Model For Survival In Patients Research Articles

Validation of a Predictive Model for Survival in Patients with Acute Myeloid Leukemia and Myelodysplastic Syndrome Receiving Hematopoietic Stem Cell Transplantation

▪IntroductionAllogeneic hematopoietic stem-cell transplantation (AHSCT) is an effective treatment for patients with high-risk or relapsed acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS). Several factors are known to influence survival post-transplant. Our group developed a model to predict survival in a small number of patients with AML/MDS using 3 disease (cytogenetics and disease status at transplant) and patient-related [Hematopoietic stem cell transplant comorbidity index (HCT-CI)] variables (Bachegowda et al.Blood.2017) . This model effectively stratified AML/MDS patients into 3 risk groups with very different survival. Here we aim to validate this model in a large cohort of AML/MDS patients receiving AHSCT with various donor types.MethodsA total of 934 patients with AML/MDS consecutively treated at the University of Texas MD Anderson Cancer Center were included in this analysis. Donors were HLA matched related (n=377, 40%), matched unrelated (n=416, 45%) mismatched unrelated (n=68, 7%) and T-cell replete haploidentical transplants (n=73, 8%). The median age was 53 years (range 18-65 years). Cytogenetics data at diagnosis was available in 920 patients and was favorable, intermediate and adverse-risk in 340 (37%), 517 (56%) and 63 (7%), respectively. Five hundred and forty patients (58%) were in first or second complete remission (CR1/2). The hematopoietic stem cell transplant comorbidity index (HCT-CI) was available in 914 patients and the median score was 2 (range 0-17). The sources of hematopoietic stem cells were peripheral blood for 586 patients (63%) and bone marrow for 348 patients (37%). Seven hundred and eighty patients (84%) received myeloablative-conditioning chemotherapy.ResultsPatients were stratified into 3 risk groups according to the previously published model: 1) low-risk group (patients in CR1/2 with intermediate/favorable risk cytogenetics, irrespective of HCT-CI (HCT-CI did not impact survival in this group) (n=351), 2) intermediate-risk group (patients beyond CR1/2 and/or with adverse-risk cytogenetics and HCT-CI score ≤4) (n=412), and 3) high-risk group (patients with at least one negative prognostic factor - beyond CR1/2 and/or with adverse-risk cytogenetics and a HCTCI score >4) (n=155).Median follow up duration for 436 survivors was 48 months. For the whole group, the 5-year progression-free survival (PFS), treatment-related mortality (TRM) and relapse rate were 39%, 21% and 39%, respectively. TRM at 5 years for low-risk group was 17%, for the intermediate-risk group was 18% (HR [95% CI] 1.12 [0.79-1.59]; P=0.52), and for the high-risk group was 40% (3.24 [2.26-4.64]; P<0.0001), respectively, whereas cumulative incidence of relapse at 5 years was 24% for low-risk, 52% for intermediate-risk (2.86 [2.20-3.71]; P<0.0001) and 38% for the high-risk group (1.88 [1.34-2.63]; P<0.0001), respectively. In multivariable model, transplantation in >CR1/2 (2.84 [2.38-3.40]; P<0.0001), adverse cytogenetic risk (1.54 [1.29-1.83]; P<0.0001), males (1.23 [1.03-1.46]; P=0.022) were significantly associated with worse PFS while higher age-adjusted HCT-CI scores (1.20 [0.99-1.47]; P=0.06) approached significance. Other factors including conditioning intensity (1.02 [0.82- 1.29]; P=0.84, age (1.01 [1.00-1.01]; P=0.17) and donor type (haploidentical (1.2 [0.83-1.72]; P=0.33), matched unrelated (1.06 [0.88-1.28]; P=0.55) and mismatched unrelated donor (1.07 [0.76-1.50]; P=0.70) as compared with matched related donor) did not significantly influence PFS. Using our previously described predictive model, the patients separated in 3 risk groups with significantly different median PFS, which was not reached for the low-risk group, 9.2 months for the intermediate-risk group (2.4 [1.97-2.99]; P<0.0001), and 6.0 months for the high-risk group (3.3 [2.57-4.21]; P<0.0001), respectively. (Figures 1A and 1B)In conclusion, we were able to stratify allogeneic stem cell transplant patients in 3 risk groups based on a combination of disease (cytogenetics and disease status at transplant) and patient-related factors (HCT-CI) with significantly different survival. HCT-CI appears to have a much higher discriminatory impact in patients with higher risk disease (not in CR1/CR2 and/or adverse risk cytogenetics) compared with those in the lower risk group (CR1/CR2 and intermediate/favorable risk cytogenetics). [Display omitted] DisclosuresOran:AROG: Research Funding; Celgene: Research Funding; Astex: Research Funding. Khouri:Novartis: Research Funding.

