Development of A Three-Gene Signature Prediction Model for Lymph Node Metastasis in Papillary Thyroid Cancer

Development of A Three-Gene Signature Prediction Model for Lymph Node Metastasis in Papillary Thyroid Cancer | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Primary research Development of A Three-Gene Signature Prediction Model for Lymph Node Metastasis in Papillary Thyroid Cancer Ziwei Huang, Yuenan Liu, Kehao Le, Ming Xu, Wenhui Li, Qiuyang Zhao, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-41157/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Thyroid cancer is one of the most prevalent endocrine cancers with a rising incidence rate over the past years. Papillary thyroid cancer (PTC) is the dominant historical type of thyroid cancer. Early lymph node metastasis happens frequently in PTC. However, some of the lymph node metastasis may be troublesome for detecting because of limited methods. Methods: Robust rank aggregation afforded us the shared differential expression genes among multiple datasets. Gene ontology analysis was performed to identify potential functions. Weighted gene co-expression network analysis was used to research the correlations between gene expression patterns with clinical characteristic. Protein-protein interaction network was performed to identify the hub genes. The least absolute shrinkage and selection operator and Logistic regression were performed to construct a prediction model. Results: We developed a three-gene signature prediction model for lymph node metastasis in PTC through transcriptomic analysis. After quality control, we collected 8 microarray datasets from GEO database and an RNA sequencing dataset from TCGA database. We found the transcriptome profiles were correlated with lymph node metastasis and 3 genes were verified to be independent prediction factors towards those statistic approach. Afterwards, we designed a predicable risk score system and effectively confirmed the model in two independent papillary thyroid cancer cohorts. Conclusions: We recommended a successful predicable model of lymph node metastasis in papillary thyroid cancer patients with moderate accuracy. Cancer Biology Prediction model Papillary thyroid cancer Lymph node metastasis Transcriptome analysis WGCNA Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Background Thyroid cancer, one of the most prevalent endocrine cancers with a rising incidence rate over the past years, is the fifth leading incidence of cancer in female [ 1 ]. Papillary thyroid cancer (PTC) is the dominant historical type which contributes to approximately 84% of all thyroid cancer [ 2 , 3 ]. However, through the low mortality and moderate prognosis were frequently mentioned, the recurrence and the complications are still perplexing those PTC patients [ 4 ]. Besides, lymphatic invasion, cervical lymph node metastasis, larger size of tumor, increasing diagnosis age and extraordinary enlargement of thyroid tissue increase the progression risk of PTC [ 5 ]. Due to early lymph node metastasis happening in PTC frequently [ 6 ], early detection and diagnosis are of great value. Currently multiple methods like thyroid and neck ultrasound, CT/MRI and fine-needle aspiration (FNA) for suspicious lateral neck nodes could effectively diagnose thyroid cancer [ 7 , 8 ]. Yet some of the lymph node metastasis may be troublesome for detecting. We therefore urgently called for a reliable and straightforward approach to determine the possibility of lymph node metastasis. Recent years, high throughput analysis afford us an advanced and efficient technique of evaluating the molecular disruptions in tumor tissues. For instance, A study of predicting chemotherapy sensitivity in cervical cancer, in which expression levels of 22 total and phosphorylated protein were analyzed in 181 frozen tissue samples, resulted in a model that was capable of predicting patients' chemotherapy sensitivity and assessing clinical outcome [ 9 ]. Thus, as more and more cancer sequencing databases are established, there is a great potential for us to acquire and analyze these data and guide clinical decisions with the findings. In order to discover the predictive value of lymph node metastasis in the papillary thyroid cancer transcriptome, we searched for several RNA-seq as well as gene microarray datasets containing both papillary thyroid cancer samples and normal thyroid samples to identify differentially expressed genes. Genes with powerful correlation to lymph nodes metastasis were selected as well. Afterwards, we designed a predicable risk score system and effectively confirmed the model in two independent papillary thyroid cancer cohorts. The entire flow of our efforts to identify predictive models was presented in Additional file 1: Fig. S1. Eventually, we recommended a successful predicable model of lymph node metastasis in papillary thyroid cancer patients with moderate accuracy. Methods Data collection, normalization and preprocessing Gene microarray datasets and the associated clinical data of PTC and normal thyroid samples were downloaded from UCSC XENA (https://xenabrowser.net/) and Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). Totally 8 GEO datasets and 1 The Cancer Genome Atlas (TCGA) dataset were involved in this research. Gene labels in other forms were all normalized into official gene symbols. Furthermore, those samples were proceeded to a quality control step called SimpleAffy to lessen bias. SimpleAffy was used to identify the 3’-to-5’ ratios of GAPDH and β-actin. Outliers were excluded from the study. Differential analysis of gene expression datasets EdgeR R package afforded us the differential expression analysis between papillary thyroid cancer tissues and normal thyroid tissues. The filter thresholds of P value were set as less than 0.05 in integrated analysis. |log2FC (fold change) | was set as more than 0.1 or more than 0.5 in GEO or TCGA microarray separately due to the different sample sizes. A robust rank aggregation (RRA) algorithm was used to determine the overall differentially expressed genes （DEGs） in the 8 GEO datasets in which they were all up-regulated or down-regulated. The most relevant genes could be revealed in the analysis. Functional annotation analysis Gene Ontology (GO) analysis was performed on the DEGs in order to identify their potential functions. The process was settled by the clusterProfiler 3.11. The result of three major GO terms called biological process (BP), cellular component (CC) and molecular function (MF) could be visualized in dot plot or pie chart form by using the ggplot2 R package. Weighted gene co-expression network analysis (WGCNA) All samples of TCGA underwent a sample clustering test to identify their relationships. Those outliers were eliminated and then the others went through soft-threshold defining procedure. A Scale-free gene co-expression network was attained when the correlation coefficient was 0.85. At this point the soft-thresholding power was equal to 12. Afterward, WGCNA algorithm of R software was performed to construct a scale-free network of all DEGs of TCGA. After the construction, cluster analysis was operated to arrange genes with similar expression patterns into gene modules. The minimum size of each module was set as 30. Besides, the cut height threshold for merging modules was defined as 0.25 and some of the modules could be integrated to analyze. Lately, we constructed the relationships between modules and phenotype. The module–trait association was defined as the correlation of module eigengenes (MEs) and traits. Accordingly, each correlation value and the significance value were calculated separately to reveal the relevant modules, which are intently related to the traits. For each gene modules, gene significance (GS) is described as the correlation level between clinical trait and expression pattern. While module membership (MM) stands for the correlation level between MEs and expression pattern. In our study, genes with GS score more than 0.8 and MM score more than 0.2 were defined as module genes with high correlation to certain phenotype. Protein–protein interaction (PPI) network and identification of hub-genes The DEGs of both GEO and TCGA datasets were uploaded to the STRING tool (http://www.string-db.org/) to look into their correlation and to discover the hub-genes in the gene network. A darker line represents a stronger edge confidence in the network. Then, Cytoscape software was utilized to visualize the outcome of STRING database and to acquire the hub-genes which have a closest relation and have a more considerable function among the DEGs. In our study, the 100 genes with the highest connectivity degree were identified as hub-genes of the PPI network. Statistical analysis of model acquisition and validation We randomly separate the TCGA microarray data and relevant clinical data into 2 parts, named a training cohort and a validation cohort. A univariate Logistic regression was applied to DEGs in TCGA training cohort, aid of finding the relationship between those genes and patient’s prognostic data. The results were then diminished by using the LASSO algorithm with the R package ‘glmnet’. The minimal partial likelihood deviance was carried out as optimal tuning parameter (λ) changed. Also some of the coefficients of gene would reduce to zero. Those genes were excluded and the others were accessed to the multivariate Logistic test. The hazard ratio (HR) and 95% confidence intervals (CI) of each gene would be calculated and only those genes which did not include 1 in the 95% CI were selected as final trait-relative genes. The coefficient of each gene in multivariate Logistic regression model was summed to calculate the risk score of lymph node metastasis. The function could be computed as follows, in which βi stands for the Logistic regression coefficient of gene i in the training cohort. By calculating risk score of each sample, we could sort them and divide the cohort into low risk group and high risk group by the median value. A student’s t test of two groups in comparing the trait demonstrated that the model could predict the risk effectively. Meanwhile, the receiver operating characteristic (ROC) curves was used to verify the sensitivity and specificity of lymph node metastasis related model risk prediction. After the model construction, the corresponding approach above was applied to TCGA validation cohort as well as GEO independent cohort in turn for risk predicting system confirmation. Results Identification of DEGs and associated GO analysis in GEO datasets Firstly, all the human tissue microarrays including papillary thyroid carcinoma samples and paired/unpaired normal thyroid samples were extracted from the GEO database. In order to make sure the quality of the research, SimpleAffy R package was utilized to determine the 3’-to-5’ ratios of β-actin and GAPDH (Additional file 2: Fig. S2). After excluding seven tumor samples and 2 normal samples in three datasets, a total of 187 tumor samples and 119 normal control samples in eight datasets (GSE33630 [10, 11], GSE60542 [12], GSE66783 [13], GSE5364 [14], GSE129562 [15], GSE97001 [16], GSE3467 [17] and GSE27155 [18, 19]) were included this research (Table 1). After performing differential analysis on the 8 datasets by the edgeR and robust rank aggregation algorithm, the DEGs between the normal tissues and PTC tissues of each dataset were obtained, with an adjusted P value 0.1 as the cut-offs. A total of 531 over-expressed genes and 474 suppressed genes were discovered in PTC tissues. The top ten genes that were overexpressed or suppressed are listed by a heatmap in Fig. 1a. GO analysis was performed on the DEGs in order to identify potential functions. Those DEGs were enriched in GO terms including 500 BPs, 70 CCs and 28 MFs with the cut-off value set as adjust P value < 0.01 and q value < 0.01. TOP 10 GO terms of each category were displayed in Fig. 1b. Among them, extracellular structure organization and extracellular matrix organization, cell-substrate adhesion and renal system development were three significant aspects in BP category. The first two functions are also remarkable in CC term. The most enriched MF terms were sulfur compound binding, glycosaminoglycan binding and serine-type peptidase activity. Identification of DEGs and associated GO analysis in TCGA dataset The TCGA-THCA dataset consisting of 56 normal thyroid tissues and 497 thyroid cancer tissues were selected by integrated analysis ( P value 0.5 set as cut-offs), from which 6492 mRNAs with steady differentially expressed patterns were identified by edgeR analysis. Similarly, these DEGs were analyzed by GO functional enrichment analysis to find out the potential biological functions. The BP term were shown in a pie chart (Fig. 2a). Activation of MAPK activity and leukocyte mediated cytoxicity are the top 2 enriched terms of BPs. While CCs and MFs were exhibited in Fig. 2b. Each category only shows the top ten functions. Discovery of a strong correlation between lymph node metastasis and DEGs in TCGA dataset by WGCNA algorithm For DEGs of thyroid cancer tissues and normal thyroid tissues in the TCGA dataset, we used WGCNA algorithm to find out which clinical traits they are significantly related to. Firstly, after screening the clinical characteristics and sample tree, a total of 482 thyroid cancer samples were included in the study. In addition, the sample tree and the clinical characteristics of each sample are also shown in Additional file 3: Fig. S3a. Then, through the soft threshold screening (Additional file 3: Fig. S3b), a gene co-expression network with the scale-free characteristics was established, in which the evaluation parameter R 2 was set as 0.85. After the network was established, genes with similar expression level were grouped in the same module and displayed in the form of a hierarchical clustering diagram (Additional file 3: Fig. S3c). Different colors indicated separate modules. It should be noted that the genes in gray color were not classified into any modules. After that, the clinical characteristics of the samples were taken out and analyzed for correlation with each module (Fig. 3). We can find that the brown module (r =-0.34, p=3e-14) and the turquoise module (r= 0.35, p=6e-15) were highly correlated with lymph node metastasis. By calculating the MM value of these two modules (cut-off set as MM P value 0.8), a total of 106 genes that were closely related to the brown module