OBJECTIVES

: PFS must be ‘reasonably likely to predict clinical benefit’ (FDA, 2007). This research aims to explore and quantify the relationship between PFS and OS in frontline CLL patients.

METHODS

: Systematic literature search conducted March 2017, on efficacy and safety for frontline CLL treatments, identified 34 randomised controlled trials. Patient characteristics were extracted from each study: age, sex, disease stage, % patients with del17p, % patients with del11q, % patients with unmutated IGVH, ECOG status, and % patients with elevated B2 microglobulin (B2M) levels (>3.5 mg/l). Pearson’s correlation was performed on the 17 studies which reported both 24 months PFS (PFS24), and OS at 36 (OS36) or 60 months (OS60) in order to correlate OS36 and OS60 with each extracted variable, and PFS24. The correlational analysis results identified covariates for the regression models. Ordinary least squares (OLS) regression analysis provided predictive equations of OS36 and OS60 from PFS24.

RESULTS

: Pearson’s correlations between PFS24 and OS36, and between PFS24 and OS60 were 0.38 (p= 0.02) and 0.39 (p= 0.06), respectively. Significant negative correlations were found between % patients with del17p and elevated B2M, and OS at both time points. Therefore, OLS regression analysis was performed to predict OS from PFS24, with del17p and elevated B2M as covariates. Eight studies reported OS36, PFS24, del17p, and B2M; five of these studies reported OS60, PFS24, del17p, and B2M. The predictive models show PFS24 is a significant predictor for OS36 and OS60 when del17p and B2M are also incorporated into the model; R: 79.7%, and R: 90.8%, respectively; 1% improvement in PFS24 results in 0.22% and 0.38% improvement in OS36 and OS60, respectively.

CONCLUSIONS

: This research supports the use of PFS as a surrogate endpoint in frontline CLL. This model can be further validated with longer follow-up OS data from trials in frontline CLL.