A Computer Tomography Radiomics-based model to Predict Survival of Patients with EGFR‑Mutated Non‑small‑cell Lung Cancer

A Computer Tomography Radiomics-based model to Predict Survival of Patients with EGFR‑Mutated Non‑small‑cell Lung 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 Research Article A Computer Tomography Radiomics-based model to Predict Survival of Patients with EGFR‑Mutated Non‑small‑cell Lung Cancer Chen Huang, Yubin Cao, Lei Zhu, Qunli Xiong, Qin Zhang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-1840484/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 To evaluate the prognosis of EGFR -mutated non-small-cell lung cancer(NSCLC) patients and accurately target the patient subgroups with best potential outcome greatest, a simpler and more effective model based on readily available laboratory parameters is needed in clinical practice. Methods Totally, computed tomographic (CT) images from 162 EGFR-mutated NSCLC patients at the time of first diagnosis in West China Hospital of Sichuan University were retrospectively collected and analysed to describe the target area of the lesion. All patients were randomly divided into a training cohort and a validation cohort. Radiomic features from images were extracted. The least absolute shrinkage selection operator (LASSO) regression method was used to screen valuable radiomic features. The logistic regression method was used to establish a radiomic model, and the nomogram was used to evaluate the discrimination ability. Receiver characteristic curve (ROC) and decision curve analysis (DCA) were used to evaluate the performance of the model. Results We established a nomogram that combined 6 important clinical factors and imaging risk scores to predict the three-year and five-year survival rates of EGFR-mutated NSCLC patients. The AUC-ROC of three-year OS was 0.715, and the five-year OS was 0.705, which indicated a favourable discrimination capability of our nomogram and its great potential in targeting clinical services and predicting patient outcomes. Conclusion Based on chest CT imaging and clinicopathological features, including age, sex, Stage_T, Stage_N, Stage_M, clinical stage, and imaging risk score, we constructed and validated a nomogram for predicting an individualized prediction of survival for patients with EGFR‑mutated NSCLC. EGFR‑Mutated NSCLC radiomics prognosis nomogram Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction Lung cancer is currently the most frequently diagnosed cancer in the world and the prominent reason for tumor-related deaths [ 1 , 2 ]. Lung cancer usually consists of two subtypes: non-small-cell lung cancer (NSCLC), which is estimated to be 85% of lung tumors, and small cell lung cancer (SCLC) [ 2 , 3 ]. Epidermal growth factor receptor (EGFR) is a 170 kDa transmembrane glycoprotein and a member of the HER/ErbB family, which is distributed extensively on the cell membrane in human tissues. The binding of EGFR to the EGFR ligand induces receptor dimerization and causes a conformational change that triggers multiple signal transduction pathways, such as the phosphoinositide 3-kinase/threonine-protein kinase (PI3K) pathway and mitogen-activated protein kinase (MARK) pathway [ 4 , 5 ]. EGFR is one of the most significantly mutated genes driving lung cancer, and approximately 50% of Asian patients with NSCLC possess EGFR gene mutations [ 6 , 7 ]. The most common activating mutations in the EGFR gene are a short deletion in exon 19 and a missense mutation (L858R) in exon 21 [ 8 , 9 ]. The overexpression of EGFR is the main reason that causes tumor cells to proliferate uncontrollably, to escape programmed death, to enhance tumor cell migration and to promote metastasis [ 4 , 5 ]. The activation of EGFR also plays a significant role in the resistance of tumor cells to chemotherapy and radiotherapy [ 10 , 11 ]. EGFR tyrosine kinase inhibitors (EGFR-TKIs) are established as the first line of treatment for EGFR‐mutated NSCLC patients; however, approximately 30% of these patients are clinically insensitive to EGFR-TKIs [ 12 , 13 ]. Thus, an individualized method is needed to develop precision therapy strategy and to predict prognosis for EGFR -mutated NSCLC patients [ 14 ]. At present, although the TNM staging system has been widely used to predict the prognosis of cancer, it is difficult to accurately predict the outcome merely based on the TNM staging system. Even within the same TNM stage, patients may have survival differences [ 18 ]. Therefore, to evaluate the prognosis of EGFR -mutated non-small-cell lung cancer(NSCLC) patients and accurately target the patient subgroups with best potential outcome greatest, a simpler and more effective model based on readily available laboratory parameters is needed in clinical practice. Radiomics is a common tool in oncologic clinical workflow, allowing for prediction of the treatment response and clinical outcome [ 19 – 21 ]. Previous studies have used radiomics combined with the nomogram model to identify imaging biomarkers for the classification of serous and mucinous types of ovarian cystadenomas [ 22 ]. In distinguishing aggressive lung adenocarcinoma from AIS/MIA, this method also showed a better diagnostic performance than 3D volume analysis with an accuracy rate of 81.2%[ 23 ]. Also, researchers have attempted to include imaging omics and multiple clinical factors to effectively predict the preoperative LNM of NSCLC patients, which greatly saved time in clinical practice [ 24 , 25 ]. Inspired by former explorations, we hypothesized that computed tomographic (CT) imaging omics with more imaging omics features might provide some valuable tumor heterogeneity information to predict the future prognosis of EGFR -mutated NSCLC patients at the time of initial diagnosis. This study presents a radiologically based analysis to assess the prognosis of NSCLC patients with EGFR mutations, with a view to pinpointing the subgroup of patients with the better potential outcomes. Methods Patients selection This retrospective and observational study was conducted in West China Hospital, a university affiliated medical center with 4,500 beds (Chengdu, Sichuan Province, China). The study population consisted of 173 lung cancer patients who were treated in West China Hospital of Sichuan University from January 2009 to December 2018 and received EGFR gene mutation detection. After screening, 162 cases were finally included. This study was approved by the Ethics Committee of West China Hospital of Sichuan University. Our inclusion criteria were as follows: (1) initial pathological diagnosis of primary NSCLC; (2) clear EGFR detection results; and (3) chest CT examination with a layer thickness of 5 mm performed at West China Hospital of Sichuan University within one month before diagnosis for imaging modelling analysis. The exclusion criteria were as follows: (1) chest local treatment or systemic treatment for other malignant tumours before CT examination; (2) massive pleural effusion completely masking the tumour; and (3) multiple-source lung cancer and inconsistent EGFR detection results. The inclusion and exclusion criteria are described in the flowchart (Fig. 1 ). Basic clinical pathological information For the included patients, the following indicators were collected from the hospital electronic clinical database: age (years), sex (male vs female), smoking history (with vs without), smoking amount (package year), family history of malignant tumours (with vs without), histological subtype (adenocarcinoma, squamous cell carcinoma, other), and tumour stage (I-III vs IV). For cases missing information regarding tumour location and tumour stage in the clinical reports, tumour location and clinical stage were evaluated by an imaging system. Clinical staging was performed according to the eighth edition of the International Association for Lung Cancer Research Tumour-Node-Metastasis (TNM) staging {Wilfried, 2015 #1037}. EGFR detection methods referred in our research included single-gene sequencing and multigene sequencing. Single-gene sequencing was performed using amplification refractory mutation system-polymerase chain reaction (ARMS-PCR); multigene sequencing was performed using next-generation sequencing (NGS). CT Acquisition IQQA software (IQQA-Chest, EDDA Technology, Princeton Junction, NJ, USA) was used one by one from the hospital PACS system to extract eligible chest CT images in a DICOM format file. Chest CT detection equipment included Siemens, Philips Brilliance Big Bore, and GE Discovery CT750 HD. All patients underwent spiral CT scans from the apex to the bottom of the lung under maximum inspiratory state. For CT scanners, tube voltage was set at 120 kV, tube current was set at 200–500 mA, rotation time was set at 0.4–0.7 s, and matrix of 512 × 512 pixel was produced. For the Siemens scanner, the convolution kernels were B31f and B30f; for the GE scanner, the reconstructed image of the STANDARD kernel was selected. Tumor Segmentation and ROI delineation Imaging omics feature extraction requires accurate delineation of tumors. Therefore, patients with unclear tumor boundaries will not be able to be included in the establishment of an imaging omics model. Tumors that cannot be accurately delineated include but are not limited to the following situations: obstructive pneumonia, unclear boundaries with pulmonary hilus, and pleural effusion. In our research, the precise delineation of tumors was realized by two physicians with clinical experience independently through the manual operation of ITK-SNAP software ( www.itksnap.org ). First, the slice of CT containing the tumor was determined, and the outline was carried out along the edge of the tumor. Then, the outline of the tumor was filled, and finally, all the slices of CT containing the tumor were plotted layer by layer. The red region in the figure was the precisely delineated region of interest for the tumor. Three-dimensional tumors could be generated in three-dimensional space by drawing the region of interest of the tumor layer by layer. Extraction of Radiomics Features The Pyradiomics toolkit ( https://github.com/Radiomics/pyradiomics ) was used to extract features. A total of 1353 radiomic features were extracted from ROIs. First-order, GLCM, GLRLM, GLSZM, GLDM and shape were extracted. Of all the features, there were 107 original shape features, 465 features using Gaussian (LoG, σ:1,2,3,4,5) Laplacian filtering and 744 wavelet (wavelet transform: LLL, LLH, LHH, LHL, HHH, HHL, HLH, HLL) features. The texture features mainly reflected the gray distribution of the image, and the wavelet features reflected the different frequency ranges of the original image. The first-order features described the intensity distribution of pixels or voxels in the region of interest, and the shape features described the shape and volume of the region of interest. Feature selection The best radiomics features were picked by a three-step feature selection process. First, to ensure reproducibility, CT