An investigation to improve a nonlinear mixed-effects approach for EC50 estimation based on multi-donor dose-response...

Area of Science:

• Pharmacology and Toxicology
• Biostatistics
• Computational Biology

Background:

• Dose-response relationships are crucial for evaluating compound efficacy and potency.
• The 4-parameter logistic (4-PL) model is commonly used, with EC50 being a key metric for potency.
• Estimating overall EC50 from multi-donor data presents challenges, with existing methods like meta-analysis and nonlinear mixed-effects having limitations.

Purpose of the Study:

• To propose a modified nonlinear mixed-effects approach for robust EC50 estimation from multi-donor dose-response data.
• To address convergence failures associated with traditional nonlinear mixed-effects models.
• To compare the performance of the proposed method against the meta-analysis approach.

Main Methods:

• Utilized the stochastic approximation expectation-maximization (SAEM) algorithm for parameter estimation.
• Implemented multiple starting points to ensure global optimum search for model parameters.
• Applied the 4-parameter logistic (4-PL) model to dose-response data.

Main Results:

• The proposed SAEM-based nonlinear mixed-effects approach significantly reduces convergence failures, even with a small number of donors (n=3).
• Achieved smaller absolute median bias and improved 95% confidence interval coverage probability compared to meta-analysis for n ≥ 7 donors.
• Demonstrated robustness and improved accuracy in EC50 estimation from multi-donor dose-response data.

Conclusions:

• The SAEM algorithm offers a viable solution for overcoming convergence issues in nonlinear mixed-effects modeling for dose-response data.
• This modified approach provides a more reliable and accurate method for estimating population EC50 values.
• The findings suggest this method is advantageous for analyzing multi-donor dose-response studies, particularly when donor numbers vary.