A novel approach to determine two optimal cut-points of a continuous predictor with a U-shaped relationship to hazard ratio in survival data: simulation and application

225 Background: Two phase III trials, COMET-1 and COMET-2, have reported that cabozantinib did not extend overall survival (OS) outcomes compared to prednisone and prednisone plus mitoxantrone respectively, in unselected post-docetaxel patients with mCRPC. We conducted a retrospective analysis of a combined dataset of these trials in an attempt to identify a benefit in subsets of patients based on prognostic risk factors. Methods: Baseline characteristics and survival data were combined for COMET-1 and COMET-2. Prognostic ability of baseline factors on survival was evaluated using Cox proportional hazards regression models, which incorporated the factors in the modern post-docetaxel nomogram and overall bone scan lesion area (BSLA) by computer-assisted analysis. Regression models included trial as a stratification factor. Evaluation of potentially benefited subsets was performed by the use of interaction terms between the factors and cabozantinib. The hazard ratio between cabozantinib and comparator arms was evaluated after adjusting for baseline prognostic factors. All tests were 2-sided and a p-value ≤0.05 was considered statistically significant. Results: A total of 1147 post-docetaxel patients with mCRPC were available (1028 from COMET-1 and 119 from COMET-2). Age, albumin, LDH and BSLA were statistically significant prognostic factors in addition to variables from the Halabi model (site of metastases, pain, docetaxel-free interval, ECOG-PS, hemoglobin, alkaline phosphatase, PSA). There was no consistent interaction effect on survival between cabozantinib vs. comparator arms for any prognostic group. After adjusting for multiple potential prognostic factors, treatment with cabozantinib vs. comparator was associated with an observed hazard ratio of 0.77 (95% CI = 0.65 to 0.92, p = 0.003) for survival. Conclusions: After combining data from both COMET-1 and COMET-2 trials, no differential effect of cabozantinib on survival was observed for any prognostic risk factor. A modest treatment effect of cabozantinib on OS was observed after controlling for multiple potential prognostic factors.

Download Full-text

Estimating the mean hazard ratio parameters for clustered survival data with random clusters

Statistics in Medicine ◽

10.1002/(sici)1097-0258(19970915)16:17<2009::aid-sim606>3.0.co;2-r ◽

1997 ◽

Vol 16 (17) ◽

pp. 2009-2020 ◽

Cited By ~ 8

Author(s):

Jianwen Cai ◽

Haibo Zhou ◽

C. E. Davis

Keyword(s):

Survival Data ◽

Hazard Ratio ◽

The Mean ◽

Clustered Survival Data

Download Full-text

A balanced hazard ratio for risk group evaluation from survival data

Statistics in Medicine ◽

10.1002/sim.6505 ◽

2015 ◽

Vol 34 (17) ◽

pp. 2528-2543 ◽

Cited By ~ 2

Author(s):

Samuel Branders ◽

Pierre Dupont

Keyword(s):

Survival Data ◽

Risk Group ◽

Hazard Ratio ◽

Group Evaluation

Download Full-text

Odds ratio, hazard ratio and relative risk

Advances in Methodology and Statistics ◽

10.51936/uwah2960 ◽

2016 ◽

Vol 13 (1) ◽

Author(s):

Janez Stare ◽

Delphine Maucort-Boulch

Keyword(s):

Relative Risk ◽

Survival Data ◽

Odds Ratio ◽

Proportional Hazards ◽

Medical Literature ◽

Cox Model ◽

Hazard Ratio ◽

Probability Of Death ◽

Time Point

Odds ratio (OR) is a statistic commonly encountered in professional or scientific medical literature. Most readers perceive it as relative risk (RR), although most of them do not know why that would be true. But since such perception is mostly correct, there is nothing (or almost nothing) wrong with that. It is nevertheless useful to be reminded now and then what is the relation between the relative risk and the odds ratio, and when by equating the two statistics we are sometimes forcing OR to be something it is not. Another statistic, which is often also perceived as a relative risk, is the hazard ratio (HR). We encounter it, for example, when we fit the Cox model to survival data. Under proportional hazards it is probably "natural" to think in the following way: if the probability of death in one group is at every time point k-times as high as the probability of death in another group, then the relative risk must be k, regardless of where in time we are. This could be hardly further from the truth and in this paper we try to dispense with this blunder.

Download Full-text

Immunological Analysis of a CCHFV mRNA Vaccine Candidate in Mouse Models

Vaccines ◽

10.3390/vaccines7030115 ◽

2019 ◽

Vol 7 (3) ◽

pp. 115 ◽

Cited By ~ 2

Author(s):

Touraj Aligholipour Farzani ◽

Katalin Földes ◽

Koray Ergünay ◽

Hakan Gurdal ◽

Aliye Bastug ◽

...