images of 30 patients were stochastically selected and resegmented by two independent doctors at an interval of 2 months. To assess the reproducibility of the extracted features, 30 patients were randomly selected and quantified by two oncologists. To obtain the robustness of radiomics features without the influence of tumor segmentation variability (ICC > 0.75), the intergroup and intraclass correlation coefficients (ICCs) were calculated [ 26 , 27 ]. Then, to construct a nomogram, the variables were observed by drawing plots, and univariate analysis was performed to select definite characteristics. The least absolute shrinkage and selection operator (LASSO) regression model is a penalizing regression method that estimates the regression coefficients by maximizing the log-likelihood function while restraining the sum of the absolute values of the regression coefficients, which could avoid overfitting and was shown to be near-minimax optimal to obtain the outcome-predictive features from patients who underwent (chemo)radiotherapy [ 28 , 29 ]. We used the “glmnet” package of R software, which is a powerful tool to apply the selected optimal variables, especially high-dimensional data [ 30 ]. Then, the characteristics were selected by a minimum λ. Finally, the prediction model was constructed by LASSO to perform multivariable logistic regression. Nomogram establishment and validation After the above variable selection, the “rms” package of R software was used to generate the nomogram [ 28 ]. The receiver operating characteristic (ROC) curve includes the false positive rate on its horizontal axis and the true positive rate on its vertical axis. To evaluate the discrimination ability, the nomogram was assessed by the area under the curve (AUC). Furthermore, we used the calibration plot to evaluate the goodness of fit between the observed values and the predicted values. Clinical Use To evaluate the clinical application value of the nomotu model and radiomics characteristics, we carried out decision curve analysis (DCA) to evaluate the net benefits of various threshold risks. Statistical analysis In this study, continuous variables were presented as median values with interquartile ranges (IQRs), while categorical variables are presented as percentages. All statistical analyses were performed using R software (version 3.4.0; www.r-project.org ). All tests were two-sided, and values of P < 0.05 were considered statistically significant. Results Patient clinical features There were 173 lung cancer patients who were treated at West China Hospital of Sichuan University from January 2009 to December 2018 and had EGFR gene mutations. Based on the inclusion/exclusion criteria, 162 patients were enrolled in this study. The former included 74(45.7%) males and 88 (54.3%) females. Median age was 59 years (IQR, 50–65). Their characteristics are presented in Table 1 . Table 1 Participant characteristics of participants Total (N=162) Gender Male Female 45.7% (74/162) 54.3% (88/162) Age Mean (SD) Median (IQR) Smoking history Current or former Never Pathological type Squamous cell carcinoma Neuroendocrine carcinoma Adenocarcinoma Adenosquamous carcinoma 58.19 (11.10) 59 (50, 65) 71.6% (116/162) 28.4% (46/162) 41.4% (67/162) 0.6% (1/162) 37.0% (60/162) 21.0% (34/162) Clinical stage I II III IV 11.7% (19/162) 6.8% (11/162) 32.5% (53/162) 48.8% (79/162) TNM stage T (primary tumor) T1 T2 T3 T4 N (regional lymph nodes) N0 N1 N2 N3 Nx M (distant metastases) M0 M1 Mx 5.6% (9/162) 41.4% (67/162) 25.3%(31/162) 34.6% (56/162) 22.2% (36/162) 8.0% (13/162) 46.9% (76/162) 20.4% (33/162) 2.5% (4/162) 51.2% (83/162) 48.1% (78/162) 0.6% (1/162) Targeted therapy Current or former Never 51.2% (83/162) 48.8% (79/162) Feature selection The univariate analysis to select definite characteristics is shown in Table 2 . Through univariate analysis, we selected 8 characteristics for the LASSO regression model. The two dashed lines in Fig. 2 represent two special λ values: lambda.min and lambda.lse. lambda.min represents the λ value when the error is the smallest, which can obtain the least target parameter; lambda.lse represents the λ value of the simplest model within a variance range of lambda.min. We selected an optimal lambda of 5 with the smallest binomial deviance, and the initial variables were reduced to five potential predictors under penalizing conditions (Fig. 2 A). Table 2 Univariate analysis to select definite characteristics. HR P value wavelet-LHL_firstorder_Kurtosis wavelet-LHH_firstorder_Kurtosis wavelet-HHL_firstorder_Kurtosis wavelet-HHL_ngtdm_Contrast wavelet-HHH_firstorder_Kurtosis 1.04(1.01–1.07) 1.06(1.01–1.12) 1.04(1.02–1.06) 0.00(0.00-0.43) 1.03(1.01–1.05) 0.008 0.026 0.000 0.041 0.011 wavelet-HHH_firstorder_Skewness 0.25(0.07–0.85) 0.027 Wavelet-HHH_gldm_DependenceNonUniformityNormalized 0.00(0.00-0.05) 0.025 wavelet-HHH_gldm_DependenceVariance 1.13(1.02–1.26) 0.025 HR, hazard ratio. Construction of the risk score model With the median risk score as the cut-off value, we divided the 162 patients into a high-risk group and a low-risk group. The Kaplan–Meier curve show that the OS of the low-risk group was significantly better than that of the high-risk group (p < 0.00) (Fig. 3 A). The risk curve and scatter plot show that the risk factors and mortality of patients in the high-risk group were higher than those in the low-risk group (Fig. 3 B, C). Nomogram establishment and validation According to the multivariable logistic regression analysis, the statistically significant features that were retained included age, sex, Stage_T, Stage_N, Stage_M, clinical stage, and imaging risk score. We then developed a prediction model that incorporated the above independent features and presented it as a nomogram to predict the three-year and five-year survival rates of EGFR-mutated NSCLC patients. (Fig. 4 ). The calibration curves shown in Fig. 5 illustrate a good consistency between the nomogram-predicted three-year and five-year survival rates of EGFR-mutated NSCLC patients. Moreover, the AUC-ROC of three-year (A) OS was 0.715, while five-year (B) OS was 0.705, as is displayed in Fig. 6 . This indicates a favourable discrimination and demonstrates a good fit for the purpose of targeting clinical services and predicting patient outcomes. Clinical Use The decision curve analysis for the radiomics nomogram and that for the model with integrated histologic grade is presented in Fig. 7 . The decision curve shows that if the threshold probability of a doctor is 10%, our radiomics nomogram to predict the three-year (Fig. 7 A) and five-year survival rates (Fig. 7 B) of EGFR-mutated NSCLC patients adds more clinical benefit than either the treat-all-patients scheme or the treat-none scheme. Within this range, net benefit was comparable. Discussion In our study, we selected 20 features by univariate analysis and further identified 7 features by multivariable logistic regression analysis. The final seven characteristics selected included six clinically important factors including age, gender, stage_T, stage_N, stage_M and clinical stage as well as one imaging risk score. Then, we developed a prediction model combining the above 7 independent features and presented it as a nomogram to predict three-year and five-year survival rates of EGFR-mutated NSCLC patients. This study explored the correlation between tumor imaging omics characteristics and patient prognosis through in-depth mining of tumor CT imaging information and constructed a stable and effective prognostic analysis model to predict the prognosis and survival of NSCLC patients. The experimental results proved that the imaging omics characteristics of tumors could be utilized to quantitatively evaluate the heterogeneity of tumors. The prognostic analysis model based on tumor imaging omics can effectively predict the prognostic survival time range of NSCLC patients and can assist doctors in analysing, diagnosing and predicting the prognosis of patients. Previous studies have attempted to explore related risk factors for the prognosis of NSCLC from clinical features, such as pathological staging [ 31 ]. However, some risk factors (such as the depth of submucosal invasion, lymphatic invasion, and vascular invasion) are related to histopathological characteristics, so they cannot be obtained before surgery, which might result in a delay in making treatment decisions. In clinical practice, CT is the most commonly used preoperative imaging method for NSCLC. However, the limitation of traditional CT examination is that it entails much professional training to interpret and lacks direct clues pointing to prognosis [ 32 ]. Wang et al. extracted the burr sign, lobular sign, ground glass nodule, cavity sign, bronchial inflation sign, pleural depression sign, and pleural adhesion sign of 79 NSCLC patients with stage Ⅰ CT images [ 33 ]. The survival analysis found that the pleural adhesion was an independent prognostic factor of stage I NSCLC. However, in the former attempts exploring the association of clinical features and prognosis in NSCLC, prognostic factors were screened out, but whether they had real clinical implications was still under doubt for the following reasons (1) simple clinical characteristics cannot be used to assess risk factors related to histopathological characteristics. (2) The obtained prognostic factors can only explore the prognosis of the patient. However, they cannot effectively help doctors predict the prognosis of NSCLC patients. To help resolve this problem, in this study, we retrospectively collected clinical data and imaging data of EGFR-mutant NSCLC patients and extracted the CT imaging features of their first diagnosis. Then, the predictive value of several known clinical factors and imaging features that affect the prognosis of predicting the survival of such patients was evaluated. Based on these results, we established a nomogram combining 6 important clinical factors and imaging risk scores to predict the three-year and five-year survival rates of EGFR-mutated NSCLC patients. If developed, validated, and used correctly, our nomogram can provide important information about patient care [ 34 ]. There are also some limitations in this study. First, this retrospective study was conducted in a single center with a limited sample size and lacked external verification, resulting from the difficulty of obtaining CT images. But given our carefully developed research design, the authenticity and validity of our research could be ensured. Second, the model still requires more biomarkers to improve its accuracy. To conclude, in order to further our exploration in enhancing this model, it is necessary to expand the sample size and to include more relevant clinical data and a wider range of biomarkers. Conclusion Utilizing chest CT imaging and clinicopathological features, included age, sex, Stage_T, Stage_N, Stage_M, clinical stage, and imaging risk score, we constructed and validated a nomogram for predicting an individualized prediction of survival for patients with EGFR‑mutated NSCLC. Declarations Ethics approval and consent to participate The study was approved by the ethics committee of West China Hospital [2019(195)] written informed consent was obtained from all patients for research purposes. All methods were carried out in accordance with relevant guidelines and regulations or declaration of helsinki. Consent for publication N.A. Availability of data and materials The datasets used or analysed during the current study are available from the corresponding author on reasonable request. Competing interests The authors declare that they have no competing interests Funding N.A. Authors' contributions QZ contributed to the guidance of this article and given the final approval of the version to be published; CH drafted the manuscript and YBC prepared the figures. LZ and QLX collected the datas. All authors read and approved the final manuscript. Acknowledgements N.A. References Barta JA, Powell CA, Wisnivesky JP. Global Epidemiology of Lung Cancer. Ann Glob Health. 