1 Health Related Quality of Life / Patient-reported Outcomes in Relapsed Ovarian Cancer:Results from a Randomized Phase III Study of Trabectedin with Pegylated Liposomal Doxorubicin (PLD) versus PLD Alone C. N. Krasner,1 A. Poveda,2 T. Herzog,3 J. Vermorken,4 B. Monk,5 P. Zintl,6 J. Li,7 Y. Su,7 R. Dhawan,7 S. Kaye8 1Massachusetts General Hospital, Boston, MA; 2Fundación Instituto Valenciano de Oncología, Valencia, Spain; 3Columbia University Medical Center, New York, NY; 4Universitair Ziekenhuis Antwerpen, Edegem, Belgium; 5Chao Family Comprehensive Cancer Center UC, Orange, CA; 6PharmaMar, S.A., Madrid, Spain; 7Johnson & Johnson Pharmaceutical Research & Development, Raritan, NJ; 8Royal Marsden Hospital, Sutton, United Kingdom ABSTRACT Statistical Methods Figure 5 shows cross-sectional cycle-by-cycle analyses of the mean Global Health Status scores per cycle, up to cycle 11, in each treatment arm. Background: In an open-label, multicenter, randomized Phase III study comparing the combination of trabectedin and PLD to PLD alone in patients with relapsed ovarian cancer, the combination demonstrated significantly improved progression free survival and response rates, manageable non- cumulative toxicity, and fewer PLD-associated adverse events. We studied the impact of the combination of trabectedin with PLD on the quality of life (QoL)/ patient-reported outcomes (PRO) evaluated as part of the trial. Methods: QoL/PRO questionnaires, EORTC-QLQ C30, OV28 and EQ-5D were completed by patients at screening and on Day 1 of every other treatment cycle starting with Cycle 1, and at the End-of-Treatment Visit. Global Health Status/QoL, Fatigue, Pain subscales from QLQ C30 and Abdominal Pain/GI symptoms scale from OV28 were chosen a priori for primary analyses. Other scales of the three questionnaires were analyzed on a supportive basis. Results: A total of 672 patients were randomized. 663 (98%) completed at least the baseline questionnaires. Median cycles of treatment was 6 (131 days) for the combination arm and 5 (143 days) for the monotherapy arm. Mixed effects models (using a covariance structure of AR[1]) predicting the score at baseline and follow-up scores as a function of treatment, days after baseline, and interaction between treatment and days after baseline showed no significant differences between the treatment arms for any of the pre-specified scales. Similar analyses of other scales, including EQ-5D Health Index scores and Health State on the Visual Analog Scale, support the findings. Conclusions: The addition of trabectedin to PLD results in superior efficacy in patients with relapsed ovarian cancer, with no added decrement to overall health status as assessed by PRO. Longitudinal Analyses Mixed-effect model with treatment (Yon), time effect (days after baseline), and a timetreatment interaction as covariants using a covariance structure of autoregressive of order 1 (AR(1))  Y=α+ β1*Trt +β2*Day+β3* Trt*Day Figure 5. QLQC30 Global Health Status Scale; Mean Score Over Time,* Cross-sectional Analyses Mean score changes from baseline at each post-baseline assessment and treatment termination was compared between the treatment arms Pre-specified PRO Endpoints QLQ-C30 QLQ-OV28 Global Health Status / QoL Fatigue Pain Abdominal / Gastrointestinal Symptoms *Higher is better Table 2 shows neither Health State from the EQ-5D Visual Analogue Scale (VAS), nor the EQ-5D Health Index scores were significantly different for the two treatment arms in the longitudinal mixed-effect model analyses. In addition to Global Health Status, other overall health status measurements support this finding. The cross-sectional cycle-by-cycle analyses confirmed the finding (data not shown). RESULTS & DISCUSSION Figure 3 shows that the PRO questionnaire completion rates were high (85% overall). Figure 3. PRO Questionnaire Overall Missing Rate Table 2. Overall Health Status – Longitudinal Analyses RATIONALE Estimated Point Change per Day P Value Mean Change During Treatment QLQ-C30 Global Health Status/QOL 0.9820 0.0 EQ-5D EQ Index 0.3385 EQ VAS Health Status 0.9431 -0.1 From mixed-effect model EQ Index score was calculated using U.S. population-based preference weights and algorithm developed by Shaw et al. 2005 Mean Change during Treatment estimated by multiplying point change per day average duration of treatment of 22 weeks or 155 days for the Trabectedin/PLD arm. Trabectedin (ET-743, Yondelis®) is a novel tetrahydroisoquinoline compound isolated originally from the marine ascidian Ecteinascidia turbinate and is now produced synthetically.1 Results from a randomized Phase