- Research article
- Open Access
A prediction model of nodal metastasis in cN0 oral squamous cell carcinoma using metabolic and pathological variables
Cancer Imaging volume 23, Article number: 34 (2023)
The efficacy of 18F-fluorodeoxyglucose (18F-FDG) Positron Emission Tomography/Computed Tomography(PET/CT) in evaluating the neck status in clinically node-negative (cN0) oral squamous cell carcinoma(OSCC) patients was still unsatisfying. We tried to develop a prediction model for nodal metastasis in cN0 OSCC patients by using metabolic and pathological variables.
Consecutive cN0 OSCC patients with preoperative 18F-FDG PET/CT, subsequent surgical resection of primary tumor and neck dissection were included. Ninety-five patients who underwent PET/CT scanning in Shanghai ninth people’s hospital were identified as training cohort, and another 46 patients who imaged in Shanghai Universal Medical Imaging Diagnostic Center were selected as validation cohort. Nodal-status-related variables in the training cohort were selected by multivariable regression after using the least absolute shrinkage and selection operator (LASSO). A nomogram was constructed with significant variables for the risk prediction of nodal metastasis. Finally, nomogram performance was determined by its discrimination, calibration, and clinical usefulness.
Nodal maximum standardized uptake value(nodal SUVmax) and pathological T stage were selected as significant variables. A prediction model incorporating the two variables was used to plot a nomogram. The area under the curve was 0.871(Standard Error [SE], 0.035; 95% Confidence Interval [CI], 0.787–0.931) in the training cohort, and 0.809(SE, 0.069; 95% CI, 0.666–0.910) in the validation cohort, with good calibration demonstrated.
A prediction model incorporates metabolic and pathological variables has good performance for predicting nodal metastasis in cN0 OSCC patients. However, further studies with large populations are needed to verify our findings.
Accurate N staging is critical for oral squamous cell carcinoma(OSCC) patients with clinically node-negative (cN0) neck. When patients have unilateral nodal metastasis, the overall survival may be reduced by 50%. If bilateral cervical nodal metastasis occurs, the survival time can be further halved . Neck dissection(ND) thus maximizes survival probability but leads to over-treatment, which is accompanied by poorer health-related quality of life in approximately 75% of patients . Recently, a prospective diagnostic study of 18F-fluorodeoxyglucose (18F-FDG) Positron Emission Tomography/Computed Tomography(PET/CT) in OSCC patients with cN0 or cN + neck showed that 18F-FDG PET/CT achieved a sensitivity and specificity of 83.3% and 84.8% in detecting cervical lymph node metastases , meanwhile, most of the OSCC patients (73.3%) in their study had no lymph node metastasis. It is no doubt that early OSCC patients without nodal metastasis can benefit from the strategy of “wait and watch surveillance” after resection of the primary tumor without ND . Therefore, it is important for us to find an efficient way to predict the occult nodal metastasis in cN0 neck accurately. Although studies indicated that 18F-FDG PET  or PET/CT  could detect occult neck metastasis better than CT/MR imaging in cN0 OSCC patients, it was reported that its utility for detecting cervical nodal metastasis is limited in early OSCC . A meta-analysis conducted by Kim et al.  showed the low sensitivity and moderate specificity of 18F-FDG PET/CT for the detection of cervical lymph node(LN) metastasis in cN0 HNSCC(head and neck SCC) patients. Therefore, 18F-FDG PET/CT is currently not sufficiently reliable to avoid elective treatment of the neck.
Previous reports indicated that pathological variables of primary tumor(OSCC) such as depth of invasion(DOI) [9, 10], pathological tumor size , perineural invasion(PNI)  were also widely used for prediction of occult nodal metastasis in cN0 OSCC patients. However, some of them still got unsatisfying results .
To the best of our knowledge, no studies tried to develop a prediction model for nodal metastasis in cN0 OSCC patients by using metabolic and pathological variables. The goal of this study was to determine whether the combination of metabolic and pathological variables could further improve the prediction of nodal metastasis in cN0 OSCC patients.
