Lors de la 22° conférence annuelle d'ISOQOL à Vancouver (Ca), Amélie Anota a présenté une communication orale et un poster. En voici les résumés :
___Impact of bevacizumab added to temozolomide-chemoradiation on time to health-related quality of life deterioration in unresectable glioblastoma: results of a phase II randomized clinical trial (ORAL COMMUNICATION)
Amélie Anota(1,2), Sandrine Dabakuyo-Yonli(1,3), Astrid Pozet(2), Brice Paquette, Loïc Feuvret(4), Luc Taillandier(5), Didier Frappaz(6), Hervé Taillia(7), Roland Schott(8), Jérôme Honnorat(9), Michel Fabbro(10), Isabelle Tennevet(11), François Ghiringhelli(12), Chantal Campello(13), Daniel Castera(14), Marc Frenay(15), Cécile Dalban(3), Jérémy Skrzypski(16), Olivier Chinot(17), Bruno Chauffert(18), Franck Bonnetain(1,2)
Two clinical trials investigating the addition of bevacizumab (BEV) to TMZ treatment (Chinot et al., NEJM 2014; Gilbert et al., NEJM 2014) for unresectable glioblastoma (UGB) highlighted an improvement of progression-free survival (PFS) but no effect on overall survival. Health-related quality of life (HRQoL) results were divergent and compromised the conclusion about clinical benefit of BEV. A phase II clinical trial evaluating BEV and irinotecan (IRI) in addition to TMZ-based chemoradiation for UGB was conducted (Chauffert et al., Annals of Oncology 2014). The primary objective 6-month PFS rate from 50% to 66% was not reached. This present study focuses on the HRQoL analysis (secondary endpoint).
EORTC QLQ-C30 and its BN20 brain cancer module were used at baseline, during treatment and every 4 weeks until progression. Time until definitive HRQoL score deterioration (TUDD) was used as a modality of longitudinal HRQoL analysis, with a 10-point Minimal Clinically Important Difference, including or not death as an event. TUDD was estimated with the Kaplan-Meier method. Cox model was used to estimate Hazard Ratios (HR) and its 95% confidence interval (CI). Multivariate Cox model investigated factors associated with TUDD.
Among the 120 patients included from 2009 to 2011, 90 patients (75%) filled the baseline HRQoL questionnaires (45 in each treatment arm). For TUDD, patients in BEV/IRI arm presented a longer TUDD than those of TMZ arm for 9/15 dimensions of the QLQ-C30 and 10/11 dimensions of the BN20. As an example, patients in BEV/IRI arm presented a longer TUDD of cognitive functioning than those of TMZ arm (Median 3.8 months (95%CI 2.5-NA) for BEV/IRI vs. 5.0 (95%CI 3.0-NA) for TMZ, HR=0.80 (0.41-1.57)).
Patients in BEV/IRI arm presented a longer TUDD than those of the TMZ arm. These HRQoL results are consistent to those obtained with AVAGLIO trial suggesting HRQoL benefit for patients with BEV.
___Impact of Response Shift effect on the longitudinal analysis of Health-related quality of life in oncology clinical trials: a simulation study (POSTER)
Amélie Anota(1,2), Antoine Barbieri(3,4), Marion Savina(5,6), Sophie Gourgou-Bourgade(3), Franck Bonnetain(1,2), Caroline Bascoul-Mollevi(3)
1 National Platform Quality of Life and Cancer, France
2 Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besançon, Besançon, France
3 Biostatistic unit, Institut régional du Cancer de Montpellier (ICM) - Val d’Aurelle, Montpellier, France
4 Institut de Mathématiques et de Modélisation de Montpellier, University of Montpellier 2, France
5 INSERM, Clinical and Epidemiological Research Unit (CIC-EC 7) – CTD INCa, Institut Bergonié, Bordeaux, France
6 INSERM CIC-EC7 Axe Cancer, Université de Bordeaux, France
An important challenge of the longitudinal analysis of Health-related quality of life (HRQoL) is the potential occurrence of a Response shift effect (RS) resulting in a recalibration, a reprioritization and a reconceptualization of HRQoL. A differential occurrence of the RS effect on each treatment arm should be investigated to assess if over- or under estimation of the treatment effect could occur due to RS. The objective is to investigate and compare the impact of the occurrence of the recalibration component of the RS effect on three statistical methods for the longitudinal analysis through a simulation study.
The methods proposed were the score and mixed model, the longitudinal partial credit model (LPCM) and time to event analysis approach based on the time to HRQoL score deterioration (TTD). Several reference scores were explored as the reference score in the TTD approach (baseline, best previous or immediately previous score) in order to take into account the occurrence of the RS effect. Simulations compared the methods regarding the bilateral type I error and statistical power of the test of an interaction effect between treatment arm and time.
Longitudinal HRQoL data were performed using a LPCM considering that the latent trait follows a multivariate normal distribution. Five measurement times were simulated assuming that a RS effect occurred since time #3. Both uniform and non-uniform recalibration effect were simulated by varying the item difficulty parameter for one or both treatment arms, in the same direction as the time effect.
When a uniform recalibration occurred in only one arm, type I error rate highly increased and the statistical power highly decreased for all statistical methods. When a non-uniform recalibration occurred in only one arm, type I error rate remained unchanged while the statistical power slightly decreased for all statistical methods except the TTD as compared to the best previous or the immediately previous score for which the statistical power increased.
The occurrence of a differential RS effect seemed to bias the results for all statistical approach except for the TTD as compared to the best previous or the immediately previous score seemed in the presence of a non-uniform recalibration.
Vous pouvez retrouver le poster ici :
A propos de Amélie ANOTA
Amélie Anota is a research engineer in biostatistics. She passed a master’s degree in mathematics for computer graphics and statistics at the University of Burgundy. She defended in 2014 a PhD in Biostatistics at the University of Franche-Comté. Her PhD research focused on the methodology of longitudinal analysis of health-related quality of life data in oncology. She is statistician for the French National Platform Quality of Life and Cancer since 2011. She realizes longitudinal statistical analysis of health-related quality of life data from oncology clinical trials and also realizes validation of the psychometrics properties of some health-related quality of life questionnaires.