CT radiomics for Predicting Survival in Patients with Unresectable Pancreatic Ductal Adenocarcinoma after Chemotherapy

CT radiomics for Predicting Survival in Patients with Unresectable Pancreatic Ductal Adenocarcinoma after Chemotherapy | 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 Article CT radiomics for Predicting Survival in Patients with Unresectable Pancreatic Ductal Adenocarcinoma after Chemotherapy Yong Zhu, Wenjing Cui, Hailin Jin, Jianhua Wang, Zhongqiu Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4284696/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 Objective: To investigate the ability of preoperative CT-based radiomics signature to be used as prognostic indicators of unresectable pancreatic ductal adenocarcinoma (PDAC) after chemotherapy and develop radiomics-based prediction models. Methods: Clinicopathological data and radiomics features of 146 pancreatic cancer patients (102 in the training cohort and 44 in the testing cohort) were retrospectively reviewed. Radiomics features were selected by the lasso cox regression model. Multivariate Cox regression analysis was used to establish the radiomics model, clinicopathologic model, and combined model in the training cohort to predict survival of PDAC after chemotherapy, and then verified in the testing cohort. Results: Of 855 extracted radiomics features in portal venous CT images, 15 most stable features were selected. Multivariate Cox regression analysis identified radiomics model riskscore, diabetes, common hepatic artery invasion, and splenic vein invasion to construct a combined model for predicting overall survival (OS) of unresectable PDAC (with AUC of 0.949 in the validation set; 95% CI 0.889-1.000). While the AUC of the radiomics model and clinicopathologic model were 0.901 (95% CI, 0.812-0.990) and 0.556 (95% CI, 0.382-0.730), respectively. Conclusion: CT radiomics signature is a powerful predictor associated with OS in patients with unresectable PDAC after chemotherapy. The combined model performed well in predicting survival. Biological sciences/Cancer/Cancer imaging Health sciences/Risk factors Health sciences/Health care/Prognosis Chemotherapy Computed Tomography Pancreatic Ductal Adenocarcinoma Radiomics Survival Figures Figure 1 Figure 2 Figure 3 Introduction Pancreatic ductal adenocarcinoma (PDAC) is a highly malignant digestive system tumor [1]. Patients with pancreatic cancer tend to be at an advanced stage at presentation, with 53% of pancreatic cancer diagnosed as metastatic and 28% with locally advanced disease, so the chance of radical surgery is lost [2]. Chemotherapy is one of the main treatment methods for unresectable pancreatic cancer due to its safety, reliability, and high efficiency, it is effective in local tumor control and pain relief while protecting the immune function of patients [3-4]. Stratification of patients before treatment according to the risk of death would allow for more aggressive treatment and guide optimal management. In recent years, the mortality rate of pancreatic cancer has been increasing, and the 5-year overall survival (OS) rate is currently less than 5% [5,6]. The prognosis of unresectable pancreatic cancer is extremely poor, with a median survival time of only 3 to 6 months [7]. According to previous reports [8-11], important indicators such as elevated carbohydrate antigen19-9 (CA 19-9), vascular invasion, lymph node metastasis, tumor stage, and tumor differentiation may affect the prognosis and OS of PDAC. However, prognostic factors related to biological behavior often require invasive histologic evaluation, and due to the variability of sampling, invasive histologic evaluation biopsy specimens cannot fully represent tumor characteristics. Radiomics is gaining importance in individualized cancer diagnosis and treatment. CT-based radiomics signature involves extracting a large number of features from digital medical images, which could noninvasively evaluate tumor differentiation, lymph node metastasis, and other prognostic factors of various types of tumors [12,13]. However, only a few studies have evaluated the important role of preoperative texture analysis in the prognosis evaluation of resectable pancreatic cancer [10,14-16] and even fewer studies with small sample sizes have been conducted for unresectable pancreatic cancer [5]. The role of CT-based radiomics in predicting outcomes of patients with unresectable pancreatic cancer deserves further investigation. Thus, this study aimed to investigate the ability of CT-based radiomics signatures to be used as prognostic indicatorsof unresectable PDAC after chemotherapy and develop radiomics-based prediction models. Materials and methods The study protocol was in complies with the Declaration of Helsinki and acts in accordance to ICH GCP guidelines. Institutional review board of Affiliated Hospital of Nanjing University of Chinese Medicine, Jiangsu Province Hospital of Chinese Medicine approved this study and explicitly waived the informed consent due to its retrospective nature. It also clarified that authors had access to identifying patient information when analyzing the data. Patients From January 2018 to December 2022, a total of 213 patients with a clinical diagnosis of pancreatic cancer were reviewed. The inclusion criteria were: (a) pathologically confirmed as PDAC and chemotherapy was performed; (b) patients were classified as unresectable disease according to the National Comprehensive Cancer Network (NCCN) guideline or present with metastatic disease; (c) preoperative contrast-enhanced CT performed within two weeks before therapy; The exclusion criteria were: (a) with a history of malignancy or combined with other malignancy (n = 7); (b) with pancreatic surgery history or with any local/systematic treatment before such as radiotherapy and chemotherapy (n =46); (c) the follow-up period was less than three months and no outcome events occurred (n=14 ) A total of 146 patients, including 80 men and 66 women, with a median age of 64 years (range, 42 – 88), were enrolled in the study. Then, 102 and 44 patients were allocated to the training and testing cohorts using random sampling at a fixed ratio of 7:3, respectively. CT examination All patients underwent contrast-enhanced CT scans on a multidetector CT scanner (brilliance iCT 256, Philips, The Netherlands). The scan ranged from the diaphragm to the pubic symphysis. After a plain CT scan, patients received 1mL/kg body weight of contrast media (Iopromide 370mg I/mL) at a rate of 3.5mL/s through the elbow. Dual-phase enhanced scanning was performed using mass injection tracking technology. An area of interest was set in the abdominal aorta. The CT value reached 100 HU and automatically triggered arterial phase scanning. Portal vein phase scanning was performed 60 to 70 seconds after triggering. The scanning parameters were as follows: tube voltage, 120 kVp; tube current, automatic milliampere technology; slice thickness, 2mm; layer spacing, 2mm; pitch, 1.0; FOV, 320mm × 320mm, collimator width, 128 mm × 0.6 mm; matrix size, 512 × 512. The medium interval between CT examination and chemotherapy was 7 days (range, 2- 13). Region-of-interest segmentation and radiomics features extraction The loading and display of CT images, delineation of the volume of interest, and feature extraction were performed using 3D Slicer (version 4.11.1; http:// www. slicer.org), an open-source medical image analysis and visualization platform. Two radiologists (with 8 and 10 years of experience in abdominal imaging diagnosis, respectively), who were blinded to the survival status of patients, delineated the boundaries of the tumor lesion independently. Region of interest (ROI) was defined as the area along the edge of each lesion layer of the tumor manually delineated on the portal vein phase transverse image. ROIs sometimes included visible necrosis and large blood vessels within the tumor, excluding adjacent pancreatic parenchyma. The software automatically reads the CT value of each pixel within the volume of interest (VOI) (mean area of VOIs, 26.86cm³± 22.80; range 2.13–186.95 cm³), generating 855 parameters including 4 types: first-order statistics, shape-based feature, texture feature, wavelet-based feature. Image preprocessing and feature extraction were performed using the open-source Pyradiomics package (version 2.2.0: http:// www. radio mics. io/ pyradiomics. html). The voxel spacing was standardized with the size of 1 × 1 × 1 mm and voxel intensity values were discretized with a bin width of 25 HU to reduce the interference of image noise and normalize intensities. Clinicopathologic data collection, chemotherapy, and follow-up For all patients, clinical symptoms, preoperative biochemical data, and pathology related information were collected from the medical records. Clinical manifestations were documented as abdominal pain (without/with), jaundice (without/with), and diabetes (with/without). The main biochemical data included total bilirubin (TB), albumin (ALB), and CA19-9. The pathology related information, including the clinical-TNM (cTNM) stage according to the American Joint Committee on Cancer (AJCC) 8th edition staging system [17] and major peripancreatic vascular invasion status, was recorded according to preoperative CT imaging. All patients underwent endoscopic ultrasound-guided tumor needle biopsy, and chemotherapy treatment was performed after pathological results were obtained. All patients received concurrent gemcitabine combined with a platinum agent, erlotinib, or a fluoropyrimidine. After treatment, patients were followed in the outpatient clinic every 3 months for the first years and every 6 months thereafter. During a median follow-up time of 8.0 months (mean, 10.3, range, 2.7 – 31.3), 38 patients were still alive, 99 patients died, and 9 patients were lost to follow-up. Statistical analysis Categorical variables were compared using χ2 validation or Wilcoxon validation. Continuous variables were compared by Student t validation or Mann-Whitney U validation, when appropriate. Interobserver agreement of radiomics features between