter of 6 mm or larger for the diagnosis of RLN metastases.

Apart from the nodal size, central necrosis is another reliable criterion for evaluation of RLN metastases. With our data, central necrosis had a specificity of 100% for the diagnosis of RLN metastases, in accordance with the results of a previous study (15). However, central necrosis is usually not visible in small lymph nodes (24,25). Only three (1.1%) of 264 metastatic RLNs had a minimal axial diameter of less than 6 mm in this series. Our data revealed that central necrosis, combined with the minimal axial diameter criterion, could lead to only a slight improvement in the sensitivity of RLN metastasis detection at a stable specificity. A grouping of three or more borderline (minimal axial diameter of 1–2 mm smaller) cervical nodes is always suggestive of a metastasis and is used as an additional radiologic criterion (15). There were no patients with groups of three or more RLNs in this series, and groups of two RLNs were only seen in five of 303 patients. RLN grouping is rare in NPC patients, and this distribution pattern of RLNs is obviously distinctive from that of cervical nodes. During the MR imaging follow-up, all 10 RLNs with grouping were confirmed to be malignant. Therefore, we believe that any MR imaging finding of groups of two or more RLNs in NPC patients is an important indicator of tumor metastasis.

Although the RLNs include median and lateral groups, the median RLNs are, unfortunately, rare. To our knowledge, there are only three published reports in which an enlarged median RLN metastasis was revealed on MR images (3,9,26). One enlarged medial RLN was also found in our study. It was regarded as a positive lymph node because of its disappearance after RT. We also support the view that any MR imaging finding of a medial RLN is regarded as abnormal.

Radiologic criteria for cervical lymph node metastasis reported in previous studies have been based on clinicopathologic correlations in other series of patients with head and neck cancers (15,17). RLNs are relatively underrepresented in previous series of lymph node metastasis characterization, because they are not routinely removed in the neck dissection and even fine-needle aspiration biopsy is rather difficult owing to their relative accessibility. Thus, similar data of RLNs that are based on clinicopathologic correlations are not available in the literature, especially in NPC patients, who are typically treated with RT rather than surgical intervention (6). In many previous studies, findings at follow-up imaging were used to determine the final diagnosis of nodal lesions because of an inability to histologically confirm the imaging findings (21,26,27). This result occurred because malignant nodal lesions usually resolved after therapy or progressed by the time of follow-up imaging, whereas benign lymph nodes remained stable in size or remained negative at repeated imaging. However, to our knowledge, the clinical importance of MR imaging follow-up for the diagnosis of RLN lesions has not been reported previously. In Southern China, most of the NPCs are nonkeratinizing carcinoma, which accounts for about 98.7% (299 of 303) of the lesions in this series. Unkeratotic carcinoma is sensitive to RT, and all of our patients were treated with curative RT, with a high dose of 68–70 Gy given to the RLNs that were visible on MR images. Thus, metastatic RLN lesions would regress after completion of RT, and the sizes of reactive nodes were usually stable. We found the results of therapy to be poor in only seven RLNs in three patients, and all of these lymph nodes progressed with the primary tumor within 3 months after RT. Therefore, follow-up MR imaging would be a credible method to use for distinguishing metastatic RLNs from benign lymph nodes.

Investigators in several previous studies have defined the radiologic criteria of metastatic RLNs according to a nodal size limit. The diagnostic criteria proposed by Mancuso et al (10) and Watarai et al (11) were based on the lower limit of abnormal RLNs. In contrast, King et al (2) and Lam et al (9) suggested a minimal axial diameter of 4 mm or larger or 5 mm, respectively, as the size criterion on the basis of the upper limit of benign RLNs. The use of a nodal size limit as a criterion is always somewhat arbitrary for a number of reasons. First, there is a large degree of overlap in size between benign reactive and malignant RLNs in head and neck cancers (12,13). The results of a clinicopathologic study in 24 patients with oropharyngeal or laryngopharyngeal carcinoma indicated that the size of metastatic RLNs ranged from 5 to 22 mm, whereas the size of normal RLNs varied between 3 and 25 mm (12). Second, the size ranges of positive or negative RLNs vary largely, depending on which patient sample was studied. The size ranges of malignant RLNs measured by Mancuso et al (10) and Watarai et al (11) were not identical, nor were the size ranges of normal RLNs measured by King et al (2) and Lam et al (9). The findings in a study by Ogura et al (28) suggested that there was a decrease in the normal RLN size with increasing age. A radiologic criterion for RLN metastasis should be determined by using an ROC curve and an AUC to evaluate the effectiveness of different size criteria with a large NPC patient sample.

Some researchers in published reports (3,7,8) have used the criterion of a minimal axial diameter of 5 mm for a radiologic diagnosis of abnormal RLNs. If the RLN involvement was classified as an N1 (unilateral) or an N2 (bilateral) lesion, as opposed to the radiologic criterion of a minimal axial diameter of 5 mm or larger, the size criterion we proposed by using the findings of our study could have resulted in a change in the N category in approximately 6.6% (20 of 303) of patients. There were 12 patients with N1 lesions that would have been downgraded to N0, two patients with N2 lesions that would have been downgraded to N0, and six patients with N2 lesions that would have been downgraded to N1. Because of the change of N category, the treatment strategy and prognostic assessment would have been altered. The 14 patients with lesions that would have been downgraded from N1 or N2 to N0 would not have received chemotherapy. To better define RLN prognostic clinical importance and the role in staging systems in NPC, a study in a larger sample of patients with lesions classified on the basis of MR imaging findings and the radiologic criteria of a minimal axial diameter of 6 mm or larger is needed in the future.

It should be stressed that this study was not a radiologic-histopathologic correlation study. The determination of the nature of RLNs, on the basis of the findings at follow-up MR imaging, could have a false-negative and false-positive value. Furthermore, no imaging modality, including MR imaging, is able to depict all of the micrometastases of small RLNs. Occasionally, because the enlarged RLNs and the primary tumor may be contiguous and show similar areas of signal intensity on images obtained with all sequences, it is difficult to discriminate between enlarged RLNs and the direct extension of the tumor. To take into account these unavoidable pitfalls, the radiologic criteria that we proposed on the basis of our results should be verified in subsequent studies.

In conclusion, we propose the use of the following radiologic criteria for assessment of RLN metastases in patients with NPC: (a) nodes with a minimal axial diameter of 6 mm or larger, (b) any nodes with central necrosis, (c) any finding of groups of two or more RLNs, and (d) any medial RLNs.

Author Contributions

Author contributions: Guarantors of integrity of entire study, G.Y.Z., L.Z.L., W.H.W., Y.Z.L.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, G.Y.Z., L.Z.L., W.H.W., Y.Z.L.; clinical studies, G.Y.Z., L.Z.L., W.H.W., Y.M.D., Y.Z.L.; statistical analysis, G.Y.Z., Y.Z.L., X.W.L.; and manuscript editing, all authors