III study comparing combination of trabectedin and PLD to PLD alone in patients with relapsed ovarian cancer have demonstrated significantly improved progression free survival (PFS) and response rates (RR) in the combination arm.2 Non-cumulative toxicity was manageable and fewer PLD-associated adverse events were observed. Note: Based on the percentage of patients who are still on treatment and who return questionnaires with valid answers. OBJECTIVE Patient drop-out rates (Figure 4) were substantially higher in the PLD arm than in the Trabectedin combo arm, especially in the early cycles (red arrows), suggesting there might be non-random missing bias, raising the possibility that PRO from the PLD arm may appear higher since more patients with worse outcomes had dropped out than in the Trabectedin combo arm. Impact of the combination of trabectedin with PLD on the quality of life (QoL)/ patient-reported outcomes (PRO) evaluated in patients with relapsed ovarian cancer. Among the pre-specified symptom scales, Fatigue shows significant or close to significant worsening in early treatment cycles (cycles 3-9) for the Trabectedin/PLD arm in cycle-by-cycle analyses (not taking into account missing data, no adjustment for multiple comparisons). However, with longer treatment cycles (cycle 11 and beyond), the worsening seems to improve and is no longer statistically significant. Overall it is not statistically significant by the longitudinal analysis (Table 1), although the early cycle worsening is consistent with clinical findings. Among the other PRO scales measured, only Body Image from QLQ-OV28 was statistically worse in the Trabectedin/PLD arm than in the PLD arm by the longitudinal analysis (p=0.025). However, the magnitude of the worsening was only point per day on a scale. For the entire average duration of treatment of 22 weeks or 155 days for the Trabectedin/ PLD arm, the average total point worsening was less than 3.5 points, not clinically meaningful. A 10-point change was generally considered the minimally important change on the EORTC-QLQ scales. Data from the cycle- bycycle analysis was inconsistent (data not shown). Considering multiple comparisons, this singular statistically significant finding was likely to be due to chance. METHODS Figure 4. Patient Retention Rates The ET743-OVA-301 study design is shown in Figure 1 Figure 1. ET743-OVA-301 Study Design PLD 50 mg/m2 90 minute infusion q28d RANDOMIZATION RANDOMIZATION PLD 30 mg/m2 90 minute infusion followed by TRABECTEDIN 1.1 mg/m2 Note: Retention rate = Patients remaining on treatment at each cycle / patients on treatment at baseline Table 1 shows results from the longitudinal mixed-effect model analyses. None of the pre- specified endpoints show statistically significant difference between the two treatment arms. The mean score difference for the Global Health Status scale is essentially zero. QoL/PRO questionnaires were completed by patients at screening and on Day 1 of every other treatment cycle starting with Cycle 1; and at the End-of -Treatment Visit. PRO Instruments used in this study are shown in Figure 2 CONCLUSION Table 1. Pre-specified PRO Endpoints – Longitudinal Analyses Addition of trabectedin to PLD results in superior efficacy (PFS & RR) in patients with relapsed ovarian cancer with no added decrement to overall health status as assessed by PRO. These data support the use of Trabectedin/PLD combination therapy as an efficacious, safe and well-tolerated treatment in relapsed ovarian cancer. Figure 2. PRO Instruments Used Estimated Point Change per Day P Value Mean Score Difference During Treatment QLQ-C30 Global Health Status/QOL 0.9820 0.0 Fatigue 0.0120 0.1928 1.9 Pain 0.0064 0.4700 1.0 QLQ-OV28 Abdominal/Gastrointestinal Symptoms 0.6058 -0.6 From mixed-effect model Mean Change during Treatment estimated by multiplying point change per day average duration of treatment of 22 weeks or 155 days for the Trabectedin/PLD arm. QLQ-C30 QLQ-OV28 EQ-5D Core questionnaire of the European Organization for Research and Treatment of Cancer (EORT C) quality of life questionnaire (QLQ) for the evaluation of HRQol of patients participating in cáncer clinical trials Supplemental QLQ questionnaire specific to ovarian cancer A standardized generic measure of health status developed by the EuroQoL Group. Applicable to a wide range of health conditions and treatments. Provides a simple descriptive profile and a single index value for health status for clinical and economic appraisal. REFERENCES Guan Y, et al. J Biomol Struct Dyn. 1993;10: Monk BJ et al. Annals of Oncology. 19:viii1–viii4, 2008.