scholarly journals Two-stage neural-network based prognosis models using pathological image and transcriptomic data: An application in hepatocellular carcinoma patient survival prediction

2020 ◽  
Author(s):  
Zhucheng Zhan ◽  
Noshad Hossenei ◽  
Olivier Poirion ◽  
Maria Westerhoff ◽  
Eun-Young Choi ◽  
...  
Keyword(s):  
Neural Network ◽  
Hepatocellular Carcinoma ◽  
Proportional Hazards Model ◽  
Patient Survival ◽  
Complex Model ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
P Value ◽  
Survival Prediction ◽  
Two Stage

AbstractPathological images are easily accessible data type with potential as prognostic biomarkers. Here we extend Cox-nnet, a neural network based prognosis method previously used for transcriptomics data, to predict patient survival using hepatocellular carcinoma (HCC) pathological images. Cox-nnet based imaging predictions are more robust and accurate than Cox proportional hazards model. Moreover, using a novel two-stage Cox-nnet complex model, we are able to combine histopathology image and transcriptomics RNA-Seq data to make impressively accurate prognosis predictions, with C-index close to 0.90 and log-ranked p-value of 4e-21 in the testing dataset. This work provides a new, biologically relevant and relatively interpretable solution to the challenge of integrating multi-modal and multiple types of data, particularly for survival prediction.

scholarly journals Randomized Open-Labeled Academic Trial Comparing Standard AZA Therapy with Combination of G-CSF with AZA in High Risk MDS Patients - Interim Analysis

Blood ◽  
2019 ◽  
Vol 134 (Supplement_1) ◽  
pp. 1729-1729
Author(s):  
Anna Jonasova ◽  
Lubomir Minarik ◽  
Vojtech Kulvait ◽  
Michal Pesta ◽  
Adel Schaffartzikova ◽  
...  
Keyword(s):  
Flow Cytometry ◽  
Mixed Model ◽  
Proportional Hazards Model ◽  
Patient Survival ◽  
Conflicts Of Interest ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
P Value ◽  
Pre Treatment ◽  
Time Varying Covariates