A Prognostic Model for Survival in Patients with Relapsed/Refractory Philadelphia Chromosome-Negative Acute Lymphoblastic Leukemia on the Combination of Low-Intensity Chemotherapy Plus Inotuzumab Ozogamicin with or without Blinatumomab

A Prognostic Model for Survival in Patients with Relapsed/Refractory Philadelphia Chromosome-Negative Acute Lymphoblastic Leukemia on the Combination of Low-Intensity Chemotherapy Plus Inotuzumab Ozogamicin with or without Blinatumomab

A novel model to predict cancer-specific survival in patients with early-stage uterine papillary serous carcinoma (UPSC)

Objective: Stage I–II uterine papillary serous carcinoma (UPSC) has aggressively biological behavior and leads to poor prognosis. However, clinicopathologic risk factors to predict cancer-specific survival of patients with stage I–II UPSC were still unclear. This study was undertaken to develop a prediction model of survival in patients with early-stage UPSC.

Development and external validation of a prediction model for survival in patients with resected ampullary adenocarcinoma

IntroductionAmpullary adenocarcinoma (AAC) is a rare malignancy with great morphological heterogeneity, which complicates the prediction of survival and, therefore, clinical decision-making. The aim of this study was to develop and externally validate a prediction model for survival after resection of AAC. Materials and methodsAn international multicenter cohort study was conducted, including patients who underwent pancreatoduodenectomy for AAC (2006–2017) from 27 centers in 10 countries spanning three continents. A derivation and validation cohort were separately collected. Predictors were selected from the derivation cohort using a LASSO Cox proportional hazards model. A nomogram was created based on shrunk coefficients. Model performance was assessed in the derivation cohort and subsequently in the validation cohort, by calibration plots and Uno's C-statistic. Four risk groups were created based on quartiles of the nomogram score. ResultsOverall, 1007 patients were available for development of the model. Predictors in the final Cox model included age, resection margin, tumor differentiation, pathological T stage and N stage (8th AJCC edition). Internal cross-validation demonstrated a C-statistic of 0.75 (95% CI 0.73–0.77). External validation in a cohort of 462 patients demonstrated a C-statistic of 0.77 (95% CI 0.73–0.81). A nomogram for the prediction of 3- and 5-year survival was created. The four risk groups showed significantly different 5-year survival rates (81%, 57%, 22% and 14%, p < 0.001). Only in the very-high risk group was adjuvant chemotherapy associated with an improved overall survival. ConclusionA prediction model for survival after curative resection of AAC was developed and externally validated. The model is easily available online via www.pancreascalculator.com.

Prognostic Value of Platelet Counts in Patients with Metastatic Prostate Cancer Treated with Endocrine Therapy.