2019;85(1). Epub 2019/02/12. doi: 10.5334/aogh.2419 . PubMed PMID: 30741509; PubMed Central PMCID: PMCPMC6724220. Hirsch FR, Scagliotti GV, Mulshine JL, Kwon R, Curran WJ, Jr., Wu YL, et al. Lung cancer: current therapies and new targeted treatments. Lancet (London, England). 2017;389(10066):299–311. Epub 2016/08/31. doi: 10.1016/s0140-6736(16)30958-8 . PubMed PMID: 27574741. Keith RL, Miller YE. Lung cancer chemoprevention: current status and future prospects. Nat Rev Clin Oncol. 2013;10(6):334–43. Epub 2013/05/22. doi: 10.1038/nrclinonc.2013.64 . PubMed PMID: 23689750; PubMed Central PMCID: PMCPMC4687734. Hess LM, Kern DM, Carter GC, Winfree K, Wang L, Sontag A, et al. Real-World Treatment Sequences and Outcomes Among Patients With Non-Small Cell Lung Cancer (RESOUNDS) in the United States: Study Protocol. JMIR Res Protoc. 2017;6(10):e195. Epub 2017/10/13. doi: 10.2196/resprot.7750 . PubMed PMID: 29021129; PubMed Central PMCID: PMCPMC5658639. Goto Y, Tanai C, Yoh K, Hosomi Y, Sakai H, Kato T, et al. Continuing EGFR-TKI beyond radiological progression in patients with advanced or recurrent, EGFR mutation-positive non-small-cell lung cancer: an observational study. ESMO Open. 2017;2(4):e000214. Epub 2017/10/12. doi: 10.1136/esmoopen-2017-000214 . PubMed PMID: 29018574; PubMed Central PMCID: PMCPMC5604715. Wang J, Sun P, Wang Q, Zhang P, Wang Y, Zi C, et al. (-)-Epigallocatechin-3-gallate derivatives combined with cisplatin exhibit synergistic inhibitory effects on non-small-cell lung cancer cells. Cancer Cell Int. 2019;19:266. Epub 2019/10/23. doi: 10.1186/s12935-019-0981-0 . PubMed PMID: 31636509; PubMed Central PMCID: PMCPMC6791019. Shi Y, Li J, Zhang S, Wang M, Yang S, Li N, et al. Molecular Epidemiology of EGFR Mutations in Asian Patients with Advanced Non-Small-Cell Lung Cancer of Adenocarcinoma Histology - Mainland China Subset Analysis of the PIONEER study. PLoS One. 2015;10(11):e0143515. Epub 2015/11/26. doi: 10.1371/journal.pone.0143515 . PubMed PMID: 26599344; PubMed Central PMCID: PMCPMC4657882. Tsao AS, Scagliotti GV, Bunn PA, Jr., Carbone DP, Warren GW, Bai C, et al. Scientific Advances in Lung Cancer 2015. J Thorac Oncol. 2016;11(5):613–38. Epub 2016/03/26. doi: 10.1016/j.jtho.2016.03.012 . PubMed PMID: 27013409. Costa DB. Kinase inhibitor-responsive genotypes in EGFR mutated lung adenocarcinomas: moving past common point mutations or indels into uncommon kinase domain duplications and rearrangements. Transl Lung Cancer Res. 2016;5(3):331–7. Epub 2016/07/15. doi: 10.21037/tlcr.2016.06.04 . PubMed PMID: 27413714; PubMed Central PMCID: PMCPMC4931120. Dong Z, Zhang J, Liu S, Su J, Zhang C, Xie Z, et al. EGFR mutation correlates with uninflamed phenotype and weak immunogenicity, causing impaired response to PD-1 blockade in non-small cell lung cancer. Oncoimmunology. 2017;6(11):e1356145. doi: 10.1080/2162402x.2017.1356145. PubMed PMID: 29147605. Singh D, Attri BK, Gill RK, Bariwal J. Review on EGFR Inhibitors: Critical Updates. Mini Rev Med Chem. 2016;16(14):1134–66. Epub 2016/03/22. doi: 10.2174/1389557516666160321114917. PubMed PMID: 26996617. Tan DS, Yom SS, Tsao MS, Pass HI, Kelly K, Peled N, et al. The International Association for the Study of Lung Cancer Consensus Statement on Optimizing Management of EGFR Mutation-Positive Non-Small Cell Lung Cancer: Status in 2016. J Thorac Oncol. 2016;11(7):946–63. Epub 2016/05/28. doi: 10.1016/j.jtho.2016.05.008 . PubMed PMID: 27229180. Taniguchi H, Yamada T, Wang R, Tanimura K, Adachi Y, Nishiyama A, et al. AXL confers intrinsic resistance to osimertinib and advances the emergence of tolerant cells. Nat Commun. 2019;10(1):259. Epub 2019/01/18. doi: 10.1038/s41467-018-08074-0 . PubMed PMID: 30651547; PubMed Central PMCID: PMCPMC6335418 Boehringer-Ingelheim, and has received speaking honoraria from AstraZeneca, Chugai Pharma, and Boehringer-Ingelheim. The remaining authors declare no competing interests. Jiang AY, Zhang J, Luo HL, Gao F, Lv YF. Icotinib and whole-brain radiotherapy for the treatment in patients with brain metastases from EGFR-mutant nonsmall cell lung cancer: A retrospective study. Medicine. 2018;97(15):e0312. Epub 2018/04/12. doi: 10.1097/md.0000000000010312 . PubMed PMID: 29642161; PubMed Central PMCID: PMCPMC5908633. Cuneo KC, Nyati MK, Ray D, Lawrence TS. EGFR targeted therapies and radiation: Optimizing efficacy by appropriate drug scheduling and patient selection. Pharmacology & therapeutics. 2015;154:67–77. Epub 2015/07/25. doi: 10.1016/j.pharmthera.2015.07.002. PubMed PMID: 26205191; PubMed Central PMCID: PMCPMC4570853. Russo A, Franchina T, Ricciardi GR, Picone A, Ferraro G, Zanghì M, et al. A decade of EGFR inhibition in EGFR-mutated non small cell lung cancer (NSCLC): Old successes and future perspectives. Oncotarget. 2015;6(29):26814–25. Epub 2015/08/27. doi: 10.18632/oncotarget.4254 . PubMed PMID: 26308162; PubMed Central PMCID: PMCPMC4694955. Improta G, Zupa A, Natalicchio MI, Sisinni L, Marinaccio A, Bozza G, et al. Uncommon frame-shift exon 19 EGFR mutations are sensitive to EGFR tyrosine kinase inhibitors in non-small cell lung carcinoma. Med Oncol. 2018;35(3):28. Epub 2018/02/02. doi: 10.1007/s12032-018-1078-7 . PubMed PMID: 29387949. Wei XL, Zhang DS, He MM, Jin Y, Wang DS, Zhou YX, et al. The predictive value of alkaline phosphatase and lactate dehydrogenase for overall survival in patients with esophageal squamous cell carcinoma. Tumour Biol. 2016;37(2):1879-87. Epub 2015/09/02. doi: 10.1007/s13277-015-3851-y . PubMed PMID: 26323257. Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology. 2016;278(2):563–77. Epub 2015/11/19. doi: 10.1148/radiol.2015151169 . PubMed PMID: 26579733; PubMed Central PMCID: PMCPMC4734157. Aerts HJ, Velazquez ER, Leijenaar RT, Parmar C, Grossmann P, Carvalho S, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014;5:4006. Epub 2014/06/04. doi: 10.1038/ncomms5006 . PubMed PMID: 24892406; PubMed Central PMCID: PMCPMC4059926. Lambin P, Leijenaar RTH, Deist TM, Peerlings J, de Jong EEC, van Timmeren J, et al. Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol. 2017;14(12):749–62. Epub 2017/10/05. doi: 10.1038/nrclinonc.2017.141 . PubMed PMID: 28975929. Pan S, Ding Z, Zhang L, Ruan M, Shan Y, Deng M, et al. A Nomogram Combined Radiomic and Semantic Features as Imaging Biomarker for Classification of Ovarian Cystadenomas. Frontiers in oncology. 2020;10(895). doi: 10.3389/fonc.2020.00895 . Yuan M, Zhang YD, Pu XH, Zhong Y, Li H, Wu JF, et al. Comparison of a radiomic biomarker with volumetric analysis for decoding tumour phenotypes of lung adenocarcinoma with different disease-specific survival. Eur Radiol. 2017;27(11):4857–65. Epub 2017/05/20. doi: 10.1007/s00330-017-4855-3 . PubMed PMID: 28523350. Zheng X, Shao J, Zhou L, Wang L, Ge Y, Wang G, et al. A Comprehensive Nomogram Combining CT Imaging with Clinical Features for Prediction of Lymph Node Metastasis in Stage I-IIIB Non-small Cell Lung Cancer. Ther Innov Regul Sci. 2022;56(1):155–67. Epub 2021/10/27. doi: 10.1007/s43441-021-00345-1 . PubMed PMID: 34699046. Lu X, Li M, Zhang H, Hua S, Meng F, Yang H, et al. A novel radiomic nomogram for predicting epidermal growth factor receptor mutation in peripheral lung adenocarcinoma. Phys Med Biol. 2020;65(5):055012. Epub 2020/01/25. doi: 10.1088/1361-6560/ab6f98 . PubMed PMID: 31978901. Cuocolo R, Stanzione A, Faletti R, Gatti M, Calleris G, Fornari A, et al. MRI index lesion radiomics and machine learning for detection of extraprostatic extension of disease: a multicenter study. European radiology. 2021. doi: 10.1007/s00330-021-07856-3 . PubMed PMID: 33792737. Weng Q, Zhou L, Wang H, Hui J, Chen M, Pang P, et al. A radiomics model for determining the invasiveness of solitary pulmonary nodules that manifest as part-solid nodules. Clinical radiology. 2019;74(12):933–43. doi: 10.1016/j.crad.2019.07.026. PubMed PMID: 31521324. Xie C, Li Q. A simple nomogram for predicting early complications in patients after primary knee arthroplasty. The Knee. 2020;27(2):518–26. doi: 10.1016/j.knee.2019.11.015. PubMed PMID: 31926676. Liu J, Huang X, Chen S, Wu G, Xie W, Franco J, et al. Nomogram based on clinical characteristics for preoperative prediction of perineural invasion in gastric cancer. The Journal of international medical research. 2020;48(1):300060519895131. doi: 10.1177/0300060519895131. PubMed PMID: 31939330. Xie W, Liu J, Huang X, Wu G, Jeen F, Chen S, et al. A nomogram to predict vascular invasion before resection of colorectal cancer. Oncology letters. 2019;18(6):5785–92. doi: 10.3892/ol.2019.10937 . PubMed PMID: 31788051. Park C, Lee I, Jang S, Lee J. Factors affecting tumor recurrence after curative surgery for NSCLC: impacts of lymphovascular invasion on early tumor recurrence. Journal of thoracic disease. 2014;6(10):1420–8. doi: 10.3978/j.issn.2072-1439. 2014.09.31. PubMed PMID: 25364519. Dighe S, Purkayastha S, Swift I, Tekkis P, Darzi A, A'Hern R, et al. Diagnostic precision of CT in local staging of colon cancers: a meta-analysis. Clinical radiology. 2010;65(9):708–19. doi: 10.1016/j.crad.2010.01.024. PubMed PMID: 20696298. Wang H, Schabath M, Liu Y, Stringfield O, Balagurunathan Y, Heine J, et al. Association Between Computed Tomographic Features and Kirsten Rat Sarcoma Viral Oncogene Mutations in Patients With Stage I Lung Adenocarcinoma and Their Prognostic Value. Clinical lung cancer. 2016;17(4):271–8. doi: 10.1016/j.cllc.2015.11.002. PubMed PMID: 26712103. Kattan M. Nomograms are difficult to beat. European urology. 2008;53(4):671-2. doi: 10.1016/j.eururo.2007.12.010 . PubMed PMID: 18093723. Additional Declarations No competing interests reported. 