and the turquoise module were identified for further analysis. Since the results of WGCNA strongly implied the relevance of these DEGs to lymph node metastasis, we subsequently concentrated on lymph node metastasis and attempted to explore models that could accurately predict lymph node metastasis in PTC. Potential hub-genes relating to lymph node metastasis revealed by PPI network The shared DEGs in both GEO datasets and TCGA dataset were considered as important genes for papillary thyroid cancer. To narrow the scope and identify the much meaningful genes among them, PPI network analysis was used to find key regulatory genes in these genes. STRING database ( https://string-db.org) was used to identify the PPI network of these 599 shared DEGs. Fig. 5a showed the overall PPI regulatory network of these differential genes. through the PPI network connection score, one cluster with the highest score was displayed in Fig. 5b. After all genes going through PPI analysis, some potential hub-genes can be found based on the score. The first 100 hub-genes were extracted as key genes. What’s more, those hub-genes were intersected with the WGCNA module genes mentioned above (Fig. 5c). A total of 9 genes were identified as important DEGs and related to lymph node metastasis. construction a risk scoring system by logistic regression analysis and LASSO algorithm We randomly divided the thyroid cancer samples in the TCGA database into two parts. The first cohort named training cohort was utilized to find the risk scoring system of lymph nodes metastasis. Meanwhile, the other one named validation cohort was used to confirm the system. Also, an independent GEO cohort with clinical traits was enrolled to verify the model. Their clinical features were listed in Table 2. Univariate logistic analysis was performed on the 9 genes to find out if they related to lymph nodes metastasis (Fig. 4d). A predictive gene was defined as a hazard ratio (HR) and 95% confidence interval (CI) greater than or less than 1 and P value less than 0.05. It is gratifying that these nine genes were all found to be associated with lymph node metastasis after the analysis. Among them, EPHB3, also called EPH Receptor B3; MET, one of the receptor tyrosine kinase; ICAM1, also named Intercellular Adhesion Molecule 1; SERPINA1, also called Serpin Family A Member 1 and FN1, also named Fibronectin 1 are the single risk factors for lymph nodes metastasis, while ITPR1, also named Inositol 1,4,5-Trisphosphate Receptor Type 1; PPARGC1A,also called PPARG Coactivator 1 Alpha; GNA14, also named G Protein Subunit Alpha 14 and BCL2,a apoptosis regulator were found as protective factors in this trait. In order to compress the model and identity the key genes, A least absolute shrinkage and selection operator (LASSO) regression model was then used to test these 9 Genes. In the LASSO model, when the value of λ increases, more coefficients (genes) will be set to zero, meaning those variables could be remove from the model due to their shrinking property (Fig. 5a). Six genes model satisfied the minimum partial likelihood deviance due to the ridge regression. The minimum log(λ) was -3.67 at this status. Finally, these six genes were subjected to multivariate logistic analysis in order to find the final risk prediction model. The result indicated that MET, ITPR1, and BCL2 were independent prognostic factors for lymph nodes metastasis (Fig. 5c, Table 3). The score of risk estimation model could be calculated as: expression of ITPR11 × 0.589 + expression of MET × 0.841 - expression of BCL2 × 0.786. The factors stand for the respective multivariate Logistic regression coefficients. Verification of the prediction system through validation cohorts from TCGA and GEO datasets TCGA training cohort used for model construction were separate into high-risk group and low-risk group based on each sample's risk score (Fig. 6a). The median risk score (3.879) was set as the cut-off. For two groups, a significant difference of lymph nodes metastasis was clearly shown in heatmap (Fig. 6b). A student’s t test also demonstrated that the high-risk group had a higher frequency of lymph nodes metastasis than the low-risk group (Fig. 6c) .The receiver operating characteristic (ROC) curve shows the efficiency of the prediction model, in which the area under the curve (AUC) was 0.744 (Fig. 6d). In order to verify the prediction accuracy of the risk prediction model, we used multiple cohorts to verify. TCGA validation cohort has been operated by the same protocol (Fig. 6e). Obviously, the result also showed that a high risk score coul d be riskier to lymph nodes metastasis by a heatmap (Fig. 6f) and student’s t test (Fig. 6g). The AUC over validation model is 0.711 (Fig. 6h). At the same time, we conducted an independent verification in another cohort combined from two GEO dataset (GSE3467 and GSE60542) with the same platform (GPL570) and corresponding clinical data, in order to show that the risk scoring system is generally applicable (Fig. 7a). In the heatmap, a strong tendency was discovered of which higher risk score indicated higher frequency of lymph nodes metastasis (Fig. 7b). A student’s t test verifies the point (Fig. 7d). Furthermore, gene expressions of each sample tissue were displayed in Fig. 7c as a heatmap. Finally, we performed a ROC analysis and a meaningful result was revealed with 0.6842 of AUC value. Accordingly, the three-genes model was verified in multiple microarray and all showed a great difference. It is a reliable, accurate and independent predictive appliance for determining lymph nodes metastasis in papillary thyroid cancer patients. Discussion Genomic analysis has an extraordinarily critical role in tumor research. For example, Paik S et al. derived a 21-genes based recurrence scoring system from a prospective analysis of multiple gene expression levels in a breast cancer population [ 20 ]. Besides, transcriptome mapping could also be applied for the determination of molecular subtypes of tumors. It has already been implemented in colorectal cancer [ 21 ], breast cancer [ 22 ], prostate cancer [ 23 ], and pancreatic cancer [ 24 ], which has a facilitating effect on clinical decision-making. Moreover, genomic studies provide an insight into the tumor immune microenvironment. Xu M et al. calculated the corresponding immune infiltration scores from the expression of immune-related molecules in breast cancer specimens and the immune score was found to perform a detrimental effect in overall survival and recurrence-free survival [ 25 ]. Chakladar J et al. also analyzed immune-related genes in PTC through the combined application of genomics and transcriptomics [ 26 ]. In the present research, we investigated multiple independent datasets of PTC, and successfully established a three-gene (MET, ITPR1 and BCL2) prediction model for lymph node metastasis in PTC patients. In our study, MET and ITPR1 expression in the predictive model was a risk factor for lymph node metastasis, whereas BCL2 was a protective factor. Previous study has reported that BCL2 was found to be highly expressed only in poorly differentiated tumors [ 27 ]. Moreover, another study also showed that a lower expression level of BCL2 could act as an early sign of oncogenesis and be a reason for the favorable prognosis [ 28 ], suggesting that BCL2 may act as a protective factor in thyroid cancer. As for ITPR1, one study demonstrated that up-expression of ITPR1 shelters renal cancer cells against natural killer cells [ 29 ]. Another study showed that ITPR1 could enhance paclitaxel toxicity in breast cancer [ 30 ]. We discovered ITPR1 is a risk factor for lymph node metastasis in thyroid cancer, but the biological mechanism still under solving. As a heterodimeric transmembrane receptor tyrosine kinase, MET mediates the activation of multiple signaling pathways, including PI3K/AKT, Ras-Rac/Rho and phospholipase C-γ pathways [ 31 ]. Significantly higher level of MET was detected in PTC [ 32 , 33 ], non-small cell lung cancer [ 34 ], bladder cancer [ 35 ] and oral cancer [ 36 ]. Those results all add up to a better proof of MET as a cancer-promoting factor and can be corroborated with our findings. In conclusion, based on the available studies, BCL2 and MET match the results more accurately, while ITPR1 in thyroid cancer has been less studied and needs to be further explored. The Robust rank aggregation algorithm was utilized to analyze multiple gene sets integrally and identify the common DEGs. This algorithm compensates for the limitations of the previous single data set analysis and minimizes bias. Since PTC samples are generally small, the RRA algorithm could be quite helpful. WGCNA is an effective method for describing associations of gene expression patterns with clinical phenotypes. In thyroid cancer research, WGCNA was widely used [ 37 , 38 ]. Based on the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) statement, studies developing new prediction models should always go through an internal validation (also called self-validation) to quantify the predictive appearance. Also, it is firmly recommended to appraise the model in other data (also called external validation) after developing a prediction model [ 39 ]. We then applied our prediction model to TCGA self-validation cohort and one independent GEO cohort. The outcomes of two dataset strongly fit in the model. The innovative feature of our study is that the combined application of RRA algorithm, WGCNA analysis and PPI methods was first used to analyze the correlation between PTC and lymph node metastasis. The degradation of samples may cause bias of the results. Therefore, we performed quality control on each sample in the dataset. Tissues that did not meet the requirements (degradation occurred) were excluded from the follow-up experiment. Furthermore, the predictive model of PTC lymph node metastasis was uniquely established, which might be useful for therapeutic decision-making and clinical monitoring. However, the inadequacy of this study is that the prognostic data in the TCGA database for papillary thyroid cancer are quite good and we failed to find significant differences in prognosis, while the GEO dataset was unable to find prognosis information. Therefore, it is regrettable that prognosis cannot be measured. For PTC, lymph node metastasis may arise at an early stage and there is a potential risk for skip metastasis [ 40 ], the mechanism of which is currently not clear. Routine diagnosis methods such as neck CT/MRI and neck lymph node ultrasound [ 7 , 8 ] may not be able to fully detect the development of lymph node metastasis. Patients who develop lymph node metastases are likely to require postoperative I 131 radiation therapy and may have a higher recurrence rate and lower survival rate [ 41 ]. Therefore, the outcome of this study may have a better role in predicting lymph node metastasis in PTC patients. Patients in the low-risk group have a lower likelihood of lymph node metastasis. With current conventional methods of measuring expression level, such as RT-qPCR or immunohistochemistry, we can easily, accurately and economically obtain risk scores for this patient, thus allowing the model to be better applied in clinical practice. Conclusions Summarily, this study is a highly scientific and accurate method with successful predictive significance for determining lymph node metastasis in PTC patient. Among the model, 2 genes were identified as risk factors and 1 genes was protective factor in PTC patients. It might be potential treatment targets and urged for further research. Abbreviations PTC, Papillary Thyroid Cancer; CT, Computed Tomography; MRI, Magnetic Resonance Imaging; GEO, Gene Expression Omnibus; TCGA, The Cancer Genome Atlas; DEGs, Differentially Expressed Genes; RRA, Robust Rank Aggregation; GO, Gene Ontology; WGCNA, Weighted Gene Co-expression Network Analysis; MEs, Module Eigengenes; GS,Gene Significance; MM, Module Membership; PPI, Protein–Protein Interaction; HR, Hazard Ratio; CI, Confidence Intervals; ROC, Receiver Operating Characteristic; FC, Fold Change; BP, Biological Process; CC, cellular component; MF, molecular functions. Declarations Acknowledgements Not applicable. Authors’ contributions PY, ZH, and YL conceived and arranged the experiments, and wrote the manuscript. KL, QZ, WL, MX, JZ, YJ, WY and LY collected and analyzed the data. All authors read and approved the final version of the manuscript. Funding Not applicable. Availability of data and materials Gene microarray datasets and the associated clinical data of PTC and normal thyroid samples in our study were all publicly available. The data of PTC patients from TCGA were downloaded from UCSC XENA (https://xenabrowser.net/). Additionally, Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) afford us 8 cohorts (GSE33630, GSE60542, GSE66783, GSE5364, GSE129562, GSE97001, GSE3467 and GSE27155) for further research in the study. Ethics approval and consent to participate This study was approved by Ethics Committee of Huazhong University of science and technology (HUST). Consent for publication All authors have agreed for this publication. Competing interests The authors declare that they have no competing interests. References Siegel RL, Miller KD, Jemal A: Cancer statistics, 2020 . CA Cancer J Clin 2020, 70 (1):7-30. Geraldo MV, Kimura ET: Integrated Analysis of Thyroid Cancer Public Datasets Reveals Role of Post-Transcriptional Regulation on Tumor Progression by Targeting of Immune System Mediators . PLoS One 2015, 10 (11):e0141726. Aschebrook-Kilfoy B, Kaplan EL, Chiu BCH, Angelos P, Grogan RH: The Acceleration in Papillary Thyroid Cancer Incidence Rates is Similar Among Racial and Ethnic Groups in the United States . Ann Surg Oncol 2013, 20 (8):2746-2753. Chrisoulidou A, Boudina M, Tzemailas A, Doumala E, Iliadou PK, Patakiouta F, Pazaitou-Panayiotou K: Histological subtype is the most important determinant of survival in metastatic papillary thyroid cancer . Thyroid Res 2011, 4 (1):12. Cheng Q, Li X, Acharya CR, Hyslop T, Sosa JA: A novel integrative risk index of papillary thyroid cancer progression combining genomic alterations and clinical factors . Oncotarget 2017, 8 (10):16690-16703. Moo TA, McGill J, Allendorf J, Lee J, Fahey T, 3rd, Zarnegar R: Impact of prophylactic central neck lymph node dissection on early recurrence in papillary thyroid carcinoma . World J Surg 2010, 34 (6):1187-1191. Grani G, Ramundo V, Falcone R, Lamartina L, Montesano