The endocrine therapy is effective for patients with advanced prostate cancer, but the disease eventually becomes refractory to treatment. The aim of this study was to investigate prognostic factors and to develop a risk stratification model for survival in patients with advanced prostate cancer undergoing endocrine therapy. This study included 197 patients with stage IV prostate cancer who were treated with endocrine therapy as primary treatment at Tokyo Medical University, Tokyo, Japan, between January 1999 and November 2012. Prognostic values including baseline clinical laboratory values before endocrine therapy for stage IV prostate cancer were examined. Patients (n = 30) who were not followed or for whom data were unavailable or who were treated with radiotherapy were excluded from the study. Excluding these patients, we retrospectively analyzed 167 patients who were treated with endocrine therapy as the primary treatment. Disease-specific survival (DSS) was evaluated using the Kaplan-Meier method, and prognostic factors were identified using the Cox proportional hazard model analysis. In univariate analyses, patients with a performance status (PS) ? 2, platelet count ? 3.0× 105 µ/L, prostate specific antigen (PSA) > 50 ng/mL, alkaline phosphatase (ALP) > 350 U/L, lactate dehydrogenase (LDH) > 240 IU/L, and Gleason score (GS) ? 8, hemoglobin (Hb) < 12 g/dL, extent of disease (EOD) ? 3 and poorly differentiated adenocarcinoma showed significantly lower DSS than their respective counterparts. Neutrophil-to-Lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR) and white blood cell (WBC) count were not significantly associated with DSS. In a multivariate Cox proportional hazard model, PS and platelet count were independent prognostic factors. Based on the hazard rate (HR) calculated by the following formula: HR = exp (0.82 × PS + 1.38 × platelet count) patients were stratified into 3 risk groups. The differences in DSS rates among the 3 groups were statistically significant. These results suggest that PS and platelet count are independent prognostic factors and that a combination of these factors can be used to stratify metastatic prostate cancer patients treated with endocrine therapy according to their DSS risk.

Development of a prediction model for survival in patients with resected ampullary adenocarcinoma: international multicenter study

Development of a prediction model for survival in patients with resected ampullary adenocarcinoma: international multicenter study

A novel model to predict cancer-specific survival in patients with early-stage uterine papillary serous carcinoma (UPSC).

ObjectiveStage I‐II uterine papillary serous carcinoma (UPSC) has aggressive biological behavior and leads to poor prognosis. However, clinicopathologic risk factors to predict cancer‐specific survival of patients with stage I‐II UPSC were still unclear. This study was undertaken to develop a prediction model of survival in patients with early‐stage UPSC.MethodsUsing Surveillance, Epidemiology, and End Results (SEER) database, 964 patients were identified with International Federation of Gynecology and Obstetrics (FIGO) stage I‐II UPSC who underwent at least hysterectomy between 2004 and 2015. By considering competing risk events for survival outcomes, we used proportional subdistribution hazards regression to compare cancer‐specific death (CSD) for all patients. Based on the results of univariate and multivariate analysis, the variables were selected to construct a predictive model; and the prediction results of the model were visualized using a nomogram to predict the cancer‐specific survival and the response to adjuvant chemotherapy and radiotherapy of stage I‐II UPSC patients.ResultsThe median age of the cohort was 67 years. One hundred and sixty five patients (17.1%) died of UPSC (CSD), while 8.6% of the patients died from other causes (non‐CSD). On multivariate analysis, age ≥ 67 (HR = 1.45, P = .021), tumor size ≥ 2 cm (HR = 1.81, P = .014) and >10 regional nodes removed (HR = 0.52, P = .002) were significantly associated with cumulative incidence of CSD. In the age ≥67 cohort, FIGO stage IB‐II was a risk factor for CSD (HR = 1.83, P = .036), and >10 lymph nodes removed was a protective factor (HR = 0.50, P = .01). Both adjuvant chemotherapy combined with radiotherapy and adjuvant chemotherapy alone decreased CSD of patients with stage I‐II UPSC older than 67 years (HR = 0.47, P = .022; HR = 0.52, P = .024, respectively). The prediction model had great risk stratification ability as the high‐risk group had higher cumulative incidence of CSD than the low‐risk group (P < .001). In the high‐risk group, patients with post‐operative adjuvant chemoradiotherapy had improved CSD compared with patients who did not receive radiotherapy nor chemotherapy (P = .037). However, there was no such benefit in the low‐risk group.ConclusionOur prediction model of CSD based on proportional subdistribution hazards regression showed a good performance in predicting the cancer‐specific survival of early‐stage UPSC patients and contributed to guide clinical treatment decision, helping oncologists and patients with early‐stage UPSC to decide whether to choose adjuvant therapy or not.