two radiologists was assessed with intraclass correlation coefficients (≥ 0.75 were kept for further analysis). Radiomics features were selected by using a parametric method, the least absolute shrinkage and selection operator (LASSO) Cox regression [18,19]. Based on the radiomics features and clinicopathological data, univariate and multivariate Cox regression were performed, and we constructed the radiomics model, clinicopathologic model, and combined model. Feature selection and model building were carried out in the training cohort, whereas model performance was examined in the testing cohort. According to the median predicted risk score, the training cohort was divided into two groups (high-risk group versus low-risk group). We subsequently applied the same cutoff to the test cohort. The log-rank test was used to compare the overall survival between groups. The prognostic performance of each model was evaluated using AUC according to the predicted risk. Comparisons between the three models were performed using the Delong validation. A p-value < 0.05 (two-tailed) was considered statistically significant. Statistical analyses were performed with SPSS, version 22.0 (IBM, Armonk, NY, USA) and R software (R Foundation for Statistical Computing, version 3.4.1; https://www.r-project.org/). Results Patient Characteristics The general characteristics of patients in training and testing cohorts are summarized in Table 1. The average OS of the training cohort and testing cohort were (9.99±6.68) months and (11.00±7.15) months, respectively. There were no significant differences in all characteristics (mainly including clinical parameters and pathology-related features, etc.) between the two groups (all p > 0.05). Radiomic Feature Selection and Signature Construction The LASSO Cox regression model was used to select radiomics features. The minimum criteria for tenfold cross-validation were applied to k selection. Using the 1 standard error (1-SE) criteria and the minimum criteria. The optimized lambda (k) was used to select features with non-zero coefficients. Of 855 extracted radiomics features in portal venous CT images, 15 most stable features were considered for subsequent analysis. Table 2 summarizes the selected features and their coefficients. The Radiomics model riskscore for the prediction of OS was calculated through a linear combination of selected features weighted by their coefficients. Compared with survival group patients, the riskscore value of death group patients significantly increased (p<0.001) (Fig 1). The AUC of the radiomics model for predicting OS in the training and testing cohorts were 0.981 (95% CI, 0.962-1.000) and 0. 0.901(95% CI, 0.812-0.990), respectively. Time-dependent ROC curve showed that the AUC values at event-time of 6, 12, and 24 months were 0.98, 0.99, and 0.91 in the training cohort, respectively, while they were 0.97, 0.94, and 0.87 in the validation cohort, respectively (Fig 2A and B). Development, Performance, and Validation of Prediction Models Univariate and multivariable Cox regression analyses for OS in the training cohort are documented in Table 3. Univariate analysis identified radiomics model riskscore, CA19-9, cTNM stage, and splenic artery invasion as being significantly associated with OS. Multivariate Cox regression analysis based on radiomics and clinicopathologic indicators identified radiomics model riskscore, diabetes, common hepatic artery invasion, and splenic vein invasion to construct a combined model for predicting OS of PDAC. Kaplan–Meier overall survival plots in the training and testing cohorts between different models were shown in Fig 3 (A-F), the OS of the high-risk group was significantly higher than that of the low-risk group. AUC estimates were compared between the radiomics model, clinicopathologic model, and combined model by using the Delong nonparametric approach in training and testing cohorts (Table 4). In the training cohort, the AUC of the combined model (0.990; 95% CI 0.977-1.000) was significantly higher than that of the radiomics model (0.549; 95% CI 0.431-0.667) (p = 0.047) and clinicopathologic model (0.981; 95% CI 0.962-1.000) (p < 0.001). In the testing cohort, the combined model yielded an excellent AUC (0.949; 95% CI 0.889-1.000), a sensitivity of 0.844, and a specificity of 0.932. Discussion Based on the radiomics signature and clinicopathologic indicators, we developed radiomics model, clinicopathologic model, and combined model to predict the OS of unresectable PDAC. Then, the prediction performance of different models was compared. The study demonstrated that CT-based radiomics signature could serve as a potential noninvasive biomarker for predicting survival in unresectable PDAC after chemotherapy. In patients with unresectable PDAC, the average OS after chemotherapy was (9.99 ± 6.68) months in the training group and (11.00 ± 7.15) months in the test group which was generally consistent with that of current chemoradiotherapy [20], making our study more clinically applicable. In the era of precision medicine, radiomics can provide a large amount of information about shape, signal strength, and texture [21, 22]. Our study fully mined the digital information of unresectable PDAC, extracted a large number of radiomics signatures, and further screened out the 15 most stable features, including morphological features and wavelet features. In our study, CT-based radiomics was found to be the significant prognostic factor for OS by univariate and multivariate Cox regression analysis with excellent performance in the training cohort (AUC = 0.981, 95% CI, 0.962-1.000) and testing cohorts (AUC = 0.901, 95% CI, 0.812-0.990). Cheng et al. [5]reported that CT imaging biomarkers from texture analysis are associated with OS in patients with unresectable PDAC who were treated with chemotherapy and developed a survival model with AUC of 0.756. Hyun et al. [23] found that the higher entropy based on PET-CT data was independently associated with worse survival. In regard to resectable PDAC, Saleh et al. suggested that several radiomics features can be used as prognostic tools for pancreatic cancer, and mean tumor density was the only variable both in 2D and 3D analyses that could reliably predict OS. Li et al. [16]developed the radiomics model in CT images that could accurately predict 1-year and 2-year recurrence in patients with PDAC after radical resection. CT radiomics and their significant correlation with prognosis and OS have also been confirmed in many previous studies [9,10,14,15]. In addition, univariate and multivariate Cox regression analysis in our study to determine the risk predictor in unresectable pancreatic cancer identified CA19-9, cTNM stage, splenic artery, diabetes, common hepatic artery invasion, and splenic vein invasion as being significantly associated with OS. CA19-9 had a mean sensitivity of 79% and a mean specificity of 90% for prognosis, providing useful preoperative prognostic information, and patients with normal levels of CA19-9 had longer median survival than those with elevated levels [24-26]. Meanwhile, we found that pancreatic cancer with diabetes significantly increases the risk of poor prognosis. Kleeff et al. [27] also suggested that diabetes status was associated with overall survival of pancreatic cancer, and the risk of death was increased in pancreatic cancer patients with diabetes. Multivariate Cox regression analysis in our study revealed that common hepatic artery invasion and splenic vein invasion were independent risk factors affecting the prognosis of unresectable PDAC patients. Studies have shown that tumor invasion of large blood vessels causes tumor cells to form micro-metastases, and neovascularization provides nutrients for tumor growth, thus forming a vicious cycle [28]. Most studies have reported that the prognosis of pancreatic cancer patients with vascular invasion is poor [29]. However, there are few reports on the correlation between the invasion of different major peripancreatic vessels and the prognosis of pancreatic cancer. Our study had some limitations. Firstly, we did not study locally advanced PDAC and distant metastatic PDAC in our study separately, and further stratification of unresectable PDAC may provide some valuable results, which need further study. Additionally, our study uses a single-center retrospective design, which led to selection and institutional biases, external validation of this study in a larger cohort will be necessary to reduce the risk of overfitting, thereby helping to improve the robustness and stability of the model. Conclusion Our study indicates that CT radiomics signature is a powerful predictor associated with OS in patients with unresectable PDAC after chemotherapy. The combined model (constructed with radiomics model riskscore, diabetes, common hepatic artery invasion, and splenic vein invasion) performed well in predicting the OS of unresectable PDAC, which helps clinicians to select the optimal treatment strategy and an individualized follow-up plan to improve clinical outcomes. Abbreviations ALB Albumin AUCArea under the curve Ca19-9 Carbohydrate antigen19-9 LASSO Least absolute shrinkage and selection operator OS Overall survival ROI Region of interest TB Total bilirubin VOI Volume of interest Declarations Ethics approval and consent to participate Not applicable Consent for publication Institutional Review Board of the Affiliated Hospital of Nanjing University of Chinese Medicine approved this study and the informed consent from patients was waived due to its retrospective nature. Data availability statement The datasets generated and/or analysed during the current study are not publicly available due local hospital policy but are available from the corresponding author on reasonable request. Competing interests The authors declare no competing interests. Funding None Acknowledgments The authors thank the Multidiscipline Team in Pancreaticoduodenal Disease of the Affiliated Hospital of Nanjing University of Chinese Medicine for professional discussion for this manuscript. References Klaiber, U. et al. Prognostic