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-1840484","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":120102633,"identity":"f36fc6d9-5086-4424-919c-cdd4bec30856","order_by":0,"name":"Chen Huang","email":"","orcid":"","institution":"West China Hospital, Sichuan University, China, Chengdu, CN","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Chen","middleName":"","lastName":"Huang","suffix":""},{"id":120102636,"identity":"d3520ba0-051c-4e90-b9f7-862f8ad52653","order_by":1,"name":"Yubin Cao","email":"","orcid":"","institution":"West China Hospital, Sichuan University, China, Chengdu, CN","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yubin","middleName":"","lastName":"Cao","suffix":""},{"id":120102638,"identity":"4cbb2a41-b073-4e93-9347-f46881331888","order_by":2,"name":"Lei Zhu","email":"","orcid":"","institution":"West China Hospital, Sichuan University, China, Chengdu, CN","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Zhu","suffix":""},{"id":120102640,"identity":"d8d81d79-eb57-4602-bb95-c23f010fdac3","order_by":3,"name":"Qunli Xiong","email":"","orcid":"","institution":"West China Hospital, Sichuan University, China, Chengdu, CN","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Qunli","middleName":"","lastName":"Xiong","suffix":""},{"id":120102641,"identity":"c702e695-e36c-42a0-9ea0-54e3f9ddd980","order_by":4,"name":"Qin Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAu0lEQVRIiWNgGAWjYDACCTBpA2MQryWNdC2HSdAiP7v52cMvZecT+2e3P/zMw2CXR1AL45xj5sYy524nzrhzxliahyG5mKAWZokEM2nJttu5GyRy2Jh5GA4kNhDSwiaR/g2o5RxQS/oz4rTwSOSYSX5sOwDUkmBGnBYJiZwyaYZzyfUzbuQYS84xSCasRX5G+jbJH2V2xvwz0h9+eFNhR1gLCDDzsMGYBsSoBwLGH2yEFY2CUTAKRsEIBgBCNTgeDoPKEAAAAABJRU5ErkJggg==","orcid":"","institution":"West China School of Medicine, Department of Postgraduate students, Chengdu, CN","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Qin","middleName":"","lastName":"Zhang","suffix":""}],"badges":[],"createdAt":"2022-07-09 01:44:07","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-1840484/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-1840484/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":24046291,"identity":"8f7e5d8a-8f25-4471-9b34-4baecb03098f","added_by":"auto","created_at":"2022-07-19 16:15:44","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":472752,"visible":true,"origin":"","legend":"\u003cp\u003eA flowchart of the study design.\u003cstrong\u003e (A)\u003c/strong\u003e patient recruitment. \u003cstrong\u003e(B)\u003c/strong\u003e radiomics implementation in the study\u003c/p\u003e","description":"","filename":"figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/6a71aec637610ba90f7044f9.png"},{"id":24045550,"identity":"bdfb394e-3184-48e6-9509-56825f09cc18","added_by":"auto","created_at":"2022-07-19 16:10:44","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":98553,"visible":true,"origin":"","legend":"\u003cp\u003eIn the stage of feature selection, we adopt a binary logistic regression model of minimum absolute shrinkage and selection operator (LASSO). \u003cstrong\u003e(A)\u003c/strong\u003e Selecting the best parameters (λ) through minimum standards and tenfold cross-validation. Draw the relationship between partial likelihood deviation (binomial deviation) curve and log (λ), and use the minimum standard and one standard error (SE) (1-SE standard) of the minimum standard to draw the virtual vertical line at the optimal value.\u003cstrong\u003e (B)\u003c/strong\u003e Tuning parameter (k) in the LASSO model is selected for ten cross validations.\u003c/p\u003e","description":"","filename":"figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/887b52263f31e781b4dde4a8.png"},{"id":24046621,"identity":"dbbf8590-1645-4cef-9046-5e45d91f2df5","added_by":"auto","created_at":"2022-07-19 16:20:44","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":157546,"visible":true,"origin":"","legend":"\u003cp\u003eNomogram for predicting the survival rates of EGFR-mutated NSCLC patients.\u0026nbsp;\u003c/p\u003e","description":"","filename":"figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/46567b15b2b7806cdfaa1ef1.png"},{"id":24045555,"identity":"edaf52e6-0e9f-49d8-a0c9-32b7a870c408","added_by":"auto","created_at":"2022-07-19 16:10:44","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":50296,"visible":true,"origin":"","legend":"\u003cp\u003eRadiomic nomogram establishment and validation. A nomogram that combines 6 important clinical factors and imaging risk scores to predict the three-year and five-year survival rates of EGFR-mutated NSCLC patients.\u003c/p\u003e","description":"","filename":"figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/6a4fd1196de2a81bd5c55585.png"},{"id":24045552,"identity":"61279dc6-a41d-47e5-8bd4-fa16c02a0e39","added_by":"auto","created_at":"2022-07-19 16:10:44","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":112677,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curve of the training cohort. Three-year \u003cstrong\u003e(A) /\u003c/strong\u003e five-year\u003cstrong\u003e (B)\u003c/strong\u003e survival rates of EGFR-mutated NSCLC patients.\u003c/p\u003e","description":"","filename":"figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/bfcdd8d9667ab652705d75be.png"},{"id":24045556,"identity":"c122a430-1f61-4e06-88ca-87eea17e058f","added_by":"auto","created_at":"2022-07-19 16:10:44","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":50221,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver operating characteristic (ROC) curve. ROC curve is usually used to evaluate the actual clinical utility of the prediction model. The area under the ROC curve (AUC) is a global measure to test whether there is a predictive ability to distinguish specific conditions. AUC of 0.5 represents the test without discrimination, while AUC of 1.0 represents the test with perfect discrimination.\u003c/p\u003e","description":"","filename":"figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/6fdfb6acf083045a438b9d03.png"},{"id":24045553,"identity":"6da98f9f-7f82-4569-a7d2-ce567b86c816","added_by":"auto","created_at":"2022-07-19 16:10:44","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":58781,"visible":true,"origin":"","legend":"\u003cp\u003eThe decision curve analysis (DCA) for the radiomics nomogram. The decision curve shows that if the threshold probability of a doctor is 10%, our radiomics nomogram to predict the three-year \u003cstrong\u003e(A)\u003c/strong\u003e and five-year survival rates\u003cstrong\u003e (B)\u003c/strong\u003e of EGFR-mutated NSCLC patients adds more clinical benefit than either the treat-all-patients scheme or the treat-none scheme.\u003c/p\u003e","description":"","filename":"figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/333bda41ee4da4f6495e07ae.png"},{"id":25854775,"identity":"b0279d37-6ffd-4e1f-8d03-edf8ee8e56b6","added_by":"auto","created_at":"2022-08-30 16:59:19","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":996480,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-1840484/v1/842aa6d0-833f-4843-a119-c64a917fb15c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Computer Tomography Radiomics-based model to Predict Survival of Patients with EGFR‑Mutated Non‑small‑cell Lung Cancer","fulltext":[{"header":"Introduction","content":"\u003cp\u003eLung cancer is currently the most frequently diagnosed cancer in the world and the prominent reason for tumor-related deaths [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Lung cancer usually consists of two subtypes: non-small-cell lung cancer (NSCLC), which is estimated to be 85% of lung tumors, and small cell lung cancer (SCLC) [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eEpidermal growth factor receptor (EGFR) is a 170 kDa transmembrane glycoprotein and a member of the HER/ErbB family, which is distributed extensively on the cell membrane in human tissues. The binding of EGFR to the EGFR ligand induces receptor dimerization and causes a conformational change that triggers multiple signal transduction pathways, such as the phosphoinositide 3-kinase/threonine-protein kinase (PI3K) pathway and mitogen-activated protein kinase (MARK) pathway [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. \u003cem\u003eEGFR\u003c/em\u003e is one of the most significantly mutated genes driving lung cancer, and approximately 50% of Asian patients with NSCLC possess \u003cem\u003eEGFR\u003c/em\u003e gene mutations [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. The most common activating mutations in the \u003cem\u003eEGFR\u003c/em\u003e gene are a short deletion in exon 19 and a missense mutation (L858R) in exon 21 [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. The overexpression of EGFR is the main reason that causes tumor cells to proliferate uncontrollably, to escape programmed death, to enhance tumor cell migration and to promote metastasis [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. The activation of EGFR also plays a significant role in the resistance of tumor cells to chemotherapy and radiotherapy [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. EGFR tyrosine kinase inhibitors (EGFR-TKIs) are established as the first line of treatment for EGFR‐mutated NSCLC patients; however, approximately 30% of these patients are clinically insensitive to EGFR-TKIs [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Thus, an individualized method is needed to develop precision therapy strategy and to predict prognosis for \u003cem\u003eEGFR\u003c/em\u003e-mutated NSCLC patients [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAt present, although the TNM staging system has been widely used to predict the prognosis of cancer, it is difficult to accurately predict the outcome merely based on the TNM staging system. Even within the same TNM stage, patients may have survival differences [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Therefore, to evaluate the prognosis of \u003cem\u003eEGFR\u003c/em\u003e-mutated non-small-cell lung cancer(NSCLC) patients and accurately target the patient subgroups with best potential outcome greatest, a simpler and more effective model based on readily available laboratory parameters is needed in clinical practice.\u003c/p\u003e \u003cp\u003eRadiomics is a common tool in oncologic clinical workflow, allowing for prediction of the treatment response and clinical outcome [\u003cspan additionalcitationids=\"CR20\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Previous studies have used radiomics combined with the nomogram model to identify imaging biomarkers for the classification of serous and mucinous types of ovarian cystadenomas [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. In distinguishing aggressive lung adenocarcinoma from AIS/MIA, this method also showed a better diagnostic performance than 3D volume analysis with an accuracy rate of 81.2%[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Also, researchers have attempted to include imaging omics and multiple clinical factors to effectively predict the preoperative LNM of NSCLC patients, which greatly saved time in clinical practice [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eInspired by former explorations, we hypothesized that computed tomographic (CT) imaging omics with more imaging omics features might provide some valuable tumor heterogeneity information to predict the future prognosis of \u003cem\u003eEGFR\u003c/em\u003e-mutated NSCLC patients at the time of initial diagnosis. This study presents a radiologically based analysis to assess the prognosis of NSCLC patients with EGFR mutations, with a view to pinpointing the subgroup of patients with the better potential outcomes.