Prediction model for survival in patients with biliary tract cancer: a development and validation study

Objective: The aim of the present study was to investigate the survival rate and its prognostic factors for patients with biliary tract cancer, and then a prognostic risk prediction model was constructed to predict the survival probability of patients. Methods: A total of 14 005 patients with biliary tract cancer (including gallbladder cancer, extrahepatic bile duct cancer, and ampulla of Vater cancer), who were diagnosed between 2010 and 2015 in the US National Cancer Institute Surveillance, Epidemiology, and End Results Program (SEER) were included in the development cohort. The prognostic risk factors of biliary tract cancer were investigated using multivariate Cox regression models. The predictive nomograms were then constructed to predict the overall survival probability of 1, 3, and 5 years, and the predictive discrimination and calibration ability of the nomograms were further evaluated. Meanwhile, 11 953 patients who were diagnosed during 2004 to 2009 from SEER Program were then selected to validate the external predictive accuracy of the prediction models. Results: The 1, 3 and 5-year cumulative survival rates of patients with biliary tract cancer were 41.9%, 20.4% and 15.3%, respectively, in the development cohort. Age greater than 50 years, African Americans and Native Americans and Alaska Natives, higher T, N and M stage and poor histological differentiation grade were risk factors for death, while married status, Asia-Pacific Islanders, insured status and surgery on primary site were protective factors. Gender was not significantly associated with the overall survival. The C statistic of the prediction model was 0.73 (95%CI: 0.72-0.74), and the calibration curve showed that the interaction curves of predictive and actual survival rates of 1, 3 and 5 years were close to the 45 degree diagonal. Results in the validation cohort were similar with those in the construction cohort, with a C statistic of 0.70 (95%CI: 0.69-0.72), indicating high external applicability of the prediction model. Findings from gallbladder cancer, extrahepatic bile duct cancer, and ampulla of Vater cancer are in consistent with the overall biliary tract cancer. Conclusions: The survival rate of patients with biliary tract cancer is relatively poor, and the survival prediction model based on prognostic factors has high prediction accuracy. In the future, this prognostic prediction model could be applied to clinical practice to guide individualized treatment for patients with biliary tract cancer.

5-factor prognostic model for survival of patients with metastatic urothelial carcinoma receiving three different post-platinum PD-L1 inhibitors.

4552 Background: A prognostic model for overall survival (OS) of metastatic urothelial carcinoma (mUC) was previously reported in the setting of post-platinum atezolizumab (Pond GR, GU ASCO 2018). This model was limited by employing only atezolizumab treated patients (pts), small size of the validation dataset and unclear applicability to other PD-1/L1 inhibitors. Hence, we constructed a robust prognostic model utilizing the combined atezolizumab cohort as the discovery dataset and used 2 separate validation datasets comprised of post-platinum avelumab or durvalumab treated pts. Methods: The discovery dataset consisted of pt level data from 2 phase I/II trials (IMvigor210 and PCD4989g) evaluating atezolizumab (n = 405). Pts enrolled on 2 separate phase I/II trials, EMR 100070-001 that evaluated post-platinum avelumab (n = 242) and CD1108 that evaluated durvalumab (n = 189) comprised the validation datasets. Cox regression analyses evaluated the association of candidate prognostic factors with OS. Factors were dichotomized and laboratory values were normalized by logarithmic transformation. Stepwise selection was employed to propose an optimal model using the discovery dataset. Discrimination and calibration were assessed in the avelumab and durvalumab datasets following the validation procedure by Royston and Altman (2013). Results: The 5 factors included in the optimal prognostic model in the discovery dataset were ECOG-PS (1 vs. 0; HR 1.80; 95% CI [1.36-2.36]), presence/absence of liver metastasis (HR 1.55; 95% CI [1.20-2.00]), number of platelets (HR 2.22; 95% CI [1.54-3.18]), neutrophil-lymphocyte ratio (NLR; HR 1.94; 95% CI [1.57-2.40]) and lactate dehydrogenase (LDH; HR 1.60; 95% CI [1.28-1.99]). There was robust discrimination of survival between low, intermediate and high-risk groups based on 0-1, 2-3 and 4 factors. The concordance of survival was 0.692 in the discovery and 0.671 and 0.775 in the avelumab and durvalumab validation datasets, respectively. Acceptable or good calibration of expected 1-year survival rate was observed. Conclusions: A 5-factor prognostic model is prognostic for survival across 3 different PD-L1 inhibitors (atezolizumab, avelumab, durvalumab) in this large study totaling 836 pts overall in the setting of post-platinum therapy for mUC. This model may assist in prognostic stratification and interpreting nonrandomized trials of post-platinum PD1/L1 inhibitors.

