Challenging from a methodological point are still adaptive trials based on time-to-event endpoints; e. In parallel to methodological developments, sponsors of clinical trials have regularly sought scientific advice SA from the EMA to obtain regulatory feedback on adaptive design proposals.
The aim of our article is to follow up these trials and to investigate whether they have actually been conducted and if they concluded.
We also retrieved the marketing authorisation application MAA supported by the results of the completed trials. Our motivation was to investigate if these trials were still planned using adaptive elements as initially proposed at the time of SA and to evaluate the impact of the adaptive elements on the trial success.
Furthermore, we analysed the time elapsed between SA and trial start. Some regulatory perspectives as regards the implementation of adaptive designs in confirmatory clinical trials are proposed before we finish with a discussion and a conclusion. In the original survey, each of the 59 SA procedures included questions relating to a given active substance whose effect was investigated in a single given adaptive trial.
At the cut-off date of 20 April , we linked each of the corresponding 59 adaptive trials to its medicine name using the EMA internal database in which all the MAAs submitted to the EMA are recorded. All the substances that could not be associated to an MAA in this database were then searched within the Springer AdisInsight database in order to enquire about their current status.
This book will examine current issues and controversies in the design of clinical trials, including topics in adaptive and sequential designs, the design of. A collection of articles that examine the current challenges in the design, performance, . of Clinical Trials Jan 05, The Large Pharmaceutical Company Perspective Ethical issues can arise in the design and conduct of clinical trials.
This registry is a curated database encompassing numerous sources of trials information such as www. The medical condition, the phase of the trial and the name of the company were used as search terms. As in the original survey [ 39 ], some details about the initial SA procedure e. The start and completion years of the completed trials were also reported. For all retrieved trials, available elements indicating the planning of adaptive techniques in the corresponding webpage of AdisInsight or related databases were collected. Substances for which SA about adaptive elements had been requested from the EMA after the trial start were systematically categorised as including adaptations even if not specified in AdisInsight or related databases.
When firstly described during the SA procedure [ 39 ], the 59 trials had been classified by the authors of the original survey in order to distinguish trials for which no concerns were raised by CHMP, trials that required further investigation from the sponsor i. This information was used to cross-tabulate the intended use of adaptive elements in the retrieved trials by their previous acceptance at SA. The time from SA letter to trial start was calculated in a time-to-event analysis for all retrieved trials. Following the ClinicalTrials. Finally, the corresponding Kaplan-Meier curve was plotted.
We investigated whether completed trials i. Some information, e. Information was also searched in ClinicalTrials. Of note, all the trials were sponsored by pharmaceutical companies. The detailed results of the SA trials matching process are given in Fig. No relevant difference which could explain why some of the trials were not retrieved was observed. No other information about the remaining 28 trials and the corresponding active substances was found in the different trials registries.
Of note, 6 of these trials were systematically categorised as including adaptive elements since corresponding SA was requested after study start. The information could not be retrieved for the other 13 remaining products. Note that one given trial could include several different types of adaptions e. A given adaptive element was considered as not implemented if only other types of adaptation were reported. Less standard adaptations, such as changing the primary analysis from a parametric to a non-parametric one or a change of the type of comparator, which were observed during the first SA survey [ 39 ], were never implemented or the corresponding trial was not retrieved.
This shows that planned and actual sample sizes were almost always very close to each other, even when an SSR was initially envisaged.
Interestingly, one of the trials planned with SSR allowed reduction of the sample size from to Dotted line represents the line of equality. Crosses represent trials planned with SSR after SA, whereas triangles represent those which were not. Footnote 2 Two trials included especially interesting adaptations.
The primary endpoint of the trial was the percentage of complete or nearly complete resolution of the haemangioma. An interim analysis was planned stage 1 in order to allow an Independent Data Monitoring Committee IDMC to discontinue the enrolment in the less promising treatment regimen s , to re-assess the sample size based on conditional power or to stop the trial for futility. The type I error rate of the overall procedure was controlled at 0. This small alpha level was suggested in an early discussion with the US FDA in order to provide stronger evidence of the treatment effect since the MA was based on a single pivotal trial see summary of the discussion with the US FDA in [ 41 ].