Introduction: Myelodysplastic syndrome (MDS) is characterized by differentiation blockade, cytopenias with commontransfusion dependency and immune defects. Upon progression the myeloblasts accumulate and the patients become vulnerable to severe infection complications. Based on the Prague Charles University General Hospital registry (N=164, median age 73), the AZA therapy in higher-risk MDS patients results in median OS 13.8 Mo with ORR 48.5%. We also noted from our retrospective data that AZA-treated patients with higher G-CSF consumption had significantly reduced occurrence of Grade 4 neutropenias and longer OS (19 vs 16 Mo, p value 0.039). Rationale: To improve poor clinical outcomes we initiated a randomized open labeled academic trial that compares standard AZA therapy (A) with novel AZA-based therapy combination involving use of G-CSF added prior AZA (GA). Both AZA and also decitabine were preclinically shown to induce myeloid differentiation upon G-CSF preincubation. G-CSF binds its receptor in granulocytic precursors and neutrophils to stimulate their survival, proliferation, and differentiation via myeloid master regulator transcription factor and leukemia-suppressor PU.1. We also have previously shown that AZA increases PU.1. expression. Study design & Methods: GA/MDS-2013 (EudraCT No 2013-001639-38). Expected for enrollment are 134 patients, currently enrolled 53 patients (GA arm N=29, A arm N=24) with median age 74 years, M:F ratio 32:21 (GA 16:16, A 13:8),median IPSS-R 6, median follow up 11.2 Mo, median cycles of therapy 6. Diagnosis included:MDS (EB1, EB 2) with IPSS int-2/high (75%), MDS/AML<30% MB (22.5%), and CMML II (2.5%). Transplant candidates were excluded. Randomization is 2:1 for GA vs A arm. Primary endpoints: OS, PFS, time to AML transformation, ORR, infections & QoL. Secondary endpoints: biomarkers. Therapy schedule: 75mg/m2 of AZA 5-2-2, in GA: G-CSF s.c. injected 48 hrs before dose 1 and dose 6. G-CSF is measured in patient sera (prior therapy), myeloid surface markers are determined by flow cytometry (day -2, day 1, and day 9 of cycle 1). Genomic libraries from whole bone marrow are prepared by NEBNext Direct Kit involving 33 gene panel, sequencing runs are performed on Illumina platform. Statistics involved longitudinal multivariate data analysis including the joint models for the OS and response. Results: The presented data include 2.5 years since the beginning of the trial. Median survival for GA arm was 11 vs 6 Mo in the A arm. ORR (CR, CRm, PR, HI) was 56% in GA arm vs 33% in the A arm. Transformation to AML for both arms was comparable. The stratified longitudinal Cox proportional hazards model containing time-varying covariates together with the ordinal multilevel logistic mixed model were utilized. From this joint fitted model, a negative coefficient for the G-AZA treatment (significant p-value 0.0442) can be noticed in the case of the Cox Proportional Hazard part of the model. This means that G-AZA treatment improves patient survival. The estimated odds for the GA arm that responded to the therapy with remission rather than progression is 12.4x higher than for the A arm, controlling for the remaining patients' characteristics (p-value 0.0016).Both the GA and A arms are comparably tolerated. Data on serious infections and neutropenia gr4 were not yet available. The levels of G-CSF in sera prior the study in both arms (GA vs A) were comparable. Flow cytometry revealed G-CSF mediated upregulation of FCgRI (CD64) in the GA but not in the A arm. Multivariate analysis indicates the following: mutated genes: DNMT3A (p-value 0.0157), EZH2 (p-value 0.0091), TP53 (borderline p-value 0.0510), & CSF3R (p-value 0.0057) shorten the overall survival. The significant negative effects on response was noted for mutated EZH2 (p-value 0.0208) and CSF3R (p-value 0.0424) genes. Conclusions: The current results supported by different methods and statistics indicates a beneficial effect of G-CSF pre-treatment to standard AZA therapy in higher risk MDS patients. G-CSF pre-treatment to AZA increases OS and ORR. In addition, we identified biomarkers that are negatively associated with patient survival and response including EZH2, DNMT3A, TP53, & CSF3R. Grant Support: Ministry of Health, #16-27790A. Institutional resources: Progres Q49 & Q26, UNCE/MED/016, LQ1604, SVV 260374/2017, RVO-64165. Disclosures No relevant conflicts of interest to declare.

scholarly journals A novel nomogram to predict the overall survival in esthesinoeroblastoma

2020 ◽  
Author(s):  
Lijie Jiang ◽  
Tengjiao Lin ◽  
Yu Zhang ◽  
Wenxiang Gao ◽  
Jie Deng ◽  
...  
Keyword(s):  
Overall Survival ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Validation Cohort ◽  
Risk Groups ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
Cancer Center ◽  
P Value ◽  
Training Cohort

Abstract Background Increasing evidence indicates that the pathology and the modified Kadish system have some influence on the prognosis of esthesioneuroblastoma (ENB). However, an accurate system to combine pathology with a modified Kadish system has not been established. Methods This study aimed to set up and evaluate a model to predict overall survival (OS) accurately in ENB, including clinical characteristics, treatment and pathological variables. We screened the information of patients with ENB between January 1, 1976, and December 30, 2016 from the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) program as a training cohort. The validation cohort consisted of patients with ENB at Sun Yat-sen University Cancer Center and The First Affiliated Hospital of Sun Yat-sen University in the same period, and 87 patients were identified. The Pearson’s chi-squared test was used to assess significance of clinicopathological and demographic characteristics. We used the Cox proportional hazards model to examine univariate and multivariate analyses. The model coefficients were used to calculate the Hazard ratios (HR) with 95% confidence intervals (CI). Prognostic factors with a p- value < 0.05 in multivariate analysis were included in the nomogram. The concordance index (c-index) and calibration curve were used to evaluate the predictive power of the nomogram. Results The c-index of training cohort and validation cohort are 0.737 (95% CI, 0.709 to 0.765) and 0.791 (95% CI, 0.767 to 0.815) respectively. The calibration curves revealed a good agreement between the nomogram prediction and actual observation regarding the probability of 3-year and 5-year survival. We used a nomogram to calculate the 3-year and 5-year growth probability and stratified patients into three risk groups. Conclusions The nomogram provided the risk group information and identified mortality risk and can serve as a reference for designing a reasonable follow-up plan.