Materials and methods
We retrieved data on all consecutive patients with oral tumor who underwent preoperative 18F-FDG PET/CT in Shanghai ninth people’s hospital and in Shanghai Universal Medical Imaging Diagnostic Center (from January 2018 to May 2022). The inclusion criteria were defined as follows: 1) 18F-FDG PET/CT was performed within two weeks before surgery. 2)Surgical resection of primary tumor and neck dissection. 3)The primary tumor was pathologically confirmed OSCC, and staged according to the latest(8th) Edition of the American Joint Committee on Cancer/Union for International Cancer Control(AJCC/UICC) TNM staging system , 4)Patients with cN0 neck, 5)Patients were followed up for at least 6 months after operation. Patients who had tumor relapse, previous treatment, or high blood glucose level (> 200 mg/dL) during the PET/CT scan were all excluded. According to the criteria, ninety-five patients imaged in Shanghai ninth people’s hospital were identified as the training cohort and another 46 patients scanned in Shanghai Universal Medical Imaging Diagnostic Center were selected as the validation cohort. All patients received surgical treatment and postoperative follow-up in Shanghai ninth people’s hospital. Details of patient selection were shown in Fig. 1.
In the training cohort, 52 patients with malignancies of tongue, followed by gingival (nineteen patients), buccal mucosa (ten patients), floor of mouth (nine patients) and palate cancers(five patients). ND level on the affected side were as follows: I-II:I-III:I-IV:I-V = 7:33:9:46, and ten patients underwent contralateral ND(I:I-II:I-III:I-IV:I-V = 1:2:4:2:1). Nodal metastases were histologically confirmed in 38 patients (40.0%), eight of them were found to have bilateral nodal metastasis. In the validation cohort, sixteen patients with malignancies of buccal mucosa, followed by tongue (fourteen patients), gingival (nine patients) and floor of mouth (seven patients). ND level on the affected side were as follows: I-II:I-III:I-IV:I-V = 1:21:3:21, and eight patients underwent contralateral ND(I-II:I-III = 2:6). Nodal metastases were histologically confirmed in 14 patients (30.4%), one of them was found to have bilateral nodal metastasis. No distant metastasis was found in the 141 patients by preoperative evaluation.
All patients were fasted at least 6 h before being injected with 18F-FDG. PET/CT scanning was performed on either Ingenuity TF (Philips Medical systems, Cleveland, USA) in Shanghai ninth people’s hospital or Biograph mCT (Siemens Healthcare, Erlangen, Germany) in Shanghai Universal Medical Imaging Diagnostic Center. PET/CT images were acquired from the base of the skull to the mid-thigh or foot approximately 60 min after the injection with 18F-FDG. The detailed acquisition parameters are given in Supplementary Material Table 1.
Metabolic variables of primary tumor and lymph node
In the training cohort, the volume of interest (VOI) of the primary OSCC tumors was automatically delineated with a threshold of 40% of maximum standardized uptake value(SUVmax)  on the Intellispace Portal workstation (Philips, Best, the Netherlands). Then, the VOIs were carefully reviewed by a nuclear medicine radiologist(F.X) with 15 years of experience, and the borders of the volume of interest(VOI) were adjusted manually to exclude adjacent physiological 18F-FDG-avid structures on PET images. SUVmax, SUVmean, metabolic tumor volume (MTV) and total lesion glycolysis (TLG = SUVmean × MTV) of primary tumors were calculated. The same radiologist visually reviewed PET/CT images of the ipsilateral and contralateral neck, the LN with the most intense uptake of 18F-FDG was carefully chosen (in case of no LN with higher SUVmax than background, the LN with the longest diameter on the affected side was chosen), and the nodal SUVmax were measured visually by placing regions of interest over the neck LNs. The short and long axis diameter of the LN was measured as well. Another nuclear medicine radiologist(L.P) with 10 years of experience performed the measurements on the syngo VB20P workstation (Siemens Healthcare, Erlangen, Germany) for the validation cohort in the same way.
