OBJECTIVES: Extrapolation of overall survival (OS) in cost-effectiveness modeling is traditionally is limited to a single intervention or line of therapy. However, there are no known models, to date, that reflect the incremental impact of a new intervention on OS at the population level. We present a method that predicts OS across lines of therapy and treatments, which represents a more comprehensive model considering real-world treatment patterns, progression-free survival (PFS), and OS.

METHODS:A cohort treatment simulation was created to predict OS across multiple treatment lines at the population level. The simulation follows population cohorts as they progress through each line of therapy (by treatment regimen) or discontinue treatment. Cohorts start on the first lines of therapy at cycle zero. For each cycle, a portion of those who are alive transition to the next line of therapy or discontinue (informed by the real-world data for treatment regimen utilization for lines of therapy and treatment-specific PFS and OS curves). As cohorts start the next line of therapy, they follow PFS and OS curves from that therapy from an effective start cycle of zero. By tracking each cohort from the cycle they progress, the start of the next line of therapy curve effectively shifts forward in time to when the progression occurs. This process continues over the simulated time horizon, after which median survival of all cohorts is calculated and reported as the median OS of the disease from diagnosis.

CONCLUSIONS: Our method of predicting OS across multiple lines of therapy and treatment regimens offers a more comprehensive simulation of real-world outcomes than current methods do. As oncology regimens become increasingly targeted, a methodology to evaluate the impact of new therapies on population-level outcomes is warranted.