PO-0880 Validation of a predictive model for survival in patients receiving radiotherapy for bone metastases

PO-0880 Validation of a predictive model for survival in patients receiving radiotherapy for bone metastases

Validated five-factor prognostic model for survival of patients (pts) with metastatic urothelial carcinoma (mUC) receiving different post-platinum PD-L1 inhibitors.

476 Background: A prognostic model for overall survival (OS) of mUC was previously reported in the setting of post-platinum atezolizumab (Pond GR, GU ASCO 2018). This model was limited by employing only atezolizumab treated pts, small size of the validation dataset and unclear applicability to other PD-1/L1 inhibitors. Hence, we constructed a robust prognostic model utilizing the combined atezolizumab cohort as the discovery dataset and used a validation dataset comprised of post-platinum avelumab-treated pts. Methods: The discovery dataset consisted of pt level data from 2 phase I/II trials (IMvigor210 and PCD4989g) evaluating atezolizumab (n = 405). Pts enrolled on a phase I/II trial that received post-platinum avelumab (n = 242) comprised the validation dataset (EMR 100070-001). Cox regression analyses evaluated the association of candidate prognostic factors with OS. Factors were dichotomized and laboratory values were normalized by logarithmic transformation. Stepwise selection was employed to propose an optimal model using the discovery dataset. Discrimination (via c-statistic) and calibration were assessed in the avelumab dataset following the validation procedure by Royston and Altman (2013). Results: The 5 factors included in the optimal prognostic model in the discovery dataset were ECOG-PS (1 vs. 0; HR 1.80; 95% CI [1.36-2.36]), presence/absence of liver metastasis (HR 1.55; 95% CI [1.20-2.00]), number of platelets (HR 2.22; 95% CI [1.54-3.18]), neutrophil-lymphocyte ratio (NLR; HR 1.94; 95% CI [1.57-2.40]) and lactate dehydrogenase (LDH; HR 1.60; 95% CI [1.28-1.99]). The c-statistic for prediction of survival was 0.692 and 0.671 in the discovery and validation datasets, respectively. Acceptable or good calibration of expected 1-year survival was observed. Conclusions: A 5-factor externally validated prognostic model for OS is proposed employing a large dataset of 647 pts overall in the setting of post-platinum PD-L1 inhibitors for mUC. This model may assist in prognostic stratification and interpreting nonrandomized trials of post-platinum PD1/L1 inhibitors.

Development of a prediction model for survival in patients with resected ampullary adenocarcinoma: an international multicenter study

Development of a prediction model for survival in patients with resected ampullary adenocarcinoma: an international multicenter study

Impact of the MELD score on the survival of hepatocellular carcinoma transplantation patients in Brazil: a systematic review.