scholarly journals Risk Factors for Mortality in Chinese Patients on Continuous Ambulatory Peritoneal Dialysis

2015 ◽  
Vol 35 (2) ◽  
pp. 199-205 ◽  
Author(s):  
Fan Zhang ◽  
Hong Liu ◽  
Xiaoli Gong ◽  
Fuyou Liu ◽  
Youming Peng ◽  
...  
Keyword(s):  
Risk Factors ◽  
Peritoneal Dialysis ◽  
Continuous Ambulatory Peritoneal Dialysis ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Patient Survival ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
Hazards Model ◽  
Kaplan Meier

ObjectiveThe intent of this study was to evaluate the clinical outcome and risk factors affecting mortality of the continuous ambulatory peritoneal dialysis (CAPD) patients in a single peritoneal dialysis (PD) center over a period of 10 years.Patients and methodsWe retrospectively analyzed patients on PD from June 2001 to June 2011. The clinical and biochemical data were collected from the medical records. Clinical variables included gender, age at the start of PD, smoking status, body mass index (BMI), cause of end-stage renal disease (ESRD), presence of diabetes mellitus and blood pressure. Biochemical variables included hemoglobin, urine volume, residual renal function (RRF), serum albumin, blood urea nitrogen (BUN), creatinine, total cholesterol, triglyceride, comorbidities, and outcomes. Survival curves were made by the Kaplan-Meier method. Univariate and multivariate analyses to identify mortality risk factors were performed using the Cox proportional hazard regression model.ResultsA total of 421 patients were enrolled, 269 of whom were male (63.9%). The mean age at the start of PD was 57.9 ± 14.8 years. Chronic glomerulonephritis was the most common cause of ESRD (39.4%). Estimation of patient survival by Kaplan-Meier was 92.5%, 80.2%, 74.4%, and 55.7% at 1, 3, 5, and 10 years, respectively. Patient survival was associated with age (hazard ratio [HR]: 1.641 [1.027 – 2.622], p = 0.038), cardiovascular disease (HR: 1.731 [1.08 – 2.774], p = 0.023), hypertriglyceridemia (HR: 1.782 [1.11 – 2.858], p = 0.017) in the Cox proportional hazards model analysis. Estimation of technique survival by Kaplan-Meier was 86.7%, 68.8%, 55.7%, and 37.4% at 1, 3, 5, and 10 years, respectively. In the Cox proportional hazards model analysis, age (HR: 1.672 [1.176 – 2.377], p = 0.004) and hypertriglyceridemia (HR: 1.511 [1.050 – 2.174], p = 0.026) predicted technique failure.ConclusionThe PD patients in our center exhibited comparable or even superior patient survival and technical survival rates, compared with reports from other centers in China and other countries.

Disparities in receipt of radiotherapy and survival by age, sex, and race among patients with nonmetastatic squamous cell carcinoma of the anus.

