Scholarlink

Background <p>Clinical trials evaluating botulinum toxin A (BoNT-A) for glabellar lines have predominantly relied on binary composite endpoints and cross-sectional comparisons, potentially underrepresenting the full spectrum of treatment efficacy. Despite increasing therapeutic diversity, most Phase 2 and 3 trials use rigid, regulator-defined metrics that may flatten clinical nuance and obscure meaningful differences in onset, duration, or patient variability.</p> Methods <p>We conducted a cross-product reanalysis of 23 pivotal Phase 2 and 3 BoNT-A trials using synthetic individual-level data (<i>n</i> ≈ 10,000), derived via Monte Carlo simulation calibrated from published aggregate statistics. Generalized linear mixed-effects models (GLMMs), parametric survival models, and Bayesian logistic regression were applied to assess efficacy trajectories, treatment durability, and uncertainty under plausible prior assumptions. Missing data were modelled under MAR and MNAR frameworks.</p> Results <p>Model-based analyses suggested higher composite responder rates at Day 30 compared to original reports. Estimated responder rates were: PrabotulinumtoxinA (PRABO) 91.2%, RelabotulinumtoxinA (RELATOBOT) 89.3%, DaxibotulinumtoxinA (DAXI) 86.4%, OnabotulinumtoxinA (ONA) 82.6%, IncobotulinumtoxinA (INCO) 79.5%, LetibotulinumtoxinA (LETI) 77.1%, and AbobotulinumtoxinA (ABO) 76.2%. Simulated median durations of response ranged from 168 days (PRABO) to 122 days (ABO). Bayesian models produced posterior odds ratios (ORs) reflecting high probabilities of efficacy relative to placebo, ranging from OR 92.4 for PRABO to OR 36.5 for ABO. GLMMs identified substantial heterogeneity in response patterns, including early-peak and sustained-response subgroups across products. Sensitivity analyses under MNAR assumptions showed reductions in estimated responder rates of 5–8% points, particularly for formulations with higher dropout in source trials.</p> Conclusions <p>Legacy BoNT-A trial designs—while regulatorily compliant—fail to reflect the complexity of aesthetic response. By applying contemporary modelling frameworks, we demonstrate the potential to surface richer efficacy profiles, reframe responder definitions, and elevate methodological standards across aesthetic trials.</p> <p>Level of evidence: Level I, therapeutic study.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Beyond binary endpoints: modernizing aesthetic trial analytics: Simulation-Based reappraisal of phase 2/3 BoNT-A studies using advanced statistical models

  • 期刊论文

  • 发表日期：

    2025年8月29日

查看全文
  • Eqram Rahman,
  • Parinitha Rao,
  • Alexander D Nassif,
  • William Richard Webb,
  • Woffles TL Wu,
  • Jean D A Carruthers

摘要

Background

Clinical trials evaluating botulinum toxin A (BoNT-A) for glabellar lines have predominantly relied on binary composite endpoints and cross-sectional comparisons, potentially underrepresenting the full spectrum of treatment efficacy. Despite increasing therapeutic diversity, most Phase 2 and 3 trials use rigid, regulator-defined metrics that may flatten clinical nuance and obscure meaningful differences in onset, duration, or patient variability.

Methods

We conducted a cross-product reanalysis of 23 pivotal Phase 2 and 3 BoNT-A trials using synthetic individual-level data (n ≈ 10,000), derived via Monte Carlo simulation calibrated from published aggregate statistics. Generalized linear mixed-effects models (GLMMs), parametric survival models, and Bayesian logistic regression were applied to assess efficacy trajectories, treatment durability, and uncertainty under plausible prior assumptions. Missing data were modelled under MAR and MNAR frameworks.

Results

Model-based analyses suggested higher composite responder rates at Day 30 compared to original reports. Estimated responder rates were: PrabotulinumtoxinA (PRABO) 91.2%, RelabotulinumtoxinA (RELATOBOT) 89.3%, DaxibotulinumtoxinA (DAXI) 86.4%, OnabotulinumtoxinA (ONA) 82.6%, IncobotulinumtoxinA (INCO) 79.5%, LetibotulinumtoxinA (LETI) 77.1%, and AbobotulinumtoxinA (ABO) 76.2%. Simulated median durations of response ranged from 168 days (PRABO) to 122 days (ABO). Bayesian models produced posterior odds ratios (ORs) reflecting high probabilities of efficacy relative to placebo, ranging from OR 92.4 for PRABO to OR 36.5 for ABO. GLMMs identified substantial heterogeneity in response patterns, including early-peak and sustained-response subgroups across products. Sensitivity analyses under MNAR assumptions showed reductions in estimated responder rates of 5–8% points, particularly for formulations with higher dropout in source trials.

Conclusions

Legacy BoNT-A trial designs—while regulatorily compliant—fail to reflect the complexity of aesthetic response. By applying contemporary modelling frameworks, we demonstrate the potential to surface richer efficacy profiles, reframe responder definitions, and elevate methodological standards across aesthetic trials.

Level of evidence: Level I, therapeutic study.