Bounded outcome score modeling: application to treating psoriasis with ustekinumab

Area of Science:

• Pharmacometrics and Statistical Modeling
• Clinical Trial Analysis
• Dermatology Research

Background:

• Bounded outcome scores (BOS) are frequently used in clinical studies but often exhibit non-standard distributions (e.g., J- or U-shaped).
• Standard statistical methods assuming normality are inappropriate for BOS data, and data transformations can be challenging.
• Existing pharmacokinetic/pharmacodynamic (PK/PD) modeling literature lacks explicit methodologies for handling non-standard BOS distributions.

Purpose of the Study:

• To apply an augmented coarsened latent variable (CO) approach with flexible transformations to analyze BOS data in psoriasis patients.
• To demonstrate the efficacy of ustekinumab in treating moderate to severe psoriasis by assessing its effects on key clinical components.
• To integrate this novel statistical methodology within a semi-mechanistic, mixed-effect exposure-response model.

Main Methods:

• Utilized a coarsened latent variable (CO) approach combined with flexible data transformations.
• Applied the methodology to analyze data from two Phase 3 clinical studies of ustekinumab in moderate to severe psoriasis.
• Integrated the CO approach with a novel semi-mechanistic, mixed-effect exposure-response model, considering placebo effects, disease progression, and dropout.

Main Results:

• The augmented CO approach successfully demonstrated the effects of ustekinumab on four clinical components of psoriasis: involved body surface area, induration, erythema, and scaling.
• An additional transformation, modifying both tails of the standard logit transformation, was found to be necessary for optimal application of the CO method.
• The exposure-response model effectively integrated placebo effects, disease progression, and investigated the influence of patient dropout.

Conclusions:

• The enhanced coarsened latent variable (CO) approach with flexible transformations provides a robust statistical framework for analyzing non-standard BOS data in PK/PD modeling.
• This methodology effectively elucidates the exposure-response relationship of ustekinumab in moderate to severe psoriasis, offering valuable insights for clinical trial analysis.
• The study highlights the importance of tailored statistical methods for complex biological data and contributes a novel analytical tool to the pharmacometric literature.