Overall Survival C-index Research Articles

Association of postoperative radiotherapy with survival in patients with completely resected pathologic N2 non-small cell lung cancer treated with chemotherapy.

e21008 Background: The current staging system for completely resected pathological N2 non-small-cell lung cancer (NSCLC) treated with chemotherapy is not suitable for predicting those patients most likely to benefit or not from post-operative radiotherapy (PORT). This study aimed to construct a survival prediction model that will enable individualized predictions of the net survival of PORT. Methods: A total of 3094 cases between 2002 and 2014 were extracted from the Surveillance, Epidemiology, and End Results databases. Patient characteristics were included as covariates, and their association for overall survival (OS) with and witout PORT was assessed. Externally validate data of 602 patients were included from China. Results: A total of 3696 patients were included for analysis. Age, gender, examined lymph node, positive lymph node, tumor size, extent of surgery, and visceral pleural invasion were significantly associated with OS, with P < .05. The calibration curve for OS showed great agreement between prediction by survival prediction model and actual observation.In the training cohort, the C-index for OS was 0.619 (95% CI, 0.598-0.641) in the PORT group and 0.627 (95% CI, 0.605-0.648) in the non-PORT group. In the externally validation cohort, the C-index for OS was 0.599 (95% CI, 0.485- 0.713) in the PORT group and 0.595 (95% CI, 0.544-0.646) in non-PORT group. The two survival prediction models were developed based on clinical variables to estimate an individual's net survival gain or not attributable to PORT. We found that PORT could improve OS (HR, 0.861; P = 0.044) for patients with a positive PORT net survival benefit. Conclusions: We established a practical survival prediction model that can be used to make individualized predictions about the expected survival of PORT or without PORT in patients with completely resected pathological N2 NSCLC, treated with chemotherapy.

Development and validation of prognostic nomograms in patients with adrenocortical carcinoma: a population-based study.

Predicting the prognosis of patients with adrenocortical carcinoma (ACC) is difficult, due to its unpredictable behavior. The aim of this study is to develop and validate a nomogram to predict survival outcomes in patients with ACC. Nomograms were established using the data collected from the Surveillance, Epidemiology, and End Results (SEER) database. Based on univariate and multivariate Cox regression analyses, we identified independent risk factors for overall survival (OS) and cancer-specific survival (CSS). Concordance indexes (c-indexes), the area under the receiver operating characteristics curve (AUC) and calibration curve were used to evaluate predictive performance of these models. The clinical use of nomogram was measured by decision curve analysis (DCA) and clinical impact curves. A total of 855 eligible patients, randomly divided into training (n = 600) and validation cohorts (n = 255), were included in this study. Based on the independent predictors, the nomograms were established and demonstrated good discriminative abilities, with C-indexes for OS and CSS were 0.762 and 0.765 in training cohorts and 0.738 and 0.758 in validation cohorts, respectively. The AUC and calibration plots also demonstrated a good performance for both nomograms. DCA indicated that the two nomograms provide clinical net benefits. We unveiled the prognostic factors of ACC and developed novel nomograms that predict OS and CSS more accurately and comprehensively, which can help clinicians improve individual treatment, making proper clinical decisions and adjusting follow-up management strategies.

Construction and Validation of Nomograms for Predicting Overall Survival and Cancer-Specific Survival in Nonmetastatic Inflammatory Breast Cancer Patients Receiving Tri-Modality Therapy: A Population-Based Study.

BackgroundAs the most aggressive breast cancer, inflammatory breast cancer (IBC) has a poor prognosis. However, analyzing the prognostic factors of IBC is challenging due to its rarity. We identified the prognostic factors to establish predictive tools for survival in nonmetastatic IBC patients who received tri-modality therapy.Material/MethodsThe data of 893 nonmetastatic IBC patients were acquired from the Surveillance, Epidemiology, and End Results (SEER) database. IBC was identified by “ICD-O-3=8530” or “AJCC T, 7th=T4d”). Patients were randomized to the training (n=668) and validation (n=225) cohorts. Prognostic factors were identified in the training cohort. Factors in the nomogram for overall survival (OS) were filtered by the least absolute shrinkage selection operator (LASSO) regression model. Factors selected by the competing-risk models were integrated to construct nomograms for breast cancer-specific survival (BCSS). Nomogram validation was performed in both cohorts.ResultsThe number of positive lymph nodes contributed the most to both nomograms. In the validation cohort, the C-indexes for OS and BCSS were 0.724 and 0.727, respectively. Calibration curves demonstrated acceptable agreement between predicted and actual survival. Risk scores were calculated from the nomograms and used to split patients into the low-risk and high-risk groups. Smooth hazard ratio (HR) curves and Kaplan-Meier curves showed a statistically significant difference in prognosis between the high-risk group and low-risk group (log-rank P<0.001).ConclusionsWe unveiled the prognostic factors of nonmetastatic IBC and formulated nomograms to predict survival. In these models, the likelihood of individual survival can be easily calculated, which may assist clinicians in selecting treatment regimens.