T, Biffoni M, Giacomelli L, Sponziello M, Verrienti A, Schlumberger M et al : Thyroid Cancer Patients With No Evidence of Disease: The Need for Repeat Neck Ultrasound . J Clin Endocrinol Metab 2019, 104 (11):4981-4989. Torlontano M, Attard M, Crocetti U, Tumino S, Bruno R, Costante G, D'Azzo G, Meringolo D, Ferretti E, Sacco R et al : Follow-up of low risk patients with papillary thyroid cancer: role of neck ultrasonography in detecting lymph node metastases . J Clin Endocrinol Metab 2004, 89 (7):3402-3407. Choi CH, Chung JY, Kang JH, Paik ES, Lee YY, Park W, Byeon SJ, Chung EJ, Kim BG, Hewitt SM et al : Chemoradiotherapy response prediction model by proteomic expressional profiling in patients with locally advanced cervical cancer . Gynecol Oncol 2020, 157 (2):437-443. Tomas G, Tarabichi M, Gacquer D, Hebrant A, Dom G, Dumont JE, Keutgen X, Fahey TJ, 3rd, Maenhaut C, Detours V: A general method to derive robust organ-specific gene expression-based differentiation indices: application to thyroid cancer diagnostic . Oncogene 2012, 31 (41):4490-4498. Dom G, Tarabichi M, Unger K, Thomas G, Oczko-Wojciechowska M, Bogdanova T, Jarzab B, Dumont JE, Detours V, Maenhaut C: A gene expression signature distinguishes normal tissues of sporadic and radiation-induced papillary thyroid carcinomas . Br J Cancer 2012, 107 (6):994-1000. Tarabichi M, Saiselet M, Tresallet C, Hoang C, Larsimont D, Andry G, Maenhaut C, Detours V: Revisiting the transcriptional analysis of primary tumours and associated nodal metastases with enhanced biological and statistical controls: application to thyroid cancer . Br J Cancer 2015, 112 (10):1665-1674. Lan X, Zhang H, Wang Z, Dong W, Sun W, Shao L, Zhang T, Zhang D: Genome-wide analysis of long noncoding RNA expression profile in papillary thyroid carcinoma . Gene 2015, 569 (1):109-117. Yu K, Ganesan K, Tan LK, Laban M, Wu J, Zhao XD, Li H, Leung CH, Zhu Y, Wei CL et al : A precisely regulated gene expression cassette potently modulates metastasis and survival in multiple solid cancers . PLoS Genet 2008, 4 (7):e1000129. Lee S, Bae JS, Jung CK, Chung WY: Extensive lymphatic spread of papillary thyroid microcarcinoma is associated with an increase in expression of genes involved in epithelial-mesenchymal transition and cancer stem cell-like properties . Cancer Med 2019, 8 (15):6528-6537. Iacobas DA, Tuli NY, Iacobas S, Rasamny JK, Moscatello A, Geliebter J, Tiwari RK: Gene master regulators of papillary and anaplastic thyroid cancers . Oncotarget 2018, 9 (2):2410-2424. He H, Jazdzewski K, Li W, Liyanarachchi S, Nagy R, Volinia S, Calin GA, Liu CG, Franssila K, Suster S et al : The role of microRNA genes in papillary thyroid carcinoma . Proc Natl Acad Sci U S A 2005, 102 (52):19075-19080. Giordano TJ, Kuick R, Thomas DG, Misek DE, Vinco M, Sanders D, Zhu Z, Ciampi R, Roh M, Shedden K et al : Molecular classification of papillary thyroid carcinoma: distinct BRAF, RAS, and RET/PTC mutation-specific gene expression profiles discovered by DNA microarray analysis . Oncogene 2005, 24 (44):6646-6656. Giordano TJ, Au AY, Kuick R, Thomas DG, Rhodes DR, Wilhelm KG, Jr., Vinco M, Misek DE, Sanders D, Zhu Z et al : Delineation, functional validation, and bioinformatic evaluation of gene expression in thyroid follicular carcinomas with the PAX8-PPARG translocation . Clin Cancer Res 2006, 12 (7 Pt 1):1983-1993. Paik S, Shak S, Tang G, Kim C, Baker J, Cronin M, Baehner FL, Walker MG, Watson D, Park T et al : A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer . N Engl J Med 2004, 351 (27):2817-2826. Guinney J, Dienstmann R, Wang X, de Reynies A, Schlicker A, Soneson C, Marisa L, Roepman P, Nyamundanda G, Angelino P et al : The consensus molecular subtypes of colorectal cancer . Nat Med 2015, 21 (11):1350-1356. Lehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, Shyr Y, Pietenpol JA: Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies . J Clin Invest 2011, 121 (7):2750-2767. Lapointe J, Li C, Higgins JP, van de Rijn M, Bair E, Montgomery K, Ferrari M, Egevad L, Rayford W, Bergerheim U et al : Gene expression profiling identifies clinically relevant subtypes of prostate cancer . Proc Natl Acad Sci U S A 2004, 101 (3):811-816. Bailey P, Chang DK, Nones K, Johns AL, Patch AM, Gingras MC, Miller DK, Christ AN, Bruxner TJ, Quinn MC et al : Genomic analyses identify molecular subtypes of pancreatic cancer . Nature 2016, 531 (7592):47-52. Xu M, Li Y, Li W, Zhao Q, Zhang Q, Le K, Huang Z, Yi P: Immune and Stroma Related Genes in Breast Cancer: A Comprehensive Analysis of Tumor Microenvironment Based on the Cancer Genome Atlas (TCGA) Database . Front Med (Lausanne) 2020, 7 :64. Chakladar J, Chu M, Gnanasekar A, Rosenberg KF, Tsai JC, Wong LM, Ongkeko WM: Computational analysis of immune-associated genomic and transcriptomic elements differentiating papillary thyroid cancer subtypes . Cancer Research 2019, 79 (13). Soda G, Antonaci A, Bosco D, Nardoni S, Melis M: Expression of bcl-2, c-erbB-2, p53, and p21 (waf1-cip1) protein in thyroid carcinomas . J Exp Clin Cancer Res 1999, 18 (3):363-367. Aksoy M, Giles Y, Kapran Y, Terzioglu T, Tezelman S: Expression of bcl-2 in papillary thyroid cancers and its prognostic value . Acta Chir Belg 2005, 105 (6):644-648. Messai Y, Noman MZ, Hasmim M, Janji B, Tittarelli A, Boutet M, Baud V, Viry E, Billot K, Nanbakhsh A et al : ITPR1 protects renal cancer cells against natural killer cells by inducing autophagy . Cancer Res 2014, 74 (23):6820-6832. Xu S, Wang P, Zhang J, Wu H, Sui S, Zhang J, Wang Q, Qiao K, Yang W, Xu H et al : Ai-lncRNA EGOT enhancing autophagy sensitizes paclitaxel cytotoxicity via upregulation of ITPR1 expression by RNA-RNA and RNA-protein interactions in human cancer . Mol Cancer 2019, 18 (1):89. Birchmeier C, Birchmeier W, Gherardi E, Vande Woude GF: Met, metastasis, motility and more . Nat Rev Mol Cell Biol 2003, 4 (12):915-925. Chitikova Z, Pusztaszeri M, Makhlouf AM, Berczy M, Delucinge-Vivier C, Triponez F, Meyer P, Philippe J, Dibner C: Identification of new biomarkers for human papillary thyroid carcinoma employing NanoString analysis . Oncotarget 2015, 6 (13):10978-10993. Wang G, Cai C, Chen L: MicroRNA-3666 Regulates Thyroid Carcinoma Cell Proliferation via MET . Cell Physiol Biochem 2016, 38 (3):1030-1039. Lutterbach B, Zeng Q, Davis LJ, Hatch H, Hang G, Kohl NE, Gibbs JB, Pan BS: Lung cancer cell lines harboring MET gene amplification are dependent on Met for growth and survival . Cancer Res 2007, 67 (5):2081-2088. Shintani T, Kusuhara Y, Daizumoto K, Dondoo TO, Yamamoto H, Mori H, Fukawa T, Nakatsuji H, Fukumori T, Takahashi M et al : The Involvement of Hepatocyte Growth Factor-MET-Matrix Metalloproteinase 1 Signaling in Bladder Cancer Invasiveness and Proliferation. Effect of the MET Inhibitor, Cabozantinib (XL184), on Bladder Cancer Cells . Urology 2017, 101 :169 e167-169 e113. Saintigny P, William WN, Jr., Foy JP, Papadimitrakopoulou V, Lang W, Zhang L, Fan YH, Feng L, Kim ES, El-Naggar AK et al : Met Receptor Tyrosine Kinase and Chemoprevention of Oral Cancer . J Natl Cancer Inst 2018, 110 (3). Zhai T, Muhanhali D, Jia X, Wu Z, Cai Z, Ling Y: Identification of gene co-expression modules and hub genes associated with lymph node metastasis of papillary thyroid cancer . Endocrine 2019, 66 (3):573-584. Tang X, Huang X, Wang D, Yan R, Lu F, Cheng C, Li Y, Xu J: Identifying gene modules of thyroid cancer associated with pathological stage by weighted gene co-expression network analysis . Gene 2019, 704 :142-148. Collins GS, Reitsma JB, Altman DG, Moons KG: Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement . BMJ 2015, 350 :g7594. Machens A, Holzhausen HJ, Dralle H: Skip metastases in thyroid cancer leaping the central lymph node compartment . Arch Surg 2004, 139 (1):43-45. Podnos YD, Smith D, Wagman LD, Ellenhorn JD: The implication of lymph node metastasis on survival in patients with well-differentiated thyroid cancer . Am Surg 2005, 71 (9):731-734. Tables Table 1 Information of enrolled PTC patients from 8 GEO datasets after quality control. Country Organization Series Platform Normal Tumor Quality control Publication Belgium ULB GSE33630 GPL570 45 47 Excluded 2 tumor samples (Collins et al. 2015, Dom et al. 2012) Belgium IRIBHM GSE60542 GPL570 30 33 Passed (Tarabichi et al. 2015) China The First Hospital of China Medical University GSE66783 GPL19850 5 5 Passed (Lan et al. 2015) Singapore National Cancer Centre GSE5364 GPL96 16 35 Passed (Yu et al. 2008) South Korea The Catholic University of Korea GSE129562 GPL10558 8 8 Passed (Lee et al. 2019) USA Center for Computational Systems Biology GSE97001 GPL10332 4 4 Passed (Iacobas et al. 2018) USA Ohio State University GSE3467 GPL570 7 5 Excluded 4 tumor and 2 normal samples (He et al. 2005) USA University of Michigan GSE27155 GPL96 4 50 Excluded 1 tumor sample ((Giordano et al. 2006; Giordano et al. 2005) Table 2 Clinical pathological characteristics of patients in the training, self-validation cohorts and the independent GEO cohort. Characteristics TCGA training cohort TCGA validation cohort GEO validation cohort (N = 248) (N = 249) (N = 42) Age at initial diagnosis (year) 46.3 ± 15.5 48.5 ± 16.1 45.11 ± 13.5 Gender Male 82(33.06%) 52(20.88%) 19(45.24%) Female 166(66.94%) 197(79.12%) 23(54.76%) Pathologic T T1 or Tx 69(27.82%) 75(30.12%) 9(21.43%) T2 81(32.66%) 81(32.53%) 4(9.52%) T3 87(35.08%) 82(32.93%) 24(57.12%) T4 11(4.44%) 11(4.42%) 5(11.90%) Pathologic N N0 or Nx 139(56.05%) 137(55.02%) 19(45.24%) N1 109(43.95%) 112(44.98%) 23(54.76%) Pathologic M M0 or Mx 244(98.39%) 245(98.39%) 39(92.86%) M1 4(1.61%) 4(1.61%) 3(7.12%) Tumor stage Stage I 144(58.06%) 135(54.22%) 19(45.24%) Stage II 27(10.89%) 25(10.04%) 0(0.00%) Stage III 52(20.97%) 59(23.69%) 10(23.81%) Stage IV 24(9.68%) 29(11.65%) 3(7.14%) Not report 1(0.40%) 1(0.40%) 10(23.81%) Overall survival status NA Alive 242(97.58%) 239(95.98%) Dead 6(2.42%) 10(4.02%) Table 3 The three independent prediction factors of lymph node metastasis in papillary thyroid cancer. Entrez ID Gene multivariate Cox regression analysis Eight GEO datasets TCGA dataset Coefficient Hazard ratio confidence interval (95%) P value Log 2 FC Adjusted P value Log 2 FC Adjusted P value 3708 ITPR1 0.589 1.801 1.133–2.864 0.013 0.322 3.532e-05 1.161 3.343e-24 4233 MET 0.841 2.320 1.606–3.350 7.000e-06 0.686 2.194e-08 0.620 4.746e-23 596 BCL2 -0.786 0.456 0.271–0.767 0.030 -0.342 2.736e-04 -0.689 5.406e-26 Supplementary Files AdditionalFigureLegends.docx Additionalfig3.tif Additionalfig2.tif Additionalfig1.tif Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-41157","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Primary research","associatedPublications":[],"authors":[{"id":817401,"identity":"a86314d8-d13d-463e-88d8-a8ca26606173","order_by":0,"name":"Ziwei Huang","email":"","orcid":"https://orcid.org/0000-0003-2070-8981","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Ziwei","middleName":"","lastName":"Huang","suffix":""},{"id":817402,"identity":"2ba6c9d3-3e7b-4482-a6a5-fb61acaae358","order_by":1,"name":"Yuenan Liu","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yuenan","middleName":"","lastName":"Liu","suffix":""},{"id":817403,"identity":"ed35e465-e6b4-4844-8450-f22538f39e4c","order_by":2,"name":"Kehao Le","email":"","orcid":"","institution":"Zhejiang University School of Medicine Sir Run Run Shaw Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Kehao","middleName":"","lastName":"Le","suffix":""},{"id":817404,"identity":"0e6b99bc-033c-4a23-944b-4635a5fe22a9","order_by":3,"name":"Ming Xu","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Ming","middleName":"","lastName":"Xu","suffix":""},{"id":817405,"identity":"455897ee-ed7f-42be-94fb-09bda3dd1617","order_by":4,"name":"Wenhui Li","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Wenhui","middleName":"","lastName":"Li","suffix":""},{"id":817406,"identity":"4e39577a-a8c5-46cc-8b46-4c5e71b5d970","order_by":5,"name":"Qiuyang Zhao","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Qiuyang","middleName":"","lastName":"Zhao","suffix":""},{"id":817407,"identity":"e5ee5808-50da-4183-bd49-95c9e1c47ae5","order_by":6,"name":"Jun Zhou","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jun","middleName":"","lastName":"Zhou","suffix":""},{"id":817408,"identity":"fdd02eb7-02ac-4bf1-9d93-64d08b482542","order_by":7,"name":"Yujia Jiang","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yujia","middleName":"","lastName":"Jiang","suffix":""},{"id":817409,"identity":"09182f7c-6ba7-4b82-9724-42a2c422ca66","order_by":8,"name":"Wen Yang","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Wen","middleName":"","lastName":"Yang","suffix":""},{"id":817410,"identity":"edea7fc9-8401-4bcd-9430-069d1834d98e","order_by":9,"name":"Li Yang","email":"","orcid":"","institution":"Wuhan Union Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Yang","suffix":""},{"id":817411,"identity":"9f66f4fb-db1d-49d0-9e74-caef2c0adac1","order_by":10,"name":"Pengfei Yi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+klEQVRIiWNgGAWjYDACZjApx8DGzHzwwYcKCTl+IrUYM7CxsyUbzjhjYSzZQJxdxgwM/Dxm0pxtFYkbCGkxOM578HHBLwM5PmYeY2PGeRKMGxiYHz66gU/LYb5k45l9BsZszGyFjwu3STCbM7AZG+fg1QJ0D2/Pn8Q2ZubNxjO3SbBZNvCwSRPQYv6bt8egvo2ZAah3jgSPwQHCWsyYeX4YJLAxswC1NEhIENQiCfQLUKWBYRszKJCPSRhINhPwC9/5swc/8/wxkJfvPwyMypq6+n725oeP8WlROMDDwMDYhizEjEc5CMg3ALUw/CGgahSMglEwCkY2AAApDkRvN5f5kwAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0003-1655-9696","institution":"","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Pengfei","middleName":"","lastName":"Yi","suffix":""}],"badges":[],"createdAt":"2020-07-11 15:02:30","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-41157/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-41157/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":1555424,"identity":"15fad247-62bf-480c-a288-7b95321a35d7","added_by":"auto","created_at":"2020-07-14 17:49:01","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":5683713,"visible":true,"origin":"","legend":"Identification of DEGs in GEO datasets along with their GO analysis. a TOP 10 up-regulated and down-regulated genes in eight GEO datasets were displayed as a heat map. Each grid represents the differential expression of these genes in each dataset with the log2FC inset, while the red color representing up-regulated genes and the green color representing down-regulated genes. b Significantly enriched GO terms of DEGs in GEO datasets. Each category including Biological process, Cellular Components and Molecular Functions were shown