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePatients selection\u003c/h2\u003e \u003cp\u003eThis retrospective and observational study was conducted in West China Hospital, a university affiliated medical center with 4,500 beds (Chengdu, Sichuan Province, China). The study population consisted of 173 lung cancer patients who were treated in West China Hospital of Sichuan University from January 2009 to December 2018 and received EGFR gene mutation detection. After screening, 162 cases were finally included. This study was approved by the Ethics Committee of West China Hospital of Sichuan University.\u003c/p\u003e \u003cp\u003eOur inclusion criteria were as follows: (1) initial pathological diagnosis of primary NSCLC; (2) clear EGFR detection results; and (3) chest CT examination with a layer thickness of 5 mm performed at West China Hospital of Sichuan University within one month before diagnosis for imaging modelling analysis. The exclusion criteria were as follows: (1) chest local treatment or systemic treatment for other malignant tumours before CT examination; (2) massive pleural effusion completely masking the tumour; and (3) multiple-source lung cancer and inconsistent EGFR detection results. The inclusion and exclusion criteria are described in the flowchart (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eBasic clinical pathological information\u003c/h2\u003e \u003cp\u003eFor the included patients, the following indicators were collected from the hospital electronic clinical database: age (years), sex (male vs female), smoking history (with vs without), smoking amount (package year), family history of malignant tumours (with vs without), histological subtype (adenocarcinoma, squamous cell carcinoma, other), and tumour stage (I-III vs IV). For cases missing information regarding tumour location and tumour stage in the clinical reports, tumour location and clinical stage were evaluated by an imaging system. Clinical staging was performed according to the eighth edition of the International Association for Lung Cancer Research Tumour-Node-Metastasis (TNM) staging {Wilfried, 2015 #1037}.\u003c/p\u003e \u003cp\u003eEGFR detection methods referred in our research included single-gene sequencing and multigene sequencing. Single-gene sequencing was performed using amplification refractory mutation system-polymerase chain reaction (ARMS-PCR); multigene sequencing was performed using next-generation sequencing (NGS).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eCT Acquisition\u003c/h2\u003e \u003cp\u003eIQQA software (IQQA-Chest, EDDA Technology, Princeton Junction, NJ, USA) was used one by one from the hospital PACS system to extract eligible chest CT images in a DICOM format file. Chest CT detection equipment included Siemens, Philips Brilliance Big Bore, and GE Discovery CT750 HD. All patients underwent spiral CT scans from the apex to the bottom of the lung under maximum inspiratory state. For CT scanners, tube voltage was set at 120 kV, tube current was set at 200\u0026ndash;500 mA, rotation time was set at 0.4\u0026ndash;0.7 s, and matrix of 512 \u0026times; 512 pixel was produced. For the Siemens scanner, the convolution kernels were B31f and B30f; for the GE scanner, the reconstructed image of the STANDARD kernel was selected.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eTumor Segmentation and ROI delineation\u003c/h2\u003e \u003cp\u003eImaging omics feature extraction requires accurate delineation of tumors. Therefore, patients with unclear tumor boundaries will not be able to be included in the establishment of an imaging omics model. Tumors that cannot be accurately delineated include but are not limited to the following situations: obstructive pneumonia, unclear boundaries with pulmonary hilus, and pleural effusion.\u003c/p\u003e \u003cp\u003eIn our research, the precise delineation of tumors was realized by two physicians with clinical experience independently through the manual operation of ITK-SNAP software (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e\u003ca href=\"http://www.itksnap.org\" target=\"_blank\"\u003ewww.itksnap.org\u003c/a\u003e\u003c/span\u003e\u003cspan address=\"http://www.itksnap.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). First, the slice of CT containing the tumor was determined, and the outline was carried out along the edge of the tumor. Then, the outline of the tumor was filled, and finally, all the slices of CT containing the tumor were plotted layer by layer. The red region in the figure was the precisely delineated region of interest for the tumor. Three-dimensional tumors could be generated in three-dimensional space by drawing the region of interest of the tumor layer by layer.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eExtraction of Radiomics Features\u003c/h2\u003e \u003cp\u003eThe Pyradiomics toolkit (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/Radiomics/pyradiomics\u003c/span\u003e\u003cspan address=\"https://github.com/Radiomics/pyradiomics\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to extract features. A total of 1353 radiomic features were extracted from ROIs. First-order, GLCM, GLRLM, GLSZM, GLDM and shape were extracted. Of all the features, there were 107 original shape features, 465 features using Gaussian (LoG, σ:1,2,3,4,5) Laplacian filtering and 744 wavelet (wavelet transform: LLL, LLH, LHH, LHL, HHH, HHL, HLH, HLL) features. The texture features mainly reflected the gray distribution of the image, and the wavelet features reflected the different frequency ranges of the original image. The first-order features described the intensity distribution of pixels or voxels in the region of interest, and the shape features described the shape and volume of the region of interest.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eFeature selection\u003c/h2\u003e \u003cp\u003eThe best radiomics features were picked by a three-step feature selection process. First, to ensure reproducibility, CT images of 30 patients were stochastically selected and resegmented by two independent doctors at an interval of 2 months. To assess the reproducibility of the extracted features, 30 patients were randomly selected and quantified by two oncologists. To obtain the robustness of radiomics features without the influence of tumor segmentation variability (ICC\u0026thinsp;\u0026gt;\u0026thinsp;0.75), the intergroup and intraclass correlation coefficients (ICCs) were calculated [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Then, to construct a nomogram, the variables were observed by drawing plots, and univariate analysis was performed to select definite characteristics. The least absolute shrinkage and selection operator (LASSO) regression model is a penalizing regression method that estimates the regression coefficients by maximizing the log-likelihood function while restraining the sum of the absolute values of the regression coefficients, which could avoid overfitting and was shown to be near-minimax optimal to obtain the outcome-predictive features from patients who underwent (chemo)radiotherapy [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. We used the \u0026ldquo;glmnet\u0026rdquo; package of R software, which is a powerful tool to apply the selected optimal variables, especially high-dimensional data [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Then, the characteristics were selected by a minimum λ. Finally, the prediction model was constructed by LASSO to perform multivariable logistic regression.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eNomogram establishment and validation\u003c/h2\u003e \u003cp\u003eAfter the above variable selection, the \u0026ldquo;rms\u0026rdquo; package of R software was used to generate the nomogram [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. The receiver operating characteristic (ROC) curve includes the false positive rate on its horizontal axis and the true positive rate on its vertical axis. To evaluate the discrimination ability, the nomogram was assessed by the area under the curve (AUC). Furthermore, we used the calibration plot to evaluate the goodness of fit between the observed values and the predicted values.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eClinical Use\u003c/h2\u003e \u003cp\u003eTo evaluate the clinical application value of the nomotu model and radiomics characteristics, we carried out decision curve analysis (DCA) to evaluate the net benefits of various threshold risks.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eIn this study, continuous variables were presented as median values with interquartile ranges (IQRs), while categorical variables are presented as percentages. All statistical analyses were performed using R software (version 3.4.0; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e\u003ca href=\"http://www.itksnap.org\" target=\"_blank\"\u003ewww.r-project.org\u003c/a\u003e\u003c/span\u003e\u003cspan address=\"http://www.r-project.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). All tests were two-sided, and values of P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered statistically significant.