Establishment of a Predictive Model of GvHD-Free, Relapse-Free Survival after Allogeneic Hematopoietic Stem Cell Transplantation Using a Machine Learning Algorithm

Background: GvHD-free, relapse-free survival (GRFS) is a useful composite endpoint that measures survival without relapse or significant morbidity after allogeneic hematopoietic stem cell transplantation (allo-HSCT). We aimed to establish a predictive model of GRFS after allo-HSCT, using a machine learning algorithm within the Kyoto Stem Cell Transplantation Group (KSCTG), a multi-institutional joint research group of 17 transplantation centers in Japan. Methods: Consecutive patients with hematologic diseases who had undergone their first allo-HSCT between 1986 and 2016, were included in the study (n = 2,565). The primary endpoint was GRFS, and the secondary endpoints were: overall survival (OS), relapse, non-relapse mortality, and grades II-IV acute GvHD. Data were randomly split into the training set (70% of the dataset) and the validation set (30% of the dataset). The model was trained and tested using a 5-fold cross validation on the training set. A stacking ensemble of multiple machine learning algorithms was applied to develop a model. Predictions were first generated using different machine learning algorithms such as AdaBoost, Random Survival Forest, and Dynamic-DeepHit. A metamodel was subsequently trained using these predictions as inputs, to generate the final prediction (Figure 1). All models were trained using 28 input variables containing information on the patient and transplant characteristics. We used Cox Proportional-Hazards (Cox-PH) model as the benchmark case and evaluated the models' performance using the Harrell's Concordance Index (C-index). The model's final prediction can be directly used as a risk score for each patient. We validated this risk score in the validation set using the competing risk hazard model. Due to the small size of the validation set (N=770), we only classified the final risk scores into 2 groups as follows: high risk (above median) and low risk (below median) for each risk category. Results: The median age of the patients at transplantation was 48 years (range from 7 to 76). The sources of stem cells were related bone marrow (BM) in 650 patients, related peripheral blood (PB) in 382, unrelated BM in 978, unrelated PB in 24, and cord blood in 522. Myeloablative conditioning was administered in 1,293 patients and reduced-intensity conditioning in 949 patients. Refined disease risk indexes were low in 310 patients, intermediate in 1,197 patients, high in 545 patients, and very high in 368 patients. Median follow-up period of survivors was 4.5 years. For GRFS, our model achieved a higher C-index across all risk categories (C-index: 0.631) than other top-of-the-art competing risk models (Cox-PH: 0.620, Random Survival Forest: 0.627, Gradient Boosting: 0.609, Component-wise Gradient Boosting: 0.599, Dynamic DeepHit: 0.603). The model also showed the highest C-index for overall survival (C-index: 0.653), relapse (C-index: 0.703), non-relapse mortality (C-index: 0.636), and grades II-IV acute GvHD (C-index: 0.594). For GRFS, the probability after 1 year and 2 year was 24.9% and 15.2% for the high risk and 48.0% and 33.6% for the low risk, respectively (hazard ratio [HR] compared to the low risk: 1.88; 95% CI: 1.60-2.20) (Figure 2). The OS at 5 years was 43.4% for the high risk and 70.6% for the low risk (hazard ratio [HR] compared to the low risk: 2.41; 95% CI: 1.96-2.97). The cumulative incidence function (CIF) for non-relapse mortality at 5 years was 32.1% for the high risk and 21.4% for the low risk (HR 2.01; 95% CI: 1.38-2.91). The CIF for relapse at 5 years was 37.1% for the high risk and 16.4% for the low risk (HR 1.67; 95% CI: 1.25-2.23). Conclusions: A predictive model of GRFS after allo-HSCT was established, which could benefit physicians and patients during the clinical decision-making process. This also forms the basis for research on new prognostic biomarkers by evaluating their predictive power based on our model. Disclosures Kanda: MSD: Honoraria; Takeda: Honoraria; Bristol-Meyers Squib: Honoraria; Otsuka: Honoraria; Daiichi Sankyo Company: Honoraria; Novartis: Honoraria; Celgene: Honoraria; Astellas: Honoraria; Chugai: Honoraria; Kyowa Hakko Kirin: Honoraria; JCR Pharmaceuticals: Honoraria; NextGeM Incorporation: Patents &amp; Royalties: 2019-011392. Imada:Bristol-Meyer Squibb K.K.: Honoraria; Celgene K.K.: Honoraria; Otsuka Pharmaceutical Co., Ltd.: Honoraria; Ono Pharmaceutical Co., Ltd.: Honoraria; Kyowa Hakko Kirin Co., Ltd.: Honoraria; Chugai Pharmaceutical Co., Ltd.: Honoraria; Nippon Shinyaku Co.,Ltd.: Honoraria; Takeda Pharmaceutical Co.,LTD.: Honoraria; Novartis Pharma K.K.: Honoraria; Astellas Pharma Inc.: Honoraria. Takaori-Kondo:Takeda: Research Funding; Ono: Research Funding; Bristol-Myers Squibb: Honoraria, Research Funding; Celgene: Honoraria, Research Funding; Novartis: Honoraria; Kyowa Kirin: Research Funding; Chugai: Research Funding; Janssen: Honoraria; Pfizer: Honoraria.

