Longitudinal analysis of quality of life (QoL) within a randomized phase III trial in patients (pts) with metastatic pancreatic adenocarcinoma (MPA)

Abstract

e17544 Background: The FFCD phase III trial in pts with MPA comparing LV5FU2-P followed by gemcitabine (arm A) versus the opposite sequence (arm B) had failed to demonstrate a significant difference in median OS. To longitudinally compare QoL according to treatment sequence we have explore definitions of time until definitive deterioration (TUDD) of QoL scores according to minimal clinically important difference (MCID) cut off. Methods: From August 2003 to May 2006, 202 pts with measurable MPA, PS 0–2, non prior CT were included in 33 centers, 102 in arm A and 100 in arm B. QoL was evaluated using EORTC QLQ-C30 every 8 wks up to death. We focused analyses on the following scores: global health (GH), emotional functioning (EF), physical functioning (PF), fatigue (FA), and pain (PA). TUDD were estimated using Kaplan Meier, and compared using log-rank tests. They were defined as the time interval between randomization and the first occurrence of a decrease in QLQ-C30 score ≥ 5 points without any further improvement in QoL score ≥ 5 points or any further available QoL data. These analyses were repeated using a 10 points MCID, and by including deaths as event. Results: Amongst the 202 pts, 179 had completed at least one QoL questionnaire (89%) and the mean number of available QoL data was 3 (SD: 2.14). Using a 5 points MCID as well as 10 points, TUDD of the 5 scores did not differ according to treatment arm. About 25% of patients reported definitive score deteriorations≥ 5 points and 15%≥ 10 points. Including death as event for a 5 points MCID, median TUDD of GH score was 5.2 months (4.3–6.2) in Arm A and 6.1 months (5.1–6.1) in Arm B (log rank p = 0.50). Median TUDD of FA score was 4.8 months (3.5–6.3) in Arm A and 5.6 months (4.2–7.7) in Arm B (log rank p = 0.76). Median TUDD of scores was reached after median PFS and at median time of second-line PFS. While treatment had no impact, multivariate Cox model showed that tumor localization and progression were independently associated with TTDU (p < 0.05). Conclusions: We have investigated QoL analyses modalities using survival techniques dealing with binding drop out missing data and with an easier clinical interpretation of results. Progression seems to negatively impact QoL. No significant financial relationships to disclose.