• Design »
  • Simulation »
  • Analysis »
  • Screenshots »
  • The accurate estimation of the maximum tolerated dose is of critical importance in the drug development process.  ADDPLAN® DF supports the decision process in the design and analysis of dose finding trials using the most innovative methodology. Uncertainty in the true underlying dose-toxicity profile may be taken into account when simulating dose escalation trials using ADDPLAN® DF. Drug developers may quickly study the operating characteristics of standard and innovative dose escalation methods under different scenarios, allowing the selection of the most appropriate method for successful dose escalation.

ADDPLAN® DF is the first validated design, simulation and analysis software for the Multiple Comparison Procedure and Modeling (MCPMod) approach.

Poor quality learning in early phase of drug development has been identified by Regulatory agencies and pharmaceutical companies as a root cause for late phase attrition. Improved dose selection in both Phase I and Phase II trials is considered to be essential to increase R&D efficiency and effectiveness.

Uncertainty about the true underlying dose-response model should be taken into account when designing Phase II dose finding studies. ADDPLAN® DF enables the study designers to answer questions on the most informative doses for the study, using the theory of optimum experimental design. Efficient test strategies allow the combination of data from different dose levels, improving the power of the study, while adding insight into the underlying dose-response relation. ADDPLAN® DF provides the fastest tool to evaluate innovative dose finding designs, including the MCPMod approach. A wide range of operating characteristics provides valuable information on limitations and benefits of the examined study designs.

Procedures for MCPMod, classical dose-response modelling and signal testing:

  • All functionality available for means, rates and count data
  • Sample size calculation for model-based trend tests in dose finding
  • Design optimization:
    • Selection of most informative study doses
    • Efficiency calculation: Comparison of candidate designs to optimal designs
    • Design calculations taking multiple possible dose-response scenarios simultaneously into account
  • Simulation functionality for the fine tuning and reporting of the study design, evaluating:
    • Precision of dose estimation
    • Probability of correct model selection
    • Modelling precision
    • Success probability
  • Analysis functionality conducting:
    • Trend tests
    • (Frequentistic/Bayesian/Bootstrap) Modelling
    • Dose estimation & prediction
  • Dose Escalation Designs:
    • Simulation of operating characteristics for dose escalation designs, including:
    • 3+3 and Modified Toxicity Probability Interval (mTPI) as comparator
    • Flexible CRMs, with/without overdose control introduced by Novartis statistical group (nCRM), 2-parametric & 1-parametric Bayesian logistic modelling given a wide range of prior-distributions.
  • Adaptive analysis of dose-escalation designs providing:
    • Recommendation for next cohort dosing
    • Posterior probability of overdosing
    • Posterior uncertainty on MTD location
  • Recruitment modelling - Given recruitment & endpoint assumptions insight on:
    • When will the interim analysis take place?
    • How many patients will have reached their endpoint?
    • How many patients are already in the study?
  • Sample size estimation / Power calculation for dose response modelling
    • Targeting length of confidence intervals and probability of correct dose estimation
  • Adaptive MCPMod Simulation and Analysis:
    • Stage-wise adaptive randomization:
  • Optimal design theory: Targeting most informative design
  • Best intention – Targeting allocation to target dose
  • Scenario based: Intuitive allocation filling the “gaps” in the curve
    • Futility stopping using:
  • Model-based trend tests
  • Pairwise comparisons
  • Power to reach targeted difference
    • Ongoing study simulation:
  • The interim analysis generates a simulation file, which may be used to predict the outcome of the study, given all available data in the interim.


1 EMA-CHMP (2014) Qualification Opinion of MCP-Mod as an efficient statistical methodology for model-based design and analysis of Phase II dose finding studies under model uncertainty. EMA/CHMP/SAWP/757052/2013.

2 FDA Drug Development tools: Fit-For-Purpose Initiative: May 26, 2016


Dose finding designs can be constructed in ADDPLAN DF using multiple criteria. ADDPLAN DF is specially designed to support internal discussions between clinicians and statisticians, to develop trial designs that take into account knowledge and uncertainties on the underlying dose-response behavior. The resulting designs will be efficient for the detection of drug related effects and the estimation of the MED for all considered dose-response scenarios.
Sample size calculations for a proof of concept (PoC) can be conducted in ADDPLAN DF with respect to different trend tests:

  • Pairwise comparisons using Dunnett-contrasts
  • Williams- and Marcus-contrasts taking monotonicity assumptions on the dose-response into account
  • Model-based contrasts, maximizing the statistical power for the detection of specified dose-response effects
  • And any other trend test of interest via User defined contrasts

Allocation ratios for an improved dose-response modelling can be optimized using different optimality criteria:

  • D-optimality for the minimization of the generalized variance of the dose-response estimation
  • TD/ED-optimality for the minimization of the variance of the  dose estimation
  • D&TD/ED-optimality for combining both criteria

The efficiency of candidate designs may be evaluated using the optimal design functionality.

Samples sizes and allocation rates are calculated taking into account the uncertainty on the underlying dose-response behavior. Different effect-scenarios may be specified and weighted to provide efficient and effective study designs.

Study designs may be optimized for a normal, binary and overdispersed-count data.


Study designs are based on dose-response assumptions. Inaccurate assumptions in the design stage can impact the appropriateness of the study design and are best verified using simulations. ADDPLAN DF provides simulation functionality for:

  • The PoC using Model-based, Dunnett, Williams, Marcus and User-defined contrasts
  • The dose-response modelling and the estimation of the targeted dose using modelling procedures
  • Both targets under one umbrella using the MCP-Mod-methodology
  • The estimation of the MTD using adaptive dose-escalation designs

ADDPLAN DF allows users to study the operating characteristics of the chosen analysis and design methods for multiple scenarios with a table of simulation outputs and plotting functionality. These computation results may also be exported to other software products. The resulting insight in the appropriateness of the selected methods leads to optimized trial designs for determining the target dose of interest. The ADDPLAN DF simulation functionality is available for normal, binary and overdispersed-count data.


ADDPLAN DF supports the analysis of Phase I and Phase II trials for determining the maximum tolerable dose (MTD), minimum effective dose (MED), modelling the dose-response relation and additionally for proving the existence of drug related effects. The analysis functionality in ADDPLAN DF allows the analysis for:

  • The PoC using Model-based, Dunnett, Williams, Marcus and User-defined contrasts.
  • The modelling of the dose-response relation and the estimation of the MED using modelling procedures, including Frequentist and Bayesian approaches
  • The combined PoC and modelling using the MCP-Mod-methodology
  • The estimation of the MTD using continual reassessment methods (CRM, including nCRM), probability toxicity interval approaches or the standard 3+3-design.

ADDPLAN DF presents test results, dose estimates and proposals for the adaptive allocation in a user friendly way for normal, binary and over disperse-count data. The graphical output provides additional insight into the estimated dose-response behavior.


Dispersion of dose-estimates in dependence on the underlying effect difference.
Calculating the required sample size for proving dose-response-effects
Dose-response assumptions and optimized design weights at dose groups.
Optimized allocation to dose groups taking uncertainty on the dose-response into account.
Dependence of the model-selection on the underlying effect difference.
Specification of the candidate set of dose-response models
Verifying the appropriateness of the design and analysis methods under different dose-response scenarios.
Analyzing the dose-response with innovative statistical methods.
Display of the fitted dose-response relation.
Adaptive analysis of dose-escalation trials.
Display of the optimized contrasts for detecting the dose-response.
Display of the selected candidate dose-response shapes.