2015 ◽  
Vol 33 (3_suppl) ◽  
pp. 721-721
Author(s):  
Doug Baughman ◽  
Krishna Bilas Ghimire ◽  
Binay Kumar Shah
Keyword(s):  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Survival Rates ◽  
Relative Survival ◽  
Population Based ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
P Value ◽  
Hazards Model ◽  
To Receive

721 Background: Combination chemoradiotherapy is the standard of care for treatment of non-metastatic squamous cell carcinoma of the anus (SCCA). This population-based study evaluated disparities in receipt of radiotherapy (RT) and its effect on survival in patients with localized and regional SCCA in the United States. Methods: The Surveillance, Epidemiology, and End Results (SEER) 18 database was used to identify patients with localized and regional SCCA diagnosed between 1998 and 2008. We used univariate and multivariate logistic regression to model the relationships between receipt of RT and age, sex, marital status, stage, and race. Relative survival rates were calculated and compared using two sample z-tests. A Cox proportional hazards model was used to find adjusted hazard ratios (HR). Results: A total of 3,971 patients with localized or regional SCCA as the only primary malignancy were included in the study, of which 3,278 (82.6%) received RT. After adjusting for covariates, those 65 years and older (adjusted OR 0.82, p=0.029) were less likely to receive RT. Females were more likely to receive RT compared to males (adjusted OR 1.54, p<0.001). We found no difference in receipt of RT by race. Comparisons of 1- and 5-year relative survival rates showed lower survival for blacks (p-value <0.01 at 1-year and <0.0001 at 5-years), those 65 years and older, and males. A 1-year survival disparity was found for those not receiving RT (p-value <0.0001 at 1-year), but no difference was observed at 5-years. A Cox proportional hazards model adjusting for all covariates showed greater hazard for blacks (adjusted HR 1.36, p=0.001), those not receiving RT (adjusted HR 1.23, p=0.03), patients 65 years or older, and males. Conclusions: This population based study identified older patients as less likely to receive RT and females as more likely to receive RT. Survival analysis identified blacks, males, older patients, and those not receiving RT as having lower rates of survival.

scholarly journals Narrowing of the Racial Disparities Gap in Survival of Patients with Mycosis Fungoides: A Longitudinal Analysis of the SEER Database (1988 to 2011)

Blood ◽  
2021 ◽  
Vol 138 (Supplement 1) ◽  
pp. 1392-1392
Author(s):  
Noha Soror ◽  
Hamid D. Ismail ◽  
Catherine Chung ◽  
Basem M. William
Keyword(s):  
Racial Disparities ◽  
Research Funding ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
P Value ◽  
Time Period ◽  
Black Patients ◽  
Over Time

