OBJECTIVES: MM-009 (N=353) and MM-010 (N=351) are, randomised, phase III clinical trials designed to evaluate the efficacy and safety of lenalidomide/ dexamethasone compared with dexamethasone alone in patients with relapsed or refractory multiple myeloma. It was necessary to extrapolate survival outcomes beyond the trial data as not all patients had died. Using standard parametric models, overall survival (OS) curves fell below the progression-free survival (PFS) curve in the extrapolated period. The objective of this research was to determine whether multi-state Markov modelling (MSM) could be applied in this scenario so the logical relationship of OS>PFS could be retained. The poster describes the steps required to apply MSM models to patient level data, including covariate adjustment, the application of time varying transition intensities and incorporating uncertainty into economic models. METHODS: MSM models describe how an individual moves between a series of health states, governed by transition intensities. A three-state MSM model was utilised for the combined data using the ‘MSM’ package in R. The model had three health states; pre-progression, progressive disease and death. Output from the model included transition probability matrices; estimating the probability of moving between health states in a defined interval. Survival curves were produced by multiplying out the transition matrices and calculating the estimated proportion of patients alive (OS), and alive and progression free (PFS) over time. RESULTS: The MSM model estimated that for lenalidomide, mean OS was 4.29 [95% CI; 2.22, 5.92] years and for PFS 2.19 [95% CI; 1.77, 2.73] years. Visually, the fit to the observed data was similar for both MSM and parametric models. However, curves no longer crossed in the extrapolated portion. CONCLUSIONS: The MSM model predicted OS and PFS beyond the trial data without the 2 curves crossing. Consequently, the MSM model predicted a higher mean OS, and lower mean PFS than the parametric model.