In total patients were randomised. On the intention-to-treat ITT sample interim analysis was conducted after the first were followed up. The sample size was not increased. More results of this trial as well the statistical analysis plan can be found in [ 40 , 42 ] and their online supplementary material. An extensive discussion about the trial design and logistic issues has been published recently as a book chapter [ 41 ]. In particular, it includes an interesting report on the discussions with regulators.
The primary objective was to choose up to two safer and more efficacious doses and demonstrate the non-inferiority of the high dose of dulaglutide to sitagliptin. The primary endpoint was glycemic control measured by change from baseline measurement of haemoglobin A 1c HbA 1c. This trial was separated into two stages encompassing several adaptive elements.
During the first stage, the randomisation probabilities were updated every 2 weeks as data accrued. When enough information was available, only the two best doses would be pursued to stage 2 with the placebo using a fixed randomisation ratio.
The trial would have to be stopped for futility if no safe and efficacious dose could be chosen. The sample sizes of the first stage and of the final analysis were not fixed and could also be adapted. All these adaptations were performed using a complex Bayesian algorithm relying on a composite measure of efficacy and safety called the clinical utility index CUI , which comprises functions modelling the effect of the medicine on HbA 1c , weight, diastolic blood pressure and heart rate as compared to placebo.
A tree gatekeeping procedure was used to control the type I error rate; however, the extensive adaptive nature of this trial required further investigations [ 44 ]. At the end of the first stage dose finding and adaptive randomisation , two dose regimens were selected. The design and results of this trial have been described extensively in the literature [ 45 , 46 , 47 , 48 ]. Its EPAR is available online [ 49 ]. In the second example this especially enabled one to choose the dose to be further re-investigated in the other jointly submitted phase III trials.
Information on the two other trials having supported an MA can be found in the literature [ 50 , 51 ] as well as on the EMA website [ 52 , 53 , 54 ]. Adaptive trials can also be useful to avoid continuing the development of ineffective drugs. For example, in the four discontinued studies from Fig.
The second trial was a phase III trial in patients with metastatic or locally advanced and unresectable chondrosarcoma with progression-free survival as an endpoint. This information was extracted from AdisInsight, which indicated that the SFF analysis was pre-planned for both trials.
Four-month moxifloxacin-based regimens for drug-sensitive tuberculosis. Statistics in Medicine. Please click here for more information on our author services. The use of adaptive trial designs in clinical research and development has become very popular since then. Search Article search Search. Plan to be flexible: a commentary on adaptive designs. The sponsor should ensure that the IDMC and TSC have members with all the skills needed to implement the AD and set up firewalls to avoid undue disclosure of sensitive information, e.
In this section we discuss the most frequent regulatory comments on the use of adaptive elements raised by the CHMP during SA or the assessment of the centralised procedure, as illustrated with the four medicines having reached MAA. Some items are not specific to adaptive designs but can apply to standard group sequential designs as well. The rationale for designing adaptive trials and the objectives of the adaptations should be made explicit by the sponsors, as in principle the acknowledging of limited knowledge contradicts the confirmatory nature of the trials in the late stage of drug development.
For example, often for phase III a broader patient population is recruited than in initial proof-of-concept and phase II clinical trials, and it may be uncertain whether promising findings of earlier trials can be replicated in the confirmatory studies to the full extent. Sample size re-calculation is then an option to make use of the accumulating information in the ongoing trial to compensate increases in variability or decreases in the treatment effect.
Although adaptive designs can be beneficial to a trial success, the use of such designs bears risks. Applicants should be able to weigh these risks as compared to those of a standard development program. The sponsor was thus requested to explain the rationale for planning an adaptive design versus a standard program of separate trials. The justification was that this would allow investigating more doses. In addition, the trial could be stopped in case of serious safety concerns more quickly since safety data would be monitored regularly across both stages of the trial.
The seamless nature of the trial would also enable one to report more information on patients randomised during the first stage in a dose group to be continued until the end of the trial with a longer duration of treatment. These considerations — which could of course also be designed into separate trials though by convention are not — have to be balanced against:. The principal option to include the outcome of a phase II trial regarding safety in the assessment of findings in the phase III trial.
Some general considerations about trial efficiency once a larger number of doses is included in the phase II part of the trial and the overall decision-making strategy has to include the outcome of the phase II part of the trial. It is seen as a strength of the adaptive approach that all these considerations can be investigated and discussed at the planning stage.