287O - Development and validation of a risk model integrating plasma Epstein-Barr virus DNA (EBV DNA) level and TNM stage for stratification of nasopharyngeal cancer (NPC) to adjuvant therapy

BackgroundClinical guidelines for treatment decision in NPC are mainly based on the anatomical classification by UICC TNM staging. The concentration of plasma EBV DNA measured after radiotherapy (RT) or chemoradiation (CRT) is highly prognostic and independent of UICC stage, which may be useful in risk stratification of NPC patients to adjuvant therapy. MethodsFor model development, we used the prospective multi-center 0502 EBV DNA screening cohort (recruitment period 2006 - 2015; n = 745). Eligible patients had histologically confirmed NPC of stage II-IVB (UICC 7th Edition) and post-RT EBV DNA measured in plasma, no loco-regional disease or distant metastasis after RT/CRT and received no adjuvant therapy. Primary endpoint was overall survival (OS). We used recursive-partitioning analysis (RPA) to classify patients into groups of low-, intermediate- and high-risk of death. For internal validation, we pooled independent patient cohorts from previous published biomarker studies (1997-2006; n = 340). For external validation, we used external cohort of NPC patients treated at Sun Yat-sen University Cancer Center (2009 - 2012; n = 837) using SYSU EBV DNA test. ResultsRPA classified NPC patients based on the post-RT plasma EBV DNA level and UICC stage into three distinct prognostic groups (Table). RPA low risk group shared similar 5-yr OS (89.4%; 95% CI = 86.4-92.5%) as UICC stage II (88.5%; 84.0-93.1%) but included 2.3x number of patients that could be potentially spared of adjuvant therapy toxicity. The overall C-index of OS was 0.7118 for RPA risk group, compared to 0.6042 for TNM stage and 0.6747 for EBV DNA (both p < 0.01). RPA risk group has improved hazard discrimination and calibration than either TNM stage or plasma EBV DNA level. The result was validated in both internal and external cohorts.Table287OTablePatient No. (%)5-yr OSHR (95% C.I.)PUICC TNM stage (7th Ed)1) Stage II209 (28.1)88.5-2) Stage III368 (49.4)81.01.50 (0.99-2.25)0.0543) Stage IVAB168 (22.6)69.43.01 (1.97-4.59)<0.0001Post-RT plasma EBV DNA (copies/ml)1) 0573 (76.9)87.3-2) 1-4974 (9.9)83.21.32 (0.76-2.27)0.32593) 50-49959 (7.9)50.53.85 (2.55-5.81)<0.00014) > =50039 (5.2)28.311.59 (7.60-17.67)<0.0001RPA risk group1) Low risk EBV DNA 0 and stage II/III 1-49 and stage II483 (64.8)89.4-2) Intermediate risk EBV DNA 0 and stage IVAB 1-49 and stage III/IVAB 50-499 and stage II176 (23.6)78.52.40 (1.65-3.50)<0.00013) High risk EBV DNA 50-499 and stage III/IVAB >500 and any stage86 (11.5)37.28.54 (5.93-12.29)<0.0001 ConclusionsIncorporation of post-RT plasma EBV DNA level into UICC TNM stage improved risk stratification of NPC patients to adjuvant therapy. Clinical trial identificationIdentifier: NCT00370890. Legal entity responsible for the studyComprehensive Cancer Trials Unit, Department of Clinical Oncology, The Chinese University of Hong Kong. FundingHas not received any funding. DisclosureE.P. Hui: Advisory / Consultancy, Research grant / Funding (institution): Merck Sharp & Dohme; Speaker Bureau / Expert testimony: Merck Serono; Research grant / Funding (institution): Pfizer. W.K.J. Lam: Shareholder / Stockholder / Stock options: Grail. K.C.A. Chan: Advisory / Consultancy, Shareholder / Stockholder / Stock options: Grail; Leadership role, Shareholder / Stockholder / Stock options: Take2; Leadership role, Shareholder / Stockholder / Stock options: DRA. Y.M.D. Lo: Advisory / Consultancy, Leadership role, Shareholder / Stockholder / Stock options: Grail; Leadership role, Shareholder / Stockholder / Stock options: Take2 Health; Leadership role, Shareholder / Stockholder / Stock options: DRA. A.T.C. Chan: Research grant / Funding (institution): Pfizer; Research grant / Funding (institution): Boehringer Ingelheim; Research grant / Funding (institution): Bristol-Myers Squibb; Research grant / Funding (institution): Merck Sharp & Dohme. All other authors have declared no conflicts of interest.

