Virtual Library
Start Your Search
L.R. Pilz
Moderator of
-
+
SC28 - Novel Clinical Trial Designs (ID 352)
- Event: WCLC 2016
- Type: Science Session
- Track: Trial Design/Statistics
- Presentations: 4
- Moderators:S. Michiels, L.R. Pilz
- Coordinates: 12/07/2016, 11:00 - 12:30, Lehar 1-2
-
+
SC28.01 - Phase I Trials of Targeted Therapies (ID 6717)
11:00 - 11:20 | Author(s): A. Adjei
- Abstract
- Presentation
Abstract not provided
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
SC28.02 - Umbrella and Basket Designs (ID 6718)
11:20 - 11:40 | Author(s): D. Tan
- Abstract
- Presentation
Abstract not provided
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
SC28.03 - Biomarker “Test and Validation” Designs (ID 6719)
11:40 - 12:00 | Author(s): M. Redman
- Abstract
- Presentation
Abstract not provided
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
SC28.04 - Adaptive Clinical Trial Designs (ID 6720)
12:00 - 12:20 | Author(s): V. Papadimitrakopoulou
- Abstract
- Presentation
Abstract:
Interest in adaptive design study methods stems from the principle that these methods hold promise for improving drug development compared to conventional study design (i.e., non-adaptive) methods. The theoretical advantages of adaptive designs are that (1) they provide similar information more efficiently by reducing sample size and total cost, (2) increase the likelihood of success on the study objective, treating more patients with more effective treatments or (3) lead to better improved appreciation of the effects of therapy such as dose-response relationship or subgroup effects, for example identifying efficacious drugs for specific subgroups of patients based on their biomarker profiles, which may also lead to more impactful subsequent studies). Adaptive designs use accumulating data to modify the ongoing trial without undermining the integrity and validity of the trial. They also hold the potential for shortening the time for drug development. Several aspects of these trials including the dose-finding scheme, interim analysis, adaptive randomization, biomarker-guided randomization, and seamless designs will be discussed. Many, but not all adaptive designs are devised under the Bayesian framework incorporating principles such as (I) obtaining the prior distribution; (II) collecting data to calculate the data likelihood; and then (III) computing the posterior distribution. The Bayesian framework provides an ideal statistical framework for adaptive trial designs (1, 2). Examples of trials conducted with adaptive designs include the BATTLE and BATTLE-2 trials and ISPY-2. The basic principle is that patients enrolling earlier in a trial are used to inform how subsequent patients are treated, thus improving the efficiency of the study; this means that fewer patients are required to achieve the same answers regarding safe dosing and/or efficacy. The BATTLE and BATTLE-2 trials are prime examples of this approach. Both trials have implemented adaptive randomization schemes to assign patients to the more efficacious treatments based on their biomarker-guided profiles, and use interim analyses to monitor the efficacy outcomes during the trial. The BATTLE trial (3, 4) enrolled patients with stage IV recurrent non-small cell lung cancer, employing a primary endpoint of eight-week disease control rate, as a binary outcome. Four targeted therapies, erlotinib, vandetanib, erlotinib plus bexarotene, and sorafenib, were evaluated, with one therapy targeting each one of four biomarker profiles and it used an adaptive randomization scheme to allocate patients to the different treatments; hence, patients had higher probabilities of being assigned to better treatments based on their biomarker profiles. The trial showed that adaptive design could work in a complex trial that assessed multiple drugs and biomarkers and required tissue collection and biomarker analysis. Based on the findings of the BATTLE trial, a follow-up BATTLE-2 trial (5) was started, that evaluated four treatment regimens, erlotinib, sorafenib, erlotinib + MK2206, and MK2206 + AZD6244, in a two-stage design with adaptive randomization. The first stage was completed with 200 patients. Biomarker selection was planned in 3 steps: training, testing and validation. In the training step, 10–15 potential prognostic and predictive markers were selected from the previous BATTLE experience, cell line data, and relevant literature information. In the testing step, the selected markers are tested using the data acquired from stage 1 of the BATTLE-2 trial. In the validation step, the markers selected in the first stage of the BATTLE-2 trial are used for adaptive randomization in the second stage of BATTLE-2. In BATTLE-2, we pre-specified an extremely limited set of markers and our intent was to use the first half of the study (200 patients) to conduct prospective testing of biomarkers/gene signatures. Predictive markers were to be used to guide patient assignments in the second half of the study. Although the design