Please activate JavaScript!
Please install Adobe Flash Player, click here for download

ISCB2014_abstract_book

ISCB 2014 Vienna, Austria • Abstracts - Poster Presentations 95Monday, 25th August 2014 • 15:30-16:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust P1.2.67 Adapted levels of evidence for small populations G Hlavin1 , F König1 , M Posch1 , C Male2 , P Bauer1 1 Medical University of Vienna, CeMSIIS, Vienna, Austria, 2 Medical University of Vienna, Paediatrics, Vienna, Austria   A full independent development programme to demonstrate efficacy may not be feasible in small populations such as paediatrics populations or or- phan indications.We propose clinical trials designs, which make use of pri- or knowledge on efficacy and safety for inference in the small population. For example, we establish how studies conducted in related (larger) popu- lations and/or studies conducted with similar drug substances can serve as basis to partly extrapolate efficacy and safety in the small population groups of interest. By using this strategy, the regulatory need for indepen- dent evidence in the small population groups of interest can be lowered. Depending on the degree of similarity and prior knowledge, e.g. between the paediatric and adult populations, the required level of evidence, test- ing strategy and sample sizes will be adopted. The less uncertainty is left, the higher the frequentist significance level for the validation study in the small population group may become. We show how such an adapted significance level can be motivated in terms of prior information in the framework of the false positive report- ing probability as well as in the Bayesian paradigm. We translate Bayesian decisions rules into classical frequentist decision making criteria. Acknowledgement: This project has received funding from the European Union’s 7th Framework Programme for research, technological develop- ment and demonstration under Grant Agreement no 602552.   P1.2.68 Retrospective evaluation of the futility analysis in LUME-Lung 2, a phase III trial of nintedanib plus pemetrexed for NSCLC patients J Hocke1 , P Glomb1 , R Kaiser1 , J Barrueco2 , B Gaschler-Markefski1 1 Boehringer Ingelheim Pharmaceuticals GmbH & Co. KG, Biberach, Germany, 2 Boehringer Ingelheim Pharmaceuticals Inc., Ridgefield, United States   Background: LUME-Lung 2 (NCT00806819; 1199.14) was a placebo-con- trolled phase III trial of nintedanib + pemetrexed for 2nd -line non-squa- mous NSCLC. The primary endpoint was progression-free survival (PFS) by central review. A preplanned futility analysis based on investigator- assessed PFS (secondary endpoint) was conducted by an independent data monitoring committee (DMC) after ~50% events. The futility crite- ria as defined in the DMC charter were met and the trial was halted with 713/1300 planned patients. Based on the DMC recommendation to stop the trial (without any safety concerns), ongoing patients were unblinded and follow-up was continued according to an amended protocol. Methods: Final study data were used for a retrospective evaluation of the conditional and predictive power over time that led to the DMC recom- mendation. Both investigator and centrally assessed PFS data were used. Results: The retrospective analysis confirmed that based on investigator- assessed PFS, the futility criteria were met at the time of the DMC analysis. However, when analyzed at other timepoints, the same statistical mea- sures remained above the predefined futility criteria. Also, when centrally reviewed PFS was analyzed over the same time period, fluctuations in con- ditional and predictive powers were observed but both remained above the formal futility criteria. Furthermore, a significant prolongation of PFS by central review was shown based on final data. Conclusions: Retrospective investigations suggest that had the DMC analysis been performed at another timepoint or had centrally reviewed PFS data been used, the futility outcome may have been different and the trial may have been continued.   P1.2.86 Cancelled Comparison of two-stage versus two separate single-stage settings for bioequivalence studies with crossover design S Knahl1 , A Seidel1 , F Fleischer1 1 Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach an der Riß, Germany   Usually the aim of conducting a bioequivalence trial is to compare the bioavailability of two or more different formulations in a statistical sense. A gold standard design for this purpose is given by a two-period, two- sequence crossover. Historically, such a design has been performed in a single-stage setting where recently it is more and more common to imple- ment two-stage designs which are also acknowledged by regulatory au- thorities [4][5]. Potvinetal.[1]proposedfourpossiblemethodsofsuchatwo-stagedesign. The properties of those methods were further evaluated by Montague et al. [2] as well as further characterized by Karalis and Macheras [3]. In this talk the different approaches are investigated and compared. The purpose is to identify scenarios and rules, which help to decide under which circumstances which specific setting is preferable. Various test cases, which vary in terms of sample size, power, geometric mean ratio, and intra-subject variability are considered and compared by means of simulations. These investigations can support the decision process on the preferred setting for further bioequivalence trials. References: [1] PotvinD. et al. Sequential design approaches for bioequivalence stud- ies with crossover design. Pharmaceutical Statistics 2008;7:245-262. [2] MontagueTH. et al. Additional results for‘Sequential design approach- es for bioequivalence studies with crossover designs’. Pharmaceutical Statistics 2012;11:8-13. [3] KaralisV, MacherasP. An insight into the properties of a two stage design in bioequivalence studies. Pharmaceutical Research 2013;30(7):1824-35. [4] FDA-Guidance for Industry: Bioavailability and Bioequivalence Studies for Orally Administered Drug Products.March2003. [5] EMA-Guideline on the Investigation of Bioequivalence.January2010   P1.2.88 Adaptive two-stage bioequivalence trials with early stopping and sample size re-estimation F König1 , M Wolfsegger2 , T Jaki3 , H Schütz4 , G Wassmer5 1 Medical University of Vienna, CeMSIIS, Vienna, Austria, 2 Baxter Innovations GmbH, Vienna, Austria, 3 Lancaster University, MPS Research Unit, Lancaster, United Kingdom, 4 BEBAC, Vienna, Austria, 5 Aptiv Solutions, Cologne, Germany   Bioequivalence between two products has to be demonstrated as an es- sential part of the generic approval process (new formulation vs. innova- tor product), bridging an innovator´s product from the formulation used in clinical phase III to the marketed formulation, or in case of major varia- tions of an approved product. The most common design of bioequiva- lence studies is a two-sequence two-period two-treatment crossover design, where inclusion of 90% confidence intervals of pharmacokinetic metrics in a pre-defined acceptance range has to be shown. Alternatively, bioequivalence can be established by using Two One-Sided Tests (TOST) each at an alpha level of 5%. %. However, a fixed sample ap- proach offers no flexibility, e.g., if there is high uncertainty about the as- sumed ratio and/or CV. We propose a two-stage adaptive design using combination tests to com- bine stagewise p-values. This will strictly control the type I error rate in case data-driven design modification have to be performed at an interim analysis. We derive 90%-repeated confidence intervals for the adaptive TOST approach. We investigate different sample size reassessment strat- egies using conditional power arguments. We discuss how futility stop-

Pages Overview