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv class=\"Section2\" id=\"Sec13\"\u003e\n \u003ch2\u003ePatient clinical features\u003c/h2\u003e\n \u003cp\u003eThere were 173 lung cancer patients who were treated at West China Hospital of Sichuan University from January 2009 to December 2018 and had EGFR gene mutations. Based on the inclusion/exclusion criteria, 162 patients were enrolled in this study. The former included 74(45.7%) males and 88 (54.3%) females. Median age was 59 years (IQR, 50\u0026ndash;65). Their characteristics are presented in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n \u003ctable border=\"1\" cellpadding=\"0\" cellspacing=\"0\"\u003e\n \u003ccaption\u003e\n \u003cp\u003eTable 1\u003c/p\u003e\n \u003cp\u003eParticipant characteristics of participants\u003c/p\u003e\n \u003c/caption\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"80.41044776119404%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"19.58955223880597%\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal (N=162)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"80.41044776119404%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGender\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Male\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Female\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"19.58955223880597%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e45.7% (74/162)\u003c/p\u003e\n \u003cp\u003e54.3% (88/162)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"80.41044776119404%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Mean (SD)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Median (IQR)\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eSmoking history\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Current or former\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Never\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003ePathological type\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Squamous cell carcinoma\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Neuroendocrine carcinoma\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Adenocarcinoma\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Adenosquamous carcinoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"19.58955223880597%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e58.19 (11.10)\u003c/p\u003e\n \u003cp\u003e59 (50, 65)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e71.6% (116/162)\u003c/p\u003e\n \u003cp\u003e28.4% (46/162)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e41.4% (67/162)\u003c/p\u003e\n \u003cp\u003e0.6% (1/162)\u003c/p\u003e\n \u003cp\u003e37.0% (60/162)\u003c/p\u003e\n \u003cp\u003e21.0% (34/162)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"80.41044776119404%\"\u003e\n \u003cp\u003e\u003cstrong\u003eClinical stage\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;I\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;II\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;III\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;IV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"19.58955223880597%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e11.7% (19/162)\u003c/p\u003e\n \u003cp\u003e6.8% (11/162)\u003c/p\u003e\n \u003cp\u003e32.5% (53/162)\u003c/p\u003e\n \u003cp\u003e48.8% (79/162)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"80.41044776119404%\"\u003e\n \u003cp\u003e\u003cstrong\u003eTNM stage\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;T (primary tumor)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; T1\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; T2\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; T3\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; T4\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003cstrong\u003eN (regional lymph nodes)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; N0\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; N1\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; N2\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; N3\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; Nx\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003cstrong\u003eM (distant metastases)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u0026nbsp;\u003c/strong\u003eM0\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; M1\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; Mx\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"19.58955223880597%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e5.6% (9/162)\u003c/p\u003e\n \u003cp\u003e41.4% (67/162)\u003c/p\u003e\n \u003cp\u003e25.3%(31/162)\u003c/p\u003e\n \u003cp\u003e34.6% (56/162)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e22.2% (36/162)\u003c/p\u003e\n \u003cp\u003e8.0% (13/162)\u003c/p\u003e\n \u003cp\u003e46.9% (76/162)\u003c/p\u003e\n \u003cp\u003e20.4% (33/162)\u003c/p\u003e\n \u003cp\u003e2.5% (4/162)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e51.2% (83/162)\u003c/p\u003e\n \u003cp\u003e48.1% (78/162)\u003c/p\u003e\n \u003cp\u003e0.6% (1/162)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"80.41044776119404%\"\u003e\n \u003cp\u003e\u003cstrong\u003eTargeted therapy\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Current or former\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Never\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"19.58955223880597%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e51.2% (83/162)\u003c/p\u003e\n \u003cp\u003e48.8% (79/162)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec14\"\u003e\n \u003ch2\u003eFeature selection\u003c/h2\u003e\n \u003cp\u003eThe univariate analysis to select definite characteristics is shown in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e. Through univariate analysis, we selected 8 characteristics for the LASSO regression model. The two dashed lines in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e represent two special \u0026lambda; values: lambda.min and lambda.lse. lambda.min represents the \u0026lambda; value when the error is the smallest, which can obtain the least target parameter; lambda.lse represents the \u0026lambda; value of the simplest model within a variance range of lambda.min. We selected an optimal lambda of 5 with the smallest binomial deviance, and the initial variables were reduced to five potential predictors under penalizing conditions (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eA).\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable border=\"1\" id=\"Tab2\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eUnivariate analysis to select definite characteristics.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eHR\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ewavelet-LHL_firstorder_Kurtosis\u003c/p\u003e\n \u003cp\u003ewavelet-LHH_firstorder_Kurtosis\u003c/p\u003e\n \u003cp\u003ewavelet-HHL_firstorder_Kurtosis\u003c/p\u003e\n \u003cp\u003ewavelet-HHL_ngtdm_Contrast\u003c/p\u003e\n \u003cp\u003ewavelet-HHH_firstorder_Kurtosis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.04(1.01\u0026ndash;1.07)\u003c/p\u003e\n \u003cp\u003e1.06(1.01\u0026ndash;1.12)\u003c/p\u003e\n \u003cp\u003e1.04(1.02\u0026ndash;1.06)\u003c/p\u003e\n \u003cp\u003e0.00(0.00-0.43)\u003c/p\u003e\n \u003cp\u003e1.03(1.01\u0026ndash;1.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003cp\u003e0.026\u003c/p\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003cp\u003e0.041\u003c/p\u003e\n \u003cp\u003e0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ewavelet-HHH_firstorder_Skewness\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.25(0.07\u0026ndash;0.85)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWavelet-HHH_gldm_DependenceNonUniformityNormalized\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.00(0.00-0.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.025\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ewavelet-HHH_gldm_DependenceVariance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.13(1.02\u0026ndash;1.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.025\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eHR, hazard ratio.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec15\"\u003e\n \u003ch2\u003eConstruction of the risk score model\u003c/h2\u003e\n \u003cp\u003eWith the median risk score as the cut-off value, we divided the 162 patients into a high-risk group and a low-risk group. The Kaplan\u0026ndash;Meier curve show that the OS of the low-risk group was significantly better than that of the high-risk group (p\u0026thinsp;\u0026lt;\u0026thinsp;0.00) (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eA). The risk curve and scatter plot show that the risk factors and mortality of patients in the high-risk group were higher than those in the low-risk group (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eB, C).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec16\"\u003e\n \u003ch2\u003eNomogram establishment and validation\u003c/h2\u003e\n \u003cp\u003eAccording to the multivariable logistic regression analysis, the statistically significant features that were retained included age, sex, Stage_T, Stage_N, Stage_M, clinical stage, and imaging risk score. We then developed a prediction model that incorporated the above independent features and presented it as a nomogram to predict the three-year and five-year survival rates of EGFR-mutated NSCLC patients. (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003eThe calibration curves shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e illustrate a good consistency between the nomogram-predicted three-year and five-year survival rates of EGFR-mutated NSCLC patients. Moreover, the AUC-ROC of three-year \u003cstrong\u003e(A)\u003c/strong\u003e OS was 0.715, while five-year \u003cstrong\u003e(B)\u003c/strong\u003e OS was 0.705, as is displayed in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e. This indicates a favourable discrimination and demonstrates a good fit for the purpose of targeting clinical services and predicting patient outcomes.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec17\"\u003e\n \u003ch2\u003eClinical Use\u003c/h2\u003e\n \u003cp\u003eThe decision curve analysis for the radiomics nomogram and that for the model with integrated histologic grade is presented in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003e. The decision curve shows that if the threshold probability of a doctor is 10%, our radiomics nomogram to predict the three-year (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eA) and five-year survival rates (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eB) of EGFR-mutated NSCLC patients adds more clinical benefit than either the treat-all-patients scheme or the treat-none scheme. Within this range, net benefit was comparable.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn our study, we selected 20 features by univariate analysis and further identified 7 features by multivariable logistic regression analysis. The final seven characteristics selected included six clinically important factors including age, gender, stage_T, stage_N, stage_M and clinical stage as well as one imaging risk score. Then, we developed a prediction model combining the above 7 independent features and presented it as a nomogram to predict three-year and five-year survival rates of EGFR-mutated NSCLC patients. This study explored the correlation between tumor imaging omics characteristics and patient prognosis through in-depth mining of tumor CT imaging information and constructed a stable and effective prognostic analysis model to predict the prognosis and survival of NSCLC patients. The experimental results proved that the imaging omics characteristics of tumors could be utilized to quantitatively evaluate the heterogeneity of tumors. The prognostic analysis model based on tumor imaging omics can effectively predict the prognostic survival time range of NSCLC patients and can assist doctors in analysing, diagnosing and predicting the prognosis of patients.