theoretically provided advantages, since clear predictive markers did not exist for any of the treatment Arms, activity was modest yielding no new predictive markers and not warranting further exploration. The ISPY-2 trial (6) is a multicenter phase II trial in the neoadjuvant setting for patients with breast cancer. The primary end point is pathologic complete response (PCR) at the time of surgery. The patient population is partitioned into ten subgroups depending on hormone-receptor (HR) status, HER2 status and Mamma Print signature. Experimental drugs are added to neoadjuvant therapy with the overall goal to prospectively learn as efficiently as possible which patients respond to each experimental treatment based on their biomarker profiles. Adaptive randomization with interim analysis is used within each biomarker subgroup, with the treatments that are performing better within a subgroup being assigned with greater probability to patients belonging to that subgroup. The phase II drug-screening stage is followed by a phase III confirmatory stage. The ISPY-2 trial has recently shown that two promising drugs improve response rates in specific biomarker subsets and has graduated these two drugs veliparib and neratinib for further development (7). The pharmaceutical industry and regulatory agencies are therefore very interested in adaptive designs because of their potential advantages and because they reflect medical practice in the real world. To recapitulate, incorporation of adaptive designs in carefully designed and executed trials can enhance drug development, provide greater benefit to the enrolled patients, and effectively address many research questions of interest. These designs require deep understanding of theoretical statistical methodology, extensive modeling with simulations, specialized software and robust databases. Continued implementation in trials with guidance from regulatory agencies and innovative methods will contribute towards progress in therapies. 1.Berry DA. Bayesian clinical trials. Nat Rev Drug Discov. 2006;5:27–36. 2.Lee JJ, Chu CT. Bayesian clinical trials in action. Stat Med. 2012;31:2955–2972. 3. Zhou X, Liu S, Kim ES, et al. Bayesian adaptive design for targeted therapy development in lung cancer-a step toward personalized medicine. Clin Trials. 2008;5:181–193. 4. Kim ES, Herbst RS, Wistuba II, et al. The BATTLE Trial: Personalizing therapy for lung cancer. Cancer Discov. 2011;1:44–53. 5.Papadimitrakopoulou V, Lee JJ, Wistuba II et al. The BATTLE-2 Study: A Biomarker-integrated targeted therapy study in previously treated patients with advanced non-small cell lung cancer. J Clin Oncol Aug 1,2016 Epub ahead of print 6.Barker AD, Sigman CC, Kelloff GJ, et al. I-SPY 2: An adaptive breast cancer trial design in the setting of neoadjuvant chemotherapy. Clin Pharmacol Ther. 2009;86:97–100. 7.Quantum Leap. I-SPY 2 Trial graduates 2 new drugs. 2013 Available online:http://www.quantumleaphealth.org/spy-2-trial-graduates-2-new-drugs-press-release/
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
Author of
-
+
YI01a - Clinical Trials & Scientific Mentoring (ID 414)
- Event: WCLC 2016
- Type: Young Investigator Session
- Track: WCLC 2016
- Presentations: 1
- Moderators:L. De Petris
- Coordinates: 12/04/2016, 08:00 - 09:45, Hall C8
-
+
YI01a.02 - Basic Statistical Considerations (ID 6736)
08:20 - 08:40 | Author(s): L.R. Pilz
- Abstract
- Presentation
Abstract:
Introduction: Published and officially approved medical research is based on evidence and subsequently, statistical methods are an essential part in proving the usefulness of results. The translation in statistical terms in most cases is to build hypotheses and their alternatives to be tested. Clearly, medical researchers need some sound understanding of statistical principles which can be taken, however, not as a matter of course. The aim of the contribution is to communicate among readers of medical journals and reports statistical matters focusing on basic statistical considerations to enable a better understanding. [1] Essentials of statistical analysis and reporting: (i) Making the information content of the research results visible in summarizing and prescinding them in tables, graphs, and figures. (ii) Assessing and quantifying any associations of reported measures like possible differences in the outcome of treatment actions etc., and using confidence intervals to express the uncertainty of those associations. (iii) Building hypotheses and their alternatives to prove that these associations have a real biomedical basis which is performed by statistical testing under a given level of significance (p-values). Important is the design of the research project: In randomized trials comparisons are an inherent part of those associations whereas in nonrandomized studies no direct conclusion can be driven that any association not due to chance indicates a causal relationship. Methods: Randomization is a process in which each of the patients has the same but not