2935 - Correlation of progression free survival-2 and overall survival in solid tumors

Background

Overall survival (OS) has become the definitive endpoint in oncology clinical trials, but may be difficult to assess due to need for lengthy follow-up, factors related to trial design or impacted by study crossover. Progression free survival (PFS) is a common primary endpoint, but, differential response to subsequent therapy may affect the correlation of PFS and OS. The European Medicines Agency encourages the use of PFS2, defined as time from initial study randomization to 2^nd disease progression or death from any cause. The surrogacy of PFS2 for OS has not been confirmed. This analysis used published data from solid tumor clinical studies to assess whether PFS2 and OS are correlated.

Methods

A systematic literature search identified solid tumor oncology studies that included data for PFS, PFS2, and OS. Two independent reviewers screened titles, abstracts and main-text of study reports for eligibility, and collected data on median time to PFS2, PFS or time from 1^st progression to 2^nd disease progression or death, and median time to OS. The correlation between PFS2 and OS was assessed and verified in two steps: (1) Kendall’s Tau (Kendall rank correlation coefficient) statistics and Pearson’s correlation coefficient in randomized controlled trials (RCTs); (2) Meta-analysis with the random effects model to compute the pooled correlation of PFS2 and OS.

Results

In total, 133 studies met the search criteria and 15 trials had complete PFS2 and OS data (28 individual treatment and control arms) in ovarian, gastric, colorectal, prostate, lung, renal and breast tumors. A positive correlation was observed between PFS2 and OS (Kendall’s Tau = 0.7 (95% CIs 0.54, 0.78) using all 15 studies. Data from the 10 RCTs showed a Pearson’s correlation coefficient = 0.86. An additional meta-analysis of 3 tumors (breast, colorectal, and lung cancer), where more than 1 study per indication was available (7 studies; 17 individual arms), identified a Spearman’s correlation coefficient = 0.84 (p = 0.0001; 95% CIs 0.71, 0.96).

Conclusions

In this retrospective analysis in solid tumors, PFS2 was positively correlated with OS. The strong correlation provides confidence in using PFS2 as a surrogate for OS before OS data are mature or when OS cannot be assessed.

Clinical trial identification

Editorial acknowledgement

Editorial assistance was provided by Ann C Sherwood, PhD with funding from Janssen Pharmaceuticals.

Legal entity responsible for the study

Janssen Pharmaceuticals.

Funding

Janssen Pharmaceuticals.

Disclosure

P.N. Mainwaring: Leadership role, Travel / Accommodation / Expenses, Shareholder / Stockholder / Stock options, Licensing / Royalties: XING Technologies P/L; Honoraria (self), Advisory / Consultancy, Speaker Bureau / Expert testimony, Travel / Accommodation / Expenses: Janssen; Honoraria (self), Advisory / Consultancy, Speaker Bureau / Expert testimony, Travel / Accommodation / Expenses: Pfizer. L. Zhang: Shareholder / Stockholder / Stock options, Full / Part-time employment: Janssen. S.D. Mundle: Shareholder / Stockholder / Stock options, Full / Part-time employment: Janssen. K. Liu: Shareholder / Stockholder / Stock options, Full / Part-time employment: Janssen. E. Pollozi: Shareholder / Stockholder / Stock options, Full / Part-time employment: IDEA Pharma; Shareholder / Stockholder / Stock options, Full / Part-time employment: Pfizer; Advisory / Consultancy: Janssen. A. Gray: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment: IDEA Pharma; Advisory / Consultancy: Janssen. M. Wildgust: Shareholder / Stockholder / Stock options, Full / Part-time employment: Janssen.