ABSTRACT

1. Individual Patient Data Analysis of Progression-Free versus Overall Survival as an Endpoint for Metastatic Colorectal Cancer in Modern Trials: Findings from the 16,700 Patient ARCAD Database Q Shi1, A de Gramont2, M Buyse3, A Grothey1, H-J Schmoll4, MT Seymour5, R Adams6, L Saltz7, RM Goldberg8, CJA Punt9, J-Y Douillard10, JR Hecht11, H Hurwitz12, E Diaz-Rubio13, Pr R Proschen14, NC Tebbutt15, C Fuchs16, J Souglakos17, A Falcone18, DJ Sargent1, For the ARCAD Group 1Mayo Clinic, Rochester MN; 2Hospital Saint Antoine, Paris, France; 3International Drug Development Institute, Louvain-la-Neuve, Belgium; 4Martin-Luther University, Halle, Germany; 5Cancer Rsrch UK Clinical Ctr, Leeds, UK; 6Cardiff University, Cardiff, UK; 7Memory Sloan Kettering Cancer Center, New York, NY; 8Ohio State University Comprehensive Cancer Center, Columbus, OH; 9Academic Medical Center, Amsterdam, Netherlands; 10Centre R Gauducheau, St Herblain, France; 11David Geffen School of Medicine at UCLA, Los Angeles, CA; 12Duke University Medical Center, Durham, NC; 13Hospital Clinico San Carlos, Madrid, Spain; 14Klinikum Bremen-Ost, Bremen, Germany; 15Austin Health, Australia; 16Dana Farber Cancer Institution, Boston, MA; 17 University of Crete, School of Medicine, Heraklion, Greece; 18University Hospital “S. Chiara”, Pisa, Italy ABSTRACT METHODS RESULTS FINDINGS Table 1: Trials Included Figure 1: Overall Treatment arm and Trial-level Surrogacy Background: Progression-free survival (PFS) has previously been established as a surrogate for overall survival (OS) based on individual patient data (IPD) from 1st line metastatic colorectal cancer )mCRC) trials conducted before 1999. As mCRC treatment (trt) has advanced in the last decade and OS has increased from 1 to 2 years, this surrogacy required re-examination. Methods: IPD from 16,762 pts, median age 62, 62% male, 53% ECOG PS 0 were available from 22 1st line mCRC studies conducted from 1997-2006; 12 tested targeted (anti-angiogenic and/or anti-EGFR) regimens. The relationship between PFS (first event of progression or death) and OS was evaluated at patient-, trt-arm-, and trial-levels using correlation (corr.) coefficients and R2 (closer to 1 the better) from weighted least square (WLS) regression of arm-specific survival rates and trial-specific hazard ratios (HRs), estimated using Cox and Copula bivariate models. The concordance of significance (CoS) of the treatment effects (TEs) on both endpoints was calculated. Results: 44% pts received a targeted regimen. Median PFS was 8 and OS was18 months. The corr. between PFS and OS was modest at all three levels with low CoS in TE comparisons (see Table). Analyses limited to trials testing targeted trts, non-strategy trials, or superiority trials did not improve surrogacy. Conclusion: In modern mCRC trials, where survival post-first progression exceeds time to first progression, PFS TEs do not reliably predict TEs on OS. Nonetheless, until alternative endpoints of clinical benefit can be validated, PFS remains a relevant primary endpoint for 1st line mCRC trials, as our data demonstrate that the ability for any agent to produce an OS benefit from a single line of therapy is challenging. • Trial Selection and Patient Characteristics • 22 trials and 43 comparisons were included • Two of 24 1st line trials in ARCAD with regimens identical in both arms before 1st PD were excluded • Trials with multiple arms were prospectively defined to generate two-arms comparisons • Where KRAS available, KRAS wildtype vs. mutated patients were treated as 2 separate comparisons • Trials with regimens identical in both arms before 1st progressive disease (PD) were excluded • The non-reported cohorts were included and treated as separate comparisons • Total of 16,762 patients were included • Age: 14% < 50; 26% 50 – 59; 35% 60 – 69; 25% ≥ 70 • ECOG PS: 53% 0; 42% 1; 5% 2+ • Gender: 61% male; 39% female • Regimen: 44% targeted; 56% non-targeted • Endpoint Definitions • OS – Time from randomization to death due