Abstract I ntroduction: Mycosis Fungoides (MF) is the most common subtype of cutaneous T-cell lymphomas. Prior Studies have identified Black race as a risk factor for earlier age at diagnosis, more advanced stages at time of diagnosis and poor prognosis in patients with MF. Data examining differences in racial disparities outcomes over time are limited. Objective: This retrospective analysis aims to examine if the racial disparities in survival outcomes of MF patients have improved over time. Subjects and Methods: Using the United States Surveillance, Epidemiology and End Results (SEER) 1988-2011 public use database, we examined survival patterns for patients with MF (with the code of 9700) between 1988 and 2011. Cases were divided into three cohorts based on the year of diagnosis; "1988 - 1995", "1996 - 2003", and "2004 - 2011". Univariable and multivariable analysis were conducted to assess for factors significantly associated with the overall survival. The nonparametric estimates of the survival distribution function, Kaplan and Meier survival curves, and Cox proportional hazards model were used to investigate the factors affecting the survival time. Results: From 1988 to 2011, a total of 2896 cases of MF were identified with a median follow-up of 60 months. The difference in the survival time between the years of diagnosis 1988-1995 and 2004-2011 is significant (p-value=0.05). The parameter estimate of the Cox proportional hazards model for the "1988-1995" and the "2004-2011" period as a reference is also significant (p-value = 0.024) and the hazard ratio (HR) is 1.407, which means that patients diagnosed in 1988-1995 were 1.4 times likely to die from the disease compared to the patients diagnosed in 2004-2011 (i.e. patients in 1988-1995 were more likely to not survive than in 2004-2011) (Table 1 and 2). There is no significant difference in the survival of the patients between "1996-2003" and "2004-2011" (p-value 0.998), Cox model estimate is not significant (p-value = 0.178), and the HR is 0.94 (Table 1 and 2). For the time period 1988-1995, the survival of Black patients was inferior to White (p= 0.0339), Asians (p=0.001), and other races (p=0.0011); Figure 2 and Table 3. For the time period 1996-2003, there was no difference in survival across races (p-value=0.7599); Figure 3 and Table 3. For the time period of 2004-2011, survival of Black patients was similar to White (p-value=1) but again inferior to Asian (p-value=0.05) and other races (p-value=0.09); Figure 4 and Table 3. Across the entire time period of 1998-2011, the survival of Black patients was inferior to White (Chi-square=6.59 and p-value=0.0084); Figure 5. The survival gap between Black and White patients seems to be obliterated in subsequent; "1996 - 2003" and "2004 - 2011" vs 1988-1995 (Figures 3 and 4) due to improvements in survival of Black patients over time (Figure 6) while the survival of White patients remained rather steady over time (Figure 7). Conclusions: Our study demonstrated that Black race was significantly correlated with poorer survival in patients with MF. The etiology of this poorer prognosis can be related to access to medical care, socioeconomic disparities, or possibly difference in disease biology and immune response. Despite the persistent pattern of lower survival across all time periods, the gap in survival between White and Black races seems to be narrowing overtime. Figure 1 Figure 1. Disclosures William: Dova Pharmaceuticals: Research Funding; Incyte: Research Funding; Kyowa Kirin: Consultancy; Merck: Research Funding; Guidepoint Global: Consultancy.

Is Obesity a Favorable Prognostic Factor in Peritoneal Dialysis Patients?

2000 ◽  
Vol 20 (6) ◽  
pp. 715-721 ◽  
Author(s):  
David W. Johnson ◽  
Karen A. Herzig ◽  
David M. Purdie ◽  
Wendy Chang ◽  
Allison M. Brown ◽  
...  
Keyword(s):  
Myocardial Infarction ◽  
Peritoneal Dialysis ◽  
Proportional Hazards Model ◽  
Patient Survival ◽  
Tertiary Care ◽  
Normal Weight ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
Adverse Cardiovascular Event ◽  
Free Survival

Objective To determine the influence of an elevated body mass index (BMI) on cardiovascular outcomes and survival in peritoneal dialysis (PD) patients. Design Prospective, observational study of a prevalent PD cohort at a single center. Setting Tertiary care institutional dialysis center. Patients The study included all patients with a BMI of at least 20 who had been receiving PD for at least 1 month as of 31 January 1996 ( n = 43). Patients were classified as overweight [BMI > 27.5; mean ± standard error of mean (SEM): 32.1 ± 1.1; n = 14] or normal weight (BMI 20 – 27.5; mean ± SEM: 23.8 ± 0.4; n = 29). Outcome Measures Patient survival and adverse cardiovascular events (myocardial infarction, congestive cardiac failure, cerebrovascular accident, and symptomatic peripheral vascular disease) were recorded over a 3-year period. Results At baseline, no significant differences were seen between the groups in clinical, biochemical, nutritional, or echocardiographic parameters, except for a lower dietary protein intake (0.97 ± 0.10 g/kg/day vs 1.44 ± 0.10 g/ kg/day, p = 0.004) and a higher proportion of well-nourished patients by subjective global assessment (100% vs 72%, p < 0.05) in the overweight group. After 3 years of follow-up, 29% of overweight patients and 69% of normal-weight patients had died ( p < 0.05). Using a Cox proportional hazards model, a BMI greater than 27.5 was shown to be an independent positive predictor of patient survival, with an adjusted hazard ratio (HR) of 0.09 [95% confidence interval (CI): 0.01 – 0.85; p < 0.05]. However, being overweight did not significantly influence myocardial infarction-free survival (adjusted HR: 0.33; 95% CI: 0.07 – 1.48; p = 0.15) or combined adverse cardiovascular event-free survival (adjusted HR: 0.67; 95% CI: 0.23 – 1.93; p = 0.46). Conclusions Obesity conferred a significant survival advantage in our PD population. Obese patients should therefore not be discouraged from receiving PD purely on the basis of BMI. Moreover, maintaining a higher-than-average BMI to preserve “nutritional reserve” may help to reduce the mortality and morbidity rates associated with PD.