Pathological variables of the primary tumor
All 141 patients had a detailed pathological diagnosis of the primary tumor in their hospital records. One oral pathologist(Y.H) checked the pathological findings of the primary tumor, including grade of tumor, surgical margin, PNI, pathological tumor size, depth of invasion, and lymphovascular invasion. Finally, according to the 8th edition of AJCC/UICC TNM staging system, the pathological T stage was determined by pathological tumor size and depth of invasion.
Standard reference for nodal metastasis
The histopathological diagnosis (immune-histochemical staining with cytokeratin and traditional hematoxylin and eosin staining) was the standard of reference for the ND side. Follow-up was taken as reference for the contralateral side without ND. Postoperative complications, primary site status and cervical nodal status were followed up every month for at least 6 months after operation.
Construction of prediction model
In the training cohort, least absolute shrinkage and selection operator(LASSO) logistic regression algorithm, with penalty parameter tuning conducted by five-fold cross-validation, was applied to select nodal-status-related variables with nonzero coefficients. Multivariable forward stepwise logistic regression modeling was then performed to identify significant variables associated with nodal metastasis. A prediction model of nodal metastasis, with combination of significant variables, was established by using logistic regression modeling. To provide a more understandable outcome measure, a nomogram was then constructed by using the significant variables derived from the training cohort.
Validation of the model and clinical utility
Calibration curves were plotted via bootstrapping with 1000 resamples to assess the prediction model, accompanied by the Hosmer–Lemeshow goodness of fit test, a P-value of the Hosmer–Lemeshow test > 0.05 indicated a good fit. Predictive accuracy of the prediction model was validated using receiver-operating-characteristic (ROC) and quantified by the area under the curve (AUC) in the training cohort and validation cohort. Finally, Clinical utility of the nomogram was assessed by decision curve analysis.
Statistical analyses were carried out using SPSS(version 17.0, SPSS Inc., Chicago, Illinois, USA), MedCal(version 19.6.1, MedCalc Software, Ostend, Belgium) and R software(version 4.2.0, http://www.r-project.org, R Foundation for Statistical Computing, Vienna, Austria). Data are presented as mean ± SD or median (followed by range) when the distribution of data was skewed. Qualitative parameters were analyzed using the Fisher’s exact test. The LASSO logistic regression algorithm was carried out by using the “glmnet” package. Nomogram and calibration curves were performed with the “rms” package. Decision curve analysis was generated using the “rmda” package. A two-tailed test with P < 0.05 represented a statistically significant difference.
Twenty clinicopathological and metabolic variables in the training cohort and validation cohort were summarized in Table 1. In the training cohort, significant differences were found in clinical T stage, grade of tumor, pathological tumor size, depth of invasion, pathological T stage, PNI, tumor SUVmax, tumor SUVmean, tumor MTV, tumor TLG, nodal short diameter, and nodal SUVmax between patients with or without nodal metastasis.
Variable selection for constructing models
Using LASSO regression in the training cohort, 20 variables in the training cohort were reduced to 7 when using the minimum error criterion (Fig. 2). The selected variables including “nodal SUVmax”, “pathological T stage”, “drinker”, “grade of tumor”, “PNI”, “nodal short diameter” and “pathological tumor size”. In the multivariable forward stepwise logistic regression model, “drinker”, “grade of tumor”, “PNI”, “nodal short diameter” and “pathological tumor size” was justifiably removed.
Construction, calibration and discrimination of prediction model
“Nodal SUVmax” and “pathological T stage” were the significant variables for predicting nodal metastasis in the regression model (Table 2). The two significant variables were used to construct a nomogram(Prediction Model) for risk prediction of nodal metastasis (Fig. 3).