Individualized immune-related gene signature predicts immune status and oncologic outcomes in clear cell renal cell carcinoma patients

ObjectiveTo develop an individualized immune-related gene signature that predicts oncologic outcomes and immune status of clear cell renal cell carcinoma (ccRCC). Materials and MethodsThe immune-related gene pair (IGP) index was constructed and validated based on pairwise comparison in 634 ccRCC patients. Association with overall survival (OS), progression-free interval (PFI), and disease-specific survival (DSS) was evaluated by Kaplan-Meier analysis, univariate and multivariate cox regression survival analysis. Prognostic values of different risk models were compared using Harrell's C-index. ResultsThe IGP index of 17 gene pairs was an adverse independent risk factor in multivariate analyses for OS (hazard ratio [HR], 1.718; P = 0.001), PFI (HR, 1.550; P = 0.006), and DSS (HR, 2.201; P = 0.001) in ccRCC patients. It showed comparable prognostic accuracy with ccA/ccB signature (C-index for OS, 0.657 vs. 0.640; P = 0.686) and better intratumor homogeneity. Immunosuppressive immune cell, markers, and pathways were all enriched in high immune-risk tumors. The integrated immune-clinical prognostic score outperformed ccA/ccB signature and University of California Integrated Staging System risk model in terms of C-index for estimation of OS, PFI, and DSS (P < 0.001). ConclusionsThe proposed IGP index is a robust and promising biomarker for estimating oncologic outcomes in ccRCC. High immune-risk tumors are immunosuppressive.

Contemporary Update on Prognostic Nomograms for Endemic Nasopharyngeal Carcinoma: Real-World Results from a Big-Data Intelligence Platform-Based Analysis

Background: To investigate the survival outcomes and update effective prognostic nomograms of nasopharyngeal carcinoma (NPC) in an endemic area. Methods: A retrospective cohort of 10126 patients who underwent radical intensity-modulated radiotherapy (IMRT) at Sun Yat-sen University Cancer Center (SYSUCC) from April 2009 to December 2015. We assigned patients into a training cohort (SYSUCC-A; n=6751) and an internal validation cohort (SYSUCC-B; n=3375) based on computer-generated random numbers. Patients collected from Wuzhou Red Cross Hospital (WZRCH) between February 2012 and July 2015 were used as the independent external validation cohort (WZRCH; n=450). Concordance index (C-index) was used to determine predictive accuracy and discriminative ability of the nomogram. Overall survival (OS) was the primary endpoint. Secondary endpoints consisted of local relapse-free survival (LRFS), regional relapse-free survival (RRFS), distant metastasis-free survival (DMFS), and disease-free survival (DFS). All statistical tests were two-sided. Findings: The three- and five-year OS, LRFS, RRFS, DMFS, and DFS rates were 92.3% and 86.5%, 95.4% and 93.4%, 96.0% and 94.3%, 89.5% and 87.0%, and 83.4% and 78.4%, respectively. Gender, age, T stage, and N stage were identified independent risk factors that predicted OS, which were all assembled into the nomogram and derived from multivariable analysis of the SYSUCC-A cohort. The nomogram c-index for OS was 0.725 (95% CI, 0.679-0.772). Calibration curve for probability of death revealed the nomogram-based predictions were in good agreement with actual observations. C-index of the nomograms for LRFS, RRFS, DMFS, and DFS were 0.684 (95% CI, 0.611-0.756), 0.650 (95% CI, 0.571-0.728), 0.680 (95% CI, 0.630-0.731), and 0.669 (95% CI, 0.629-0.710), respectively. All results were confirmed in the validation cohorts. Interpretation: Patients with NPC had satisfactory locoregional survival outcomes undergoing IMRT in an endemic area, and distant metastasis predominates as the pattern of disease relapse. These validated nomograms could serve as useful tool for predicting survival outcomes and providing patient counseling. Funding Statement: This work was supported by the Special Support Program of Sun Yat-sen University Cancer Center (16zxtzlc06), the Health & Medical Collaborative Innovation Project of Guangzhou City, China (201604020003), the Natural Science Foundation of Guang Dong Province (No. 2017A030312003), Health & Medical Collaborative Innovation Project of Guangzhou City, China (201803040003), the Innovation Team Development Plan of the Ministry of Education (No. IRT_17R110), the Overseas Expertise Introduction Project for Discipline Innovation (111 Project, B14035) and the National Natural Science Foundation of China (No. 81802707). Declaration of Interests: The authors declare no potential conflicts of interest. Ethics Approval Statement: All patients provided informed consent, and consent waiver for the rest of the cohort was approved by the respective ethical review boards. Patients’ clinical and demographic information were anonymized before analysis.

Nomograms for predicting long-term overall survival and cancer-specific survival in patients with primary urethral carcinoma: a population-based study.

Our aim was to identify the independent prognostic factors in patients with primary urethral carcinoma (PUC) and to predict their overall survival (OS) and cancer-specific survival (CSS) at 3, 5, and 8years. Patients with PUC identified in the Surveillance, Epidemiology, and End Results (SEER) database were divided into training and validation cohorts. Nomograms were constructed based on the results of Cox regression analysis. The predictive performance of each nomogram was evaluated using the consistency index (C-index), the area under the receiver operating characteristics curve (AUC), and calibration plots. Decision-curve analysis (DCA) was used to test the clinical value of the predictive models. Our study screened 822 patients with PUC. Multivariate analysis showed that the age at diagnosis, race, histology, American Joint Committee on Cancer (AJCC) stage, and surgery status were independent prognostic factors for CSS and age at diagnosis, race, histology, AJCC stage, surgery status, and chemotherapy for OS (all P < 0.05). We used these prognostic factors to construct nomograms. The C-indexes for OS and CSS were 0.713 and 0.741 in training cohorts and 0.714 and 0.738 in validation cohorts, respectively. The AUC and calibration plots demonstrated the good performance of both nomograms. The DCA indicated the presence of clinical net benefits in both the training and validation cohorts. We developed and validated nomograms for predicting OS and CSS in patients with PUC, which can help clinicians make treatment decisions.