scholarly journals Survival of Patients with Hepatocellular Carcinoma in Renal Insufficiency: Prognostic Role of Albumin-Bilirubin Grade

Cancers ◽  
2020 ◽  
Vol 12 (5) ◽  
pp. 1130
Author(s):  
Shu-Yein Ho ◽  
Chia-Yang Hsu ◽  
Po-Hong Liu ◽  
Chih-Chieh Ko ◽  
Yi-Hsiang Huang ◽  
...  
Keyword(s):  
Hepatocellular Carcinoma ◽  
Renal Insufficiency ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Performance Status ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
Prognostic Role ◽  
The Impact

Renal insufficiency (RI) is commonly seen in patients with hepatocellular carcinoma (HCC). The prognostic role of albumin-bilirubin (ALBI) grade in this special setting is unclear. We aimed to investigate the role of ALBI grade associated with the impact of RI on HCC. A prospective cohort of 3690 HCC patients between 2002 and 2016 were retrospectively analyzed. The Kaplan–Meier method and multivariate Cox proportional hazards model were used to determine survival and independent prognostic predictors. Of all patients, RI was an independent predictor associated with decreased survival. In multivariate Cox analysis for patients with RI, α-fetoprotein level ≥20 ng/mL, tumor size >3 cm, vascular invasion, distant metastasis, presence of ascites, performance status 1–2, performance status 3–4, and ALBI grade 2 and grade 3 were independent predictors of decreased survival (all p < 0.05). In subgroup analysis of patients with RI undergoing curative and non-curative treatments, the ALBI grade remained a significant prognostic predictor associated with decreased survival (p < 0.001). In summary, HCC patients with RI have decreased survival compared to those without RI. The ALBI grade can discriminate the survival in patients with RI independent of treatment strategy and is a feasible prognostic tool in this special patient population.

MULTIMODALITY TREATMENT OF GLIOBLASTOMA PATIENTS USING DIFFERENT TEMOZOLOMIDE REGIMENS IN POSTOPERATIVE CHEMORADIOTHERAPY

2017 ◽  
Vol 63 (6) ◽  
pp. 915-919
Author(s):  
Edvard Zhavrid ◽  
Pavel Demeshko ◽  
Nataliya Artemova ◽  
V. Sinayko
Keyword(s):  
Overall Survival ◽  
Median Overall Survival ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Patient Survival ◽  
Radiation Treatment ◽  
Multimodality Treatment ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
Hazards Model

The outcomes of multimodality treatment of 227 glioblastoma (grade IV) patients were evaluated in relation to the postoperative chemoradiotherapy (ChRT) regimen. No statistically significant differences were found in groups with conventional ChRT (temozolomide 75 mg/m2 per os 1 hour before the radiation treatment during the whole course of radiotherapy, n=111) and modified ChRT (temozolomide 75 mg/m2 per os 5 days a week 1 hour before the radiation treatment in the first and last two weeks of radiotherapy, n=116), the median overall survival being 16 months and 16 months respectively (P=0,889). The Cox proportional hazards model demonstrated that the postoperative ChRT regimen was not a prognostic factor affecting patient survival.

Sign in / Sign up

Export Citation Format

Share Document