A multi-state model approach for prediction in chronic myeloid leukaemia

• Verfasst von: Lauseker, Michael [VerfasserIn]
• Verfasst von: Müller, Martin Christian [VerfasserIn]
• Verfasst von: Hehlmann, Rüdiger [VerfasserIn]
• Titel: A multi-state model approach for prediction in chronic myeloid leukaemia
• Verf.angabe: Michael Lauseker, Joerg Hasford, Verena S. Hoffmann, Martin C. Müller, Rüdiger Hehlmann, Markus Pfirrmann, German CML Study Group
• E-Jahr: 2015
• Jahr: 3 December 2014
• Jahr des Originals: 2014
• Umfang: 9 S.
• Fussnoten: Gesehen am 08.01.2018
• Titel Quelle: Enthalten in: Annals of hematology
• Ort Quelle: Berlin : Springer, 1955
• Jahr Quelle: 2015
• Band/Heft Quelle: 94(2015), 6, Seite 919-927
• ISSN Quelle: 1432-0584
• DOI: doi:10.1007/s00277-014-2246-2
• Datenträger: Online-Ressource
• Sprache: eng
• K10plus-PPN: 1566920949

Abstract

Multi-state models support prediction in medicine. With different states of disease, chronic myeloid leukaemia (CML) is particularly suited for the application of multi-state models. In this article, we tried to find a model for CML that allows predicting the prevalence of three different states (initial state of disease, remission and progression) in dependence on treatment, adjusted for age, sex and risk score. Based on the German CML Study IV, one of the largest randomised studies in CML, the model was able to represent the known effects of age and risk score on the probabilities of remission and progression. Patients achieving a major molecular remission had a better chance of surviving without progression, but this effect was not significant. Comparing treatments, patient of the high-dose arm had the greatest chance to be in the state “remission” at 5 years but did not seem to have an advantage considering “progression”. The proposed illness-death model can be useful for predicting the course of CML based on the patient’s individual covariates.