B-cell indirect-response estimation

Why EC50 is hard to estimate for a B-cell depleting agent

Index of the documents in the bcell-indresp-est project: the identifiability analysis with its simulations, and its references.

Published indirect-response models of B-cell depletion report an \(E_{\max}\) near 150 and an \(EC_{50}\) that is either wide, fixed from in vitro data, or abandoned for a slope. This project works out why: the information that separates the two parameters sits in the first three days after dosing and in a concentration band two orders of magnitude below \(EC_{50}\), and a study that samples B cells weekly at therapeutic doses has neither. The simulations run on every render.

Documents

• EC50 identifiability. The model, the arithmetic of a large \(E_{\max}\), where the information lives, expected precision by design from the Fisher information, the likelihood ridge, the published record, the slope parameterization, and what a study needs if it wants \(EC_{50}\). Its In brief section is one screen and says what the rest argues.
• References. The two classic papers on \(E_{\max}\) estimation from truncated data, the indirect-response model sources, and the five published B-cell depletion models, each with a marker recording whether the claim drawn from it has been checked.

The ITP-PK-Platelet project is where the question came from; its Section 6.2 is the short version.