in separate charts. A bigger size of circle informed more genes were enriched. Respectively, the color of each circle represented the adjust P value.","description":"","filename":"Fig1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Fig1.jpg"},{"id":1555425,"identity":"a0bbae5f-2df8-45d1-86f3-874465dfd3cb","added_by":"auto","created_at":"2020-07-14 17:49:02","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":4078815,"visible":true,"origin":"","legend":"Plots of the enriched GO terms of DEGs in TCGA dataset. a Biological process, one of the GO categories, was presented in the form of a circle chart. TOP 5 enriched terms were listed below. b Cellular Components and Molecular Functions were shown in dot plot. The size of circle represented the numbers which genes in this term.","description":"","filename":"Fig2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Fig2.jpg"},{"id":1555426,"identity":"5e938f6e-5d90-4052-a969-85f5790b4093","added_by":"auto","created_at":"2020-07-14 17:49:02","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":3234319,"visible":true,"origin":"","legend":"A strong correlation between lymph node metastasis and DEGs found by WGCNA. The Pearson correlation coefficient was shown in the box and the P value was shown in the brackets below. Values with significant differences were marked in red color. Abbreviations: T, Primary tumor; N, Regional lymph nodes; M, Metastasis.","description":"","filename":"Fig3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Fig3.jpg"},{"id":1555427,"identity":"1f2b6cce-850e-4677-a427-c24ea8f82acc","added_by":"auto","created_at":"2020-07-14 17:49:02","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":7851963,"visible":true,"origin":"","legend":"Revealing of potential hub-genes relating to lymph node metastasis. a The overall PPI network made of 599 DEGs in both TCGA and GEO datasets. b One of the clusters that had the highest score in overall PPI network. c Venn gram revealed the intersection between top 100 genes in PPI network and the genes with high relationship to modules that strongly related to lymph node metastasis. d Forest illustration of the univariate logistic analysis. P value and Hazard ratio of each gene were shown on the left side. Hazard ratio less than 1 indicated that the gene was a protective factor. Meanwhile a bigger-than-1’s hazard ratio indicated a harmful factor.","description":"","filename":"Fig4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Fig4.jpg"},{"id":1555428,"identity":"8bd701d1-02d1-4223-a080-d7f47e9d37b4","added_by":"auto","created_at":"2020-07-14 17:49:02","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":7844529,"visible":true,"origin":"","legend":"Drawing three genes most relevant to lymph node metastasis by LASSO algorithm and multivariate Logistic regression. a The optimal parameter (λ) was chosen by cross validation. The dashed vertical line on the left intersects over the best log λ, corresponding to the maximum value of AUC. b LASSO coefficient plot of 9 DEGs. Each curve represents a coefficient and the x-axis represents the regularization penalty parameter. Those coefficients that do not become zero as x changes are included in the LASSO regression model. c HRs and 95% CIs of the three genes based on multivariate Logistic regression analysis of the training cohort from TCGA-THCA dataset.","description":"","filename":"Fig5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Fig5.jpg"},{"id":1555429,"identity":"8761189c-ba20-4755-a203-d330f5b2e4af","added_by":"auto","created_at":"2020-07-14 17:49:03","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":2505027,"visible":true,"origin":"","legend":"Obtainment of the risk score system from TCGA cohort. The median risk score, which served as the cut-off value for dividing the high-risk and low-risk groups, was presented as a horizontal dashed line. The vertical dashed line separated patients on the standard of the high-risk (red) and low-risk (blue) (a, e). The distribution of lymph node metastasis in the training (b) and validation (f) cohorts from TCGA was illustrated. Patients with lymph node metastasis were shown in red, while patients without lymph node metastasis were shown in blue. The result of student’s t test of patients predicted to be at risk for poor outcomes in the training (c) and validation (g) cohorts from TCGA. The number of patients remaining at a particular timepoint was shown at the bottom. ROC curves for predicting Lymph node metastasis in the training (d) and validation (h) cohorts from TCGA. ****, P\u003c0.0001. Error bars indicate mean ± SD.","description":"","filename":"Fig6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Fig6.jpg"},{"id":1555430,"identity":"fa93cc29-a109-4670-8011-c1de8142135a","added_by":"auto","created_at":"2020-07-14 17:49:03","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":6167162,"visible":true,"origin":"","legend":"Verification of the risk scoring system through the GEO independent datasets. a The distribution of risk scores in independent GEO dataset. b The distribution of patients with (red) or without (blue) lymph node metastasis. Patients were ranked in ascending order of risk score. c The expression of the three genes in valuating system of all the papillary thyroid cancer patients in GEO datasets. Student's t test of the risk score between N0 and N1 patients in GEO dataset (d) and TCGA total dataset (f). ROC curves for predicting Lymph node metastasis in the GEO dataset (e) and TCGA total dataset (g). *, P value\u003c0.05, ****, P value\u003c0.0001. Error bars indicate mean ± SD.","description":"","filename":"Fig7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Fig7.jpg"},{"id":13551084,"identity":"478c9896-e16a-4526-b8a4-c2866fec3f3d","added_by":"auto","created_at":"2021-09-17 02:27:17","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2475802,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/a4aae288-2f56-47b0-9a38-07fdf3ce7a6d.pdf"},{"id":1555432,"identity":"5e134a44-c34e-4b63-bf99-d629f9faecd3","added_by":"auto","created_at":"2020-07-14 17:49:04","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":12662,"visible":true,"origin":"","legend":"","description":"","filename":"AdditionalFigureLegends.docx","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/AdditionalFigureLegends.docx"},{"id":1555433,"identity":"b3fd1bd0-0859-4509-a234-c3359a1b3bf6","added_by":"auto","created_at":"2020-07-14 17:49:04","extension":"tif","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":9020008,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalfig3.tif","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Additionalfig3.tif"},{"id":1555434,"identity":"51ed08ee-03ae-430c-8aa5-c168ba23c591","added_by":"auto","created_at":"2020-07-14 17:49:05","extension":"tif","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":5821788,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalfig2.tif","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Additionalfig2.tif"},{"id":1555435,"identity":"02855e8e-0431-49e2-8b59-342977a35ea6","added_by":"auto","created_at":"2020-07-14 17:49:05","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":5159932,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalfig1.tif","url":"https://assets-eu.researchsquare.com/files/rs-41157/v1/Additionalfig1.tif"}],"financialInterests":"","formattedTitle":"\u003cp\u003eDevelopment of A Three-Gene Signature Prediction Model for Lymph Node Metastasis in Papillary Thyroid Cancer\u003c/p\u003e","fulltext":[{"header":"Background","content":" \u003cp\u003eThyroid cancer, one of the most prevalent endocrine cancers with a rising incidence rate over the past years, is the fifth leading incidence of cancer in female [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Papillary thyroid cancer (PTC) is the dominant historical type which contributes to approximately 84% of all thyroid cancer [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. However, through the low mortality and moderate prognosis were frequently mentioned, the recurrence and the complications are still perplexing those PTC patients [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Besides, lymphatic invasion, cervical lymph node metastasis, larger size of tumor, increasing diagnosis age and extraordinary enlargement of thyroid tissue increase the progression risk of PTC [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Due to early lymph node metastasis happening in PTC frequently [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e], early detection and diagnosis are of great value. Currently multiple methods like thyroid and neck ultrasound, CT/MRI and fine-needle aspiration (FNA) for suspicious lateral neck nodes could effectively diagnose thyroid cancer [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Yet some of the lymph node metastasis may be troublesome for detecting. We therefore urgently called for a reliable and straightforward approach to determine the possibility of lymph node metastasis.\u003c/p\u003e \u003cp\u003eRecent years, high throughput analysis afford us an advanced and efficient technique of evaluating the molecular disruptions in tumor tissues. For instance, A study of predicting chemotherapy sensitivity in cervical cancer, in which expression levels of 22 total and phosphorylated protein were analyzed in 181 frozen tissue samples, resulted in a model that was capable of predicting patients' chemotherapy sensitivity and assessing clinical outcome [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Thus, as more and more cancer sequencing databases are established, there is a great potential for us to acquire and analyze these data and guide clinical decisions with the findings.\u003c/p\u003e \u003cp\u003eIn order to discover the predictive value of lymph node metastasis in the papillary thyroid cancer transcriptome, we searched for several RNA-seq as well as gene microarray datasets containing both papillary thyroid cancer samples and normal thyroid samples to identify differentially expressed genes. Genes with powerful correlation to lymph nodes metastasis were selected as well. Afterwards, we designed a predicable risk score system and effectively confirmed the model in two independent papillary thyroid cancer cohorts. The entire flow of our efforts to identify predictive models was presented in Additional file 1: Fig. S1. Eventually, we recommended a successful predicable model of lymph node metastasis in papillary thyroid cancer patients with moderate accuracy.\u003c/p\u003e "},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eData collection, normalization and preprocessing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGene microarray datasets and the associated clinical data of PTC and normal thyroid samples were downloaded from UCSC XENA (https://xenabrowser.net/) and Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). Totally 8 GEO datasets and 1 The Cancer Genome Atlas (TCGA) dataset were involved in this research. Gene labels in other forms were all normalized into official gene symbols. Furthermore, those samples were proceeded to a quality control step called SimpleAffy to lessen bias. SimpleAffy was used to identify the 3\u0026rsquo;-to-5\u0026rsquo; ratios of GAPDH and \u0026beta;-actin. Outliers were excluded from the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDifferential analysis of gene expression datasets\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEdgeR R package afforded us the differential expression analysis between papillary thyroid cancer tissues and normal thyroid tissues. The filter thresholds of \u003cem\u003eP\u003c/em\u003e value were set as less than 0.05 in integrated analysis. |log2FC (fold change) | was set as more than 0.1 or more than 0.5 in GEO or TCGA microarray separately due to the different sample sizes. A robust rank aggregation (RRA) algorithm was used to determine the overall differentially expressed genes （DEGs） in the 8 GEO datasets in which they were all up-regulated or down-regulated. The most relevant genes could be revealed in the analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunctional annotation analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGene Ontology (GO) analysis was performed on the DEGs in order to identify their potential functions. The process was settled by the clusterProfiler 3.11. The result of three major GO terms called biological process (BP), cellular component (CC) and molecular function (MF) could be visualized in dot plot or pie chart form by using the ggplot2 R package.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eWeighted gene co-expression network analysis (WGCNA)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll samples of TCGA underwent a sample clustering test to identify their relationships. Those outliers were eliminated and then the others went through soft-threshold defining procedure. A Scale-free gene co-expression network was attained when the correlation coefficient was 0.85. At this point the soft-thresholding power was equal to 12. Afterward, WGCNA algorithm of R software was performed to construct a scale-free network of all DEGs of TCGA. After the construction, cluster analysis was operated to arrange genes with similar expression patterns into gene modules. The minimum size of each module was set as 30. Besides, the cut height threshold for merging modules was defined as 0.25 and some of the modules could be integrated to analyze. Lately, we constructed the relationships between modules and phenotype. The module\u0026ndash;trait association was defined as the correlation of module eigengenes (MEs) and traits. Accordingly, each correlation value and the significance value were calculated separately to reveal the relevant modules, which are intently related to the traits. For each gene modules, gene significance (GS) is described as the correlation level between clinical trait and expression pattern. While module membership (MM) stands for the correlation level between MEs and expression pattern. In our study, genes with GS score more than 0.8 and MM score more than 0.2 were defined as module genes with high correlation to certain phenotype.