OBJECTIVES: Diffuse large b-cell lymphoma is an aggressive disease with most of the disease progression happening within the first 2 years with R-CHOP treatment. GO39942 trial is an ongoing Phase III randomized trial (RCT; POLARIX; NCT03274492) that compared the efficacy and safety of Polatuzumab+R-CHP against R-CHOP in untreated DLBCL patients. The aim of this study was to compare two statistical models (mixture-cure model versus standard parametric survival model) in order to determine which model can predict the long-term PFS based on an ongoing trial for 1L DLBCL.

METHODS: The study developed two statistical models in order to extrapolate the PFS data from POLARIX trial: a mixture-cure model and a standard parametric survival model. In order to validate the extrapolations, the curves were compared with the Kaplan Meier curves from a finalized Phase 3 randomized clinical trial (RCT; GOYA; NCT0128774) adjusted with propensity score weighting. The GOYA trial was chosen to validate both models since this finalized phase III trial also evaluated R-CHOP regimen in untreated DLBCL patients.

RESULTS: Although the standard parametric survival model had a good fit to the KM of POLARIX trial, this model could not predict the long-term PFS that is expected to be seen in R-CHOP in the GOYA trial adjusted. Therefore, the standard parametric survival model has demonstrated to underestimate the long-term PFS in 1L DLBCL with the current POLARIX data. Whereas the mixture-cure model had not only a good fit to the KM of POLARIX trial, but it could also predict the long-term PFS in R-CHOP from the GOYA trial.

CONCLUSIONS: Based on the current data available in this ongoing clinical trial, it is possible to conclude that the mixture-cure model have a better prediction of the long-term PFS in 1L DLBCL.