This study aimed to analyse the predictive value of Model For End-Stage Liver Disease (MELD) score on medium- and long-term survival in transplanted hepatocellular carcinoma (HCC) patients in Brazil. The study was registered with International Prospective Register of Systematic Reviews (PROSPERO) under N# 152,363. Inclusion criteria were based on Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) recommendations. The search was performed on the indexed databases of Lilacs, SciELO, PubMed, and Cochrane Library, and used as search strategy the following Medical Subject Headings (MeSH) terms: ((("MELD Score") OR "Model For End-Stage Liver Disease") AND "Hepatocellular Carcinoma") AND ("Brazil"). We included full-text articles published from January 2006 to October 2019. The initial search found 162 articles. After reading the available abstracts and full texts, 156 articles were excluded, totaling six articles for qualitative analysis. Although the small number of eligible articles was a limiting factor of the study, our results partially corroborated those found in the United States, United Kingdom, and Ireland. In these countries, unlike Brazil, MELD prognostic model has shown a strong association with post-liver transplant (LT) survival. However, the low predictive capacity of the model in medium- and long-term has been similar to the one of our study. The urgency of the development and validation of a post-transplant survival model for patients with HCC is set, improving the organ allocation system in Brazil.

Biochemical Variables are Predictive for Patient Survival after Surgery for Skeletal Metastasis. A Prediction Model Development and External Validation Study

Background: Predicting survival for patients with metastatic bone disease in the extremities (MBDex) is important for ensuring the implant will outlive the patient. Hitherto, prediction models for these patients have been constructed using subjective assessments, mostly lacking biochemical variables. Objectives: To develop a prediction model for survival after surgery due to MBDex using biochemical variables and externally validate the model. Methods: We created Bayesian Belief Network models to estimate likelihood of survival 1, 3, 6, and 12 months after surgery using 140 patients. We validated the models using the data of 130 other patients and calculated the area under the Receiver Operator Characteristic curve (ROC). Variables included: hemoglobin, neutrophil-count, C-reactive protein, alkaline phosphatase, primary cancer, Karnofsky-score, ASA-score, visceral metastases, bone metastases, days from diagnose of primary cancer to index surgery for MBDex, ischemic heart disease, diabetes, fracture/impending-fracture and age. Results: Survival probabilities were influenced by all biochemical variables. Validation showed ROC for the 1, 3, 6, and 12-months model: 68% (C.I.: 55%-80%), 69% (C.I.: 60%-78%), 81% (C.I.: 74%-87%) and 84% (C.I.: 77%-90%). Conclusion: Biochemical markers can be incorporated into a prediction model for survival in patients having surgery for MBDex allowing surgeons to offer more objective and individualized treatment options.

Combined performance of screening and variable selection methods in ultra-high dimensional data in predicting time-to-event outcomes

BackgroundBuilding prognostic models of clinical outcomes is an increasingly important research task and will remain a vital area in genomic medicine. Prognostic models of clinical outcomes are usually built and validated utilizing variable selection methods and machine learning tools. The challenges, however, in ultra-high dimensional space are not only to reduce the dimensionality of the data, but also to retain the important variables which predict the outcome. Screening approaches, such as the sure independence screening (SIS), iterative SIS (ISIS), and principled SIS (PSIS), have been developed to overcome the challenge of high dimensionality. We are interested in identifying important single-nucleotide polymorphisms (SNPs) and integrating them into a validated prognostic model of overall survival in patients with metastatic prostate cancer. While the abovementioned variable selection approaches have theoretical justification in selecting SNPs, the comparison and the performance of these combined methods in predicting time-to-event outcomes have not been previously studied in ultra-high dimensional space with hundreds of thousands of variables.MethodsWe conducted a series of simulations to compare the performance of different combinations of variable selection approaches and classification trees, such as the least absolute shrinkage and selection operator (LASSO), adaptive least absolute shrinkage and selection operator (ALASSO), and random survival forest (RSF), in ultra-high dimensional setting data for the purpose of developing prognostic models for a time-to-event outcome that is subject to censoring. The variable selection methods were evaluated for discrimination (Harrell’s concordance statistic), calibration, and overall performance. In addition, we applied these approaches to 498,081 SNPs from 623 Caucasian patients with prostate cancer.ResultsWhen n = 300, ISIS-LASSO and ISIS-ALASSO chose all the informative variables which resulted in the highest Harrell’s c-index (> 0.80). On the other hand, with a small sample size (n = 150), ALASSO performed better than any other combinations as demonstrated by the highest c-index and/or overall performance, although there was evidence of overfitting. In analyzing the prostate cancer data, ISIS-ALASSO, SIS-LASSO, and SIS-ALASSO combinations achieved the highest discrimination with c-index of 0.67.ConclusionsChoosing the appropriate variable selection method for training a model is a critical step in developing a robust prognostic model. Based on the simulation studies, the effective use of ALASSO or a combination of methods, such as ISIS-LASSO and ISIS-ALASSO, allows both for the development of prognostic models with high predictive accuracy and a low risk of overfitting assuming moderate sample sizes.

