Abstract
Background/Aim: We evaluated the diagnostic value of functional imaging with [18F]-fluoro-2-deoxy-D-glucose (FDG)-positron emission tomography/computed tomography (PET/CT) for the identification of extranodal extension (ENE) in patients with head and neck squamous cell carcinoma (HNSCC). Patients and Methods: In this study, 94 patients with HNSCC who underwent FDG-PET/CT were enrolled. We recorded the maximum standardized uptake value (SUVmax), compared the results with pathologic findings, and evaluated the diagnostic performance of using a SUVmax cut-off value for ENE. Results: Of the 566 dissected levels examined, 53 (9.4%) exhibited ENE. The mean SUVmax of LN with and without ENE were 6.67 and 1.64, respectively (p<0.001). A receiver operating characteristics (ROC) curve analysis for SUVmax showed an area under the ROC curve of 0.913. A SUVmax cut-off of 3.0 achieved diagnostic performance for identifying ENE with sensitivity, specificity, and accuracy of 81.1%, 94.3% and 93.1%, respectively. Conclusion: FDG-PET/CT findings using a SUVmax cut-off of 3.0 provides appropriate diagnostic value in identifying ENE.
- Head and neck cancer
- squamous cell carcinoma
- positron emission tomography
- extranodal extension
- lymph node metastasis
Intensity-modulated radiation therapy (IMRT) has become the standard technique for radiation treatment of head and neck squamous cell carcinoma (HNSCC). IMRT has been shown to reduce the dose distribution to critical organs at risk, which has led to a significant reduction in treatment-related morbidity, however, the rapid dose drop-off beyond the target volume makes treatment success highly dependent on the accurate determination of the target (1).
Extranodal extension (ENE), which occurs in approximately 60% of HNSCC patients with regional lymph node metastases (LNM), has a strongly negative impact on prognosis and outcome (2, 3). After the 2016 revision of the Union for International Cancer Control (UICC) TNM staging system, ENE has been included in nodal staging. Nodal clinical target volume (CTV) in HNSCC takes ENE into account during radiotherapy treatment planning. A 0-5 mm expansion of the nodal gross tumour volume (GTV) is typically proposed for CTV in patients with LNM without ENE (1). When LNM with ENE is present, a 10 mm nodal CTV margin around the GTV is recommended (4). Therefore, it is important to identify the existence of ENE before treatment.
Functional imaging with [18F]-fluoro-2-deoxy-D-glucose (FDG)-positron emission tomography (PET) provides information on glucose metabolism and assists management of HNSCC patients with respect to staging and prediction of treatment outcome (5, 6). However, the value of FDG-PET in identifying ENE is not fully discussed (7-10). In this study, we evaluated whether FDG-PET provides diagnostic value for identification of ENE in patients with HNSCC.
Patients and Methods
Patients. This retrospective study received institutional review board approval. Between April 2008 and March 2017, 94 patients with pathologically confirmed HNSCC who underwent FDG-PET/computed tomography (CT) imaging in our hospital before surgery were included. Patients with a prior history of head and neck cancer, neck surgery, chemoradiotherapy, and those who underwent PET/CT imaging more than 6 weeks prior to the operative date were excluded (9, 11). Prior informed consent was obtained from all the patients for the use of their images in future studies. The primary cancer sites were oral cavity in 58 patients, oropharynx in 6, hypopharynx in 20, and larynx in 16. Six patients had cancer involving multiple sites in the head and neck. All patients underwent a conventional preoperative work up, including endoscopy, ultrasound, contrast-enhanced CT, and magnetic resonance imaging (MRI), in addition to FDG-PET/CT imaging. The decision to perform selective or radical neck dissection (ND) was made by experienced head and neck surgeons, based on results of the diagnostic workup and patient performance status. ND specimens were removed en bloc and divided by nodal level; pathologic findings for the existence of LNM and ENE were recorded at each anatomic level (5).