Validation of the GRade, Age, Nodes and Tumor (GRANT) score within the Surveillance Epidemiology and End Results (SEER) database: A new tool to predict survival in surgically treated renal cell carcinoma patients

The purpose of the present study was to validate the new GRade, Age, Nodes and Tumor (GRANT) score for renal cell carcinoma (RCC) prognostication within a large population of patients. Within the Surveillance, Epidemiology, and End Results database, we identified patients with either clear-cell or papillary RCC, who underwent nephrectomy between 2001 and 2015. Harrell’s C-Index, calibration plot and decision curve analysis were used to validate the GRANT model using a five-risk group stratification (0 vs. 1 vs. 2 vs. 3 vs. 4 risk factors). The primary endpoint was overall survival (OS) at 60 months. The analyses were repeated according to the histologic subgroup. The overall population included 73217 cases; 60900 with clear-cell RCC and 12317 with papillary histology, respectively. According to a five-risk group stratification, 23985 patients (32.8%) had no risk factor (0), 35019 (47.8%) had only one risk factor (1), 13275 (18.1%) had risk score 2, 854 (1.2%) had 3 risk factors and 84 (0.1%) of cases had a GRANT score of 4, respectively. At 60 months, OS rates as determined by the GRANT score were respectively 94% (score 0) vs. 86% (score 1) vs. 76% (score 2) vs. 46% (score 3) vs. 16% (score 4). In both histologic subtypes, the GRANT score yielded good calibration and high net benefit. OS C-Index values were 0.677 and 0.650 for clear-cell and papillary RCC at 60 months after surgery, respectively. In conclusion, the GRANT score was validated with a five-risk group stratification in a huge population from the SEER database, offering a further demonstration of its reliability for prognostication in RCC.

Nomograms Predict Cancer-Specific and Overall Survival of Patients With Primary Limb Leiomyosarcoma.

To date, there have been no data to predict the survival of patients with leiomyosarcoma from soft limb tissue because of the rarity of this disease. Nomograms have been widely applied in clinical oncology to precisely predict the survival of individual patients. This was a retrospective study to construct and validate nomograms to predict the cancer-specific survival (CSS) and overall survival (OS) of patients with primary limb leiomyosarcoma (PL-LMS). A total of 1,208 patients with LMS from limb soft tissue were collected from the Surveillance, Epidemiology, and End Results database from 1975 to 2015. We identified independent prognostic factors using univariate and multivariate Cox analyses. These prognostic factors were then included in the nomograms to predict 3- and 5-year CSS and OS rates. Finally, we validated the nomograms internally and externally. A total of 1208 patients were collected and divided into validation (N = 604) and training (N = 604) groups. Age, race, grade, tumor size, stage, and surgical types were demonstrated as independent prognostic factors for CSS and OS (all p < 0.05) and further used to construct the nomograms. The concordance index (C-index) for CSS was 0.857 for internal validation and 0.727 for external validation. The C-index for OS and CSS both demonstrated that the nomogram prediction agreed perfectly with actual survival. We developed nomograms to predict CSS and OS in PL-LMS patients and can benefit from using them to identify patients' mortality risk and make more precise assessments regarding survival. © 2019 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 37:1649-1657, 2019.

A nomogram to predict prognosis in Ewing sarcoma of bone

ObjectiveThis study was designed to develop a nomogram for assessing the survival of patients with Ewing sarcoma (ES). MethodsData from patients diagnosed with ES between 2004 and 2013 were collected from the Surveillance, Epidemiology, and End Results (SEER) database. Based on patient registration, the primary cohort was divided into a training set (n = 479, data from 17 cancer registries) and a validation set (n = 137, data from 1 cancer registry). Then, the prognostic effects of variables were analyzed using Kaplan–Meier method and Cox proportional hazard model. Moreover, nomograms were established for estimating 3- and 5-year overall survival (OS) and cancer-special survival (CSS) based on Cox regression model. Last, nomogram was validated by training set and validation set. ResultsAccording to the multivariate analysis of training set, nomogram which combined age, race, stage, tumor site, tumor size and chemotherapy was identified. The internal bootstrap resampling approach suggested the nomogram had sufficient discriminatory power with the C-index of OS: 0.754 (95% CI, 0.705–0.802) and CSS: 0.759 (95% CI, 0.700–0.800). The calibration plots also demonstrated good consistence between the prediction and the observation. ConclusionOur nomogram is a reliable and powerful tool for distinguishing and predicting the survival of ES patients, thus helping to better select medical examinations and optimize treatment options in collaboration with medical oncologists and surgeons.