\u003c/p\u003e \u003cp\u003ePrevious studies have attempted to explore related risk factors for the prognosis of NSCLC from clinical features, such as pathological staging [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. However, some risk factors (such as the depth of submucosal invasion, lymphatic invasion, and vascular invasion) are related to histopathological characteristics, so they cannot be obtained before surgery, which might result in a delay in making treatment decisions. In clinical practice, CT is the most commonly used preoperative imaging method for NSCLC. However, the limitation of traditional CT examination is that it entails much professional training to interpret and lacks direct clues pointing to prognosis [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Wang et al. extracted the burr sign, lobular sign, ground glass nodule, cavity sign, bronchial inflation sign, pleural depression sign, and pleural adhesion sign of 79 NSCLC patients with stage Ⅰ CT images [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. The survival analysis found that the pleural adhesion was an independent prognostic factor of stage I NSCLC. However, in the former attempts exploring the association of clinical features and prognosis in NSCLC, prognostic factors were screened out, but whether they had real clinical implications was still under doubt for the following reasons (1) simple clinical characteristics cannot be used to assess risk factors related to histopathological characteristics. (2) The obtained prognostic factors can only explore the prognosis of the patient. However, they cannot effectively help doctors predict the prognosis of NSCLC patients. To help resolve this problem, in this study, we retrospectively collected clinical data and imaging data of EGFR-mutant NSCLC patients and extracted the CT imaging features of their first diagnosis. Then, the predictive value of several known clinical factors and imaging features that affect the prognosis of predicting the survival of such patients was evaluated. Based on these results, we established a nomogram combining 6 important clinical factors and imaging risk scores to predict the three-year and five-year survival rates of EGFR-mutated NSCLC patients. If developed, validated, and used correctly, our nomogram can provide important information about patient care [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThere are also some limitations in this study. First, this retrospective study was conducted in a single center with a limited sample size and lacked external verification, resulting from the difficulty of obtaining CT images. But given our carefully developed research design, the authenticity and validity of our research could be ensured. Second, the model still requires more biomarkers to improve its accuracy. To conclude, in order to further our exploration in enhancing this model, it is necessary to expand the sample size and to include more relevant clinical data and a wider range of biomarkers.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eUtilizing chest CT imaging and clinicopathological features, included age, sex, Stage_T, Stage_N, Stage_M, clinical stage, and imaging risk score, we constructed and validated a nomogram for predicting an individualized prediction of survival for patients with EGFR‑mutated NSCLC.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was approved by the ethics committee of West China Hospital [2019(195)] written informed consent was obtained from all patients for research purposes. All methods were carried out in accordance with relevant guidelines and regulations or declaration of helsinki.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eN.A.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used or analysed during the current study are available from the corresponding author on reasonable request.\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\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eN.A.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eQZ contributed to the guidance of this article and given the final approval of the version to be published; CH drafted the manuscript and YBC prepared the figures. LZ and QLX collected the datas. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eN.A.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBarta JA, Powell CA, Wisnivesky JP. Global Epidemiology of Lung Cancer. Ann Glob Health. 2019;85(1). Epub 2019/02/12. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5334/aogh.2419\u003c/span\u003e\u003cspan address=\"10.5334/aogh.2419\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 30741509; PubMed Central PMCID: PMCPMC6724220.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHirsch FR, Scagliotti GV, Mulshine JL, Kwon R, Curran WJ, Jr., Wu YL, et al. Lung cancer: current therapies and new targeted treatments. Lancet (London, England). 2017;389(10066):299\u0026ndash;311. Epub 2016/08/31. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s0140-6736(16)30958-8\u003c/span\u003e\u003cspan address=\"10.1016/s0140-6736(16)30958-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 27574741.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKeith RL, Miller YE. Lung cancer chemoprevention: current status and future prospects. Nat Rev Clin Oncol. 2013;10(6):334\u0026ndash;43. Epub 2013/05/22. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrclinonc.2013.64\u003c/span\u003e\u003cspan address=\"10.1038/nrclinonc.2013.64\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 23689750; PubMed Central PMCID: PMCPMC4687734.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHess LM, Kern DM, Carter GC, Winfree K, Wang L, Sontag A, et al. Real-World Treatment Sequences and Outcomes Among Patients With Non-Small Cell Lung Cancer (RESOUNDS) in the United States: Study Protocol. JMIR Res Protoc. 2017;6(10):e195. Epub 2017/10/13. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2196/resprot.7750\u003c/span\u003e\u003cspan address=\"10.2196/resprot.7750\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 29021129; PubMed Central PMCID: PMCPMC5658639.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoto Y, Tanai C, Yoh K, Hosomi Y, Sakai H, Kato T, et al. Continuing EGFR-TKI beyond radiological progression in patients with advanced or recurrent, EGFR mutation-positive non-small-cell lung cancer: an observational study. ESMO Open. 2017;2(4):e000214. Epub 2017/10/12. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/esmoopen-2017-000214\u003c/span\u003e\u003cspan address=\"10.1136/esmoopen-2017-000214\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 29018574; PubMed Central PMCID: PMCPMC5604715.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang J, Sun P, Wang Q, Zhang P, Wang Y, Zi C, et al. (-)-Epigallocatechin-3-gallate derivatives combined with cisplatin exhibit synergistic inhibitory effects on non-small-cell lung cancer cells. Cancer Cell Int. 2019;19:266. Epub 2019/10/23. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12935-019-0981-0\u003c/span\u003e\u003cspan address=\"10.1186/s12935-019-0981-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 31636509; PubMed Central PMCID: PMCPMC6791019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShi Y, Li J, Zhang S, Wang M, Yang S, Li N, et al. Molecular Epidemiology of EGFR Mutations in Asian Patients with Advanced Non-Small-Cell Lung Cancer of Adenocarcinoma Histology - Mainland China Subset Analysis of the PIONEER study. PLoS One. 2015;10(11):e0143515. Epub 2015/11/26. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0143515\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0143515\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 26599344; PubMed Central PMCID: PMCPMC4657882.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTsao AS, Scagliotti GV, Bunn PA, Jr., Carbone DP, Warren GW, Bai C, et al. Scientific Advances in Lung Cancer 2015. J Thorac Oncol. 2016;11(5):613\u0026ndash;38. Epub 2016/03/26. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jtho.2016.03.012\u003c/span\u003e\u003cspan address=\"10.1016/j.jtho.2016.03.012\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 27013409.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCosta DB. Kinase inhibitor-responsive genotypes in EGFR mutated lung adenocarcinomas: moving past common point mutations or indels into uncommon kinase domain duplications and rearrangements. Transl Lung Cancer Res. 2016;5(3):331\u0026ndash;7. Epub 2016/07/15. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21037/tlcr.2016.06.04\u003c/span\u003e\u003cspan address=\"10.21037/tlcr.2016.06.04\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 27413714; PubMed Central PMCID: PMCPMC4931120.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDong Z, Zhang J, Liu S, Su J, Zhang C, Xie Z, et al. EGFR mutation correlates with uninflamed phenotype and weak immunogenicity, causing impaired response to PD-1 blockade in non-small cell lung cancer. Oncoimmunology. 2017;6(11):e1356145. doi: 10.1080/2162402x.2017.1356145. PubMed PMID: 29147605.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSingh D, Attri BK, Gill RK, Bariwal J. Review on EGFR Inhibitors: Critical Updates. Mini Rev Med Chem. 2016;16(14):1134\u0026ndash;66. Epub 2016/03/22. doi: 10.2174/1389557516666160321114917. PubMed PMID: 26996617.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTan DS, Yom SS, Tsao MS, Pass HI, Kelly K, Peled N, et al. The International Association for the Study of Lung Cancer Consensus Statement on Optimizing Management of EGFR Mutation-Positive Non-Small Cell Lung Cancer: Status in 2016. J Thorac Oncol. 2016;11(7):946\u0026ndash;63. Epub 2016/05/28. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jtho.2016.05.008\u003c/span\u003e\u003cspan address=\"10.1016/j.jtho.2016.05.008\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 27229180.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTaniguchi H, Yamada T, Wang R, Tanimura K, Adachi Y, Nishiyama A, et al. AXL confers intrinsic resistance to osimertinib and advances the emergence of tolerant cells. Nat Commun. 2019;10(1):259. Epub 2019/01/18. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41467-018-08074-0\u003c/span\u003e\u003cspan address=\"10.1038/s41467-018-08074-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 30651547; PubMed Central PMCID: PMCPMC6335418 Boehringer-Ingelheim, and has received speaking honoraria from AstraZeneca, Chugai Pharma, and Boehringer-Ingelheim. The remaining authors declare no competing interests.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang AY, Zhang J, Luo HL, Gao F, Lv YF. Icotinib and whole-brain radiotherapy for the treatment in patients with brain metastases from EGFR-mutant nonsmall cell lung cancer: A retrospective study. Medicine. 2018;97(15):e0312. Epub 2018/04/12. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/md.0000000000010312\u003c/span\u003e\u003cspan address=\"10.1097/md.0000000000010312\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 29642161; PubMed Central PMCID: PMCPMC5908633.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCuneo KC, Nyati MK, Ray D, Lawrence TS. EGFR targeted therapies and radiation: Optimizing efficacy by appropriate drug scheduling and patient selection. Pharmacology \u0026amp; therapeutics. 2015;154:67\u0026ndash;77. Epub 2015/07/25. doi: 10.1016/j.pharmthera.2015.07.002. PubMed PMID: 26205191; PubMed Central PMCID: PMCPMC4570853.