Relationship between response rates and median progression-free survival in non-Hodgkin's lymphoma: A meta-analysis of published clinical trials

Demonstration of clinical effectiveness of a non-Hodgkin's lymphoma (NHL) treatment generally involves determination of progression-free survival (PFS). However, the long evaluation time of PFS limits its utility to make timely decisions in drug development. Therefore, the objective of this analysis was to determine the relationship between response rates and median PFS in NHL. A database was systematically developed from 513 identified NHL trials reported from 1996 to 2015. Potential predictors of the relationship between response rates and PFS were evaluated, including age, sex, treatment, percentage of treatment-naive patients, and subtype of NHL. Seventy-three trials involving 86 cohorts were included in the meta-analysis. Linear regression analysis using logit of response rates and logarithm of median PFS indicated that the correlation between overall response rate (ORR) and median PFS was higher (R-2=0.70) when compared to that of complete response (CR) rate and median PFS (R-2=0.57). Furthermore, the correlation was improved with the addition of percentage of treatment-naive patients and percentage of patients with follicular lymphoma (FL) (P<.005) between ORR and median PFS (R-2=0.78), and between CR rate and median PFS relationship (R-2=0.74). Treatment was not found to alter this relationship. In summary, ORR is as good as CR rate in predicting median PFS. Moreover, longer median PFS is expected in the trials including treatment-naive and/or FL patients at a given ORR/CR rate. The determined relationship can be used to project the median PFS based on ORR or CR rate.