Automated Survival Prediction in Metastatic Cancer Patients Using High-Dimensional Electronic Medical Record Data.

Oncologists use patients' life expectancy to guide decisions and may benefit from a tool that accurately predicts prognosis. Existing prognostic models generally use only a few predictor variables. We used an electronic medical record dataset to train a prognostic model for patients with metastatic cancer. The model was trained and tested using 12 588 patients treated for metastatic cancer in the Stanford Health Care system from 2008 to 2017. Data sources included provider note text, labs, vital signs, procedures, medication orders, and diagnosis codes. Patients were divided randomly into a training set used to fit the model coefficients and a test set used to evaluate model performance (80%/20% split). A regularized Cox model with 4126 predictor variables was used. A landmarking approach was used due to the multiple observations per patient, with t0 set to the time of metastatic cancer diagnosis. Performance was also evaluated using 399 palliative radiation courses in test set patients. The C-index for overall survival was 0.786 in the test set (averaged across landmark times). For palliative radiation courses, the C-index was 0.745 (95% confidence interval [CI] = 0.715 to 0.775) compared with 0.635 (95% CI = 0.601 to 0.669) for a published model using performance status, primary tumor site, and treated site (two-sided P < .001). Our model's predictions were well-calibrated. The model showed high predictive performance, which will need to be validated using external data. Because it is fully automated, the model can be used to examine providers' practice patterns and could be deployed in a decision support tool to help improve quality of care.

1512O - Automated survival prediction in metastatic cancer patients using high-dimensional electronic medical record data

Background: Oncologists use patients’ life expectancy to guide medical decisions and may benefit from a tool that provides accurate, unbiased assessments of prognosis. Existing prognostic models generally use only a few predictor variables. We used a large electronic medical record dataset to train a prognostic model for patients with metastatic cancer. Methods: The model was trained and tested using data from 12,588 patients treated for metastatic cancer in the Stanford Health Care system from 2008-2017. Data sources included provider note text, labs, vital signs, procedures, medication orders, and diagnosis codes. Patients were divided randomly into training and test sets (80%/20% split). A regularized Cox proportional hazards model with 4,126 predictor variables was fit to the training set and evaluated on the test set. A landmarking approach was used due to the multiple observations per patient, with t0 set to the time of metastatic cancer diagnosis. Performance was also evaluated using 399 palliative radiation courses in test set patients. An existing published model that uses performance status, primary tumor site, and treated site was used as a baseline [Chow, JCO 2008:20;26(36)]. Results: From the first visit after metastatic cancer diagnosis, median follow-up was 14.5 months and median overall survival was 20.9 months. Patients were seen for 384,402 daily visits. The prognostic model’s C-index for overall survival was 0.79 in the test set (averaged across landmark times). For palliative radiation courses, the C-index was 0.75 (95% CI 0.72-0.78), compared to 0.64 (95% CI 0.60-0.67) for an existing published model (p < 0.001).Table: 1512OPredicted vs actual survival for 2,518 test set patients at landmark time t0 (first visit after diagnosis of metastatic cancer)Predicted median survival in monthsActual median survival in months (95% CI)0-3 (n = 106)1.3 (0.9-2.0)3.1-6 (n = 172)3.7 (2.5-5.2)6.1-12 (n = 382)6.7 (5.6-7.9)>12 (n = 1858)35.7 (31.8-39.1) Open table in a new tab Conclusions: The model showed high predictive performance, which was significantly better than that of an existing model. Because it is fully automated, the model can be used to examine providers’ practice patterns and could deployed in a decision support tool to help improve quality of care. Legal entity responsible for the study: Michael Gensheimer. Funding: National Institutes of Health. Disclosure: M.F. Gensheimer: Research funding: Varian Medical Systems, Philips Healthcare. E. Cho: Employee of Genentech. D. Rubin: Research funding: Philips Healthcare. D.T. Chang: Research funding, honoraria: Varian Medical Systems; Stock ownership: ViewRay. All other authors have declared no conflicts of interest.

HBV-DNA Load-Related Peritumoral Inflammation and ALBI Scores Predict HBV Associated Hepatocellular Carcinoma Prognosis after Curative Resection.

Background Both persistent inflammatory activity and liver function damage contribute to a poor prognosis of hepatocellular carcinoma (HCC). This study aimed to develop nomograms that incorporate hepatitis virus B (HBV)-related peritumoral inflammation score (PIS) and liver function based on ALBI score to predict postoperative outcomes of HCC. Methods The prognostic roles of HBV-related preoperative PIS and ALBI scores in HCC recurrence were examined, and then two nomograms were constructed. The predictive accuracy and discriminative ability of the nomograms were compared with AJCC and BCLC staging systems of HCC. Results PIS (HBV-PIS) and ALBI scores (HBV-ALBI) with different HBV-DNA loads had association with overall survival (OS) and/or recurrence-free survival (RFS) of HCC. The independent predictors of OS and RFS were incorporated into the corresponding nomograms. In the training cohort, the C-indexes of OS and RFS nomograms were 0.751 and 0.736, respectively. ROC analyses showed that both OS and RFS nomograms had larger AUC (0.775 and 0.739, respectively) than AJCC and BCLC staging systems. These results were verified by the internal and external validation cohorts. Conclusion The proposed nomograms, including HBV-DNA load-related PIS and ALBI scores, were accurate in predicting survival for HCC after curative resection.