to any cause • PFS – Time from randomization to 1st PD or death due to any cause • When possible, PFS centrally redefined to allow consistent calculations across trials • Statistical Methods • Patient level • Landmark analysis (evaluating prognostic value of PFS at 6 and 12 months for OS) • Rank correlation coefficient, ρ, between PFS and OS (Copula bivariate survival model) • Treatment arm level • Correlation between short-term (6 months) PFS rates and long-term (12 & 18 months) OS rates • Estimated based on Kaplan-Meier estimates • Coefficient of determination from weighted least square (WLS) linear regression, r2WLS • Trial (i.e. comparison) level • Correlation between hazard ratios (HRs) on PFS and OS – Coefficient of determination from linear regression based on HRs estimated by • Cox model (& WLS regression), R2WLS • Copula bivariate survival model, R2Copula • Concordance of significance of the treatment effect • Although early PFS rate is a strong predictor of long-term OS (patient level landmark analysis), treatment effects observed on PFS at the trial-level do not provide sufficiently accurate prediction of treatment effect on OS in the first-line advanced CRC trials. • Trials testing targeted therapies demonstrated only modestly higher PFS/OS correlations at patient-, treatment arm, and trial-level between PFS and OS than non-targeted trials. • Restricting to non-strategy trials, overall results remain: correlation measures between PFS and OS by treatment arm and at trial-level improved by ~ 0.1 in absolute values. • Restricting trials with superiority design, the above results remain. • Excluding one outlier improved the correlation measures only slightly DISCUSSION • PFS has its own merits as an endpoint, and is considered by some a clinical benefit endpoint. However, based on modern trials, PFS treatment effects do not reliably translate into treatment effects on OS. • The lack of association between PFS and OS in modern trials is likely due to extensive use of later-line therapies. • The lack of ability for PFS to predict OS emphasizes rather than detracts from its importance as a clinical trials endpoint to demonstrate activity of a new agent. • PFS remains a relevant primary endpoint for 1st line mCRC trials, as our data demonstrate that the ability for any agent to produce an OS benefit from a single line of therapy is challenging. • 1 HYPOTHESES • Short-term PFS predicts the clinical benefit on long-term OS at the patient-level • Treatment effect measured on PFS predicts the treatment effect on OS at trial-level • Type of treatment (targeted vs. non-targeted) affects the predictive relationship between PFS and OS • Superiority trials will have stronger PFS/OS surrogacy than strategy/non-inferiority trials Table 2: Surrogacy Estimation Abbreviations: C, control arm; E, experimental arm; IRI, irinotecan; OX, oxaliplatin; LV, leucovorin; Bev, bevacizumab; Cap, capecitabine; Pmab, Pamtumumab; WT, wild type; MT, mutated; T, targeted; NT, non-targeted; ANG, Anti-angiogenic regimen; EGFR, Anti-EGFR regimen; OS, overall survival; PFS, progression-free survival; HR, hazard ratio; PD, progressive disease 1.Cross-over reflects the actual cross-over from one arm to another due to toxicity, PD and other reasons. 2Strategy trial refers more treatment specified by protocol beyond the first per protocol regimen. For example, a sequence of treatment was specified per protocol (i.e., the treatment effect of a whole strategy of experimental arm is compared to the control arm), or cross over after PD was specified per protocol. For these studies, the PFS (1st PD-PFS) with the 1st PD occurred during study follow-up as one of the events may not be the appropriate endpoint if the regimen before 1st PD was same for both experimental and control arms. 3The design refers to the intended primary comparison between arms, i.e., non-inferiority (NI) or superiority (Sup) comparison. The statistical sample size/power consideration section may not reflect NI or Sup design.