Factors of Survival After Neoadjuvant Treatment and Resection for Initially Unresectable Pancreatic Cancer[J]. Ann Surg, 273 (1):154-162 (2021). Sheahan, A.V., Biankin, A.V., Parish, C.R., Khachigian, L.M. Targeted therapies in the management of locally advanced and metastatic pancreatic cancer: a systematic review[J]. Oncotarget, 9 (30):21613-21627(2018). D'Haese, J.G. et al. Pain sensation in pancreatic diseases is not uniform: the different facets of pancreatic pain[J]. World J Gastroenterol, 20 (27):9154-9161(2014). Barcellini, A., Peloso, A., Pugliese, L., Vitolo, V., Cobianchi, L. Locally Advanced Pancreatic Ductal Adenocarcinoma: Challenges and Progress[J]. Onco Targets Ther, 10 (13):12705-12720 (2020). Cheng, S.H., Cheng, Y.J., Jin, Z.Y., Xue, H.D. Unresectable pancreatic ductal adenocarcinoma: Role of CT quantitative imaging biomarkers for predicting outcomes of patients treated with chemotherapy[J]. Eur J Radiol, 113 :188-197 (2019). Malla, M. et al. The evolving role of radiation in pancreatic cancer[J]. Front Oncol, 12 :1060885(2022). Yousaf, M. N. et al. Role of Radiofrequency Ablation in the Management of Unresectable Pancreatic Cancer[J]. Front Med (Lausanne), 7 :624997(2020). Dell'Aquila, E. et al. Prognostic and predictive factors in pancreatic cancer[J]. Oncotarget, 11 (10):924-941(2020). Shi, H. et al. Survival prediction after upfront surgery in patients with pancreatic ductal adenocarcinoma: Radiomic, clinic-pathologic and body composition analysis[J]. Pancreatology, 21 (4):731-737(2021). Healy, G. M. et al. Pre-operative radiomics model for prognostication in resectable pancreatic adenocarcinoma with external validation[J]. Eur Radiol, 32 (4):2492-2505(2022). Tong, Z. et al. Development, Validation, and Comparison of Artificial Neural Network Models and Logistic Regression Models Predicting Survival of Unresectable Pancreatic Cancer[J]. Front Bioeng Biotechnol, 8 :196(2020). Hosny, A. et al. Deep learning for lung cancer prognostication: A retrospective multi-cohort radiomics study[J]. PLoS Med, 15 (11): e1002711 (2018). Wu, S. et al. A Radiomics Nomogram for the Preoperative Prediction of Lymph Node Metastasis in Bladder Cancer[J]. Clin Cancer Res, 23 (22):6904-6911 (2017). Zhang, Y. et al. Improving prognostic performance in resectable pancreatic ductal adenocarcinoma using radiomics and deep learning features fusion in CT images[J]. Sci Rep, 11 (1):1378 (2021). Cen, C. et al. Pancreatic Ductal Adenocarcinoma at CT: A Combined Nomogram Model to Preoperatively Predict Cancer Stage and Survival Outcome[J]. Front Oncol, 11 :594510 (2021). Li, X. et al. Preoperative recurrence prediction in pancreatic ductal adenocarcinoma after radical resection using radiomics of diagnostic computed tomography[J]. EClinicalMedicine, 43 :101215 (2022). Chun, Y. S., Pawlik, T. M., Vauthey, J. N. 8th Edition of the AJCC Cancer Staging Manual: Pancreas and Hepatobiliary Cancers[J]. Ann Surg Oncol, 25 (4):845-847 (2018). Vasquez, M.M., Hu, C., Roe, D.J., Halonen, M., Guerra, S. Measurement error correction in the least absolute shrinkage and selection operator model when validation data are available[J]. Stat Methods Med Res, 28 (3):670-680 (2019). Vasquez, M. M. et al. Least absolute shrinkage and selection operator type methods for the identification of serum biomarkers of overweight and obesity: simulation and application[J]. BMC Med Res Methodol, 16 (1):154 (2016). Girelli, R. et al. Results of 100 pancreatic radiofrequency ablations in the context of a multimodal strategy for stage III ductal adenocarcinoma[J]. Langenbecks Arch Surg, 398 (1):63-69 (2013). Wang, J. et al. Radiomics analysis of contrast-enhanced CT for staging liver fibrosis: an update for image biomarker[J]. Hepatol Int, 16 (3):627-639 (2022). Lin, Z. et al. Preoperative prediction of clinically relevant postoperative pancreatic fistula after pancreaticoduodenectomy[J]. Eur J Radiol, 139 :109693 (2021). Hyun, S.H. et al. Intratumoral heterogeneity of (18)F-FDG uptake predicts survival in patients with pancreatic ductal adenocarcinoma[J]. Eur J Nucl Med Mol Imaging, 43 (8):1461-1468 (2016). Ballehaninna, U.K., Chamberlain, R.S. The clinical utility of serum CA 19-9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: An evidence based appraisal[J]. J Gastrointest Oncol, 3 (2):105-119 (2012). Rofi, E. et al. The emerging role of liquid biopsy in diagnosis, prognosis and treatment monitoring of pancreatic cancer[J]. Pharmacogenomics, 20 (1):49-68 (2019). Amaral, M.J., Oliveira, R.C., Donato, P., Tralhão, J.G. Pancreatic Cancer Biomarkers: Oncogenic Mutations, Tissue and Liquid Biopsies, and Radiomics-A Review[J]. Dig Dis Sci, 68 (7):2811-2823 (2023). Kleeff, J. et al. The impact of diabetes mellitus on survival following resection and adjuvant chemotherapy for pancreatic cancer[J]. Br J Cancer, 115 (7):887-894 (2016). Viallard, C., Larrivee, B. Tumor angiogenesis and vascular normalization: alternative therapeutic targets[J]. Angiogenesis, 20 (4):409-426 (2017). Huang, H. et al. Risk factors and prognostic index model for pancreatic cancer[J]. Gland Surg, 11 (1):186-195 (2022). Tables Table 1. General characteristics of patients in the training and testing cohorts. Variable Training Cohort (n = 102) Testing Cohort (n = 44) P value Overall survival, months 9.99 ± 6.68 11.00 ± 7.15 0.410 Number of deaths (%) 67(65.7%) 32 (72.7%) 0.403 Age (mean ± SD) 63.42 ± 9.15 62.27 ± 9.96 0.499 Sex (male/female) 55/47 25/19 0.747 Abdominal pain (with/without) 89/13 37/7 0.610 Jaundice (with/without) 20/82 11/33 0.465 Diabetes (with/without) 21/81 7/37 0.510 TB (μmol/L) 31.65 ± 55.01 45.51 ± 65.70 0.190 ALB (g/L) 41.01 ± 4.01 40.21 ± 3.94 0.761 CA19-9 (≤/>37 U/mL) 15/87 8/36 0.597 cTNM stage (Ⅲ/Ⅳ) 66/36 32/12 0.344 Celiac artery invasion (with/without) 40/62 18/26 0.848 Common hepatic artery invasion (with/without) 49/53 24/20 0.471 Splenic artery invasion (with/without) 52/50 21/23 0.718 Superior mesenteric artery invasion (with/without) 35/67 19/25 0.308 Portal vein invasion (with/without) 44/58 19/25 0.996 Superior mesenteric vein invasion (with/without) 51/51 21/23 0.801 Splenic vein invasion (with/without) 58/44 21/23 0.309 TB, total bilirubin; HDL, high-density lipoprotein; ALB, albumin Table 2. Radiomics features associated with overall survival and coefficients selected by LASSO cox regression. Survival Outcome and Radiomic Features Coefficient Overall survival Original_ shape_Maximum2D_DiameterSlice 3.358070×10 -4 Wavelet-LHL_ glcm _ Contrast 2.403567×10 0 Wavelet-LHL_ gldm _ LargeDependenceLowGrayLevelEmphasis -1.976730×10 -2 Wavelet-LHL_ glszm _ GrayLevelNonUniformityNormalized -1.921688×10 0 Wavelet-LHH_ glcm_ ClusterShade 5.702145×10 -1 Wavelet-LHH_ glcm_Imc2 -4.611064×10 0 Wavelet-LHH_ gldm_HighGrayLevelEmphasis 4.004665×10 -4 Wavelet-LHH_ glszm _LargeAreaLowGrayLevelEmphasis 1.307349×10 -5 Wavelet-LHL_ firstorder_Mean -2.430718×10 0 Wavelet-HLH_ glcm_Contrast -1.577605×10 1 Wavelet-HLH_ glcm_Idmn -2.907880×10 0 Wavelet-HLH_ ngtdm_Busyness 3.876596×10 -5 Wavelet-HLL_ firstorder_Mean -6.142849×10 0 Wavelet-HHH_ glszm_LargeAreaLowGrayLevelEmphasis 1.837997×10 -4 Wavelet-HHH_ ngtdm_Busyness 1.758318×10 -2 Abbreviations: glszm—gray-level size zone matrix; glcm—gray-level co-occurrence matrix; and ngtdm—neighboring gray tone difference matrix. LHH, LHL, HLH, HLL, and HHH denote the high- and low-pass filters on the x, y, and z dimensions, respectively (H—high; L—low). Table 3. Uni- and Multivariable Cox Regression Analyses for OS in the training cohort. Variable Univariable HR P Value Multivariable HR P Value Radiomics model riskscore 15.788 (8.123, 30.688) < 0.001* 30.184 (12.830, 71.009) < 0.001* Diabetes 1.553 (0.271, 2.047) 0.068 1.389 (1.157, 2.967) 0.042* CA19-9 2.959 (1.074, 8.147) 0.036* … … cTNM stage 1.675 (1.006, 2.788) 0.047* … … Celiac artery invasion 1.073 (0.656, 1.754) 0.779 … … Common hepatic artery invasion 1.056 (0.651, 1.713) 0.824 5.046 (1.385, 18.388) 0.014* Splenic artery invasion 1.669 (1.024, 2.720) 0.040* … … Superior mesenteric artery invasion 1.105 (0.675, 1.809) 0.690 … … Portal vein invasion 1.175 (0.723, 1.909) 0.515 … … Superior mesenteric vein invasion 0.817 (0.536, 1.414) 0.576 … … Splenic vein invasion 1.299 (0.796, 2.118) 0.295 7.163 (1.044, 9.606) 0.007* *, p < 0.05 Table 4. Performance of different models in training and testing cohorts for the prediction of overall survival. Variables and models AUC (95% CI) P value 1vs.2 1vs.3 2vs.3 Training Cohort 1. Radiomics model 0.981 (0.962-1.000) < 0.001 0.047 < 0.001 2. Clinicopathologic model 0.549 (0.431-0.667) 3. Combined model 0.990 (0.977-1.000) Testing Cohort 1. Radiomics model 0.901 (0.812-0.990) 0.004 0.311 < 0.001 2. Clinicopathologic model 0.556 (0.382-0.730) 3. Combined model 0.949 (0.889-1.000) 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-4284696","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":295600028,"identity":"50691772-98b5-4ee7-86e5-25785ced9b7f","order_by":0,"name":"Yong Zhu","email":"","orcid":"","institution":"Affiliated Hospital of Nanjing University of Chinese Medicine, Jiangsu Province Hospital of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yong","middleName":"","lastName":"Zhu","suffix":""},{"id":295600029,"identity":"5fce0f97-77d9-42d5-a8d7-3e5516a3fd5a","order_by":1,"name":"Wenjing Cui","email":"","orcid":"","institution":"Affiliated Hospital of Nanjing University of Chinese Medicine, Jiangsu Province Hospital of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Wenjing","middleName":"","lastName":"Cui","suffix":""},{"id":295600030,"identity":"8c537cf0-0fe2-4a72-911f-cf0d8a1ef869","order_by":2,"name":"Hailin Jin","email":"","orcid":"","institution":"Affiliated Hospital of Nanjing University of Chinese Medicine, Jiangsu Province Hospital of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Hailin","middleName":"","lastName":"Jin","suffix":""},{"id":295600031,"identity":"0eaaf5a8-70f5-49b8-a1dd-e68cf6f18429","order_by":3,"name":"Jianhua Wang","email":"","orcid":"","institution":"Affiliated Hospital of Nanjing University of Chinese Medicine, Jiangsu Province Hospital of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Jianhua","middleName":"","lastName":"Wang","suffix":""},{"id":295600032,"identity":"cb399422-e7c5-40cc-9452-b3c80e5391bb","order_by":4,"name":"Zhongqiu Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA20lEQVRIiWNgGAWjYDCCA8icDwY2csRoYWyAsRlnFKQZk6aFmefD4USCOviONz9/8HHPYXlz/uUPb9sYMCcwsB8+ugGfFskzxwwbZzw7bLhzxoNk6xwDtjwGnrS0G/i0GNzIYWzmOXCbccONA8ekcwx4ihkkeMyI0mK/4cbBNmkLA4nEBmK1JG4438wmzWBgQFgLyC8zZxz4n7zhBhuzZY9BgjEbIb8AQ+zBhw8H0mw3nD/+8MaPP//l+NkPH8OrBQEkEhgkQDQbccpBgP8ARMsoGAWjYBSMAnQAAIxnU8mUcm+8AAAAAElFTkSuQmCC","orcid":"","institution":"Affiliated Hospital of Nanjing University of Chinese Medicine, Jiangsu Province Hospital of Chinese Medicine","correspondingAuthor":true,"prefix":"","firstName":"Zhongqiu","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2024-04-18 02:59:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4284696/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4284696/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":55536919,"identity":"9b365719-b647-4d73-85b8-69d4b2117b51","added_by":"auto","created_at":"2024-04-29 16:39:18","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":64741,"visible":true,"origin":"","legend":"\u003cp\u003eRadiomics modelrisk scores for patients with unresectable pancreatic ductal adenocarcinoma in different survival statuses, where “0” represents survival and “1” represents death.