In the training cohort, when nodal metastasis prediction performance was analyzed using ROC curve analysis, the AUC was 0.774(Standard Error [SE], 0.050; 95% Confidence Interval [CI], 0.667–0.859) for nodal SUVmax, 0.758 (SE, 0.042; 95% CI, 0.659–0.840) for pathological T stage, and 0.871 (SE, 0.035; 95% CI, 0.787–0.931) for prediction model (Fig. 4a). When choosing the prediction of 0.34 as optimal cutoff value, the prediction model achieved a sensitivity of 89.5%(34/38), a specificity of 70.2%(40/57), an accuracy of 77.9%(74/95), a positive predictive value of 66.7%(34/51), and a negative predictive value of 90.9%(40/44), calibration plot of the nomogram showed good agreement between predicted and observed nodal metastasis (Fig. 4b), the Hosmer–Lemeshow test yield a P value of 0.532, demonstrating a good fit.
The good calibration was further confirmed in the validation cohort with AUC of 0.809(SE, 0.069; 95% CI, 0.666–0.910) (Fig. 5), the Hosmer–Lemeshow test yield a P value of 0.694.
Decision curve showed that the prediction model presented with a better performance than the “nodal SUVmax”, “pathological T stage”, and “two extreme lines” (treat-none and treat-all) when the threshold probability was 0–0.8 (Fig. 6). Representative images for patients with cN0 neck correctly diagnosed by the prediction model were shown in Fig. 7.
In the present study, we tried to establish an efficient prediction model for nodal status based on 20 nodal status-related variables in cN0 OSCC patients. Finally, one metabolic variable(nodal SUVmax) and one pathological variable(pathological T stage) were selected to build the prediction model.
There were various reports on the association between metabolic variables of the primary tumor and nodal metastasis. Morand et al. . found that higher SUVmax (≥ 9.5) of the primary tumor was associated with higher occurrence of metastatic nodal disease(AUC 0.651) in 71 cN0 or cN + OSCC patients. Bae et al. . suggested that MTV(AUC 0.629) and TLG(AUC 0.626) of the primary tumor independently predicted nodal metastasis in 130 cN0 OSCC patients. However, some limitations in their studies should be mentioned. First, the nodal SUVmax, which represent the metabolism of LN, was not analyzed, second, the physiologic 18F-FDG uptake in the oral cavity, which may cause overestimation of tumor MTV and TLG, was not concerned, furthermore, their results were based on two different PET scanners. Lastly, most of all, the AUCs of the metabolic variables of the primary tumor were relatively low. In this study, patients in the training cohort (or validation cohort) underwent a same PET/CT scanner with same protocol, the borders of the VOI of the primary tumor were adjusted manually to exclude adjacent physiological 18F-FDG-avid structures on PET images, and nodal SUVmax were considered as well. Significant differences were observed in tumor SUmax, SUVmean, MTV, TLG and nodal SUVmax between cN0 OSCC patients with and without nodal metastasis in the training cohort (Table 1). LASSO regression indicated that nodal SUVmax was the only significant metabolic variable in predicting nodal metastasis, unfortunately, its AUC(0.774) was still dissatisfied. Yamazaki et al.  noted that detection of smaller volumes was limited by the capabilities of the PET/CT scanner resolution. PET/CT was found invalid if the lesion was smaller than 5 mm. Furthermore, some occult nodal metastases harbored a small amount of 18F-FDG uptake, while some inflammatory LNs or LNs close to the submandibular gland may got relative higher value of nodal SUVmax . Therefore, nodal SUVmax should not be the sole criterion for predicting the presence of neck metastasis.