necessarily the equal chance to be assigned to predefined treatment arms ensuring that the treatment arms are comparable with respect to known or unknown risk factors. Hence, it is a method to remove selection and accidental bias and to guarantee the validity of statistical tests. Main design issues of studies are the formulation of the primary aim, the question of blinding, and the boundary conditions of sample size calculations. [2] Tables of baseline data and outcome events are part of most medical journal papers concerning treatments. Generally the first table displays the patients’ characteristics including some demographic variables and variables related to the primary aim. The main outcome events are forming the key table of every paper stratified by treatment groups. Categorical variables are shown as number and percent by group. Continuous variables can either be presented by mean and the standard deviation or by median and the interquartile range. Latter is preferred if the data are scattered and far from normal distribution with the implication that in the sequel non-parametric tests should be favored. For composite events like severe toxicities, progression of disease, and death the number of patients experiencing any of them plus the number in each component should be given, since we have the effect of multiple events. In focus are often variables displaying the time to the first event (e.g. progression of disease which can happen more than once during treatment history). For time driven events in the sequel analysis of general survival times are applied leading to special statistics and graphs. The Kaplan-Meier plot is the most used graph to show time-to-event outcomes as death, time to progression, disease free interval etc. In general the graph displays the steadily increasing difference in incidence rates of the outcome for two or more treatment arms. To make the process clearer, the numbers at risk in each group should be shown at regular time intervals in the time axis. Individuals who did not reach the endpoint are censored (e.g. still alive, lost to follow-up) and should be marked in the plot. The conditional probabilities of Kaplan-Meier statistics indicate the probability of experiencing the endpoint under consideration beyond a certain length of follow-up. Estimation of treatment effects is to measure the magnitude of the difference between treatments on patient outcomes. Normally this is done by a point estimate showing the actual difference observed. Inherent in this kind of statistics is that the bigger the trial, the more precise the point estimate will be. Such uncertainty is usually expressed by a 95% confidence interval in which this percentage of the sample will be found. The primary aim of the study determines the type of estimate required. Namely, there are three main types of outcomes: (a) Binary (dichotomous) response, e.g. dead or alive, progressive or non-progressive, success or failure, respectively. (b) Time to event outcome most measured in intervals, e.g. time from randomization to death, time of inclusion in the study to treatment failure. (c) Quantitative outcome as the reduction of a certain percentage of tumor loads at a given time point (e.g. a seen reduction of 30% after exactly 6 months). Estimates based in percentage are indicated if a binary outcome has to be judged in terms of absence or presence. Then a confidence interval of the proportion of interest can be given. Relative risks are the ratio of two percentages and can be converted to relative risk reduction. Alternatively relative odds can be applied which is a cross-product relationship and shows the relation of chance. Relative risk and relative odds are sometimes called risk ratio and odds ratio instead. The absolute difference in percentage is taken as a measure of absolute risk reduction. Estimates for time-to-event outcomes are used in all survival statistics as time to death, time to progression etc. The Kaplan-Meier plot depicts the first time of the occurrence of the event but does not in itself provide a simple estimate summarizing the treatment difference. The Kaplan-Meier estimate at the end of plotted time or at any other time between can be taken as cumulative rate of the leading event. That is only a time point estimate. Instead, the most common approach is to use a Cox proportional hazards model to obtain a hazard ratio and its 95% confidence interval. The hazard ratio can be thought of as the hazard rate in one group divided by the hazard rate in the other group averaged over the whole follow-up period. Examples from medical trials will be used to explain the statistical principles shown here. References [1] Pocock SJ, McMurray JJV, and Collier TJ: Making sense of statistics in clinical trial reports. J Am Coll Cardiol 2015; 66(23):2648-2662. [2] Pilz LR, Manegold C: Endpoints in lung cancer trials: Today's challenges for clinical statistics. MEMO 2013; 6(2): 92-97.
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