Keyword(s):

Survival Data ◽

Booster Dose ◽

Sandwich Elisa ◽

Hemorrhagic Fever ◽

Ifn Gamma ◽

Hemorrhagic Fever Virus ◽

Novel Approach ◽

Protection Rate ◽

Protection Potential ◽

Immunological Assays

Development of new vaccine platforms against viral diseases is considered urgent. In recent years, mRNA constructs have attracted great interest in this field due to unique advantages over conventional gene transfer platforms. In the present study, we developed a new naked conventional mRNA vaccine expressing the non-optimized small (S) segment of the Ank-2 strain of Crimean-Congo Hemorrhagic Fever virus (CCHFV). We then analyzed its single and booster dose immunogenicity and protection potential in the challenge assay in two mice models, including IFNα/β/γR−/− and C57BL/6. The results obtained from the immunological assays, namely IL-4 and IFN-gamma ELISPOT, intracellular IFN-gamma staining, in-house sandwich ELISA, and survival data, demonstrated that our construct elicited the production of anti-nucleocapsid (N) specific immune responses in both mice models. A 100% protection rate was only obtained in the booster dose group of IFNα/β/γR−/− mice, indicating that this platform needs further optimization in future studies. In conclusion, we assessed a novel approach in CCHFV vaccination by introducing a conventional mRNA platform which can be considered in future experiments as an efficient and safe way to battle this disease.

Download Full-text

Data-Derived Treatment Duration Goal for Cervical Cancer: Should 8 Weeks Remain the Target in the Era of Concurrent Chemoradiation?

JCO Clinical Cancer Informatics ◽

10.1200/cci.16.00072 ◽

2017 ◽

pp. 1-15 ◽

Cited By ~ 7

Author(s):

Julian C. Hong ◽

Jonathan Foote ◽

Gloria Broadwater ◽

Julie A. Sosa ◽

Stephanie Gaillard ◽

...

Keyword(s):

Cervical Cancer ◽

Treatment Duration ◽

Cross Validation ◽

Recursive Partitioning ◽

Hazard Ratio ◽

Recursive Partitioning Analysis ◽

Rank Test ◽

Cut Points ◽

Cut Point ◽

Lower Stage

Purpose Prior studies have demonstrated the importance of treatment duration (TD) in radiation therapy (RT) for cervical cancer, with an 8-week goal based primarily on RT alone. This study uses a contemporary cohort to estimate the time point by which completion of chemoradiation therapy is most critical. Patients and Methods The National Cancer Database was queried for women with nonmetastatic cervical cancer diagnosed from 2004 to 2012 who underwent chemotherapy, external beam RT, and brachytherapy. Data-derived TD cut points for overall survival (OS) were computed by using recursive partitioning analysis with bootstrapped aggregation (bagging) and 10-fold cross-validation. Models were independently trained with 70% of the population and validated on 30% of the population by log-rank test with and without propensity matching. Multivariable Cox proportional hazards regression was performed for the entire cohort. Results In all, 7,355 women were identified with a median TD of 57 days. Bagged recursive partitioning analysis converged to a mean cut point of 66.6 days (median, 64.5 days; interquartile range, 63.5 to 68.5 days). Cross-validation yielded a cut point of 63.3 days. Both cut points differentiated OS in validation. Younger age, recent diagnosis, geographic region, nongovernment insurance, shorter distance to treatment facility, metropolitan location, lower comorbidity, squamous cell carcinoma, lower stage, negative lymph nodes, and shorter TD were independently associated with longer OS. With adjustment, TD within the mean cut point (64.9 days; hazard ratio, 0.79; 95% CI, 0.73 to 0.87) and 56 days (hazard ratio, 0.87; 95% CI, 0.80 to 0.95) were associated with longer OS. Exploratory stratification suggested increasing OS detriment beyond 64 days. Conclusion Shorter chemoradiation TD in cervical cancer is associated with longer survival, and TD should be minimized as much as possible. The data-derived cut point was distributed around 64 days, with a continuous relationship between shorter TD and longer OS.

Download Full-text

Bayesian varying coefficient mixed-effects joint models with asymmetry and missingness

Statistical Modelling ◽

10.1177/1471082x16678543 ◽

2017 ◽

Vol 17 (3) ◽

pp. 117-141 ◽

Cited By ~ 1

Author(s):

Tao Lu ◽

Chunyan Cai ◽

Minggen Lu ◽

Jun Zhang ◽

Guang-Hui Dong ◽

...

