Kari Chansky1, Dragan Subotic2, Nathan R Foster3, Torsten Blum4. 1. Cancer Research and Biostatistics, Seattle, USA. 2. Clinic for Thoracic Surgery, Clinical Center of Serbia, University of Belgrade School of Medicine, Belgrade, Serbia. 3. Division of Biomedical Statistics and Informatics, Mayo Clinic and Mayo Foundation, Rochester, Minnesota, USA. 4. Lungenklinik Heckeshorn, Belin, Germany.
Abstract
BACKGROUND: Although survival analyses represent one of the cornerstones in oncology in general, some aspects of the reported survival data in lung cancer patients are still not fully elucidated. METHODS: After having defined several open questions, an evidence based approach was applied in order to answer these questions. Areas of interest were: (I) possible uncertainties in reported survival data; (II) survival surrogates; (III) recommended methods for evaluating progression free survival (PFS) as a surrogate endpoint in future datasets; (IV) postoperative lung cancer recurrence and survival. RESULTS: In recent years, PFS has seen increasing use as a primary endpoint, particularly in phase II trials. This article focuses on the statistical aspects, and particularly on evaluating the ability of PFS to accurately predict the overall survival (OS) outcome. If the data are available from randomized trials, then the evaluation of trial level surrogacy should be carried out, in addition to the methods described in the paper. If it is not a case, the patient-level methods should be applied. Suggestions for "landmark analysis" are also given: (I) classify your cases according to progression status (progressed, progression-free, or unknown) at one or more time points of interest; (II) perform a separate Cox proportional hazards regression analysis for each time point; (III) determine and report the landmark time point where progression status best predicts survival according to the hazard ratios and P values; (IV) calculate the concordance index for each landmark analysis model. The concordance index (or "c-Index") is essentially the probability that for any two randomly selected cases, the case that is predicted to have the worst outcome, does in fact have the worst outcome. CONCLUSIONS: the widening spectrum of diagnostic and treatment in pulmonary oncology imposes the need for an updated knowledge about statistical method that would fit best for the analysed problem.
BACKGROUND: Although survival analyses represent one of the cornerstones in oncology in general, some aspects of the reported survival data in lung cancerpatients are still not fully elucidated. METHODS: After having defined several open questions, an evidence based approach was applied in order to answer these questions. Areas of interest were: (I) possible uncertainties in reported survival data; (II) survival surrogates; (III) recommended methods for evaluating progression free survival (PFS) as a surrogate endpoint in future datasets; (IV) postoperative lung cancer recurrence and survival. RESULTS: In recent years, PFS has seen increasing use as a primary endpoint, particularly in phase II trials. This article focuses on the statistical aspects, and particularly on evaluating the ability of PFS to accurately predict the overall survival (OS) outcome. If the data are available from randomized trials, then the evaluation of trial level surrogacy should be carried out, in addition to the methods described in the paper. If it is not a case, the patient-level methods should be applied. Suggestions for "landmark analysis" are also given: (I) classify your cases according to progression status (progressed, progression-free, or unknown) at one or more time points of interest; (II) perform a separate Cox proportional hazards regression analysis for each time point; (III) determine and report the landmark time point where progression status best predicts survival according to the hazard ratios and P values; (IV) calculate the concordance index for each landmark analysis model. The concordance index (or "c-Index") is essentially the probability that for any two randomly selected cases, the case that is predicted to have the worst outcome, does in fact have the worst outcome. CONCLUSIONS: the widening spectrum of diagnostic and treatment in pulmonary oncology imposes the need for an updated knowledge about statistical method that would fit best for the analysed problem.
Authors: Victor Chang; Jayang Narang; Lonni Schultz; Ahmad Issawi; Rajan Jain; Jack Rock; Mark Rosenblum Journal: Acta Neurochir (Wien) Date: 2012-04 Impact factor: 2.216
Authors: Binsheng Zhao; Lawrence H Schwartz; Chaya S Moskowitz; Michelle S Ginsberg; Naiyer A Rizvi; Mark G Kris Journal: Radiology Date: 2006-12 Impact factor: 11.105
Authors: G L Walsh; M O'Connor; K M Willis; M Milas; R S Wong; J C Nesbitt; J B Putnam; J J Lee; J A Roth Journal: Ann Thorac Surg Date: 1995-12 Impact factor: 4.330
Authors: Nathan R Foster; Lindsay A Renfro; Steven E Schild; Mary W Redman; Xiaofei F Wang; Suzanne E Dahlberg; Keyue Ding; Penelope A Bradbury; Suresh S Ramalingam; David R Gandara; Taro Shibata; Nagahiro Saijo; Everett E Vokes; Alex A Adjei; Sumithra J Mandrekar Journal: J Thorac Oncol Date: 2015-07 Impact factor: 15.609
Authors: E A Eisenhauer; P Therasse; J Bogaerts; L H Schwartz; D Sargent; R Ford; J Dancey; S Arbuck; S Gwyther; M Mooney; L Rubinstein; L Shankar; L Dodd; R Kaplan; D Lacombe; J Verweij Journal: Eur J Cancer Date: 2009-01 Impact factor: 9.162