Cumulative Intracranial Tumor Volume Augments the Prognostic Value of Diagnosis-Specific Graded Prognostic Assessment Model for Survival in Patients with Melanoma Cerebral Metastases.

The diagnosis-specific graded prognostic assessment scale (ds-GPA) for patients with melanoma brain metastasis (BM) utilizes only 2 key prognostic variables: Karnofsky performance status and the number of intracranial metastases. We wished to determine whether inclusion of cumulative intracranial tumor volume (CITV) into the ds-GPA model for melanoma augmented its prognostic value. To determine whether or not CITV augments the ds-GPA prognostic scale for melanoma. We analyzed the survival pattern of 344 melanoma patients with BM treated with stereotactic radiosurgery (SRS) at separate institutions and validated our findings in an independent cohort of 201 patients. The prognostic value of ds-GPA for melanoma was quantitatively compared with and without the addition of CITV using the net reclassification index (NRI>0) and integrated discrimination improvement (IDI) metrics. The incorporation of CITV into the melanoma-specific ds-GPA model enhanced its prognostic accuracy. Addition of CITV to the ds-GPA model significantly improved its prognostic value, with NRI>0 of 0.366 (95% CI: 0.125-0.607, P= .002) and IDI of 0.024 (95% CI: 0.008-0.040, P= .004). We validated these findings that CITV improves the prognostic utility of melanoma ds-GPA in an independent cohort of 201 melanoma cohort. The prognostic value of the ds-GPA scale for melanoma BM is enhanced by the incorporation of CITV.

Development and validation of a prognostic model for survival in patients treated with venoarterial extracorporeal membrane oxygenation: the PREDICT VA-ECMO score

Several scoring systems have been introduced for prognostication after initiating venoarterial extracorporeal membrane oxygenation (VA-ECMO) therapy. However, static scores offer limited guidance once VA-ECMO is implanted, although continued allocation of healthcare resources is critical. Patients requiring continued VA-ECMO support are extremely unstable, with minimal heart function and multi-organ failure in most cases. The aim of the present study was to develop and validate a dynamic prognostic model for patients treated with VA-ECMO. A derivation cohort included 205 all-comers undergoing VA-ECMO implantation at a tertiary referral hospital (51% received VA-ECMO during resuscitation and 43% had severe shock). Two prediction models based on point-of-care biomarkers were developed using penalised logistic regression in an elastic net approach. A validation cohort was recruited from an independent tertiary referral hospital. Comparators for the prediction of hospital survival were the SAVE score (area under the receiver operation characteristic curve (AUC) of 0.686), the SAPS score (AUC 0.679), the APACHE score (AUC 0.662) and the SOFA score (AUC 0.732) in 6-hour survivors. The 6-hour PREDICT VA-ECMO score (based on lactate, pH and standard bicarbonate concentration) outperformed the comparator scores with an AUC of 0.823. The 12-hour PREDICT VA-ECMO integrated lactate, pH and standard bicarbonate concentration at 1 hour, 6 hours and 12 hours after ECMO insertion allowed even better prognostication (AUC 0.839). Performance of the scores in the external validation cohort was good (AUCs 0.718 for the 6-hour score and 0.735 for the 12-hour score, respectively). In patients requiring VA-ECMO therapy, a dynamic score using three point-of-care biomarkers predicts hospital mortality with high reliability. Furthermore, the PREDICT scores are the first scores for extracorporeal cardiopulmonary resuscitation patients.