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-4284696/v1/a66b5e041ccfd8cf1df77527.png"},{"id":55536921,"identity":"aeec8e92-be2a-483e-8d2d-faaed420acac","added_by":"auto","created_at":"2024-04-29 16:39:18","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":713453,"visible":true,"origin":"","legend":"\u003cp\u003eTime-dependent ROC curves of the radiomics model in the training (A) and testing (B) cohorts at different event-time (6, 12, and 24 months).\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-4284696/v1/e7946d02b82942cd6ee92a1f.png"},{"id":55536920,"identity":"d9b8fa7d-1fb0-48d0-9fe9-cca24a6cd181","added_by":"auto","created_at":"2024-04-29 16:39:18","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":5847812,"visible":true,"origin":"","legend":"\u003cp\u003eKaplan–Meier overall survival plots in the training (A, C, E) and testing (B, D, F) cohorts stratified according to the predicted risk derived from radiomic models (A, B), clinicopathologic models (C, D), and combined models (E, F). p values were calculated using log-rank tests.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-4284696/v1/880a8f54f7801767891d2f5e.png"},{"id":57186008,"identity":"5c6c02f5-44a9-4a2c-bdc7-1a8d5f880245","added_by":"auto","created_at":"2024-05-27 05:52:27","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8854884,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4284696/v1/3a64285f-2e9c-473a-b9b4-53be5335ef64.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"CT radiomics for Predicting Survival in Patients with Unresectable Pancreatic Ductal Adenocarcinoma after Chemotherapy","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePancreatic ductal adenocarcinoma (PDAC) is a highly malignant digestive system tumor [1]. Patients with pancreatic cancer tend to be at an advanced stage at presentation, with 53% of pancreatic cancer diagnosed as metastatic and 28% with locally advanced disease, so the chance of radical surgery is lost [2]. Chemotherapy is one of the main treatment methods for unresectable pancreatic cancer due to its safety, reliability, and high efficiency, it is effective in local tumor control and pain relief while protecting the immune function of patients [3-4]. Stratification of patients before treatment according to the risk of death would allow for more aggressive treatment and guide optimal management.\u003c/p\u003e\n\u003cp\u003eIn recent years, the mortality rate of pancreatic cancer has been increasing, and the 5-year overall survival (OS) rate is currently less than 5% [5,6]. The prognosis of unresectable pancreatic cancer is extremely poor, with a median survival time of only 3 to 6 months [7]. According to previous reports [8-11], important indicators such as elevated carbohydrate antigen19-9 (CA 19-9), vascular invasion, lymph node metastasis, tumor stage, and tumor differentiation may affect the prognosis and OS of PDAC. However, prognostic factors related to biological behavior often require invasive histologic evaluation, and due to the variability of sampling, invasive histologic evaluation biopsy specimens cannot fully represent tumor characteristics.\u003c/p\u003e\n\u003cp\u003eRadiomics is gaining importance in individualized cancer diagnosis and treatment. CT-based radiomics signature involves extracting a large number of features from digital medical images, which could noninvasively evaluate tumor differentiation, lymph node metastasis, and other prognostic factors of various types of tumors [12,13]. However, only a few studies have evaluated the important role of preoperative texture analysis in the prognosis evaluation of resectable pancreatic cancer [10,14-16] and even fewer studies with small sample sizes have been conducted for unresectable pancreatic cancer [5]. The role of CT-based radiomics in predicting outcomes of patients with unresectable pancreatic cancer deserves further investigation.\u003c/p\u003e\n\u003cp\u003eThus, this study aimed to investigate the ability of CT-based radiomics signatures to be used as prognostic indicatorsof unresectable PDAC after chemotherapy and develop radiomics-based prediction models.\u0026nbsp;\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cp\u003eThe study protocol was in complies with the Declaration of Helsinki and acts in accordance to ICH GCP guidelines. Institutional review board of Affiliated Hospital of Nanjing University of Chinese Medicine, Jiangsu Province Hospital of Chinese Medicine approved this study and explicitly waived the informed consent due to its retrospective nature. It also clarified that authors had access to identifying patient information when analyzing the data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003ePatients\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFrom January 2018 to December 2022, a total of 213 patients with a clinical diagnosis of pancreatic cancer were reviewed. The inclusion criteria were: (a) pathologically confirmed as PDAC and chemotherapy was performed; (b) patients were classified as unresectable disease according to the National Comprehensive Cancer Network (NCCN) guideline or present with metastatic disease; (c) preoperative contrast-enhanced CT performed within two weeks before therapy; The exclusion criteria were: (a) with a history of malignancy or combined with other malignancy (n = 7); (b) with pancreatic surgery history or with any local/systematic treatment before such as radiotherapy and chemotherapy (n =46); (c) the follow-up period was less than three months and no outcome events occurred (n=14 )\u003c/p\u003e\n\u003cp\u003eA total of 146 patients, including 80 men and 66 women, with a median age of 64 years (range, 42 \u0026ndash; 88), were enrolled in the study. Then, 102 and 44 patients were allocated to the training and testing cohorts using random sampling at a fixed ratio of 7:3, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCT examination\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll patients underwent contrast-enhanced CT scans on a multidetector CT scanner (brilliance iCT 256, Philips, The Netherlands). The scan ranged from the diaphragm to the pubic symphysis. After a plain CT scan, patients received 1mL/kg body weight of contrast media (Iopromide 370mg I/mL) at a rate of 3.5mL/s through the elbow. Dual-phase enhanced scanning was performed using mass injection tracking technology. An area of interest was set in the abdominal aorta. The CT value reached 100 HU and automatically triggered arterial phase scanning. Portal vein phase scanning was performed 60 to 70 seconds after triggering. The scanning parameters were as follows: tube voltage, 120 kVp; tube current, automatic milliampere technology; slice thickness, 2mm; layer spacing, 2mm; pitch, 1.0; FOV, 320mm \u0026times; 320mm, collimator width, 128 mm \u0026times; 0.6 mm; matrix size, 512 \u0026times; 512.\u003c/p\u003e\n\u003cp\u003eThe medium interval between CT examination and chemotherapy was 7 days (range, 2- 13).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRegion-of-interest segmentation and radiomics features extraction\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe loading and display of CT images, delineation of the volume of interest, and feature extraction were performed using 3D Slicer (version 4.11.1; http:// www. slicer.org), an open-source medical image analysis and visualization platform.\u003c/p\u003e\n\u003cp\u003eTwo radiologists (with 8 and 10 years of experience in abdominal imaging diagnosis, respectively), who were blinded to the survival status of patients, delineated the boundaries of the tumor lesion independently.\u0026nbsp;Region of interest\u0026nbsp;(ROI) was defined as the area along the edge of each lesion layer of the tumor manually delineated on the portal vein phase transverse image. ROIs sometimes included visible necrosis and large blood vessels within the tumor, excluding adjacent pancreatic parenchyma.\u003c/p\u003e\n\u003cp\u003eThe software automatically reads the CT value of each pixel within the volume of interest (VOI)\u0026nbsp;(mean area of VOIs, 26.86cm\u0026sup3;\u0026plusmn; 22.80; range 2.13\u0026ndash;186.95 cm\u0026sup3;), generating 855 parameters including 4 types: first-order statistics, shape-based feature, texture feature, wavelet-based feature. Image preprocessing and feature extraction were performed using the open-source Pyradiomics package (version 2.2.0: http:// www. radio mics. io/ pyradiomics. html). The voxel spacing was standardized with the size of 1 \u0026times; 1 \u0026times; 1 mm and voxel intensity values were discretized with a bin width of 25 HU to reduce the interference of image noise and normalize intensities.