Pathological variables of primary tumor(OSCC) were also widely used to predict nodal metastasis in cN0 OSCC patients. As reported in the literature, the pathological tumor size , DOI [9, 10], and PNI  were important pathological variables which were associated with nodal metastasis in OSCC patients. In the latest(8th) Edition of the AJCC/UICC TNM staging system , DOI was introduced as a fundamental staging criterion to define T1, T2, and T3 categories as it correlates well with the risk of nodal metastasis and loco-regional recurrence, especially in tongue SCC [9, 10]. In the present study, pathological T stage was determined by pathological tumor size and DOI, as introduced by the 8th Edition of the AJCC/UICC TNM staging system . Interestingly, it was proved to be a significant predictor of nodal metastasis, with an AUC of 0.758, rather than pathological tumor size and DOI. We infer that pathological T stage, which combined pathological tumor size and DOI, maybe potentially more useful in predicting the nodal metastasis than the two variables separately. We believed it could serve as useful complementary predictive variables for nodal metastasis.
In the present study, a prediction model which was based on nodal SUVmax and pathological T stage was successfully built by using logistic regression modeling. It achieved a higher AUC(0.871) compared to nodal SUVmax and pathological T stage. Reports on the prediction model of nodal metastasis in cN0 OSCC patients were various, most of them were based on clinicopathological [20,21,22,23,24] and radiologic parameters . Bur et al.  developed machine learning algorithms by using clinicopathologic data from 782 cT1-2N0 OSCC patients to predict pathologic nodal metastasis, they found the best classification performance was achieved with a decision forest algorithm (AUC = 0.840). Wang et al. . constructed a prediction model for nodal metastasis based on molecular panel and clinicopathological factors in 112 cN0 OSCC patients, the model with the combination of CDKN2A, PLAU, T stage and pathological grade was the best in predicting lymph node metastasis (AUC = 0.807). Yuan et al. . developed several machine learning models to predict occult nodal metastasis in 116 early-stage oral tongue SCC patients from preoperative MRI texture features, they found naïve Bayes model gave the best overall performance with an AUC of 0.802. Farrokhian et al. . built a predictive model with XGBoost architecture in 634 early stage OSCC patients, they achieved an AUC of 0.840. Unfortunately, these methods described above were relatively complicated and they were not routinely used in clinical practice. Recently, Kallarakkal et al. . established a predictive model based on the clinicopathological parameters only, including gender, tumor size, site, pattern of invasion and DOI, whereas, the AUC(0.752) they achieved was relatively low. We believe the combination of metabolic and pathological variables may be another useful way to predict the nodal metastasis in cN0 OSCC patients. The advantages of this model were as follows, first, the prediction model in the present study was trained and validated by data acquired from two different imaging centers, the subsequent calibration plot and ROC analysis illustrating a good discriminatory ability and predictive accuracy. Second, the nodal SUVmax and pathological T stage were easy to be obtained, which make the model simple to be established. Besides, preoperative PET/CT is also a good method for distant metastasis screening in OSCC patients. Finally, the model was useful to predict the probability of nodal metastasis for cN0 OSCC patients who underwent simple resection of the primary tumor without ND. Patients with negative results could rely on the high negative predictive value of the model (90.9%) to rule out nodal metastasis. However, caution should be exercised in patients with positive results as the positive predictive value (66.7%) was relatively low, leading to potential unnecessary neck dissections. Therefore, such patients should undergo more careful follow-up to confirm neck status.
Our study had several limitations. First, it is a retrospective study including selection bias. Second, the partial volume effect was not corrected for the LN, the value of nodal SUVmax may be underestimated. Third, the result of pathological T stage can only be obtained after surgery, which makes the pre-operative evaluation impossible. Lastly, the number of patients involved in the study was relatively small, further study with large number of patients is needed.
We proposed a prediction model that incorporates the nodal SUVmax and pathological T stage to facilitate evaluation of LN status in OSCC patients with cN0 neck. To our knowledge, this is the first study which combined metabolic and pathological variables in predicting nodal metastasis in OSCC patients with cN0 neck. However, further studies with large populations are needed to verify our findings.
Availability of data and materials
Data available at supplementary file.