99 - Validation of a Predictive Model for Survival in Patients with Acute Myeloid Leukemia and Myelodysplastic Syndrome Receiving Hematopoietic Stem Cell Transplantation

99 - Validation of a Predictive Model for Survival in Patients with Acute Myeloid Leukemia and Myelodysplastic Syndrome Receiving Hematopoietic Stem Cell Transplantation

The NEAT Predictive Model for Survival in Patients with Advanced Cancer.

PurposeWe previously developed a model to more accurately predict life expectancy for stage IV cancer patients referred to radiation oncology. The goals of this study are to validate this model and to compare competing published models.Materials and MethodsFrom May 2012 to March 2015, 280 consecutive patientswith stage IV cancerwere prospectively evaluated by a single radiation oncologist. Patients were separated into training, validation and combined sets. TheNEAT model evaluated number of active tumors (“N”), Eastern Cooperative Oncology Group performance status (“E”), albumin (“A”) and primary tumor site (“T”). The Odette Cancer Center model validated performance status, bone only metastases and primary tumor site. The Harvard TEACHH model investigated primary tumor type, performance status, age, prior chemotherapy courses, liver metastases, and hospitalization within 3 months. Cox multivariable analyses and logisticalregressionwere utilized to compare model performance.ResultsNumber of active tumors, performance status, albumin, primary tumor site, prior hospitalizationwithin the last 3 months, and liver metastases predicted overall survival on uinvariate and multivariable analysis (p < 0.05 for all). The NEAT model separated patients into four prognostic groups with median survivals of 24.9, 14.8, 4.0, and 1.2 months, respectively (p < 0.001). The NEAT model had a C-index of 0.76 with a Nagelkerke’s R2 of 0.54 suggesting good discrimination, calibration and total performance compared to competing prognostic models.ConclusionThe NEAT model warrants further investigation as a clinically useful approach to predict survival in patients with stage IV cancer.

Predictive model for survival in patients with gastric cancer.

Background and aimGastric cancer is one of the most prevalent cancers in the world. Characterized by poor prognosis, it is a frequent cause of cancer in Iran. The aim of the study was to design a predictive model of survival time for patients suffering from gastric cancer.MethodsThis was a historical cohort conducted between 2011 and 2016. Study population were 277 patients suffering from gastric cancer. Data were gathered from the Iranian Cancer Registry and the laboratory of Emam Reza Hospital in Mashhad, Iran. Patients or their relatives underwent interviews where it was needed. Missing values were imputed by data mining techniques. Fifteen factors were analyzed. Survival was addressed as a dependent variable. Then, the predictive model was designed by combining both genetic algorithm and logistic regression. Matlab 2014 software was used to combine them.ResultsOf the 277 patients, only survival of 80 patients was available whose data were used for designing the predictive model. Mean ?SD of missing values for each patient was 4.43?.41 combined predictive model achieved 72.57% accuracy. Sex, birth year, age at diagnosis time, age at diagnosis time of patients’ family, family history of gastric cancer, and family history of other gastrointestinal cancers were six parameters associated with patient survival.ConclusionThe study revealed that imputing missing values by data mining techniques have a good accuracy. And it also revealed six parameters extracted by genetic algorithm effect on the survival of patients with gastric cancer. Our combined predictive model, with a good accuracy, is appropriate to forecast the survival of patients suffering from Gastric cancer. So, we suggest policy makers and specialists to apply it for prediction of patients’ survival.