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProtein\u0026ndash;protein interaction (PPI) network and identification of hub-genes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe DEGs of both GEO and TCGA datasets were uploaded to the STRING tool (http://www.string-db.org/) to look into their correlation and to discover the hub-genes in the gene network. A darker line represents a stronger edge confidence in the network. Then, Cytoscape software was utilized to visualize the outcome of STRING database and to acquire the hub-genes which have a closest relation and have a more considerable function among the DEGs. In our study, the 100 genes with the highest connectivity degree were identified as hub-genes of the PPI network.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis of model acquisition and validation \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe randomly separate the TCGA microarray data and relevant clinical data into 2 parts, named a training cohort and a validation cohort. A univariate Logistic regression was applied to DEGs in TCGA training cohort, aid of finding the relationship between those genes and patient\u0026rsquo;s prognostic data. The results were then diminished by using the LASSO algorithm with the R package \u0026lsquo;glmnet\u0026rsquo;. The minimal partial likelihood deviance was carried out as optimal tuning parameter (\u0026lambda;) changed. Also some of the coefficients of gene would reduce to zero. Those genes were excluded and the others were accessed to the multivariate Logistic test. The hazard ratio (HR) and 95% confidence intervals (CI) of each gene would be calculated and only those genes which did not include 1 in the 95% CI were selected as final trait-relative genes. The coefficient of each gene in multivariate Logistic regression model was summed to calculate the risk score of lymph node metastasis. The function could be computed as follows,\n\u003cp\u003e\u003cimg src=\"data:image/jpeg;base64,/9j/4AAQSkZJRgABAQEAYABgAAD/4RD6RXhpZgAATU0AKgAAAAgABAE7AAIAAAAQAAAISodpAAQAAAABAAAIWpydAAEAAAAgAAAQ0uocAAcAAAgMAAAAPgAAAAAc6gAAAAgAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAFNhY2hpbiBNYWhhcm51cgAABZADAAIAAAAUAAAQqJAEAAIAAAAUAAAQvJKRAAIAAAADMDYAAJKSAAIAAAADMDYAAOocAAcAAAgMAAAInAAAAAAc6gAAAAgAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAADIwMjA6MDc6MTQgMTk6Mzg6MDIAMjAyMDowNzoxNCAxOTozODowMgAAAFMAYQBjAGgAaQBuACAATQBhAGgAYQByAG4AdQByAAAA/+ELImh0dHA6Ly9ucy5hZG9iZS5jb20veGFwLzEuMC8APD94cGFja2V0IGJlZ2luPSfvu78nIGlkPSdXNU0wTXBDZWhpSHpyZVN6TlRjemtjOWQnPz4NCjx4OnhtcG1ldGEgeG1sbnM6eD0iYWRvYmU6bnM6bWV0YS8iPjxyZGY6UkRGIHhtbG5zOnJkZj0iaHR0cDovL3d3dy53My5vcmcvMTk5OS8wMi8yMi1yZGYtc3ludGF4LW5zIyI+PHJkZjpEZXNjcmlwdGlvbiByZGY6YWJvdXQ9InV1aWQ6ZmFmNWJkZDUtYmEzZC0xMWRhLWFkMzEtZDMzZDc1MTgyZjFiIiB4bWxuczpkYz0iaHR0cDovL3B1cmwub3JnL2RjL2VsZW1lbnRzLzEuMS8iLz48cmRmOkRlc2NyaXB0aW9uIHJkZjphYm91dD0idXVpZDpmYWY1YmRkNS1iYTNkLTExZGEtYWQzMS1kMzNkNzUxODJmMWIiIHhtbG5zOnhtcD0iaHR0cDovL25zLmFkb2JlLmNvbS94YXAvMS4wLyI+PHhtcDpDcmVhdGVEYXRlPjIwMjAtMDctMTRUMTk6Mzg6MDIuMDYxPC94bXA6Q3JlYXRlRGF0ZT48L3JkZjpEZXNjcmlwdGlvbj48cmRmOkRlc2NyaXB0aW9uIHJkZjphYm91dD0idXVpZDpmYWY1YmRkNS1iYTNkLTExZGEtYWQzMS1kMzNkNzUxODJmMWIiIHhtbG5zOmRjPSJodHRwOi8vcHVybC5vcmcvZGMvZWxlbWVudHMvMS4xLyI+PGRjOmNyZWF0b3I+PHJkZjpTZXEgeG1sbnM6cmRmPSJodHRwOi8vd3d3LnczLm9yZy8xOTk5LzAyLzIyLXJkZi1zeW50YXgtbnMjIj48cmRmOmxpPlNhY2hpbiBNYWhhcm51cjwvcmRmOmxpPjwvcmRmOlNlcT4NCgkJCTwvZGM6Y3JlYXRvcj48L3JkZjpEZXNjcmlwdGlvbj48L3JkZjpSREY+PC94OnhtcG1ldGE+DQogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgIAogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgIAogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgIAogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgIAogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgIAogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgIAogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKICAgICAgICAgICAgICAgICAgICAgICAgICAgIDw/eHBhY2tldCBlbmQ9J3cnPz7/2wBDAAcFBQYFBAcGBQYIBwcIChELCgkJChUPEAwRGBUaGRgVGBcbHichGx0lHRcYIi4iJSgpKywrGiAvMy8qMicqKyr/2wBDAQcICAoJChQLCxQqHBgcKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKir/wAARCABAARADASIAAhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAAAAECAwQFBgcICQoL/8QAtRAAAgEDAwIEAwUFBAQAAAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEII0KxwRVS0fAkM2JyggkKFhcYGRolJicoKSo0NTY3ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlqc3R1dnd4eXqDhIWGh4iJipKTlJWWl5iZmqKjpKWmp6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW19jZ2uHi4+Tl5ufo6erx8vP09fb3+Pn6/8QAHwEAAwEBAQEBAQEBAQAAAAAAAAECAwQFBgcICQoL/8QAtREAAgECBAQDBAcFBAQAAQJ3AAECAxEEBSExBhJBUQdhcRMiMoEIFEKRobHBCSMzUvAVYnLRChYkNOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElKU1RVVldYWVpjZGVmZ2hpanN0dXZ3eHl6goOEhYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3uLm6wsPExcbHyMnK0tPU1dbX2Nna4uPk5ebn6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwD6RooooAKKKKACiiigAooooAKKKKACiiigAooqhq2vaPoEMc2u6rY6ZFI21HvLlIVc9cAsRk0AR6t4k0nQ5reHVb1LeS6LLChUkyEKWwMA8kKcDqcYGTVuwv7bVNOt7+wlE1rcxrLFIAQGUjIPNcjqXizwFql/Y3U3jjREaxMrQhNVg+V3Qpv5bqFZgP8AeNS+HfFngjSNH0/RrTxto14beNLeIvqduXfHCgBSB6AACgDsaKKKACiiigAooooAKKKKACiiigAooooAKKKKACiiigAooooAKKK4z4t3E9n8L9VurS6urSaHymSa0mkikX96gOChDcgkY70AdnRXnXgy5e48Za1/wjd5qlxo9rZxxvaa1c3HmJeklhgXGZkUpjJIwc5UHBqno/xBfSvh34RnjgadtWtmK3Wu6qURCuMLLc+UcyNu+XKDO09MUAeo0V5ZqPiW28P/ABBOqXdnaWutan4dthFpz3SK1zcmZgsIkUEMQSF3gEYweBXpdjJdy2ML6lbxW10y5lhhmMqIfQOVUt9dooAsUVwHjk3tv4hjutUGtnwyLHa0uiXMkclncBmLSyLEQ7ps24xuAKnK85qxB4o1GCGx0Tw5DD4q1C30iC8uL2a8Fqk6MCqOCFfLSFGYDoB1YZGQDt6K4nUPHmowtq76b4fS7g0GJW1MvfiN0fyhK0cShGEjKjDOWQEnANMvPH2qS65Npnhrw2usOdIg1a1Y6gsAmjkZlCtuT5G+Q45IPGdtAHc0V53ZfF7TL7xLBp8P9ntb3Govp0e3U0N55isyh2ttuRGWXAbdnkHbg5rq/FGu/wDCPaQL3ztJhzKqbtW1H7FDyD/y02P83HAxzzzxQBs1z/jfwdp3jvwndaJqgwso3QzKPmglH3XX3B/MZHesLSfiP/aWr2tl9u8Ev58oTbZ+KvPmOT/BH9nXe3oMjPrXe0AecfCvxhqN0bzwZ4xOzxPoWI5GJ/4/YOiTqT1yCMn3B74HVeMPC1t4v0EadcssTJcQ3EM5j3mJ45FfIGR1AK9ejGuP+MXh+SCwt/Hmg3EVj4g8O/vI5ZOFuoScNbt67t2FHqxA+9keh6ZdT3ukWd1d2r2c88CSS20hy0LMoJQkdwTj8KALVFcf8S7h08N2NmrukepaxY2M5jbaTFJcIsi59CuVPsa4x5YPD+gfEObTLCbQhb2cTrolpcC0kt0VXzcxuiuiNIM4KBhmIZOScAHsdFQ2b77C3f5juiU/M2T07nuabqENzcadPFYXX2S5eMiKfyw/lt2O08EZ7UAWKK4LQfGep+KvELaJbLb2Fzo8gOsyo6yrKQxAjg7lWKnczDKfdxuyV72gAorhPizNcQeHtJa0k1BXfW7KFo9OvHtpZ0eUK0YZXT7wJHLAAkHIxkRaf4Xk1jw1rka3Hirw5NdHyYVvdammktWj3FZkfzGwGL/MA7KQg96APQKK8y8HXs3i+wvdXvr67t5tOsP7NNlbahOqF/LLNckbhu3hgY35O3DZJxtwvh/carqMPgm80W68Szyywh9em1OS7e0kjMRyVNwdpYvt2mLjrnigD2qivO9N+Jt/qN5ocg0Sxh0rWo5p4bl9VJmiiiG5y8QhOGx/CGIB4LA07wt8WbLxRrFlY20enZ1K3kns0g1RJ5k2ANtuIlXMJIJPBfGCDg8UAehUVwWgfEa+1eHw7d3nh9LOx124ltI5UvvNeKZFkYZXYMoRE2GyDnqveoNJ+L2mav4is7CD+z2t769ksoDHqaPdhk34eS2C5SNthwdxPzLkDNAHolFcJZ3Euq2finxFMrTTWc93ZafGZvLECQAxsUYBtjs6yEvtJwVHQVQub8af8LvDnjazadLu3stPed7iYyy3NvJ5avHM4A8xsSFgxUfOMgDJFAHpVFFFABWP4q8OQeLPDtxo15dXNrBcFC8lqUD/ACsGAG9WHUDtWxRQBz8/hC3k8SQ69b6jfWmopafZJ5YDEBdp1UyqUKllOSCAMZI6cVmWvw3hsfDVhoVj4j1u3s7K3e1ADW7edExztdWhKkjoGChgM88muzooA46T4ZaPKWjku75rP+yE0dLJmjaJIU+4wJQtvU/MGLHmuk0jT30rSYLGTULvUTAuwXN4UMrgdNxVVBPvjJ75PNXaKAOc1bwcmpaxPqVtrWqaXNdW6W10tk8WyeNSxUESRttPzt8y7Tz1qM+BLGC4s5dGv7/R2tbGPT/9DdP31vHnYj+YjdMnDDDDJ5rp6KAOU1L4f2OoXd/JFqeqWMGqRrHqNrazJ5d4AgTLFkZ1JQBSyMpIHJq7beErWz8YN4gt7y6SQ2EenizHl+QsSMWUAbN2QWY/e7+mMb1FAHPaV4Ph0bUJJbDVdSjsXnkuRpnmp9nSR2LMRhN+CzE7d+3JPFdDRRQAUUVxvxPk8Vt4VNl4I0qW/u7x/Knkiu4rdoIf4yrSHhyPlBAOM57DIBBBnx94tF2fm8N6FcEW4/hv71Dgye8cR4XsXyf4RXc15bpviT4h6Ppdtp2m/BtLe0tY1ihiTxJbYVQMAfdqHWfEHxX13TTptr8PH0M3EkayahH4gt5Hgj3guVVdpztyOD370Adb8QtCvvEGk6Vbaajs0Os2VzK0cio0cSTBncEnqACeMn0Bpl18O7K/0rW7XUNV1O6udbgW2u9QkMIm8lc4jULGI1HzMOEz8xOc4I66igCGzt/sdjb2xmkn8mJY/Nl27nwMbjtAGTjPAA9AKbqFn/aGnT2hnntxPGUMtu+yRAeMq3Y+9WKKAMX/AIRLSI10r7Fb/YW0k/6I9qdjKn8UZP8AEjfxA5yefvAEbVFFAGJ4p8LxeKrG0t59QvLD7JeRXsUln5e7zI23JnzEcYDYOMc49OKybj4efbINVS98Va9PLqsUMFxOxtlYRRGQiNQsAUK3mtu+XJ9euexooA5m48EW8viefXLbVdQsp7iw+wTQ2/kiKRAGCsVMZJdSxIOeMYxtyDe8P6CPCvhO20XTLia8Sxg8q2a8ZdxAHyqxRQMDgZxnHrWxRQB5F4M8Ha9o+qWDjTLq0kl+XWp76LTWiukKtuCSQr9oclyCDIen3smu80Dwinh2BLa01nVZrO3i8m0tZ5Y2S1TGAFwgZsDgeYXxiuhooA46y+G9pYaXoFhBrereToN495b5MGZHbcMP+65GJJBxg/OecgEaGjeD4dCvC1jqupixEsk0WmtKn2eJnJLYwgcrliQpYqM8DgV0NFAHH6fos9jPr+gTrcxWGr3E91a3tsoby/OGZUOQwVg+9gWG0hgOSCKrap4YdvCmk+A9N+1XFlD9mS4vbpFxHawuGC7gqqzsIwgAGQDuPYnuaKACiiigAooooAKKKKACiiigAooooAKKKKACiiigArF8Y3d1Y+DNVudPbbdR2zeSfLMmHxgZUckZPOO1bVFAHHa1Y6lcPZNosGqtai0QL9n1AWgHXAMbru3YxnP07VU17UItL8QeCVu7m+iuVuMXMRkkkURm0nXMm0bGPmmMbiM55GBmu8ooAKKKKACiiigAooooAKKKKACiiigAooooAKKKKACiiigAooooA//Z\"\u003e\u003c/p\u003e\nin which \u0026beta;i stands for the Logistic regression coefficient of gene i in the training cohort.\u003c/p\u003e\n\u003cp\u003eBy calculating risk score of each sample, we could sort them and divide the cohort into low risk group and high risk group by the median value. A student\u0026rsquo;s t test of two groups in comparing the trait demonstrated that the model could predict the risk effectively. Meanwhile, the receiver operating characteristic (ROC) curves was used to verify the sensitivity and specificity of lymph node metastasis related model risk prediction. After the model construction, the corresponding approach above was applied to TCGA validation cohort as well as GEO independent cohort in turn for risk predicting system confirmation.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eIdentification of DEGs and associated GO analysis in GEO datasets\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFirstly, all the human tissue microarrays including papillary thyroid carcinoma samples and paired/unpaired normal thyroid samples were extracted from the GEO database. In order to make sure the quality of the research, SimpleAffy R package was utilized to determine the 3\u0026rsquo;-to-5\u0026rsquo; ratios of \u0026beta;-actin and GAPDH (Additional file 2: Fig. S2). After excluding seven tumor samples and 2 normal samples in three datasets, a total of 187 tumor samples and 119 normal control samples in eight datasets (GSE33630 [10, 11], GSE60542 [12], GSE66783 [13], GSE5364 [14], GSE129562 [15], GSE97001 [16], GSE3467 [17] and GSE27155 [18, 19]) were included this research (Table 1). After performing differential analysis on the 8 datasets by the edgeR and robust rank aggregation algorithm, the DEGs between the normal tissues and PTC tissues of each dataset were obtained, with an adjusted \u003cem\u003eP\u003c/em\u003e value \u0026lt; 0.05 and |log2FC (fold change) | \u0026gt; 0.1 as the cut-offs. A total of 531 over-expressed genes and 474 suppressed genes were discovered in PTC tissues. The top ten genes that were overexpressed or suppressed are listed by a heatmap in Fig. 1a. GO analysis was performed on the DEGs in order to identify potential functions. Those DEGs were enriched in GO terms including 500 BPs, 70 CCs and 28 MFs with the cut-off value set as adjust \u003cem\u003eP\u003c/em\u003e value \u0026lt; 0.01 and q value \u0026lt; 0.01. TOP 10 GO terms of each category were displayed in Fig. 1b. Among them, extracellular structure organization and extracellular matrix organization, cell-substrate adhesion and renal system development were three significant aspects in BP category. The first two functions are also remarkable in CC term. The most enriched MF terms were sulfur compound binding, glycosaminoglycan binding and serine-type peptidase activity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIdentification of DEGs and associated GO analysis in TCGA dataset\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe TCGA-THCA dataset consisting of 56 normal thyroid tissues and 497 thyroid cancer tissues were selected by integrated analysis (\u003cem\u003eP \u003c/em\u003evalue \u0026lt; 0.05 and |log2FC | \u0026gt; 0.5 set as cut-offs), from which 6492 mRNAs with steady differentially expressed patterns were identified by edgeR analysis. Similarly, these DEGs were analyzed by GO functional enrichment analysis to find out the potential biological functions. The BP term were shown in a pie chart (Fig. 2a). Activation of MAPK activity and leukocyte mediated cytoxicity are the top 2 enriched terms of BPs. While CCs and MFs were exhibited in Fig. 2b. Each category only shows the top ten functions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDiscovery of a strong correlation between lymph node metastasis and DEGs in TCGA dataset by WGCNA algorithm\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor DEGs of thyroid cancer tissues and normal thyroid tissues in the TCGA dataset, we used WGCNA algorithm to find out which clinical traits they are significantly related to. Firstly, after screening the clinical characteristics and sample tree, a total of 482 thyroid cancer samples were included in the study. In addition, the sample tree and the clinical characteristics of each sample are also shown in Additional file 3: Fig. S3a. Then, through the soft threshold screening (Additional file 3: Fig. S3b), a gene co-expression network with the scale-free characteristics was established, in which the evaluation parameter R\u003csup\u003e2\u003c/sup\u003e was set as 0.85. After the network was established, genes with similar expression level were grouped in the same module and displayed in the form of a hierarchical clustering diagram (Additional file 3: Fig. S3c). Different colors indicated separate modules. It should be noted that the genes in gray color were not classified into any modules. After that, the clinical characteristics of the samples were taken out and analyzed for correlation with each module (Fig. 3). We can find that the brown module (r =-0.34, p=3e-14) and the turquoise module (r= 0.35, p=6e-15) were highly correlated with lymph node metastasis. By calculating the MM value of these two modules (cut-off set as MM\u003cem\u003e P\u003c/em\u003e value \u0026lt; 0.05 and Gene MM value \u0026gt; 0.8), a total of 106 genes that were closely related to the brown module and the turquoise module were identified for further analysis. Since the results of WGCNA strongly implied the relevance of these DEGs to lymph node metastasis, we subsequently concentrated on lymph node metastasis and attempted to explore models that could accurately predict lymph node metastasis in PTC.