FDG-PET/CT imaging. FDG-PET/CT images were obtained using a 3D PET/CT scanner (Gemini GXL 16; Philips Medical Systems, Cleveland, OH, USA). All patients fasted for 6 hours before the imaging procedure and underwent two routine whole-body PET/CT scans in a single session after receiving an injection of FDG (185-370 MBq): one at 60-90 min (early scan) and a second at 120-150 min (delayed scan). In the early scan, we acquired CT images [63 mA, 120 kV, 512×512 matrix, 600-mm field of view (FOV), 5-mm slice thickness] and then performed emission measurements in 3D mode with a 144×144 matrix. The emission scan time per bed position was 2 min; 10-12 bed positions (FOV 576 mm) were acquired. Attenuation correction was with CT transmission data; emission images were reconstructed using the line of response-row-action maximum likelihood algorithm. Reconstructed images had a 4 mm slice thickness. Emission scan time per bed position of the delayed scan was 3 min; 4-6 bed positions were acquired. Other imaging protocols were the same as those of the early scan (12).
FDG-PET/CT data and statistical analysis. One radiation oncologist with 15 years of experience in diagnosing and treating HNSCC reviewed PET and CT images of the early scan. The maximum standardized uptake value (SUVmax) of a represented LN or the area of the highest SUV were recorded at each nodal level (I-V) and compared with pathologic findings of ENE based on the nodal level (5, 9).
The Mann–Whitney U-test was performed to evaluate the association of SUVmax, with ENE. Receiver operating characteristics (ROC) analysis with Youden index was performed to determine the best SUVmax cut-off for the identification of ENE. Statistical calculations were performed with SPSS software, version 24.0 (IBM, Armonk, NY, USA). Differences with p-values of <0.05 were considered statistically significant.
Results
Of the 94 patients, 54 (57.4%) patients had LNM. Of these, 23 (24.4%) patients had LNM with ENE. The other 40 (42.6%) patients were recorded as N0. Dissection of 566 nodal levels was performed; correlation between SUVmax and histologic data was analyzed. ENE was present in 53 levels (9.4%) of 106 (18.7%) pathologically positive nodal levels. The mean SUVmax of all nodal levels was 2.11±2.44. The mean SUVmax of LN with and without ENE was 6.67±4.74 (range=0.7-18.9) and 1.64±1.38 (range=0.5-23.4), respectively (p<0.001, Figure 1).
The ROC curve analysis of SUVmax for the differentiation of LN with ENE from LN without ENE showed an area under the ROC curve (AUC) value of 0.913±0.28 (p<0.001, Figure 2). At the best discriminative SUVmax cut-off of 3.0, the sensitivity and specificity were 81.1% and 94.3%, respectively (Youden index = 0.755, Tables I and II).
Discussion
ENE is one of the most important factors for clinical decision making in radiotherapy planning in HNSCC (3, 4). However, there is currently no established method for the diagnosis of ENE. The previously reported respective sensitivities and specificities are 85% and 88% with SUV cut-off of 2.25 for the oral cavity (AUC value of 0.864), 74% and 71% with SUV cut-off of 3.85 for the oropharynx (AUC value of 0.814), 80% and 74% with SUV cut-off of 2.65 for the hypopharynx (AUC value of 0.857), and 86% and 86% with SUV cut-off of 2.8 for the larynx (AUC value of 0.923) (7-10). Our results using the SUV cut-off method were similar with the high AUC value and Youden index. One possible reason is that we evaluated the diagnostic performance of FDG-PET/CT imaging with an adequate 6-week interval between imaging and surgery (11). Furthermore, our results suggested that FDG-PET using the single SUV cut-off method provides appropriate diagnostic value for identifying ENE in patients with HNSCC regardless of the primary subsite. Because HNSCC often involves more than one head and neck subsite, it may be reasonable to utilize the single SUV cut-off method regardless of the primary subsite.