Oral squamous cell carcinoma
Positron Emission Tomography/Computed Tomography
Depth of invasion
Least absolute shrinkage and selection operator
Maximum standardized uptake value
Area under the curve
American Joint Committee on Cancer/Union for International Cancer Control
Ordered subsets expectation maximization
Volume of interest
Metabolic tumor volume
Total lesion glycolysis
Mukherji S, Armao D, Joshi V. Cervical nodal metastases in squamous cell carcinoma of the head and neck: what to expect. Head Neck. 2001;23:995–1005.
Spalthoff S, Zimmerer R, Jehn P, Gellrich N, Handschel J, Krüskemper G. Neck Dissection’s Burden on the Patient: Functional and Psychosocial Aspects in 1,652 Patients With Oral Squamous Cell Carcinomas. J Oral Maxillofac Surg. 2017;75:839–49.
Linz C, Brands R, Herterich T, Hartmann S, Müller-Richter U, Kübler A, et al. Accuracy of 18-F Fluorodeoxyglucose Positron Emission Tomographic/Computed Tomographic Imaging in Primary Staging of Squamous Cell Carcinoma of the Oral Cavity. JAMA netw open. 2021;4:e217083.
McDonald C, Lowe D, Bekiroglu F, Schache A, Shaw R, Rogers S. Health-related quality of life in patients with T1N0 oral squamous cell carcinoma: selective neck dissection compared with wait and watch surveillance. Br J Oral Maxillofac Surg. 2019;57:649–54.
Ng S, Yen T, Chang J, Chan S, Ko S, Wang H, et al. Prospective study of [18F]fluorodeoxyglucose positron emission tomography and computed tomography and magnetic resonance imaging in oral cavity squamous cell carcinoma with palpably negative neck. J Clin Oncol. 2006;24(27):4371–6.
Bae M, Roh J, Kim J, Lee J, Cho K, Choi S, et al. F-FDG PET/CT versus CT/MR imaging for detection of neck lymph node metastasis in palpably node-negative oral cavity cancer. J Cancer Res Clin Oncol. 2020;146:237–44.
Yamaga E, Toriihara A, Nakamura S, Asai S, Fujioka T, Yoshimura R, et al. Clinical usefulness of 2-deoxy-2-[18F] fluoro-d-glucose-positron emission tomography/computed tomography for assessing early oral squamous cell carcinoma (cT1-2N0M0). Jpn J Clin Oncol. 2018;48:633–9.
Kim S, Pak K, Kim K. Diagnostic accuracy of F-18 FDG PET or PET/CT for detection of lymph node metastasis in clinically node negative head and neck cancer patients; A systematic review and meta-analysis. Am J Otolaryngol. 2019;40:297–305.
Shinn J, Wood C, Colazo J, Harrell F, Rohde S, Mannion K. Cumulative incidence of neck recurrence with increasing depth of invasion. Oral oncol. 2018;87:36–42.
Huang S, Hwang D, Lockwood G, Goldstein D, O’Sullivan B. Predictive value of tumor thickness for cervical lymph-node involvement in squamous cell carcinoma of the oral cavity: a meta-analysis of reported studies. Cancer. 2009;115:1489–97.
Sagheb K, Blatt S, Rahimi-Nedjat R, Eigenbrodt S, Al-Nawas B, Walter C. Cervical metastases behavior of T1–2 squamous cell cancer of the oral mucosa. Clin Oral Investig. 2017;21:931–5.
Chatzistefanou I, Lubek J, Markou K, Ord R. The role of neck dissection and postoperative adjuvant radiotherapy in cN0 patients with PNI-positive squamous cell carcinoma of the oral cavity. Oral oncol. 2014;50:753–8.
Melchers L, Schuuring E, van Dijk B, de Bock G, Witjes M, van der Laan B, et al. Tumour infiltration depth ≥4 mm is an indication for an elective neck dissection in pT1cN0 oral squamous cell carcinoma. Oral oncol. 2012;48:337–42.
Lydiatt W, Patel S, O’Sullivan B, Brandwein M, Ridge J, Migliacci J, et al. Head and Neck cancers-major changes in the American Joint Committee on cancer eighth edition cancer staging manual. CA Cancer J Clin. 2017;67(2):122–37.