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePotential hub-genes relating to lymph node metastasis revealed by PPI network\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe shared DEGs in both GEO datasets and TCGA dataset were considered as important genes for papillary thyroid cancer. To narrow the scope and identify the much meaningful genes among them, PPI network analysis was used to find key regulatory genes in these genes. STRING database (\u003ca href=\"https://string-db.org)\"\u003ehttps://string-db.org)\u003c/a\u003e was used to identify the PPI network of these 599 shared DEGs. Fig. 5a showed the overall PPI regulatory network of these differential genes. through the PPI network connection score, one cluster with the highest score was displayed in Fig. 5b. After all genes going through PPI analysis, some potential hub-genes can be found based on the score. The first 100 hub-genes were extracted as key genes. What\u0026rsquo;s more, those hub-genes were intersected with the WGCNA module genes mentioned above (Fig. 5c). A total of 9 genes were identified as important DEGs and related to lymph node metastasis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003econstruction a risk scoring system by logistic regression analysis and LASSO algorithm\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe randomly divided the thyroid cancer samples in the TCGA database into two parts. The first cohort named training cohort was utilized to find the risk scoring system of lymph nodes metastasis. Meanwhile, the other one named validation cohort was used to confirm the system. Also, an independent GEO cohort with clinical traits was enrolled to verify the model. Their clinical features were listed in Table 2. Univariate logistic analysis was performed on the 9 genes to find out if they related to lymph nodes metastasis (Fig. 4d). A predictive gene was defined as a hazard ratio (HR) and 95% confidence interval (CI) greater than or less than 1 and \u003cem\u003eP\u003c/em\u003e value less than 0.05. It is gratifying that these nine genes were all found to be associated with lymph node metastasis after the analysis. Among them, EPHB3, also called EPH Receptor B3; MET, one of the receptor tyrosine kinase; ICAM1, also named Intercellular Adhesion Molecule 1; SERPINA1, also called Serpin Family A Member 1 and FN1, also named Fibronectin 1 are the single risk factors for lymph nodes metastasis, while ITPR1, also named Inositol 1,4,5-Trisphosphate Receptor Type 1; PPARGC1A,also called PPARG Coactivator 1 Alpha; GNA14, also named G Protein Subunit Alpha 14 and BCL2,a apoptosis regulator were found as protective factors in this trait. In order to compress the model and identity the key genes, A least absolute shrinkage and selection operator (LASSO) regression model was then used to test these 9 Genes. In the LASSO model, when the value of \u0026lambda; increases, more coefficients (genes) will be set to zero, meaning those variables could be remove from the model due to their shrinking property (Fig. 5a). Six genes model satisfied the minimum partial likelihood deviance due to the ridge regression. The minimum log(\u0026lambda;) was -3.67 at this status. Finally, these six genes were subjected to multivariate logistic analysis in order to find the final risk prediction model. The result indicated that MET, ITPR1, and BCL2 were independent prognostic factors for lymph nodes metastasis (Fig. 5c, Table 3). The score of risk estimation model could be calculated as: expression of ITPR11 \u0026times; 0.589 + expression of MET \u0026times; 0.841 - expression of BCL2 \u0026times; 0.786. The factors stand for the respective multivariate Logistic regression coefficients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eVerification of the prediction system through validation cohorts from TCGA and GEO datasets\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTCGA training cohort used for model construction were separate into high-risk group and low-risk group based on each sample's risk score (Fig. 6a). The median risk score (3.879) was set as the cut-off. For two groups, a significant difference of lymph nodes metastasis was clearly shown in heatmap (Fig. 6b). A student\u0026rsquo;s t test also demonstrated that the high-risk group had a higher frequency of lymph nodes metastasis than the low-risk group (Fig. 6c) .The receiver operating characteristic (ROC) curve shows the efficiency of the prediction model, in which the area under the curve (AUC) was 0.744 (Fig. 6d). In order to verify the prediction accuracy of the risk prediction model, we used multiple cohorts to verify. TCGA validation cohort has been operated by the same protocol (Fig. 6e). Obviously, the result also showed that a high risk score coul\u003cu\u003ed\u003c/u\u003e be riskier to lymph nodes metastasis by a heatmap (Fig. 6f) and student\u0026rsquo;s t test (Fig. 6g). The AUC over validation model is 0.711 (Fig. 6h). At the same time, we conducted an independent verification in another cohort combined from two GEO dataset (GSE3467 and GSE60542) with the same platform (GPL570) and corresponding clinical data, in order to show that the risk scoring system is generally applicable (Fig. 7a). In the heatmap, a strong tendency was discovered of which higher risk score indicated higher frequency of lymph nodes metastasis (Fig. 7b). A student\u0026rsquo;s t test verifies the point (Fig. 7d). Furthermore, gene expressions of each sample tissue were displayed in Fig. 7c as a heatmap. Finally, we performed a ROC analysis and a meaningful result was revealed with 0.6842 of AUC value. Accordingly, the three-genes model was verified in multiple microarray and all showed a great difference. It is a reliable, accurate and independent predictive appliance for determining lymph nodes metastasis in papillary thyroid cancer patients.\u003c/p\u003e"},{"header":"Discussion","content":" \u003cp\u003eGenomic analysis has an extraordinarily critical role in tumor research. For example, Paik S et al. derived a 21-genes based recurrence scoring system from a prospective analysis of multiple gene expression levels in a breast cancer population [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Besides, transcriptome mapping could also be applied for the determination of molecular subtypes of tumors. It has already been implemented in colorectal cancer [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], breast cancer [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], prostate cancer [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e], and pancreatic cancer [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e], which has a facilitating effect on clinical decision-making. Moreover, genomic studies provide an insight into the tumor immune microenvironment. Xu M et al. calculated the corresponding immune infiltration scores from the expression of immune-related molecules in breast cancer specimens and the immune score was found to perform a detrimental effect in overall survival and recurrence-free survival [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Chakladar J et al. also analyzed immune-related genes in PTC through the combined application of genomics and transcriptomics [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. In the present research, we investigated multiple independent datasets of PTC, and successfully established a three-gene (MET, ITPR1 and BCL2) prediction model for lymph node metastasis in PTC patients.\u003c/p\u003e \u003cp\u003eIn our study, MET and ITPR1 expression in the predictive model was a risk factor for lymph node metastasis, whereas BCL2 was a protective factor. Previous study has reported that BCL2 was found to be highly expressed only in poorly differentiated tumors [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Moreover, another study also showed that a lower expression level of BCL2 could act as an early sign of oncogenesis and be a reason for the favorable prognosis [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], suggesting that BCL2 may act as a protective factor in thyroid cancer. As for ITPR1, one study demonstrated that up-expression of ITPR1 shelters renal cancer cells against natural killer cells [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Another study showed that ITPR1 could enhance paclitaxel toxicity in breast cancer [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. We discovered ITPR1 is a risk factor for lymph node metastasis in thyroid cancer, but the biological mechanism still under solving. As a heterodimeric transmembrane receptor tyrosine kinase, MET mediates the activation of multiple signaling pathways, including PI3K/AKT, Ras-Rac/Rho and phospholipase C-γ pathways [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Significantly higher level of MET was detected in PTC [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], non-small cell lung cancer [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e], bladder cancer [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e] and oral cancer [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. Those results all add up to a better proof of MET as a cancer-promoting factor and can be corroborated with our findings. In conclusion, based on the available studies, BCL2 and MET match the results more accurately, while ITPR1 in thyroid cancer has been less studied and needs to be further explored.\u003c/p\u003e \u003cp\u003eThe Robust rank aggregation algorithm was utilized to analyze multiple gene sets integrally and identify the common DEGs. This algorithm compensates for the limitations of the previous single data set analysis and minimizes bias. Since PTC samples are generally small, the RRA algorithm could be quite helpful. WGCNA is an effective method for describing associations of gene expression patterns with clinical phenotypes. In thyroid cancer research, WGCNA was widely used [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Based on the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) statement, studies developing new prediction models should always go through an internal validation (also called self-validation) to quantify the predictive appearance. Also, it is firmly recommended to appraise the model in other data (also called external validation) after developing a prediction model [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. We then applied our prediction model to TCGA self-validation cohort and one independent GEO cohort. The outcomes of two dataset strongly fit in the model.\u003c/p\u003e \u003cp\u003eThe innovative feature of our study is that the combined application of RRA algorithm, WGCNA analysis and PPI methods was first used to analyze the correlation between PTC and lymph node metastasis. The degradation of samples may cause bias of the results. Therefore, we performed quality control on each sample in the dataset. Tissues that did not meet the requirements (degradation occurred) were excluded from the follow-up experiment. Furthermore, the predictive model of PTC lymph node metastasis was uniquely established, which might be useful for therapeutic decision-making and clinical monitoring. However, the inadequacy of this study is that the prognostic data in the TCGA database for papillary thyroid cancer are quite good and we failed to find significant differences in prognosis, while the GEO dataset was unable to find prognosis information. Therefore, it is regrettable that prognosis cannot be measured.\u003c/p\u003e \u003cp\u003eFor PTC, lymph node metastasis may arise at an early stage and there is a potential risk for skip metastasis [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e], the mechanism of which is currently not clear. Routine diagnosis methods such as neck CT/MRI and neck lymph node ultrasound [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] may not be able to fully detect the development of lymph node metastasis. Patients who develop lymph node metastases are likely to require postoperative I\u003csub\u003e131\u003c/sub\u003e radiation therapy and may have a higher recurrence rate and lower survival rate [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Therefore, the outcome of this study may have a better role in predicting lymph node metastasis in PTC patients. Patients in the low-risk group have a lower likelihood of lymph node metastasis. With current conventional methods of measuring expression level, such as RT-qPCR or immunohistochemistry, we can easily, accurately and economically obtain risk scores for this patient, thus allowing the model to be better applied in clinical practice.\u003c/p\u003e "},{"header":"Conclusions","content":" \u003cp\u003eSummarily, this study is a highly scientific and accurate method with successful predictive significance for determining lymph node metastasis in PTC patient. Among the model, 2 genes were identified as risk factors and 1 genes was protective factor in PTC patients. It might be potential treatment targets and urged for further research.\u003c/p\u003e "},{"header":"Abbreviations","content":"\u003cp\u003ePTC, Papillary Thyroid Cancer; CT, Computed Tomography; MRI, Magnetic Resonance Imaging; GEO, Gene Expression Omnibus; TCGA, The Cancer Genome Atlas; DEGs, Differentially Expressed Genes; RRA, Robust Rank Aggregation; GO, Gene Ontology; WGCNA, Weighted Gene Co-expression Network Analysis; MEs, Module Eigengenes; GS,Gene Significance; MM, Module Membership; PPI, Protein\u0026ndash;Protein Interaction; HR, Hazard Ratio; CI, Confidence Intervals; ROC, Receiver Operating Characteristic; FC, Fold Change; BP, Biological Process; CC, cellular component; MF, molecular functions.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePY, ZH, and YL conceived and arranged the experiments, and wrote the manuscript. KL, QZ, WL, MX, JZ, YJ, WY and LY collected and analyzed the data. All authors read and approved the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGene microarray datasets and the associated clinical data of PTC and normal thyroid samples in our study were all publicly available. The data of PTC patients from TCGA were downloaded from UCSC XENA (https://xenabrowser.net/). Additionally, Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) afford us 8 cohorts (GSE33630, GSE60542, GSE66783, GSE5364, GSE129562, GSE97001, GSE3467 and GSE27155) for further research in the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by Ethics Committee of Huazhong University of science and technology (HUST).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors have agreed for this publication.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSiegel RL, Miller KD, Jemal A: \u003cstrong\u003eCancer statistics, 2020\u003c/strong\u003e. \u003cem\u003eCA Cancer J Clin \u003c/em\u003e2020, \u003cstrong\u003e70\u003c/strong\u003e(1):7-30.\u003c/li\u003e\n\u003cli\u003eGeraldo MV, Kimura ET: \u003cstrong\u003eIntegrated Analysis of Thyroid Cancer Public Datasets Reveals Role of Post-Transcriptional Regulation on Tumor Progression by Targeting of Immune System Mediators\u003c/strong\u003e. \u003cem\u003ePLoS One \u003c/em\u003e2015, \u003cstrong\u003e10\u003c/strong\u003e(11):e0141726.