Table 2

Main features of trial designs applied to personalised medicine

Feature domainsFeatures
Inference frameworkBayesian
Model*Disease progression*
Control groupCommon/shared†
RandomisationWith treatment randomisation in both biomarker-positive and biomarker-negative subgroups
Without treatment randomisation in the biomarker-negative subgroup¶
Only for patients with discordant clinical and genomic risk evaluation**
Randomisation in the non-biomarker based strategy armWith treatment randomisation
Without treatment randomisation††
Reverse biomarker strategy‡‡
Subgroup specificSequential subgroup specific§§
Parallel subgroup specific¶¶
Biomarker positive and overall strategies***With sequential assessment
With parallel assessment
With fall-back analysis†††
Marker sequential test‡‡‡
Biomarker assessmentWith biomarker assessment in the entire population
Without biomarker assessment in the control arm
Personalised medicine (PM) specific adaptive aspects§§§Adaptive enrichment
Adaptive signature
Threshold determination¶¶¶
Generic adaptive aspectsAdding a new arm
Early stopping****
Interim analysis††††
Outcome-based adaptive randomisation
Sample size reassessment
Treatment tailoring aspectsPharmacodynamic biomarker assessment after run-in phase period‡‡‡‡
Dynamic treatment regime§§§§
  • *Model used for analysis. A disease progression model takes into account the patient disease state and other patient baseline characteristics for charactering patient clinical outcome(s).44 Longitudinal model permits including in the analysis the partial information of patients who have not yet reached their final outcome at an interim analysis.44

  • †A common/shared control group can be used in a trial design in which multiple treatments are being tested, instead of each treatment having its own control arm.

  • ‡If patients in the common/shared control group receive a ‘Standard of care’ that may change over time or the profile of the patients enrolled on the trial may change over time, a trial design can use a contemporaneous control group meaning that the comparison of treatment’s effects may be restricted to those patients who were enrolled/randomised in the same period as those patients who were allocated to the treatment.

  • §If a comparison group is not available in the existing trial or substudy or at the same time but in a different setting, a trial design can use a historical control consisted of a group of individuals treated in the past.

  • ¶Patients in the biomarker-negative subgroup receive the control treatment.

  • **Only patients with discordant results (ie, either high clinical risk an low genomic risk or low clinical risk and high genomic risk) are randomly assigned to either the control or intervention arm.

  • ††Patients, which are randomly assigned to the non-biomarker-based strategy arm, receive the control treatment.

  • ‡‡Patients which are randomly assigned to reverse-based strategy receive the control treatment if they are biomarker-positive and the experimental treatment if they are biomarker-negative.

  • §§Study designs testing the treatment effect first in the biomarker-positive subpopulation and if the result is positive in the biomarker-negative subgroup.

  • ¶¶Study designs testing the treatment effect in both biomarker-positive and biomarker negative subgroups simultaneously.

  • ***Study designs testing the treatment effect in the entire study population and in the biomarker-positive subgroup separately.

  • †††Study designs testing the treatment effect in the overall population and in the biomarker-positive subgroup sequentially.

  • ‡‡‡Study designs testing the treatment effect not only in the biomarker-positive and biomarker-negative subgroups but also in the entire population sequentially.

  • §§§PM-specific adaptive aspects could be used to stratify the patients to the treatment. Generic adaptive aspects could be considered when planning a PM trial, but they could be also found in fields outside PM.

  • ¶¶¶A threshold is used to divide the population into ‘biomarker positive’ and ‘biomarker negative’.

  • ****A trial arm or clinical trial is stopped early due to pre-specified rules related to treatment efficacy and safety risk.

  • ††††Interim analyses are pre-planned analyses, which use accumulating data in order to make an early decision or adaptation.

  • ‡‡‡‡All patients receive the new treatment for a run-in period and then are classified as either biomarker positive or negative using a pharmacodynamics biomarker.45

  • §§§§A dynamic treatment regime consists of a sequence of individually tailored therapies during the course of a treatment.

  • ¶¶¶¶Models to suggest optimal dosage regimes of drugs for individual patients.46

  • PK/PD, Pharmacokinetic/pharmacodynamic.