Dibble E, Alvarez A, Truong M, Mercier G, Cook E, Subramaniam R. 18F-FDG metabolic tumor volume and total glycolytic activity of oral cavity and oropharyngeal squamous cell cancer: adding value to clinical staging. J Nucl Med. 2012;53:709–15.
Morand G, Vital D, Kudura K, Werner J, Stoeckli S, Huber G, et al. Maximum Standardized Uptake Value (SUV) of Primary Tumor Predicts Occult Neck Metastasis in Oral Cancer. Sci Rep. 2018;8:11817.
Bae M, Roh J, Kim J, Choi S, Nam S, Kim S. Prediction of cervical metastasis and survival in cN0 oral cavity cancer using tumour F-FDG PET/CT functional parameters. J Cancer Res Clin Oncol. 2020;146:3341–8.
Yamazaki Y, Saitoh M, Notani K, Tei K, Totsuka Y, Takinami S, et al. Assessment of cervical lymph node metastases using FDG-PET in patients with head and neck cancer. Ann nucl med. 2008;22:177–84.
Niu L, Zheng D, Wang D, Zhang J, Fei J, Guo C. Accuracy of F-FDG PET/CT in Detection of Neck Metastases of Oral Squamous Cell Carcinoma in Patients Without Large Palpable Lymph Nodes. Oral Surg Oral Med Oral Pathol Oral Radiol. 2020;129:418–26.
De Silva R, Siriwardena B, Samaranayaka A, Abeyasinghe W, Tilakaratne W. A model to predict nodal metastasis in patients with oral squamous cell carcinoma. PLoS ONE. 2018;13:e0201755.
Bur A, Holcomb A, Goodwin S, Woodroof J, Karadaghy O, Shnayder Y, et al. Machine learning to predict occult nodal metastasis in early oral squamous cell carcinoma. Oral oncol. 2019;92:20–5.
Wang S, Li T, Liu H, Wei W, Yang Y, Wang C, et al. A Combined Prediction Model for Lymph Node Metastasis Based on a Molecular Panel and Clinicopathological Factors in Oral Squamous Cell Carcinoma. Front Oncol. 2021;11:660615.
Kallarakkal T, Siriwardena B, Samaranayaka A, De Silva R, Tilakaratne W. A validated predictive model for risk of nodal metastasis in node negative oral squamous cell carcinoma of the buccal mucosa and tongue. J Oral Pathol Med. 2022;51:436–43.
Farrokhian N, Holcomb AJ, Dimon E, Karadaghy O, Ward C, Whiteford E, et al. Development and Validation of Machine Learning Models for Predicting Occult Nodal Metastasis in Early-Stage Oral Cavity Squamous Cell Carcinoma. JAMA Netw Open. 2022;5:e227226.
Yuan Y, Ren J, Tao X. Machine learning-based MRI texture analysis to predict occult lymph node metastasis in early-stage oral tongue squamous cell carcinoma. Eur Radiol. 2021;31:6429–37.
The authors would like to acknowledge the support from the colleagues of nuclear medicine department, Shanghai ninth people's hospital.
Ethical approval and consent to participate
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This retrospective study was approved by the institutional review board of Shanghai Ninth People's hospital, Shanghai Jiaotong University, School of medicine. Written informed consent was waived for the use of patients’ medical and imaging data. All patients’ information was anonymized prior to the analysis.
Consent for publication
All authors named in this manuscript gave their consent for this publication and take full responsibility for its content. All authors read and approved the final manuscript.
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This article has been updated to correct the author contribution section.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
About this article
Cite this article
Xu, F., Peng, L., Feng, J. et al. A prediction model of nodal metastasis in cN0 oral squamous cell carcinoma using metabolic and pathological variables. Cancer Imaging 23, 34 (2023). https://doi.org/10.1186/s40644-023-00552-z
- 18F-FDG PET/CT
- Oral squamous cell cancer
- Nodal metastasis