\u003c/li\u003e\n\u003cli\u003eAschebrook-Kilfoy B, Kaplan EL, Chiu BCH, Angelos P, Grogan RH: \u003cstrong\u003eThe Acceleration in Papillary Thyroid Cancer Incidence Rates is Similar Among Racial and Ethnic Groups in the United States\u003c/strong\u003e. \u003cem\u003eAnn Surg Oncol \u003c/em\u003e2013, \u003cstrong\u003e20\u003c/strong\u003e(8):2746-2753.\u003c/li\u003e\n\u003cli\u003eChrisoulidou A, Boudina M, Tzemailas A, Doumala E, Iliadou PK, Patakiouta F, Pazaitou-Panayiotou K: \u003cstrong\u003eHistological subtype is the most important determinant of survival in metastatic papillary thyroid cancer\u003c/strong\u003e. \u003cem\u003eThyroid Res \u003c/em\u003e2011, \u003cstrong\u003e4\u003c/strong\u003e(1):12.\u003c/li\u003e\n\u003cli\u003eCheng Q, Li X, Acharya CR, Hyslop T, Sosa JA: \u003cstrong\u003eA novel integrative risk index of papillary thyroid cancer progression combining genomic alterations and clinical factors\u003c/strong\u003e. \u003cem\u003eOncotarget \u003c/em\u003e2017, \u003cstrong\u003e8\u003c/strong\u003e(10):16690-16703.\u003c/li\u003e\n\u003cli\u003eMoo TA, McGill J, Allendorf J, Lee J, Fahey T, 3rd, Zarnegar R: \u003cstrong\u003eImpact of prophylactic central neck lymph node dissection on early recurrence in papillary thyroid carcinoma\u003c/strong\u003e. \u003cem\u003eWorld J Surg \u003c/em\u003e2010, \u003cstrong\u003e34\u003c/strong\u003e(6):1187-1191.\u003c/li\u003e\n\u003cli\u003eGrani G, Ramundo V, Falcone R, Lamartina L, Montesano T, Biffoni M, Giacomelli L, Sponziello M, Verrienti A, Schlumberger M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThyroid Cancer Patients With No Evidence of Disease: The Need for Repeat Neck Ultrasound\u003c/strong\u003e. \u003cem\u003eJ Clin Endocrinol Metab \u003c/em\u003e2019, \u003cstrong\u003e104\u003c/strong\u003e(11):4981-4989.\u003c/li\u003e\n\u003cli\u003eTorlontano M, Attard M, Crocetti U, Tumino S, Bruno R, Costante G, D'Azzo G, Meringolo D, Ferretti E, Sacco R\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eFollow-up of low risk patients with papillary thyroid cancer: role of neck ultrasonography in detecting lymph node metastases\u003c/strong\u003e. \u003cem\u003eJ Clin Endocrinol Metab \u003c/em\u003e2004, \u003cstrong\u003e89\u003c/strong\u003e(7):3402-3407.\u003c/li\u003e\n\u003cli\u003eChoi CH, Chung JY, Kang JH, Paik ES, Lee YY, Park W, Byeon SJ, Chung EJ, Kim BG, Hewitt SM\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eChemoradiotherapy response prediction model by proteomic expressional profiling in patients with locally advanced cervical cancer\u003c/strong\u003e. \u003cem\u003eGynecol Oncol \u003c/em\u003e2020, \u003cstrong\u003e157\u003c/strong\u003e(2):437-443.\u003c/li\u003e\n\u003cli\u003eTomas G, Tarabichi M, Gacquer D, Hebrant A, Dom G, Dumont JE, Keutgen X, Fahey TJ, 3rd, Maenhaut C, Detours V: \u003cstrong\u003eA general method to derive robust organ-specific gene expression-based differentiation indices: application to thyroid cancer diagnostic\u003c/strong\u003e. \u003cem\u003eOncogene \u003c/em\u003e2012, \u003cstrong\u003e31\u003c/strong\u003e(41):4490-4498.\u003c/li\u003e\n\u003cli\u003eDom G, Tarabichi M, Unger K, Thomas G, Oczko-Wojciechowska M, Bogdanova T, Jarzab B, Dumont JE, Detours V, Maenhaut C: \u003cstrong\u003eA gene expression signature distinguishes normal tissues of sporadic and radiation-induced papillary thyroid carcinomas\u003c/strong\u003e. \u003cem\u003eBr J Cancer \u003c/em\u003e2012, \u003cstrong\u003e107\u003c/strong\u003e(6):994-1000.\u003c/li\u003e\n\u003cli\u003eTarabichi M, Saiselet M, Tresallet C, Hoang C, Larsimont D, Andry G, Maenhaut C, Detours V: \u003cstrong\u003eRevisiting the transcriptional analysis of primary tumours and associated nodal metastases with enhanced biological and statistical controls: application to thyroid cancer\u003c/strong\u003e. \u003cem\u003eBr J Cancer \u003c/em\u003e2015, \u003cstrong\u003e112\u003c/strong\u003e(10):1665-1674.\u003c/li\u003e\n\u003cli\u003eLan X, Zhang H, Wang Z, Dong W, Sun W, Shao L, Zhang T, Zhang D: \u003cstrong\u003eGenome-wide analysis of long noncoding RNA expression profile in papillary thyroid carcinoma\u003c/strong\u003e. \u003cem\u003eGene \u003c/em\u003e2015, \u003cstrong\u003e569\u003c/strong\u003e(1):109-117.\u003c/li\u003e\n\u003cli\u003eYu K, Ganesan K, Tan LK, Laban M, Wu J, Zhao XD, Li H, Leung CH, Zhu Y, Wei CL\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA precisely regulated gene expression cassette potently modulates metastasis and survival in multiple solid cancers\u003c/strong\u003e. \u003cem\u003ePLoS Genet \u003c/em\u003e2008, \u003cstrong\u003e4\u003c/strong\u003e(7):e1000129.\u003c/li\u003e\n\u003cli\u003eLee S, Bae JS, Jung CK, Chung WY: \u003cstrong\u003eExtensive lymphatic spread of papillary thyroid microcarcinoma is associated with an increase in expression of genes involved in epithelial-mesenchymal transition and cancer stem cell-like properties\u003c/strong\u003e. \u003cem\u003eCancer Med \u003c/em\u003e2019, \u003cstrong\u003e8\u003c/strong\u003e(15):6528-6537.\u003c/li\u003e\n\u003cli\u003eIacobas DA, Tuli NY, Iacobas S, Rasamny JK, Moscatello A, Geliebter J, Tiwari RK: \u003cstrong\u003eGene master regulators of papillary and anaplastic thyroid cancers\u003c/strong\u003e. \u003cem\u003eOncotarget \u003c/em\u003e2018, \u003cstrong\u003e9\u003c/strong\u003e(2):2410-2424.\u003c/li\u003e\n\u003cli\u003eHe H, Jazdzewski K, Li W, Liyanarachchi S, Nagy R, Volinia S, Calin GA, Liu CG, Franssila K, Suster S\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe role of microRNA genes in papillary thyroid carcinoma\u003c/strong\u003e. \u003cem\u003eProc Natl Acad Sci U S A \u003c/em\u003e2005, \u003cstrong\u003e102\u003c/strong\u003e(52):19075-19080.\u003c/li\u003e\n\u003cli\u003eGiordano TJ, Kuick R, Thomas DG, Misek DE, Vinco M, Sanders D, Zhu Z, Ciampi R, Roh M, Shedden K\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eMolecular classification of papillary thyroid carcinoma: distinct BRAF, RAS, and RET/PTC mutation-specific gene expression profiles discovered by DNA microarray analysis\u003c/strong\u003e. \u003cem\u003eOncogene \u003c/em\u003e2005, \u003cstrong\u003e24\u003c/strong\u003e(44):6646-6656.\u003c/li\u003e\n\u003cli\u003eGiordano TJ, Au AY, Kuick R, Thomas DG, Rhodes DR, Wilhelm KG, Jr., Vinco M, Misek DE, Sanders D, Zhu Z\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eDelineation, functional validation, and bioinformatic evaluation of gene expression in thyroid follicular carcinomas with the PAX8-PPARG translocation\u003c/strong\u003e. \u003cem\u003eClin Cancer Res \u003c/em\u003e2006, \u003cstrong\u003e12\u003c/strong\u003e(7 Pt 1):1983-1993.\u003c/li\u003e\n\u003cli\u003ePaik S, Shak S, Tang G, Kim C, Baker J, Cronin M, Baehner FL, Walker MG, Watson D, Park T\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer\u003c/strong\u003e. \u003cem\u003eN Engl J Med \u003c/em\u003e2004, \u003cstrong\u003e351\u003c/strong\u003e(27):2817-2826.\u003c/li\u003e\n\u003cli\u003eGuinney J, Dienstmann R, Wang X, de Reynies A, Schlicker A, Soneson C, Marisa L, Roepman P, Nyamundanda G, Angelino P\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe consensus molecular subtypes of colorectal cancer\u003c/strong\u003e. \u003cem\u003eNat Med \u003c/em\u003e2015, \u003cstrong\u003e21\u003c/strong\u003e(11):1350-1356.\u003c/li\u003e\n\u003cli\u003eLehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, Shyr Y, Pietenpol JA: \u003cstrong\u003eIdentification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies\u003c/strong\u003e. \u003cem\u003eJ Clin Invest \u003c/em\u003e2011, \u003cstrong\u003e121\u003c/strong\u003e(7):2750-2767.\u003c/li\u003e\n\u003cli\u003eLapointe J, Li C, Higgins JP, van de Rijn M, Bair E, Montgomery K, Ferrari M, Egevad L, Rayford W, Bergerheim U\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eGene expression profiling identifies clinically relevant subtypes of prostate cancer\u003c/strong\u003e. \u003cem\u003eProc Natl Acad Sci U S A \u003c/em\u003e2004, \u003cstrong\u003e101\u003c/strong\u003e(3):811-816.\u003c/li\u003e\n\u003cli\u003eBailey P, Chang DK, Nones K, Johns AL, Patch AM, Gingras MC, Miller DK, Christ AN, Bruxner TJ, Quinn MC\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eGenomic analyses identify molecular subtypes of pancreatic cancer\u003c/strong\u003e. \u003cem\u003eNature \u003c/em\u003e2016, \u003cstrong\u003e531\u003c/strong\u003e(7592):47-52.\u003c/li\u003e\n\u003cli\u003eXu M, Li Y, Li W, Zhao Q, Zhang Q, Le K, Huang Z, Yi P: \u003cstrong\u003eImmune and Stroma Related Genes in Breast Cancer: A Comprehensive Analysis of Tumor Microenvironment Based on the Cancer Genome Atlas (TCGA) Database\u003c/strong\u003e. \u003cem\u003eFront Med (Lausanne) \u003c/em\u003e2020, \u003cstrong\u003e7\u003c/strong\u003e:64.\u003c/li\u003e\n\u003cli\u003eChakladar J, Chu M, Gnanasekar A, Rosenberg KF, Tsai JC, Wong LM, Ongkeko WM: \u003cstrong\u003eComputational analysis of immune-associated genomic and transcriptomic elements differentiating papillary thyroid cancer subtypes\u003c/strong\u003e. \u003cem\u003eCancer Research \u003c/em\u003e2019, \u003cstrong\u003e79\u003c/strong\u003e(13).\u003c/li\u003e\n\u003cli\u003eSoda G, Antonaci A, Bosco D, Nardoni S, Melis M: \u003cstrong\u003eExpression of bcl-2, c-erbB-2, p53, and p21 (waf1-cip1) protein in thyroid carcinomas\u003c/strong\u003e. \u003cem\u003eJ Exp Clin Cancer Res \u003c/em\u003e1999, \u003cstrong\u003e18\u003c/strong\u003e(3):363-367.\u003c/li\u003e\n\u003cli\u003eAksoy M, Giles Y, Kapran Y, Terzioglu T, Tezelman S: \u003cstrong\u003eExpression of bcl-2 in papillary thyroid cancers and its prognostic value\u003c/strong\u003e. \u003cem\u003eActa Chir Belg \u003c/em\u003e2005, \u003cstrong\u003e105\u003c/strong\u003e(6):644-648.\u003c/li\u003e\n\u003cli\u003eMessai Y, Noman MZ, Hasmim M, Janji B, Tittarelli A, Boutet M, Baud V, Viry E, Billot K, Nanbakhsh A\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eITPR1 protects renal cancer cells against natural killer cells by inducing autophagy\u003c/strong\u003e. \u003cem\u003eCancer Res \u003c/em\u003e2014, \u003cstrong\u003e74\u003c/strong\u003e(23):6820-6832.\u003c/li\u003e\n\u003cli\u003eXu S, Wang P, Zhang J, Wu H, Sui S, Zhang J, Wang Q, Qiao K, Yang W, Xu H\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAi-lncRNA EGOT enhancing autophagy sensitizes paclitaxel cytotoxicity via upregulation of ITPR1 expression by RNA-RNA and RNA-protein interactions in human cancer\u003c/strong\u003e. \u003cem\u003eMol Cancer \u003c/em\u003e2019, \u003cstrong\u003e18\u003c/strong\u003e(1):89.\u003c/li\u003e\n\u003cli\u003eBirchmeier C, Birchmeier W, Gherardi E, Vande Woude GF: \u003cstrong\u003eMet, metastasis, motility and more\u003c/strong\u003e. \u003cem\u003eNat Rev Mol Cell Biol \u003c/em\u003e2003, \u003cstrong\u003e4\u003c/strong\u003e(12):915-925.\u003c/li\u003e\n\u003cli\u003eChitikova Z, Pusztaszeri M, Makhlouf AM, Berczy M, Delucinge-Vivier C, Triponez F, Meyer P, Philippe J, Dibner C: \u003cstrong\u003eIdentification of new biomarkers for human papillary thyroid carcinoma employing NanoString analysis\u003c/strong\u003e. \u003cem\u003eOncotarget \u003c/em\u003e2015, \u003cstrong\u003e6\u003c/strong\u003e(13):10978-10993.\u003c/li\u003e\n\u003cli\u003eWang G, Cai C, Chen L: \u003cstrong\u003eMicroRNA-3666 Regulates Thyroid Carcinoma Cell Proliferation via MET\u003c/strong\u003e. \u003cem\u003eCell Physiol Biochem \u003c/em\u003e2016, \u003cstrong\u003e38\u003c/strong\u003e(3):1030-1039.\u003c/li\u003e\n\u003cli\u003eLutterbach B, Zeng Q, Davis LJ, Hatch H, Hang G, Kohl NE, Gibbs JB, Pan BS: \u003cstrong\u003eLung cancer cell lines harboring MET gene amplification are dependent on Met for growth and survival\u003c/strong\u003e. \u003cem\u003eCancer Res \u003c/em\u003e2007, \u003cstrong\u003e67\u003c/strong\u003e(5):2081-2088.\u003c/li\u003e\n\u003cli\u003eShintani T, Kusuhara Y, Daizumoto K, Dondoo TO, Yamamoto H, Mori H, Fukawa T, Nakatsuji H, Fukumori T, Takahashi M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe Involvement of Hepatocyte Growth Factor-MET-Matrix Metalloproteinase 1 Signaling in Bladder Cancer Invasiveness and Proliferation. Effect of the MET Inhibitor, Cabozantinib (XL184), on Bladder Cancer Cells\u003c/strong\u003e. \u003cem\u003eUrology \u003c/em\u003e2017, \u003cstrong\u003e101\u003c/strong\u003e:169 e167-169 e113.\u003c/li\u003e\n\u003cli\u003eSaintigny P, William WN, Jr., Foy JP, Papadimitrakopoulou V, Lang W, Zhang L, Fan YH, Feng L, Kim ES, El-Naggar AK\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eMet Receptor Tyrosine Kinase and Chemoprevention of Oral Cancer\u003c/strong\u003e. \u003cem\u003eJ Natl Cancer Inst \u003c/em\u003e2018, \u003cstrong\u003e110\u003c/strong\u003e(3).\u003c/li\u003e\n\u003cli\u003eZhai T, Muhanhali D, Jia X, Wu Z, Cai Z, Ling Y: \u003cstrong\u003eIdentification of gene co-expression modules and hub genes associated with lymph node metastasis of papillary thyroid cancer\u003c/strong\u003e. \u003cem\u003eEndocrine \u003c/em\u003e2019, \u003cstrong\u003e66\u003c/strong\u003e(3):573-584.\u003c/li\u003e\n\u003cli\u003eTang X, Huang X, Wang D, Yan R, Lu F, Cheng C, Li Y, Xu J: \u003cstrong\u003eIdentifying gene modules of thyroid cancer associated with pathological stage by weighted gene co-expression network analysis\u003c/strong\u003e. \u003cem\u003eGene \u003c/em\u003e2019, \u003cstrong\u003e704\u003c/strong\u003e:142-148.\u003c/li\u003e\n\u003cli\u003eCollins GS, Reitsma JB, Altman DG, Moons KG: \u003cstrong\u003eTransparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement\u003c/strong\u003e. \u003cem\u003eBMJ \u003c/em\u003e2015, \u003cstrong\u003e350\u003c/strong\u003e:g7594.\u003c/li\u003e\n\u003cli\u003eMachens A, Holzhausen HJ, Dralle H: \u003cstrong\u003eSkip metastases in thyroid cancer leaping the central lymph node compartment\u003c/strong\u003e. \u003cem\u003eArch Surg \u003c/em\u003e2004, \u003cstrong\u003e139\u003c/strong\u003e(1):43-45.\u003c/li\u003e\n\u003cli\u003ePodnos YD, Smith D, Wagman LD, Ellenhorn JD: \u003cstrong\u003eThe implication of lymph node metastasis on survival in patients with well-differentiated thyroid cancer\u003c/strong\u003e. \u003cem\u003eAm Surg \u003c/em\u003e2005, \u003cstrong\u003e71\u003c/strong\u003e(9):731-734.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003eInformation of enrolled PTC patients from 8 GEO datasets after quality control.