Radiological evaluation using CT and MRI are commonly performed for the diagnosis of ENE. Steinkamp et al. (13) evaluated the diagnostic performance of CT for ENE in 165 patients with HNSCC. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 80.9%, 72.7%, 69.4%, 83.3%, and 76.3%, respectively. Prabhu et al. (14) evaluated diagnostic performance of CT for ENE in 432 patients with oral cavity or locally advanced/nonfunctional laryngeal cancer. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 43.7%, 97.7%, 82.6%, 87.3%, and 86.8%, respectively. Steinkamp et al. (15) evaluated diagnostic performance of MRI for ENE in 110 patients with HNSCC. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 74.4%, 72.2%, 76.1%, 70.3%, and 73.4%, respectively. Another group (16) compared the diagnostic performance of MRI with that of CT for ENE in 17 patients with HNSCC. The sensitivity, specificity, and accuracy were 65%, 93%, and 73% for CT, and 78%, 86%, and 80% for MRI, respectively. They found no significant difference between the two modalities for either sensitivity (p=0.1317) or specificity (p=0.3137). Our results suggested that diagnostic performance of FDG-PET/CT imaging using the SUV cut-off method was comparable. The commonly-used diagnostic criteria for ENE on CT and MRI are the presence of an indistinct nodal margin, irregular nodal capsular enhancement, or infiltration into the adjacent fat or muscle (16, 17). However, there is a wide interobserver variation in the interpretation of these criteria. King et al. (16) reported that interobserver variation for the detection of ENE based on CT and MRI showed a kappa coefficient of 0.6077 and 0.4966, respectively. The diagnostic performance of the nodal size criteria, which is a more objective method than these radiological features, is not sufficient for the diagnosis of ENE. Previous studies suggested that ENE exists in more than 40% of LNs with a size of <10 mm (18, 19). Zoumalan et al. (20) evaluated the relationship between nodal size and ENE based on CT imaging and ND specimens of 17 patients. LN diameter was not significantly different between LNs with ENE and those without (p=0.83). We recommend the SUV cut-off method as an objective and simplified method to identify ENE in patients with HNSCC, regardless of subsite. The CTV should be defined to cover microscopic spread around LNs with high SUVmax.
Relationship between maximum standardized uptake value and lymph nodes with and without extranodal extension.
Comparison between diagnosis of extranodal extension on FDG-PET/CT with SUVmax cut-off of 3.0 and pathological results.
Diagnostic accuracy of FDG-PET/CT with SUVmax cut-off of 3.0 for extranodal extension.
Receiver operating characteristics curve of maximum standardized uptake value for differentiating lymph nodes with and without extranodal extension.
There are some limitations to our study. First, it is retrospective in nature, therefore the potential for selection bias exists, as preoperative imaging may have influenced the decision to perform surgery. Second, we determined an SUVmax of 3.0 as the optimal cut-off to identify ENE. However, this cut-off value was based on PET/CT imaging at a single institution. Previous studies have suggested that SUV may change between individual PET/CT scanners, acquisition mode, and image reconstruction and processing parameters (21). Therefore, optimal cut-off value may be slightly different among each institution's PET/CT imaging protocol.
In conclusion, FDG-PET findings using a SUVmax cut-off of 3.0 provides appropriate diagnostic value in identifying ENE in patients with HNSCC regardless of the primary subsite. The CTV should be defined to cover microscopic spread around LNs with SUVmax above this value due to their greater risk of harbouring unknown ENE.
Acknowledgements
This work was supported by the JSPS KAKENHI (grant number 18K07758).
Footnotes
Authors' Contributions
RT developed the study design, collected, analyzed, and interpreted data, performed statistical analysis, and wrote the manuscript. TS, TM, SS, and TW developed the study design, analyzed and interpreted data. YK analyzed data and performed statistical analysis. DM, RY, FS, and NT collected data. HA, YO, HN, and NO developed the study design and interpreted data. All Authors have read and approved the final manuscript.
This article is freely accessible online.
Conflicts of Interest
The Authors declare that they have no conflicts of interest related to this study.
- Received February 9, 2020.
- Revision received February 15, 2020.
- Accepted February 19, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- International Multi-Specialty Expert Physician Preoperative Identification of Extranodal Extension in Oropharyngeal Cancer Patients using Computed Tomography: Prospective Blinded Human Inter-Observer Performance Evaluation
- Concurrent Chemoradiotherapy With Docetaxel, Cisplatin, and 5-Fluorouracil for T3 N0 Glottic Carcinoma Without Vocal Cord Fixation
- High Spatial Resolution Digital Positron Emission Tomography Images With Dedicated Source-to-background Algorithm for Radiotherapy Planning