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eClinicopathologic\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003edata collection, chemotherapy, and follow-up\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor all patients, clinical symptoms, preoperative biochemical data, and pathology related information were collected from the medical records. Clinical manifestations were documented as abdominal pain (without/with), jaundice (without/with), and\u0026nbsp;diabetes\u0026nbsp;(with/without). The main biochemical data included total bilirubin (TB), albumin (ALB), and CA19-9. The pathology related information, including the clinical-TNM (cTNM) stage according to the American Joint Committee on Cancer (AJCC) 8th edition staging system [17] and major peripancreatic vascular invasion status, was recorded according to preoperative CT imaging.\u003c/p\u003e\n\u003cp\u003eAll patients underwent endoscopic ultrasound-guided tumor needle biopsy, and chemotherapy treatment was performed after pathological results were obtained. All patients received concurrent gemcitabine combined with a platinum agent, erlotinib, or a fluoropyrimidine. After treatment, patients were followed in the outpatient clinic every 3 months for the first years and every 6 months thereafter. During a median follow-up time of 8.0 months (mean, 10.3, range, 2.7 \u0026ndash; 31.3), 38 patients were still alive, 99 patients died, and 9 patients were lost to follow-up.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eStatistical analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCategorical variables were compared using \u0026chi;2 validation or Wilcoxon validation. Continuous variables were compared by Student t validation or Mann-Whitney U validation, when appropriate. Interobserver agreement of radiomics features between two radiologists was assessed with intraclass correlation coefficients (\u0026ge; 0.75 were kept for further analysis). Radiomics features were selected by using a parametric method, the least absolute shrinkage and selection operator (LASSO) Cox regression [18,19]. Based on the radiomics features and clinicopathological data, univariate and multivariate Cox regression were performed, and we constructed the radiomics model, clinicopathologic model, and combined model. Feature selection and model building were carried out in the training cohort, whereas model performance was examined in the testing cohort. According to the median predicted risk score, the training cohort was divided into two groups (high-risk group versus low-risk group). We subsequently applied the same cutoff to the test cohort. The log-rank test was used to compare the overall survival between groups. The prognostic performance of each model was evaluated using AUC according to the predicted risk. Comparisons between the three models were performed using the Delong validation. A p-value \u0026lt; 0.05 (two-tailed) was considered statistically significant. Statistical analyses were performed with SPSS, version 22.0 (IBM, Armonk, NY, USA) and R software (R Foundation for Statistical Computing, version 3.4.1; https://www.r-project.org/).\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003ePatient Characteristics\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe general characteristics of patients in training and testing cohorts are summarized in Table 1. The average OS of the training cohort and testing cohort were (9.99±6.68) months and (11.00±7.15) months, respectively. There were no significant differences in all characteristics (mainly including clinical parameters and pathology-related features, etc.) between the two groups (all p \u0026gt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRadiomic Feature Selection and Signature Construction\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe LASSO Cox regression model was used to select radiomics features. The minimum criteria for tenfold cross-validation were applied to k selection. Using the 1 standard error (1-SE) criteria and the minimum criteria. The optimized lambda (k) was used to select features with non-zero coefficients. Of 855 extracted radiomics features in portal venous CT images, 15 most stable features were considered for subsequent analysis. Table 2 summarizes the selected features and their coefficients. The\u0026nbsp;Radiomics model riskscore\u0026nbsp;for the prediction of OS was calculated through a linear combination of selected features weighted by their coefficients. Compared with survival group patients, the riskscore value of death group patients significantly increased (p\u0026lt;0.001) (Fig 1). The AUC of the radiomics model for predicting OS in the training and testing cohorts were 0.981\u0026nbsp;(95% CI, 0.962-1.000) and 0. 0.901(95% CI, 0.812-0.990), respectively. Time-dependent ROC curve showed that the AUC values at event-time of 6, 12, and 24 months were 0.98, 0.99, and 0.91 in the training cohort, respectively, while they were 0.97, 0.94, and 0.87 in the validation cohort, respectively (Fig 2A and B).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDevelopment, Performance, and Validation of Prediction Models\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUnivariate\u0026nbsp;and multivariable Cox regression analyses for OS in the training cohort are documented in Table 3. Univariate analysis identified\u0026nbsp;radiomics model riskscore,\u0026nbsp;CA19-9, cTNM stage, and\u0026nbsp;splenic artery invasion\u0026nbsp;as being significantly associated with OS. Multivariate Cox regression analysis based on radiomics and clinicopathologic\u0026nbsp;indicators identified\u0026nbsp;radiomics model riskscore,\u0026nbsp;diabetes,\u0026nbsp;common hepatic artery invasion, and splenic vein invasion\u0026nbsp;to construct a combined model for predicting OS of PDAC.\u003c/p\u003e\n\u003cp\u003eKaplan–Meier overall survival plots in the training and testing cohorts between different models were shown in Fig 3 (A-F), the OS of the high-risk group was significantly higher than that of the low-risk group. AUC estimates were compared between the radiomics model, clinicopathologic model, and combined model by using the Delong nonparametric approach in training and testing cohorts (Table 4). In the training cohort, the AUC of the combined model (0.990; 95% CI 0.977-1.000) was significantly higher than that of the radiomics model (0.549; 95% CI 0.431-0.667) (p = 0.047) and clinicopathologic model (0.981; 95% CI 0.962-1.000) (p \u0026lt; 0.001). In the testing cohort, the combined model yielded an excellent AUC (0.949; 95% CI 0.889-1.000), a sensitivity of 0.844, and a specificity of 0.932.\u003c/p\u003e"},{"header":"Discussion ","content":"\u003cp\u003eBased on the radiomics signature and clinicopathologic indicators, we developed radiomics model, clinicopathologic model, and combined model to predict the OS of unresectable PDAC. Then, the prediction performance of different models was compared. The study demonstrated that CT-based radiomics signature could serve as a potential noninvasive biomarker for predicting survival in unresectable PDAC after chemotherapy.\u003c/p\u003e\n\u003cp\u003eIn patients with unresectable PDAC, the average OS after chemotherapy was (9.99 ± 6.68) months in the training group and (11.00 ± 7.15) months in the test group which was generally consistent with that of current chemoradiotherapy\u0026nbsp;[20], making our study more clinically applicable. In the era of precision medicine, radiomics can provide a large amount of information about shape, signal strength, and texture [21, 22]. Our study fully mined the digital information of unresectable PDAC, extracted a large number of radiomics signatures, and further screened out the 15 most stable features, including morphological features and wavelet features.\u003c/p\u003e\n\u003cp\u003eIn our study, CT-based radiomics was found to be the significant prognostic factor for OS by univariate and multivariate Cox regression analysis with excellent performance in the training cohort (AUC = 0.981,\u0026nbsp;95% CI, 0.962-1.000) and testing cohorts\u0026nbsp;(AUC = 0.901, 95% CI, 0.812-0.990). Cheng et al. [5]reported that CT imaging biomarkers from texture analysis are associated with OS in patients with unresectable PDAC who were treated with chemotherapy and developed a survival model with AUC of 0.756. Hyun et al. [23] found that the higher entropy based on PET-CT data was independently associated with worse survival. In regard to resectable PDAC, Saleh et al. suggested that several radiomics features can be used as prognostic tools for pancreatic cancer, and mean tumor density was the only variable both in 2D and 3D analyses that could reliably predict OS. Li et al. [16]developed the radiomics model in CT images that could accurately predict 1-year and 2-year recurrence in patients with PDAC after radical resection. CT radiomics and their significant correlation with prognosis and OS have also been confirmed in many previous studies [9,10,14,15].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn addition, univariate and multivariate Cox regression analysis in our study to determine the risk predictor in unresectable pancreatic cancer identified CA19-9, cTNM stage,\u0026nbsp;splenic artery,\u0026nbsp;diabetes,\u0026nbsp;common hepatic artery invasion, and splenic vein invasion\u0026nbsp;as being significantly associated with OS. CA19-9 had a mean sensitivity of 79% and a mean specificity of 90% for prognosis, providing useful preoperative prognostic information, and patients with normal levels of CA19-9 had longer median survival than those with elevated levels [24-26]. Meanwhile, we found that pancreatic cancer with diabetes significantly increases the risk of poor prognosis. Kleeff et al. [27] also suggested that diabetes status was associated with overall survival of pancreatic cancer, and the risk of death was increased in pancreatic cancer patients with diabetes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMultivariate Cox regression analysis in our study revealed that common hepatic artery invasion and splenic vein invasion were independent risk factors affecting the prognosis of unresectable PDAC patients. Studies have shown that tumor invasion of large blood vessels causes tumor cells to form micro-metastases, and neovascularization provides nutrients for tumor growth, thus forming a vicious cycle [28]. Most studies have reported that the prognosis of pancreatic cancer patients with vascular invasion is poor [29]. However, there are few reports on the correlation between the invasion of different major peripancreatic vessels and the prognosis of pancreatic cancer.