\u003c/b\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCountry\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOrganization\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSeries\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePlatform\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eNormal\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eTumor\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eQuality control\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003ePublication\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBelgium\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eULB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE33630\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eExcluded 2 tumor samples\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(Collins et al. 2015, Dom et al. 2012)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBelgium\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eIRIBHM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE60542\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePassed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(Tarabichi et al. 2015)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChina\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eThe First Hospital of China Medical University\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE66783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL19850\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePassed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(Lan et al. 2015)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSingapore\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNational Cancer Centre\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE5364\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePassed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(Yu et al. 2008)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSouth Korea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eThe Catholic University of Korea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE129562\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL10558\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePassed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(Lee et al. 2019)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCenter for Computational Systems Biology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE97001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL10332\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePassed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(Iacobas et al. 2018)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOhio State University\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE3467\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eExcluded 4 tumor and 2 normal\u003c/p\u003e \u003cp\u003esamples\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(He et al. 2005)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUniversity of Michigan\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGSE27155\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGPL96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eExcluded 1 tumor sample\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e((Giordano et al. 2006; Giordano et al. 2005)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \n\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003eClinical pathological characteristics of patients in the training, self-validation cohorts and the independent GEO cohort.\u003c/b\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eCharacteristics\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTCGA training cohort\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTCGA validation cohort\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGEO validation cohort\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;248)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;249)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;42)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge at initial diagnosis (year)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e46.3\u0026thinsp;\u0026plusmn;\u0026thinsp;15.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e48.5\u0026thinsp;\u0026plusmn;\u0026thinsp;16.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e45.11\u0026thinsp;\u0026plusmn;\u0026thinsp;13.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e82(33.06%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e52(20.88%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e19(45.24%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e166(66.94%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e197(79.12%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e23(54.76%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePathologic T\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eT1 or Tx\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e69(27.82%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e75(30.12%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9(21.43%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eT2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e81(32.66%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e81(32.53%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4(9.52%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eT3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e87(35.08%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e82(32.93%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e24(57.12%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eT4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11(4.44%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11(4.42%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5(11.90%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePathologic N\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN0 or Nx\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e139(56.05%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e137(55.02%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e19(45.24%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e109(43.95%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e112(44.98%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e23(54.76%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePathologic M\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eM0 or Mx\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e244(98.39%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e245(98.39%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e39(92.86%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eM1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4(1.61%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4(1.61%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3(7.12%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor stage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage I\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e144(58.06%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e135(54.22%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e19(45.24%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage II\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27(10.89%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25(10.04%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0(0.00%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage III\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e52(20.97%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e59(23.69%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10(23.81%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage IV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e24(9.68%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e29(11.65%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3(7.14%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNot report\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1(0.40%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1(0.40%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10(23.81%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOverall survival status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNA\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAlive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e242(97.58%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e239(95.98%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDead\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6(2.42%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10(4.02%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eThe three independent prediction factors of lymph node metastasis in papillary thyroid cancer.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"10\"\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eEntrez ID\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eGene\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c6\" namest=\"c3\"\u003e \u003cp\u003emultivariate Cox regression analysis\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003eEight GEO datasets\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e \u003cp\u003eTCGA dataset\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCoefficient\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHazard ratio\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003econfidence interval (95%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eLog\u003csub\u003e2\u003c/sub\u003eFC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eAdjusted \u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eLog\u003csub\u003e2\u003c/sub\u003eFC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eAdjusted \u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3708\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eITPR1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.589\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.801\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.133\u0026ndash;2.864\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.013\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.322\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e3.532e-05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e1.161\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3.343e-24\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4233\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMET\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.841\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.320\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.606\u0026ndash;3.350\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7.000e-06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.686\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e2.194e-08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.620\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.746e-23\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e596\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBCL2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.786\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.456\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.271\u0026ndash;0.767\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.030\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e-0.342\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e2.736e-04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e-0.689\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e5.406e-26\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Prediction model, Papillary thyroid cancer, Lymph node metastasis, Transcriptome analysis, WGCNA","lastPublishedDoi":"10.21203/rs.3.rs-41157/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-41157/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e Thyroid cancer is one of the most prevalent endocrine cancers with a rising incidence rate over the past years. Papillary thyroid cancer (PTC) is the dominant historical type of thyroid cancer. Early lymph node metastasis happens frequently in PTC. However, some of the lymph node metastasis may be troublesome for detecting because of limited methods.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eRobust rank aggregation afforded us the shared differential expression genes among multiple datasets. Gene ontology analysis was performed to identify potential functions. Weighted gene co-expression network analysis was used to research the correlations between gene expression patterns with clinical characteristic. Protein-protein interaction network was performed to identify the hub genes. The least absolute shrinkage and selection operator and Logistic regression were performed to construct a prediction model.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e We developed a three-gene signature prediction model for lymph node metastasis in PTC through transcriptomic analysis. After quality control, we collected 8 microarray datasets from GEO database and an RNA sequencing dataset from TCGA database. We found the transcriptome profiles were correlated with lymph node metastasis and 3 genes were verified to be independent prediction factors towards those statistic approach. Afterwards, we designed a predicable risk score system and effectively confirmed the model in two independent papillary thyroid cancer cohorts.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e We recommended a successful predicable model of lymph node metastasis in papillary thyroid cancer patients with moderate accuracy.\u003c/p\u003e","manuscriptTitle":"Development of A Three-Gene Signature Prediction Model for Lymph Node Metastasis in Papillary Thyroid Cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2020-07-14 17:49:00","doi":"10.21203/rs.3.rs-41157/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f5a9765d-cc40-481d-a73c-22768f32a3e0","owner":[],"postedDate":"July 14th, 2020","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":152973,"name":"Cancer Biology"}],"tags":[],"updatedAt":"2020-07-15T22:28:22+00:00","versionOfRecord":[],"versionCreatedAt":"2020-07-14 17:49:00","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-41157","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-41157","identity":"rs-41157","version":["v1"]},"buildId":"WrCJVZZCHTDjtuVLN7oU0","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}