Combined analysis of randomized controlled trial (RCT) and patient-preference trial (PPT) evaluating second-line chemotherapy (SLC) in advanced gastric cancer (AGC).

Abstract

4064 Background: While prolonged overall survival (OS) from SLC compared to best supportive care (BSC) has recently been demonstrated for AGC (Park et al, ASCO 2011), the prognosis of pretreated AGC remains poor. We assessed whether patient preference, willing to participate onto RCT, or other parameters contributed to this OS improvement. Methods: In the phase III trial, pretreated AGC patients (n=372) were first offered RCT. If a patient agreed to participate, randomly assigned 2:1 to SLC or BSC (n=202). If refused RCT, but agreed to receive treatment of their preference, they were offered SLC or BSC (PPT; n=170). OS of RCT participants was compared to that of whole patient population according to prognostic subgroups. In addition, analyses of OS unadjusted for multiple comparisons were conducted across predefined subgroups. Results: Median OS was 6.3 months among 254 patients treated with SLC and 3.3 months among 118 patients treated with BSC (HR, 0.507; 95% CI, 0.405 to 0.637; one-sided p<0.001). Compared to the RCT patients, younger age group, those with an ECOG performance status (PS) of 0, and with one prior chemotherapy were under-represented in the PPT patients. OS was comparable between RCT (5.0 months) and PPT (4.4 months) patients even after controlling for major prognostic factors (log-rank p=0.322). The OS benefit for SLC was preserved when analyzed according to baseline parameters. Parameters that were associated significantly with a prolonged OS included a PS of 0, the chemotherapy-free interval ≥3 months, and one prior chemotherapy. When the analysis was adjusted for these three parameters, patients who received SLC still had improved OS (HR, 0.554; 95% CI, 0.441 to 0.696; p<0.001). Conclusions: The 54% participation rate obtained in the current study represents the best achievable rate for this kind of phase III RCT involving BSC only arm. Even if we consider differences in the baseline characteristics between RCT and PPT patients, it is concluded that the RCT results can be generalized for the patient population and provided additional evidence supporting the use of SLC in patients with pretreated AGC.