\u003c/p\u003e\n\u003cp\u003eOur study had some limitations. Firstly, we did not study locally advanced PDAC and distant metastatic PDAC in our study separately, and further stratification of unresectable PDAC may provide some valuable results, which need further study. Additionally, our study uses a single-center retrospective design, which led to selection and institutional biases, external validation of this study in a larger cohort will be necessary to reduce the risk of overfitting, thereby helping to improve the robustness and stability of the model.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eOur study indicates that CT radiomics signature is a powerful predictor associated with OS in patients with unresectable PDAC after chemotherapy. The combined model (constructed with radiomics model riskscore, diabetes, common hepatic artery invasion, and splenic vein invasion) performed well in predicting the OS of unresectable PDAC, which helps clinicians to select the optimal treatment strategy and an individualized follow-up plan to improve clinical outcomes.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eALB \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Albumin\u003c/p\u003e\n\u003cp\u003eAUCArea under the curve\u003c/p\u003e\n\u003cp\u003eCa19-9 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Carbohydrate antigen19-9\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLASSO \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Least absolute shrinkage and selection operator\u003c/p\u003e\n\u003cp\u003eOS \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Overall survival\u003c/p\u003e\n\u003cp\u003eROI \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Region of interest\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTB \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Total bilirubin\u003c/p\u003e\n\u003cp\u003eVOI \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Volume of interest\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInstitutional Review Board of the Affiliated Hospital of Nanjing University of Chinese Medicine approved this study and the informed consent from patients was waived due to its retrospective nature.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and/or analysed during the current study are not publicly available due local hospital policy but 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 no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNone\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank the Multidiscipline Team in Pancreaticoduodenal Disease of the Affiliated Hospital of Nanjing University of Chinese Medicine for professional discussion for this manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eKlaiber, U. et al. Prognostic Factors of Survival After Neoadjuvant Treatment and Resection for Initially Unresectable Pancreatic Cancer[J]. Ann Surg,\u003cstrong\u003e273\u003c/strong\u003e(1):154-162 (2021). \u003c/li\u003e\n\u003cli\u003eSheahan, A.V., Biankin, A.V., Parish, C.R., Khachigian, L.M. Targeted therapies in the management of locally advanced and metastatic pancreatic cancer: a systematic review[J]. Oncotarget, \u003cstrong\u003e9 \u003c/strong\u003e(30):21613-21627(2018). \u003c/li\u003e\n\u003cli\u003eD\u0026apos;Haese, J.G. et al. Pain sensation in pancreatic diseases is not uniform: the different facets of pancreatic pain[J]. World J Gastroenterol, \u003cstrong\u003e20\u003c/strong\u003e(27):9154-9161(2014). \u003c/li\u003e\n\u003cli\u003eBarcellini, A., Peloso, A., Pugliese, L., Vitolo, V., Cobianchi, L. Locally Advanced Pancreatic Ductal Adenocarcinoma: Challenges and Progress[J]. Onco Targets Ther, \u003cstrong\u003e10\u003c/strong\u003e(13):12705-12720 (2020). \u003c/li\u003e\n\u003cli\u003eCheng, S.H., Cheng, Y.J., Jin, Z.Y., Xue, H.D. Unresectable pancreatic ductal adenocarcinoma: Role of CT quantitative imaging biomarkers for predicting outcomes of patients treated with chemotherapy[J]. Eur J Radiol,\u003cstrong\u003e113\u003c/strong\u003e:188-197 (2019). \u003c/li\u003e\n\u003cli\u003eMalla, M. et al. The evolving role of radiation in pancreatic cancer[J]. Front Oncol,\u003cstrong\u003e12\u003c/strong\u003e:1060885(2022). \u003c/li\u003e\n\u003cli\u003eYousaf, M. N. et al. Role of Radiofrequency Ablation in the Management of Unresectable Pancreatic Cancer[J]. Front Med (Lausanne),\u003cstrong\u003e7\u003c/strong\u003e:624997(2020). \u003c/li\u003e\n\u003cli\u003eDell\u0026apos;Aquila, E. et al. Prognostic and predictive factors in pancreatic cancer[J]. Oncotarget,\u003cstrong\u003e11\u003c/strong\u003e(10):924-941(2020). \u003c/li\u003e\n\u003cli\u003eShi, H. et al. Survival prediction after upfront surgery in patients with pancreatic ductal adenocarcinoma: Radiomic, clinic-pathologic and body composition analysis[J]. Pancreatology,\u003cstrong\u003e21\u003c/strong\u003e(4):731-737(2021). \u003c/li\u003e\n\u003cli\u003eHealy, G. M. et al. Pre-operative radiomics model for prognostication in resectable pancreatic adenocarcinoma with external validation[J]. Eur Radiol,\u003cstrong\u003e32\u003c/strong\u003e(4):2492-2505(2022). \u003c/li\u003e\n\u003cli\u003eTong, Z. et al. Development, Validation, and Comparison of Artificial Neural Network Models and Logistic Regression Models Predicting Survival of Unresectable Pancreatic Cancer[J]. Front Bioeng Biotechnol,\u003cstrong\u003e8\u003c/strong\u003e:196(2020). \u003c/li\u003e\n\u003cli\u003eHosny, A. et al. Deep learning for lung cancer prognostication: A retrospective multi-cohort radiomics study[J]. PLoS Med,\u003cstrong\u003e15\u003c/strong\u003e(11): e1002711 (2018). \u003c/li\u003e\n\u003cli\u003eWu, S. et al. A Radiomics Nomogram for the Preoperative Prediction of Lymph Node Metastasis in Bladder Cancer[J]. Clin Cancer Res,\u003cstrong\u003e23\u003c/strong\u003e(22):6904-6911 (2017). \u003c/li\u003e\n\u003cli\u003eZhang, Y. et al. Improving prognostic performance in resectable pancreatic ductal adenocarcinoma using radiomics and deep learning features fusion in CT images[J]. Sci Rep,\u003cstrong\u003e11\u003c/strong\u003e(1):1378 (2021). \u003c/li\u003e\n\u003cli\u003eCen, C. et al. Pancreatic Ductal Adenocarcinoma at CT: A Combined Nomogram Model to Preoperatively Predict Cancer Stage and Survival Outcome[J]. Front Oncol,\u003cstrong\u003e11\u003c/strong\u003e:594510 (2021). \u003c/li\u003e\n\u003cli\u003eLi, X. et al. Preoperative recurrence prediction in pancreatic ductal adenocarcinoma after radical resection using radiomics of diagnostic computed tomography[J]. EClinicalMedicine,\u003cstrong\u003e43\u003c/strong\u003e:101215 (2022). \u003c/li\u003e\n\u003cli\u003eChun, Y. S., Pawlik, T. M., Vauthey, J. N. 8th Edition of the AJCC Cancer Staging Manual: Pancreas and Hepatobiliary Cancers[J]. Ann Surg Oncol,\u003cstrong\u003e25\u003c/strong\u003e(4):845-847 (2018). \u003c/li\u003e\n\u003cli\u003eVasquez, M.M., Hu, C., Roe, D.J., Halonen, M., Guerra, S. Measurement error correction in the least absolute shrinkage and selection operator model when validation data are available[J]. Stat Methods Med Res,\u003cstrong\u003e28\u003c/strong\u003e(3):670-680 (2019). \u003c/li\u003e\n\u003cli\u003eVasquez, M. M. et al. Least absolute shrinkage and selection operator type methods for the identification of serum biomarkers of overweight and obesity: simulation and application[J]. BMC Med Res Methodol,\u003cstrong\u003e16\u003c/strong\u003e(1):154 (2016). \u003c/li\u003e\n\u003cli\u003eGirelli, R. et al. Results of 100 pancreatic radiofrequency ablations in the context of a multimodal strategy for stage III ductal adenocarcinoma[J]. Langenbecks Arch Surg,\u003cstrong\u003e398\u003c/strong\u003e(1):63-69 (2013). \u003c/li\u003e\n\u003cli\u003eWang, J. et al. Radiomics analysis of contrast-enhanced CT for staging liver fibrosis: an update for image biomarker[J]. Hepatol Int,\u003cstrong\u003e16\u003c/strong\u003e(3):627-639 (2022). \u003c/li\u003e\n\u003cli\u003eLin, Z. et al. Preoperative prediction of clinically relevant postoperative pancreatic fistula after pancreaticoduodenectomy[J]. Eur J Radiol,\u003cstrong\u003e139\u003c/strong\u003e:109693 (2021). \u003c/li\u003e\n\u003cli\u003eHyun, S.H. et al. Intratumoral heterogeneity of (18)F-FDG uptake predicts survival in patients with pancreatic ductal adenocarcinoma[J]. Eur J Nucl Med Mol Imaging,\u003cstrong\u003e43\u003c/strong\u003e(8):1461-1468 (2016). \u003c/li\u003e\n\u003cli\u003eBallehaninna, U.K., Chamberlain, R.S. The clinical utility of serum CA 19-9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: An evidence based appraisal[J]. J Gastrointest Oncol,\u003cstrong\u003e3\u003c/strong\u003e(2):105-119 (2012). \u003c/li\u003e\n\u003cli\u003eRofi, E. et al. The emerging role of liquid biopsy in diagnosis, prognosis and treatment monitoring of pancreatic cancer[J]. Pharmacogenomics,\u003cstrong\u003e20\u003c/strong\u003e(1):49-68 (2019). \u003c/li\u003e\n\u003cli\u003eAmaral, M.J., Oliveira, R.C., Donato, P., Tralh\u0026atilde;o, J.G. Pancreatic Cancer Biomarkers: Oncogenic Mutations, Tissue and Liquid Biopsies, and Radiomics-A Review[J]. Dig Dis Sci,\u003cstrong\u003e68\u003c/strong\u003e(7):2811-2823 (2023). \u003c/li\u003e\n\u003cli\u003eKleeff, J. et al. The impact of diabetes mellitus on survival following resection and adjuvant chemotherapy for pancreatic cancer[J]. Br J Cancer,\u003cstrong\u003e115\u003c/strong\u003e(7):887-894 (2016). \u003c/li\u003e\n\u003cli\u003eViallard, C., Larrivee, B. Tumor angiogenesis and vascular normalization: alternative therapeutic targets[J]. Angiogenesis,\u003cstrong\u003e20\u003c/strong\u003e(4):409-426 (2017). \u003c/li\u003e\n\u003cli\u003eHuang, H. et al. Risk factors and prognostic index model for pancreatic cancer[J]. Gland Surg,\u003cstrong\u003e11\u003c/strong\u003e(1):186-195 (2022). \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"630\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"4\" valign=\"top\"\u003e\n \u003cp\u003eTable 1. General characteristics of patients in the training and testing cohorts.