Pathological staging criteria for carcinoma in the remnant stomach: an exploratory study

To explore the prognostic value of the tumor-ratio-metastasis (TRM) staging system for carcinoma in the remnant stomach(CRS). Clinicopathological data of 91 CRS patients who underwent surgery at Peking University People's Hospital between March 1992 and December 2017 were retrospectively analyzed. According to the ratio of metastatic lymph node to dissected lymph node, the R staging was obtained, and the pN staging was replaced by the R staging to create the TRM staging. To compare the predictive accuracy of TRM and tumor-node-metastasis (TNM, UICC version 7), the R staging and pN staging were included in the prognostic factor analysis model, and the survival curve, c-index, and 95% confidence interval (CI) of the TRM staging and TNM staging system were compared. A higher c-index value means higher prediction accuracy. Of 91 CRS patients, 77 were male and 14 were female with the mean onset age of (65.2±10.4) years. The mean interval from the first operation to CRS onset was 156(6-600) months. The primary diseases of 49(53.8%) cases were benign and of 42(46.2%) cases were malignant. The median number of retrieved lymph node (RLN) was 8 (0-38), and 64 patients (70.3%) had an RLN ≤15. Lymph node metastasis occurred in 50 patients (54.9%). pN staging result was as follows: 41 cases in N0 stage, 14 in N1 stage, 19 in N2 stage, and 17 in N3 stage. R staging result was as follows: 41 cases in R0 stage, 4 in R1 stage, 19 in R2 stage, and 27 in R3 stage. TNM staging result was as follows: 13 cases in stage I(, 25 in stage II(, 10 in stage III(a, 23 in stage III(b, and 6 in stage III(c. TRM staging result was as follows: 13 cases in stage I(, 24 in stage II(, and 4 in stage III(a, 18 in stage III(b, and 18 in stage III(c. Univariate analysis showed that tumor diameter ≥7 cm (HR=2.696, 95%CI: 1.307-5.563, P=0.007), T3-4 stage (HR=4.350, 95%CI: 1.949-9.707, P=0.000), N2-3 stage (HR=1.883, 95%CI: 1.167-3.038, P=0.009), R2-3 stage (HR=1.642, 95%CI: 1.026-2.628, P=0.039), TNM III(-IIII( stage (HR=2.448, 95%CI:1.490-4.021, P=0.000), and TRM III(-IIII( stage (HR=2.504, 95%CI:1.515-4.137, P=0.000) were related to prognosis. Tumor diameter, pT staging, and pN staging were included in the Cox multivariate analysis, and the result showed that pT staging (HR=5.507, 95%CI:2.254-13.454, P=0.000) and pN staging (HR=1.698, 95%CI: 1.022-2.789, P=0.041) were independent risk factors for overall survival of CRS in this group. While R staging replaced pN staging and was included in the Cox multivariate analysis together with tumor diameter and pT staging, the result showed that R staging was not an independent risk factor for CRS in this group (HR=1.622, 95%CI: 0.866-2.329, P=0.164). Survival curve revealed pN and TNM staging systems provided better stratified curves according to each staging than R and TRM staging systems. The overall survival c-index of TNM and TRM staging systems was 0.813(95%CI: 0.732-0.826) and 0.809(95%CI: 0.741-0.847) respectively, and no significant difference in predictive accuracy was found (P=0.693). In 42 patients with primary malignance, the overall survival c-index of TNM and TRM staging systems was 0.774(95%CI: 0.589-0.901) and 0.761(95%CI: 0.596-0.912) respectively, and there was no significant difference in predictive accuracy as well (P=0.881). TRM staging is not superior to TNM staging (7th UICC) in evaluating the resected samples of CRS.

Elevated expression of IFN-inducible CXCR3 ligands predicts poor prognosis in patients with non-metastatic clear-cell renal cell carcinoma.

IFN-inducible CXCR3 ligands (ICL), namely CXCL9, CXCL10 and CXCL11, exhibit pleiotropic roles in orchestrating immunity and angiogenesis. However, the prognosis value of them in renal cell carcinoma (RCC) was still obscure. Thus, we retrospectively used immunohistochemistry approach to evaluate the impact of these ligands on recurrence and survival of non-metastatic clear cell RCC (ccRCC) patients after nephrectomy. We systemically built a prespecified ICL score based on these ligands, and found specimens with high ICL score were prone to possess high Fuhrman grade, necrosis, and high-risk level of SSIGN. Moreover, ICL score stratified patients into different risk subgroups, and remained an independent adverse prognosticator for overall survival (OS) and recurrence-free survival (RFS). Meanwhile, in TCGA database, the increasing ICL mRNA predicted poor survival and early recurrence. Furthermore, after adding ICL score into SSIGN, the C-index for OS and RFS increased from 0.705 to 0.746 and 0.712 to 0.765, respectively. In conclusion, the ICL score based on expression of CXCL9, CXCL10 and CXCL11 stratified non-metastatic ccRCC patients into different risk subgroups of recurrence and death, which might benefit preoperative risk stratification and guide immune therapy in the future.