Keyword(s):

Clinical Data ◽

Survival Data ◽

Mixed Effects ◽

Model Errors ◽

Estimation Of Parameters ◽

Joint Models ◽

Data Simulation ◽

Varying Coefficient ◽

Longitudinal Outcome ◽

Important Relationship

Abstract: Longitudinal and survival data are often collected from clinical studies. Mixed-effects joint models are commonly used for the analysis of such data. Nevertheless, the following issues may arise in longitudinal survival data analysis: (a) most joint models assume a simple parametric mixed-effects model for longitudinal outcome, which may obscure the important relationship between response and covariates; (b) clinical data often exhibits asymmetry so that symmetric assumption for model errors may lead to biased estimation of parameters; (c) response may be missing and missingness may be informative. There is little work concerning all of these issues simultaneously. We develop a Bayesian varying coefficient mixed-effects joint model with skewness and missingness to study the simultaneous influence of these features. The proposed methods are applied to an AIDS clinical data. Simulation studies are conducted to assess the performance of the method.

Download Full-text

Subtyping Cutaneous Melanoma Matters

JNCI Cancer Spectrum ◽

10.1093/jncics/pkaa097 ◽

2020 ◽

Vol 4 (6) ◽

Author(s):

Mary-Ann El Sharouni ◽

Paul Johannes van Diest ◽

Arjen Joost Witkamp ◽

Vigfús Sigurdsson ◽

Carla Henrica van Gils

Keyword(s):

Survival Data ◽

Cutaneous Melanoma ◽

Hazard Ratio ◽

Breslow Thickness ◽

Lentigo Maligna ◽

Acral Lentiginous Melanoma ◽

Netherlands Cancer Registry ◽

Hazard Ratios ◽

Increased Risk ◽

Lentigo Maligna Melanoma

Abstract Background Our aim was to investigate the role of melanoma subtype on survival and focus on the effects stratified by Breslow thickness and ulceration status. Methods Patients with cutaneous melanoma stage I, II, or III diagnosed between 2000 and 2014 were derived from the Dutch Nationwide Pathology Registry and overall survival data from the Netherlands Cancer Registry. Patients were followed until 2018. Using multivariable Cox proportional hazards models, hazard ratios were calculated for each melanoma subtype, per Breslow thickness category and ulceration status, and adjusted for age, sex, stage, and localization. Results A total of 48 361 patients were included: 79.3% had superficial spreading melanoma (SSM), 14.6% nodular melanoma (NM), 5.2% lentigo maligna melanoma, and 0.9% acral lentiginous melanoma (ALM). In the total patient group, using SSM as the reference category, adjusted hazard ratios were 1.06 (95% confidence interval [CI] = 1.01 to 1.12) for NM, 1.02 (95% CI = 0.93 to 1.13) for lentigo maligna melanoma, and 1.26 (95% = CI 1.06 to 1.50) for ALM. Among patients with 1.0 mm or less Breslow thickness and no ulceration, NM showed a twofold increased risk (hazard ratio = 1.96, 95% CI = 1.58 to 2.45) compared with SSM. Compared with 1.0 mm or less SSM without ulceration, the hazard ratio for 1.0 mm or less SSM with ulceration was 1.94 (95% CI = 1.55 to 2.44), and the hazard ratio for 1.0 mm or less NM with ulceration was 3.46 (95% CI = 2.17 to 5.50). NM patients with tumors greater than 1.0 mm did not show worse survival than SSM patients with tumors greater than 1.0 mm. Conclusions In this large nationwide study, ALM patients showed worse survival than SSM patients. Among patients with melanomas that were thin (1.0 mm or less), NM subtype patients also showed worse survival than SSM patients.

Download Full-text

Piecewise Cause-Specific Association Analyses of Multivariate Untied or Tied Competing Risks Data

The International Journal of Biostatistics ◽

10.1515/ijb-2013-0023 ◽

2014 ◽

Vol 10 (2) ◽

Author(s):

Hao Wang ◽

Yu Cheng

Keyword(s):

Competing Risks ◽

Survival Data ◽

Asymptotic Properties ◽

Hazard Ratio ◽

The Other ◽

Rounding Errors ◽

Association Analyses ◽

Pseudo Likelihood ◽

Competing Risks Data ◽

Event Times

AbstractIn this paper we extend the bivariate hazard ratio to multivariate competing risks data and show that it is equivalent to the cause-specific cross hazard ratio. Two approaches are proposed to estimate these two equivalent association measures. One extends the plug-in estimator, and the other adapts the pseudo-likelihood estimator for bivariate survival data to multivariate competing risks data. The asymptotic properties of the extended estimators are established by using empirical processes techniques. The extended plug-in and pseudo-likelihood estimators have comparable performance with the existing U-statistic when the data have no tied events. However, in many applications, there are tied events in which all the three estimators are found to produce biased results. To our best knowledge, we are not aware of any association analysis for multivariate competing risks data that has considered tied events. Hence we propose a modified U-statistic to specifically handle tied observations. The modified U-statistic clearly outperforms the other estimators when there are rounding errors. All methods are applied to the Cache County Study to examine mother–child and sibship associations in dementia among this aging population, where the event times are rounded to the nearest integers. The modified U performs consistently with our simulation results and provides more reliable results in the presence of tied events.

Download Full-text