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003eTraining Cohort\u003c/p\u003e\n \u003cp\u003e(n = 102)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003eTesting\u0026nbsp;Cohort\u003c/p\u003e\n \u003cp\u003e(n = 44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eOverall survival, months\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e9.99 \u0026plusmn; 6.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e11.00 \u0026plusmn; 7.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.410\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eNumber of deaths (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e67(65.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e32 (72.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.403\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eAge (mean \u0026plusmn; SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e63.42 \u0026plusmn; 9.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e62.27 \u0026plusmn; 9.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.499\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eSex (male/female)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e55/47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e25/19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.747\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eAbdominal pain\u0026nbsp;(with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e89/13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e37/7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.610\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eJaundice\u0026nbsp;(with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e20/82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e11/33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.465\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eDiabetes\u0026nbsp;(with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e21/81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e7/37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.510\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eTB (\u0026mu;mol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e31.65 \u0026plusmn; 55.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e45.51 \u0026plusmn; 65.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.190\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eALB (g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e41.01 \u0026plusmn; 4.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e40.21 \u0026plusmn; 3.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.761\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eCA19-9 (\u0026le;/\u0026gt;37 U/mL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e15/87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e8/36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.597\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003ecTNM stage (Ⅲ/Ⅳ)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e66/36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e32/12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.344\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eCeliac artery invasion (with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e40/62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e18/26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.848\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eCommon hepatic artery invasion (with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e49/53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e24/20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.471\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eSplenic artery invasion (with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e52/50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e21/23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.718\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eSuperior mesenteric artery invasion (with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e35/67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e19/25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.308\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003ePortal vein invasion (with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e44/58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e19/25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.996\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eSuperior mesenteric vein invasion (with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e51/51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e21/23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.801\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"47.86053882725832%\" valign=\"top\"\u003e\n \u003cp\u003eSplenic vein invasion\u0026nbsp;(with/without)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.17591125198098%\" valign=\"top\"\u003e\n \u003cp\u003e58/44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.870047543581617%\" valign=\"top\"\u003e\n \u003cp\u003e21/23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.09350237717908%\" valign=\"top\"\u003e\n \u003cp\u003e0.309\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"4\" valign=\"top\"\u003e\n \u003cp\u003eTB, total bilirubin; HDL, high-density lipoprotein; ALB, albumin\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"601\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003eTable 2. Radiomics features associated with overall survival and coefficients selected by LASSO cox regression.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\" valign=\"top\"\u003e\n \u003cp\u003eSurvival Outcome and Radiomic Features\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\" valign=\"top\"\u003e\n \u003cp\u003eCoefficient\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\" valign=\"top\"\u003e\n \u003cp\u003eOverall survival\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eOriginal_\u0026nbsp;shape_Maximum2D_DiameterSlice\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e3.358070\u0026times;10\u003csup\u003e-4\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHL_ glcm _ Contrast\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e2.403567\u0026times;10\u003csup\u003e0\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHL_\u0026nbsp;gldm _ LargeDependenceLowGrayLevelEmphasis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e-1.976730\u0026times;10\u003csup\u003e-2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHL_ glszm _ GrayLevelNonUniformityNormalized\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e-1.921688\u0026times;10\u003csup\u003e0\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHH_ glcm_ ClusterShade\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e5.702145\u0026times;10\u003csup\u003e-1\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHH_ glcm_Imc2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e-4.611064\u0026times;10\u003csup\u003e0\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHH_\u0026nbsp;gldm_HighGrayLevelEmphasis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e4.004665\u0026times;10\u003csup\u003e-4\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHH_ glszm _LargeAreaLowGrayLevelEmphasis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e1.307349\u0026times;10\u003csup\u003e-5\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-LHL_\u0026nbsp;firstorder_Mean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e-2.430718\u0026times;10\u003csup\u003e0\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-HLH_ glcm_Contrast\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e-1.577605\u0026times;10\u003csup\u003e1\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-HLH_\u0026nbsp;glcm_Idmn\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e-2.907880\u0026times;10\u003csup\u003e0\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-HLH_ ngtdm_Busyness\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e3.876596\u0026times;10\u003csup\u003e-5\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-HLL_\u0026nbsp;firstorder_Mean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e-6.142849\u0026times;10\u003csup\u003e0\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-HHH_ glszm_LargeAreaLowGrayLevelEmphasis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e1.837997\u0026times;10\u003csup\u003e-4\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.7906976744186%\"\u003e\n \u003cp\u003eWavelet-HHH_\u0026nbsp;ngtdm_Busyness\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.04318936877076%\"\u003e\n \u003cp\u003e1.758318\u0026times;10\u003csup\u003e-2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16611295681063123%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"99.83361064891847%\" colspan=\"2\"\u003e\n \u003cp\u003eAbbreviations: glszm\u0026mdash;gray-level size zone matrix; glcm\u0026mdash;gray-level co-occurrence matrix; and ngtdm\u0026mdash;neighboring gray tone difference matrix. LHH, LHL, HLH, HLL, and HHH denote the high- and low-pass filters on the x, y, and z dimensions, respectively (H\u0026mdash;high; L\u0026mdash;low).\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.16638935108153077%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"638\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"6\" valign=\"top\"\u003e\n \u003cp\u003eTable 3. Uni- and Multivariable Cox Regression Analyses for OS in the training cohort.