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRusso A, Franchina T, Ricciardi GR, Picone A, Ferraro G, Zangh\u0026igrave; M, et al. A decade of EGFR inhibition in EGFR-mutated non small cell lung cancer (NSCLC): Old successes and future perspectives. Oncotarget. 2015;6(29):26814\u0026ndash;25. Epub 2015/08/27. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.18632/oncotarget.4254\u003c/span\u003e\u003cspan address=\"10.18632/oncotarget.4254\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 26308162; PubMed Central PMCID: PMCPMC4694955.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImprota G, Zupa A, Natalicchio MI, Sisinni L, Marinaccio A, Bozza G, et al. Uncommon frame-shift exon 19 EGFR mutations are sensitive to EGFR tyrosine kinase inhibitors in non-small cell lung carcinoma. Med Oncol. 2018;35(3):28. Epub 2018/02/02. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s12032-018-1078-7\u003c/span\u003e\u003cspan address=\"10.1007/s12032-018-1078-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 29387949.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei XL, Zhang DS, He MM, Jin Y, Wang DS, Zhou YX, et al. The predictive value of alkaline phosphatase and lactate dehydrogenase for overall survival in patients with esophageal squamous cell carcinoma. Tumour Biol. 2016;37(2):1879-87. Epub 2015/09/02. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s13277-015-3851-y\u003c/span\u003e\u003cspan address=\"10.1007/s13277-015-3851-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 26323257.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology. 2016;278(2):563\u0026ndash;77. Epub 2015/11/19. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1148/radiol.2015151169\u003c/span\u003e\u003cspan address=\"10.1148/radiol.2015151169\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 26579733; PubMed Central PMCID: PMCPMC4734157.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAerts HJ, Velazquez ER, Leijenaar RT, Parmar C, Grossmann P, Carvalho S, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014;5:4006. Epub 2014/06/04. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/ncomms5006\u003c/span\u003e\u003cspan address=\"10.1038/ncomms5006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 24892406; PubMed Central PMCID: PMCPMC4059926.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLambin P, Leijenaar RTH, Deist TM, Peerlings J, de Jong EEC, van Timmeren J, et al. Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol. 2017;14(12):749\u0026ndash;62. Epub 2017/10/05. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrclinonc.2017.141\u003c/span\u003e\u003cspan address=\"10.1038/nrclinonc.2017.141\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 28975929.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePan S, Ding Z, Zhang L, Ruan M, Shan Y, Deng M, et al. A Nomogram Combined Radiomic and Semantic Features as Imaging Biomarker for Classification of Ovarian Cystadenomas. Frontiers in oncology. 2020;10(895). doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2020.00895\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2020.00895\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYuan M, Zhang YD, Pu XH, Zhong Y, Li H, Wu JF, et al. Comparison of a radiomic biomarker with volumetric analysis for decoding tumour phenotypes of lung adenocarcinoma with different disease-specific survival. Eur Radiol. 2017;27(11):4857\u0026ndash;65. Epub 2017/05/20. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00330-017-4855-3\u003c/span\u003e\u003cspan address=\"10.1007/s00330-017-4855-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 28523350.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng X, Shao J, Zhou L, Wang L, Ge Y, Wang G, et al. A Comprehensive Nomogram Combining CT Imaging with Clinical Features for Prediction of Lymph Node Metastasis in Stage I-IIIB Non-small Cell Lung Cancer. Ther Innov Regul Sci. 2022;56(1):155\u0026ndash;67. Epub 2021/10/27. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s43441-021-00345-1\u003c/span\u003e\u003cspan address=\"10.1007/s43441-021-00345-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 34699046.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu X, Li M, Zhang H, Hua S, Meng F, Yang H, et al. A novel radiomic nomogram for predicting epidermal growth factor receptor mutation in peripheral lung adenocarcinoma. Phys Med Biol. 2020;65(5):055012. Epub 2020/01/25. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1088/1361-6560/ab6f98\u003c/span\u003e\u003cspan address=\"10.1088/1361-6560/ab6f98\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 31978901.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCuocolo R, Stanzione A, Faletti R, Gatti M, Calleris G, Fornari A, et al. MRI index lesion radiomics and machine learning for detection of extraprostatic extension of disease: a multicenter study. European radiology. 2021. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00330-021-07856-3\u003c/span\u003e\u003cspan address=\"10.1007/s00330-021-07856-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 33792737.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeng Q, Zhou L, Wang H, Hui J, Chen M, Pang P, et al. A radiomics model for determining the invasiveness of solitary pulmonary nodules that manifest as part-solid nodules. Clinical radiology. 2019;74(12):933\u0026ndash;43. doi: 10.1016/j.crad.2019.07.026. PubMed PMID: 31521324.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXie C, Li Q. A simple nomogram for predicting early complications in patients after primary knee arthroplasty. The Knee. 2020;27(2):518\u0026ndash;26. doi: 10.1016/j.knee.2019.11.015. PubMed PMID: 31926676.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu J, Huang X, Chen S, Wu G, Xie W, Franco J, et al. Nomogram based on clinical characteristics for preoperative prediction of perineural invasion in gastric cancer. The Journal of international medical research. 2020;48(1):300060519895131. doi: 10.1177/0300060519895131. PubMed PMID: 31939330.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXie W, Liu J, Huang X, Wu G, Jeen F, Chen S, et al. A nomogram to predict vascular invasion before resection of colorectal cancer. Oncology letters. 2019;18(6):5785\u0026ndash;92. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3892/ol.2019.10937\u003c/span\u003e\u003cspan address=\"10.3892/ol.2019.10937\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 31788051.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePark C, Lee I, Jang S, Lee J. Factors affecting tumor recurrence after curative surgery for NSCLC: impacts of lymphovascular invasion on early tumor recurrence. Journal of thoracic disease. 2014;6(10):1420\u0026ndash;8. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3978/j.issn.2072-1439.\u003c/span\u003e\u003cspan address=\"10.3978/j.issn.2072-1439.\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e2014.09.31. PubMed PMID: 25364519.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDighe S, Purkayastha S, Swift I, Tekkis P, Darzi A, A'Hern R, et al. Diagnostic precision of CT in local staging of colon cancers: a meta-analysis. Clinical radiology. 2010;65(9):708\u0026ndash;19. doi: 10.1016/j.crad.2010.01.024. PubMed PMID: 20696298.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang H, Schabath M, Liu Y, Stringfield O, Balagurunathan Y, Heine J, et al. Association Between Computed Tomographic Features and Kirsten Rat Sarcoma Viral Oncogene Mutations in Patients With Stage I Lung Adenocarcinoma and Their Prognostic Value. Clinical lung cancer. 2016;17(4):271\u0026ndash;8. doi: 10.1016/j.cllc.2015.11.002. PubMed PMID: 26712103.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKattan M. Nomograms are difficult to beat. European urology. 2008;53(4):671-2. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.eururo.2007.12.010\u003c/span\u003e\u003cspan address=\"10.1016/j.eururo.2007.12.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 18093723.\u003c/span\u003e\u003c/li\u003e\u003c/ol\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":"EGFR‑Mutated NSCLC, radiomics, prognosis, nomogram","lastPublishedDoi":"10.21203/rs.3.rs-1840484/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-1840484/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eTo evaluate the prognosis of \u003cem\u003eEGFR\u003c/em\u003e-mutated non-small-cell lung cancer(NSCLC) patients and accurately target the patient subgroups with best potential outcome greatest, a simpler and more effective model based on readily available laboratory parameters is needed in clinical practice.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eTotally, computed tomographic (CT) images from 162 EGFR-mutated NSCLC patients at the time of first diagnosis in West China Hospital of Sichuan University were retrospectively collected and analysed to describe the target area of the lesion. All patients were randomly divided into a training cohort and a validation cohort. Radiomic features from images were extracted. The least absolute shrinkage selection operator (LASSO) regression method was used to screen valuable radiomic features. The logistic regression method was used to establish a radiomic model, and the nomogram was used to evaluate the discrimination ability. Receiver characteristic curve (ROC) and decision curve analysis (DCA) were used to evaluate the performance of the model.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eWe established a nomogram that combined 6 important clinical factors and imaging risk scores to predict the three-year and five-year survival rates of EGFR-mutated NSCLC patients. The AUC-ROC of three-year OS was 0.715, and the five-year OS was 0.705, which indicated a favourable discrimination capability of our nomogram and its great potential in targeting clinical services and predicting patient outcomes.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eBased on chest CT imaging and clinicopathological features, including age, sex, Stage_T, Stage_N, Stage_M, clinical stage, and imaging risk score, we constructed and validated a nomogram for predicting an individualized prediction of survival for patients with EGFR‑mutated NSCLC.\u003c/p\u003e","manuscriptTitle":"A Computer Tomography Radiomics-based model to Predict Survival of Patients with EGFR‑Mutated Non‑small‑cell Lung Cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2022-07-19 16:10:42","doi":"10.21203/rs.3.rs-1840484/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":"3ba3a154-88e8-4809-9269-029d941ab4b6","owner":[],"postedDate":"July 19th, 2022","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2022-08-30T16:59:15+00:00","versionOfRecord":[],"versionCreatedAt":"2022-07-19 16:10:42","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-1840484","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-1840484","identity":"rs-1840484","version":["v1"]},"buildId":"WrCJVZZCHTDjtuVLN7oU0","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}