D P Modest1, I Ricard2, S Stintzing3, L Fischer von Weikersthal4, T Decker5, A Kiani6, U Vehling-Kaiser7, S-E Al-Batran8, T Heintges9, C Kahl10, G Seipelt11, F Kullmann12, W Scheithauer13, M Moehler14, C B Westphalen15, J W Holch15, J C von Einem15, S Held16, V Heinemann3. 1. Department of Medicine III and Comprehensive Cancer Center, University Hospital, LMU Munich, Germany; German Cancer Consortium (DKTK), German Cancer Research Centre (DKFZ), Heidelberg, Germany. Electronic address: dominik.modest@med.uni-muenchen.de. 2. Institute of Medical Informatics, Biometry, and Epidemiology, University of Munich, Munich, Germany. 3. Department of Medicine III and Comprehensive Cancer Center, University Hospital, LMU Munich, Germany; German Cancer Consortium (DKTK), German Cancer Research Centre (DKFZ), Heidelberg, Germany. 4. Gesundheitszentrum St. Marien, Amberg, Germany. 5. Oncological Practice, Ravensburg, Germany. 6. Medizinische Klinik IV, Klinikum Bayreuth, Bayreuth, Germany. 7. Oncological Practice, Landshut, Germany. 8. Department of Hematology and Oncology, Krankenhaus Nordwest Frankfurt/Main, Germany. 9. Department of Medicine II, Städtisches Klinikum Neuss, Germany. 10. Haematology and Oncology, Staedtisches Klinikum Magdeburg, Germany. 11. Oncological Practice, Bad Soden, Germany. 12. Department of Medicine I, Klinikum Weiden, Germany. 13. Department of Internal Medicine I and Comprehensive Cancer Center, Medical University Vienna, Austria. 14. German Cancer Consortium (DKTK), German Cancer Research Centre (DKFZ), Heidelberg, Germany; Medical Department 1, Johannes-Gutenberg Universität Mainz, Mainz, Germany; University Cancer Center Frankfurt/Mainz, Germany. 15. Department of Medicine III and Comprehensive Cancer Center, University Hospital, LMU Munich, Germany. 16. ClinAssess GmbH, Leverkusen, Germany.
Abstract
BACKGROUND: We explored the impacts of sequential application of various treatment lines on survival kinetics. Therefore, differences in overall survival (OS) observed in FIRE-3 were investigated in the context of time and exposure to applied treatment. PATIENTS AND METHODS: OS analyses (stratified by treatment with FOLFIRI plus either cetuximab or bevacizumab) were performed according to time intervals as well as using a Cox model to define changes of hazard ratio (HR) over time. RESULTS: The fraction of patients with systemic treatment and time on treatment markedly decreases over treatment lines and time. OS evaluation by a Cox model indicated a trend towards a non-proportional hazard between treatment arms (P = 0.12/P = 0.09 for KRAS-intention-to-treat (ITT)/all-RAS wild-type populations, respectively). To improve the fit of the model, a change-point (point of curve separation) was estimated at 22.6 months (day 687) after randomisation. The HR between the two arms before 22.6 months was not significantly different from one. However, markedly different survival kinetics in favour of the cetuximab arm were apparent after the change-point (KRAS-ITT: P = 0.0018; HR, 0.60 [95% confidence interval [CI], 0.44-0.83] and RAS: P = 0.0006; HR, 0.51 [95% CI, 0.35-0.75]). CONCLUSION: The differences in OS favouring the cetuximab arm become apparent about 22.6 months after randomisation, indicating that only those patients who survive 22.6 months after randomisation benefit from the superiority of the cetuximab arm. When OS curves separate, only few patients receive active systemic treatment in short courses, suggesting that earlier treatment effects are responsible for later kinetics of survival curves.
RCT Entities:
BACKGROUND: We explored the impacts of sequential application of various treatment lines on survival kinetics. Therefore, differences in overall survival (OS) observed in FIRE-3 were investigated in the context of time and exposure to applied treatment. PATIENTS AND METHODS: OS analyses (stratified by treatment with FOLFIRI plus either cetuximab or bevacizumab) were performed according to time intervals as well as using a Cox model to define changes of hazard ratio (HR) over time. RESULTS: The fraction of patients with systemic treatment and time on treatment markedly decreases over treatment lines and time. OS evaluation by a Cox model indicated a trend towards a non-proportional hazard between treatment arms (P = 0.12/P = 0.09 for KRAS-intention-to-treat (ITT)/all-RAS wild-type populations, respectively). To improve the fit of the model, a change-point (point of curve separation) was estimated at 22.6 months (day 687) after randomisation. The HR between the two arms before 22.6 months was not significantly different from one. However, markedly different survival kinetics in favour of the cetuximab arm were apparent after the change-point (KRAS-ITT: P = 0.0018; HR, 0.60 [95% confidence interval [CI], 0.44-0.83] and RAS: P = 0.0006; HR, 0.51 [95% CI, 0.35-0.75]). CONCLUSION: The differences in OS favouring the cetuximab arm become apparent about 22.6 months after randomisation, indicating that only those patients who survive 22.6 months after randomisation benefit from the superiority of the cetuximab arm. When OS curves separate, only few patients receive active systemic treatment in short courses, suggesting that earlier treatment effects are responsible for later kinetics of survival curves.