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003eUnivariable HR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003eP\u0026nbsp;Value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003eMultivariable HR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003eP\u0026nbsp;Value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eRadiomics model riskscore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e15.788 (8.123, 30.688)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026lt; 0.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e30.184 (12.830, 71.009)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026lt; 0.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eDiabetes\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.553 (0.271, 2.047)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.068\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e1.389 (1.157, 2.967)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e0.042*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eCA19-9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e2.959 (1.074, 8.147)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.036*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003ecTNM stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.675 (1.006, 2.788)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.047*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eCeliac artery invasion\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.073 (0.656, 1.754)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.779\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eCommon hepatic artery invasion\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.056 (0.651, 1.713)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.824\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e5.046 (1.385, 18.388)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e0.014*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eSplenic artery invasion\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.669 (1.024, 2.720)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.040*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eSuperior mesenteric artery invasion\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.105 (0.675, 1.809)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003ePortal vein invasion\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.175 (0.723, 1.909)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.515\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eSuperior mesenteric vein invasion\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e0.817 (0.536, 1.414)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.576\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e\u0026hellip;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.48982785602504%\" valign=\"top\"\u003e\n \u003cp\u003eSplenic vein invasion\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.317683881064163%\" valign=\"top\"\u003e\n \u003cp\u003e1.299 (0.796, 2.118)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.015649452269171%\" valign=\"top\"\u003e\n \u003cp\u003e0.295\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.69170579029734%\" valign=\"top\"\u003e\n \u003cp\u003e7.163 (1.044, 9.606)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.172143974960877%\" valign=\"top\"\u003e\n \u003cp\u003e0.007*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.3129890453834116%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"99.68652037617555%\" colspan=\"5\" valign=\"top\"\u003e\n \u003cp\u003e*, p \u0026lt; 0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"0.31347962382445144%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"637\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"6\" valign=\"top\"\u003e\n \u003cp\u003eTable 4.\u0026nbsp;Performance of different models in\u0026nbsp;training and\u0026nbsp;testing cohorts\u0026nbsp;for the prediction of overall survival.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.3265306122449%\" rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"27.001569858712717%\" rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eVariables and models\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.29199372056515%\" rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eAUC (95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"34.37990580847724%\" colspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"33.027522935779814%\" valign=\"top\"\u003e\n \u003cp\u003e1vs.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"29.357798165137616%\" valign=\"top\"\u003e\n \u003cp\u003e1vs.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"37.61467889908257%\" valign=\"top\"\u003e\n \u003cp\u003e2vs.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.352201257861637%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eTraining\u0026nbsp;Cohort\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"27.044025157232703%\" valign=\"top\"\u003e\n \u003cp\u003e1. Radiomics model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.32704402515723%\" valign=\"top\"\u003e\n \u003cp\u003e0.981 (0.962-1.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.320754716981131%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026lt;\u0026nbsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.062893081761006%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e0.047\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.89308176100629%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026lt;\u0026nbsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"54.77707006369427%\" valign=\"top\"\u003e\n \u003cp\u003e2. Clinicopathologic\u0026nbsp;model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"45.22292993630573%\" valign=\"top\"\u003e\n \u003cp\u003e0.549 (0.431-0.667)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"54.77707006369427%\" valign=\"top\"\u003e\n \u003cp\u003e3. Combined model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"45.22292993630573%\" valign=\"top\"\u003e\n \u003cp\u003e0.990 (0.977-1.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.352201257861637%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eTesting Cohort\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"27.044025157232703%\" valign=\"top\"\u003e\n \u003cp\u003e1. Radiomics model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"22.32704402515723%\" valign=\"top\"\u003e\n \u003cp\u003e0.901 (0.812-0.990)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.320754716981131%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.062893081761006%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e0.311\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.89308176100629%\" rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026lt;\u0026nbsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"54.77707006369427%\" valign=\"top\"\u003e\n \u003cp\u003e2. Clinicopathologic model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"45.22292993630573%\" valign=\"top\"\u003e\n \u003cp\u003e0.556 (0.382-0.730)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"54.77707006369427%\" valign=\"top\"\u003e\n \u003cp\u003e3. Combined model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"45.22292993630573%\" valign=\"top\"\u003e\n \u003cp\u003e0.949 (0.889-1.000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"6\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\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":"Chemotherapy, Computed Tomography, Pancreatic Ductal Adenocarcinoma, Radiomics, Survival","lastPublishedDoi":"10.21203/rs.3.rs-4284696/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4284696/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cem\u003eObjective: \u003c/em\u003eTo investigate the ability of preoperative CT-based radiomics signature to be used as prognostic indicators of unresectable pancreatic ductal adenocarcinoma (PDAC) after chemotherapy and develop radiomics-based prediction models.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMethods: \u003c/em\u003eClinicopathological data and radiomics features of 146 pancreatic cancer patients (102 in the training cohort and 44 in the testing cohort) were retrospectively reviewed. Radiomics features were selected by the lasso cox regression model. Multivariate Cox regression analysis was used to establish the radiomics model, clinicopathologic model, and combined model in the training cohort to predict survival of PDAC after chemotherapy, and then verified in the testing cohort.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eResults: \u003c/em\u003eOf 855 extracted radiomics features in portal venous CT images, 15 most stable features were selected. Multivariate Cox regression analysis identified radiomics model riskscore, diabetes, common hepatic artery invasion, and splenic vein invasion to construct a combined model for predicting overall survival (OS) of unresectable PDAC (with AUC of 0.949 in the validation set; 95% CI 0.889-1.000). While the AUC of the radiomics model and clinicopathologic model were 0.901 (95% CI, 0.812-0.990) and 0.556 (95% CI, 0.382-0.730), respectively.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eConclusion: \u003c/em\u003eCT radiomics signature is a powerful predictor associated with OS in patients with unresectable PDAC after chemotherapy. The combined model performed well in predicting survival.\u003c/p\u003e","manuscriptTitle":"CT radiomics for Predicting Survival in Patients with Unresectable Pancreatic Ductal Adenocarcinoma after Chemotherapy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-29 16:39:13","doi":"10.21203/rs.3.rs-4284696/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":"aaa051d2-2284-4bc5-b2b2-cd950f46a366","owner":[],"postedDate":"April 29th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":31174027,"name":"Biological sciences/Cancer/Cancer imaging"},{"id":31174028,"name":"Health sciences/Risk factors"},{"id":31174029,"name":"Health sciences/Health care/Prognosis"}],"tags":[],"updatedAt":"2024-05-27T05:44:14+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-29 16:39:13","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4284696","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4284696","identity":"rs-4284696","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}