Abstract PD6-5: Pooled analysis of circulating tumor cells in metastatic breast cancer: Findings from 1944 individual patients data

Abstract Background: Clinical validity of CTCs (CellSearch®) in metastatic breast cancer (MBC) patients has previously been assessed in studies with limited statistical power. We aimed to pool all European studies to obtain high-level evidence on the prognostic value of CTCs, to investigate their effects across different clinico-pathological characteristics and therapies and to further validate the MD Anderson/Institut Curie/Fox Chase CTC-based prognostic nomogram established in first-line treated MBC patients (Giordano et al, Clin Cancer Res 2013). Material and methods: Methods were predefined in a written protocol. In December 2012, we searched for eligible studies that accrued patients in 2003-2012. We contacted all European laboratories using CellSearch®. We used likelihood ratio tests (LR) in Cox regression models stratified by study to assess the independent prognostic value of CTC when added to a clinicopathological (CP) model for progression-free (PFS) and overall survival (OS). Landmark analyses were used to assess the prognostic effect of early changes in CTC. The CTC-based nomogram (http://cancernomograms.com/CTCOnline.html) score was retrieved for every patient; we calculated C-indices, drew calibration plots and Kaplan-Meier curves according to quintiles of the nomogram score. Results: We collected individual data of 1944 MBC patients, from 20 different studies (some unpublished), from 17 centers in 7 European countries. We observed 1507 PFS events and 929 deaths. Baseline CTC count was significantly associated with several patient characteristics, such as performance status (PS, p&amp;lt;10-4), synchronous metastasis (p&amp;lt;10- 2) tumor subtype (p&amp;lt;10-4), liver &amp; bone metastases (p&amp;lt;10-4), CEA &amp; CA15-3 levels (p&amp;lt;10-4). The CP model for OS included PS, MBC subtypes, number of previous lines of treatment, patient's age, metastasis-free interval, metastatic sites (p&amp;lt;0.01 for all). In a multivariate analysis containing the CP model parameters and CTC count at baseline, elevated CTC count (≥5) was a significant independent predictor of OS (n = 1444, HR = 2.7, 95%CI [2.2-3.2], LR p&amp;lt;10-4). Baseline serum markers added either no or marginal effect to the CP plus baseline CTC model for OS. In contrast, early changes in CTC status at week 3-5 significantly added prognostic information for OS to the model with CP factors and baseline CTC+ (n = 569, HR = 1.8 [2.2-3.2], LR p&amp;lt;0.001). In the population of interest (MBC treated by first line chemotherapy, n = 402 patients, 176 deaths), the CTC-based nomogram exhibited a good C-index for OS (0.69), was well calibrated and showed clear separation of the survival curves. Additional results, including subgroup analyses by tumor subtype and treatments will be presented at the meeting. Conclusions: This pooled analysis is the largest study ever reported on CTC in MBC, with a previously unreached statistical power. It provides a clear level-of-evidence 1 on the independent prognostic value of CTCs before and during treatment in MBC. Also, the CTC-based prognostic nomogram is independently validated. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr PD6-5.

Comparison of predictive accuracy of four prognostic models for nonmetastatic renal cell carcinoma after nephrectomy

The objective of the current study was to compare, in a large multicenter study, the discriminating accuracy of four prognostic models developed to predict the survival of patients undergoing nephrectomy for nonmetastatic renal cell carcinoma (RCC). A total of 2404 records of patients from 6 European centers were retrospectively reviewed. For each patient, prognostic scores were calculated according to four models: the Kattan model, the University of California at Los Angeles integrated staging system (UISS) model, the Yaycioglu model, and the Cindolo model. Survival curves were estimated by the Kaplan-Meier method and compared by the log-rank test. Discriminating ability was assessed by the Harrell c-index for censored data. The primary end point was overall survival (OS), and the secondary end points were cancer-specific survival (CSS) and disease recurrence-free survival (RFS). At last follow-up, 541 subjects had died of any causes, with a 5-year OS rate of 80%. The 5-year CSS and RFS rates were 85% and 78%, respectively. All models discriminated well (P < 0.0001). The c-indexes for OS were 0.706 for the Kattan nomogram, 0.683 for the UISS model, and 0.589 and 0.615 for the Yaycioglu and Cindolo models, respectively. The Kattan nomogram was found to improve discrimination substantially in the UISS intermediate-risk patients. The current study appears to better define the general applicability of prognostic models for predicting survival in patients with nonmetastatic RCC treated with nephrectomy. The results suggest that postoperative models discriminate substantially better than preoperative ones. The Kattan model was consistently found to be the most accurate, although the UISS model was only slightly less well performing. The Kattan model can be useful in the UISS intermediate-risk patients.